Regulation of Skeletal Muscle Oxidative Capacity and Insulin Signaling by the Mitochondrial Rhomboid Protease PARL

PARL mRNA is associated with reduced mitochondrial mass in T2DM and aging

To investigate whether variation in PARL mRNA expression was associated with mitochondrial dysfunction and insulin resistance in humans, we studied healthy, young individuals (23±1y, n=11) and older subjects (64.4±1.2 y, n=11) matched for body composition, plasma free fatty acids and glucose concentrations as well as physical activity levels (Suppl. Table 1). Relative to younger participants, elderly individuals had elevated plasma insulin (p<0.01), lower insulin sensitivity (p<0.01) and a 35% reduction in resting metabolic rate (RMR; p<0.01; Suppl. Table 1). Analysis of the skeletal muscle transcriptome from muscle biopsies collected under fasting conditions revealed lower mRNA content of peroxisome proliferator-activated receptor gamma coactivator-β (PGC1β), nuclear respiratory factor (NRF1) and mitochondrial transcription factor A (TFAM) in elderly individuals compared to young controls, suggesting a reduction in mitochondrial number and function (Suppl. Table 2). Consistent with these observations, PARL mRNA, mtDNA (Fig 1a-b) and citrate synthase activity (Suppl. Table 2) (the latter two being markers of mitochondrial content (Wang et al., 1999)) were reduced in elderly subjects by 51%, 48% and 33% respectively (all, p<0.05). In addition, PARL mRNA was strongly correlated with TFAM mRNA (r=0.87, p<0.001) and mtDNA content (r-=0.61, p=0.001), suggesting PARL may be involved in the regulation of mitochondrial mass.

Functional studies by Petersen et al (Petersen et al., 2003) and Conley et al. (Conley et al., 2007) have observed reduced resting rates of ATP synthesis in elderly subjects with insulin resistance, suggesting that mitochondrial dysfunction with aging may be relevant to mitochondrial dysfunction seen in T2DM (Petersen et al., 2005). We therefore next identified that PARL mRNA and mtDNA content were reduced in subjects with T2DM relative to lean healthy controls (Fig 1a-b, both p<0.05), and in Pima Indians with a family history of T2DM (FH+; n=14) compared to those without a family history (FH−; n=7) (50±6 vs. 86±15 PARL mRNA/cyclophilin mRNA, p=0.05). Collectively, these observations establish that PARL mRNA and mitochondrial content in skeletal muscle are reduced in subjects with a genetic predisposition to develop T2DM, those presenting with T2DM, and in insulin resistant elderly individuals.

mCK-CAR-PARL(−) transgenic mice have reduced mitochondrial mass in muscle

Tissues with high demand for aerobic respiration such as the brain, skeletal muscle and heart, have the most prominent requirement for mitochondrial reticulum networks (Bach et al., 2003; Kirkwood et al., 1986). Accordingly, in Psammomys obesus, PARL protein was highly expressed in the brain, heart and skeletal muscle and predominantly in oxidative muscle fibers of lean healthy animals (Fig 1c). Similarly, in humans PARL mRNA is mainly expressed in skeletal muscle, brain, placenta, testis and heart (Suppl. Fig 1a). Given that skeletal muscle is the major site of insulin-stimulated glucose disposal and therefore a key tissue in the development of whole-body insulin resistance (Petersen et al., 2005), we set out to examine the functional consequences of reduced PARL protein on mitochondrial function and glucose metabolism in the skeletal muscle of a transgenic rodent model.

Mice expressing the human coxsackie /adenovirus receptor (CAR) under the control of the muscle creatine kinase promoter (mCK) were used to create mCK-CARPARL(−) mice muscle, herein referred to as PARL(−) mice. mCK-CAR mice have a dramatic enhancement of skeletal muscle transduction with site-specific injections of AdV-shuttle because they transgenetically overexpress CAR on mature myofibers (Nalbantoglu et al., 2001). We treated mCK-CAR mice (16 weeks of age, n=7) with a single intramuscular transverse injection into each lobe of the gastrocnemius (gastroc) and tibialis anterior/extensor digitorum longus (TA/EDL) muscle groups with AdV PARL(−) siRNA injected into the left leg and an “empty” vector with a GFP expression cassette in the right leg to serve as a collateral control (Supple Fig 1b). Analysis of PARL expression in the TA/EDL muscle of the PARL(−) leg revealed lower PARL protein that was associated with reduced PGC1α and CS protein levels relative to the control leg (mCK-GFP) (Fig 2a). Similar reductions in PARL protein was observed in gastroc muscle (data not shown). MtDNA (Fig 2d; p=0.03) content and CS activity (Fig 2e; p=0.007) were lower in PARL(−) muscle relative to GFP control muscle. Thus, it appears that mitochondrial content is affected by PARL protein content in muscle.

The key regulators of mitochondrial fusion, mitofusin 1 and 2 (Mfn1/2) and OPA1 (long and short isoform of OPA1 (Guillery et al., 2007)) were examined next. Surprisingly, MFN2 and OPA1 were lower in PARL(−) muscle, whereas, MFN1 was unchanged (Fig 2a). The reductions in OPA1 protein are consistent with previous reports demonstrating enhanced degradation of OPA1 protein (constitutive cleavage) and increased susceptibility to apoptosis (Cipolat et al., 2006; Griparic et al., 2007) with reduced PARL and DRP-1 (Mopert et al., 2009) expression. Therefore, lower OPA1 protein may result from reduced mitochondrial mass or dysfunction as observed in PARL-loss of function myotubes (see below). OPA1 functions include mitochondrial fusion and mitochondrial cristae remodeling (Cipolat et al., 2006; Griparic et al., 2007). To determine the effects of PARL depletion on cellular morphology, we examined mCK-CAR mice muscle by transmission electron microscopy. Electron micrographs demonstrated normal mitochondrial distribution (Fig 3 a-b) in both PARL(−) and GFP-mice muscle, however in PARL(−) (Fig 3 d) muscle mitochondrial cristae appeared “disorganized”, with some mitochondria totally lacking clearly defined cristae. This mitochondrial phenotype is characteristic of HeLa cells (Griparic et al., 2007) and “class II” remodeled mitochondria from mouse embryonic fibroblasts depleted of OPA1 (Frezza et al., 2006). We also observed increased activity of the pro-apoptopic protein Caspase-9 in PARL(−) TA/EDL muscle (1.8±0.4 vs 2.9±0.6 rfu/mg protein; p=0.048), consistent with a role for PARL in the anti-apoptotic function of OPA1 (Frezza et al., 2006). Interestingly, global PARL knockout mice display normal intrauterine development but from the fourth postnatal week undergo progressive multi-systemic atrophy leading to accelerated cellular apoptosis and cachectic death (Cipolat et al., 2006). Skeletal muscle loss and the development of sarcopenia is a hallmark feature of the aging process and is associated with elevated levels of apoptosis (Whitman et al., 2005), suggesting that lower PARL protein may be a contributing factor to muscle atrophy during aging.

The inner mitochondrial membrane cristae are easily damaged by ROS as a large portion of protein complexes associated with oxidative phosphorylation are embedded in this membrane (Gilkerson et al., 2003). Analysis of cysteine oxidation was performed by measuring protein disulfides and TBARS concentration (marker of lipid peroxidation) in PARL(−) muscle and revealed elevated oxidative damage relative to the control muscle (Fig 2f-g; both p<0.05). Superoxide dismutase (SOD) catalyses the dismutation of superoxide radicals into hydrogen peroxide and oxygen and is considered to be one of the primary antioxidant responses to elevated ROS production (Luo et al., 2006). SOD protein was elevated in the muscle of PARL(−) muscle when compared to control muscle, whereas there was no difference in glutathione peroxidase content (Fig 2a). Thus, PARL protein appears to influence total OPA1 protein levels and a low level of PARL protein expression is associated with increased oxidative damage in skeletal muscle. Importantly, reduced muscle PARL protein content results in abnormal mitochondrial morphology consistent with the mitochondrial structural abnormalities observed in skeletal muscle of subjects with T2DM (Kelley et al., 2002; Ritov et al., 2005).

PARL(−) mouse muscle has elevated lipid content and impaired insulin-signaling

Insulin resistant individuals have excess triglyceride and intramyocellular lipid (IMCL) accumulation (Forouhi et al., 1999; Jacob et al., 1999; Krssak et al., 1999; Perseghin et al., 1999) in skeletal muscle (Goodpaster et al., 2000a; Goodpaster et al., 2000b; Pan et al., 1997) that has been postulated to accumulate in part, due to deficient mitochondrial oxidative capacity (Befroy et al., 2007; McGarry, 1992; Morino et al., 2005; Petersen et al., 2003; Petersen et al., 2005; Zhang et al., 2007). We next measured the triglyceride content in PARL(−) mice muscle homogenates. PARL(−) muscle had elevated triglyceride content relative to the GFP-null muscle, although this did not reach significance (65±15 vs. 48±9 μM/mg protein; p=0.077). Several groups have shown the accretion of long chain fatty acyl-CoAs (LCFCoA), diacylglycerols (DAG) and ceramides (Adams et al., 2004; Adams et al., 2009; Chavez and Summers, 2003; Yu et al., 2002) is more closely associated with impaired insulin signaling than triglycerides or IMCL per se. Among these bioactive lipid intermediates, there is evidence to support a direct role for fatty acyl CoAs in the resistance to insulin action (Adams et al., 2004; Chavez and Summers, 2003; Yu et al., 2002). Consistently, knock down of PARL protein in skeletal muscle resulted in increased concentration of all fatty acyl CoAs examined (Fig 2h-total fatty aycl CoAs). In particular, the saturated long-chain myristoyl-CoA (14:0; p=0.031), palmitoyl-CoA (16:0; p=0.016) and stearoyl-CoA (18:0; p=0.004) were all elevated in PARL(−) muscle relative to the collateral GFP control.

We then tested whether elevated LCFCoA concentration in PARL(−) muscle was associated with impaired insulin signaling. Animal models and in vitro experiments have shown that the phosphorylation of serine residues (^{307, 636, 639, 1101}) of insulin-receptor substrate-1 (IRS-1) blocks the activity of IRS-1 on critical tyrosine sites that are required for phosphatidylinositol 3-kinase (PI3K) activation (at phosphorylation site PI3K p85-Tyr⁴⁵⁸ (Luo et al., 2007; Yi et al., 2007). Downstream of PI3K, Akt phosphorylation at Ser⁴⁷³ has been shown to be a key step for the activation of glucose transport in skeletal muscle (Morino et al., 2005). In fasting conditions, we observed reduced basal phosphorylation of PI3K^−Tyr458 and AKt-Ser⁴⁷³ in PARL(−) TA/EDL muscle relative to control muscle (Fig 2b). In the insulin-stimulated state, phosphorylation at IRS-1^Ser-1101 was elevated in PARL(−) muscle relative to control, in parallel with lower insulin-stimulated Akt^−Ser473 phosporylation (Fig 2b). We next isolated the cytosolic and plasma membrane fraction of PARL(−) and AdV-GFP red (oxidative) gastrocnemius muscle under insulin-stimulated conditions. No change was observed in the plasma membrane content of the insulin receptor-β between PARL(−) and GFP control muscle(Fig 2c). However, reductions in muscle PARL protein resulted in lower insulin-stimulated translocation of GLUT4 protein to the plasma membrane (Fig 2c). Together, these data indicate that skeletal muscle-specific reduction of PARL protein results in elevated saturated long chain fatty acyl CoAs, impaired insulin signaling and reduced GLUT4 translocation to the plasma membrane.

HFD lowers muscle PARL and PGC1α mRNA along with glucose tolerance in mice

Impairments in mitochondria transcription and function (Civitarese et al., 2006; Patti et al., 2003; Petersen et al., 2004) along with lower PARL mRNA (see above) are present in the pre-diabetic state, suggesting that PARL and mitochondrial abnormalities may be primary factors in the development of skeletal muscle insulin resistance. Because a high fat diet (HFD) causes insulin resistance and a decrease in the expression of multiple genes involved in mitochondrial biogenesis and oxidative phosphorylation (Koves et al., 2008; Sparks et al., 2006; Sparks et al., 2005; Tunstall and Cameron-Smith, 2005), we examined whether the regulation of PARL transcription, precedes or coincides with changes in glucose tolerance, mtDNA, fusion and fission gene expression in C57Bl6/J mice fed a HFD for 4 or 10 wk. After 4 wk of the HFD, there was no difference in fasting glucose levels between groups, however, blood glucose levels were significantly elevated in mice fed a HFD for 10wk (Suppl. Table 1, <0.05). By 4 wk, HFD-mice had glucose intolerance and resistance to the glucose lowering effects of insulin that worsened by the end of the study (Suppl. Table 3). Glucose homeostasis remained normal in control diet mice.

Despite the insulin resistance in HFD mice, there was no change in the expression of the majority of genes examined at 4 wk (Suppl. Table 3). DRP1 was the only gene modified, with an up regulation in HFD group (Suppl. Table 3, p<0.05). The increase in DRP1 expression in the HFD mice may relate to the role of DRP1 in the tethering and biogenesis of peroxisomes (Koch et al., 2003), that participate in the β-oxidation of fatty acids and the adaptation to nutritional stimuli that elevates lipid flux (Reddy and Mannaerts, 1994). After 10wk of HFD, there was a down regulation in mtDNA that was mirrored by a suppression of PARL and PGC1α mRNA (Suppl. Table 3, p<0.05). No other significant changes were observed, although there was a trend for elevated PGC1β and OPA1 mRNA (Suppl. Table 1). Together, these data suggest that in the progression of insulin resistance to T2DM, the suppression of PARL and PGC1α parallel the impairment in mitochondrial biogenesis.

Abnormal fusion and fission protein expression in primary cultures of human muscle

To address the biological significance of reduced PARL protein expression in human muscle, we first investigated mitochondrial content and PARL protein expression and key proteins of mitochondrial energetics, fusion and fission in primary human skeletal muscle cell model derived from vastus lateralis samples from our clinical cohorts. Consistent with our mRNA analysis in whole skeletal muscle, PARL mRNA (Suppl. Fig 2a) and protein levels (Fig 4a), were strikingly reduced in primary myotubes derived from both T2DM and elderly subjects vs. myotubes from control subjects (both p<0.001). In mammals, the separation of adjoined mitochondria by the fission process is chiefly controlled by the cytosolic dynamin-related GTPase protein 1 (DRP1) that is recruited to the mitochondrial outer membrane by FIS1 (Stojanovski et al., 2004). FIS1 protein was lowest in myotubes from T2DM individuals relative to myotubes from elderly (p=0.01), which in turn had lower FIS1 protein than in controls (p=0.005) (Fig 4a). DRP1 was not different between the three cohorts (Fig 4a) possibly reflecting high cytosolic abundance of DRP1 (Smirnova et al., 1998). The fusion of mitochondria requires MFN-1 and -2 to conjoin the “tethered” outer membranes of juxtaposed mitochondria (Bach et al., 2003; Pellegrini and Scorrano, 2007), whereas OPA1 unifies the inner mitochondria membranes (Frezza et al., 2006). However, despite lower protein content of MFN2 (p=0.04) and OPA1 (Fig 4a, p<0.001) in myotubes from T2DM/elderly subjects compared to those of control subjects, there was no significant difference in MFN1 (Fig 4a) between the 3 groups. Similarly, reduced MFN2 protein was observed in skeletal muscle samples from obese humans (Bach et al., 2003). Thus, we show for the first time that primary skeletal muscle cells derived from elderly and T2DM subjects are characterized by low expression of PARL, and mitochondrial fusion and fission proteins (FIS1, MFN2, and OPA1).

Low PARL protein expression was also associated with impaired muscle energetics, with lower levels of mtDNA (Fig 4b), CS protein (Fig 3a) and impaired insulin-stimulated glycogen synthesis (Fig 4c) in insulin resistant groups vs controls. However, we observed no difference in PGC1α and TFAM mRNAs (Suppl. Fig 2a) between the three cohorts. Despite the fact that sirtuin-1 (SIRT1) activates PGC1α and mitochondrial biogenesis (Nemoto et al., 2005; Rodgers et al., 2005), we observed a surprising up-regulation of SIRT1 mRNA (Suppl. Fig 2a). However, SIRT1 protein was lower in myotubes of T2DM participants (p=0.04) and elderly individuals (p=0.03) relative to healthy controls (Fig 4a). The mitochondrial fractions derived from myotubes of insulin resistant subjects (T2DM and elderly) demonstrated reduced levels of DTT-reducible cysteines (measure of irreversible cysteine oxidation; Fig 4d) and SOD2 activity (Fig 4e) compared with controls. Together, these data show myotubes derived from insulin resistant subjects are characterized by low expression of PARL, FIS1, MFN2, and OPA1 protein, reduced mitochondrial mass, non-oxidative glucose metabolism and elevated oxidative stress. Furthermore, the reductions in mitochondrial mass in elderly and T2DM myotubes were independent of PGC1α, a known regulator of mitochondrial mass.

PARL protein is a determinant of mitochondrial mass in primary human muscle cells

To investigate whether reduced PARL may be causative, rather than a consequence, of perturbed mitochondrial function in human-derived muscle cells, we treated myotubes from the lean-young cohort with AdV-PARL(−)-siRNA and AdV-GFP control shuttles. Myotubes transduced with AdV-PARL-siRNA began to senesce (detachment of adherent cells) on day 7 of differentiation (data not shown) in conjunction with lower OPA1 protein and elevated protein expression of the stress-induced, pro-apoptotic protein BAX (Fig 5a) suggesting an elevated rate of apoptosis (Frezza et al., 2006). No obvious morphological defects were observed between the two types of myotubes (decreased PARL vs normal PARL). Mitochondrial mass (p<0.001) and mtDNA (p=0.04) were both reduced in PARL depleted myotubes (Fig 5b).

We next examined potential mechanisms by which PARL may regulate mitochondrial biogenesis. Regulated intramembrane proteolysis (RIP) is a signaling mechanism that involves the generation of biologically active peptides from membrane-tethered precursor proteins (Sik et al., 2004). PARL is cleaved by RIP to generate a smaller 33KDa inner mitochondrial membrane protein termed PARL-C terminal product of β-cleavage (PACT) and in turn liberates a nuclear targeted 25-amino acid peptide named PARL-β (Pβ) (Sik et al., 2004). Fluorescent microscopy analysis of healthy human myotubes transfected with a synthetic Pβ-construct with a GFP (Pβ-GFP) flag in the 5′position was used to study Pβ subcellular localization. Image overlay analysis demonstrated a small fraction of Pβ-GFP localized within the nucleus (white arrows), with the majority of Pβ-GFP in the perinuclear region (pound sign), with mitochondria evenly distributed through out the myotubes (as indicated by Mitotracker Red) (Supple. Fig 2b, i-iv). It has been hypothesized that Pβ may mediate mitochondria-to-nucleus cross-talk and regulate transcription factor(s) to alter the expression of mitochondrial specific genes (Sik et al., 2004). We next treated myotubes with naive synthetic Pβ with no GFP flag. Consistently, Pβ (0.1μg) robustly increased the mRNA expression of PARL and genes involved in mitochondrial biogenesis (PGC1β and NRF1) and mitochondrial fusion (MFN-1 & -2; Supple Table 4 and OPA1 protein expression (Fig 6a). No changes were observed in PGC1α, NRF2, and RPLPO (internal control) mRNA (Suppl. Table 4). Importantly, Pβ increased SIRT1 protein expression (Fig 6 a,d) in conjunction with elevated mitochondrial biogenesis (Fig 6b), suggesting a potential Pβ-SIRT1 regulation of mitochondrial mass. To confirm Pβ effects on cellular energetics and SIRT1 protein content we treated RMS-13 human myoblasts with Pβ-GFP plasmid containing the wild type Pβ (PβWt) sequence (amino acids 53-77 of human PARL gene), Pβ-GFP plasmid with coded mutation of the nuclear localization sequence (PβNLSM), the backbone pEGFPN1 plasmid as control and synthetic naïve-Pβ peptide (no GFP-flag). Expression of PβWt and transfection of naïve-Pβ peptide increased cellular oxygen consumption above that of PβNLSM (Fig 6e) and coincided with an increase in SIRT1 protein (Fig 6d). Consistent with the ability of SIRT1 to regulate mitochondrial mass (Lagouge et al., 2006), PβWt in RMS13 cells robustly increased mitochondrial biogenesis (Fig 6b) however, the expression of PARL cDNA in myoblasts did not alter mitochondrial mass (data not shown). Conversely, expression of PARL in myoblasts increased cellular oxygen consumption (Fig 6f). This suggests that Pβ is able to modulate mitochondrial energetics by increasing the transcriptional machinery involved in regulating mitochondrial mass, whereas PARL effect on cellular oxygen consumption may relate to its effects on cristae remodeling or direct interaction with OXPHOS protein(s).

PARL protein depletion in myotubes lower energetics and increases ROS production

Consistent with lower mitochondrial mass, PARL(−) myotubes displayed lower rates of oxygen consumption (Fig 5c; p<0.01), diminished COX II activity (Fig 5d, p<0.01), and lower rates of palmitate oxidation (Fig 5e, p=0.04), indicating a reduction in oxidative capacity in PARL(−) myotubes. PARL(−) myotubes were also characterized by impairments in non-oxidative glucose disposal with reduced rates of insulin-stimulated glycogen synthesis (Fig 4c) and AKT-phosphorylation (Fig 4c-upper panel). The effects on oxygen consumption and glucose metabolism is similar to lower cellular respiration and glucose oxidation in mammalian fibroblasts deficient in MFN2 protein (Bach et al., 2003; Pellegrini and Scorrano, 2007). Analysis of mitochondrial morphology by electron microscopy in PARL(−) myotubes showed mitochondria to be uniformly distributed throughout the myotubes (Fig 3e-f). Confocal microscopy reconfirmed distribution results with mitochondria contained within reticulum networks (Fig 3g-h). Therefore, PARL protein depletion in primary human muscle cells results in reduced mitochondrial mass, impaired oxidative capacity and lower glycogen storage. However, PARL does not appear to play a role in mitochondrial distribution in human muscle cells-consistant with recent reports in other mammalian models (Cipolat et al., 2006; Griparic et al., 2007). Therefore, PARL appears to be involved in mitochondrial bioenergetics by modulating cristae remodeling and/or participation in the regulation of mitochondrial mass or cellular apoptosis.

Disturbance in mitochondrial cristae morphology can result in elevated ROS production (Yu et al., 2006) as a significant proportion of the oxidative phosphorylation machinery is embedded in the inner mitochondrial membrane. Analysis of ROS production using carboxy-dichloro-fluorescence (Civitarese et al., 2006) in PARL(−) myotubes revealed a 84% increase in H₂0₂ levels (p=0.02) (Fig 5f), along with elevated lipid peroxidation (Fig 5g), cysteine oxidation and SOD2 activity (Fig 4d-e; all p<0.05) relative to AdV-GFP. In addition, acute (3 hour) exposure of PARL(−) myotubes to 20 mM glucose resulted in a 96% increase in H₂O₂ production compared with basal levels (p=0.04, data not shown). Together, this indicates that reduced PARL protein may explain a proportion of the elevated oxidative stress observed in myotubes from elderly and T2DM subjects.

In summary, we have shown that lowering PARL protein in human muscle cells results in lower mitochondrial oxidative capacity, coupled with reduced mitochondrial mass, increased protein oxidation and ROS production and impaired insulin signaling — all known metabolic defects in T2DM and aging (Conley et al., 2007; Imoto et al., 2006; Kelley et al., 2002; Morino et al., 2005; Petersen et al., 2003; Petersen et al., 2005; Ritov et al., 2005).