MuRF1 deficiency prevents age-related fat weight gain, possibly through accumulation of PDK4 in skeletal muscle mitochondria in older mice

2.1 Animals

MuRF1^–/– mice, generated by inserting a neomycin-resistance gene into exon 2 of MuRF1, were kindly provided by Dr. Sorimachi.^{12, 25} MuRF1^–/– mice were backcrossed with C57BL/6 mice more than eight times. MuRF1^+/+ and MuRF1^–/– male mice were housed in a room maintained at 23 ± 2°C on a 12:12-h light/dark cycle and allowed free access to standard chow and water. A small amount of blood for measuring glucose and insulin was collected after overnight fast 1 week before euthanasia. Body weight (BW) was measured monthly for 24 months. Fat mass of mice was quantified by computed tomography (LATheta LCT-200; Aloka Inc.).

Respiratory quotient (RQ) and energy expenditure (EE) were recorded every 10 min for 48 h using an Ox Imax™ sensor (Covidien-Nellcor). Mean values were calculated hourly, and these values were averaged at the same Zeitgeber time (ZT). Locomotor activity was measured using an ACTIMO system (Shintechno). The number of movements of mice recognized by infrared beams was counted for 48 h. Mean values were calculated at the same ZT.

All protocols were implemented according to the Guide for the Care and Use of Laboratory Animals at Tokushima University. Experimental protocols described in this study were approved by the Tokushima University Ethics Review Committee for Animal Experimentation.

2.2 Isolation of mitochondria

Mitochondria from COS7 and skeletal muscles from MuRF1^+/+ and MuRF1^–/– mice were prepared following previously established protocols.²⁶ Briefly, cells or tissues were harvested and immediately minced in ice-cold CP-1 buffer (100 mM KCl, 50 mM Tris-HCl, 2 mM EGTA, pH 7.4). After grinding in a glass homogenizer, supernatants were passed through a 40-μm cell strainer. Mitochondria were then separated by differential centrifugation and lysed with RIPA™ buffer (50 mM Tris-HCl, pH 8.0, containing 1% NP-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfate (SDS), 150 mM NaCl, 5 mM ethylenediaminetetraacetic acid [EDTA], and protease inhibitors) (Nacalai Tesque Inc.).

2.3 Plasmid constructs

Mouse MuRF1, PDK4, PDHα, PDHβ, and SUMO3 complementary DNAs (cDNAs) were isolated from a mouse skeletal muscle cDNA library by PCR. PCR products were subcloned into mammalian expression plasmid vectors pcDNA3.1-V5, pcDNA6.1-myc, and pcDNA-FLAG (MuRF1-V5, PDK4-myc, PDHα-myc, PDHβ-myc, and SUMO3-FLAG). RING domains of truncation mutants of MuRF1 (ΔRING-MuRF1-V5) and mitochondrial targeting signal (MTS) truncation mutants of PDK4 (ΔMTS-PDK4-myc) were constructed using a modified PCR technique (Stratagene Cloning Systems), as previously described.²⁷ Construction of the expression plasmid for green fluorescence protein (GFP) containing the MTS of PDK4 (MTS-GFP) used an MTS mutant of PDK4 (residues 1–10, pcDNA6.1-MTS-myc) constructed as described above, and GFP was subcloned from the pCDNA3.1-GFP vector and ligated in-frame into the pcDNA6.1-MTS-myc vector.

2.4 Cell culture and transfection

COS7 cells were maintained in Dulbecco's modified Eagle medium containing 10% fetal bovine serum and penicillin-streptomycin (Nacalai Tesque Inc.) at 37°C in the presence of 5% CO₂. Cells were transfected with plasmid vectors containing indicated genes using a FuGENE HD™ lipofection reagent (Roche Diagnostics) for 24 h before experiments.

2.5 SDS-polyacrylamide gel electrophoresis (PAGE) and Immunoblotting

Cells were homogenized with a sonicator in 50 mM Tris-HCl, pH 7.5, containing 150 nM NaCl, 1% Triton X-100, and protease inhibitors with EDTA. Protein samples were combined with 4× sample buffer (250 mM Tris-HCl, 8% SDS, 40% glycerol, 8% β-mercaptoethanol, 0.02% bromophenol blue) and separated on a polyacrylamide gel. Proteins were transferred to a polyvinylidene difluoride membrane and probed with the following primary antibodies, following the manufacturer's instructions: anti-V5 (R96025; Invitrogen), anti-myc (#2276; Upstate), anti-FLAG M2 (F3165; Sigma-Aldrich), anti-PDK4 (ab38242; Abcam), anti-PDH (ab110416; Abcam), anti-MFN2 (ab50838; Abcam), antimitochondrial cytochrome c oxidase subunit IV (COX IV) (#4844s; Upstate), anti-β-actin (A1978; Calbiochem), and anti-glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (sc-25778; Santa Cruz Biotechnology). Secondary antibodies were donkey anti-rabbit 1:5000 and sheep anti-mouse 1:5000 (NA934-1ML, Amersham Biosciences and NA-9310, Piscataway). Membranes were developed using Amersham™ ECL™ or Amersham™ ECLTM Prime Western blotting detection reagents (GE Healthcare).

2.6 Immunoprecipitation

Immunoprecipitation samples were prepared using 100 μg protein and adjusted to 300 μl with lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, plus protease inhibitors with or without EDTA). Samples were then incubated with 0.5 μl of primary antibody at 4°C for 3 h under rotation. Thirty microliters of Protein G Sepharose™ 4 Fast Flow (GE Healthcare) was added to the samples, which were then rotated at 4°C for 16 h. The samples were washed in wash buffer six times and then prepared for SDS-PAGE as described above.

2.7 Quantitative reverse-transcription RT-qPCR

Total RNA was extracted from mouse skeletal muscle with an acid guanidinium thiocyanate-phenol-chloroform mixture (ISOGEN; Nippongene). RT-qPCR was performed using gene-specific primers and SYBR™ Green dye in an ABI StepOne™ PCR system (Applied Biosystems), as previously described.²⁸ Oligonucleotide primers used for amplification are listed in Table 1.

2.8 Measurement of the number of nuclei and fiber size of skeletal muscle

Sections of TA were stained with hematoxylin-eosin. Sections were photographed using a BIOREVO BZ-9000 fluorescence microscope (Keyence). The number of nuclei per muscle fiber and fiber size-frequency were measured in four sections of each muscle.

2.9 Measurement of biochemical parameters

Protein concentrations were determined using a bicinchoninic acid assay kit (Pierce).²⁹ Serum concentrations of triglycerides and nonesterified fatty acid (NEFA) were measured using appropriate kits (Fuji Film Wako Pure Chemical Co.), following the manufacturer's protocol. Serum glucose concentrations were measured using a glucose meter for laboratory animals (Research and Innovation Japan Inc.). Serum insulin concentration was measured using an enzyme-linked immunosorbent assay kit (Funakoshi Co., Ltd.). Serum concentrations of pyruvate and lactate were determined using commercial assay kits (Funakoshi Co., Ltd.).

2.10 Cycloheximide (CHX) treatment

CHX is an inhibitor of protein synthesis, particularly of translation. CHX is widely used for evaluation of activity in protein degradation independent of protein synthesis. COS7 cells, transfected with the plasmid vectors containing the indicated genes, were treated with CHX for indicated periods.

2.11 Epoxomicin treatment

Epoxomicin is a strong and selective inhibitor of proteasomes and does not affect other protein degradation systems. COS7 cells, transfected with the plasmid vectors containing the indicated genes, were treated with epoxomicin for 2–4 h.

2.12 Statistical analysis

Data are expressed as mean ± SD (n = 3–7). Differences between two groups were assessed using Scheffé's test. All statistical analyses were conducted using SPSS 6.1 software (SPSS Japan). Differences were considered significant at p < .05.

3.1 Depletion of MuRF1 prevents body weight gain

Gene expression of MuRF1 was measured in 3-, 12-, and 24-month-old mice in wild-type (MuRF1^+/+) by RT-qPCR and increased with age (Figure 1A). Our RT-qPCR did not detect MuRF1 expression in any skeletal muscles of MuRF1-deficient (MuRF1^–/–) mice (data not shown), indicating complete knockout of MuRF1 expression. We measured BW of MuRF1^+/+ and MuRF1^–/– mice monthly for 2 years. Interestingly, BW of MuRF1^–/– mice more than 8 months old was significantly less than age-matched MuRF1^+/+ mice (Figure 1B). However, no significant difference in food intake between MuRF1^+/+ and MuRF1^–/– mice was noted during the study period (Figure 1C). We assessed body composition using computed tomography scans to explore this finding in MuRF1^–/– mice. The percentage of fat mass was significantly lower in MuRF1^–/– mice than in MuRF1^+/+ mice (Figure 1D,E). The wet weight of epididymal fat in 12- and 24-month-old MuRF1^–/– mice was also reduced compared with MuRF1^+/+ mice (Figure 1F), indicating that aged (>7 months old) MuRF1^–/– mice were leaner.

3.2 MuRF1 deficiency partially protects from age-related muscle atrophy

We elucidated MuRF1 deficiency on age-dependent muscle atrophy, consistent with the association between muscle atrophy and ubiquitin ligase.^{8, 12, 13} We measured wet muscle weight of four hindlimb muscles (TA, extensor digitorum longus [EDL], gastrocnemius [GA], and soleus [SO]) at 3 and 24 months of age. Age-induced decreases in MW compensated by BW were significantly inhibited in TA and SO in MuRF1^–/– mice (Figure 2A).

We next measured the number of nuclei per muscle fiber and cross-sectional area (CSA) of TA, which is sensitive to aging^{3, 4, 30} in mice at 3 and 24 months of age. The number of nuclei per muscle fiber significantly decreased in TA of MuRF1^+/+ mice during aging. On the other hand, no significant change was observed in TA from MuRF1 deficient mice during aging (Figure 2B). In addition, the fiber size-frequency distribution of TA muscle fibers was measured. The fiber size-frequency distribution shifted to larger size in MuRF1^–/– mice than in age-matched MuRF1^+/+ mice (Figure 2C).

3.3 MuRF1^–/– mice preferentially metabolize lipid rather than glucose

We measured serum and muscle (GA) parameters of lipid and glucose metabolism in 3 months old MuRF1^+/+ and MuRF1^–/– mice (Table 2) to determine why MuRF1 deficiency prevented BW and fat mass gain in mice. Already, at 3 months of age, the serum glucose levels in fasted MuRF1^–/– mice were significantly lower than those in MuRF1^+/+ mice, although the fasting serum insulin level of MuRF1^+/+ mice was similar to that of MuRF1^–/– mice. In contrast, serum NEFA levels in MuRF1^–/– mice were significantly lower than that in MuRF1^+/+ mice. Further, serum triglyceride levels tended to be lower in MuRF1^–/– mice, although the difference with MuRF1^+/+ mice was not significant. Interestingly, serum and muscle lactate levels were higher in MuRF1^–/– mice than in MuRF1^+/+ mice, but serum pyruvate levels were lower in MuRF1^–/– mice. Thus, lipolysis and anaerobic glycolysis were promoted in the skeletal muscle of MuRF1^–/– mice.

We also analyzed whole-body lipid utilization and locomotor activity in young adult mice using indirect calorimetry and an animal movement analysis system, respectively. We use ZT, which shows time of lights on as ZT 0 and time of lights off as ZT 12. The RQ of 3-month-old MuRF1^–/– mice was significantly lower than that of MuRF1^+/+ mice (Figure 3A). In addition, EE of 3-month-old MuRF1^–/– mice was significantly higher than that of MuRF1^+/+ mice during ZT 0 to ZT 12 (light periods) in spite of similar locomotor activity. Even during ZT 12 to ZT 0 (dark periods), EE of MuRF1^−/− animals was about the same as for MuRF1^+/+ mice in spite of less locomotor activity (Figure 3A). These results indicate that MuRF1 deficiency enhanced EE, especially via lipid oxidation. Next, we measured gene expression of genes involved in lipogenesis and lipolysis and associated transcription factors in GAs of MuRF1^+/+ and MuRF1^–/– mice. MuRF1 deficiency showed little effect on expression of lipogenesis- and lipolysis-associated genes (Figure 3B).

3.4 PDK4 is accumulated in the mitochondria of MuRF1^–/– skeletal muscle

A previous study showed that MuRF1 interacts with glucose and fatty acid metabolism-related molecules, such as PDHs and their negative regulators, PDKs, through yeast two-hybrid screens.¹⁸ We performed coimmunoprecipitation assays in MuRF1-V5 and empty (Mock)-, PDHα-, PDHβ-, or PDK4-myc cotransfected COS7 cells to confirm MuRF1 binding to these molecules. The assays revealed that PDK4 specifically interacts with MuRF1 (Figure 4A). We focused on PDK4 as a crucial enzyme for switching from glucose to fatty acid energy sources.^{19, 20}

No significant difference in the gene expression of PDK4 was observed between the GA of MuRF1^+/+ and MuRF1^–/– mice (Figure 4B). Thus, we examined mitochondrial PDK4 protein levels in the skeletal muscle of MuRF1^–/– mice; PDK4 translocated into the mitochondria after translation. Surprisingly, PDK4 significantly accumulated in the mitochondrial fraction of GA of 3-month-old MuRF1^–/– mice (Figure 4C). In contrast, MuRF1 deficiency did not affect protein levels of PDHs, including PDH E1α, E1β, E2, and E3, in the mitochondrial fraction of GA (Figure 4C). Further, the increase of PDK4 protein levels of the muscle mitochondrial fraction in MuRF1^–/– mice was sustained for 24 months. To confirm our hypothesis, we examined total PDK4 level by Western blotting. We used mitofusin 2 (MFN2), which is one of the mitochondrial fusion proteins, as internal standard, because its protein levels in human vastus lateralis muscle were not changed in a wide age range (21–88 years).³¹ We clearly demonstrated that the value of mitochondrial PDK4/total PDK4 in 12 and 24-month-old MuRF1^–/– mice were higher than those in the respective month-old MuRF1^+/+ mice. These results indicate that the accumulation of PDK4 was enhanced during aging (Figure 4D). These results suggest that MuRF1 affects translocation of PDK4 into skeletal muscle mitochondria in mice.

3.5 PDK4 interacts with MuRF1 possibly in cytosol

PDK4 is targeted to mitochondria via its N-terminal MTS (Met-Lys-Ala-Ala-Arg-Phe-Val-Met-Arg-Ser-).³² However, MuRF1 is in the cytosol.^{8, 12, 13} We thus examined the interaction of MuRF1 and PDK4 to explore mechanisms of interaction. We constructed an MTS-truncated mutant of PDK4 (ΔMTS-PDK4), which exhibits appropriate cytosolic localization, and a GFP containing the MTS of PDK4 (MTS-GFP), which exhibits appropriate mitochondrial localization (Figure 5A). We confirmed that PDK4 (-myc) localizes in the mitochondria; PDK4 was detected in the mitochondrial fraction containing MFN2, a mitochondrial protein. As expected, ΔMTS-PDK4 was detected in the cytosolic fraction containing cytosolic GAPDH (Figure 5B).

Coimmunoprecipitation assays revealed that MuRF1 interacted with ΔMTS-PDK4 and PDK4, but MTS-GFP failed to interact with MuRF1 (Figure 5C). Deletion of MTS in PDK4 enhanced interaction between MuRF1 and PDK4 (Figure 5C). Since MuRF1 interacted with native PDK4, which is preferentially localized in the mitochondria (Figure 5A), MuRF1 likely interacted with PDK4 during translocation from cytosol to mitochondria.

3.6 MuRF1 induces SUMOylation of PDK4, but not ubiquitination

MuRF1 is suggested to trigger muscle protein degradation through substrate ubiquitination.^{8, 12, 13} PDK4 accumulation in the mitochondria in MuRF1^–/– skeletal muscle (Figure 4) led us to hypothesize a posttranslational modification by MuRF1. Therefore, we examined MuRF1 induction of ubiquitination and proteasomal degradation of PDK4. Cell lysates from cells transfected with mock vector or vectors containing MuRF1 or an inactive mutant with E2-binding domain deletion (ΔRING-MuRF1) were used in a PDK4 ubiquitination assay. Ubiquitination of PDK4 was measured in the presence of epoxomicin, a selective and strong proteasome inhibitor, to enhance the accumulation of ubiquitin conjugates. Overexpression of MuRF1 and ΔRING-MuRF1 failed to induce PDK4 ubiquitination (Figure 6A). Western blotting for PDK4 after CHX treatment showed that PDK4 degradation rate in MuRF1-expressing cells did not differ from degradation in mock- and ΔRING-MuRF1-expressing cells (Figure 6B). MuRF1 is probably not involved in ubiquitin-dependent degradation of PDK4.

As described in introduction, MuRF1 reportedly interacts with SUMO3 involved in SUMOylation via its RING domain,^{22, 23} and PDK4 contains the conserved SUMOylation motif -ψ-K-X-E/D- (ψ: bulky hydrophobic residue), ²⁸²Gly-Lys-Glu-Asp, which binds SUMOs 1, 2, and 3.²⁴ Thus, we examined whether MuRF1 induces SUMOylation of PDK4. The RING finger domain contains the SUMO2/3 binding site,^{23, 33} and ΔRING-MuRF1 was used as a negative control. Although SUMOylated PDK4 was weakly detected in the Mock and ΔRING-MuRF1 lane, we found that MuRF1 induced SUMOylation of PDK4 more than Mock and ΔRING-MuRF1 (Figure 6C).