Insulin signaling includes generation of low levels of H2O2; however, its origin and contribution to insulin-stimulated glucose transport are unknown. We tested the impact of H2O2 on insulin-dependent glucose transport and GLUT4 translocation in skeletal muscle cells. H2O2 increased the translocation of GLUT4 with an exofacial Myc-epitope tag between the first and second transmembrane domains (GLUT4myc), an effect additive to that of insulin. The anti-oxidants N-acetyl L-cysteine and Trolox, the p47phox–NOX2 NADPH oxidase inhibitory peptide gp91-ds-tat or p47phox knockdown each reduced insulin-dependent GLUT4myc translocation. Importantly, gp91-ds-tat suppressed insulin-dependent H2O2 production. A ryanodine receptor (RyR) channel agonist stimulated GLUT4myc translocation and insulin stimulated RyR1-mediated Ca2+ release by promoting RyR1 S-glutathionylation. This pathway acts in parallel to insulin-mediated stimulation of inositol-1,4,5-trisphosphate (IP3)-activated Ca2+ channels, in response to activation of phosphatidylinositol 3-kinase and its downstream target phospholipase C, resulting in Ca2+ transfer to the mitochondria. An inhibitor of IP3 receptors, Xestospongin B, reduced both insulin-dependent IP3 production and GLUT4myc translocation. We propose that, in addition to the canonical α,β phosphatidylinositol 3-kinase to Akt pathway, insulin engages both RyR-mediated Ca2+ release and IP3-receptor-mediated mitochondrial Ca2+ uptake, and that these signals jointly stimulate glucose uptake.
Reactive oxygen species (ROS) are relevant second messengers in physiological and pathological cellular processes ranging from the immune response to cell growth and differentiation (Bashan et al., 2009; Horie et al., 2008; Paulsen and Carroll, 2010). Cellular ROS include singlet oxygen, superoxide anion, hydrogen peroxide (H2O2), the hydroxyl radical and other partially reduced oxygen derivatives. H2O2 is the most suitable to act as a second messenger because of its high specificity, stability and regulation (Forman et al., 2010). In particular, insulin-dependent H2O2 production contributes to Akt activation and glucose uptake in 3T3-L1 adipocytes (Mahadev et al., 2001).
The ROS-generating NOX2 enzyme is a multi-protein complex formed by three cytoplasmic subunits (p47phox, p67phox and p40phox), a G protein (Rac1 or Rac2), and two membrane-bound subunits (p22phox and NOX2, formerly known as gp91phox) (Bedard and Krause, 2007). Adult muscle fibers express NOX2 (Javeshghani et al., 2002) in transverse tubules (T-tubules) (Hidalgo et al., 2006), suggesting that in this tissue NOX2-mediated ROS generation and the ensuing redox effects might be highly localized. We have previously shown that NOX2 generates ROS in skeletal muscle myotubes in response to insulin (Espinosa et al., 2009). A rise in cellular ROS promotes redox modifications of skeletal muscle proteins (Hidalgo et al., 2004) and a global ‘phosphatase tone’ (Wright et al., 2009) in both skeletal and cardiac muscle. The thiol (-SH) group of cysteine residues undergoes covalent reactions with oxidants, which produce modifications, such as S-glutathionylation, that give rise to functional alterations. Several intracellular ROS target proteins relate to muscle excitation–contraction coupling (Barreiro and Hussain, 2010). In particular, the open probability of ryanodine receptor type 1 (RyR1) increases upon H2O2-dependent oxidation (Donoso et al., 2000; Hidalgo and Donoso, 2008).
In striated muscle cells, two Ca2+ channels, the inositol-1,4,5-triphosphate (IP3) receptor (IP3R) and RyR1, mediate intracellular Ca2+ release from the sarcoplasmic reticulum (SR). The IP3R channels respond to IP3 generation by phospholipase C (PLC), producing cytoplasmic Ca2+ signals related to metabolism (Pacher et al., 2008) and gene expression (Jaimovich and Carrasco, 2002; Liberona et al., 2008). The RyR1 channels, which release Ca2+ in response to plasma membrane depolarization, contribute to the characteristic excitation–contraction coupling process of skeletal muscle cells (Protasi et al., 2002). A recent report proposes that RyR1 channels, by controlling passive Ca2+efflux from the SR to the cytoplasm, represent key factors in the management of the resting muscle cytoplasmic Ca2+concentration (Eltit et al., 2010).
Interestingly, our groups have reported that: (1) insulin elevates intracellular Ca2+ levels via NOX2 activation in primary neonatal myotubes (Espinosa et al., 2009), (2) NOX2-dependent ROS production promotes RyR1 S-glutathionylation and stimulates RyR1-mediated Ca2+ release from triad-enriched vesicles isolated from adult skeletal muscle (Hidalgo et al., 2006), and (3) intracellular Ca2+ contributes to the insulin-dependent increase in glucose uptake in cardiomyocytes (Contreras-Ferrat et al., 2010). The physiological mechanism whereby insulin activates NOX2, however, remains unknown.
An insulin-dependent Ca2+ increase, detected with the Ca2+ fluorescent dye Indo-1, occurs near the plasma membrane in single muscle fibers isolated from the Flexor digitorum brevis (FDB) muscle (Bruton et al., 1999), and this has been associated with insulin-stimulated Ca2+ influx via L-type voltage-dependent Ca2+ channels (Cav1.1). In cultured myotubes from rat skeletal muscle, insulin induces RyR1-mediated Ca2+ transients measured with the fluorescent Ca2+ indicator Fluo3 (Espinosa et al., 2004). In skeletal muscle fibers, 2-aminoethoxydiphenyl borate (2-APB) inhibits both basal and insulin-dependent Ca2+ influx (Lanner et al., 2006). A contribution of intracellular Ca2+ stores to glucose uptake was proposed on the basis that dantrolene, an inhibitor of the interaction between Cav1.1 and RyR1 that inhibits depolarization-induced Ca2+ release (Bannister, 2013), prevents insulin-dependent glucose uptake in adipocytes (Li et al., 2006). In those cells, decreasing intracellular Ca2+ signals decreased insulin-stimulated glucose uptake and Akt phosphorylation (Whitehead et al., 2001). In skeletal muscle fibers, Ca2+ influx is important for full stimulation of glucose uptake (Lanner et al., 2009; Lanner et al., 2006).
Notably, it remains unknown how insulin elicits intracellular Ca2+ signals and what are the sources of these signals. Given the prominence of insulin action in skeletal muscle, unraveling this mechanism is paramount to understand the regulation of glucose uptake by this hormone.
ROS induce GLUT4myc translocation and glucose uptake in skeletal muscle myotubes
We used L6-GLUT4myc myotubes (myotubes expressing GLUT4 with an exofacial Myc-epitope tag between the first and second transmembrane domains) to investigate the effect of ROS on glucose uptake. L6-GLUT4myc myotubes stimulated with insulin displayed an increase in exofacial exposure of the Myc epitope (Fig. 1A), as previously reported (Ishikura and Klip, 2008). L6-GLUT4myc myotubes stimulated for 20 min with different concentrations of H2O2 displayed a concentration-dependent increase of GLUT4myc translocation to the cell surface; as little as 0.01 mM H2O2 produced a stimulatory effect, which reached its maximum at 0.1 mM H2O2; treatment with 1 mM H2O2 did not produce an extra increase (Fig. 1A). The effects of 0.1 mM H2O2 were additive to those of insulin (Fig. 1B); of note, insulin-dependent GLUT4myc translocation to the cell surface was prevented by 1 mM H2O2, indicating a biphasic, concentration-dependent role of ROS as enhancers or inhibitors of insulin signaling.
To evaluate whether the increase in surface GLUT4myc reflects an increase in glucose uptake, we used the conventional pulse and chase activity assay using [3H]-2-deoxy glucose. A low concentration of H2O2 (0.1 mM) boosted insulin-dependent glucose uptake, whereas a higher H2O2 concentration (1 mM) prevented this response (Fig. 1C). To study the possible participation of ROS in insulin-dependent glucose uptake, we pre-incubated L6-GLUT4myc myotubes with two types of widely used antioxidant agents: N-acetylcysteine (NAC) or Trolox, a water-soluble vitamin E derivative. Interestingly, both antioxidant agents partially inhibited the stimulation of GLUT4myc surface exposure produced by insulin in myotubes (Fig. 1D) but not in myoblasts, in which insulin effectively stimulated GLUT4myc translocation (data not shown). These results suggest that ROS-dependent stimulation of GLUT4myc surface exposure requires a more mature state of differentiation.
In order to establish causality using molecular tools, we tested the participation of cytoplasmic antioxidant enzymes on insulin-induced GLUT4myc translocation to the cell surface. To this aim, L6-GLUT4myc myotubes were transduced with adenoviruses encoding superoxide dismutase 1 (AdSOD1) or the wild-type form of catalase (AdCat). Superoxide dismutase (SOD1) produces H2O2 from superoxide species, increasing the available basal levels of H2O2, whereas catalase rapidly induces H2O2 dismutation reducing its effective half-life. Transduced myotubes displayed in vitro enzyme activities over 15-fold higher compared to non-transduced cells (data not shown). Myotubes overexpressing SOD1 displayed a bigger increase in GLUT4myc translocation in response to insulin; in contrast, myotubes overexpressing catalase displayed reduced insulin-induced stimulation of GLUT4myc translocation to the cell surface (Fig. 1E). These results suggest a role for oxidative signaling in GLUT4myc translocation and glucose uptake downstream of insulin stimulation in L6-GLUT4myc skeletal myotubes.
Insulin increases cytoplasmic H2O2 production through PKC-dependent NOX2 activation, which promotes GLUT4myc translocation to the cell surface
As shown in Fig. 1, H2O2 induced an increase in both GLUT4myc translocation and glucose uptake; the molecular entities involved in these responses are unknown, as is the mechanism responsible for ROS generation downstream of insulin signaling in skeletal muscle cells. To determine whether insulin enhances cytoplasmic H2O2 production, we transiently transfected L6-GLUT4myc myotubes with a plasmid vector encoding the molecular H2O2 biosensor HyPer targeted to the cytoplasm (HyPer-cyto). Recording in real time revealed that Hyper-cyto effectively sensed low H2O2 concentrations in L6-GLUT4myc myotubes (Fig. 2A); addition of vehicle did not change the basal fluorescence levels (Fig. 2A). Addition of insulin induced a fast and transient increase in HyPer-cyto fluorescence emission, which reached a maximum at 7±3 s post stimuli and took 143±32 s to decay thereafter (mean±s.d.; Fig. 2B, black trace).
NOX2 is a protein complex canonically activated by translocation of p47phox, p67phox and the small GTP-binding protein Rac1 to the plasma membrane, where they form the active enzyme by interacting with the membrane integral proteins gp91 and p22 (Maghzal et al., 2012). Indirect immunofluorescence determinations showed p47phox subunits located in the cytoplasm of myotubes, showing poor expression in myoblasts (data not shown). To explore NOX2 participation in insulin-induced H2O2 generation, we inhibited NOX2 with the specific peptide gp91-ds-tat (Wei et al., 2008), which contains the gp91phox sequence linked to the HIV tat peptide. L6-GLUT4myc myotubes pre-incubated for 1 h with gp91-ds-tat or its inactive scrambled sequence sc-gp91-ds-tat were stimulated with insulin. Incubation with the inhibitory peptide abolished insulin-dependent H2O2 production (Fig. 2B, red trace), whereas the inactive peptide did not inhibit this response (data not shown). In order to explore whether protein kinase C (PKC) contributes to NOX2 activation, L6-GLUT4myc myotubes were pre-incubated with 1 µM bisindolylmaleimide (BIM), a conventional inhibitor of PKCα and PKCβ, also described to inhibit PKCδ. This PKC inhibitor completely suppressed insulin-dependent H2O2 production (Fig. 2B, green trace). Moreover, addition of 5 mM NAC drastically reduced basal HyPer-cyto fluorescence and prevented insulin-induced H2O2 production; NAC also moderately lowered the fluorescent signal induced by addition of 100 µM H2O2, used as positive control (data not shown). These combined results strongly suggest that NOX2 mediates insulin-dependent ROS generation through PKC activation.
To elucidate whether NOX2 contributes to insulin-dependent GLUT4myc translocation, we inhibited NOX2 with gp91-ds-tat. To this aim, L6-GLUT4myc myotubes were pre-incubated for 1 h with gp91-ds-tat or its inactive scrambled sequence sc-gp91-ds-tat and were stimulated with insulin for 20 min in the presence of these synthetic peptides. Pre-incubation with the gp91-ds-tat peptide partially reduced insulin-dependent GLUT4myc translocation; in contrast, transfection with the scrambled sequence did not affect this response (Fig. 2C). Furthermore, the pharmacological NOX2 inhibitors apocynin or diphenyliodonium (DPI) reduced but did not fully suppress insulin-dependent GLUT4myc translocation to the cell surface (supplementary material Fig. S1A). Myotubes transiently transfected with siRNA (50 nM) against the regulatory NOX2 subunit p47phox displayed significantly reduced p47phox levels (supplementary material Fig. S1C) and lower GLUT4myc translocation following insulin addition than either control myotubes or myotubes transfected with a scrambled siRNA (Fig. 2D).
Different PKC enzymes contribute to insulin signaling
Previous studies indicate that a diacylglycerol-mimicking phorbol ester (Wright et al., 2004) and compounds that increase intracellular Ca2+ (Wright et al., 2005) stimulate glucose uptake in cultured cells and isolated rodent muscles. In muscle cells, it has been suggested that activation of PKCζ and PKCλ increases GLUT4 translocation, although this finding has been somewhat controversial (Liu et al., 2006; Powell et al., 2003; Stretton et al., 2010). Moreover, PKCα and PKCβ promotes p47phox activation and plasma membrane translocation to assemble the NOX2 activated complex in macrophages (San José et al., 2009). It has been widely accepted that phosphorylation via PKC activates NOX2 in many systems. To test whether direct PKC activation promotes NOX2-dependent H2O2 production, we stimulated myotubes with the phorbol ester 12-myristate 13-acetate (PMA, a universal activator of classical and novel PKCs). In L6-GLUT4myc myotubes, PMA induced a fast and transient increase in HyPer-cyto fluorescence, which reached a maximum value 3±2 s post stimuli (Fig. 2E), with a decay half time of 126±41 s (mean±s.d.; presumably reflecting the breakage kinetics of the H2O2-induced HyPer-cyto disulfide bridge by cellular reducing systems). Pre-incubation of myotubes with gp91-ds-tat but not with the inactive peptide abolished PMA-dependent H2O2 production (Fig. 2E). Furthermore, incubation of myotubes with PMA for 20 min strongly increased GLUT4myc translocation to the cell surface; the synthetic inhibitory peptide gp91-ds-tat abolished this increase (Fig. 2F). Co-stimulation of myotubes with PMA and insulin resulted in a significant extra increase in exofacial exposure of the Myc epitope over that produced by insulin alone (supplementary material Fig. S1B). Both the PKC inhibitor Gö6976 (specific for Ca2+-dependent PKCs) and more so BIM (a general inhibitor of types α, β, γ and δ PKCs), significantly reduced GLUT4 translocation induced by insulin (Fig. 2G). Taken together, these results suggest that PKC is involved in the increased NOX2-mediated cytoplasmic H2O2 generation and GLUT4myc translocation stimulated by insulin.
Exofacial exposure of GLUT4myc induced by insulin requires RyR1-mediated Ca2+ release
L6-GLUT4myc myotubes pre-incubated with the Ca2+ dye indicator Fluo4 am were mounted in a chamber and intracellular Ca2+ levels were recorded frame-by-frame in a confocal microscope. After 1 min recording, we replaced the Ca2+-containing resting solution with a Ca2+-free solution (Fig. 3A–C, blue dashed lines). Addition of 100 µM H2O2 produced a rapid increase in intracellular Ca2+ levels that did not occur in cells pre-treated with 50 µM ryanodine, strongly suggesting that ROS induced RyR1-mediated Ca2+ release (Fig. 3A). Insulin addition in Ca2+-free resting medium generated whole-cell Ca2+ transients, which were not present in cells pre-incubated with 50 µM ryanodine for 3 h or 5 µM gp91-ds-tat for 30 min (Fig. 3B). Addition of the Ca2+ ionophore ionomycin to cells kept in Ca2+-free solution induced a rapid rise in fluorescence in all conditions, presumably by inducing Ca2+ release from intracellular stores (Fig. 3A,B). To induce RyR1-mediated Ca2+ release, we used the specific RyR agonist 4-chloro-m-cresol (4-CMC). In Ca2+-free resting medium, 4-CMC induced a fast increase in intracellular Ca2+, which slowly decreased to near basal levels after 5 min (Fig. 3C); ionomycin addition was used as a positive control at the end of the recording.
4-CMC also induced an increase in GLUT4myc translocation to the cell surface, and this effect was more evident in insulin co-stimulated myotubes (Fig. 3D). Translocation of GLUT4myc induced by 4-CMC was completely inhibited in myotubes pre-incubated with inhibitory doses of ryanodine (data not shown). Moreover, insulin-dependent GLUT4myc translocation was also significantly lower in myotubes pre-incubated with ryanodine (Fig. 3D), whereas myotubes pre-incubated jointly with the NOX2 gp91-ds-tat inhibitor and ryanodine did not show additional inhibition over that produced by ryanodine alone (Fig. 3D). We used ionomycin to determine whether an increase in intracellular Ca2+ levels stimulates GLUT4myc translocation to the cell surface. Myotubes stimulated with 1 µM ionomycin for 5 min showed increased exofacial exposure of the Myc epitope that was additive to the stimulation induced by insulin (Fig. 3E). Insulin-dependent GLUT4myc translocation did not occur in L6-GLUT4myc myotubes pre-incubated with BAPTA-AM for 30 min; this intracellular Ca2+ chelator did not affect the basal levels of translocation (Fig. 3E). This strategy was complemented by overexpressing the parvalbumin Ca2+ buffer protein fused to DsRed, as the cell tracker (PV–DsRed). A plasmid encoding only DsRed was used as control. Cells were fixed in non-permeabilized conditions and the exofacial exposure of the Myc epitope was assayed in single cells. PV–DsRed-expressing cells were unresponsive to insulin and exhibited decreased basal exposure of the Myc epitope (Fig. 3F). Moreover, non-transfected myotubes showed a notable increase in the amount of insulin-induced cell surface GLUT4myc stain (Fig. 3F, arrowheads). These combined experiments suggest that the cytoplasmic Ca2+ increase produced by stimulation of Ca2+ release through ROS-mediated RyR1 activation forms part of the insulin signaling pathways that promote GLUT4myc translocation to the cell surface.
Insulin-dependent H2O2 production induced RyR1 S-glutathionylation in skeletal muscle cells
ROS generation induces reversible S-glutathionylation of reactive cysteine residues (Pastore and Piemonte, 2012). To detect whether S-glutathionylation of proteins increases in insulin-stimulated cells, myotubes were stimulated with insulin or H2O2 for 1 min and the indirect co-immunofluorescence against S-glutathionylated protein adducts or RyR1 was detected with specific antibodies. In basal conditions, myotubes showed very low fluorescence when tested with the antibody against S-glutathionylated protein adducts, indicating low basal levels of this redox modification (Fig. 4A). In contrast, myotubes stimulated with 100 µM H2O2 exhibited significantly higher S-glutathionylated protein levels, which partially overlapped with RyR1 immunofluorescence yielding a Pearson coefficient of 0.44±0.06 and Mander's coefficients M1 and M2 of 0.68±0.09 and 0.61±0.09 (mean±s.d.), respectively (Fig. 4A). Of note, 1 min exposure to 100 nM insulin increased S-glutathionylated protein adducts with a clear overlap with the RyR1 stain, reaching a Pearson coefficient of 0.782±0.07 and Mander's coefficients M1 and M2 of 0.92±0.07 and 0.91±0.02, respectively (Fig. 4A). As the resolution provided by co-immunofluorescence experiments cannot ascertain the actual RyR1 S-glutathionylation levels, we used a novel high-resolution technique that yields positive results for molecules located less than 40 nm apart. Specific RyR1 S-glutathionylation was ascertained using in situ proximity ligation assay (PLA) probes, which detect closely positioned antibodies with an optimal distance of 20–30 nm (Söderberg et al., 2006). To this aim, cultured myotubes stimulated with insulin or H2O2 for 1 min and rapidly fixed were probed both with anti-RyR1 and anti-S-glutathionylated protein adducts; RyR1 S-glutathionylation is indicated by the appearance of fluorescent dots (Fig. 4B). The basal levels of RyR1 S-glutathionylation were 5±1.4 dots/1000 µm3, whereas in insulin-stimulated myotubes these levels increased to 24±2.5 dots/1000 µm3. Cells treated with 100 µM H2O2 displayed 37±4.2 dots/1000 µm3 (mean±s.d.; Fig. 4C).
Insulin-dependent exofacial exposure of GLUT4myc and glucose uptake in skeletal myotubes require IP3R activation
IP3R-dependent Ca2+ signaling plays a relevant role in cell physiology (Foskett et al., 2007). The IP3R increases its open probability in response to IP3 produced by PLC activation. We explored whether IP3R-generated Ca2+ signals also contribute to insulin-dependent GLUT4myc translocation in L6 and primary rat neonatal myotubes. L6-GLUT4myc myotubes express all three IP3R isoforms (supplementary material Fig. S2A). The IP3R1 isoform displayed a cytoplasmic and nuclear distribution showing diffuse stain with condensed highly fluorescent dots. Interestingly, the IP3R2 isoform presented a striated transversal pattern in the cytoplasm. The IP3R3 isoform displayed poor fluorescence intensity compared to the other isoforms, detected with the same microscope settings (supplementary material Fig. S2A).
Insulin addition produced a fast increase in intracellular IP3 levels, which increased from the basal values of 12.3±1.6 to 24.8±2.4 pg/mg of protein in 30 s (mean±s.d.; Fig. 5A). Both, LY294002 and U-73122 [broad inhibitors of phosphoinositide 3-kinase (PI3K) and PLC, respectively] inhibited insulin-stimulated IP3 production, whereas U-73343, an inactive analog of U-73122, did not affect IP3 production (Fig. 5A). Pre-incubation of myotubes with the PI3K inhibitor LY294002 did not affect the basal myc epitope levels but abolished insulin-dependent GLUT4myc translocation (Fig. 5B). We found that the PLC inhibitor U73122 partially inhibited, whereas the inactive analogue U73343 had no effect on, insulin-dependent GLUT4myc translocation in myotubes (Fig. 5B). The last step in IP3-dependent Ca2+ release is the interaction of IP3 with the IP3-binding site of IP3R. Myotubes pre-incubated with Xestospongin B, a specific IP3R inhibitor, displayed a partial reduction of insulin-dependent GLUT4myc translocation (Fig. 5C). Combined treatment with both intracellular Ca2+ channels inhibitors, Xestospongin B and ryanodine, strongly inhibited insulin-dependent Myc epitope externalization, which reached values not significantly different from the controls not treated with insulin (Fig. 5C).
To further test the involvement of IP3R2, L6-GLUT4myc myotubes were transfected with siRNA against IP3R2, which produced a >90% reduction in protein levels compared with non transfected cells or with a control non-targeting oligonucleotide (Fig. 5D). Insulin addition to L6-GLUT4myc myotubes transfected with the IP3R2 siRNA produced a lower stimulation of Myc epitope exposure compared to the stimulation produced in control cells or in cells transfected with the control siRNAs (Fig. 5E).
Myotubes overexpressing IP3R2 wild type exhibited increased basal levels of myc epitope exposure compared to control or DsRed-transfected cells, detected by indirect immunofluorescence in non-permeabilized cells (Fig. 5F). Moreover, insulin-stimulated myoblasts overexpressing IP3R2 displayed significantly increased surface GLUT4myc stain compared to control cells or to cells transfected only with DsRed (supplementary material Fig. S2B).
Because it is important to validate these results in skeletal muscle primary cultures, we used the well-characterized myotube preparation from mice neonatal muscle. Insulin addition stimulated glucose transport in primary myotubes by over 2-fold, as determined by measurement of the uptake of the fluorescent glucose analog 2-NBDG (Fig. 5G). LY294002 suppressed and U73122 partially decreased insulin-dependent 2-NBDG uptake, whereas the inactive analog U73343 had no effect (Fig. 5G). Xestospongin B reduced insulin-dependent 2-NBDG uptake, whereas co-incubation with Xestospongin B and ryanodine suppressed this response (Fig. 5H). The stimulatory effect of insulin on 2-NBDG uptake was partially reduced in primary myotubes transfected with siRNA against IP3R2; transfection with the control siRNA did not induce changes (Fig. 5I). These combined results show that IP3R2 has a relevant participation in insulin-dependent GLUT4myc translocation and 2-NBDG uptake in skeletal muscle cells.
Insulin-dependent IP3R activation regulates mitochondrial Ca2+ and pH levels in L6-GLUT4myc cells
To investigate whether insulin promotes mitochondrial Ca2+ uptake, and hence stimulates mitochondrial function, we used the mitochondrial-targeted ratiometric PeriCaM (mito-PeriCaM) as a Ca2+ probe. This probe displays two excitation maxima (410–440 nm and 480–490 nm), each one presenting a maximal emission wavelength of 535 nm (Fonteriz et al., 2010). The 410–440 nm peak displays high Ca2+ sensitivity, as decreasing fluorescence emission, whereas the 480–490 nm is highly responsive to [H+], decreasing fluorescence emission when the pH goes down. Following insulin addition, L6-GLUT4myc myotubes transfected with mito-PeriCaM and mounted 24 h later for viewing under a microscope, displayed fast and transient increases in both mitochondrial Ca2+ levels and matrix pH (Fig. 6A). Ca2+ levels decreased towards the basal level at 10 min post-stimuli (data not shown). The area under the curve, corresponding to the first 5 min post insulin was measured (Fig. 6B). The serine/threonine kinase Akt inhibitor Akti1/2 did not affect insulin-dependent mitochondrial Ca2+ uptake of L6-GLUT4myc myotubes transiently transfected with mito-PeriCaM, whereas pre-incubation with inhibitory doses of ryanodine for 3 h produced a partial decrease (Fig. 6C). Pre-incubation with Xestospongin B or Ruthenium Red (RuRed) to inhibit the mitochondrial Ca2+ uniporter (MCU), or co-transfection with siRNA against IP3R2 significantly reduced insulin-dependent mitochondrial Ca2+ uptake (Fig. 6C). To associate the insulin-dependent mitochondrial Ca2+ handling to GLUT4myc translocation, myotubes were pre-incubated for 30 min with RuRed or Akti1/2; both inhibitors reduced insulin-dependent GLUT4myc translocation (Fig. 6D). Interestingly, co-incubation with Akti1/2 and RuRed, or with Akti1/2 and Xestospongin B, did not increase the inhibitory effects of RuRed or Xestospongin B alone (Fig. 6D).
To test the participation of PI3K signaling on insulin-dependent mitochondrial Ca2+ uptake, primary myotubes were co-transfected with Δp85dn, a dominant-negative form of the PI3K IA regulatory subunit. These cells displayed significantly reduced insulin-dependent mitochondrial Ca2 uptake (Fig. 6E). Co-transfection of the PH domain of centaurin α1 fused to eGFP [a molecular inhibitor of phosphatidylinositol 3,4,5-trisphosphate (PIP3) signaling] together with mito-PeriCaM or of an IP3 sponge corresponding to the IP3-binding site of IP3R (M49-IP3 sponge), acting as competitive inhibitors of PIP3 signaling and IP3 production respectively, suppressed insulin-dependent mitochondrial Ca2+ uptake (Fig. 6E). Co-transfection with a mutated construct without sponge capacity (M30-IP3 sponge) did not alter the stimulatory effect of insulin (Fig. 6E). Additional experiments to inhibit mitochondrial Ca2+ uptake (RuRed) or IP3R function (M49-IP3 sponge) reduced insulin-dependent 2-NBDG uptake in primary myotubes (Fig. 6F). All these experiments were made in labeled single living cells, as described in the Materials and Methods, in order to avoid dealing with the low transfection efficiency (less than 5%) observed in primary myotubes. These data suggest that the increased mitochondrial Ca2+ and pH levels induced by insulin are associated with GLUT4 translocation to the cell surface and glucose uptake in muscle cells.
Insulin binding to its receptor on muscle cells promotes tyrosine phosphorylation of insulin receptor substrate-1, inducing PI3K activation, which in turn activates the serine/threonine kinase Akt and the small G protein Rac1 to mobilize translocation of GLUT4 to the muscle membrane (Zaid et al., 2008). For years, reports have proposed that additional input occurs from discrete and possibly localized changes in H2O2 and cytoplasmic Ca2+ concentration, as well as from PLC-derived signals, but the origin and precise mode of action of these putative mediators has remained elusive. Here, we show how these two signals originate, and that they arise in parallel and jointly contribute to insulin-induced GLUT4 translocation in skeletal myotubes. Specifically, we report that insulin leads to activation of NOX2-dependent H2O2 production to promote RyR1 S-glutathionylation and RyR1-mediated Ca2+ release; in parallel, there is sequential activation of PI3K and PLC resulting in IP3 generation and activation of IP3R-mediated Ca2+ release, which promotes mitochondrial Ca2+ uptake. Collectively, these two signaling modes contribute to the net gain in surface GLUT4 levels elicited by insulin. We propose a new integrative mechanism of insulin signaling that comprises RyR1 Ca2+ channel redox modifications and ligand-dependent IP3R activation in skeletal muscle cells (Fig. 7).
The ROS–RyR1 branch
As for many actions of ROS, the impact of H2O2 on insulin action and on glucose transport in particular is biphasic (Mahadev et al., 2001; Rudich et al., 1998; Bashan et al., 2009). Using a battery of molecular and pharmacological tools, we show that insulin promotes NOX2-dependent H2O2 generation. Furthermore, we show that exogenous H2O2, induces, at all concentrations, the translocation of the GLUT4 transporter to the surface of myotubes, and also either potentiates or decreases, respectively, the effects of insulin in myotubes co-stimulated with low or high H2O2 concentrations. These evidences suggest that Ca2+ release mediates the stimulatory effect of H2O2 on insulin action, but the inhibitory effect of high doses of H2O2 on insulin action likely implies an additional mechanism. A previous report has shown that mice lacking glutathione peroxidase 1 (Gpx−/−), a key enzyme for physiological ROS removal, and fed a high-fat diet are protected from insulin resistance; their increased insulin sensitivity correlates with enhanced oxidation of the protein tyrosine phosphatase PTEN and is reversed by NAC (Loh et al., 2009). Consistent with those findings, we report that chemical (NAC, Trolox) and molecular (AdCat) antioxidants decrease insulin-dependent GLUT4 translocation in L6-GLUT4myc myotubes, whereas the potentiated effect displayed by myotubes overexpressing SOD1 presumably relates to a high pro-oxidative state. In neonatal cardiac myocytes, acute H2O2 increases cytoplasmic Ca2+ levels and GLUT4 translocation to the cell surface (Horie et al., 2008). A role of Ca2+/calmodulin-dependent protein kinase kinase (CaMKK) in H2O2-dependent ERK1/2 and Akt activation has also been reported in smooth muscle cells (Bouallegue et al., 2009).
Chronic exposure of adipocytes to H2O2 reduces IRS1-asociated PI3K activity and impairs insulin-stimulated GLUT4 translocation (Rudich et al., 1998) and Akt and Rac1 activation (JeBailey et al., 2007). This is consistent with oxidative stress as a contributor factor to insulin resistance. Indeed, obese mice have a higher intracellular oxidative environment (Anderson et al., 2009) that could produce a decrease in glucose uptake, such as that presented in Fig. 1, at high H2O2 concentration.
The participation of NOX2 in insulin signaling has begun to emerge in a number of cellular systems. We previously reported that NOX2 activation mediates insulin-dependent Ca2+ release in primary myotubes (Espinosa et al., 2009); this study suggested that PKC participates in NOX2 activation, and it is known PKC promotes NOX2 activation in other systems (Gupte et al., 2009). In the present work, we confirmed the participation of PKC in insulin signaling. In fact, PKC inhibitors completely prevented insulin-induced H2O2 production, whereas PKC activation with PMA induced GLUT4 translocation. The fact that the stimulation produced by PMA was higher than that produced by insulin suggests that several different PKC isoforms participate in this process, not all of them sensitive to insulin signaling.
We further describe a specific role of insulin-dependent H2O2 production on RyR1 redox state. Sulfhydryl reagents modify selective RyR1 cysteine residues known as the ‘hyper-reactive’ cysteine residues (Zable et al., 1997). Modification of these hyper-reactive cysteine residues increases the open probability of RyR1 single channels (Marengo et al., 1998) and enhances RyR1-mediated Ca2+ release (Hidalgo et al., 2006). These hyper-reactive cysteine residues on RyR1 are targets for disulfide cross-linking, S-nitrosylation and/or S-glutathionylation (Aracena-Parks et al., 2006). In particular, a selective decrease in RyR1 channel inhibition by Mg2+ underlies the increased RyR1 activity induced by S-glutathionylation (Aracena et al., 2003). Protein modifications by S-glutathionylation have been implicated in the regulation of gene expression, cell signaling, ion channels, energy metabolism, mitochondria function, cell death and survival, cytoskeleton, folding and degradation (Pastore and Piemonte, 2012).
Based on those findings, it became important to assess the potential contribution of Ca2+ to downstream insulin action. Indeed, there are reports that GLUT4 exocytosis requires intracellular Ca2+ in adipocytes, allowing activation of Akt (Whitehead et al., 2001) and CDP138, a Ca2+-dependent protein downstream of Akt (Xie et al., 2011). In skeletal muscle, however, direct evidence showing intracellular Ca2+-dependent GLUT4 translocation was missing, although results have suggested that it occurs (Lanner et al., 2006; Wijesekara et al., 2006), and localized, sub-membrane changes in Ca2+ levels have been reported in response to the hormone (Bruton et al., 1999).
Here, we show that RyR1-mediated Ca2+ release, presumably through enhanced RyR1 S-glutathionylation, contributes solely to the cytoplasmic Ca2+ increase produced by insulin, which promotes insulin-induced GLUT4 translocation to the cell surface. Given that NOX2-dependent ROS generation leading to RyR1 S-glutathionylation and activation occurs in the skeletal muscle T-tubule membranes (Hidalgo et al., 2006), and GLUT4 preferentially inserts into the T-tubules in response to insulin (Marette et al., 1992), we speculate that short-term insulin-mediated cytoplasmic Ca2+ spikes near the T-tubule membrane allow vesicles containing the GLUT4 transporter to fuse into the T-tubule muscle membranes.
The IP3R–mitochondria branch
Intriguingly, we found that insulin also stimulates IP3 production in myotubes, raising the question as to whether this intracellular mediator might also contribute to insulin-induced GLUT4 translocation. The observed reduction in GLUT4 translocation brought about by strategies designed to inhibit IP3R or dampen IP3 levels support this prediction (Figs 5 and 6). Inhibition of RyR1 abolished the insulin-dependent Ca2+ spikes, indicating that IP3R-mediated Ca2+ release does not make a substantial contribution to the global cytoplasmic Ca2+ increase induced by insulin. Instead, our results show significant increases in mitochondrial Ca2+ originating from insulin-stimulated IP3R-mediated Ca2+ release, suggesting that mitochondria rapidly buffer the local Ca2+ changes produced by this pathway. We have not explored, however, whether H2O2 affects IP3-dependent mitochondrial Ca2+ signals; this interesting possibility should be the subject of future studies. In addition, the impact of the IP3R–mitochondrial shunt might contribute to maintenance of mitochondrial function, adding to the overall anabolic function of insulin. It is interesting to note that incubation with ryanodine also has a partial effect on insulin-dependent mitochondria Ca2+ uptake (Fig. 6C); this can be interpreted in terms of a modulation of IP3Rs by Ca2+ coming from RyR, or a direct Ca2+ transfer from RyR to the mitochondria.
In conclusion, we find that insulin-activated NOX2 leads to S-glutathionylation and consequent activation of RyR1 that, in conjunction with activation of IP3R, contribute to insulin-dependent GLUT4 translocation. Future work should examine whether the transient increase in cytoplasmic Ca2+ potentiates elements in the canonical insulin signaling pathway constituted by PI3K and Akt and/or whether it directly promotes vesicle fusion with the surface membrane.
MATERIALS AND METHODS
Penicillin-streptomycin and amphotericin B were obtained from Sigma-Aldrich. Dulbecco's modified Eagle's medium-F12, αMEM, bovine serum and fetal bovine serum (FBS) were from Invitrogen. Collagenase type II was from Worthington Biochemical Corp. Mini protease inhibitors were from Roche Applied Science. Secondary horseradish-peroxidase-conjugated anti-rabbit and anti-mouse Ig antibodies were from Pierce Biotechnology. Enhanced chemiluminescence reagents were from Amersham Pharmacia Biotech (Amersham, UK). Polyvinylidenedifluoride (PVDF) membranes were from Millipore. All other reagents were obtained from Sigma, Merck (Darmstadt, Germany) or Invitrogen. 2-NBDG and anti-rabbit or anti-mouse conjugated to Alexa Fluor 488 or 546 were from Molecular Probes. LY-294002 and Akt inhibitor VIII were from Calbiochem. Anti-Myc antibodies used were both polyclonal (Sigma-Aldrich) or monoclonal (sc40 clone, Santa Cruz Biotechnology). siRNA against p47phox was from Santa Cruz Biotechnology, siRNA against IP3R2 and Dharmafect was from Thermo Scientific.
Newborn rats were bred in the Animal Breeding Facility, Faculty of Medicine, Universidad de Chile (Santiago, Chile). Studies were approved by the Institutional Bioethical Committee, Faculty of Medicine, Universidad de Chile, in accordance with the ‘Guide for the Care and Use of Laboratory Animals’ (Bayne, 1996).
Primary cultures of skeletal muscle cells were prepared from Sprague-Dawley neonatal rats as previously reported (Jaimovich et al., 2000). Six- to seven-day-old cultures were employed for the experiments. L6 muscle cells stable expressing GLUT4 with an exofacial Myc epitope (L6-GLUT4myc) were cultured as described previously (Wang et al., 1998).
Western blot analysis was performed as previously reported (Contreras-Ferrat et al., 2010). The following primary antibodies and dilutions were used: anti-β-actin (1∶3000 Cell Signaling) and anti-IP3R2 (1∶2000; ABR, PA1-904). After scanning the films, densitometry analysis of the bands was performed with the Image J program (NIH, Bethesda, MD, USA).
Myotubes were pre-incubated with inhibitors for 30 min and with insulin for 20 min in presence of each inhibitor. Glucose uptake was measured using 10 µM [3H]2-deoxyglucose (3H-2DG). To quantify 3H-2DG uptake (0.1 µCi/well), myotubes were grown on 12-well plates (106 cells per well) and treated with insulin in different conditions as detailed in the text. After stimulation, cells were washed with cold HEPES buffer solution and were later incubated with 3H-2DG for 10 min on ice. Cold glucose solution (30 mM) was added to stop 3H-2DG uptake. Cells were washed with 30 mM HEPES/glucose buffer and incubated with 0.05 M NaOH. Radioactivity was measured by liquid scintillation counting. Glucose uptake was expressed as fold over basal.
Intracellular Ca2+ measurements
Images were collected for 10 min using laser scanning confocal microscope in the frame-by-frame mode as previously reported (Contreras-Ferrat et al., 2010). L6 myotubes were preloaded with Fluo4 am (5 µM) at 37° for 30 min in Krebs Ringer buffer and stimulus were added to the microscopy chamber as indicated by arrows. BAPTA-AM (30 µM) was pre-incubated together with Fluo3 am whereas ryanodine (50 µM) was pre-incubated during 3 h. The dashed line in Fig. 3 represents medium replacement from Ca2+-containing to Ca2+-free conditions. Data were analyzed using ImageJ software.
Single-cell fluorescent hexose uptake assay
Analysis uptake of 2-NBDG was performed as previously reported (Osorio-Fuentealba et al., 2012). Approximately 100 cells in each condition were analyzed in different myotube cultures. The ImageJ software (NIH, Bethesda, MD, USA) was used to quantify 2-NBDG uptake in stimulated and non-stimulated cells. Glucose uptake is expressed as fold over basal (non-stimulated cells).
Adenovirus for catalase (AdCat) (Lam et al., 1999), cytosolic superoxide dismutase 1 (AdSOD1) (Zwacka et al., 1998) and empty construct (AdEmpty) were used to transduce myotubes at a multiplicity of infection (MOI) equivalent to 10,000 adenoviral particles per myotube for 24 h. Transduction efficiency was over 95% as monitored with adenovirus. Myotubes were infected with adenoviral vectors at a multiplicity of infection (MOI) of 1000 at least 24 h before use.
Quantification of cell surface GLUT4myc
A previously described assay was used (Wang et al., 1998). Briefly, cells grown in 12- or 24-well plates and serum starved for 3 h were treated with 100 nM insulin. The reaction was stopped with 0.25 ml per well of 3 M HCl. Supernatants were collected and absorbance was measured at 492 nm. To determine the level of GLUT4myc stain in single cells, the experiments were performed as previously described by (Wang et al., 1999).
Plasmid transfection, immunofluorescence microscopy and intracellular H2O2 detection
Myotubes were transfected with 3 µg of NES-PV-DsRed, Hyper-cyto, M49-IP3 sponge, M30-IP3 sponge, Δp85dn, eGFP-tagged centaurin α1, RFP or mitoPeriCaM using 6 µL/ml Lipofectamine™ 2000 (Invitrogen). Transfection efficiency was over 80%. Plasmid-treated cells were incubated with transfection mixture for 4 h in penicillin-streptomycin and FBS-free OPTI-MEM (Invitrogen) medium. Myotubes were incubated in αMEM or F12-DMEM (1∶1) for 24 h, and stimulated and processed according to each experimental protocol. The level of cell surface GLUT4myc was determined as described previously (Contreras-Ferrat et al., 2010; Wang et al., 1999). 1 µm z-stack images were acquired by laser scanning confocal microscope (CarlZeissPascal5, 63× NA1.4 objectives, Oberkochen, Germany). All materials used were free of detergent to avoid cell permeabilization. For GLUT4myc cell surface detection, >20 cells were analyzed in each condition from four different cultures. The HyPer-cytoplasmid has cpYFP inserted into OxyR-RD and displays submicromolar affinity for H2O2 (Belousov et al., 2006). Addition of 100 µM H2O2 to myotubes 24 h post-transfection caused a rapid and transient increase in intracellular HyPer-cyto fluorescence (ratio 490∶420 nm).
Determination of intracellular IP3 production
Control or experimental L6-GLUT4myc myotubes were quickly frozen in liquid nitrogen and were homogenized in 20 mM Tris-HCl pH 7.5, 2 mM EDTA, 150 mM NaCl and 0.5% Triton X-100. Determinations of IP3 production were performed with an IP3 ELISA Kit (Cusabio Biotech) following the manufacturer's instructions.
Data from at least four independent experiments are expressed as mean±s.d. The significance of difference among treatments was evaluated using a Student's t-test for unpaired data or by analysis of variance followed by Tukey post-hoc test. P<0.01 was considered statistically significant.
A.C.F. proposed the original idea; A.E. and E.J. contributed to the experimental design; A.C.F., P.L.l., C.V. performed Ca2+ measurements, GLUT4myc translocations and transfections; A.C.F., C.O.F. performed 2-NBDG uptake; A.C.F., M.A.C. and E.J. performed P.L.A. imaging and analysis; A.C.F., A.E., S.L., A.K., C.H. and E.J. analyzed results and wrote the manuscript.
This work was supported by FONDECYT [grant numbers 3110170 and 11130267 to A.C.-F., 3110105 to P.L., ACT1111 to E.J., S.L., A.E., 1100052 and BNI P-09-015 to C.H.]; and Canadian Institutes of Health Research [grant number MT 7307 to A.K.]. C.V. and M.A-C. hold PhD fellowships from CONICYT, Chile.
This work was supported by FONDECYT [grant numbers 3110170 and 11130267 to A.C.-F., 3110105 to P.L., ACT1111 to E.J., S.L., A.E., 1100052 and BNI P-09-015 to C.H.]; and Canadian Institutes of Health Research [grant number MT 7307 to A.K.]. C.V. and M.A-C. hold PhD fellowships from CONICYT, Chile.
The authors declare no competing interests.