Activation of a Novel Calcineurin-mediated Insulin-like Growth Factor-1 Receptor Pathway, Altered Metabolism, and Tumor Cell Invasion in Cells Subjected to Mitochondrial Respiratory Stress*

We have previously shown that disruption of mitochondrial membrane potential by depletion of mitochondrial DNA (mtDNA) or treatment with a mitochondrial ionophore, carbonyl cyanide m-chlorophenylhydrazone, initiates a stress signaling, which causes resistance to apoptosis, and induces invasive behavior in C2C12 myocytes and A549 cells. In the present study we show that calcineurin (Cn), activated as part of this stress signaling, plays an important role in increased glucose uptake and glycolysis. Here we report that, although both insulin and insulin-like growth factor-1 receptor levels (IR and IGF1R, respectively) are increased in response to mitochondrial stress, autophosphorylation of IGF1R was selectively increased suggesting a shift in receptor pathways. Using an approach with FK506, an inhibitor of Cn, and mRNA silencing by small interference RNA we show that mitochondrial stress-activated Cn is critical for increased GLUT 4 and IGF1R expression and activation. The importance of the IGF1R pathway in cell survival under mitochondrial stress is demonstrated by increased apoptosis either by IGF1R mRNA silencing or by treatment with IGF1R inhibitors (AG1024 and picropodophyllin). This study describes a novel mechanism of mitochondrial stress-induced metabolic shift involving Cn with implications in resistance to apoptosis and tumor proliferation.

Mitochondria play important roles in energy production, cell metabolism, and both integration and execution of apoptotic signals. Dysfunctional mitochondria in mammalian cells trigger a stress signaling (1)(2)(3) analogous to the mitochondrial retrograde signaling pathway described in °yeast cells (4). A distinct difference between the yeast and mammalian cell systems is the involvement of RTG signaling pathway in the former, as against a retrograde Ca 2ϩ signaling cascade in the latter (5). In both C2C12 myocytes, and A549 lung carcinoma cells, disruption of mitochondrial membrane potential (⌬⌿m) either due to mitochondrial genetic stress (partial depletion of mtDNA) or treatment with mitochondrial ionophore, CCCP, 2 caused a sustained increase in cytosolic free Ca 2 and activation of Ca 2ϩsensitive factors, calcineurin (Cn), NFAT, C/EBP␦ (1,2,6), and a novel IB␤-dependent NFB pathway (7). In 143B osteosarcoma and other cells, mitochondrial stress resulting from mtDNA depletion or protein misfolding caused activation of CREB, and a C/EBP family factor CHOPS, respectively (8,9). Indeed these same transcription factors have been implicated in the regulation of a number of nuclear marker genes that respond to mitochondrial respiratory stress signaling in different cells (10,11).
Currently there is increasing evidence that mtDNA content directly reflects on the metabolic state of cells (11)(12)(13). Some of the °cells, lacking mtDNA, exhibit slow growth and higher sensitivity to apoptotic stimuli (14,15). In contrast, mtDNA depletion induced cell proliferation and resistance to apoptosis in C2C12 and A549 cells (1,6), 143B osteosarcoma cells (16), and SK-Hep1 cells (17). Similarly, a cybrid cell line carrying mutations in mtDNA-encoded ATPase6 gene exhibited increased tumorigenicity and reduced apoptosis (18,19), indicating a role for mtDNA mutations in tumor progression. An increasing number of recent studies shows that proliferating tumors in both human patients and animal models contain mutated or deleted mtDNA and/or dysfunctional mitochondria (19 -21). In mtDNA-depleted C2C12 and A549 cells, increased resistance to apoptosis was associated with higher expression of Bcl2 and/or increased phosphorylation of Bid, Bad, and Bax (22). Rapidly growing tumors meet their metabolic demand by increased expression of genes encoding glucose transporters and glycolytic enzymes (23). In the present study we investigated the mechanism of cell proliferation and induced glycolysis in C2C12 cells subjected to mitochondrial respiratory stress. Our results show that selective inhibition of IR autophosphorylation and Cn-dependent activation of the IGF1R pathway is the basis for increased glucose utilization and cell proliferation. Interestingly, mitochondrial stress-induced metabolic change appears to be an important survival factor in these cells, because blocking the IGF1R function caused increased cell death. Results also show that elevated GLUT 4 and IGF1R levels are directly related to increased Cn activity, which is an essential component of the mitochondria-to-nucleus stress signaling pathway.

EXPERIMENTAL PROCEDURES
Cell Lines and Culture Conditions-Murine C2C12 skeletal myoblasts (ATCC CRL1772), human pulmonary carcinoma A549 cells (ATTC CCL 185), mouse fibroblasts NIH 3T3 (ATCC CRL-1658), and rat H9C2 cardiac myocytes (ATCC CRL 1446) were grown in Dulbecco's modified Eagle's medium (Life Technology Inc.) supplemented with 10% FBS and 0.1% gentamycin. In some specified experiments charcoal-treated FBS was used. Depletion of mtDNA was carried out by EtBr treatment (100 ng/ml, for 30 passages) as described before (2). Selected clones containing ϳ20% mtDNA contents were grown in presence of 1 mM sodium pyruvate and 50 g/ml uridine. Reverted cells represent mtDNA-depleted cells (with 85% DNA depleted) grown for 30 cycles in the absence of EtBr until the mtDNA content reached 80% of control cells. The mtDNA contents were estimated by Southern hybridization using a DNA probe from cytochrome oxidase I ϩ II coding region as described before (2).
Calcineurin A␣, IGF1R, and IR mRNA Knockdown by siRNA-Predesigned siRNAs for mouse CnA␣ (ID #292199), IGF1R (ID #159115), and IR (ID #67808) and negative controls (scrambled siRNA) were purchased from Ambion Inc. (Austin, TX). Control and mtDNA-depleted cells (10 4 cells/well) were transfected with pre-annealed double-stranded siRNAs at a final concentration of 30 nM by the method of reverse transfection. Transient transfections were carried out in triplicate using siPORT NeoFX reagent (Ambion Inc). RNA was isolated 48 h after transfections using TRIzol reagent (Invitrogen), and the level of silencing of CnA␣, IGF1R, and IR mRNA was quantitated by real-time PCR as described later in this section.
Uptake of 2-Deoxyglucose-Glucose uptake was measured as described previously (25). Briefly, 10 6 cells grown in 6-well plates were serum-starved for 6 h, and incubated in the presence or absence of indicated levels of insulin for 30 min in glucose-free medium. 1 Ci of 2-[ 3 H]DOG (1 mCi/0.1 mmol, American Radiolabeled Chemicals, St. Louis, MO) was added to each well containing 2 ml of medium and incubated for an additional 15 min. The transport was terminated by washing cells rapidly three times with ice-cold PBS and lysed in 1 N NaOH. Aliquots of the lysates were counted in liquid scintillation counter (Beckman Instruments, Fullerton, CA).
Measurement of Cellular ATP Levels-Total cellular ATP levels were determined by bioluminescence using ATP-bioluminescent somatic cell assay kit (Sigma) as described before (6). Briefly, 10 6 cells were lysed using the "somatic cell releasing reagent," which caused the release of cellular ATP by altering membrane permeability. ATP-dependent formation of light was measured in a Chameleon Multilabel Microplate Reader (Bioscan Inc., Washington, D. C.) with appropriate ATP standards. In these experiments, cells were treated with 25 M CCCP and/or 25 M oligomycin for 1 h.
Measurement of Hexokinase, Phosphofructokinase, and Pyruvate Kinase Activities-Hexokinase assay was carried out as described by Miccoli et al. (26). Glucose 6-phosphate produced in the hexokinase reaction was coupled to glucose-6-phosphate dehydrogenase. The reaction was started by the addition of 50 g of the cell lysate, and the rate of NADPH production was measured by the increase in absorbance at 340 nm for 20 min in a Cary Varian spectrophotometer. Phosphofructokinase activity was measured spectrophotometrically by coupling the enzyme with aldolase, triosephosphate isomerase, and glyceraldehyde-3-phosphate dehydrogenase as described before (27). The reaction was started by the addition of 50 g of the cell lysate, and the rate of NADH metabolized over a time of 20 min was measured by the decrease in absorbance at 340 nm. Pyruvate kinase reaction was coupled to lactate dehydrogenase, and the activity was measured by the oxidation of NADH at 340 nm for 20 min as described before (28). The reaction was started by the addition of 0.15 mM phosphoenolpyruvate and 50 g of the cell lysate.
Calcineurin Activity Assay-Calcineurin activity was assayed according to Antoni et al. (29) with the following modifications. Total serine/threonine phosphatase activity was first determined by incubating cell lysate (2 g) from untreated, AICAR (2 mM, 2 h)-treated, and FK506 (10 nM, 30 min)-treated cells with threonine phosphopeptide (Upstate, Cat. no. 12-219) in a final volume of 25 l for 20 min at 30°C in 50 M Tris-Cl, pH 7.0, 0.1 mM CaCl 2 . The phosphate released was quantitated by incubating with 100 l of malachite green and measuring the absorbance at 660 nm after 15 min. In parallel, control incubations containing phosphatase inhibitor mix I (NaF, 5 mM; okadaic acid, 500 nM; orthovanadate, 100 M; FK506, 1 nM; calyculin, 0.1 M; and cyclosporine A, 100 M) and phosphatase inhibitors mix II (FK506, 1 nM; calyculin, 0.1 M; and cyclosporine A, 100 M) were also carried out. Nonspecific phosphatase activity under these conditions was subtracted from total serine/threonine phosphatase activity referred to as corrected activity. Calcineurin activity was calculated by subtracting the activity obtained from incubation with phosphatase inhibitor mix II from the corrected activity.
Immunoprecipitation and Receptor Autophosphorylation-Plasma membrane fractions from control and mtDNA-depleted C2C12 cells were isolated as described (30). Autophosphorylation of the IR and IGF1R in the plasma membrane fractions was carried out at 37°C in the presence of 10 mM MnCl 2 , [␥-32 P]ATP (0.2 mM and 2-15 cpm/fmol), 0.1 M insulin for 30 min. The reaction was terminated by adding 15 mM EDTA. The autophosphorylated receptors (100 g of plasma membrane protein) were immunoprecipitated overnight at 4°C with the IR or IGF1R␤ antibody (2 g/ml). The immune complexes were collected on Protein A-agarose beads (Sigma) and washed extensively with PBS. The immunoprecipitates were extracted from the beads with 2ϫ Laemmli buffer devoid of ␤-mercaptoethanol at 95°C for 5 min for further analysis.
Immunoblot Analysis-The protein contents were estimated by the Lowry method (31). Cytosolic, nuclear, and plasma membrane proteins (30 g) solubilized in Laemmli sample buffer were resolved by electrophoresis on 10 or 12% SDS-polyacrylamide gels (32) and subjected to immunoblot analysis. The immunoblots were developed using SuperSignal West Femto maximum sensitivity substrate (Pierce). In each case, blots presented here are a representative of two to three independent experiments.
Plasma Membrane Lawn Assay-The level of plasma membrane GLUT 4 was determined using the plasma membrane lawn assay as described earlier (33). Cells grown on poly-L-lysine-coated coverslips were treated with or without insulin (100 nM for 30 min). The coverslips were rinsed in PBS and subjected to hypotonic shock with three washes (10 s each) in membrane buffer (70 mM KCl, 5 mM MgCl 2 , 3 mM EGTA, 1 mM dithiothreitol, 30 mM HEPES, pH 7.2). The coverslips were submerged in membrane buffer and sonicated using a Probe sonicator, to generate a lawn of plasma membrane fragments that remained attached to the coverslip. The fragments were then immunolabeled with anti-GLUT 4 polyclonal antibody (1:100, U.S. Biologicals, Swampscott, MA) and Alexa 488 goat anti-rabbit secondary antibody (Molecular Probes, 1:300). The membrane fragments were viewed through an immunofluorescence microscope. At least six fields were examined within any one experiment for each condition.
mRNA Quantitation by Real-time PCR-Total RNA was isolated using TRIzol reagent as per the supplier's protocol (Invitrogen). cDNA was generated from 5 g of RNA using the cDNA Archive kit from Applied Biosystems (Foster City, CA), and 50 ng of this cDNA was used as template per reaction. Relative quantification of GLUT 4, IGF1R, and IR mRNA by real-time PCR was done using a TaqMan assay probe (Applied Biosystems), and CnA␣ mRNA was quantified using SYBR Green (Applied Biosystems) in an ABI PRISM 7300 sequence detection system (Applied Biosystems). Data were normalized to 18 S RNA (for TaqMan assay) and ␤-actin contents (for SYBR Green) as endogenous controls.
Annexin V Assay-The assay for cells undergoing apoptosis was performed using a Nexin Kit (Guava Technologies, Hayward, CA) according to the manufacturer's suggested protocol. Cells (10 6 each) treated without or with inhibitors (Genistein, 0.1 mg/ml; AG1024, 10 M; and PPP, 2.5 M) for 2 h were washed with 1 ml of 1ϫ Nexin Buffer (Guava Technologies) and resuspended in the same buffer. After labeling with Annexin V-PE, the percentage of apoptotic cells was quantified using the Personal Cytometer (Guava Technologies). In the case of the siRNA experiments, cells were harvested 48 h after the siRNA transfections and counted as described above.
Matrigel Invasion Assay-The in vitro invasion assays were carried out as described previously (34). The Matrigel invasion chambers were prepared at 1:2 dilution of Matrigel (BD Biosciences, Belford, MA) as described before (1). Equal numbers of viable cells (4 ϫ 10 4 ) were seeded on top of the Matrigel layer. After incubation for 24 h at 37°C, non-invading cells in the Matrigel layer were quantitatively removed, and the micro-porous membrane containing invaded cells was stained and viewed under a Olympus BX 61 bright field microscope as described before (22). At least six fields were examined within any one experiment for each condition.
Statistical Analysis-Data on enzyme activity, glucose uptake, mRNA quantitation, and immunoblot analysis have been presented as means Ϯ S.D. of three to five independent measurements. Differences between paired variables were determined using two-way analysis of variance. p values Ͻ0.05 were considered statistically significant, and p values Ͻ0.001 were highly significant.

RESULTS
Increased Glucose Uptake by mtDNA-depleted Cells-Initially, we evaluated glucose uptake by control and mtDNA-depleted C2C12 cells containing 20% of control cell mtDNA content, and reverted cells with mtDNA content restored to ϳ80% of control cells (Fig. 1A, left panel). Fig. 1A (right panel) shows , and the uptake was measured as described under "Experimental Procedures." Nonspecific background counts with heat-killed cells were subtracted from experimental values. Averages Ϯ S.E. of three independent estimates (each taken in triplicate) are presented. In B the difference in glucose uptake between control and mtDNA-depleted cells is indicated: **, p Ͻ 0.001. Statistical differences between mtDNA-depleted cells without added insulin and with added 1 M insulin are indicated: *, p Յ 0.05. In control and reverted cells, the values with added insulin (*) compared with or without added insulin show a significance of p Յ 0.05. In C and D: *, p Յ 0.05; **, p Յ 0.001. the relative levels of nuclear DNA as assessed by the cytochrome c oxidase IVil amplicons using 10 ng of genomic DNA template in each case. Aliquots of cells from the same parent stock culture as shown in Fig. 1A were used to ascertain uniform cellular mtDNA contents in all experiments reported herein.
We studied the uptake of 2-DOG in the three cell types under uninduced and insulin stimulation, with insulin concentrations ranging from 0 to 1 M (Fig. 1B). The mtDNA-depleted cells showed Ͼ2-fold higher glucose uptake in the absence of added insulin (basal) compared with control cells (p Յ 0.001). The reverted cells showed glucose uptake closer to the control cells. With added 0.1 M insulin, the mtDNA-depleted cells showed nearly 2-fold higher glucose uptake than control cells. With 0.5 M and 1 M insulin, the increase was only 60 and 40%, respectively, suggesting a saturating effect. Results also showed that all three cell lines take up higher 2-DOG in response to increasing levels of insulin, although the increase was markedly apparent in control and reverted cells (p Ͻ 0.05), but less apparent in mtDNA-depleted cells. The glucose uptake in mtDNA-depleted cells with added 1 M insulin as compared with no insulin controls was also significant (p Ͻ 0.05). Although not shown, a similar pattern of 3-O-methyl-D-glucose uptake was observed confirming altered transport.
The mitochondrial stress-induced signaling in different cell lines involves activation of Cn and its related factors (2,5). Fig.  1C shows that FK506, an inhibitor of Cn, inhibited glucose uptake by control and reverted cells only marginally (15 and 20%, respectively), but to a markedly higher extent of 70% in mtDNA-depleted cells (p Յ 0.001). These results suggested the role of Cn in glucose uptake. A reduced glucose uptake by reverted cells also suggested a direct link between mitochondrial genetic stress and glucose uptake.
Previously we proposed that disruption of ⌬⌿m by CCCP treatment also activates the mitochondrial stress signaling similar to mtDNA depletion (2). In accordance with these observations, CCCP treatment indeed caused ϳ2.5-fold higher glucose uptake by C2C12 cells, which was further increased by ϳ40% in response to added insulin (Fig. 1D). These results further support the role of mitochondrial stress in inducing cellular glucose uptake.
The predominant isoforms of glucose transporters in skeletal muscle cells are GLUT 4 and GLUT 1 (35). There was no difference in the levels of GLUT 1 between control and mtDNAdepleted cells (data not shown). In keeping with the results of glucose uptake, immunoblot in Fig. 2A shows that mtDNAdepleted cells contain Ͼ2-fold higher GLUT 4 protein as compared with control cells, which is in accordance with 3-to 4-fold higher mRNA level (Fig. 2B). GLUT 4 protein ( Fig. 2A) and mRNA (Fig. 2B) levels in reverted cells were markedly lower than that of mtDNA-depleted cells and closer to that of control cells. These results show that increased GLUT 4 expression is directly related to mitochondrial stress. GLUT 4 mRNA level in mtDNA-depleted cells, but not in control cells, was sensitive to AG1024, an inhibitor of IGF1R. Although not shown, PPP, a more specific inhibitor of IGF1R (36, 37) also decreased GLUT 4 protein levels in mtDNA-depleted cells. These results suggest that increased GLUT 4 expression in response to mitochondrial stress may be related to stress-mediated activation of IGF1R pathway. In keeping with increased glucose uptake (Fig. 1D), addition of CCCP increased the level of GLUT 4 in control cells. Additionally, FK506 added to mtDNA-depleted cells resulted in a marked reduction of both GLUT 4 protein and mRNA level (Fig. 2, A and B) further suggesting the role of Cn pathway in increased glucose uptake.
Relative contributions of IGF1R and Cn activation on GLUT 4 expression was further investigated by overexpression of IGF1R in mtDNA-depleted cells treated with Cn inhibitor FK506. As seen from Fig. 2C, FK506 treatment caused a 60% reduction of GLUT 4 mRNA in mock transfected mtDNA-depleted cells. Transfection of mtDNA-depleted cells with IGF1R cDNA caused a near 2-fold increase in GLUT 4 mRNA level, which was reduced to near basal level by FK506. These results suggest that IGF1R and Cn pathways may cooperatively induce the expression of GLUT 4 mRNA. The results also show that GLUT 4 protein and mRNA contents were increased by 2.5-to 4.0-fold in mtDNA-depleted cells that returned close to the control cell level in reverted cells.
We carried out the lawn assay for determining the level of GLUT 4 protein in the plasma membrane compartment of control and mtDNA-depleted cells. Fig. 2D shows that the basal level of GLUT 4 in the plasma membrane fraction was higher in were subjected to immunoblot analysis, and the blot was co-developed with antibody to GLUT 4 (1:1000 dilution) and Na ϩ /K ϩ -ATPase (1:1000 dilution). The latter was used as a loading control. B, RT-PCR analysis of total RNA (50 ng each of RT product) from control, mtDNA-depleted, and reverted C2C12 cells. The values represent average Ϯ S.E. of three independent assays. *, p Յ 0.05; **, p Յ 0.001. C, real-time PCR analysis of total RNA (50 ng each of RT product) from mtDNA-depleted C2C12 cells. Cells were transfected with 2 g of IGF1R cDNA for 48 h, and total RNA was isolated. Real-time PCR analysis of GLUT 4 mRNA was done using a TaqMan probe as described under "Experimental Procedures." *, p Յ 0.05; **, p Յ 0.001. D, lawn assay for GLUT 4 contents of plasma membrane fractions from control and mtDNA-depleted cells before or after treatment with insulin (100 nM, 30 min). Details of immunoblot analysis, real-time PCR, and the lawn assay are as described under "Experimental Procedures." mtDNA-depleted cells as compared with control cells. Furthermore, insulin treatment increased the plasma membrane contents of GLUT 4 in both cell types. These results are consistent with GLUT4 mRNA and protein levels in mtDNA-depleted cells.
Increased Expression of GLUT 4 and IGF1R mRNAs and Activation of IGF1 Receptor in mtDNA-depleted Cells-IR and IGF1R both activate GLUT 4 and glucose uptake in C2C12 cells. To gain insight into the mechanism of increased glucose uptake and induced expression of GLUT 4 in mtDNA-depleted cells, we studied the effects of specific inhibitors of receptor pathways that regulate glucose uptake. Fig. 3A shows that both wortmannin and LY294002, general inhibitors of the phosphatidylinositol 3-kinase pathway (38,39), inhibited glucose uptake both in control and mtDNA-depleted cells by ϳ70 -85%. Genistein, a known inhibitor of IR (40) inhibited glucose uptake by ϳ30% in control cells and by 15% in mtDNA-depleted cells. AG1024, which inhibits IGF1R more preferentially at the concentration used (10 M), had no significant effect on glucose uptake in control cells, whereas it inhibited glucose uptake by nearly 85% in mtDNA-depleted cells (Fig. 3A). PPP, a specific inhibitor of IGF1R (36,37), is known to specifically inhibit the phosphorylation of Tyr 1136 , which is critical for the activation of IGF1 receptor, but has no effect on IR activation (41). PPP inhibited glucose uptake to ϳ46% in control cells and 88% in mtDNA-depleted cells. The differential effects of these inhibitors suggest that insulin receptor plays a more prominent role in glucose uptake in control cells, whereas IGF1R plays a more prominent role in mtDNA-depleted cells.
We then evaluated IR and IGF1R levels and their autophosphorylation in these cells. The immunoblot in Fig. 3B shows that the IR level was increased by ϳ1.5-fold in mtDNA-depleted cells, which was back to near control cell level in reverted cells. Treatment with FK506 had no significant effect on IR protein level in control and reverted cells, whereas it inhibited the level by ϳ50% in mtDNA-depleted cells. The autoradiogram in Fig. 3C shows the level of autophosphorylation of IR in control, depleted, and reverted cells. Results show high levels of autophosphorylation in control and reverted cells but a markedly lower level in mtDNA-depleted cells. These levels were not significantly affected by FK506, suggesting marked differences in the level of IR activation in control and mtDNA-depleted cells.
The immunoblot in Fig. 3D shows that the IGF1R level in mtDNA-depleted cells is ϳ4.5-fold higher than in control cells. Reverted cells, on the other hand, contained a IGF1R level closer to the control cells. CCCP treatment in control cells induced the IGF1R level by ϳ3-fold. The level of IGF1R protein in mtDNA-depleted cells was reduced by treatment with FK506 suggesting a possible role of Cn in increased expression or translocation of receptor protein. The autoradiogram in Fig. 3E shows that the receptor is autophosphorylated at low levels in control and reverted cells, which is not significantly affected by FK506. In mtDNA-depleted cells, however, the level of autophosphorylation was markedly higher (ϳ6-fold) and was sensitive to FK506. Consistent with the IGF1R protein levels in different cell types, real-time PCR data in Fig. 3F shows that the IGF1R mRNA level was increased 7-fold in mtDNA-depleted cells, and the level returned to near control cell level in reverted cells. The results collectively suggest that increased glucose uptake in mtDNA-depleted cells was probably due to increased expression and/or increased autophosphorylation of IGF1 receptor. Measurement of IGF1 in culture medium showed a  Fig. 1. Shown are the differences between control (basal) and mtDNA-depleted cells (**, p Յ 0.001) and between mtDNA-depleted cells with and without added genistein (*, p Ͻ 0.05). Shown also are the levels of inhibition in mtDNA-depleted cells with added wortmannin, LY294002, AG1024, and PPP compared with no inhibitor control (**, p Յ 0.001). B, basal plasma membrane levels of IR␤ in control, mtDNA-depleted, and reverted cells treated with or without of FK506 (10 nM, 2 h). Cell fractions (30 g each) were subjected to immunoblot analysis with antibody to IR␤ as described under "Experimental Procedures." The blot was co-developed with Na ϩ /K ϩ -ATPase antibody to assess equal protein loading levels. C, levels of basal IR␤ autophosphorylation. Autophosphorylation was carried out in plasma membranes of cells treated with or without FK506 as described under "Experimental Procedures." An autoradiogram of gels showing 32 P-labeled IR␤ is presented. D, immunoblot analysis of plasma membranes with IGF1R antibody following treatment with or without FK506. The blots were co-developed with antibody to Na ϩ /K ϩ -ATPase to assess loading levels. E, autoradiograms of gels showing 32 P-labeled IGF1R protein are presented. Levels of IGF1R autophosphorylation in cells treated with or without FK506. Details of treatment are as in C. F, IGF1R mRNA levels were quantified by real-time PCR in control, mtDNA-depleted, and reverted C2C12 cells as described under "Experimental Procedures." In some cases cells were treated with FK506 (10 nM) for 30 min before isolating total RNA. *, p Յ 0.05; **, p Յ 0.001. G, autophosphorylation of IGF1R (upper panel) and IR (lower panel) in cells grown in charcoal-treated serum as described under "Experimental Procedures." In B-E and G, values of relative band intensities presented in parentheses represent averages of two to three independent blots. small increase in IGF1 level in mtDNA-depleted cells (ϳ75 ng/ml) compared with control cells (ϳ50 ng/ml). To evaluate if the marked difference in the receptor autophosphorylation was due to extracellular or intracellular factors, we cultured the cells in medium containing charcoal pre-adsorbed FBS. As shown in Fig. 3G, the IGF1R autophosphorylation was ϳ3-fold higher in mtDNA-depleted cells and only marginally higher in control cells grown in charcoal-treated FBS. Similarly, the level of IR autophosphorylation was reduced marginally in mtDNA-depleted cells in comparison to control cells.
Possible Role of AMPK in Increased Glucose Uptake in mtDNA-depleted Cells-5ЈAMP-activated protein kinase is known to play an important role in insulin signaling, glucose uptake, and energy homeostasis (42). Recently AMPK has also been implicated to be involved in PGC1␣-mediated mitochondrial biogenesis (43). For these reasons we assessed the role of AMPK in the respiratory stress-mediated activation of Cn and glucose uptake in mtDNA-depleted C2C12 cells. We evaluated the effects of AICAR, a well known activator of AMPK (42), and FK506, an inhibitor of Cn, on the calcineurin activity in cell extracts. As shown in Fig. 4A AICAR had no effect on the Cn activity in mtDNA-depleted cells, whereas FK506 inhibited the activity. As seen from Fig. 4B, AICAR marginally increased glucose uptake by mtDNA-depleted cells. Similarly, AICAR was able to partially restore FK506-mediated inhibition of glucose uptake. These results suggest that, although AMPK is activated in response to respiratory stress, it does not have a role in increased Cn activity in mtDNA-depleted cells. Additionally, the contribution of AMPK to increased glucose uptake in mtDNA-depleted cells appears to be only marginal. These results also suggest that the two pathways may not be dependent on each other.

Activation of IGF1R and Glucose Uptake Is a General Phenomenon in Cells Subjected to Mitochondrial Respiratory
Stress-The generality of increased glucose uptake and elevated levels of GLUT 4 and IGF1R mRNA observed in mtDNAdepleted C2C12 cells was investigated using three other cell lines. For this purpose we chose mtDNA-depleted A549 cells (6) and H9C2 cardiomyocytes, both of which are transformed cells and also non-transformed 3T3 cells, all of which express GLUT 4 and are insulin-responsive. In the latter two cell types, CCCP treatment was used to induce mitochondrial stress. Fig.  5A shows that in all three cell lines mitochondrial stress either due to mtDNA depletion or CCCP treatment caused 2-to 3-fold increased 2-DOG uptake. As shown for mtDNA-depleted C2C12 cells, 2-DOG uptake in H9C2 cells was inhibited by FK506 (Fig. 5A). Although not shown, a similar pattern of inhibition was seen in 3T3 and A549 cells. Immunoblots in Fig.  5B show that IGF1R protein levels were 2-to 5-fold higher in all three cells either in response to mtDNA depletion or CCCP treatment. Similarly, GLUT 4 protein levels (Fig. 5C) were increased 2.0-to 2.5-fold in A549 and 3T3 cells in response to added CCCP. Consistent with increased protein levels, Fig. 5D shows that GLUT 4 and IGF1R mRNA levels were increased by ϳ2.5to 3.5-fold in H9C2 cells in response to CCCP treatment. The CCCP-mediated increase of both GLUT 4 and IGF1R mRNAs was sensitive to FK506 confirming the role of Cn. These results collectively show that activation of IGF1R, increased GLUT 4, and glucose uptake are general features of mitochondrial stress response in many cell types.
Distinctive Patterns of Metabolic Shifts in Response to Mitochondrial Stress-With a view to understanding the significance of IGF1R activation and increased glucose uptake, we  , 2 h). A, total cell lysates (1 g of protein) were assayed for calcineurin activity as described under "Experimental Procedures." B, 2-DOG uptake in mtDNA-depleted C2C12 cells was performed as described under "Experimental Procedures." *, p Յ 0.05; **, p Յ 0.001. FIGURE 5. Increased glucose uptake, IGF1R, and GLUT 4 levels in A549, H9C2, and 3T3 cells subjected to mitochondrial stress. mtDNA-depleted A549 cells were generated by ethidium bromide treatment as described before (1). Mitochondrial respiratory stress was induced in H9C2 cells and 3T3 cells by treatment with CCCP (25 mM) for 30 min. A, 2-DOG uptake by control, mtDNA-depleted A549 cells, control, and CCCP-treated H9C2 and 3T3 cells was carried out as described under "Experimental Procedures." **, p Յ 0.001. B and C, plasma membrane fractions (30 g each) were subjected to immunoblot analysis, and the blot was developed with antibody to GLUT 4, IGF1R␤, and IR␤, respectively. Antibody to Na ϩ /K ϩ -ATPase (1:1000 dilution) was used as a loading control. D, GLUT 4 and IGF1R mRNA levels were quantified by real-time PCR in control and CCCP-treated H9C2 cells as described under "Experimental Procedures." In some cases cells were treated with FK506 (10 nM) for 30 min before isolating total RNA. *, p Յ 0.05; **, p Յ 0.001. In B and C, the relative band intensities represent averages of two to three independent blots.
assayed the levels of the three irreversible and regulated steps of glycolysis. Table 1 shows that the hexokinase, phosphofructokinase, and pyruvate kinase activities were increased 5-to 28-fold in mtDNA-depleted C2C12 and A549 cells, except that the unit increase of PFK in mtDNA-depleted A549 cells was only marginal. The activities of these enzymes were also increased 2.5-to 4.5-fold in CCCPtreated 3T3 cells. The level of increase in CCCP-treated H9C2 cells was significantly lower. In particular, the PK activity was marginally reduced, which is not statistically significant. Although the reason for this difference remains unknown, it may be related to cell-specific differences in the inducibility of glycolytic pathways. Consistent with induced glycolytic pathways, results in Table 2 show that the levels of oligomycin-insensitive ATP synthesis is increased nearly 2-fold in cells subjected to mitochondrial genetic (mtDNAdepleted) and metabolic (CCCP-treated) stress. These results collectively show that mitochondrial respiratory stress induces a shift in cellular metabolic pattern.
Effects of Calcineurin A␣ mRNA Silencing on GLUT 4, IGF1R, and IR mRNA Levels-To confirm the role of increased Cn activity on increased steady-state GLUT 4 and IGF1R mRNA levels in cells subjected to mitochondrial stress, we used an siRNA approach to knock down CnA␣ mRNA in C2C12 cells. Fig. 6A shows that transient expression of siRNA reduced CnA␣ mRNA levels by ϳ60% in control cells and nearly 80% in mtDNA-depleted cells compared with mock transfections. Results in Fig. 6 (B and C) show that Cn mRNA silencing in control cells did not have a significant effect on either IGF1R or GLUT 4 mRNA levels. However, in mtDNA-depleted cells, Cn mRNA silencing had a profound effect, because both IGF1R and GLUT 4 mRNA levels were reduced by 75-85%. Further, Fig. 6D shows that Cn mRNA silencing had only a marginal effect on IR mRNA levels. These results provide confirmatory evidence that Cn and its target factors are involved in the regulation of IGF1R and GLUT 4 genes. Results also suggest that, in control cells, basal levels of expression of these two genes are not regulated by Cn.

TABLE 1 Mitochondrial respiratory stress increases activity of glycolytic enzymes
The values shown represent the mean Ϯ S.D. from three independent assays. Statistical analyses were done between control and depleted or control and CCCP-treated cells.

IGF1R Inactivation Induces Apoptosis and Inhibits
Invasive Property of mtDNA-depleted C2C12 Cells-Previously we showed that mitochondrial stress in otherwise non-tumorigenic C2C12 cells renders them highly invasive (2,6). Here we ascertained the role of IGF1R activation on increased glucose transport and invasive property of cells by mRNA silencing. Fig.  7A shows that transfection with siRNA constructs reduced IR and IGF1R mRNA contents in mtDNA-depleted C2C12 cells by 80 and 60%, respectively, of mock transfected cells (p Ͻ 0.001). As shown in Fig. 7B, in control cells IR mRNA silencing reduced 2-DOG uptake by ϳ50%, but IGF1R mRNA silencing had only a marginal effect. In mtDNA-depleted cells, IR mRNA silencing reduced 2-DOG uptake only marginally (ϳ20%), whereas IGF1R mRNA silencing was more effective by inhibiting 2-DOG uptake by ϳ60% (p Ͻ 0.001). This level of inhibition was consistent with the level of mRNA silencing (see Fig. 7A). As seen from Fig. 7C, both IR and IGF1R mRNA silencing in control cells marginally affected GLUT 4 mRNA levels. In mtDNAdepleted cells, IR mRNA silencing reduced GLUT 4 mRNA levels by ϳ20%. IGF1R mRNA silencing in mtDNA-depleted cells, however, had a profound effect in that GLUT 4 mRNA levels were reduced by ϳ75%. These results show that IGF1R activation as part of mitochondrial respiratory stress plays a direct role in increased GLUT 4 mRNA and increased glucose uptake.
Because IGF1R activation is associated with tumorigenic property of cells, we tested the effects of mRNA silencing on in vitro invasive property and rate of apoptosis. Fig. 7D shows that IR mRNA silencing had no significant effect on the ability of mtDNA-depleted cells to invade through the Matrigel membrane, whereas IGF1R mRNA silencing reduced the number of invading cells by ϳ85%. An opposite effect was observed on the propensity of cells to undergo apoptosis. The number of cells undergoing apoptosis was measured using the Nexin assay, which utilizes annexin V-PE to detect phosphatidylserine on the plasma membrane signifying apoptotic cells. The Nexin assay in Fig.  7E showed only a marginal increase in apoptosis by IR mRNA silencing, whereas IGF1R mRNA silencing caused a vastly increased 50% apoptosis.
We also assessed the effects of known inhibitors of IGF1R on the invasive property and rate of apoptosis on control, mtDNAdepleted, and reverted C2C12 cells. The cell distribution pattern in Fig. 8A and the quantitation in Fig. 8B show that control, depleted, and reverted cells contained only 8 -10% apoptotic cells. Treatment with AG1024 and PPP, specific inhibitors of IGF1R, however induced apoptosis in mtDNA-depleted cells to ϳ28% and 40%, respectively (p Ͻ 0.001), whereas these agents had no significant effect on control cells. In reverted cells the number of cells undergoing apoptosis in response to added AG1024 and PPP increased to 15-17% (p Ͻ 0.05), further supporting the possible role of mitochondrial stress signaling in altered metabolic state of cells. Genistein had only marginal effects on all three cell types. These results therefore indicate that increased glucose utilization induced by mitochondrial stress functions as a survival factor in mtDNA-depleted cells. Fig. 8C shows the patterns of in vitro invasion in a Matrigel invasion assay system. It is seen that PPP, a specific inhibitor of IGF1R had no significant effect on the rate of invasion of control cells. As reported before, mtDNA-depleted cells showed vastly increased invasion, which was effectively inhibited by PPP. These results are consistent with the view that activation of the IGF1R pathway as part of mitochondrial stress signaling provides survival advantage to mtDNA-depleted cells as they become more resistant to apoptosis and gain the invasive ability.

DISCUSSION
The IR and IGF1R family of receptors bears both structural and functional similarities (44), and IGF1R can functionally replace IR in inducing glucose uptake and utilization in many cell types. Although IGF1 is the preferred ligand, insulin also binds to and activates IGF1R, at a lower efficiency than IR (45). Both IR-and IGF1R-mediated signaling involves autophosphorylation of tandem activation loop Tyr residues at the kinase domain of the receptor facing the cytoplasmic side, which is critical for the activation of signaling cascades through cytosolic IRS1 or IRS2 kinase systems (44). In both cases the signaling is propagated through activation of phosphatidylinositol 3-kinase, Akt, and protein kinase C, ultimately leading to transcriptional activation of many genes involved in glucose uptake and utilization. Induction and activation of IGF1R pathways through a number of external and internal factors in many tumor cell models have been described (46,47). In this study we describe a novel mechanism of mitochondrial respiratory stress-induced switching from IR to IGF1R pathway leading to induced glucose uptake and resistance to apoptosis in C2C12 rhabdomyocytes. Mitochondrial stress, due to reduced mtDNA content and/or treatment with respiratory inhibitors (2), is the major contributing factor in this switch, because restoration of mtDNA content and thus the ⌬⌿m effectively reverses the level of glucose uptake and the entire signaling cascade.
Our results show that both GLUT 4 and IGF1R genes are overexpressed and/or activated in mtDNA-depleted cells, and the levels are reduced to control cell level in reverted cells suggesting the role of mitochondrial stress in these changes. Three separate lines of experiments provide supporting evidence for this possibility: 1) The mitochondrial ionophore, CCCP, which is known to induce mitochondrial stress signaling, also induced the steady-state levels of these two proteins and glucose uptake in four different cell lines. 2) FK506, an inhibitor of Cn, not only inhibited stress-induced glucose uptake but also reduced the steady-state levels of IGF1R and GLUT 4 proteins as well as mRNAs. 3) Further confirmatory evidence on the role of Cn comes from the mRNA silencing experiments, which show that knockdown of CnA␣ mRNA by ϳ80% also caused reduced steady-state levels of GLUT 4 and IGF1R mRNAs by a similar magnitude in mtDNA-depleted cells.
Results on overexpression of IGF1R in FK506-treated cells suggest that IGF1R plays a direct role in GLUT4 mRNA expression. Thus, Cn appears to exert a multimodal effect on GLUT 4 gene expression, namely through activation of IGF1R and through activation of other Cn target factors. Unpublished studies in our laboratory show that GLUT 4 expression is transcriptionally up-regulated by Cn-activated NFB, C/EBP␦, CREB, and NFAT, in addition to IGF1R. It is likely that yet unknown IGF1R-activated factors cooperatively interact with other Cn-activated factors in GLUT 4 expression.
The steady-state levels of both IR and IGF1R proteins in the plasma membrane compartment were increased in mtDNAdepleted cells, although only the latter remained functional probably because of selective inhibition of autophosphorylation of IR (Fig. 3). A possible receptor switch and the role of IGF1R pathway are further confirmed by results showing that IGF1R mRNA silencing selectively reduced the glucose uptake, reduced the cell invasion through Matrigel membrane, and increased the rate of apoptosis. IR mRNA silencing affected all these cellular parameters, although at a vastly reduced level. These results provide firm evidence for the Cn-mediated recep- tor switch for the observed invasive behavior and resistance to apoptosis in mtDNA-depleted cells. The molecular basis of preferentially higher autophosphorylation of IGF1R currently remains unclear. Preliminary results, however, suggest that increased PTP1B activity and possible post-translational modification of this enzyme in mtDNA-depleted cells may be responsible for this selectivity.
Consistent with the observed receptor switch, cells subjected to mitochondrial stress also show a distinct metabolic shift. An increasing number of studies suggest that mitochondrial mutations and/or dysfunction are closely associated with a wide variety of neuromuscular, retinal degenerative disease, diabetes, ischemic injury, cancer, and aging (19,48,49). In a large majority of these cases, the heteroplasmic nature of mutations with an age-dependent increase in the ratio of mutated to wild-type mtDNA is implicated as an important factor. In this regard, partial mtDNA depletion used in this study reflects the steady decline of functional mtDNA levels in many diseases (50). The respiratory stress-induced metabolic shift with induced glycolysis is not restricted to either mtDNA depletion or C2C12 cells. We show that mitochondrial stress induced either by mtDNA depletion in A549 cells or disruption of ⌬⌿m by CCCP treatment in H9C2 cardiomyocytes and 3T3 fibroblasts causes increased IGF1R activity, increased GLUT 4 mRNA levels, and induced glucose uptake. The relative increase in activities of different glycolytic enzymes in the four cell types varied markedly suggesting cell-specific effects. Further, although the precise molecular mechanism of this increase remains unclear, unpublished results with hexokinase promoter suggests transcriptional activation as the main factor. Additionally, the level of oligomycin-insensitive ATP synthesis, which probably reflects the contribution of the glycolytic pathway, is increased Ͼ2-fold in all cases, confirming the metabolic switch. Based on these observations we propose that stress-mediated activation of the IGF1R pathway is closely linked to the higher resistance to apoptosis and invasive properties of mtDNA-depleted C2C12 and A549 cells.
As proposed by Warburg (51), the capacity to sustain high rates of glycolysis under aerobic conditions appears to be an important feature of rapidly growing tumors. The Warburg effect has been observed in many solid tumors, and accordingly increased expression of genes encoding glucose transporter genes and glycolytic enzymes has been suggested as the basis for an altered metabolic profile of tumor cells (52). Other studies show that the high metabolic activity of tumor cells is associated with activation of Akt, an Ser/Thr kinase (53,54). Akt has been shown to induce anabolic metabolism and suppress apoptosis by multiple mechanisms (55), thus providing a molecular basis for an altered metabolic state of proliferating tumors. Warburg also predicted that impaired mitochondrial oxidative function may be a factor in increased glucose utilization of tumor cells. Our results show that mtDNA depletion and other types of mitochondrial respiratory dysfunction induce stress signaling, which culminates in altered glucose uptake and metabolism. In this regard our results provide a potential mechanistic explanation for the Warburg hypothesis.
Cn is known to play critical roles in many physiological processes, including T-cell differentiation, myogenesis, switch of muscle fiber types, and muscle contraction/glucose homeostasis (56 -58). Here we show that mitochondrial respiratory stress-activated Cn plays a direct role in the altered metabolic property of tumor cells through an induced IGF1R pathway and increased expression of GLUT 4. In summary, we describe a novel Cn-mediated IGF1R activation pathway that is likely to play an important role in tumor progression and tumor invasion.