Tissue‐Specific Regulation of Fatty Acid Metabolism in a Mouse Model of Isolated Complex I Deficiency

ABSTRACT

Isolated complex I deficiency (ICD) is commonly associated with mitochondrial diseases and closely mimics subacute necrotising encephalomyelopathy. This disorder is characterised by metabolic perturbations that affect energy metabolism pathways, including fatty acid metabolism. Here, we examined the tissue‐specific changes in fatty acid metabolism in the Ndufs4 KO mice by employing mass‐spectrometry‐based proteomics as a hypothesis‐generating approach. We investigated proteomic changes in six tissues, including brain regions (brainstem, cerebellum, olfactory bulb), heart, kidney and liver, focusing on proteins involved in fatty acid metabolism. Although it is expected that most tissues, except for the brain, will utilise fatty acids as alternative energy sources when oxidative phosphorylation (OXPHOS) is deficient, our data revealed a more complex response. In the liver, fatty acid consumption (oxidation) was favoured as expected, but in the heart, fatty acid synthesis was favoured. In the kidney, proteins involved in almost all fatty acid metabolic processes (oxidation and synthesis) were downregulated. Our data demonstrate that metabolic adaptations in fatty acid metabolism to ICD were tissue‐specific and often in opposing directions. Understanding the differential adaptations across tissues could inform future treatment targets for mitochondrial disorders.

1. Introduction

Mitochondria are double‐membrane organelles essential for the production of energy for aerobic cellular respiration via the oxidative phosphorylation (OXPHOS) system. In addition to their primary role of energy production, mitochondria serve as the bioenergetics hubs of numerous metabolic pathways, including amino acid metabolism, β‐oxidation of fatty acids, ketogenesis, the TCA cycle and the urea cycle. The critical role of mitochondria in regulating these metabolic pathways is evident from the wide array of mitochondrial diseases that result from mitochondrial dysfunction [1].

Among these diseases, isolated complex I deficiency (ICD) is one of the most common disorders and is associated with variable onset and clinical symptom manifestation [2]. ICD most frequently manifests as Leigh syndrome (LS, OMIM 256000), a subacute necrotising encephalomyelopathy‐like phenotype, characterised by neuromuscular decline and spongiform lesions in specific basal ganglia and brainstem [3, 4]. In addition, multi‐systemic clinical manifestations such as cardiomyopathy [5, 6], renal tubulopathy [7, 8] and hepatopathy have been associated with ICD [9, 10, 11]. Mutations in the gene encoding NADH: ubiquinone oxidoreductase subunit 4 (NDUFS4) of complex I (CI) of the OXPHOS system have been associated with LS [12, 13]. Previous metabolomics studies have proposed that defects in CI result in a decrease in NAD⁺/ NADH ratio, as observed in mitochondrial disease animal models [14] and patients presenting with Leigh syndrome, the French Canadian variant [15] which in turn leads to the dysregulation of a number of bioenergetics pathways involved in energy metabolism [16]. Among the energy metabolic processes and pathways, fatty acid metabolism is frequently altered in ICD patients and disease models [17, 18, 19, 20]. In a recent study on Ndufs4 knock‐out (KO) mice [21], demonstrated metabolic adaptation to fatty acid metabolism in specific brain regions (brain slice, cerebellum, cerebral cortex, hippocampus, inferior colliculus, superior colliculus), compared to wild‐type (WT) mice. Fatty acid metabolism comprises several distinct processes of fatty acid transport, anabolism and catabolism, found in the cytoplasm and organelles such as the peroxisomes, mitochondria, lysosomes and endoplasmic reticulum (ER). Fatty acid transportation between organelles and the cytoplasm is facilitated by proteins such as acyl‐coenzyme A thioesterases (ACOTs) that remove acyl‐CoA [22], and carnitine O‐octanoyltransferases (carnitine shuttles) that add carnitines to fatty acids before transport into the mitochondrion [23]. Generally, long‐chain fatty acids are shortened during α‐ and/or β‐oxidation in peroxisomes to medium or short‐chain fatty acids, which can then either continue with peroxisomal β‐oxidation or enter the mitochondria to undergo further β‐oxidation in the matrix. The intermediates and end‐products of β‐oxidation can be utilised in various anabolic processes or serve as alternative fuel sources [24]. Recent studies have reported alterations in fatty acid metabolism of brain regions in LS models [21, 25], however, proteomic changes in other organ systems remain unexplored. Since ICD and LS frequently present as multi‐systemic disorders, we focus here on tissue‐specific adaptive changes of proteins involved in fatty acid metabolism in brain regions not previously reported on, as well as the heart, liver and kidney. To do this, we extracted data for a list of proteins related to fatty acid metabolism from an untargeted proteomics dataset. We show varied tissue‐specific metabolic adaptations to ICD, with compensatory responses in the olfactory bulb, liver, and heart to some extent, and dysfunctions in the kidney of the Ndufs4 KO mice. Understanding how different tissues are affected by ICD could help in unravelling the complicated clinical presentation of mitochondrial disorders and help to identify potential therapeutic targets.

2. Methods

2.1. Animals and Tissues Sampling

All animal experiments were approved by the NWU Animal Research Ethics Committee (ethics clearance number: NWU‐00764‐23‐A5). Mice were housed in a specific pathogen‐free unit of the vivarium (SAVC reg. no. FR15/13458) of the Pre‐Clinical Drug Development Platform (PCDDP; NWU, RSA). Mice were maintained at a temperature of 20°C, exposed to a 12‐h day/light rhythm, and given ad libitum access to food and water. Ndufs4 KO breeding pairs were obtained from Jackson Laboratory (JAX stock #027058) and heterozygous males and females Ndufs4 mice (B6.129S4‐Ndufs4tm1.1Rpa/J) were interbred to generate Ndufs4 KO and WT mice according to previously described methods [26]. Genotype confirmation was carried out on DNA isolated from tail snips using a polymerase chain reaction (PCR). For proteomics analyses, nine male Ndufs4 KO mice along with nine age‐ and sex‐matched WT controls were euthanised on postnatal Day 45–48 by cervical dislocation under the supervision of trained veterinarians. Tissues excised included biological replicates from three brain regions (brainstem, cerebellum and olfactory bulb), heart, kidney and liver. The collected tissues were immediately snap‐frozen and transported in liquid nitrogen (LN₂), before being stored at −80°C until further use. Only male animals were used to mitigate sex‐related variations, which may affect downstream interpretation.

2.2. Protein Preparation and On‐Bead Protein Digestion

Sample processing and analysis were conducted at the Council for Industrial and Scientific Research, Pretoria, South Africa. Frozen tissue samples were thawed and resuspended in a lysis buffer (4% sodium dodecyl sulphate [SDS] in Tris‐HCl, pH 8.0, with protease inhibitor) and sonicated for 30 min. The lysate was centrifuged at 15,000 x g for 10 min. After centrifugation, total protein concentration was determined by the BCA assay, according to the manufacturer's instructions. Total proteins from each tissue sample were reduced with 5 mM TCEP and alkylated with 10 mM CAA at room temperature for 20 min. Thereafter, proteins were desalted and cleared of debris using the MagResyn HILIC beads (Resyn Biosciences) according to methods previously described by Nweke et al. [27]. Proteins were digested with a sequencing‐grade trypsin at 37°C for approximately 5 h using a protease‐to‐protein ratio of 1:20. The generated peptides were vacuum‐dried overnight and reconstituted with formic acid. Data analysis included nine individual mice per genotype, except for the olfactory bulb, which included only eight samples for Ndufs4 KO due to contamination that occurred during sample preparation.

2.3. MS/MS Analysis

Tandem mass spectrometry (MS/MS) was performed using the EvoSep One LC System, coupled with a 5600 Sciex Triple TOF mass spectrometer. Approximately 750 ng of the peptides were loaded on a CSH C18 NanoEase analytical column (15 cm length, 75 µm diameter, 1.7 µm particle size, Water Acquity) and separated at a flow rate of 100 nL/min over a 30 min gradient (solvent A: 0.1% FA; solvent B 80%MeCN/0.1% FA). The separated peptides were infused into the ion source for electrospray ionisation using a potential energy of 3000 V. The mass spectrometer was run under the Sequential Window Acquisition of all THeoretical Mass Spectra (SWATH‐MS) data acquisition method. For the SWATH‐MS, precursor scans were acquired from 400 to 1100 m/z with a 50 ms accumulation time. Fragment ions were acquired from 200 to 1800 m/z for 48 variable‐width precursor windows with 0.5 Da overlap between windows and 20 ms accumulation time per window.

2.4. MS/MS Data Processing

Raw SWATH‐MS data were processed and analysed using Spectronaut v17 Software (Biognosys) with default direct DIA identification and quantification settings. For database searching, Swiss‐Prot Mouse sequences (downloaded on 16 August 2023 from www.uniprot.org), along with common contaminating proteins, were used. Carbamidomethylation on cysteine was selected as the fixed modification, whilst oxidation on N‐terminal and methionine was chosen as variable modifications, respectively. For identification, a q‐value ≤0.01 cut‐off was applied at the precursor, peptide and protein levels, and with at least one unique peptide. Quantification was performed at the MS1 and MS2 levels, and label‐free cross‐run normalisation was employed using a global normalisation strategy. This involved normalising individual runs based on their global quantity metric towards the overall experimental quantity. Normalisation was performed independently for each tissue. All Spectronaut parameters used for the data processing are included in the .sne file, which has been submitted and deposited to the ProteomeXchange Consortium via PRIDE partner repository with the dataset identifier PXD061439.

2.5. Statistical Analysis

A list of proteins involved in fatty acid and related metabolism was compiled from pathways found in Fatty acid metabolism (R‐MMU‐8978868) and related (R‐MMU‐74182) events from Reactome (https://reactome.org/), as well as previous studies [17, 28]. The list of identified proteins from an untargeted dataset from each tissue was searched for proteins related fatty acid metabolism. We performed quantitative statistical analysis by a two‐tailed unpaired Students t‐test on the Log₂ transformed average ratios of the peptide intensities for each protein to compare the protein expression levels between the Ndufs4 KO and WT mice across different tissues. The resulting p values from this analysis were adjusted for multiple‐hypothesis testing utilising the Benjamini & Hochberg method. Differentially expressed proteins were identified as those with an adjusted p value ≤0.05 and Log₂ fold change (FC) ≥0.58. Comprehensive results from this analysis are provided in Table S1. Subsequently, the proteomic expression profiles were compared across the tissues. Volcano plots were generated using R (version 4.4.1) with the Enhanced Volcano package.

3. Results

3.1. Comparative Expression of Proteins Involved in Fatty Acid Metabolism

We compared the expression profiles of proteins involved in fatty acid metabolism of various tissues of Ndufs4 KO mice to WT mice, comprising several processes where fatty acids are either synthesised (fatty acid biosynthesis and peroxisome fatty acid synthesis pathways), broken down (mitochondrial‐ and peroxisomal fatty acid oxidation pathways) or shuttled between different cellular compartments (carnitine and peroxisome carnitine shuttle systems). As expected, the NDUFS4 subunit, along with a majority of other CI subunits, was less abundant except for the assembly factor, NDUFAF2, which showed increased abundance (Figure 1) in all the tissues of Ndufs4 KO mice compared to WT mice, corroborating findings from a previous quantitative proteomics study [29]. Since CI is the major oxidation site of NADH, this decrease in functional CI units is expected to lead to a disturbed redox balance (NAD⁺/NADH ratio). The NAD⁺/NADH ratio is a major regulatory factor, particularly of the TCA cycle [30], which recruits several NAD⁺ dependent dehydrogenases and, under normal conditions, produces NADH and FADH₂ in a 3:1 ratio. In comparison, mitochondrial β‐oxidation results in a 1:1 ratio of these two reducing equivalents. During CI deficiency, when NADH cannot be oxidised sufficiently, a shift to β‐oxidation of fatty acids could lessen the NADH redox burden, whilst still supporting oxidative ATP production via FADH₂ oxidation by acyl‐CoA dehydrogenases, which donate electrons directly to the coenzyme Q₁₀ (CoQ) cycle. We thus expected perturbations in fatty acid catabolism in Ndufs4 KO mice. We observed a tissue‐specific differential expression of proteins involved in both fatty acid oxidation (catabolic) and fatty acid synthesis (anabolic) in response to Ndufs4 deficiency (Table 1).

FIGURE 1.

Volcano plots of differentially expressed proteins across tissues of the Ndufs4 KO mice. The dotted vertical and horizontal lines indicate the cut‐off values of Log₂ fold change ≥ 0.58 and Q‐value ≤ 0.05, respectively. Subunits of complex I (CI) of the OXPHOS system and some proteins related to fatty acid metabolism are labeled in the plots.

TABLE 1.

Fatty acid metabolism protein expression profiles in different tissues of the Ndufs4 KO mice.

Pathway	Protein (enzyme)	Description	BS	CB	OB	Heart	Kidney	Liver
Peroxisomal fatty acid oxidation	HACL1	2‐hydroxyacyl‐CoA lyase 1	nd	nd	nd	nd	↑	↑
	HAO2	2‐Hydroxyacid oxidase 2	nd	nd	nd	nd	↑	nd
	ALDH3A2	Aldehyde dehydrogenase family 3 member A2	ns	ns	ns	↑	ns	ns
	AMACR	Alpha‐methylacyl‐CoA racemase	nd	nd	nd	↑	↓	ns
	ACOX1	Peroxisomal acyl‐coenzyme A oxidase 1	ns	nd	ns	↑	↓	↑
	ACOX2	Peroxisomal acyl‐coenzyme A oxidase 2	—	—	—	nd	↓	ns
	ACOX3	Peroxisomal acyl‐coenzyme A oxidase 3	nd	nd	nd	nd	↓	nd
	EHHADH	Peroxisomal bifunctional enzyme	ns	nd	ns	↑	↓	↑
	ACAA1B	3‐ketoacyl‐CoA thiolase B, peroxisomal	—	—	—	—	ns	↑
	HSD17B4	Peroxisomal multifunctional enzyme type 2	ns	ns	ns	↑	ns	↓
	SCP2	Sterol carrier protein 2	ns	nd	ns	↑	ns	↓
	NUDT19	Acyl‐coenzyme A diphosphatase	—	—	—	ns	↓	↑
	ACOT4	Peroxisomal succinyl‐coenzyme A thioesterase	—	—	—	—	ns	↑
	ACOT8	Acyl‐coenzyme A thioesterase 8	nd	nd	nd	nd	↓	nd
	ACOT3	Acyl‐coenzyme A thioesterase 3	nd	nd	nd	nd	nd	↑
	ACOT12	Acetyl‐coenzyme A thioesterase 12	nd	nd	nd	nd	↓	↑
	ACOT13	Acyl‐coenzyme A thioesterase 13	ns	nd	ns	ns	ns	↑
Peroxisome carnitine shuttle	CROT	Peroxisomal carnitine O‐octanoyltransferase	nd	nd	nd	nd	↓	↑
Carnitine shuttle system	SLC22A5	Organic cation/carnitine transporter 2	—	—	—	—	↑	—
	CPT1B	Carnitine O‐palmitoyltransferase 1, muscle isoform	nd	nd	nd	ns	↑	nd
	CPT2	Carnitine O‐palmitoyltransferase 2, mitochondrial	ns	ns	↑	ns	ns	ns
Mitochondrial fatty acid oxidation	PDK1	Pyruvate dehydrogenase kinase isozyme 1, mitochondrial	ns	nd	ns	ns	↑	↑
	ACOT2	Acyl‐coenzyme A thioesterase 2, mitochondrial	ns	nd	ns	ns	ns	↑
	ACADL	Long‐chain specific acyl‐CoA dehydrogenase, mitochondrial	ns	ns	↑	ns	ns	↑
	ACADM	Medium‐chain specific acyl‐CoA dehydrogenase, mitochondrial	ns	ns	ns	ns	↓	↑
	ACAA2	3‐ketoacyl‐CoA thiolase, mitochondrial	ns	—	—	ns	↑	ns
	ACADS	Short‐chain specific acyl‐CoA dehydrogenase, mitochondrial	ns	↑	ns	ns	ns	ns
	ACAD10	Acyl‐CoA dehydrogenase family member 10	ns	nd	nd	ns	ns	↑
	ACAD11	Acyl‐CoA dehydrogenase family member 11	nd	nd	nd	↑	ns	↑
Citrate shuttle	SLC25A1	Tricarboxylate transport protein, mitochondrial	ns	ns	↑	↑	ns	ns
Cytosolic fatty acid biosynthesis	ACACA	Acetyl‐CoA carboxylase α	ns	nd	ns	↑	nd	ns
	ACACB	Acetyl‐CoA carboxylase β	nd	nd	nd	↑	ns	ns
	ACLY	ATP‐citrate synthase	ns	nd	ns	↑	ns	ns
	FASN	Fatty acid synthase	ns	nd	ns	↑	↓	ns
Mitochondrial fatty acid biosynthesis	MECR	Enoyl‐[acyl‐carrier‐protein] reductase, mitochondrial	ns	ns	↑	↑	ns	ns
	CBR4	3‐oxoacyl‐[acyl‐carrier‐protein] reductase	ns	ns	↑	ns	ns	ns
	ACSF2	Medium‐chain acyl‐CoA ligase, mitochondrial	ns	ns	ns	↑	↑	ns
Peroxisome fatty acid synthesis	HSD17B12	Very‐long‐chain 3‐oxoacyl‐CoA reductase	ns	ns	ns	ns	ns	↓
	PECR	Peroxisomal trans‐2‐enoyl‐CoA reductase	nd	nd	nd	↑	↓	ns
	TECRL	Very‐long‐chain enoyl‐CoA reductase	nd	nd	ns	↑	nd	nd
	ACBD7	Acyl‐CoA‐binding domain‐containing protein 7	nd	nd	↑	—	—	—
	HACD2	Very‐long‐chain (3R)‐3‐hydroxyacyl‐CoA dehydratase 2	ns	↓	↑	ns	nd	ns
Cytochrome P450 monooxygenase	CYP1A2	Cytochrome P450 1A2	nd	nd	nd	nd	nd	↓
	CYP2J6	Cytochrome P450 2J6	—	—	—	—	↑	nd
	CYP4A10	Cytochrome P450 4A10	—	—	—	—	↑	↑
	CYP4A14	Cytochrome P450 4A14	nd	nd	nd	nd	↑	↑
	CYP4B1	Cytochrome P450 4B1	nd	nd	nd	ns	↓	nd
	CYP8B1	7‐alpha‐hydroxycholest‐4‐en‐3‐one 12‐alpha‐hydroxylase	—	—	—	—	nd	↓
Ketone body metabolism	HMGCS2	Hydroxymethylglutaryl‐CoA synthase	ns	nd	ns	nd	↑	↑
	ACSS3	Acyl‐CoA synthetase short‐chain family member 3	nd	nd	nd	ns	↑	↑
	BDH1	D‐beta‐hydroxybutyrate dehydrogenase	ns	ns	ns	↑	ns	ns
	AACS	Acetoacetyl‐CoA synthetase	↑	nd	ns	—	↓	↓
	OXCT1	Succinyl‐CoA:3‐ketoacid coenzyme A transferase 1	ns	ns	↑	ns	ns	nd

Note: Upward and downward represent differentially upregulation and downregulated, respectively, based on q‐value ≤ 0.05 and Log2 fold change (FC) ≥0.58. Dashes denote proteins that are not expressed in the corresponding tissue (https://www.ebi.ac.uk/gxa/experiments/E‐PROT‐13).

Abbreviations: BS, brainstem; CB, cerebellum; nd, not detected; ns, not significant; OB, olfactory bulb.

3.2. In the Liver, Peroxisomal and Mitochondrial β‐oxidation Lead to Ketogenesis

In Ndufs4 KO mice liver tissues, catabolic processes were increased whilst anabolic processes were mostly not significantly affected compared to WT mice. In peroxisomes, enzymes involved in α‐oxidation (HACL1) were over‐expressed, as were proteins involved in peroxisomal β‐oxidation of very long chain fatty acids (ACOX1, EHHADH, ACAA1B). However, enzymes that catalyse the last three steps of peroxisomal β‐oxidation of long‐chain fatty acids (HSD17B4, SCP2) were under‐expressed in Ndufs4 KO mice, possibly indicating that mitochondrial β‐oxidation of these and shorter fatty acids are preferred. Indeed, several members of the ACOTs (ACOT2‐4, 12, 13) as well as the peroxisomal carnitine shuttle (CROT) were over‐expressed. ACOT proteins are located in different cellular compartments; ACOT3, ACOT4, ACOT12 and ACOT 13 are mainly located in the peroxisome, whilst ACOT2 is found in the mitochondrial matrix [22]. The upregulation of the ACOTs and CROT in Ndufs4 KO mice suggests that the final products of peroxisomal fatty acid oxidation are actively being shuttled from peroxisomes to adjacent subcellular compartments to serve as intermediates in different metabolic pathways. NUDT19 is thought to regulate mitochondrial fatty acid utilisation by determining the rate of acyl‐CoA formation, with decreased NUDT19 levels leading to increased mitochondrial TCA activity [31]. We observed increased levels of NUDT19 in Ndufs4 KO liver tissues compared to WT, along with PDK1 overexpression. PDK inhibits the formation of acetyl‐CoA from pyruvate by pyruvate dehydrogenase (PDH) [32]. The overexpression of both these regulatory factors suggests a shift away from glycolysis and oxidative TCA towards β‐oxidation in mitochondria. Indeed, we observed overexpression of acyl‐CoA dehydrogenases involved in mitochondrial β‐oxidation and specific to different chain lengths and configurations, namely, ACADL, ACADM, ACAD10 and ACAD11 in Ndufs4 KO mice compared to WT mice. An end product of mitochondrial β‐oxidation is acetyl‐CoA, a substrate for ketogenesis. In Ndufs4 KO liver tissue, we observed the overexpression of two enzymes involved in ketogenesis (HMGCS2, ACSS3), suggesting that the liver is producing ketone bodies as an alternative fuel source for extrahepatic tissues (Figure 2).

FIGURE 2.

Upregulated fatty acid β‐oxidation leads to enhanced ketone body synthesis in the liver of the Ndufs4 KO mice. Proteins involved in these pathways are illustrated, with green indicating upregulation and red indicating downregulation. The products of peroxisomal fatty acid β‐oxidation are exported to the mitochondrion for further oxidation or to the cytosol to be utilised as substrate. Acetyl‐CoA derived from mitochondrial fatty acid oxidation is converted to the ketone bodies. ACAA1 indicates 3‐ketoacyl‐CoA thiolase B, peroxisomal; ACAD10, acyl‐CoA dehydrogenase family member 10; ACAD11, acyl‐CoA dehydrogenase family member 10; ACADL, long‐chain specific acyl‐CoA dehydrogenase; ACADM, medium‐chain specific acyl‐CoA dehydrogenase; ACOT12, acetyl‐coenzyme A thioesterase 12; ACOX1, Peroxisomal acyl‐coenzyme A oxidase 1; ACOX3, Peroxisomal acyl‐coenzyme A oxidase 3; ACSS3, acyl‐CoA synthetase short‐chain family member 3; ALDH3A2, Aldehyde dehydrogenase family 3 member A2; AMACR, alpha‐methylacyl‐CoA racemase; CROT, peroxisomal carnitine O‐octanoyltransferase; EHHADH, peroxisomal bifunctional enzyme; FAD; oxidised flavin adenine dinucleotide; FADH₂, reduced flavin adenine dinucleotide; HMGCS2, Hydroxymethylglutaryl‐CoA synthase, mitochondrial; HSD17B4, peroxisomal multifunctional enzyme type 2; NAD⁺, oxidised nicotinamide dinucleotide; NADH, reduced nicotinamide adenine dinucleotide.

3.3. In the Heart, Anabolic Pathways Are Prioritised, and CI Is Bypassed in OXPHOS via FADH_2‐Driven ATP Production

In humans, the heart is highly reliant on energy generated from the mitochondrial β‐oxidation of long‐chain fatty acids to sustain its contractile activity and maintain ion homeostasis [33]. In mice, the heart is more reliant on glucose, lactate and ketone bodies [34, 35]. In the heart, we observed no significant changes in the expression levels of proteins involved in mitochondrial β‐oxidation, except for ACAD11. Conversely, proteins involved in fatty acid biosynthesis (ACACA, ACACB, ACLY, FASN, MECR and ACSF2) were markedly overexpressed in Ndufs4 KO compared to WT mice. ACACA and ACLY, together with SLC25A1, which was also overexpressed, catalyse rate‐limiting steps in fatty acid biosynthesis. In addition, proteins involved in peroxisome fatty acid synthesis (PECR and TECRL) were also increased. ACACB activity acts as a negative regulator of mitochondrial fatty acid β‐oxidation by inhibition of the carnitine shuttle (CPT) [28]. Although CPT1B and CPT2 levels were not significantly altered in the heart, only one mitochondrial fatty acid β‐oxidation protein (ACAD11) was elevated, which oxidises the very long‐chain fatty acid, behenoyl‐CoA [36]. However, increased levels of proteins involved in peroxisomal α‐oxidation (ALDH3A2) and peroxisomal β‐oxidation of branched chain, very long and long to medium chain fatty acids (AMACR, ACOX1, EHHADH, HSD17B4 and SCP2) were overexpressed in Ndufs4 KO mice compared to WT. Products of peroxisomal fatty acid β‐oxidation are usually transported into the mitochondria, where they are further oxidised into acetyl‐CoA, which enters the TCA cycle, generating reducing equivalents (NADH and FADH₂) to fuel ATP generation. However, since CI of the OXPHOS system is dysfunctional in Ndufs4 KO mice, NADH is not oxidised, whilst FADH₂ can still feed electrons into the OXPHOS system via complex II and other enzymes. The observed increased expression (Figure 3) of flavin‐linked ACAD11 in the Ndufs4 KO heart may represent this metabolic adaptation to ICD. ACAD11 is one of the 14 flavin‐dependent dehydrogenase enzymes that donates electrons of FADH₂ to the electron transfer flavoprotein (ETF), which subsequently transfers it to CoQ of the OXPHOS system via the electron transfer flavoprotein‐ubiquinone oxidoreductase (ETF‐QO) pathway [37]. This metabolic adaptation effectively bypasses the dysfunctional CI in Ndufs4 KO mice, ensuring the continuity of ATP generation in ICD.

FIGURE 3.

Upregulated fatty acid biosynthesis and bypassing of defective CI in the heart of Ndufs4 KO mice. Proteins involved in these pathways are illustrated, with green indicating upregulation and red indicating downregulation. In the presence of isolated complex I deficiency (ICD), proteins involved in fatty acid synthesis are upregulated along with the ACAD11 involved in fatty acid oxidation. Upregulated ACAD11 facilitates the transfer of electrons to the OXPHOS system via the ETF; QO pathway. ACACA indicates acetyl‐CoA carboxylase α; ACACB, acetyl‐CoA carboxylase β; ACAD11, acyl‐CoA dehydrogenase family member 11; ACLY, ATP‐citrate synthase; CoQ, reduced ubiquinone; ETF: QO, electron transfer flavoprotein: ubiquinone oxidoreductase; FAD, oxidised flavin amide adenine dinucleotide; FADH₂, reduced nicotine amide adenine dinucleotide; FASN, fatty acid synthase; NAD⁺, oxidised nicotine amide adenine dinucleotide; NADH, reduced nicotine amide adenine dinucleotide; NADP⁺; oxidised nicotinamide dinucleotide phosphate; NADPH, reduced nicotinamide dinucleotide phosphate; SLC25A1, tricarboxylate transport protein; TECRL, very‐long‐chain enoyl‐CoA reductase.

3.4. In the Brain, the Olfactory Bulb Is Most Affected and Favours Fatty Acid Synthesis Pathways

Lipids account for more than 50% of the total brain dry weight. These include phospholipids and cholesterol, which make up the cellular membranes of the brain. In the brain, glucose is the obligatory energy substrate; however, under metabolic stress, the brain can utilise alternative substrates such as ketones, pyruvate and lactate to sustain its cognitive function [38, 39]. Out of the body's total metabolic energy consumption, 20% is utilised by the brain to sustain proper brain function and is therefore highly susceptible to defects in energy metabolism [40, 41].

In the different brain regions investigated here, a relatively small number of fatty acid metabolism proteins were affected by Ndufs4 KO, with one, two and eight of these proteins differentially regulated in the brainstem, cerebellum and olfactory bulb, respectively. Among the eight differentially expressed proteins in the olfactory bulb of Ndufs4 KO mice, we observed an increased abundance of key proteins of mitochondrial fatty acid synthesis, which included the mitochondrial enoyl‐CoA reductase, MECR, and carbonyl reductase 4, NADPH‐dependent (CBR4). Fatty acid synthesis pathways are mainly in the cytosol. Thus, the exact role of the mitochondrial fatty acid synthesis pathway remains a matter of debate, however, some studies have suggested that the product of this pathway plays a role in regulating mitochondrial energy metabolism, mitochondrial respiratory chain assembly and mitochondrial translation [42, 43, 44]. Additionally, hydroxyacyl‐CoA dehydratase 2 (HACD2), which catalyses the third step of very long chain fatty acid synthesis, was also increased in the Ndufs4 KO mice compared to WT.

Furthermore, carnitine palmitoyltransferease 2 (CPT2) which is involved in the import of fatty acids into the mitochondrion, was overexpressed. Only acyl‐CoA dehydrogenase long‐chain (ACADL) of the mitochondrial fatty acid oxidation pathways was overexpressed, whilst all others were not significantly affected. No proteins involved in peroxisomal fatty acid oxidation were significantly affected either. This is not surprising as fatty acids are not a typical alternative energy source in brain tissues. Intriguingly, whilst most of the proteins expressed in these brain regions showed increased abundance in Ndufs4 KO mice, this was not the case for the HACD2, which was downregulated in the cerebellum. The limited number of expressed proteins in the brainstem and cerebellum of the Ndufs4 KO mice suggest a minimal or absent adaptive response to energy deprivation caused by NDUFS4 deficiency in these regions.

3.5. In the Kidney, Reduced Expression of Proteins Involved in Mitochondrial and Peroxisomal β‐Oxidation Suggests Impaired Fatty Acid Degradation

The kidney is a metabolically demanding tissue that requires a constant supply of energy for cellular functions such as filtration of waste products from the blood, reabsorption of nutrients, and regulation of an acid‐base balance [45]. In the kidney, the preferred energy substrate is cell‐type specific: the proximal tubules utilise fatty acids as the primary energy substrate whilst the glomerular cells prefer glucose [46, 47]. Differentially expressed proteins in the kidney are involved in the peroxisomal metabolism, which include ALOX12, ACOX1‐3, ACOT8, AMACR, CROT, EHHADH, NUDT19 and PECR. However, in the kidney, all were downregulated, except for the HACL1, which was upregulated. Two proteins of the carnitine shuttle systems, SLC22A5 and CPT1B, respectively catalyse the uptake and conversion of fatty‐acyl‐CoA into acyl‐carnitine for subsequent catabolism via β‐oxidation in the mitochondrial matrix [48]. Both were overexpressed, suggesting that fatty acids are transferred into the mitochondria at an increased rate. However, proteins involved in the initial steps of mitochondrial β‐oxidation were not significantly affected, except for that of medium chain fatty acids, ACADM, which was less abundant in Ndufs4 KO mice compared to WT mice. Concurrently, the fatty acid biosynthesis protein, ACFS2, which utilises medium‐chain fatty acids as substrate, is overexpressed in Ndufs4 KO mice. Together, this suggests that medium‐chain fatty acids, at least, are directed towards fatty acid synthesis rather than oxidation in the kidney of Ndufs4 KO mice.

Of particular interest is a subclass of cytochrome P450 monooxygenase proteins, particularly CYP4A14 and CYP4A10, which catalyse the ω‐hydroxylation of medium‐chain fatty acids into dicarboxylic acids [49] were found to be overexpressed in the kidney and liver of the Ndufs4 KO mice. The resulting dicarboxylic acids generated are exported from the microsomes to the peroxisomes, where they are degraded into acetyl‐CoA. From our results (Figure 4), we observed a downregulation of acyl‐CoA thioesterase, specifically ACOT8 and ACOT12. These proteins are responsible for the hydrolysis of acyl‐CoA esters into free fatty acids and CoA‐SH [50, 51]. The decreased expression of carnitine O‐octanoyl transferase, CROT, in the kidney, along with the decreased expression of almost all other proteins involved in fatty acid oxidation, suggests that these pathways are not utilised in the kidney of Ndufs4 KO mice.

FIGURE 4.

Alterations of proteins involved in peroxisomal and mitochondrial fatty acid β‐oxidation in the kidney of the Ndufs4 KO mice. Proteins involved in these pathways are illustrated, with green indicating upregulation and red indicating downregulation. Proteins involved in the β‐oxidation of fatty acids in both organelles, as well as the transport of fatty acids between organelles, are underexpressed in Ndufs4 KO mice. Medium‐chain fatty acids that enter the mitochondrion are utilised in fatty acid synthesis, rather than undergoing β‐oxidation. ACSF2 indicates medium‐chain acyl‐CoA ligase; ACOT8, acyl‐coenzyme A thioesterase 8; ACOT12, acetyl‐coenzyme A thioesterase 12; ACOX1; peroxisomal acyl‐coenzyme A oxidase 1; ACOX2; peroxisomal acyl‐coenzyme A oxidase 2; CPT1, carnitine O‐palmitoyl transferase 1; CROT, peroxisomal carnitine O‐octanoyltransferase; EHHADH; peroxisomal bifunctional enzyme.

4. Discussion

Fatty acids are crucial energy nutrients acquired from various sources, including dietary intake, de novo synthesis and the hydrolysis of triglycerides [52, 53, 54]. The homeostasis of fatty acid metabolism depends on the complex crosstalk between uptake, synthesis, oxidation, and release of free fatty acids from triacylglycerol. Dysregulation of this metabolic crosstalk has been linked to several metabolic diseases, including obesity, diabetes, heart failure, kidney injury, neurodegeneration and non‐alcoholic fatty liver disease [34, 44, 55–57]. Cellular uptake of fatty acids occurs through passive diffusion (for short‐chain fatty acids) or by the action of fatty acid binding proteins (for long‐chain fatty acids), which are specific to each tissue. Once inside the cell, fatty acids are esterified into acyl‐CoA esters, which can either be utilised for lipogenesis or oxidised for energy generation, depending on the metabolic state [44, 58]. Fatty acid oxidation serves as a vital energy source across various tissues.

In the present study, we employed proteomics to evaluate the expression patterns of fatty acid metabolism proteins in various tissues of Ndufs4 KO mice.

In the Ndufs4 KO liver, there was an observed increase in the expression of most of the proteins encoding enzymes of mitochondrial and peroxisomal β‐oxidation. However, proteins involved in the metabolism of branched‐chain and ω‐fatty acids, including, SCP2, HSD17B12, CYP8B1 and CYP1A2 showed a significant decrease in expression. Elevated levels of fatty acids act as ligands that bind to PPARα thereby activating it and subsequently increasing the transcription of genes encoding proteins involved in peroxisomal and mitochondrial fatty acid β‐oxidation [59, 60]. The upregulation of proteins involved in fatty acid oxidations may represent a broader metabolic adaptive response to compensate for elevated fatty acid levels. These elevated levels result from a progressive decline in liver fat droplets, which has previously been shown in Ndufs4 KO mice [14]. Liver fat droplets function as reservoirs for excess fatty acids, and defects in these reservoirs would consequently lead to increased delivery of fatty acids into liver cells (Gluchowski et al., 2017; Seebacher et al., 2020). Given the impaired energy production via the OXPHOS system and the disrupted TCA cycle caused by ICD in the Ndufs4 KO mice, acetyl‐CoA generated from the enhanced mitochondrial fatty acid β‐oxidation is diverted toward ketogenesis, resulting in the production of ketone bodies that can be used as an alternative energy source to fuel extrahepatic tissues. This is also supported by the observed upregulation of two proteins involved in ketogenesis (Table  1 ). Such changes in fatty acid metabolism in the liver might warrant additional disease monitoring protocols that include liver function assessments. The use of ketogenic diets to manage mitochondrial disease is considered a promising treatment strategy [61] but also has serious adverse effects in some patients [62, 63]. Our findings suggest that careful consideration is required for this strategy, as such a diet could over‐burden the liver as well as extra‐hepatic tissues [64].

For the brain regions, our results demonstrated that only a few proteins in fatty acid metabolism were altered in the brainstem and cerebellum of the Ndufs4 KO mice, with one and two proteins being differentially expressed, respectively. This suggests a limited or absent adaptive response in these regions, potentially indicating that they lack the capacity to compensate for energy deprivation caused by Ndufs4 deficiency. In contrast, the olfactory bulb exhibited a compensatory response as evidenced by the significant upregulation of fatty acid metabolism proteins; these included proteins associated with mitochondrial fatty acid synthesis and mitochondrial fatty acid oxidation. It has previously been shown that upregulated MECR of mitochondrial fatty acid synthesis has a key regulatory role in gene transcription via the PPARα pathway in mammals [65]. Different models have been proposed regarding the mechanism by which MECR modulates the PPARα [66, 67]. In one model by Parl et al. [67], it was proposed that upregulation of MECR leads to increased generation of mitochondrial fatty acid synthesis products such as palmitate, a well‐recognised inducer of the PPARα pathway. Our results in the olfactory bulb of the Ndufs4 KO mice are consistent with the Parl model, as the upregulation of MECR and CBR4 is suggestive of an increase in mitochondrial fatty acid synthesis products, which subsequently activate PPARα, thereby increasing the transcription of genes encoding proteins involved in mitochondrial fatty acid β‐oxidation. However, aside from ACADL, no other fatty acid oxidation proteins were upregulated. The exact role of the upregulation of ACADL in the olfactory bulb is unclear since fatty acids are not typical alternative energy sources in brain tissues. Taken together, our findings revealed differential regulation of fatty acid metabolism proteins in the olfactory bulb, but not in the brainstem or cerebellum, which indicates region‐specific adaptive response to Ndufs4 deficiency.

Furthermore, distinct expression of proteins involved in fatty acid synthesis was observed across the heart, kidney and liver. In the Ndufs4 KO heart, as in the olfactory bulb, mitochondrial fatty acid synthesis was increased, as evidenced by the increased expression of MECR. However, whilst increased MECR expression may represent an adaptive metabolic compensation in the olfactory bulb, this is not the case in mice heart where overexpression of the gene encoding MECR has been associated with the development of contractile dysfunction [68]. Unexpectedly, most of the proteins catalysing the various steps of mitochondrial fatty acid β‐oxidation did not reach statistical significance in the heart, except for ACAD11. This observation corroborates a previous metabolomics study which showed that acylcarnitine levels (i.e., conjugates of fatty acyl‐CoA) of different carbon chain lengths were not significantly altered in the heart of a different Ndufs4 KO mouse model [19]. Collectively, these results suggest that in the Ndufs4 KO heart, energy deprivation resulting from ICD does not lead to the expected upregulation of mitochondrial β‐oxidation as alternative energy and source and likely compensates by utilising alternative ATP production pathways such as increased glycolysis. Further investigation is thus needed. The possible increased utilisation of ETF‐QO pathway discussed above lends further support to the use of alternative Q‐cycle fuels in MD patients [62].

In the kidney, the downregulation of proteins involved in β‐oxidation (except for HSD17B4 and SCP2) was evident, at both mitochondrial and peroxisomal levels, a recognised metabolic hallmark of MD [69]. Such downregulation is known to result in the accumulation of lipids beyond physiological levels within the mitochondrial matrix and peroxisomal compartments [69]. The accumulation is further exacerbated by the downregulation of CROT and ACOT proteins. Taken together, these alterations point to an increase in fatty acid import and uptake without a parallel increase in peroxisomal and mitochondrial β‐oxidation would ultimately lead to the accumulation of lipid conjugates such as acylcarnitines in the kidney, potentially contributing to renal complications and tubular injury observed in LS patients [7, 8]. Therefore, our findings would suggest early monitoring of kidney function in patients with ICD or LS.

5. Conclusion

In summary, the present study provides insight into the effect of ICD in the regulation of fatty acid metabolism. Our findings highlighted the distinct regulation of fatty acid metabolism as either adaptive or maladaptive response to ICD in various tissues of Ndufs4 KO mice. The olfactory bulb and liver exhibited compensatory mechanisms, other tissues, such as the brainstem, cerebellum and kidney, and to a certain extent the heart showed maladaptive responses. Although this study provides insights into the effects of ICD on fatty acid metabolism and the potential contributions of dysregulated fatty acid metabolism to the phenotypes observed in mitochondrial disease, it is important to acknowledge certain limitations. Our data reflects quantitative changes in protein abundances, which may not always correlate with metabolic flux. Therefore, these findings should be interpreted as indicative of potential pathway alterations, rather than definitive measure of metabolic activity. Our findings suggest that a simple disease mechanism initiated by ICD is not sufficient for multi‐system organisms, as each organ is affected differently. Such observations are especially important when considering therapeutic development, since a specific intervention strategy might be beneficial in some tissues whilst being counter‐productive in others. These possible differences between tissues in response to treatments might help to explain the inconsistent outcomes of interventions in individual patients [62]. Furthermore, our study further highlights the importance of investigating complicated disorders with multi‐systemic consequences, such as MDs, in disease models that allow for the simultaneous investigation of several organ systems. To our knowledge, this is the first study to do so in a model for NDUFS4 deficiency. This study provides some preliminary mechanistic explanations underlying the well‐known tissue specificity seen in the presentation of mitochondrial disorders and provides several key points from which follow‐up functional studies can be pursued.

Author Contributions

Glen Khumalo: investigation, methodology, formal analysis, writing—original draft, review and editing. Jeremie Z Lindeque: validation, methodology, investigation, writing—review and editing. Marianne Venter: conceptualisation, methodology, investigation, resources, writing—review and editing, supervision, project administration, funding acquisition.

Ethics Statement

Ethical clearance for the study (NWU‐00764‐23‐A5) was provided by the Animal Research Ethics Committee of the North‐West University (NWU‐AnimCareREC), and all the experimental procedures were conducted in accordance with the code of ethics in research, training and testing of drugs in South Africa and complied with the national legislation.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting Information

Supporting Information

Data Availability Statement

The mass spectrometry proteomics data have been deposited in the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via PRIDE partner repository with the dataset identifier PXD061439. Data will be shared upon reasonable request to the corresponding author.