NADPH Oxidase and Hydrogen Peroxide Mediate Insulin-induced Calcium Increase in Skeletal Muscle Cells*

Skeletal muscle is one of the main physiological targets of insulin, a hormone that triggers a complex signaling cascade and that enhances the production of reactive oxygen species (ROS) in different cell types. ROS, currently considered second messengers, produce redox modifications in proteins such as ion channels that induce changes in their functional properties. In myotubes, insulin also enhances calcium release from intracellular stores. In this work, we studied in myotubes whether insulin stimulated ROS production and investigated the mechanisms underlying the insulin-dependent calcium increase: in particular, whether the late phase of the Ca2+ increase induced by insulin required ROS. We found that insulin stimulated ROS production, as detected with the probe 2′,7′-dichlorofluorescein diacetate (CM-H2DCFDA). We used the translocation of p47phox from the cytoplasm to the plasma membrane as a marker of the activation of NADPH oxidase. Insulin-stimulated ROS generation was suppressed by the NADPH oxidase inhibitor apocynin and by small interfering RNA against p47phox, a regulatory NADPH oxidase subunit. Additionally, both protein kinase C and phosphatidylinositol 3-kinase are presumably involved in insulin-induced ROS generation because bisindolylmaleimide, a nonspecific protein kinase C inhibitor, and LY290042, an inhibitor of phosphatidylinositol 3-kinase, inhibited this increase. Bisindolylmaleimide, LY290042, apocynin, small interfering RNA against p47phox, and two drugs that interfere with inositol 1,4,5-trisphosphate-mediated Ca2+ release, xestospongin C and U73122, inhibited the intracellular Ca2+ increase produced by insulin. These combined results strongly suggest that insulin induces ROS generation trough NADPH activation and that this ROS increase is required for the intracellular Ca2+ rise mediated by inositol 1,4,5-trisphosphate receptors.

A transient intracellular Ca 2ϩ increase is a key component of the excitation-coupling mechanism in skeletal muscle cells. Intracellular Ca 2ϩ can also increase in response to stimuli other than membrane potential depolarization, including hormones such as insulin. We have previously reported that in myotubes, the addition of insulin produces a fast intracellular Ca 2ϩ concentration transient, which requires external Ca 2ϩ and is inhibited by the L-type Ca 2ϩ channel blocker nifedipine and by ryanodine (1). Other reports show that the Ca 2ϩ increase evoked by insulin in skeletal muscle fibers depends on Ca 2ϩ influx and is related to Glut-4 translocation (2). As steep changes in insulin concentration are unlikely to occur physiologically, we decided to explore changes in intracellular Ca 2ϩ after longer exposure to insulin. We further investigated whether ROS 3 play a role in this process. It is known that ROS modulate the activity of Ca 2ϩ release channels (3), and in particular, the activity of skeletal muscle ryanodine receptors is regulated by NADPH oxidase-dependent redox modifications (4). On the other hand, in target tissues, insulin is known to generate ROS, which participate in signaling processes triggered by the hormone (5).
There are many intracellular sources of ROS in mammalian cells, such as mitochondria, xanthine oxidase, and NADPH oxidase (6,7). We have reported the presence of NADPH oxidase subunits in myotubes (1) and in adult skeletal muscle transverse tubules (4). Recently, the NADPH oxidase has been involved in modifications of tyrosine phosphates that shut down signals evoked by insulin in adipocytes (8). The phagocytic NADPH oxidase is composed of five subunits: two catalytic subunits located in the plasma membrane (p22 phox and gp91 phox ) and three cytoplasmic subunits (p40 phox , p47 phox , and p67 phox ), plus the small GTP-binding protein rac-1. Upon activation, these subunits translocate to the plasma membrane to form the active enzyme (9). The regulatory p47 phox subunit has an autoinhibitory domain; phosphorylation of p47 phox in serine residues causes its translocation to the plasma membrane to form the active enzymatic complex. Activation of p47 phox depends on several serine/threonine and tyrosine kinases such as protein kinase C (PKC) (10,11), c-Src (12), and phosphatidylinositol 3-kinase (PI3K) (13).
By inducing phosphorylation of its receptor on tyrosine residues, insulin initiates a well described signal transduction cascade (14), which involves PI3K activation. Insulin can also activate phospholipase C and generate IP 3 and diacylglycerol (DAG). These second messengers can induce Ca 2ϩ release and PKC activation, respectively. The main goal of this work was to investigate the possible correlation between the increase in Ca 2ϩ and ROS induced by insulin. We found that both PI3K and PKC are needed to activate NADPH oxidase to produce ROS and that ROS are involved in the delayed Ca 2ϩ increase induced by insulin, which involves IP 3 receptor (IP 3 R) activation by ROS.
The plasmids used encode the fusion protein between the pleckstrin homology (PH) domain of Bruton's tyrosine kinase (BTK) and the enhanced green fluorescent protein (GFP) ((PH)BTK-GFP) or the (PHmut)BTK-GFP (punctual mutation R28C of the PH domain). Both plasmids were kindly provided by Dr. Tamas Balla (National Institutes of Health, Bethesda, MD). A plasmid that encodes for HyPer protein was acquired from Evrogen Joint Stock Company, Moscow, Russia.
Cell Culture-Primary cultured cells were obtained from rat neonatal hind limbs. The muscle tissue was mechanically dispersed and then incubated under mild agitation with 10% (w/v) collagenase for 15 min at 37°C. The suspension was filtered through a Nytex (Sartorius, Göttingen, Germany) membrane and spun down at low speed, preplating was used to partially eliminate fibroblasts, and finally, cells were plated onto dishes (60 mm) with coverslips at a density of 0.6 ϫ 10 6 /dish. The culture medium used was Dulbecco's modified Eagle's medium/F-12, 10% bovine serum, 2.5% fetal calf serum, 100 mg/liter penicillin, 50 mg/liter streptomycin, and 2.5 mg/liter amphotericin B. To eliminate remaining fibroblasts, on the third day of culture, 10 M cytosine arabinoside was added for 24 h. The medium was then replaced with serumfree medium. 6 -8-day-old cultures were used.
Ca 2ϩ Measurement-Cytoplasmic Ca 2ϩ images were obtained using inverted confocal microscopy (Carl Zeiss Pascal 5 LSM) from single, non-spontaneously contracting myotubes preloaded with fluo-3AM. Myotubes were washed with Krebs buffer (10 mM Hepes-Na, pH 7.4, 145 mM NaCl, 5 mM KCl, 2.6 mM CaCl 2 , 1 mM MgCl 2 , 5.6 mM glucose,) and loaded with 5.4 M fluo-3AM (added from a stock solution in 20% pluronic acid-DMSO) for 30 min at room temperature. Cell-containing coverslips were mounted in a 1-ml capacity plastic chamber and placed in the microscope for fluorescence measurements (band pass 505-530 nm) after excitation with a 488 nm argon laser.
Determination of ROS Production-ROS generation was determined in skeletal muscle cells using CM-H 2 DCFA. Myotubes were cultured on glass coverslips and incubated with 5 M CM-H 2 DCFA for 15 min at 37°C. Cells on coverslips were washed with PBS and placed at the bottom of an incubation chamber, which was transferred to the confocal microscope. CM-H 2 DCFA fluorescence was detected using excitation/ emission wavelengths exc / em ϭ 488/505-530 nm. In all measurements a control of the effect of laser excitation alone was performed. Noise in the images was removed by the Image J software (National Institutes of Health).
Inmunofluorescence-Cells were placed on coverslips and fixed with methanol at Ϫ20°C for 15 min. The blockade was performed in 1% bovine serum albumin for 60 min, and the primary antibodies were incubated overnight at 4°C. Cells were washed and incubated with secondary antibody during 2 h at room temperature. Coverslips were mounted in VECTASH-IELD (Vector Laboratories, Inc.) for confocal microscopy, and representative images were acquired. Negative controls used only secondary antibodies.
Image Capture and Quantification of Fluorescence-Confocal image stacks were captured with a Zeiss LSM-5, Pascal 5 Axiovert 200 microscope, using LSM 5 3.2 image capture, analysis software, and a Plan-Apochromat ϫ63/1.4 oil differential interference contrast objective. Two-channel fluorescent image stacks (intensity I(x,y,z), voxel size ⌬x/⌬y/⌬z ϭ 50/50/ 300 nm) were recorded in the multitrack mode. Channel-1 (Alexa Fluor 488) was used with excitation/emission wavelengths exc / em ϭ 488/505-530 nm, and channel-2 (Alexa Fluor 633) was used with exc / em ϭ 635/Ͼ650 nm. We made sure that I (x,y,z) did not saturate and that the image background was slightly above zero by carefully adjusting the laser power, the detector gain, and the detector offset.
Image stacks were deconvoluted with Huygens Scripting (Scientific Volume Imaging, Hilversum, Netherlands). All following image processing routines were developed in our laboratory on the basis of IDL (Interactive Data Language, ITT, Boulder, CO). To determine p47 phox in the plasma membrane and the cytoplasm of myotubes, segmentations were performed to define different regions of interest (ROIs). First, the cross-section of myotubes was segmented by an intensity threshold in the green fluorescence channel (see Fig. 3D, center). Remaining holes inside the cells and artifacts outside the cells were filled or removed by morphological filters. Second, we used a border detection filter to identify the plasma membrane with a constant thickness of 400 nm (see Fig. 3D, right). The nuclear region was defined manually using the bright field image and excluded from the segmented cells. The fluorescence intensities of the cytoplasm and the plasma membrane were determined in the ROIs and divided by the respective surfaces (I/m 2 ). For all experiments, the protocols remained constant, and the quality of the segmentation was controlled interactively by overlaying the original fluorescent images with the segmented ROIs.
Co-localization Analysis-A reliable segmentation of cellular structures containing p47 phox or ␣-actinin could not be obtained by a simple threshold value in fluorescence intensity because background intensities vary within the cell thickness across the image. We solved this problem by applying gradient filters and selecting threshold values in the gradient histograms, which resulted in a homogeneous definition of p47 phox or ␣-actinin signals (see Fig. 3A).
For the quantification of co-localization between p47 phox and ␣-actinin (Fig. 3A, left, ROI1/green regions and ROI2/red regions), we calculated the Manders co-localization coefficients M 1 and M 2 (15). M 1 and M 2 sum up the contribution of the respective fluorescence intensities in the co-localizing region I Ch1/2 (ROI1 പ ROI2), and we divided the number by the sum of the fluorescence intensities inside of the segmented regions I Ch1 (ROI1) or I Ch2 (ROI2). M 1 and M 2 can be interpreted directly as the amount of co-localizing p47 phox / ␣-actinin labeled structures (Fig. 3A, lower, yellow regions) with respect to the total amount of segmented p47 phox /␣actinin signals (Fig. 3A, lower left panel, yellow region plus green/red region). To reveal the significance of the co-localization coefficients, we implemented a defined displacement algorithm (16). Defined displacement algorithm shifts one fluorescent channel with respect to the second channel in a radial manner and calculates M 1 and M 2 successively for each displacement. When the radial displacement is larger than the size of the segmented ROIs, the calculated Manders coefficients quantify random scenarios of fluorescent structures inside the cellular borders, which were used to calculate the M 1 /M 2 minima values in Fig. 3B.
Data Analysis-All experiments were performed in at least three different cultures, and at least three cells per culture were measured in each case. Results are expressed as the mean Ϯ S.E. Significance was evaluated using Student's t test for paired data and one-way analysis of variance followed by Dunnett's post hoc test for multiple column comparison to control column; p Ͻ 0.05 was considered significant.

RESULTS
Insulin Induced ROS Generation through NADPH Oxidase Stimulation-Cultured myotubes preincubated with CM-H 2 DCFDA were stimulated with insulin to investigate insulin-induced ROS generation. Fig. 1A shows a representative experiment where images were acquired at different times after the addition of 50 nM insulin. This concentration was chosen because of its clear-cut effects on calcium transient generation (see below). The time dependence of the mean fluorescence increase mediated by the hormone is presented in Fig. 1B. A significant increase in intracellular ROS was evident 5 min after insulin stimulation. Cells incubated with apocynin, a specific inhibitor of p47 phox -dependent NADPH oxidase (EC 50 ϭ 10 M (7)) displayed significantly reduced ROS production, and this effect was evident 5 min after insulin addition (Fig. 1B). To ascertain whether H 2 O 2 is one of the ROS involved in this response, we used the HyPer protein, which is a genetically encoded fluorescent sensor for the specific detection of intracellular H 2 O 2 (17). We measured the increase of HyPer fluorescence in myotubes after insulin stimulation. The increase in HyPer fluorescence was homogeneous in the cytoplasm of the myotube (Fig.  1C) and increased steadily in time, reaching a plateau of about 25-fold increase at 300 s; apocynin prevented this increase and even produced a small decrease of the initial endogenous levels of H 2 O 2 (Fig. 1D).
Role of the p47 phox Subunit in ROS Generated by Insulin-The results obtained with apocynin strongly suggest that a  JANUARY 23, 2009 • VOLUME 284 • NUMBER 4 p47 phox -dependent NADPH oxidase is involved in ROS generation. To test this hypothesis, we measured insulin-dependent CM-H 2 DCFDA fluorescence in cells transfected with siRNA against p47 phox , using siRNA with a random sequence as control ( Fig. 2A). To identify the transfected cells, they were cotransfected with a plasmid encoding DsRed protein. Red-labeled cells, transfected with the random sequence, responded with a fluorescence increase 10 min after insulin stimulation ( Fig. 2A). In contrast, red-labeled cells transfected with siRNA against p47 phox did not respond to insulin, and the fluorescence signal was similar to that of cells expressing random siRNA in the basal state (Fig. 2, A and B). To evaluate whether p47 phox expression was decreased in myotubes transfected with siRNA, we performed an immunofluorescence study against the p47 phox subunit. The p47 phox subunit was not detected in cells transfected with siRNA p47 phox (Fig. 2C).

Insulin Stimulates NADPH Oxidase in Myotubes
To study the insulin-dependent activation of NADPH oxidase, we determined by immunofluorescence the translocation of p47 phox . In basal conditions, p47 phox showed a particular distribution in myotubes; the label was present in striations and displayed a significant co-localization with ␣-actinin, a Z line marker protein (Fig. 3, A and B). In the first minute after insulin addition, some p47 phox label already appeared near the membrane (Fig. 3C, arrow). 15 min after stimulation, most of the label appeared in the periphery of myotubes, probably representing the location of the enzyme near the plasma membrane. At 60 min, the distribution of the fluorescence signal showed a pattern similar to the basal condition prior to insulin addition (Fig.  3C). We measured fluorescence intensity in regions of the cytoplasm and near the plasma membrane (Fig. 3D). We defined the region near the plasma membrane with a constant diameter of 400 nm and the region inside the cytoplasm excluding cell nuclei (Fig. 3D, right  and center). In the region near the plasma membrane, a significant fluorescence increase was detected after insulin stimulation (Fig. 3E). This membrane location was seen only in stimulated sections, never in basal conditions (Fig. 3C shows different times). Cytoplasmic fluorescence decreases upon stimulation (Fig. 3E) concomitant to the increase in near membrane fluorescence; as both myotubes (basal and stimulated) were measured in identical conditions, these observed differences suggest p47 phox translocation to the membrane region.
Role of PI3K in ROS Generation-PI3K activation is a key event in insulin-dependent signal transduction, so we studied whether ROS generation depends on PI3K activity. Using a plasmid that encodes a fusion protein between the PH domain of BTK and the enhanced GFP ((PH)-BTK-GFP), we monitored the activation of PI3K as changes in probe fluorescence, considering that when PI3K is activated, it can recruit proteins with the PH domain to membranes. Cells transfected with (PH)-BTK-GFP showed basal fluorescence in nucleus and cytoplasm; 10 s after insulin stimulation, a distinct fluorescence increase appeared near the plasma membrane and in regions that may correspond to transverse tubules. This particular fluorescence increase disappeared after 90 s (Fig. 4A). These results indicate that activation of PI3K is an early event of the signal transduction cascade triggered by insulin that occurs prior to ROS generation. We next measured ROS generation induced by 50 nM insulin 15 min after stimulation in controls or in cells preincubated with different drugs. ROS production was completely prevented by genistein, an inhibitor of insulin receptor (Fig. 4B). LY290042, an inhibitor of the PI3K pathway, also produced a complete inhibition of fluorescence increase with respect to the control stimulated condition (Fig. 4B), suggesting that PI3K is a key player in ROS generation triggered by insulin. Previous reports suggest that PKB can phosphorylate p47 phox . To probe the role of PKB in ROS generation, we used a PKB inhibitor, PKBI (Fig. 4C, PKBI). We measured ROS generation 30 min after insulin stimulation and found similar values for ROS production both in the presence of inhibitor and in control conditions (Fig. 4C). Accordingly, we can conclude that insulin-stimulated ROS generation is dependent on PI3K but  (upper panel, arrow). A.U., arbitrary units. B, mean fluorescence in myotubes loaded with CM-H 2 DCFDA. Cells were preincubated with LY290042 (50 M, Ly) for 30 min and were then stimulated with insulin. Fluorescence was measured 15 min after insulin stimulation (n ϭ 5-9). C, myotubes were incubated with an inhibitor of PKB (PKBI), and then fluorescence was measured 15 min after insulin stimulation (n ϭ 8 -10). D, myotubes, control and incubated with LY290042, were stimulated with insulin (Ins) and then fixed with methanol for 15 min to label with specific antibody against p47 phox (representative images, n ϭ 5). E, the relative number of myotubes in which p47 phox translocated from cytoplasm to membrane. Values are expressed as the percentage of cells in which translocation was evident (n ϭ 15 from three independent cultures, p Ͻ 0.05 (*)). not on PKB. To evaluate the participation of PI3K in the activation of NADPH oxidase, we studied the insulin-dependent translocation of p47 phox to the plasma membrane in the presence of LY290042. As shown above, p47 phox is present in the cytoplasm in basal conditions; after 10 min of insulin stimulation in the presence of LY290042, most of the label still remained in striations (Fig. 4C). There was some translocation in the presence of the inhibitor but clearly much less than in the control conditions illustrated in Fig. 3C.
Role of PKC in ROS Generation-To study the possible role of PKC in insulin-dependent ROS generation, we preincubated myotubes with BIM, a general inhibitor of PKC. Incubation with BIM inhibited insulin-induced ROS generation (Fig. 5A). In addition, 0.5 M BIM inhibited the increase in HyPer fluorescence induced by insulin (Fig. 5C) and also prevented the translocation of p47 phox to the plasma membrane induced by insulin (Fig. 5B).
Insulin Increased the Intracellular Ca 2ϩ Concentration in Cultured Myotubes-We had some evidence for delayed Ca 2ϩ oscillations occurring after the fast Ca 2ϩ transient induced by insulin in myotubes (1); these oscillations were seen at insulin concentrations as low as 1-10 nM but were consistently larger (80-fold increase in fluo-3 fluorescence) at 50 nM insulin. Accordingly, in this report, we performed a detailed study of these delayed Ca 2ϩ signals in cultured myotubes using the fluorescence probe fluo-3AM. The insulin-dependent Ca 2ϩ signals studied in this work are independent of the fast Ca 2ϩ transients that require extracellular Ca 2ϩ and functional ryanodine receptors (1). Thus, in conditions in which the fast transient is absent (i.e. in the absence of external Ca 2ϩ ), the delayed signal remained unchanged, whereas inhibitors of the slow Ca 2ϩ signal (see below) did not affect the fast Ca 2ϩ signal (for example, Fig. 6C). We investigated the possible role of NADPH oxidasegenerated ROS on the slower Ca 2ϩ signals produced by insulin. To that aim, we measured the effect of both diphenyliodonium (Fig. 6A, DPI) and apocynin on insulin-evoked signals. Preincubation of myotubes with both inhibitors abolished the insulindependent intracellular Ca 2ϩ increases (Fig. 6A). On the same line, we studied the role of NADPH oxidase using siRNA against p47 phox . Fig. 6B shows a representative response of transfected myotubes in which the Ca 2ϩ increase produced by insulin was reduced significantly in myotubes transfected with siRNA against p47 phox . In addition, the PKC inhibitor BIM decreased significantly the insulin-dependent Ca 2ϩ increase, whereas the PI3K inhibitors wortmannin and LY290042 completely prevented this response (Fig. 6C). To test the involvement of IP 3 receptors on Ca 2ϩ signals, we preincubated the myotubes with two inhibitors of the IP 3 pathway, U73122 and xestospongin C; the insulin-stimulated Ca 2ϩ increase was abolished in both conditions (Fig. 6D).

DISCUSSION
In this work, we provide evidence favoring a role for insulin in ROS generation due to NADPH oxidase activation in skeletal muscle cells. We show, in addition, that enhanced ROS generation forms part of the signaling mechanisms whereby insulin increases intracellular Ca 2ϩ concentration. The fast increase of intracellular ROS produced by insulin was detected using both CM-H 2 DCFDA as fluorescence probe and a molecular recombinant protein that in vitro displays half-maximal increase in fluorescence at 10 M H 2 O 2 . CM-H 2 DCFDA is a probe susceptible to oxidation by H 2 O 2 and other ROS and nitric oxidederived species, but transfection of myotubes with HyPer plasmid provided us with a specific tool to ascertain that 50 nM insulin produced H 2 O 2 . An increase in intracellular peroxide concentration was detected with both probes as early as 1 min after insulin addition.
In various cell types, the NADPH oxidase has been described as the intracellular source of H 2 O 2 (9). This enzyme has defensive functions in phagocytic cells, where it catalyzes the oneelectron reduction of oxygen to release superoxide anion that rapidly dismutates to H 2 O 2, a microbicide agent. However, in non-phagocytic cells, H 2 O 2 is considered a second messenger due to its capacity to produce reversible post-translational oxidative modification of proteins (18). We have reported previously that myotubes possess NADPH oxidase activity that is stimulated by membrane depolarization (1). Here, we provide evidence indicating that insulin also stimulates a p47 phox -dependent NADPH oxidase in skeletal muscle cells, as indicated by our findings that insulin-stimulated ROS generation was inhibited in myotubes preincubated with apocynin or transfected with an siRNA against p47 phox . In addition, our current results, showing p47 phox translocation from cytoplasmic striations to the plasma membrane from the first minute after stimulation with insulin ( Fig. 3C), provide further support for NADPH oxidase activation by insulin.
Ca 2ϩ signals mediate a variety of physiological processes in skeletal muscle cells; in particular, muscle depolarization elicits fast Ca 2ϩ signals that initiate muscle contraction and slower signals that regulate gene expression (19). Insulin-dependent Ca 2ϩ influx has been implicated in glucose uptake in adult muscle fibers (2); in that case, a delayed effect (after 20 min) of insulin that required external Ca 2ϩ was reported. A direct link between NADPH-mediated ROS production and glucose uptake seems unlikely because a recent report showed no inhibition of glucose uptake by apocynin in insulin-stimulated skeletal muscle cell lines and primary cultures (20). Insulin has been involved in the regulation of the expression of a number of skeletal muscle genes related to carbohydrate metabolism and protein synthesis (21). In skeletal muscle cells, Ca 2ϩ release from IP 3 -dependent intracellular stores mediates the regulation of signaling cascades, leading to expression or repression of both early and late genes (22)(23)(24). IP 3 receptors are highly dependent on its regulators and modulators (25). A redox sensor protein has been shown to regulate type 1 IP 3 R (26), redox-sensitive free thiol groups have been identified in the IP 3 R molecule (27), and direct redox modifications of isolated IP 3 R channels have been described (28). Accordingly, the Ca 2ϩ signals elicited by insulin, which require NADPH oxidase and IP 3 receptor activation, may also be  Proposed mechanism for insulin-induced Ca 2؉ increase. Insulin binds to its receptor in the plasma membrane, and binding promotes phosphorylation of its ␤ subunit in tyrosine residues, recruiting IRS-1; binding of PI3K to IRS-1 activates inositol 1,4,5-trisphosphate (PIP 3 ) production, which stimulates phospholipase C ␥ (PLC␥) in the membrane to produce IP 3 and DAG. DAG activates PKC␦, and this enzyme phosphorylates p47 phox , allowing this subunit to migrate to the membrane to form the activated NADPH oxidase complex; the resulting H 2 O 2 modifies IP 3 receptors and favors IP 3 -induced Ca 2ϩ release. PIP 2 , phosphatidylinositol 4,5-bisphosphate; pTyr, phosphorylated tyrosine.