Mitochondrial fatty acid synthesis is essential for coordinated energy transformation

Loss of adipocyte Mecr leads to impaired energy breakdown and insulin resistance

Adipose tissues are the major sources of Mecr expression (Fig.S1a). To characterize the impact of Mecr deficiency in adipose tissues, we generated Mecr^flox/flox Adipoq-Cre mice. MECR is described to be involved in lipoic acid synthesis in vitro (Nowinski et al., 2018) and indeed, in white adipose tissue, loss of MECR leads to decreased protein lipoylation (Fig.1a), but also to reduced succinylation, indicating disturbed mitochondrial succinate metabolism. Lipoic acid is a critical cofactor for PDH activity, which was decreased in adipose tissue (Fig.1b). Of note, PDH E1 subunit alpha 1 (PDHA1), mirroring total PDH level, was increased on protein level and its inhibitory phosphorylation was nearly absent (Fig.1a), indicating compensatory effects to restore the attenuated PDH activity. Further, electron micrographs of WAT showed a disturbed mitochondrial structure in the Adipoq Cre⁺ mice characterized by reduced cristae structure and a cigar-like shaped, electron-dense mitochondrial form (Fig.1c). To analyze any impact on the lipidomic signature of the mice, we performed lipidomic analysis of plasma, white and brown adipose tissues as well as liver. Except for slight reductions in DAG/TAG, the lipidome of plasma and liver remained mostly unaffected by adipocyte-specific loss of Mecr. In contrast, in brown and even more pronounced in white adipose tissue all lipid classes were reduced in Adipoq-Cre⁺ mice compared to the respective organ of the Adipoq-Cre^- littermates (control), emphasizing a general cellular lipid metabolism dysfunction (Fig.1d). We further applied a targeted metabolomics approach to analyze other potentially changed metabolites. Here, we additionally identified changes in amino acids, and sulfonic acids – in particular taurine (FigS1b) – indicating that MECR affects various metabolites and might be involved in many metabolic pathways in WAT or that the cells adapt by rearrangement of metabolic pathways (Fig.1e). Gene expression analysis revealed a considerable increase of cytosolic lipid synthesis-related genes in WAT upon Mecr deficiency as well as slightly elevated markers of fibrosis (Timp1, Cxcl10) and cell stress (Ddit3, Ccl2) (Fig.1f). However, this was not accompanied by an inflammatory cell infiltration, indicated by neither altered inflammatory markers (Tnf, Cd68) on gene expression level (Fig.1f) nor innate/adaptive immune cell population-enrichement in the WAT of the Adipoq-Cre⁺ mice(Fig.3g). Similarly, among the “metabolically active macrophages” defined as CD45⁺Ly6G^-CD11b⁺F4/80⁺CD11c⁺ (Hill et al., 2018; Jaitin et al., 2019; Shaul et al., 2010), no differences in the expression of markers which better define their identity (CD68 and Ly6C) as well as their polarization status (ARG1, CCR2, CD80 and CX3CR1) were observed between Adipoq-Cre⁺ and Adipoq-Cre^- mice. To further investigate the ability of functional cellular lipid metabolism of WAT cells, we performed in vitro lipolysis assays in primary adipocytes. This showed that both basal as well as norepinephrine-induced lipolysis and free fatty acid release were increased in primary adipocytes derived from Adipoq-Cre⁺ mice (Fig.1i). At the same time, while the release of free fatty acids could be suppressed by insulin in Adipoq-Cre^- adipocytes, this effect was attenuated in Adipoq-Cre⁺ adipocytes (Fig.1j) indicating cellular insulin resistance. This was also accompanied by reduced phospho-AKT levels, a surrogate marker of impaired insulin signaling (Fig.S1c). In line, in vivo VLDL production, which is highly dependent on the amount of free fatty acids released from white adipose tissue, was higher in Adipoq-Cre⁺ mice (Fig.1k). To further estimate the role of Mecr for the development of insulin resistance, we fed Adipoq-Cre mice with a high fat diet for 12 weeks and performed a combined radioactive tracer-spiked oral glucose and fat tolerance test. Interestingly, despite the fact that Adipoq-Cre⁺ mice similarly gained body weight (Fig.S1d), they exhibited a decreased glucose and lipid tolerance (Fig.1l-m), which might be explained by lower uptake of ¹⁴C-triolein tracer in WAT (Fig.1n) and the reduced PDH activity in obese WAT (Fig.1o) needed for efficient glucose breakdown.

• Download figure
• Open in new tab

Figure 1: Loss of adipocyte Mecr leads to impaired energy breakdown and insulin resistance.

(a) Western blot (n=3), (b) WAT PDH activity (n=3), (c) electron microscopy (n=3) of white adipose tissue from Mecr^flox/flox Adipoq-Cre⁺ mice and Cre^- littermate controls. (d) Lipidomics of liver, white adipose tissue (WAT), brown adipose tissue (BAT) or plasma from Mecr^flox/flox Adipoq-Cre mice (n=4). (e) Enrichment plot of targeted metabolomics (n=6) from gonadal white adipose tissue of Mecr^flox/flox Adipoq-Cre mice. (f) Gene expression (n=5-9) and (g-h) flow cytometry analysis (n=6) of white adipose tissue of Mecr^flox/flox Adipoq-Cre mice. (i) Primary adipocytes were differentiated for 10 days and were treated with norepinephrine or mock (n=4) and (j) insulin or mock (n=4) for the indicated time period and free fatty acids as a marker of lipolysis was measured from cell culture supernatant. (k) VLDL synthesis experiment in Mecr^flox/flox Adipoq-Cre mice by injection of lipase inhibitor tyloxapol and measurement of the accumulating triacylglycerols in plasma in Mecr^flox/flox Adipoq-Cre mice during fasting over time. (l-o) Mecr^flox/flox Adipoq-Cre mice were fed a high fat diet for 12 weeks and were gavaged with a combined radioactive tracer-spiked (l) lipid and (m) glucose load where (n) ¹⁴C-triolein uptake as well as (o) postprandial WAT PDH activity was measured (n=9-10). Error bars indicate standard error of the mean (SEM). *= p<0.05 by two-way ANOVA and Fishers-LSD Test (i, j) or Student’s T-Test.

Loss of adipocyte Mecr leads to brown adipose tissue inflammation and dysfunctional thermogenesis

Compared to other organs and most likely due to its high mitochondrial density, brown adipose tissue (BAT) exhibits the highest gene expression of Mecr (Fig.S1a). Of note, when mice are exposed to cold conditions (7 days at 4°C ambient temperature), gene expression levels of mtFAS genes (Mcat, Mecr, Oxsm, Ndufab1, Cbr4) increased in BAT (Fig.S2a), suggesting a role during cold adaptation. Besides the drop in most membrane lipid classes (shown in Fig.1d), targeted metabolomic analysis indicated severe changes in amino acid and biotin metabolism in BAT of Adipoq-Cre⁺ mice (Fig.2a; Fig.S2b). Furthermore, similarly as observed in WAT (Fig.1a), BAT tissue showed that loss of Mecr strongly diminished protein lipoylation and the inhibitory PDHA1 phosphorylation (Fig.2b), again suggesting compensatory loss of PDH inhibiting/inactivating mechanisms. In addition, mitochondrial morphology was highly affected in BAT of Adipoq-Cre⁺ mice. Electron micrographs revealed a different ultrastructure and reduced mitochondria to ER contact sites (marked in green), while ER-lipid droplet contacts (marked in red) appeared to be enriched (Fig.2c). Based on these observations, we hypothesized that glucose catabolism and uptake as well as lipid metabolism would be disturbed in Adipoq-Cre⁺ mice. Indeed, when we performed radioactive tracer studies, BAT uptake of ³H-deoxyglucose (DOG) (Fig.1d) and ¹⁴C-triolein (Fig.1e) was reduced in Adipoq-Cre⁺ mice on chow diet at conventional housing temperature, suggesting lower activity of the tissue or a diminished capacity to handle nutrients for thermogenesis. To evaluate thermogenic capacity, we performed indirect calorimetry experiments. When injected with the beta3-adrenergic agonist CL316,243 to specifically activate thermogenic adipose tissues, Adipoq-Cre⁺ mice exhibited lower energy expenditure and oxygen consumption (Fig.1f, Fig.S2c,d), indicative of mitigated thermogenic function. When we housed the mice at cold conditions (4°C), Adipoq-Cre⁺ mice appeared to be cold intolerant as to our surprise their survival was critically reduced (Fig.2g) and the experiment had to be terminated 3 days after the initiation of cold exposure. Cold-exposed BAT of the Adipoq-Cre⁺ mice presented lower protein levels of UCP1 and OXPHOS complexes (Fig.1h), indicating that impaired non-shivering thermogenesis and mitochondrial respiration might be culpable for their lower survival. Gene expression analysis of BAT at conventional housing temperature showed that genes related to cytosolic lipid synthesis were downregulated while several inflammatory and stress markers were markedly induced. Interestingly, an almost 25-fold increase was noted for the fibrosis marker Timp1 (Fig.1i). To evaluate whether the inflammation was attributed to brown but not white adipocytes, similar gene expression analysis was performed using Ucp1-Cre⁺ mice, which resembled the phenotype of Adipoq-Cre⁺ (Fig.S3a-d). Flow cytometry analysis of BAT showed largely elevated cell counts of several immune cell populations in the Adipoq-Cre⁺ mice, including CD4+ cells and macrophages (CD11b⁺F4/80⁺CD11c⁺ and CD11b⁺F4/80⁺CD11c^-) eosinophils (SiglecF⁺). (Fig.2j). Further phenotypical characterization of a population of metabolically active macrophages (CD11b⁺F4/80⁺CD11c⁺) revealed differences in the frequency of cells expressing CD68 and Ly6C antigens as well as alteration in their polarization status with high levels of molecules associated to a tissue remodeling/pro fibrotic phenotype (CX3CR1 and ARG-1) and antigen presentation capacity (CD80) in Adipoq-Cre⁺ mice compared to wild type littermates (Fig.2k). Of note, the increased number of non-adipocytes in BAT of Adipoq-Cre⁺ mice might explain some of the differences in lipidomic and metabolomic analysis due to the massive infiltration of immune cells and a dilution of brown adipocyte analytes. In sum, both white and brown adipose tissue of mice lacking Mecr in adipocytes showed a severe parenchymal dysfunction, which was accompanied by higher markers of inflammation and fibrosis in BAT, indicating that mtFAS is essential for regular adipose tissue function.

• Download figure
• Open in new tab

Figure 2: Loss of adipocyte Mecr leads to brown adipose tissue inflammation and dysfunctional thermogenesis.

(a) Metabolic pathways enriched based on targeted metabolomics data of brown adipose tissue (n=6; for single metabolites see also Fig.S2b). (b) Western blot of brown adipose tissues (n=3). (c) Electron micrographs of brown adipose tissue (n=3) of Mecr^flox/flox Adipoq-Cre mice. (d) ³H-deoxyglucose (DOG) and (e) ¹⁴C-triolein tracer uptake into indicated organs after combined oral glucose and lipid gavage (n=5-6). (f) Energy expenditure of Mecr^flox/flox Adipoq-Cre mice injected with the beta3-adrenergic agonist CL316,243 at thermoneutrality (n=6). (g) Survival of Mecr^flox/flox Adipoq-Cre mice during cold exposure (experiment stopped at d3 of cold exposure). (h) Western blot of brown adipose tissues after 3d of cold exposure (n=3). (i) Gene expression of brown adipose at conventional housing temperature as well as (j) flow cytometry analysis of infiltrating cells and (k) phenotypic characterisation of BAT macrophages (n=6). Error bars indicate standard error of the mean (SEM). *= p<0.05 by Student’s T-Test.

Liver-specific loss of Mecr leads to lipoylation depletion, impairment of PDH activity and early death

As Mecr is highly expressed in liver tissue (Fig.S1a) and the liver is a central organ for glucose and lipid handling, we generated Mecr^flox/flox Alb-Cre mice, to study the role of MECR for systemic lipid and glucose metabolism, Interestingly, the hepatocyte-specific deletion of Mecr led to globally affected mice with lower body weight (Fig.S4a), lower fasting blood glucose (Fig.S4b), cachexia-like adipose tissue wasting (Fig.S4c) as well as early death around the age of 14 weeks. For this reason, experimental interventions with the Cre⁺ homozygous floxed mice were performed before the 12^th week of age. For breeding, heterozygous floxed Cre⁺ mice were paired with homozygous floxed Cre^- mice because homozygous floxed Cre⁺ females did not survive until the end of pregnancy and homozygous floxed Cre⁺ males showed low fertility, likely due to failure to thrive. Interestingly, hepatocyte-specific Mecr deletion resulted in hepatomegaly, which was evident by an almost 2-fold increase of the percent liver-to-body weight ratio (Fig.S4c). Lipid analysis unveiled a rearranged hepatic lipidomic profile in the adult Alb-Cre⁺ mice (Fig.2a), with marked changes on triacylglycerol (TAG) (>70% reduction) and CE (>50% reduction), but also ceramides (CER) (25% reduction), SM (30% reduction) and to a lower extend all other investigated lipid classes (LPC, FFA, LPE, PC, PE ranging from 15%-25% reduction). As expected, in Alb-Cre⁺ mice with white adipose tissue wasting, white adipose tissue had lower TAG and higher phospholipids, most likely because the ratio of membrane-associated lipids to storage lipids is higher in their depots (Fig.3a). As observed by H&E staining, the livers of the Alb-Cre⁺ mice were characterized by enlarged hepatocytes in comparison to the wild type littermates (Fig.3b). Further, electron microscopy revealed that the hepatocytes were completely filled with swollen mitochondria (Fig.3c), features of early oncosis and cell death, which is described for acute liver failure (Bantel and Schulze-Osthoff, 2012). Following this observation, we next estimated the effect of MECR on mitochondrial function by seahorse-based oxygen consumption rate analysis in HepG2 cells treated with MECR siRNA, as primary hepatocytes of Alb-Cre⁺ mice were not viable. HepG2 cells treated with siRNA against MECR showed reduced oxygen consumption rates during basal respiration and uncoupled respiration already after 24h of siRNA treatment (Fig.3d), indicating a deleterious mitochondrial phenotype. Mitochondrial dysfunction is associated with cellular stress responses and indeed, upon hepatic deletion of MECR, gene expression analysis revealed high liver stress indicated by Ddit3, interestingly without markers of ER-stress (Xbp1, sXbp1). In line, we detected an increased expression of the pro-inflammatory cytokine Tnfa and the macrophage marker Cd68 in the Alb-Cre⁺ mice, whereas other genes usually highly abundant in liver were hardly detectable (Fasn, Ppara, Fgf21) in adult mice (Fig.3e).

• Download figure
• Open in new tab

Figure 3: Liver-specific loss of Mecr leads to lipoylation depletion, impairment of PDH activity and early death.

(a) LC/MS-based targeted lipidomics of liver, inguinal white adipose tissue (iWAT) or plasma from Mecr^flox/flox Alb-Cre mice (n=3-5).(b) H&E staining and (c) electron microscopy of livers from Mecr^flox/flox Alb-Cre mice (n=3). (d) Seahorse analysis of HepG2 cells transfected with either siMECR or control (n=12). (e) Hepatic gene expression analysis (n=3-9), (f) western blot of liver from Mecr^flox/flox Alb-Cre mice (n=3). (g) PDH activity (n=3-5), (h) liver pyruvate (n=3-9), (i) liver lactate (n=3-9), (j) NAD/NADH ratio of livers (n=3-9), (k) blood glucose (n=3-9), (l) plasma insulin (n=3-9), (m) plasma triglycerides and ¹⁴C-deoxyglucose uptake from Mecr^flox/flox Alb-Cre mice gavaged with a combined radioactive glucose and lipid load (n=3-9). Error bars indicate standard error of the mean (SEM). *= p<0.05 by Student’s T-Test.

Western blot analysis of the livers showed that, despite the massive enlargement of the mitochondria, oxidative phosphorylation complex levels were only mildly affected on protein level (Fig.S2e). However, markers of intracellular energy availability such as lysine-malonylation and –succinylation were dysregulated on protein level with reductions in Alb-Cre⁺ mice (Fig.S4d). Of note, the reduced phosphorylation of AKT in the Alb-Cre⁺ mice, pointed towards an attenuated insulin signaling. Importantly, as indicated by the impaired protein lipoylation pattern in the liver of Alb-Cre⁺ mice (Fig.3f), loss of liver MECR most likely decreased lipoic acid synthesis. Additionally, and in line with the formation of megamitochondria, the cleaved form of OPA1, an important protein that regulates mitochondrial fusion/fission dynamics, was particularly elevated in the livers of Alb-Cre⁺ (Fig.3f).

Since lipoic acid is a crucial cofactor for PDH, the levels of this enzyme were further investigated. Despite elevated hepatic protein levels of the PDHA1 in the Alb-Cre⁺ mice (Fig.3f), the PDH activity was significantly diminished (Fig.3g), most likely due to almost absent protein lipoylation (Fig.3f). Since the phosphorylated form of PDHA1 was not significantly altered (Fig.3f), the elevated protein levels of PDHA1 might stem from a feedback mechanism to counteract its attenuated activity. Despite the lower PDH activity, hepatic pyruvate did not accumulate in Alb-Cre⁺ mice, but was rather decreased (Fig.3h), which might in part be explained by the hypoglycemia of the mice, which generally leads to lower glycolysis (Unger, 1985). Besides its degradation by PDH, pyruvate can alternatively be converted into lactate by lactate dehydrogenase, which regenerates NAD⁺ for ongoing glycolysis. Hepatic lactate levels were also lower in Alb-Cre⁺ mice (Fig.2i) overall suggesting lower glycolytic activity. We next speculated, that glycolysis might be reduced due to a lack of NAD⁺ which cannot be regenerated from pyruvate. Nevertheless, the hepatic ratio of NAD⁺ to NADH was critically increased (Fig.3j) in Alb-Cre⁺ mice, arguing that enough NAD⁺ would be available for ongoing glycolysis and that NAD⁺ levels are not the reason for the failure of glycolytic activity. Of note, also NADP⁺/NADPH was increased and therefore in principle the pentose phosphate pathway was impaired as well (Fig.S4e), whereas ATP levels were similar in both genotypes (Fig.S4f). In addition, enzyme activity of alpha-ketoglutarate dehydrogenase, which like PDH also relies on lipoic acid as cofactor, was also significantly reduced in Alb-Cre⁺ mice (Fig.S4g).

To characterize the postprandial metabolic capacity, we performed a combined oral glucose and fat tolerance test. Interestingly, Alb-Cre⁺ mice showed not only a lower basal glucose level, but also reduced levels during the oral glucose tolerance test (Fig.3k). Surprisingly, the insulin levels of the Alb-Cre⁺ mice were almost unchanged during the glucose load (Fig.3l), indicating a lower insulin dependency of glucose clearance or an increased insulin sensitivity. Plasma triglyceride levels were elevated during lipid load in Alb- Cre^-, but not in Alb-Cre⁺ mice, which also presented lower basal triglyceride (Fig.3m) and cholesterol (Fig.S4h) levels.

Overall, our data suggest that the reduced NADH/NADPH formation capacity and reduced dehydrogenase activities observed in Alb-Cre⁺ mice might be important contributors to fatal cause of liver dysfunction ultimately resulting in disturbances in systemic glucose and lipid metabolism.

Functional analysis of human MECR gene variants reveal disturbed glucose and lipid homeostasis

A male patient in his 20s presented clinically with childhood onset optic nerve atrophy and gait ataxia, dystonia, generalized myoclonia and choreoathetoid movements. Trio-whole exome sequencing revealed a biallelic mutation in MECR. It uncovered a loss-of-function mutation c.553delG (p.D185Ifs*25), and a c.772C>T (p.Arg258Trp) previously published pathogenic missense variant (Heimer et al., 2016) in compound-heterozygous state. To characterize the impact of these patient variants on cellular function, we overexpressed wild-type MECR, as well as both variants in HEK cells. First, we assessed PDH activity as PDH is known to depend on lipoic acid as a cofactor. Cells transfected with wild-type MECR, but not with one of the two patient variants, showed increased PDH activity (Fig.4a). Next, we assessed glucose uptake upon overexpression of wild-type and variant MECR. Of note, overexpression of both variants, but not the wild-type transcript, led to higher uptake of ³H-deoxyglucose into the cells (Fig.4b), suggesting a dominant negative effect of each MECR variant on mitochondrial catabolism and a subsequent higher need for anaerobic glycolysis to compensate. To assess the functional impact of mtFAS on cellular lipid metabolism, we performed LC/MS-based targeted lipidomic analysis of HEK293T cells transfected with wild-type MECR and the patient MECR variants. Interestingly, we found that membrane lipids, such as (lyso)-phosphatidylcholine ((L)PC), (lyso-) phosphatidylethanolamine ((L)PE), and sphingomyelin (SM), were lower upon overexpression of the patients MECR variants (Fig.4c; Fig.S5a). Overexpression of wild type MECR caused higher abundance of cholesterylester (CE) species containing palmitate and oleate residues (Fig.4d), which reflect fatty acid species described to be specifically derived from cytosolic de novo lipogenesis (Heeren and Scheja, 2021; Worthmann et al., 2022). While the frameshift variant p.D185Ifs*25 showed similar lipid species concentrations compared to cells transfected with the empty vector control, the missense variant p.Arg258Trp showed a tendency towards higher levels of indicated CE species as well, indicating a potential residual activity of the MECR^R258W variant. Evaluation of the mouse models revealed that cholesterylester species in plasma were not influenced by adipose tissue mtFAS deficiency (Fig.4e), but liver (Fig.4f). However, the lipid class itself might not be specific to CE, as e.g., in BAT lower carbon fatty acid residues in PC were decreased, whereas longer and higher desaturated fatty acids were increased (Fig.S5b). In line, the patient with biallelic MECR variants showed reduced levels of palmitate and palmitoleate residues in TAGs and CEs compared to plasma of healthy sex- and age-matched controls (Fig.S5c), which might be partially influenced by the low mtFAS level. Based on these observations, we hypothesized that mtFAS and subsequently identified mtFAS-associated lipids might also be disturbed in patients with confirmed other mitochondriopathies.

• Download figure
• Open in new tab

Figure 4: Functional characterization of variants in MECR reveal disturbed glucose and lipid homeostasis.

HEK293T cells overexpressed either MECR wild type or indicated variants for 24h and (a) PDH activity (n=7-8), (b) ³H-deoxyglucose uptake (n=5-6) or (c-d) targeted lipidomics (n=3) were performed. (e) Plasma lipidomics of Adipoq-Cre or (f) Alb-Cre mice. (g) Lipidomic analysis of 300 plasma samples from patients with genetically confirmed mitochondriopathies were analyzed with regard to MECR overexpression-associated lipid species found in (d) and were grouped based on the genes MT-ND6 (low), ETFDH and POLG (high) into low and high mtFAS-associated CE patients, which were grouped in (f) with regard to their plasma lipid classes and the respective length and desaturation. Error bars indicate standard error of the mean (SEM). *= p<0.05 by two-way ANOVA and Fishers-LSD Test.

Therefore, we screened a cohort of over 2000 patients with confirmed or suspected mitochondrial disease from the mitoNET consortium and selected adult cases, who had a genetically confirmed diagnosis and took part in biobanking (n=300). We performed lipidomic analysis of their plasma samples focusing on the respective mtFAS-associated CEs identified to be increased by MECR expression. We found that compared to the cohorts mean CE abundance especially patients with pathogenic variants in MT-ND6 (n=28) showed lower mtFAS-associated CEs whereas patients with pathogenic variants in POLG and ETFDH showed higher mtFAS-associated CEs (n=41) (Fig.4g). Based on the abundance of these CEs we defined subgroups with low and high mtFAS-associated CE amongst these patients, which of note, are also influenced by many other factors. Next, plotting the plasma concentration of all lipid species analyzed (sorted by their length and degree of desaturation) in the low and high mtFAS subgroup revealed that especially CEs, free fatty acids (FFA), hexosyl ceramides (HCER), lactosyl ceramides (LCER), PC and SM were distinguishing these two groups (Fig.4h). In sum, this indicates that multiple lipids are affected by the MECR, that specific CE lipid species may be affected in vivo by mtFAS and thus it is tempting to speculate that also other, currently non-mtFAS-associated mitochondriopathies may show an intrinsic dysregulation in mtFAS activity.