CLPP deficiency protects against metabolic syndrome but hinders adaptive thermogenesis

Abstract

Mitochondria are fundamental for cellular metabolism as they are both a source and a target of nutrient intermediates originating from converging metabolic pathways, and their role in the regulation of systemic metabolism is increasingly recognized. Thus, maintenance of mitochondrial homeostasis is indispensable for a functional energy metabolism of the whole organism. Here, we report that loss of the mitochondrial matrix protease CLPP results in a lean phenotype with improved glucose homeostasis. Whole‐body CLPP‐deficient mice are protected from diet‐induced obesity and insulin resistance, which was not present in mouse models with either liver‐ or muscle‐specific depletion of CLPP. However, CLPP ablation also leads to a decline in brown adipocytes function leaving mice unable to cope with a cold‐induced stress due to non‐functional adaptive thermogenesis. These results demonstrate a critical role for CLPP in different metabolic stress conditions such as high‐fat diet feeding and cold exposure providing tools to understand pathologies with deregulated Clpp expression and novel insights into therapeutic approaches against metabolic dysfunctions linked to mitochondrial diseases.

Introduction

Mitochondrial function is intimately linked to cellular metabolic homeostasis as many metabolic pathways are hosted within the organelle. An emerging role of mitochondria in the regulation and maintenance of systemic energy metabolism has been increasingly recognized 1. Even more evidence also highlights the significance of adaptive responses activated by changes in organelle homeostasis that influence cell, tissue, and ultimately whole‐body metabolism 1, 2. Currently, we witness a move away from a simplistic view on mitochondria as mainly energy‐producing machineries to increasingly complex signaling hubs of cellular and organismal metabolism 2.

Mitochondrial proteases are essential for maintaining mitochondrial function and therefore cell homeostasis 3. This does not only involve protein quality control by degrading misfolded or damaged proteins, mitochondrial proteases also take part in highly regulated proteolytic activities 3. The selective degradation of substrates, which are no longer required, represents an important means to regulate mitochondrial metabolic pathways upon environmental insults, for example, nutrient excess or starvation. Mitochondria harbor a large set of proteases involved in these quality control and surveillance functions, including the major mitochondrial matrix, ATP‐dependent Clp protease (CLPP).

CLPP possesses only peptidase activity and therefore assembles with the chaperone CLPX to form a proteasome‐like complex. CLPX functions in substrate recognition, ATP‐dependent protein unfolding, and translocation of the substrate into the proteolytic chamber of CLPP 4. Although ClpXP has been extensively characterized in prokaryotes, the role of mammalian ClpXP is far less understood. In humans, recessive CLPP mutations, causing a complete loss of the protein, cause Perrault syndrome, characterized by premature ovarian failure, sensorineural hearing loss, and growth retardation 5. We recently showed that CLPP regulates mitochondrial translation by modulating levels of its substrate, ERAL1 that in turn coordinates mitoribosomal assembly and function 6. In a separate study, we disputed a role of CLPP in activation or maintenance of mitochondrial unfolded protein response (UPR^mt) while showing that loss of CLPP alleviates the mitochondrial cardiomyopathy by increasing synthesis rates of OXPHOS subunits and reducing levels of toxic aborted peptides 7. Through a proteomic approach and the use of a catalytically inactive CLPP, we also produced the first comprehensive list of possible mammalian ClpXP substrates that contains a number of metabolic enzymes primarily involved in fatty acid β‐oxidation (FAO) 6.

With this result and the notion that both Perrault syndrome patients and CLPP‐deficient mice show smaller body size, we aimed to understand the role of CLPP in control of cellular and systemic metabolism. CLPP‐deficient mice were fed normal chow diet (NCD) or high‐fat diet (HFD) to elucidate the function of CLPP under different metabolic conditions in order to establish its relevance for whole‐body physiology.

Results

Loss of CLPP does not affect life span but results in growth retardation and increased energy expenditure

To unravel the in vivo function of CLPP in mammals, we recently generated whole‐body CLPP‐deficient mice (Clpp ^−/−) 6. Although CLPP function seems to be necessary in some critical period during development 6, the deficiency did not influence the postnatal life span, as no changes in the median or maximal survival for both sexes of Clpp ^−/− mice were detected (Fig EV1A), in contrast to a previous study on a different CLPP‐deficient model that reported a strong reduction in life span to about 1 year of age 8. Possible reasons that might account for the discrepancy could be the different techniques used to generate the CLPP‐deficient mice (gene trapping versus gene targeting) or differences in the hygienic status of the animal facilities. It seems that the effect of CLPP on the life span is species‐dependent, as in the filamentous fungus Podospora anserina CLPP deficiency leads to an increased life span 9.

Figure EV1. Phenotypical and metabolic characterization of CLPP‐deficient mice fed NCD.

1. Kaplan–Meyer survival curves of male (n = 9–14) and female (n = 4–7) mice. Median survival of male control (Clpp ^+/+) mice was 117 weeks and Clpp ^−/− 121 weeks. Median survival of female control mice was 131 weeks and of Clpp ^−/− is 110 weeks.
2. Body weight of female mice monitored till 15 weeks of age (n = 8–9).
3. Total bone marrow density (n = 6–7).
4. Relative Igf1 transcript levels in liver of 5‐ and 15‐week‐old mice (n = 4).
5. Food intake normalized to lean body mass and water intake of male mice during the light and dark phase (n = 7).
6. Average activity and rearing activity per hour during the light and dark phase (n = 7).

Data information: Bars represent mean ± SD (unpaired two‐tailed Student's t‐test; **P < 0.01, ***P < 0.001).

At the time of weaning, it became obvious that Clpp ^−/− mice are smaller with a weight reduction of approximately 25–40%, when fed normal chow diet (NCD) that persisted throughout their life span (Figs 1A and B, and EV1B), in agreement with results from human patients with Perrault syndrome 5. The overall fat content of Clpp ^−/− mice was reduced by more than 50% (Fig 1C), while the lean mass decreased to a lesser extent (around 25–30%) in both genders (Fig 1D). The significant loss of body weight was accompanied by a reduced size of Clpp ^−/− mice as confirmed by the measure of femur length (Fig 1E). The growth defect also manifested as a significant decrease in the femur bone mineral density (BMD) of around 8–14% (Fig EV1C). To determine whether alterations in the growth hormone/insulin‐like growth factor 1 (GH/IGF1) axis were contributing to the Clpp ^−/− growth defect, the Igf1 transcript levels were analyzed 10, 11. We did not observe a decrease in Igf1 transcript levels in liver, which is the main source of circulating IGF1 that promotes the linear growth effect of pituitary GH, suggesting a normal GH function in Clpp ^−/− mice at 5 or 15 weeks of age (Fig EV1D).

Figure 1. CLPP‐deficient mice display a favorable metabolic phenotype when fed NCD.

• A
  Representative image of male control (Clpp ^+/+) and Clpp ^−/− mice revealing the difference in body size.
• B
  Body weight of male mice fed normal chow diet (NCD) (n = 9–10).
• C, D
  Body fat mass (C) and lean body mass (D) determined by NMR at 15 weeks of age (male n = 8 and female n = 7).
• E
  Femur length measured between femur head and lateral condyle (male n = 6–7, female n = 6).
• F
  Respiratory exchange ratio (RER) over a period of 24 h (n = 7).
• G
  Energy expenditure corrected for the lean body mass during the dark and light phase (n = 7).
• H
  Energy expenditure adjusted by ANCOVA during the dark and light phase (n = 7).

Data information: Bars represent mean ± SD (unpaired two‐tailed Student's t‐test; ***P < 0.001).

To further investigate the energy homeostasis of Clpp ^−/− mice, indirect calorimetry was performed. No significant difference in the food and water consumption was detected, ruling out a decreased caloric intake as a reason for the lean phenotype of Clpp ^−/− mice (Fig EV1E). Clpp ^−/− mice displayed a strong decrease in locomotor and rearing behavior during the active/dark phase, indicating that physical activity also does not contribute to the lean phenotype (Fig EV1F).

Normal respiratory exchange ratio (RER = CO₂ produced/O₂ consumed) suggested overall unaffected metabolic fuel preference in Clpp ^−/− mice (Fig 1F). As Clpp ^−/− mice are substantially lighter than controls, we initially plotted energy expenditure (EE) divided into lean mass and detected an increased EE in the active/dark phase in Clpp ^−/− mice (Fig 1G). However, the significance disappeared upon regression adjustment for lean mass using ANCOVA tool 12, prompting us to conclude that there is no difference in the EE between Clpp ^−/− and control mice (Fig 1H).

Clpp ^−/− mice have improved glucose homeostasis and a fasting‐like phenotype

We next examined the glucose metabolism of Clpp ^−/− mice and detected elevated insulin sensitivity and improved glucose tolerance (Fig 2A and B) that was not accompanied by lower circulating insulin levels (Fig 2C). Blood glucose levels were significantly reduced in Clpp ^−/− mice, in both fed and fasted condition, without an effect on hepatic gluconeogenesis (Fig 2D and E) or on the PEPCK levels (a key rate‐limiting gluconeogenic enzyme) (Fig EV2A). The observed fasting phenotype was accompanied by a strong increase in protein levels of the insulin‐dependent glucose transporter 4 (GLUT4) in skeletal muscle and epididymal white adipose tissue (EWAT), whereas no change was observed in brown adipose tissues (BAT) (Figs 2F and EV2B). In contrast, the abundance of the insulin‐independent glucose transporter 1 (GLUT1) was not changed in the liver and skeletal muscle (Fig 2F and Appendix Fig S1A).

Figure 2. Loss of CLPP leads to enhanced glucose homeostasis.

• A, B
  (A) Insulin tolerance test (n = 10) and (B) glucose tolerance test (n = 10) on control (Clpp ^+/+) and Clpp ^−/− mice fed NCD.
• C
  Insulin serum concentration in the fed state at 16 weeks of age (n = 10).
• D
  Fed and fasted blood glucose levels at 15 weeks of age (n = 5).
• E
  Glucose output from primary hepatocytes after stimulation with glucagon and cAMP for 6 h in the presence of gluconeogenic precursors pyruvate and lactate (n = 3–4).
• F
  Left: Western blot quantification of GLUT4 in skeletal muscle (SkM), BAT, and EWAT (n = 4). Right: Western blot quantification of GLUT1 in liver and SkM (n = 4).
• G
  Ratio of phosphorylated (AMPK^Thr172) and non‐phosphorylated AMPK in liver, SkM, and BAT as determined by Western blot quantification (n = 4).

Data information: Bars represent mean ± SD (unpaired two‐tailed Student's t‐test; *P < 0.05, **P < 0.01, ***P < 0.001).

Figure EV2. Serum markers of CLPP‐deficient mice during NCD and HFD feeding.

1. Western blot analysis of PEPCK in liver from control (Clpp ^+/+) and Clpp ^−/− mice (n = 4).
2. Western blot analysis of GLUT4 in SkM, BAT, and EWAT from control (Clpp ^+/+) and Clpp ^−/− mice fed NCD (n = 4).
3. Analysis of serum metabolic markers of 16‐week‐old mice fed NCD (n = 10).
4. Analysis of serum metabolic markers in HFD fed mice at 16 weeks of age (n = 6–8).

Data information: Bars represent mean ± SD (unpaired two‐tailed Student's t‐test; *P < 0.05, **P < 0.01, ***P < 0.001).Source data are available online for this figure.

Energy deficiency is sensed by the AMP‐activated protein kinase (AMPK) that plays a key role in the regulation of glucose and lipid metabolism in response to various stimuli. Intriguingly, the pAMPK/AMPK ratio was significantly increased in skeletal muscle, but not liver or BAT of Clpp ^−/− mice (Fig 2G and Appendix Fig S1B), indicating various energy statuses for these tissues, possibly due to a different level of respiratory chain deficiency 6.

Analysis of different hormones and cytokines involved in the energy balance revealed no changes in serum ghrelin, glucagon‐like peptide 1 (GLP‐1), glucose‐dependent insulinotropic polypeptide (GIP), and resistin levels in Clpp ^−/− mice (Fig EV2C). Normal GLP‐1 and GIP levels correspond to comparable insulin levels in control and Clpp ^−/− mice and suggest a functional intestinal nutrient absorption 13. Consistent with a fasting phenotype, serum leptin, a central satiety agent reflecting a decrease in fat mass, was diminished, whereas glucagon, a hormone that is supposed to increase blood glucose levels, was increased in Clpp ^−/− mice (Fig EV2C). Unexpectedly, levels of plasminogen activator inhibitor‐1 (PAI‐1) that is linked to glucose intolerance and inflammation 14, were also elevated (Fig EV2C). Taken together, these data suggest that loss of CLPP leads to an enhanced glucose homeostasis facilitated by increased peripheral glucose uptake, leading to a fasting‐like phenotype with altered hormone signature.

CLPP‐deficient mice are resistant to diet‐induced obesity and retain enhanced glucose metabolism upon high‐fat feeding

To examine how Clpp ^−/− mice are able to cope with metabolic stress, they were fed high‐fat diet (HFD) with 60% calories from fat from age 2 to 4 months. CLPP‐deficient mice gained significantly less weight in comparison with control mice irrespective of the gender, resulting in around 50% weight difference at the end of the treatment (Fig 3A). Consistent with the preserved lean phenotype, the enhanced insulin sensitivity and glucose tolerance were retained in Clpp ^−/− mice fed HFD (Fig 3B and C, respectively). In contrast, blood glucose levels of control mice fed HFD were only marginally reduced by insulin administration, pointing out to an insulin resistant state (Fig 3B). To link the enhanced glucose metabolism with insulin signaling, the level of insulin‐stimulated phosphorylation of AKT (serine/threonine protein kinase) was analyzed. Although no difference was observed in the insulin‐stimulated AKT phosphorylation between the groups on HFD, Clpp ^−/− mice had lower basal phosphorylation levels and retained responsiveness to insulin, unlike control mice where high‐fat feeding induced insulin resistance (Fig 3D and Appendix Fig S1C). Levels of GLUT1, used as a negative control, were not changed between the groups or upon insulin treatment (Appendix Fig S1C).

Figure 3. Loss of CLPP protects against HFD‐induced obesity.

1. Body weight of male (n = 12–16) and female (n = 3–12) control (Clpp ^+/+) and Clpp ^−/− mice fed HFD from 8 weeks of age (indicated with the arrow).
2. Insulin tolerance test performed with 15‐week‐old mice fed HFD (n = 5).
3. Glucose tolerance test performed with 16‐week‐old mice fed HFD (n = 5).
4. Ratio of phosphorylated (AKT^Ser473) and non‐phosphorylated AKT in liver lysates during basal and insulin‐stimulated conditions (n = 2–4).
5. Fed and fasted blood glucose levels of 15‐week‐old mice fed HFD (n = 6).
6. Insulin serum concentration of 16‐week‐old mice fed HFD (n = 10).

Data information: Bars represent mean ± SD (unpaired two‐tailed Student's t‐test; *P < 0.05, **P < 0.01, ***P < 0.001).

Fasting and non‐fasting circulating glucose levels were significantly lower in CLPP‐deficient mice during HFD feeding (Fig 3E), as previously shown for NCD. Whereas HFD exposure induced a strong increase in circulating insulin and leptin and a milder upregulation of GIP, resistin, and PAI‐1 levels in control mice, it did not have an effect on Clpp ^−/− mice, further corroborating their resistance to diet‐induced obesity (Figs 2C and 3F, and EV2C and D).

Liver‐ or muscle‐restricted loss of CLPP does not affect body weight and glucose homeostasis

To elucidate the contribution of major metabolic organs to the whole‐body metabolism of CLPP‐deficient mice, we generated two models where conditional Clpp mice (Clpp ^fl/fl) were bred with mice that express the Cre recombinase either specifically in liver or muscle. CLPP ablation in liver was mediated by Cre recombinase under control of the albumin promoter and enhancer and the α‐fetoprotein enhancer (Afp‐Cre) 15, resulting in Clpp ^fl/fl ; Afp‐Cre mice (referred to as Clpp ^LKO). To deplete CLPP in both cardiac and skeletal muscle cells, we used mice that express Cre recombinase under the muscle creatine kinase promoter (Ckmm‐Cre) 16, generating Clpp ^fl/fl; Ckmm‐Cre animals (referred to as Clpp ^MKO mice). A complete loss of CLPP was confirmed in targeted tissues of both models, with normal steady‐state levels in reciprocal organs (Fig EV3A). In contrast to whole‐body CLPP‐deficient mice, Clpp ^LKO and Clpp ^MKO were born at the expected Mendelian ratio, were fertile, and did not display any signs of growth retardation. No significant difference in body weight was detected for either Clpp ^LKO or Clpp ^MKO until 4 months of age (Fig EV3B). Consistent with the unaltered body weight, the insulin sensitivity and glucose tolerance were also not altered in either CLPP tissue‐specific mutant as compared to control mice (Fig 4A and B). Given the absence of a whole‐body phenotype of Clpp ^LKO or Clpp ^MKO mice, we examined molecular defects by evaluating mitochondrial respiratory deficiency on the level of OXPHOS complexes.

Figure EV3. Loss of CLPP in liver and muscle is negligible for the regulation of whole‐body metabolism.

1. Western blot analysis of CLPP levels in control (Clpp ^+/+), liver‐specific (Clpp ^LKO, left panel), or muscle‐specific (Clpp ^MKO, right panel) CLPP‐deficient (n = 3–4) mice.
2. Body weight of male Clpp ^LKO mice (left panel) and Clpp ^MKO mice (right panel) and control mice fed NCD (n = 6–7).
3. Relative transcript levels of Fgf21 in liver and SkM of Clpp ^MKO mice and control mice (n = 4).

Data information: Bars represent mean ± SD (unpaired two‐tailed Student's t‐test; ***P < 0.001).Source data are available online for this figure.

Figure 4. Tissue‐specific loss of CLPP does not recapitulate the metabolic phenotype of whole‐body CLPP ablation.

1. Insulin tolerance test performed with liver‐specific (Clpp ^LKO, left panel) and muscle‐specific (Clpp ^MKO, right panel) Clpp mice at 15 weeks of age fed NCD (n = 6).
2. Glucose tolerance test performed with Clpp ^LKO (left panel) and Clpp ^MKO (right panel) mice at 16 weeks of age fed NCD (n = 6).
3. Western blot analysis of OXPHOS complexes separated by BN‐PAGE. Antibodies against individual OXPHOS subunits (indicated in parenthesis) were used to detect OXPHOS complexes in liver (left panel) and SkM (right panel) (n = 3–6).

Data information: Data points represent mean ± SD (unpaired two‐tailed Student's t‐test).Source data are available online for this figure.

Depletion of hepatic CLPP resulted in a strong deficiency of respiratory supercomplexes, while levels of monomers or dimers of individual respiratory complexes were not altered (Fig 4C). Interestingly, loss of CLPP in skeletal muscle led to a milder decrease in the steady‐state level of supercomplexes (Fig 4C). A mitochondrial dysfunction of Clpp ^MKO mice was confirmed by increased expression of Fgf21, a marker for mitochondrial dysfunction that functions as a mitokine and elicits a global response (Fig EV3C) 17. Thus, moderate hepatic or skeletal muscle mitochondrial dysfunction caused by the isolated CLPP deficiency does not alter body weight and glucose homeostasis.

Loss of CLPP severely impairs cold tolerance, causing a whitening of BAT while promoting a browning of WAT

Increased energy expenditure, as observed in Clpp ^−/− mice, could be caused by higher mitochondrial uncoupling due to enhanced adaptive thermogenesis. However, CLPP‐deficient interscapular BAT was more lipid‐laden, having a lighter appearance in contrast to the control (Fig EV4A). Histological analysis revealed a number of cells with large unilocular lipid droplets, instead of the classic multilocular lipid droplets, in Clpp ^−/− BAT, and the phenotype was exaggerated upon HFD (Fig EV4A). Steady‐state levels of respiratory chain subunits were downregulated, although we did not observed a reduction in the respiration rate in Clpp ^−/− BAT mitochondria in steady‐state conditions when different substrates were used (OXPHOS, Fig 5A and Appendix Fig S2A). Remarkably, a lower basal rate, limited by the mitochondrial membrane proton leak (LEAK), and a strong reduction in maximal respiration rate (ETS) were detected, despite normal UCP1 levels (Figs 5A and EV4B, and Appendix Fig S3A). These results show that the lower levels of OXPHOS complexes do not affect steady‐state oxygen consumption rates (OCR) in Clpp ^−/− BAT mitochondria, but strongly influence the uncoupled and maximal respiration, likely compromising the thermogenic capacity of BAT.

Figure EV4. Morphological and molecular alterations of adipose tissues upon loss of CLPP during metabolic stress conditions.

1. Gross morphology and histological analysis of BAT of in control (Clpp ^+/+) and Clpp ^−/− mice fed NCD or HFD, H&E—hematoxylin and eosin staining (n = 4). Scale bar 50 μm.
2. Western blot quantification of mitochondrial respiratory chain subunits and UCP1 levels in BAT lysates (n = 3).
3. Western blot analysis of mitochondrial respiratory chain subunits in EWAT and IWAT (n = 3).
4. Western blot quantification of mitochondrial respiratory chain subunits in IWAT (n = 3).

Data information: Bars represent mean ± SD (unpaired two‐tailed Student's t‐test; *P < 0.05, **P < 0.01).Source data are available online for this figure.

Figure 5. Differential effect of CLPP depletion on brown and white adipose tissues in metabolic stress conditions.

1. Respiratory flux of isolated BAT mitochondria from control (Clpp ^+/+) and Clpp ^−/− mice fed NCD. The ADP‐stimulated respiration (OXPHOS) was measured in the presence of: C I substrates—pyruvate, malate, glutamate (PMG)—and CI + C II substrates—pyruvate, malate, glutamate + succinate (PMG+S). The proton leak (LEAK) was measured upon addition of oligomycin. The maximal respiration rate (ETS) was measured after FCCP titration (n = 4).
2. Body temperature of mice exposed to 5°C for 2 h (n = 3–4).
3. Western blot analysis and quantification of UCP1 levels in mitochondrial lysates from BAT of mice exposed to 5°C for 2 h (n = 3).
4. Gross morphology of EWAT and IWAT of NCD fed mice.
5. Hematoxylin and eosin staining of EWAT and IWAT tissue sections of mice fed NCD (left panel) or HFD (right panel). Scale bar 50 μm (n = 5).
6. Relative transcript levels of thermogenic genes in EWAT, IWAT, and BAT (n = 4).
7. Western blot quantification of mitochondrial respiratory chain subunits in EWAT lysates (n = 3).
8. Respiratory flux of EWAT tissue lysates from control (Clpp ^+/+) and Clpp ^−/− mice in the presence of fatty acid oxidation substrates: palmitoylcarnitine and malate (PC+M) OXPHOS—ADP‐stimulated respiration, LEAK—upon addition of oligomycin, and ETS—maximal respiration after FCCP titration (n = 4).

Data information: Bars represent mean ± SD (unpaired two‐tailed Student's t‐test; *P < 0.05, **P < 0.01, ***P < 0.001).Source data are available online for this figure.

To analyze whether the observed lipid accumulation in BAT combined with a respiratory chain deficiency has an impact on cold‐induced thermogenesis, control and Clpp ^−/− mice were exposed to 5°C. Although body temperature was not affected by CLPP deficiency when animals are kept at 22°C, cold exposure induced severe hypothermia in Clpp ^−/− mice with a rapid drop of body temperature from 36 to 27°C within 2 h, at what point the experiment needed to be terminated for the welfare of the animals (Fig 5B). This was accompanied by reduced levels of UCP1 but no further decline in the levels of OXPHOS supercomplexes (Fig 5C and Appendix Fig S3B, respectively). The observed phenotype is referred to as “whitening”, a decline of BAT function resulting in a more WAT‐like appearance 18. In agreement with the BAT “whitening” phenotype and a decreased uncoupling activity, cold exposure also resulted in increased CV levels in Clpp ^−/− mice (Appendix Fig S3B). We excluded skeletal muscle dysfunction, as a major contributor to the compromised cold‐induced thermogenesis of Clpp ^−/− mice, since sk. muscle function, measured by maximal speed and distance of the run on a treadmill, or the grip strength, was unchanged (Appendix Fig S3C).

Consistent with the overall fat decrease, EWAT (visceral) and IWAT (subcutaneous) depots were smaller in Clpp ^−/− mice and particularly EWAT displayed more brown‐like appearance (Fig 5D and E). A decrease in Clpp ^−/− adipocyte size was more obvious for EWAT than IWAT when mice are fed NCD and became even more apparent upon HFD feeding as CLPP‐deficient EWAT showed obvious resistance to diet‐induced fat accumulation (Fig 5E). The “browning” of EWAT and to a lesser extent IWAT of Clpp ^−/− mice was confirmed by showing the upregulation of thermogenic genes transcript levels, such as Ucp1, Dio2, and Cidea (Fig 5F). In contrast, Ucp1 and Dio2 levels were decreased in BAT, supporting the detected “whitening” in the absence of CLPP (Fig 5F). Remarkably, steady‐state levels of C II and C III subunits were strongly upregulated in EWAT of Clpp ^−/− mice suggesting an increase in FAO‐dependent respiration (Figs 5G and EV4C). Concurrently, levels of most C I subunits were not significantly changed in EWAT, but we observed a decrease in some C I and C IV subunits in IWAT of Clpp ^−/− mice (Figs 5G and EV4C and D). Curiously, the observed increases or decreases seem not to be a consequence of altered expression, as transcript levels of C I subunits in both EWAT and IWAT were normal, unlike BAT where a robust downregulation was detected (Appendix Fig S3D). In agreement, higher FAO activity was detected in CLPP‐deficient EWAT, as measured by the increase in oxygen consumption rates, when mitochondria were fed palmitoylcarnitine (PC or PC+G). Remarkably, the maximal capacity of the electron transfer system (ETS) in these conditions was doubled in Clpp ^−/− EWAT (Fig 5H and Appendix Fig S2B).

Effect of CLPP deficiency on fatty acid β‐oxidation

To further understand the molecular mechanism behind the metabolic phenotype of Clpp ^−/− mice, we analyzed FAO in tissues other than WAT. The β‐oxidation was measured using [1‐¹⁴C]‐palmitate and the incorporation into CO₂ and acid‐soluble metabolites in liver, skeletal muscle, and BAT mitochondria. Remarkably, the FAO rate was significantly reduced by about 25% in liver and skeletal muscle and was not changed in mitochondria isolated from BAT of CLPP‐deficient mice (Fig 6A). Consequently, the ADP‐stimulated respiration of long‐chain palmitoylcarnitine (PC or PC+M) was decreased even though mitochondrial uncoupling (LEAK) and the maximal capacity of the electron transfer system (ETS) were unchanged in CLPP‐deficient liver mitochondria (Fig EV5A). Remarkably, no difference in respiratory states was detected in Clpp ^−/− mitochondria when substrates directly feeding electrons to C I and C II were applied (Fig EV5B). Thus, CLPP deficiency in liver and likely skeletal muscle leads to a decreased rate of FAO, independent of the electron transfer capacity (Fig EV5A and B).

Figure 6. CLPP regulates energy metabolism by modulating fatty acid oxidation.

1. Ex vivo fatty acid oxidation of long‐chain [1‐¹⁴C]‐palmitate in liver, SkM, and BAT mitochondria from control (Clpp ^+/+) and Clpp ^−/− mice fed NCD (n = 6).
2. Western blot quantification of enzymes involved in FAO in liver, SkM, and BAT lysates in Clpp ^−/− mice relative to control mice (set to 1) (n = 4).
3. Western blot quantification of enzymes involved in FAO in EWAT and IWAT lysates in Clpp ^−/− mice relative to control mice (set to 1) (n = 4).
4. Relative Acadvl and Cpt2 transcript levels in liver (n = 4).
5. Cycloheximide (CHX) chase experiment following the VLCAD stability in the absence of CLPP. MEFs were harvested at 0, 2, 4, 6, 8, 10, and 12 h after starting the treatment with CHX and lysed for subsequent Western blot analysis (n = 3).
6. mtDNA levels determined by qPCR using TaqMan probes for Cytb and ATP6 in liver, SkM, BAT, EWAT, and IWAT (n = 4).

Data information: Bars represent mean ± SD (unpaired two‐tailed Student's t‐test; *P < 0.05, **P < 0.01, ***P < 0.001).Source data are available online for this figure.

Figure EV5. Loss of CLPP decreases fatty acid oxidation without altering respiratory chain capacity.

• A, B
  Respiratory flux of isolated liver mitochondria from control (Clpp ^+/+) and Clpp ^−/− mice in the presence of (A) fatty acid oxidation substrates: palmitoylcarnitine and malate (PC+M) or (B) C I substrates —pyruvate, malate, glutamate (PMG)—and C II substrates—succinate (S) OXPHOS—ADP‐stimulated respiration, LEAK—upon addition of oligomycin, and ETS—maximal respiration after FCCP titration (n = 4).
• C
  Western blot analysis of enzymes involved in FAO in liver, SkM, and BAT lysates (n = 4).
• D
  Western blot analysis of enzymes involved in FAO in EWAT and IWAT lysates (n = 4).

Data information: Bars represent mean ± SD (paired two‐tailed Student's t‐test; *P < 0.05).Source data are available online for this figure.

Recently, we reported the first comprehensive list of possible mammalian ClpXP substrates involved in the regulation of mitochondrial homeostasis, including different acyl‐CoA dehydrogenases (ACADs) that are involved in the first step of FAO inside mitochondria 6. The acyl‐CoA dehydrogenases differ in their specificity for very long‐ (VLCAD), long‐ (LCAD), medium‐ (MCAD), and short‐chain acyl‐CoAs (SCAD) 19. Although both MCAD and SCAD were detected as possible CLPP substrates in the previous screen 6, the overall striking upregulation of VLCAD suggests it to be a likely CLPP substrate (Fig 6B and C). The strongest upregulation of VLCAD levels was observed in EWAT (25‐fold) in addition to low LCAD and high MCAD levels (Fig 6C). Only a mild VLCAD upregulation was detected in Clpp ^−/− BAT accompanied by MCAD downregulation in agreement with previous results on diminished mitochondrial function and overall “whitening” of this tissue (Figs 6B and EV5C). No significant difference in Acadvl (gene encoding VLCAD) levels was detected, suggesting that increased protein stability, rather than expression, contributes to the observed increase in steady‐state levels (Fig 6D). Despite strong upregulation of VLCAD in CLPP‐deficient MEFs, we could not detect a clear difference in the turnover rate of these proteins upon cycloheximide chases, owing to their high stability even in wild‐type mitochondria (Fig 6E). Therefore, while these data suggest that VLCAD is a bonafide CLPP substrate, additional studies are still needed to prove the protease role in the VLCAD turnover.

Mitochondrial fatty acid β‐oxidation comprises different rate‐limiting steps, with the uptake of long‐chain acyl‐CoAs via the carnitine palmitoyltransferase system (CPT), being one of the most important ones. Analysis of the steady‐state levels of these enzymes revealed a strong downregulation of the carnitine palmitoyltransferase 2 (CPT2), whereas no difference was observed for carnitine palmitoyltransferase 1 (CPT1A or CPT1M) in CLPP‐deficient liver, skeletal muscle, and BAT, leaving EWAT and IWAT as the only tissues with unaltered CPT1 and CPT2 levels (Figs 6B and C, and EV5C and D). The observed changes do not correlate with the compensatory increase in mitochondrial mass (measured by the amount of mtDNA or steady‐state levels of mitochondrial house‐keeping genes) that was comparable in sk. muscle, EWAT, and IWAT, mild in liver, and not detected in BAT (Fig 6F and Appendix Fig S4).

Overall, these results suggest that downregulation of CPT2 might be the primary cause for decreased FAO in sk. muscle and liver. The decrease in the CPT2 levels is likely a compensatory mechanism against high FAO that might arise from an increase in VLCAD levels, as observed in CLPP‐deficient EWAT.

Discussion

Our results demonstrate that ubiquitous loss of CLPP protease results in a lean phenotype with normal energy expenditure, despite decreased physical activity. CLPP depletion leads to improved glucose metabolism, as evidenced by increased insulin‐stimulated glucose disposal and reduced blood glucose levels. This could be attributed to the observed increase of GLUT4 levels in different tissues of CLPP‐deficient mice. In skeletal muscle, the upregulation likely relates to the increased activation of the metabolic regulator AMPK that was shown to enhance glucose uptake by facilitating GLUT4 translocation to the plasma membrane 20. Enhanced glucose metabolism is commonly seen in mouse models with OXPHOS dysfunction, as a switch to glycolysis is an attempt to quickly provide an alternative source of ATP 21, 22, 23. In Clpp ^−/− mice, this modification of metabolism leads to a fasting‐like phenotype, evident by changes in blood glucose and metabolic hormone levels, and renders CLPP‐deficient mice utterly unaffected by diet‐induced obesity and insulin resistance 24, 25.

Paradoxically, a large number of studies, with the first one published over 40 years ago 26, have implicated mitochondrial dysfunction in the etiology of type 2 diabetes as a major cause of insulin resistance 27, 28, 29, 30. The opposing metabolic outcomes resulting from mitochondrial dysfunction that have been described might be dependent on the localization and/or the degree of mitochondrial defect. For example, the deletion of mitochondrial transcription factor A (TFAM), which causes a rapid depletion of mtDNA, in pancreatic β‐cells leads to impaired mitochondrial function, deficient insulin secretion, and glucose intolerance 31, whereas TFAM loss in skeletal muscle results in enhanced glucose metabolism 21. The importance of the degree of mitochondrial dysfunction becomes evident by comparing homozygous and heterozygous mtDNA mutator mice that accumulate somatic mtDNA mutations leading to respiratory chain dysfunction and a premature aging phenotype 32. Curiously, while homozygous mtDNA mutators display resting hypoglycemia and enhanced glucose tolerance, heterozygous mutants are glucose‐intolerant and exhibit an obesity phenotype 33, 34. Here, we show that a global deletion of CLPP leads to varying degrees of dysfunction in different tissues, yet systemic glucose metabolism is rather enhanced, further questioning the causal role for mitochondrial deficiency in insulin resistance.

The loss of CLPP in BAT resulted in moderate respiratory chain deficiency, higher accumulation of lipids, leading to a “whitening” of BAT, and the inability of mice to adequately respond to cold‐induced stress. A “whitening” of BAT was also observed in mouse models deficient for CPT2 35 and COX7RP 36, suggesting that both FAO and mitochondrial supercomplex assembly are required for BAT homeostasis. Remarkably, we showed that both of these pathways are affected in CLPP‐deficient BAT leaving the mice unable to maintain body temperature upon cold exposure. A thermogenic defect in BAT was also observed in mtDNA mutator mice, but unlike CLPP‐deficient mice, this phenotype was fully reversed upon HFD feeding 25, arguing that OXPHOS deficiency could not be the only contributing factor to the observed BAT insufficiency.

Although liver and muscle are both involved in the regulation of glucose and lipid metabolism, we showed that an isolated loss of CLPP in either tissue has no significant impact on systemic metabolism, despite moderate mitochondrial dysfunction. Changes in body weight as indicator for the alteration of systemic metabolism were reported for other models generated by Ckmm‐Cre‐mediated deletion of different essential mitochondrial genes 37, 38, 39, 40. The difference in phenotypes might however stem from much higher level of mitochondrial dysfunction in these mutants, which all primarily developed a strong cardiomyopathy, with lesser dysfunction in skeletal muscle 37, 38, 39, 40.

Studies of OXPHOS dysfunction specifically in liver induced by depletion of AIF or COX10, both of which primarily affect OXPHOS function, also show improved glucose metabolism, despite unchanged body weight 23, 41. A plausible explanation for the discrepancy in phenotypes between these models and Clpp ^−/− mice might again be the level of OXPHOS (dys)function, because we detected lower levels of supercomplexes, but no decrease in respiration. Therefore, the metabolic phenotype in CLPP‐deficient mice probably depends more on specific CLPP substrates or adaptive responses activated by deregulated homeostasis, than on OXPHOS dysfunction.

Our results further suggest that changes in adipocytes, especially EWAT, primarily contribute to the altered metabolic signature of CLPP‐deficient mice. In contrast to other CLPP‐deficient tissues, EWAT showed upregulation of respiratory chain subunits and evidence for the increased FAO rate consistent with the “browning” phenotype. This obvious disparity in the effect of CLPP deficiency on various adipose tissues suggests a complex interplay of different CLPP substrates in the development of the observed phenotypes that again cannot be attributed simply to the respiratory chain deficiency that usually leads to a much more uniform effect and correlates to the level of mitochondrial dysfunction 24, 42. For example, a moderate reduction of TFAM in IWAT and BAT, but not EWAT, resulted in higher energy expenditure and protection from age‐ and diet‐induced obesity, insulin resistance, and hepatosteatosis, despite greater food intake 24, 42. A broader depletion of TFAM in all adipocyte tissues led to increased cell death and inflammation in WAT and a whitening of BAT resulting in a lipodystrophic syndrome characterized by insulin resistance on both NCD and HFD 42.

Our study also establishes VLCAD, an enzyme that catalyzes the initial rate‐limiting step in mitochondrial fatty acid beta‐oxidation, as a novel CLPP substrate. Indeed, VLCAD was ubiquitously upregulated in all tissues and cell types of CLPP‐deficient mice despite normal transcript levels. Although VLCAD upregulation might lead to an increased FAO, in the majority of Clpp ^−/− tissues, this was counterbalanced by a strong decrease in CPT2 resulting in a lower FAO rate in skeletal muscle and liver. Curiously, a reverse adaptation, with upregulated CPT2 levels, was reported in VLCAD‐deficient mice 43.

Downregulation of CPT2 in CLPP‐deficient mice is likely an adaptive response aimed to prevent high FAO upregulation that could have detrimental impact. Indeed, muscle‐specific overexpression of PGC‐1α (peroxisome proliferator‐activated receptor‐γ coactivator‐1α) resulted in increased mitochondrial density and higher ATP synthesis rate, likely due to upregulated levels of many FAO enzymes, including acetyl coenzyme A carboxylase 2 (ACC2), CPT1, CPT2, VLCAD, LCAD, and MCAD 44. Paradoxically, these mice were also more prone to fat‐induced muscle insulin resistance, associated with reduced insulin‐stimulated glucose uptake in skeletal muscle 44.

Taken together, our results suggest that in the absence of CLPP, VLCAD accumulates in mitochondria, leading to a compensatory downregulation of CPT2, thereby reducing FAO rate in liver, skeletal muscle, and possibly even BAT. In parallel, robust carbohydrate usage mediated by increased GLUT4 levels in skeletal muscle lowers circulating glucose concentration and provides Clpp ^−/− mice with high insulin sensitivity. In WAT, especially EWAT, CLPP deficiency also causes upregulation of its putative substrate VLCAD that is not compensated for by decreased CPT2 levels and therefore leads to increased FAO and likely strongly contributes to the improved energy metabolism of Clpp ^−/− mice. Although loss of CLPP has an overall positive effect on systemic metabolism and renders animals unaffected by insulin resistance and diet‐induced obesity, it also strongly diminishes adaptive thermogenesis through BAT insufficiency.

Finally, these results depict important changes in molecular, cellular, and systemic metabolism caused by CLPP deficiency that allow us to better understand the underlying mechanisms and changes that contribute to the development of Perrault syndrome in patients. They also send a strong message by highlighting the opposing tissue‐specific outcomes of the mitochondrial defect that have to be taken into account when attempting to treat the metabolic dysfunction in patients.

Materials and Methods

Generation and genotyping of transgenic mice

The generation of Clpp ^fl/fl mice has been described elsewhere 6, 7. To generate full body Clpp knockout mice, Clpp ^fl/fl were mated with transgenic mice ubiquitously expressing Cre recombinase under control of the β‐actin promoter, resulting in Clpp ^+/− mice. Heterozygous Clpp ^+/− mice were further intercrossed to obtain homozygous Clpp ^−/− mice. Liver‐specific knockout animals were generated by mating Clpp^fl/fl animals with transgenic mice expressing Cre recombinase under control of the albumin enhancer and promoter and the α‐fetoprotein enhancer (Afp‐Cre). Heart and skeletal muscle‐specific knockout animals were generated by mating Clpp ^fl/fl animals with transgenic mice expressing Cre recombinase under control of the muscle creatine kinase promoter (Ckmm‐Cre).

The genotyping primers used to determine Clpp alleles were Clpp^forw (5′‐GTGGATGATGGTCAGTAGAATCC‐3′) and Clpp^rev (5′‐CCCAGACATGATTCCTAGCAC‐3′).

For isolation of mouse embryonic fibroblasts (MEFs), Clpp ^+/− heterozygous mice were intercrossed to obtain the Clpp ^−/− homozygous embryos.

Mice had ad libitum access to either normal chow diet (R/M‐H‐V1554, ssniff Spezialdiäten GmbH) containing 55.1% carbohydrates, 19.3% proteins, and 3.3% fat (9% calories from fat) or high‐fat diet (ssniff EF acc. D12492 (I) mod., ssniff Spezialdiäten GmbH) containing 27.1% carbohydrates, 24.1% protein, and 34.0% fat (60% calories from fat) and drinking water. Mice were fed normal chow diet (NCD), unless otherwise indicated in the text. The high‐fat diet was fed starting from 8 weeks of age until the termination of the experiment with 16 weeks. Mice were killed by cervical dislocation at the end of the study. Animal protocols were in accordance with guidelines for humane treatment with animals and were reviewed and approved by the Animals Ethics Committee of North Rhine‐Westphalia, Germany (AZ84‐02.04.2012.A407).

Body composition

The body fat content and lean mass were measured using the nuclear magnetic resonance (NMR) analyzer minispec mq7.5 (Bruker Optik).

Indirect calorimetry

Energy expenditure and respiratory exchange ratio (RER) were determined with indirect calorimetry using the PhenoMaster (TSE systems). Animals were acclimated in metabolic chambers (7.1 l) for 3 days prior to data acquisition to adapt them to single housing, food, and water dispensers. During this time, animals were monitored daily and the body weight was determined to ensure proper adaption. Calorimetric measurements were conducted at 22°C for 72 h assessing oxygen consumption, carbon dioxide production, and locomotor activity (infrared light beam frame, TSE systems). In addition, food intake and water intake were assessed with automated measuring devices (TSE systems).

The energy expenditure ANCOVA was provided by the NIDDK Mouse Metabolic Phenotyping Centers (MMPC, http://www.mmpc.org) using their Energy Expenditure Analysis page (http://www.mmpc.org/shared/regression.aspx) and supported by grants DK076169 and DK115255.

Treadmill

Exercise experiments were performed on a treadmill (TSE Systems GmbH). The following protocol was used. On day 1, the mice were adapted and familiarized to the sound and movement of the treadmill in two training sessions. On day 2, the mice were again trained in the morning and in the afternoon the experimental run was performed. Mice had to run for 25 min. The velocity was started at 0.05 m/s and reached a final pace of 0.23 m/s. The experiment was terminated (exhaustion) when the mice were falling off the treadmill more than 10 times per min.

Blood and serum analyses

Blood glucose levels were determined from whole venous tail blood using an automatic glucose monitor (Contour Next, Bayer) from either random fed or 6‐h fasted animals.

Serum was obtained by collecting blood from the submandibular vein. Blood samples were then incubated at room temperature (RT) for 45 min and centrifuged at 1,500 g for 15 min. Resulting serum was stored at −80°C for subsequent analysis.

Serum hormone levels were determined in 16‐week‐old mice fed NCD or HFD using the Bio‐Plex Pro Mouse Diabetes 8‐Plex Assay (Bio‐Rad) with the Bio‐Plex™ 200 system (Bio‐Rad).

Glucose tolerance test, insulin tolerance test, and insulin signaling

Glucose tolerance tests were carried out in 15‐week‐old animals following a 6‐h fast. After measuring fasted blood glucose levels, mice were injected intraperitoneally (i.p.) with 2 mg/g body weight glucose (20% glucose solution, Sigma‐Aldrich). Blood glucose levels were determined at 15, 30, 60 and 90 min postglucose injection. Insulin tolerance tests were performed with ad libitum fed 16‐week‐old animals. After recording baseline glucose levels, each animal was administered i.p. with 0.6 U/kg body weight insulin (Insuman Rapid, Sanofi‐Aventis) and blood glucose levels were monitored at 15, 30, 60 and 90 min after injection.

For insulin signaling analysis, mice were injected with 0.6 U/kg body weight insulin (Insuman Rapid, Sanofi‐Aventis) or saline following a 2‐h fast. Mice were then sacrificed 30 min postinjection, and tissues were collected and frozen in liquid nitrogen.

Acute cold exposure and measurement of rectal body temperature

For acute cold exposure, Clpp ^−/− and control animals were individually housed in home cages with bedding material, however, without nesting material and exposed to 5°C. Rectal body temperature was measured using a rectal thermometer (Bioseb) prior to the experiment at ambient temperature. Afterward, body temperature was measured every hour for 7 h or until core body temperature dropped below 26°C.

Femur length

The mouse hind leg was carefully removed, and then, muscle and ligaments were trimmed away to dissect the femur. The femur was cleaned from connective tissue with paper cloth. In the end, the length of the femur was measured between femur head and the lateral condyle with a sliding caliper.

Micro‐computed tomography

The measurement of bone mineral density was performed with the LaTheta Micro‐CT Scanner (Aloka) in anaesthetized mice using 2.5% Isoflurane in oxygen. The sections were set to 0.6 μm, and analysis was done with the LaTheta software.

Primary hepatocyte isolation and measurement of gluconeogenesis

Mice were anesthetized, and livers were perfused with Earle's balanced salt solution (EBSS) containing 0.5 mM EGTA, followed by perfusion with 0.03% collagenase (Sigma) and 0.004% Trypsin inhibitor in EBSS containing 10 mM HEPES. The liver was then detached and filtered through a 100‐μm nylon mesh, and cells were sedimented by Percoll centrifugation. Cells were plated onto six‐well plates (600,000 cells/well) and grown in low‐glucose Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal calf serum, 1% L‐Glut, 100 μg/ml penicillin, and 100 μg/ml streptomycin, when kept at 37°C in humidified atmosphere (95% air and 5% CO₂).

On the next day, gluconeogenic stimuli were induced by 0.1 μM glucagon (Sigma) and 1 μM Bt2cAMP (Sigma) added to the culture medium for 6 h. Culture medium was replaced with 1 ml of glucose‐free DMEM (pH 7.4), without phenol red, supplemented with 20 mM sodium lactate and 2 mM sodium pyruvate. After 1 h, the medium was collected and the glucose concentration was measured by means of the Glucose (GO) Assay Kit (Sigma).

Real‐time quantitative PCR

Isolated RNA was treated with DNase (DNA‐free Kit, Ambion) and subsequently reverse‐transcribed with the High‐Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Mouse expression levels were determined using Brilliant III Ultra‐Fast SYBR Green QPCR Master Mix (Agilent Technologies) with the following primer pairs:

Gene	Forward primer 5′–3′	Reverse primer 5′–3′
Acadvl	CTACTGTGCTTCAGGGACAAC	CAAAGGACTTCGATTCTGCCC
Cidea	TGACATTCATGGGATTGCAGAC	GGCCAGTTGTGATGACTAAGAC
Cpt2	CAGCACAGCATCGTACCCA	TCCCAATGCCGTTCTCAAAAT
Dio2	CAGCTTCCTCCTAGATGCCTA	CTGATTCAGGATTGGAGACGTG
Hprt	TCAGTCAACGGGGGACATAAA	GGGGCTGTACTGCTTAACCAG
Igf1	AAATCAGCAGCCTTCCAACTC	GCACTTCCTCTACTTGTGTTCTT
Fgf21	GTGTCAAAGCCTCTAGGTTTCTT	GGTACACATTGTAACCGTCCTC
Ndufa9	GGCCAGCTTACCTTTCTGGAA	GCCCAATAAGATTGATGACCACG
Ndufs2	TCGTGCTGGAACTGAGTGGA	GGCCTGTTCATTACACATCATGG
Ndufv2	GGCTACCTATCTCCGCTATGA	TCCCAACTGGCTTTCGATTATAC
Sdha	GAACACTCCAAAAACAGACCTGC	TCCACCACTGGGTATTGAGTAG
Tbp	GGGAGAATCATGGACCAGAA	TTGCTGCTGCTGTCTTTGTT
Ucp1	AGCCATCTGCATGGGATCAAA	GGGTCGTCCCTTTCCAAAGTG

Samples were adjusted for total RNA content by Hprt or Tbp. The relative expression of mRNAs was determined with a comparative method (2^−ΔΔCT).

mtDNA quantification

DNA was isolated using the Qiagen Blood and Tissue Kit. Probes for target genes were from TaqMan Assay‐on‐Demand kits (Applied Biosystems). Samples were adjusted for total DNA content by 18S. The relative level of mtDNA was determined with a comparative method (2^−ΔΔCT). The TaqMan probes were used and their accession numbers were Cytb/ATP6 (Mm03649417_g1) and 18S (Hs999999_s1).

Western blot analysis

Protein lysates were obtained from either homogenized tissue or isolated mitochondria and subsequently subjected to Western blot analysis as described previously 45. All antibodies were used in dilutions indicated or otherwise recommended by the manufacturers: ACADS (1:2,000, sc‐365953, Santa Cruz), ACADL (1:2,000, ab196655, Abcam), ACADVL (1:2,000, sc‐98338, Santa Cruz), ACTIN (1:5,000, A5441, Sigma‐Aldrich), AFG3L2 (1:1,000, Polyclonal antisera made by Prof. Dr. Elena I. Rugarli), AKT (1:2,000, 9272, Cell Signaling), AMPK (1:1,000, 2532, Cell Signaling), ATP5A (1:1,000, MS507, Mitosciences), CLPP (1:1,000, WH0008192‐M1, Sigma‐Aldrich), CLPX (1:1,000, HPA040262, Sigma‐Aldrich), COX1 (1:2,000, 459600, Invitrogen), CPT1 (1:1,000, sc‐31128, Santa Cruz), CPT1M (1:1,000, CPT1M11‐A, Alpha Diagnostic), CPT2 (1:1,000, ab181114, Abcam), GLUT1 (1:300, sc‐7903, Santa Cruz), GLUT4 (1:1,000, 07‐1404, Millipore), HSC70 (1:5,000, sc‐7298, Santa Cruz), MCAD (1:2,000, sc‐49047, Santa Cruz), NDUFA9 (1:5,000, 459100, Invitrogen), NDUFB6 (1:5,000, A21359, Invitrogen), NDUFS2 (1:2,000, ab96160, Abcam), NDUFS3 (1:1,000, MS112, Mitosciences), NDUFV1 (1:1,000, 11238‐1‐AP, Proteintech), NDUFV2 (1:2,000, 15301‐1‐AP, Proteintech), pAKT (Ser473) (1:1,000, 4060, Cell Signaling), pAMPK (Thr172) (1:1,000, 2535, Cell Signaling), SDHA (1:10,000, 459200, Invitrogen), TFAM (1:2,000, Polyclonal antisera made by Prof. Dr. Nils‐Göran Larsson), TOMM20 (1:1,000, sc‐11415, Santa Cruz Biotechnology), UCP1 (1:1,000, Polyclonal antisera made by Prof. Dr. Nils‐Göran Larsson), UQCRC1 (1:5,000, 459140, Invitrogen), UQCRFS1 (1:1,000, MS305, Mitosciences), VDAC (1:1,000, 4661, Cell Signaling).

BN‐PAGE analysis

BN‐PAGE was carried out using the NativePAGE Novex Bis‐Tris Mini Gel System (Invitrogen) according to the manufacturer's specifications. Proteins were transferred onto a PVDF membrane, and immunodetection of mitochondrial OXPHOS complexes was performed.

Mitochondrial isolation & oxygen consumption measurements

Mitochondria were isolated from liver or BAT 46, or crude homogenate of EWAT was prepared, and oxygen consumption rates were measured with Oxygraph‐2k (Oroboros Instruments). Two different substrate uncoupler inhibitor titration (SUIT) protocols were applied. After air equilibration of the mitochondrial respiration buffer (120 mM sucrose, 50 mM KCl, 20 mM Tris–HCl, 1 mM EGTA, 4 mM potassium dihydrogen phosphate, 2 mM magnesium chloride, 0.1% BSA), 150 μg liver mitochondria was added. For the carbohydrate SUIT protocol, first the OXPHOS state for complex I was determined by adding 5 mM pyruvate, 2 mM malate, 20 mM glutamate, and 2 mM ADP. The convergent electron flow through complexes I and II was measured by adding 10 mM succinate. Next, LEAK respiration was assessed by the inhibition of the ATP synthase with oligomycin (1.5 μg/ml). Titration of carbonyl cyanide‐p‐trifluoromethoxyphenylhydrazone (FCCP, 0.5 μM) was performed to determine the maximal electron transfer system (ETS) capacity. Subsequently, complex I was inhibited by 0.5 μl Rotenone to determine the maximal ETS capacity with electron flow through complex II only. Finally, residual oxygen consumption (ROX) was measured upon inhibition of complex III by 2.5 μM Antimycin A. For the FAO SUIT protocol, first the endogenous FAO was determined by adding 2 mM malate and 2 mM ADP. Next, the OXPHOS state for electron transferring flavoprotein and complex I was determined by adding 50 μM palmitoylcarnitine and 10 mM glutamate. As in the carbohydrate SUIT protocol, the state of LEAK and ETS was determined.

Palmitate oxidation rate ex vivo

Fatty acid β‐oxidation in isolated mitochondria was performed as described previously 47. Briefly, crude mitochondria were isolated from 200 μg liver, 100 μg SkM, and 80 μg BAT in STE buffer (0.25 M sucrose, 10 mM Tris–HCl pH 7.4, 1 mM EDTA). Isolated mitochondria were then incubated in oxidation reaction mixture containing 0.4 μCi BSA‐conjugated [1‐¹⁴C]‐palmitate for 1 h at 37°C. After the incubation, the reaction mixture was transferred to a tube with 200 μl of 1 M perchloric acid and piece of Whatman paper disk in the cap soaked with 20 μl 1 M NaOH. After 1‐h incubation at RT, the paper disk was transferred to a scintillation vial for the determination of ¹⁴CO₂. The acid solution was centrifuged, and the supernatant afterward transferred to a scintillation vial for the measurement of the acid‐soluble metabolites using a liquid scintillation counter.

Histological analysis of tissue sections

BAT, EWAT, and IWAT were embedded in paraffin and sectioned at 5 μm. Deparaffinized and rehydrated tissue sections were stained with Mayer's hematoxylin and counterstained with eosin.

Densitometry analysis

For quantification of scanned films, densitometry analysis using public domain software ImageJ (a Java image processing program inspired by National Institutes of Health (NIH) Image for Macintosh) was performed.

Determination of protein turnover in mouse embryonic fibroblasts (MEFs)

Primary MEFs were isolated from E13.5 embryos obtained as a result of intercrossed Clpp ^+/− animals as described previously 48. Immortalized cell lines were generated upon transformation with the SV40 T antigen 49. Cells were grown to 80–90% confluency in a DMEM–high‐glucose‐containing media (Invitrogen). Cytoplasmic protein synthesis was blocked by addition of cycloheximide (100 μg/ml), and cells were collected at indicated time points, counted, and washed with PBS. The cell lysates were obtained in the RIPA buffer, proportional to the number of living cells, and subjected to a Western blot analysis.

Statistical analysis

All numerical data are expressed as mean ± SD. Student's t‐test and Kaplan–Meier distribution were used for statistical analysis. In order to detect outlier in our data sets, we have used Grubbs’ test (http://graphpad.com/quickcalcs/Grubbs1.cfm). Differences were considered statistically significant for P < 0.05. Most experiments were performed using at least four biological replicates.

Author contributions

AT conceived the project and together with CB designed the experiments, analyzed the data, and wrote the manuscript. CB, AK, KSz, SH, KSe, CPB, and PM performed the experiments, interpreted, and analyzed data.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Source Data for Expanded View

Review Process File

Source Data for Figure 4

Source Data for Figure 5

Source Data for Figure 6

EMBO Reports (2018) 19: e45126