Oxidative phosphorylation and fatty acid oxidation in slow-aging mice

Abstract

Oxidative metabolism declines with aging in humans leading to multiple metabolic ailments and subsequent inflammation. In mice, there is evidence of age-related suppression of fatty acid oxidation and oxidative phosphorylation in the liver, heart, and muscles. Many interventions that extend healthy lifespan of mice have been developed, including genetic, pharmacological, and dietary interventions. In this article, we review the literature on oxidative metabolism changes in response to those interventions. We also discuss the molecular pathways that mediate those changes, and their potential as targets for future longevity interventions.

Graphical Abstract

1. Introduction

Aging leads to gradual structural and functional decline of multiple biological systems. Aging is usually accompanied by increased incidence of multiple forms of pathology and age-related diseases, which in humans include diabetes, dementia, cataracts, sarcopenia, decline in protective immunity, cancer, and osteoarthritis, among many others. In humans, many age-related diseases are accompanied by, and in some cases seemingly caused by, metabolic dysregulation, increased inflammation, or both. Metabolic syndrome is one of the most common phenotypes seen in aged people, and is characterized by high blood pressure, hypertriglyceridemia, obesity, and elevated blood glucose. Other mammals, including rodents, have a related age-dependent dysregulation of nutrient sensing and decline of systemic oxidative metabolism.

Oxidative metabolism is made up of two closely associated processes; fatty acid oxidation (FAO) and oxidative phosphorylation (OXPHOS) [1, 2]. FAO takes place in the mitochondria or the peroxisome [3], and there are significant physical and functional interactions between the two processes. For example, inhibition of FAO activity can impair biogenesis of OXPHOS proteins, and alter their stability and activity [4]. Conversely, genetic disorders of OXPHOS proteins can impede different steps of FAO. Enzymes involved in both pathways have been reported to physically interact in the cell, and to coprecipitate in polyacrylamide gel electrophoresis experiments [5–7].

FAO and OXPHOS both decline with age [8–11]. In humans, aging is associated with increased body fat percentage and decreased rate of postprandial fat oxidation [8]. The rate of lipid oxidation during exercise declines with age, even with equivalent total caloric expenditure, indicating an age-related shift in fuel dependency in humans from fat oxidation to carbohydrate metabolism [12]. OXPHOS is impaired in multiple tissues with age. In humans and rats, the rate of OXPHOS in liver declines with aging [13–15]. This is partially explained by the age-related decline in activities of complex I and II, and in respiration with NADH-dependent substrates [13,16]. In both humans and mice, nuclear encoded OXPHOS genes in liver are reduced with aging as a conserved transcriptomic signature [17,18]. The decline of OXPHOS with aging is not limited to the liver. In the heart of aged rats, the mitochondrial yield, and the enzyme activities of some electron transport chain (ETC) complexes decline with aging [19,20]. This phenomenon, however, is exclusive to one of two mitochondrial populations found in the cardiac tissue of rats, i.e. the interfibrillar mitochondria, but is not seen in the subsarcolemmal mitochondria [21]. The mitochondrial yield, but not maximal OXPHOS, also declines in aged hindlimb muscles from rats [22]. In rat brain, similar to liver, cytochrome oxidase activity declines with age [23,24].

Drugs and genetic interventions that can extend lifespan have now been developed for multiple model organisms, including mice. Elucidating the molecular mechanisms that underlie the lifespan and health benefits of such drugs can lead to further discoveries and optimization of lifespan extending interventions. The correlation between aging and declining FAO and OXPHOS does not necessarily imply a causal link between these two pathways and enhanced longevity.

In this article, we will review current data on the changes that occur in FAO and OXPHOS in long-lived mice, and the underlying molecular mechanisms that can tie those two pathways to longevity. Since different tissues differ in their reliance on oxidative metabolism and their energy needs, we review the current literature for each of the tissues studied, focusing on mammalian and especially murine data.

2. Oxidative metabolism phenotypes in various tissues of long-lived mouse models

2.1. Global and hepatic metabolism

The liver is frequently studied in murine longevity studies, partly due to its importance in regulating systemic metabolism and inflammation. Multiple transcriptomic and proteomic screens have been analyzed to study shared mechanisms of lifespan extension across different murine longevity models. In one study, many interventions that extend lifespan have been shown to upregulate FAO, OXPHOS, and tricarboxylic acid (TCA) functions according to functional enrichment analysis of liver RNA-seq data [17,25]. Upregulation was noted in mice treated with acarbose, rapamycin, or 40 % caloric restriction (CR), as well as in growth hormone receptor knockout mice (GHRKO), and Snell dwarf mice. The same changes were noted in male (but not female) mice given 17-a-estradiol or placed on a methionine-restricted diet. This is particularly interesting since 17-α-estradiol only extends lifespan in male mice. It is important to note that the MR experiment study was done using seven-weeks old male C57BL/6J mice, while CR and drug treatments were started on four-months old male and female UM-HET3 mice, in accordance with the protocol of the Interventions Testing Program (ITP) protocol [26–29]. Snell Dwarf and GHRKO liver samples were taken from five-months old male mice [30,31]. In the same study [17], the authors grouped the different longevity interventions into three main categories; rapamycin treatment, growth hormone (GH) deficiency, and CR. Gene set enrichment analysis (GSEA) revealed that upregulated OXPHOS and TCA were among few shared molecular pathways in response to the three categories. NRF2, PPARα, and PPARγ, three pathways that regulate FAO and OXPHOS, in addition to mitochondrial biogenesis, were found to be upregulated in response to CR or GH deficiency mutations [17].

The same study [17] also identified genes associated with degree of lifespan extension as a quantitative measurement of increased longevity. They utilized a mixed-effect regression model with three metrics: median lifespan ratio, maximum lifespan ratio, and median hazard ratio. They defined median hazard ratio as “the ratio of slopes of survival curves at the timepoint when 50 % of cohort is alive”. Using those metrics, mRNA from genes related to OXPHOS and TCA were also significantly positively associated with lifespan extension [17].

Another study investigated systemic changes in the molecular regulation of biological processes in mice treated with acarbose, 17-α-estradiol, or rapamycin, or undergoing CR [32]. This study used the transcriptomic dataset from the previously mentioned paper from the Gladyshev lab [17], and two mouse liver proteomic datasets from the NIA Longevity Consortium. The analyses focused on data from calorically restricted mice, and rapamycin, 17-α-estradiol, or acarbose treated mice. The experiments were started using 4-month-old male and female UM-HET3 mice. While those four lifespan-extending interventions have distinct mechanisms of action, they all led to tightening the regulation of biological modules including FAO, protein transport to peroxisome, and TCA cycle. Acarbose and CR particularly shared significant enrichment in FAO and central energy metabolism, while 17-α-estradiol and rapamycin similarly modulated mitochondrial energy metabolism proteins [32]. Overall, the four interventions showed consistent tightening of fatty acid β-oxidation at both protein and transcript levels, while transport of proteins into peroxisomes was tightened at the protein level. Those findings reinforced the conclusion from the previous transcriptomic screen in implying that mechanistically distinct lifespan-extending interventions lead to rerouting of energy metabolism to FAO via different mechanisms [17,32]. It is important to note that the analyses from this paper are different from previous studies in that they only highlight which pathways are commonly regulated in response to the four different longevity interventions, with no indication of the functional consequences of this regulation. In other words, this study demonstrates that FAO is an important target for different longevity interventions but does not indicate whether it is consistently upregulated or downregulated.

Since CR is one of the most consistently effective lifespan-extending interventions, many groups focus on studying its metabolic consequences. One such study focused on shared transcriptomic signatures in cerebral cortex, hippocampus, cochlea, heart, liver, kidney, white adipose tissue, and gastrocnemius of CR mice [33]. They found that the most robustly induced gene sets were involved in mitochondrial structure, OXPHOS, mitochondrial carboxylic acid and ketone metabolism. This transcriptomic signature was not limited to ETC transcripts alone, and instead indicated a broad reprogramming of energy metabolism. This was also corroborated by another study that showed that a shift in systemic fatty acid metabolism occurred within just two days of starting CR [34]. This shift had an initial phase of increased fatty acid synthesis, followed by a prolonged period (18–20 h) of FAO, indicated by a respiratory exchange ratio of 0.7. Although CR mice consumed less fat from food, they oxidized four times as much fat per day as ad libitum mice [34]. Furthermore, studying intermediate metabolites of hepatic FAO revealed a significant change in liver acylcarnitine profile [35]. CR also upregulated hepatic expression of enzymes involved in FAO including acyl-CoA dehydrogenase long-chain and very-long-chain Acyl-CoA (ACADL and ACADVL) and enzymes involved in peroxisomal FAO and acyl carnitine conversion including carnitine acetyltransferase (CrAT). Blood ketone bodies were increased under CR indicating a shift towards ketogenesis, which could stem from elevated Acetyl-CoA generated from FAO [35].

GHRKO mice also share some of these distinct metabolic phenotypes. Metabolic assays indicated that male and female GHRKO mice display increased oxygen consumption and have a shift in energy metabolism towards higher FAO as indicated by lower respiratory quotient (RQ) values [36–38]. The increase in oxygen consumption of male GHRKO mice was shown to be significant whether they were fed low or high fat diet. Similarly, Ames dwarf mice, which are also long-lived and deficient in growth hormone signaling, displayed significantly lower RQ compared to their controls. In contrast, transgenic mice with elevated growth hormone signaling had significantly higher RQ [37].

Additionally, our group has recently reported consistent and significant upregulation of hepatic mitochondrial FAO and OXPHOS proteins in GHRKO and Snell Dwarf mice [39]. We found evidence for increases of many key enzymes in the FAO pathway including the enzyme responsible for fatty acid transport into the mitochondria, carnitine palmitoyltransferase 2 (CPT2), and the enzymes responsible for the four steps of mitochondrial β-oxidation: acyl-CoA dehydrogenase medium chain (ACADM), enoyl-CoA hydratase, short Chain 1 (ECHS1), hydroxyacyl-CoA dehydrogenase (HADH) and acetyl-CoA acyltransferase 2 (ACAA2) (Fig. 1A). This is consistent with previous reports that ACADL is elevated in Snell Dwarf livers, and hepatic ACADM and ACADL are upregulated in CR mice. Proteins from different complexes of ETC were also elevated, including subunits from complex I (NDUFAB1, NDUFAF7, NDUFB11, NDUFS1), complex II (SDHA), complex III (UQCRB, UQCRC1), complex IV (COX IV), and complex V (ATP5a) (Fig. 1B). Furthermore, enzymes involved in peroxisomal FAO were also upregulated in both models, including the enzyme responsible for fatty acid transfer into the peroxisome, ATP binding cassette subfamily D member 2 (ABCD2), and the enzymes responsible for the four peroxisomal FAO steps: acyl-CoA oxidase 1 (ACOX1), enoyl-CoA hydratase (ECH1), enoyl-CoA hydratase And 3-hydroxyacyl CoA dehydrogenase (EHHADH), and acetyl-CoA acyltransferase 1 (ACAA1), with a few exceptions in which upregulation of a specific enzyme was significant in one model, but not the other (Fig. 1C). Finally, qRT-PCR showed upregulated of mRNA for most of these enzymes, implying a high degree of transcriptional control for these proteins [39].

Fig. 1.

Pathways involved in oxidative metabolism in the mitochondria: (A) mitochondrial fatty acid oxidation, (B) oxidative phosphorylation, and in the peroxisome: (C) peroxisomal fatty acid oxidation.

While metabolic and fuel utilization phenotypes of GHRKO and CR mice have been well documented, other longevity models have not been as thoroughly studied. PTEN overexpression in mice (Pten^tg) leads to lifespan extension [40]. Similar to CR and GHRKO mice, Pten^tg mice have higher insulin sensitivity, higher metabolic rate, and higher resting metabolic rate relative to lean mass. However, RQ was not significantly changed in Pten^tg mice. No data on hepatic OXPHOS or FAO enzyme levels or activities have been reported for this model [40].

Treatment of mice with 14.4 ppm 17-α-estradiol also led to metabolic benefits in male mice including suppression of de novo lipogenesis and activation of β-oxidation [41]. This phenotype was stronger in male mice than in females and was accompanied by an upregulation of PPARα gene expression. Interestingly, those effects were abolished in mice with a global knockout of the estrogen receptor α (ERα) gene. Treatment of male C57BL/6 mice on normal or high fat diet with canagliflozin for 14 weeks also increased fatty acid oxidation, reduced body weight and fat mass, and increased hepatic expression of CPT1a, ACOX1, ACADM, and PGC-1α [42,43]. Canagliflozin also upregulated CPT1a, ACOX1, and ACADM in AML12 hepatocytes [43].

Studies of rapamycin effects on FAO and OXPHOS proteins have led to inconsistent results. Rapamycin regulates mitochondrial proteins involved in energy metabolism. In one study, involving responses to a high fat diet, rapamycin treatment led to leaner mice with higher energy expenditure and less insulin resistance [44]. This study was conducted on male and female C57BL/6JRj mice that were fed a high fat diet starting at 6 weeks of age for 5 weeks before starting the rapamycin treatment, given as a weekly intraperitoneal injection at a dose of 2 mg/kg body weight [44]. In a separate study, rapamycin treatment was started at four months of age in male and female UM-HET3 mice, at a dose (14 ppm) that increased lifespan in both sexes, and led to significantly increased plasma glucose level (both sexes), increased circulating FGF-21 level (significant only in males), and increased leptin levels/Each of these changes is in the opposite direction of those induced by CR [29]. Another study, by Liu et al. reported that chronic rapamycin treatment in both inbred (C57BL/6J) and genetically heterogeneous (UM-HET3) male mice led to changes in glucose metabolism including glucose intolerance, hyperglycemia, and insulin resistance. The treatment in this study was also administered as a diet containing 14 ppm encapsulated rapamycin. The age of the C57BL/6J mice was 2–3 months old, while the age of UT-HET3 mice was 10–12 months old. The detrimental metabolic effects in this study are, again, in contrast to other longevity models including GHRKO, Snell Dwarf, and CR mice. Those impairments in glucose metabolism were induced in mice fed low-fat or high-fat diets. T Liu et al. found that these changes were reversible and were quickly alleviated once rapamycin treatment was stopped [45].

Early treatment of male and female UM-HET3 mice with rapamycin for the first 45 days of life led to enhanced glucose and insulin resistance, lower frailty index score, and extended lifespan in male mice [46]. In this study, a diet containing 42 ppm of encapsulated rapamycin was given to the dams for the first 45 days after giving birth. The health and lifespan benefits were accompanied by hepatic transcriptomic signatures that were opposite to age-related changes in young (2 months-old) and middle-aged (20-22 months-old) mice. Overall, rapamycin led to increased expression of liver metabolic genes and decreased expression of pro-inflammatory genes. Gene Set Enrichment Analysis (GSEA) revealed that the metabolic pathways that were upregulated in response to early rapamycin treatment included fatty acid metabolism, OXPHOS, and peroxisome. This effect was noted in both male and female rapamycin-treated mice but was stronger and lasted for a longer period in male mice. The reported upregulation of oxidative metabolism gene expression profile in this study was consistent with other longevity interventions mentioned earlier including growth hormone signaling deficiency, CR, and rapamycin at adult age. The results from this study can be reconciled with the Liu et al., 2014 paper since the rapamycin treatment in this study was introduced early in life then stopped, which may have allowed the metabolic side effects of rapamycin to subside [45,46].

2.2. Heart

The heart has a high metabolic rate and depends on carbohydrates and fat for energy, with FAO being the major source of energy in the adult heart. While it is not as well studied as liver, high-density oligonucleotide microarrays revealed that aging in mouse hearts leads to transcriptional changes that indicate a shift from fat to carbohydrate metabolism [47]. Several genes involved in FAO were downregulated with aging, including CPT1, CPT2, ECHS1, and CRAT. CR, when started at 14 months of age, completely or partially prevented most of the aging-related transcriptional changes in FAO [47].

Acarbose also leads to protective effects in the cardiac tissue of mice when started at 4 months old [48]. For example, acarbose attenuated male-specific age-related cardiac hypertrophy, reduced activation of growth-promoting pathways, and increased expression of peroxisomal proteins functionally involved in lipid metabolism. Acarbose also inhibited age-associated changes in cardiac lipids that are also associated with cardiovascular diseases and cardiac disfunction in humans [48].

CR and rapamycin treatment has both been shown to treat age-dependent diastolic dysfunction and cardiac hypertrophy in old mice [49]. Additionally, hearts from 27-month-old female C57BL/6 mice demonstrated age-related decreases in proteins involved in mitochondrial function, electron transport chain, citric acid cycle, and fatty acid metabolism when compared to hearts from 5-month-old mice. Those proteomic changes were significantly reversed after 10 weeks of CR or rapamycin treatment rapamycin (2.24 mg/kg/day) [49]. Another study by the same group showed that rapamycin has beneficial effects on the heart start just 2–4 weeks after beginning treatment. Rapamycin, in addition to inhibiting mTORC1 signaling in the heart, remodeled cardiac energy metabolism via upregulation of mitochondrial biogenesis markers (PGC-1α and TFAM), induced autophagy reversing the age-related decline in cardiac FAO, and restored a fuel substrate utilization that resembles younger mice [50]. This study also used 14 ppm encapsulated rapamycin in the diet and was done on young (3-5 months-old) and old (24–26 months-old) female C57BL/6 mice.

2.3. Adipose tissue

Although adipose tissue is critically involved in regulation of lipid metabolism, there are relatively little data on abundance or activity of FAO enzymes or OXPHOS complex proteins in this tissue. A recent study utilized GSEA to study the available data on transcriptomic responses of white adipose tissue (WAT) to different lifespan-extending interventions. They found that the top shared upregulated genetic signatures included OXPHOS, amino acid metabolism, and ribosome structural genes [17]. An interesting caveat here is that similar to genetic screens of the liver, different longevity interventions affected different sets of genes that converged to the same pathways.

GHRKO mice seem to have a significant increase in the ratio of WAT and brown adipose tissue (BAT) whole body weight [36,51–53]. Compared to WT littermates, GHRKO mice had an increase of ≈10 % in ratio of WAT weight to body weight, and a 75 % increases in the ratio of BAT to body weight. The increase in BAT mass is accompanied by upregulated uncoupling protein-1 (UCP1), which is usually associated with increased energy expenditure and OXPHOS activity [36,54].

Other functions of adipose tissue may also contribute to the longevity phenotype. For example, since CR mice oxidize more fat than ad libitum mice, they need to balance this process by a significant increase in adipose tissue fatty acid synthesis [34]. This was accompanied by increased levels of sterol regulatory element-binding protein-1c (Srebp-1c), a master regulator of fatty acid biosynthesis. In this context, Srebp-1c enhanced fatty acid synthesis by upregulating mRNA levels of fatty acid synthase (Fasn), acetyl-CoA carboxylase (Acc), ATP citrate lyase (Acly), and Me-1 proteins, in WAT of both fed and fasted CR mice. Srebp-1c also elevated mitochondrial biogenesis via upregulation of peroxisome proliferator-activated receptor gamma coactivator-1α (Pgc-1α) and COX IV mRNAs [34,55].

Adipose tissue plays a key role in oxidative metabolism. In many longevity models, WAT has more beiging compared to controls, BAT is larger in size [54,56–58]. GHRKO, Snell Dwarf, and CR mice all have higher WAT and BAT weight relative to body weight compared to their WT and control littermates [36,54]. BAT is especially important because it plays an important role in thermogenesis and systemic regulation of body temperature, which is mostly dependent on the actions of uncoupling protein 1 (UCP1). UCP1 uncouples the ETC from the last step of OXPHOS, utilizing the produced energy in thermal regulation instead of storing it for future use. Upregulated activity of UCP1 is often correlated with increased levels of FAO and increased activities of initial OXPHOS complexes [54,59,60].

In contrast to the relative high weight of WAT in Snell dwarf mice and CR mice, Pten^tg mice have a relative decrease in fat mass and epididymal WAT compared to WT controls. Pten^tg mice, however, shared with CR and GHRKO mice the phenotype of increased energy expenditure, and displayed higher glucose uptake in BAT, along with higher UCP1 and Pgc-1α expression [40].

2.4. Skeletal muscles

Skeletal muscles are also highly metabolic and depend on oxidative metabolism for function. There are no data available on the relative levels or activities of FAO enzymes or OXPHOS subunits in skeletal muscles of long-lived mice compared to their controls. Human data show a strong correlation between aging and a decline in mitochondrial metabolism and OXPHOS in skeletal muscle. For example, in human vastus lateralis muscles, oxidative capacity per muscle volume significantly declined from adult to elderly subjects [61]. The age-associated decline in OXPHOS capacity in muscles has been implicated in elevated muscle insulin resistance, intramyocellular lipid accumulation, and reduced resting metabolic rate (RMR) in muscles [62,63]. Additionally, muscles with more oxidative fibers (e.g., soleus muscle) maintained their mitochondrial homeostasis and were more protected from mitochondrial aging compared to muscles with more glycolytic fibers (e.g., tibialis anterior) [64].

Despite the lack of data on FAO and OXPHOS markers or activity in long-lived mice, there are data that suggest that muscles play an important endocrine role in regulating oxidative metabolism in adipose tissue. Deletion of growth hormone receptor in fat or liver tissue did not recapitulate the anti-inflammatory effects of global GHRKO mice in WAT and BAT tissues, but deletion of growth hormone receptor in muscle tissues led to changes in BAT and WAT, including increases in UCP1, decreases in pro-inflammatory M1 macrophages, and increases in anti-inflammatory M2 macrophages. The indirect effect of GHR in muscle on fat tissue was explained, at least partially, by increases in muscle FNDC5 and in plasma levels of its cleavage product, irisin, in mice with global or muscle-specific deletion of growth hormone receptor. Plasma irisin is thought to mediate multiple benefits in adipose tissue including anti-inflammatory macrophage polarization as well as UCP1 and PGC-1α upregulation [54].

2.5. Cultured Cells

Studies of anti-aging drugs on cultured cells may provide some insights into the effects of these agents on cellular FAO and OXPHOS.

Rapamycin, which acts via inhibition of mTORC1, led to increased FAO in skeletal muscle cells (L6 myotubes) [65]. This was accompanied by a significant increase in CPT1 and CPT2 activities. Rapamycin also significantly decreased glucose transport capacity, glycogen synthesis, and glycolysis, indicating a shift in cellular fuel utilization from carbohydrates to fat. This is achieved mechanistically by rapamycin-mediated inhibition of mTOR phosphorylation at S2448 and S2481, as well as inhibition of p70 S6 kinase phosphorylation. Rapamycin also prevented high levels of insulin (110 nmol/L for 24h) from suppressing FAO in L6 myotubes [65].

The results of rapamycin on skeletal muscle cells are consistent with its effects in other mammalian cell types [65]. In primary culture of rat hepatocytes, rapamycin significantly increased FAO and inhibited esterification of exogenous fatty acids. It also led to inhibition of lipogenic pathways, non-insulin-dependent glucose transport, and glycogen synthesis [66]. In contrast to the effects on cultured muscle cells, glucose utilization in hepatocytes was unaffected by rapamycin.

In 3T3-L1 adipocytes, canagliflozin upregulated the mRNA and protein levels of many genes involved in oxidative metabolism, including PFC-1α, TFAM, COX5b, and CPT1b, in a dose-dependent manner. Interestingly, those effects were inhibited by the PPARα antagonist GW6471 [42]. Consistently, canagliflozin also upregulated FAO in renal tubular epithelial cells (RTECs) via upregulating FOX1A and CPT1a [67].

Research on PTEN in primary bovine hepatocytes suggests a role for PTEN in regulation of FAO. Overexpression of PTEN using an adenoviral vector significantly upregulated CPT1, ACSL, and HADH expression. Genes involved in assembly and secretion of VLDL, including ApoB100, ApoE, and MTP, were also significantly upregulated. Additionally, levels of triglycerides were decreased, while levels of VLDL were increased, confirming the modulation of the FAO pathway [68]. Moreover, knockdown of PTEN in the same cells led to downregulation of FAO enzymes ACSL, CPT1, CPT2, and HADH on the transcript and protein levels. PTEN knockdown also decreased mRNA and protein levels of VLDL components including ApoB100, ApoE, MTP, and LDLR. Finally, PTEN knockdown also led to increased triglyceride accumulation and reduced VLDL secretion [69].

3. The link between oxidative metabolism and longevity pathways

We turn next to the molecular pathways that could link increased oxidative metabolism to lifespan extension. Tissue-specific FAO and OXPHOS can interact with various nutrient-sensing and metabolic signaling cascades, including mTORC1, AMPK, nuclear receptors, etc.). Elucidating which of those signaling pathways contribute to extension of longevity in these mouse models will help in the development of novel interventions with beneficial effects on health and lifespan. The pathways discussed below are connected, and the mechanisms proposed are not mutually exclusive.

3.1. PGC-1α and NCOR1

One relevant pathway involves PGC-1α and NCOR1. PGC-1α is a positive regulator of PPARs and other nuclear receptors that mediate oxidative metabolism, including estrogen-related receptors (ERRs) [70, 71]. NCOR1 opposes the function of PGC-1α, acting as a transcription factor repressor along with its co-factor histone deacetylase 3 (HDAC3) [72–74]. Together, those two nuclear receptor regulators modulate the levels of many enzymes and proteins involved in FAO and OXPHOS. Indirect (via FGF-21 overexpression) or direct upregulation of PGC-1α in vivo has been shown to activate an increased oxidative metabolism phenotype in the liver and skeletal muscles, respectively [75–78]. Muscle-specific knockout of NCOR1 in mice, and in vitro knockout of NCOR1 in adipocytes produced a similar phenotype [72,79,80].

Studies of mice that overexpress fibroblast growth factor 21 (FGF21) suggest a causal relationship between PGC-1α and longer lifespan [75, 81,82]. FGF21 increases energy expenditure and induces thermogenesis in BAT via upregulation of UCP1. It also upregulates PGC-1α protein level and its subsequent oxidative metabolism [75,81], upregulates autophagy genes and possesses an anti-inflammatory effect. A recent study also found FGF21 essential for mediating the lifespan extension induced by protein restriction. Mice lacking FGF21 failed to display the metabolic or the lifespan benefits of protein restriction [83,84].

In GHRKO livers, PGC-1α and forkhead box protein O1 (Foxo 1) are both upregulated, as well as the gluconeogenic enzymes phosphoenolpyruvate carboxykinase (PEPCK) and glucose-6-phosphatase (G6Pase), which are downstream targets of PGC-1α [85]. Furthermore, at 21 months of age, GHRKO livers display higher levels of PPARγ and PPARα than WT mice at both mRNA and protein levels [85]. At the same age, cardiac muscle of GHRKO mice showed no change in PPARγ level and a decrease in PPARα level, while skeletal muscles of GHRKO mice had lower levels of the PPARγ and PPARα. Additionally, the upregulation of hepatic PPARγ in GHRKO liver appears to start later in life, since research done on 6-month-old mice showed no significant change in PPARγ or ERRα protein levels [39]. It is worth noting that PPARγ activity does not solely depend on its abundance, and PGC-1α upregulation or NCOR1 downregulation led to elevated PPARγ activity that was not necessarily reflected in its protein level [39,70–72,86].

CR has also been hypothesized to work, at least partially, through activation of nuclear receptors. The basis for this hypothesis was that many compounds that mimic CR transcriptional effects in mouse liver were found to be nuclear receptor agonists, including agonists for PPARα and liver x receptor (LXR) [87]. In fact, almost 20 % of all gene expression changes induced by CR in mouse liver were PPARα targets. Many of those transcriptional changes were involved in fatty acid omega oxidation, lipid metabolism, cell fate, and inflammation [87]. Another reason for connecting CR to PGC-1α and PPARs was that the ability of CR to protect liver against carcinogens like thioacetamide was lost in mice lacking PPARα [87]. Interestingly CR led to higher PPARα protein and mRNA levels in livers of WT and GHRKO mice but did not have an effect of PPARγ in either WT or GHRKO livers [88].

CR in mice also leads to upregulation of SIRT1, a positive regulator of PGC-1α, in WAT [70]. SIRT1 acts via increasing PGC-1α activity and stability against proteasomal degradation. This is consistent with the increase in PGC-1α levels -as well as many of its downstream targets-in WAT of CR mice. In CR mice treated with the oxidative stressor paraquat, skeletal muscles displayed higher nuclear PGC-1α accumulation after 1 h and 5 h of treatment compared to ad libitum mice. Consistently, PGC-1α targets COX IV and UCP3 were also accumulated in the nucleus of skeletal muscle cells of CR mice [70].

3.2. mTORC1 inhibition

Nutrient sensing and nutrient homeostasis are tightly regulated by interconnected networks of signaling pathways including those involving AMPK and mTORC1 [89,90]. These pathways maintain the balance between energy utilization and energy production. One of the important nutrient sensors is the mTORC1-dependent protein S6 kinase 2 (S6K2). Interestingly, mice lacking S6K2 display enhanced PPARα activity and enhanced ketogenesis under feeding conditions [91]. Ketogenesis provides an important pathway for energy production that is related to oxidative metabolism, correlates with inhibition of mTORC1 activity, and declines with aging. S6K2 inhibits PPARα transcriptional activity via associating with NCOR1, underscoring the important crosstalk between nutrient sensing and oxidative metabolism signaling pathways [91]. Consistent with these observations, ob/ob mice exhibit significantly higher S6K2 activity, while knockdown of S6K2 increases PPARα transcriptional activity [91].

Another study also found that mTORC1 signaling also controls ketogenesis under fasting conditions. In this study, loss of the mTORC1 inhibitor TSC1 led to a decrease in ketogenesis and ketogenic gene expression, and to an increase in liver size [92]. Moreover, loss of the essential mTORC1 component raptor led to the opposite effect. Inhibition of mTORC1 was again found to be necessary for PPARα ketogenic activity. This study also corroborated the role of NCOR1 in mediating mTORC1 actions, where Inhibition of NCOR1 reactivated hepatic ketogenesis in mice with hyperactive mTORC1 signaling [92].

3.3. Anti-inflammatory signaling pathways

Inflammation and macrophage polarization are also modulated by metabolic reprogramming. Free fatty acids are essential precursors for synthesizing inflammatory mediators, and pro-inflammatory macrophages (M1) display higher levels of fatty acid biosynthesis compared to anti-inflammatory macrophages (M2) [93,94]. Additionally, M2 macrophages rely on increased fatty acid uptake, FAO, and OXPHOS to produce the energy required to repair tissues and resolve inflammation [94]. On the other hand, M1 macrophages rely mainly on glycolysis to generate the energy necessary to kill pathogens and sustain inflammation [95].

To establish a causal link between FAO and macrophage polarization, one study showed that expressing a constitutively active CPT1 enzyme in a macrophage cell culture prevented palmitic acid mediated M1 phenotype activation [96]. Etomoxir, a CPT1 inhibitor that blocks FAO, abolishes IL-4 mediated M2 polarization [96]. Blocking OXPHOS with oligomycin or the mitochondrial uncoupler carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone) also blocks IL-4 mediated M2 activation [97]. Furthermore, IL-4 was shown in an independent study to upregulate mRNA expression of FAO genes in addition to genes involved in fatty acid uptake and PGC1β. FAO itself was also upregulated, an effect that was partially inhibited by knockdown of PGC1β [98]. Both FAO and OXPHOS were essential for IL-4 to block LPS-induced expression of pro-inflammatory cytokines.

In addition to PGC1β, the switch to the oxidative metabolism phenotype in M2 macrophages occurs via activation of PPARγ [99]. Upon stimulation with IL4 or oleic acid, PPARγ was activated and led to upregulating of M2 signature genes in addition to promoting glutamine oxidation which fueled OXPHOS [100–102].

These data suggest an important role for oxidative metabolism in regulating innate and adaptive immunity. However, there are some caveats that require further analyses before reaching a conclusion [98]. First, most of the data presented so far was produced in murine systems. Studies from human macrophages do not suggest a strong correlation or reliance of M2 macrophages on oxidative metabolism [103]. Second, even in murine macrophages, recent data suggest a more complex relationship between macrophage metabolism and polarization. For example, FAO was also required for inflammasome activation in pro-inflammatory M1 macrophages [95]. Finally, an independent study failed to reproduce the inhibitory effect of etomoxir on IL-4 induced gene expression [104].

With respect to connections between inflammation and macrophage polarization in long-lived mice, GHRKO mice have been shown to express higher levels of endothelial nitric oxide (NO) synthase (eNOS) in large vessels, a protein that is known to activate M2 macrophage polarization and suppress M1 phenotype [105]. CR also promoted mitochondrial biogenesis in various tissues of male mice which was achieved via inducing eNOS expression. Consistently, CR in male mice lacking eNOS did not produce the same mitochondrial phenotype [106].

4. Conclusion

The work reviewed above suggests that age-related decline in oxidative metabolism is a shared aging phenotype among humans, rats, and mice, whereas upregulation of transcripts, proteins, and/or activities of OXPHOS and FAO enzymes is a shared feature of many longevity interventions. This effect of age is seen more consistently in highly metabolic organs such as liver and heart. Less information is available on adipose tissue, skeletal muscles, and brain. The murine longevity models that have been shown to demonstrate this transcriptomic and proteomic signature are models with growth hormone signaling disruption (GHRKO and Snell Dwarf mice), CR mice, and to lesser extent rapamycin- and acarbose-treated mice. Other interventions, like PTEN overexpression, have been shown to increase FAO in cultured hepatocytes. This phenotype seems to shift fuel reliance of long-lived mice from carbohydrates to fat oxidation, which preserves metabolic health, mediates metabolic benefits, and inhibits pro-inflammatory phenotypes.

The data on the molecular mechanisms mediating those effects in long-lived mice are limited, with few articles highlighting an upregulation of PGC-1α or a downregulation of NCOR1, two mechanisms that lead to activation of nuclear receptors and transcription factors such as PPARγ, PPARα, and ERRα. Other studies point to increased expression levels of those transcription factors, rather than their regulators. Due to lack of mechanistic data from some long-lived models, we turned to in vitro assays to gain mechanistic insights from interventions such as rapamycin treatment, NCOR1 knockdown, and CPT1 or PTEN overexpression. Those interventions suggest a causal link between mTORC1 inhibition, or PTEN overexpression, and upregulated FAO. They also revealed a mechanistic link between increased FAO in macrophages and inhibition of pro-inflammatory M1 phenotype activation.

The available data suggest that upregulated FAO and OXPHOS correlate with extended lifespan and may contribute to it. In contrast, however, transcriptomic signatures compared across species reveal that genes encoding TCA and OXPHOS proteins negatively correlate with species maximum lifespan. For example, NDUFA9, a gene that encodes an OXPHOS complex I protein, is negatively correlated with maximum lifespan in the brain, liver, and kidney, with the effect being statistically significant in the latter two tissues [25]. Within a species, OXPHOS proteins typically decline with age and increase after anti-aging interventions. In contrast, the OXPHOS proteins are typically diminished in long-lived species compared to short-lived ones. The authors of this study [25] argue that such discrepancy indicates that not all age-related transcriptomic changes are harmful, and that some of those changes are adaptive and may lead to health benefits. This point, while provocative, is difficult to reconcile with data showing that increased oxidative metabolism mediates multiple metabolic benefits [75,81,84,107]. Additionally, the pro-longevity transcriptomic signatures across species and within the same species were poorly correlated throughout the study, including other pathways. For example, Rela (p65) was upregulated with aging within species, downregulated in response to murine pro-longevity interventions, and still positively correlated with longevity across species [25].

The work presented suggests two major questions about the relationship between lifespan extension and upregulated oxidative metabolism. The first is whether upregulated oxidative metabolism is required for the longevity and health benefits of different interventions. The available data imply a strong correlation between lifespan extension and increased oxidative metabolism, particularly in the liver. Analyses of mice lacking proteins essential for FAO (e.g. CPT1 or CPT2) or for OXPHOS (e.g. COX IV) could help to evaluate whether upregulation of either pathway is required for lifespan extension.

A second major question is whether upregulated FAO and/or OXPHOS is sufficient to confer lifespan extension. Studies using interventions that directly upregulate either FAO or OXPHOS would be helpful here. These could include studies of overexpression of FGF21, which increases hepatic FAO, ketogenesis, and insulin sensitivity [75,83,84]. Male and female FGF21-Tg mice demonstrated a 36 % increase in the median survival time compared to WT littermates. The effects of FGF21 on longevity might, however, also reflect its effects on glucose metabolism, NAD + metabolism, downregulation of STAT5, or inhibition of growth hormone signaling [82,108]. A more specific regulator of oxidative metabolism could be required to establish that oxidative metabolism is sufficient for lifespan extension. PGC-1α overexpression is a good candidate for such a study. Muscle-specific overexpression of PGC-1α confers many metabolic and health benefits in skeletal muscles and protects against aging-related molecular changes [77,109,110]. PGC-1α overexpression produces to a small (5 %) but significant increase in lifespan when male and female mice were combined [77,109, 110]. A separate study also suggests a strong link between oxidative metabolism and longevity. In this report [111], treatment of middle-age (14-months-old) mice with the PPARγ agonist rosiglitazone (1 mg/kg/day) led to improved mitochondrial health, metabolism, and cognitive function, and decreased inflammation and tissue atrophy. In this study rosiglitazone significantly extended median lifespan by 11 %. A drawback of this study is that rosiglitazone was evaluated only in male C57BL/6J mice. Pioglitazone, another PPARγ agonist, was also shown to enhance systemic glucose and lipid metabolism in rats, and to protect against aging-related renal injury [111]. The ITP is currently evaluating pioglitazone for lifespan effects, using mice born in 2022. Fenofibrate, a PPARα agonist, was also shown to promote health benefits and protect against aging-related kidney dysfunction, proteinuria, inflammation, glomerulosclerosis, and tubular interstitial fibrosis [112,113]. While no studies on lifespan extension in response to fenofibrate have been conducted in mice, this agent has been shown to extend lifespan in C. elegans [112–114].

Overall, upregulated oxidative metabolism offers an attractive pathway to study in longevity research. The data on whether it can consistently lead to increased lifespan and whether it can apply to humans are promising, but incomplete.