H₂O₂-Activated Mitochondrial Phospholipase iPLA₂γ Prevents Lipotoxic Oxidative Stress in Synergy with UCP2, Amplifies Signaling via G-Protein–Coupled Receptor GPR40, and Regulates Insulin Secretion in Pancreatic β-Cells

Abstract

Aims: Pancreatic β-cell chronic lipotoxicity evolves from acute free fatty acid (FA)–mediated oxidative stress, unprotected by antioxidant mechanisms. Since mitochondrial uncoupling protein-2 (UCP2) plays antioxidant and insulin-regulating roles in pancreatic β-cells, we tested our hypothesis, that UCP2-mediated uncoupling attenuating mitochondrial superoxide production is initiated by FA release due to a direct H₂O₂-induced activation of mitochondrial phospholipase iPLA₂γ. Results: Pro-oxidant tert-butylhydroperoxide increased respiration, decreased membrane potential and mitochondrial matrix superoxide release rates of control but not UCP2- or iPLA₂γ-silenced INS-1E cells. iPLA₂γ/UCP2-mediated uncoupling was alternatively activated by an H₂O₂ burst, resulting from palmitic acid (PA) β-oxidation, and it was prevented by antioxidants or catalase overexpression. Exclusively, nascent FAs that cleaved off phospholipids by iPLA₂γ were capable of activating UCP2, indicating that the previously reported direct redox UCP2 activation is actually indirect. Glucose-stimulated insulin release was not affected by UCP2 or iPLA₂γ silencing, unless pro-oxidant activation had taken place. PA augmented insulin secretion via G-protein–coupled receptor 40 (GPR40), stimulated by iPLA₂γ-cleaved FAs (absent after GPR40 silencing). Innovation and Conclusion: The iPLA₂γ/UCP2 synergy provides a feedback antioxidant mechanism preventing oxidative stress by physiological FA intake in pancreatic β-cells, regulating glucose-, FA-, and redox-stimulated insulin secretion. iPLA₂γ is regulated by exogenous FA via β-oxidation causing H₂O₂ signaling, while FAs are cleaved off phospholipids, subsequently acting as amplifying messengers for GPR40. Hence, iPLA₂γ acts in eminent physiological redox signaling, the impairment of which results in the lack of antilipotoxic defense and contributes to chronic lipotoxicity. Antioxid. Redox Signal. 23, 958–972.

Introduction

A significant antioxidant role in pancreatic β-cells (1, 2, 9, 13, 23, 28, 29, 31, 42, 45, 48, 54) or α-cells (3) is provided by mitochondrial uncoupling protein-2 (UCP2). This was evidenced for UCP2 KO mice of three highly congenic strain backgrounds, all of which exhibit oxidative stress (decreased ratios of reduced-to-oxidized glutathione in blood or tissues), elevated levels of antioxidant enzymes, and increased nitrotyrosine content in their islets (42). Pancreatic β-cells from UCP2 KO mice showed chronically higher reactive oxygen species (ROS) when compared with wild-type mice (29). Mice with selective knockout of UCP2 in pancreatic β-cells exhibited increased glucose-induced inner mitochondrial membrane (IMM) potential (ΔΨ_m) and elevated intracellular ROS (48).

Innovation.

Fatty acid (FA)–stimulated and redox-stimulated insulin releases have not been fully understood as well as acute lipotoxicity, instantly decreasing insulin secretion in pancreatic β-cells. We describe a feedback antioxidant mechanism based on redox signaling initiated by FA β-oxidation, and promoted plus amplified by mitochondrial phospholipase iPLA₂γ. Not only the antioxidant synergy of iPLA₂γ with mitochondrial UCP2 is demonstrated, but also the iPLA₂γ role in the amplifying mechanism, since further free FAs cleaved by iPLA₂γ serve as “messengers” for G-protein–coupled receptor 40 (GPR40). Consequently, the iPLA₂γ/UCP2 synergy regulates glucose-stimulated, redox-, and FA-stimulated insulin release in pancreatic β-cells.

Superoxide formation is an inevitable side reaction at Complex I and III of mitochondrial respiratory chain (24) and in 2-oxoacid dehydrogenases (41, 46). Mitochondrial superoxide formation increases with an increasing substrate (NADH) load, represented by increasing glucose in pancreatic β-cells (10). Similarly, in numerous situations of local or global electron transfer retardation within the respiratory chain, superoxide production is specifically elevated. This serves for redox signaling, for example, during initiation of hypoxic gene expression remodeling (27).

Mitochondrial H⁺ pumping is usually tightly coupled to the H⁺ backflow via the ATP synthase. Since any uncoupling of this accelerates electron transfer within the respiratory chain (and hence respiration), the superoxide formation is attenuated by mitochondrial uncoupling. This represents the key mechanism exerted by UCP2, although it slightly attenuates ATP synthesis. In pancreatic β-cells, the increase in oxidative phosphorylation (OXPHOS) substantiates the canonical mechanism of glucose sensing. The increasing ATP/ADP ratio at higher glucose initiates the glucose-stimulated insulin secretion (GSIS) (5, 26, 47).

By shifting ROS homeostasis, UCP2 may participate in redox signaling in β-cells (31, 48), which may be easily transmitted due to the low capacity of redox buffers (23). H₂O₂-responsive gene expression is manifested for both major differentiation factors of β-cells, PDX-1 and MafA (47). Impaired antioxidant defense leading to chronic oxidative stress may affect insulin secretion machinery that is finely tuned for optimum GSIS in β-cells, as recognized in type 2 diabetes patients (16, 39, 40) and rodent diabetic models (30, 33). ROS may further accelerate diabetic development by promoting apoptosis, thus decreasing β-cell mass (51). Consequently, oxidative stress serves as a mediator of β-cell remission.

The function of UCP1 (12) and recombinant UCP2 (6, 7, 20, 53) is essentially dependent on its anionic transport substrates, nonesterified fatty acids (FAs) (6, 7, 18, 20, 53). However, FAs augment GSIS in β-cells, when exposed for hours (8, 15, 19), but chronically excessive saturated FAs suppress insulin secretion (32, 43, 52), the phenomenon termed lipotoxicity (15, 19). As a simplifying scheme, UCP2 might counteract acute lipotoxicity arising from oxidative stress due to the incoming FAs. Nevertheless, its role should be further clarified.

The role of phospholipases A₂ (PLA₂) (21, 22, 25, 34, 35, 38) residing in (such as iPLA₂β) (34) or recruited to mitochondria of pancreatic β-cells should also be explained in relation to their activation. PLA₂ may amplify lipotoxicity, but in concert with UCP2 a hypothetical synergic antioxidant activity may prevail, such as with mitochondrial iPLA₂γ in heart (25) and lung tissues (21). Both iPLA₂γ and iPLA₂β belong to the group VI of PLA₂s (38) ascribed to the cytosolic Ca²⁺-independent iPLA₂s. They are also termed patatin-like phospholipase domain-containing lipases (PNPLAs), which, besides the release of unsaturated FAs by cleaving the sn-2 ester bond of membrane phospholipids, also cleave saturated FAs at the sn-1 ester bond (21, 38). The β-isoform is sensitive to a stereo-selective inhibitor s-bromoenol lactone (s-BEL), whereas the γ-isoform (PNPLA8) is sensitive to r-BEL (22). Alternative splice isoforms of iPLA₂γ 88-, 77-, 74-, and 63-kDa potentially exist, with the 63-kDa being peroxisomal (35).

The FAs were also implicated in the insulin release in pancreatic β-cells. Although the mechanism is not decisively elucidated (15), the G-protein–coupled receptor 40 (GPR40) has been recently implicated in insulin-stimulating responses (11). In addition, the threshold FA concentration, above which acute lipotoxicity is exerted and below which insulin release is stimulated, is not known.

In this work, we aimed at studying how UCP2-mediated antioxidant action may be initiated and regulated and at elucidating its functional relationship with PLA₂s. We revealed the antioxidant synergy of UCP and iPLA₂γ and describe in detail its impact on acute lipotoxicity and insulin secretion in model pancreatic β-cells, insulinoma INS-1E cells.

Results

Uncoupling manifested by respiration increase after tert-butylhydroperoxide addition to INS-1E cells is absent in cells silenced for UCP2 or iPLA₂γ

Nontransfected INS-1E cells with glucose lowered to 3 mM still exhibited a substantial respiration of 23.6±6.3 pmol O₂·s⁻¹×10⁻⁶ cells (n=49; number of preparations N=10). The corresponding state may be regarded as the phosphorylating state-3, although with low substrate levels available for OXPHOS. Tert-butylhydroperoxide (TBHP) addition resulted in the acceleration of respiration, so that the differences (Δ_TBHPJ_O2) between respiration rates before (J⁰_O2) and after TBHP exhibited a saturating dependence on glucose.

INS-1E cells transfected with nontargeting microRNA (miRNA) (ntgINS-1E cells) and with glucose lowered to 3 mM exhibited virtually unchanged respiration of J⁰_O2=25.3±6.3 pmol O₂·s⁻¹×10⁻⁶cells (n=356; N=51) and responded to TBHP in the same way (Fig. 1A). UCP2-silenced cells (Supplementary Fig. S1A, C; Supplementary Data are available online at www.liebertpub.com/ars) with glucose lowered to 3 mM did not change respiration (J⁰_O2=24.2±7.4 pmol O₂·s⁻¹×10⁻⁶ cells; n=358; N=58). However, they responded much less to TBHP addition with increasing glucose (Fig. 1A). The residual response predominantly reflects contribution of the ADP/ATP carrier (25), since it was prevented by bongkrekic acid, which blocked the whole response in UCP2-silenced cells (Fig. 1A).

FIG. 1.

Tert-butylhydroperoxide (TBHP)-induced respiratory elevation and IMM potential decrease in INS-1E cells. ntgINS-1E cells: black circles or bars; UCP2-silenced cells: dark red squares or bars; iPLA₂γ-silenced cells: green diamonds or bars; iPLA₂β-silenced cells: yellow bars (I). ANOVA *p<0.05; **p<0.01; ***p<0.001. (A, B) Glucose-dependent respiration elevations: Glucose addition to INS-1E cells respiring at state-3 was performed after cell preincubation in a culture medium with 3 mM glucose for 16 h. Subsequently, 250 μM TBHP was added to cells assayed in this medium. The resulting respiration elevation responses Δ_TBHPJ_O2 are plotted versus final glucose concentration (n=3–7 estimations for each concentration, 8–12 for 25 mM glucose). Red-edge symbols: 100 μM bongkrekic acid was present during the assay. (C, D) Glucose-dependent ΔΨ_m decrease: TBHP-induced decreases in ΔΨ_m, δ_TBHPΔΨ_m, were assayed as TMRE fluorescence increase (F_TBHP–F₀)/F₀ between the initial (F₀) and fluorescence after TBHP addition (F_TBHP) in relative units (n=3). Uncoupling of mitochondria by 2.5 μM FCCP prevented δ_TBHPΔΨ_m decreases, which then reached values <1 (data not shown). This also confirms that plasma membrane potential did not affect the assay. (E–I) r-BEL versus s-BEL inhibition of TBHP-induced uncoupling assayed as respiration elevations Δ_TBHP J_O2 (E, F, I); or ΔΨ_m decreases (G, H). The effects of 40 μM r-BEL (“rBEL”) and 40 μM s-BEL (“sBEL”) were quantified for cells remaining at 3 mM glucose (“Glc 3” in E, F, I) and cells subjected to 25 mM glucose (“25” in E, F, I) (n=3, 11 for ntg controls and UCP2-silenced cells). (J) Effect of antioxidants on TBHP-induced uncoupling of INS-1E cell respiration: The assay with 25 mM glucose was conducted as in (A) and with antioxidants: 0.5 μM SkQ1 (“SkQ”); 2 mM N-acetyl-l-cysteine (“NAC”); 7.5 μM ebselen; 200 μM Trolox; and 100 μM MnTBAP. “Catalase” represents catalase overexpression. It should be noted that MnTBAP in the presence of TBHP increased oxygen consumption even in the absence of cells, indicating that the observed effects of MnTBAP are not caused by interacting with cells or mitochondria, but by the formation of butylperoxy radicals, initiating peroxidation of the porphyrin organic groups at the expense of oxygen. ANOVA, analysis of variance; FCCP, trifluoromethoxy carbonylcyanide phenylhydrazone; IMM, inner mitochondrial membrane; MnTBAP, Mn(III)tetrakis(4-benzoic acid)porphyrin; s-BEL, s-bromoenol lactone; TMRE, tetramethylrhodamine ethyl ester; UCP2, uncoupling protein-2.

In addition, iPLA₂γ-silenced cells (Supplementary Fig. S1B, C) with glucose lowered to 3 mM did not significantly change respiration (J⁰_O2=19.4±3.9 pmol O₂·s⁻¹×10⁻⁶ cells; n=116; N=23; compared with own ntg shRNA controls having J⁰_O2=22.4±3.3 pmol O₂·s⁻¹×10⁻⁶ cells; n=117; N=24). They also exhibited suppressed respiratory responses to TBHP addition with increasing glucose and complete inhibition by bongkrekic acid (Fig. 1B).

To verify that the observed respiratory elevations represent a mild mitochondrial uncoupling, we monitored the IMM potential decrease (δ_TBHPΔΨ_m) at various glucose concentrations, after TBHP addition to UCP2- (Fig. 1C) and iPLA₂γ-silenced INS-1E cells (Fig. 1D). We found the same pattern as for respiration increases. Since the accelerated respiration, together with simultaneous ΔΨ_m decline, strictly defines mitochondrial uncoupling, we conclude that TBHP induces uncoupling in INS-1E cells in a UCP2- and iPLA₂γ-dependent manner.

The iPLA₂γ contribution also stems from the sensitivity to its stereo-selective inhibitor r-BEL, which prevented both TBHP-induced J_O2 elevations (Fig. 1E, F and Supplementary Fig. S2A) and ΔΨ_m drops (Fig. 1G, H), whereas both were unaffected by s-BEL (Fig. 1E–H). In contrast, cells silenced for iPLA₂β responded to TBHP normally and this was not inhibited by s-BEL, the stereo-selective inhibitor of iPLA₂β (Fig. 1I). Similar UCP2- (Supplementary Fig. S2B) and iPLA₂γ-dependent uncoupling (not shown) was induced by TBHP at state-4 set by oligomycin. We have also confirmed that the state-3 respiration of mitochondria isolated from INS-1E cells can be uncoupled in a UCP2-dependent manner (Supplementary Fig. S3).

TBHP-induced uncoupling is prevented by antioxidants

As expected, UCP2-dependent TBHP-induced mitochondrial uncoupling in ntgINS-1E cells at 25 mM glucose was completely prevented by matrix-targeted antioxidant SkQ1 and substantially blocked by catalase overexpression (Fig. 1J). Among nonmitochondrial antioxidants, only N-acetyl-l-cysteine and ebselen exhibited moderate prevention. Trolox, an alkyl-peroxyl free radical scavenger, had no effect (Fig. 1J), excluding involvement of lipoperoxidation. A superoxide dismutase (SOD) mimetic, Mn(III)tetrakis(4-benzoic acid)porphyrin (MnTBAP), by direct interaction with TBHP, accelerated TBHP-induced respiration independently of UCP2 presence (Fig. 1J).

iPLA₂γ is directly activated by TBHP or H₂O₂

Reconstituted recombinant iPLA₂γ was reversibly activated by H₂O₂ (Fig. 2A–E) with a half-maximum activation constant AC₅₀ corresponding to ∼1.5 nmol H₂O₂ per nmol of enzyme (Fig. 2E). The H₂O₂-induced FA release in proteoliposomes containing iPLA₂γ was fully reversed by membrane-permeable dithiothreitol (DTT) and partially by membrane-impermeable glutathione (Fig. 2D, E). These results indicate that iPLA₂γ activation by H₂O₂ is dependent on reversible oxidation of accessible cysteine residues. In cells, AC₅₀ was 200 μM for TBHP (Supplementary Fig. S2A).

FIG. 2.

Redox-dependent free FA release in proteoliposomes with recombinant iPLA₂γ (A–E) and in INS-1E cells (F, G). (A, B, D, E) Fluorimetrically indicated free FA release: Passive diffusion (flip) of FAs into the liposome interior was monitored by sulfopropylquinolinium fluorescence, which increases due to intraliposomal acidification concomitant to FA release. The fluorescence increase was induced by 25 μM H₂O₂ (“ +H₂O₂”) in (A, C, D) and was absent without H₂O₂ (“no H₂O₂” in A). Inhibition by 1 μM r-BEL (“ +H₂O₂+RBEL” in A), 10 mM glutathione (“GSH” in D, E), or 10 mM dithiothreitol (“DTT” in D, E) is also shown. (B) Without iPLA₂γ (“no iPLA₂γ”), H₂O₂ caused no intraliposomal acidification (“ +H₂O₂”) and the response was identical to the baseline (“no H₂O₂”) in contrast to two additions of 10 μM linoleic acid (“LA”), causing the instant intraliposomal acidification. Five micromolar bovine serum albumin (“BSA”), removing FAs, led to subsequent alkalization back toward the initial pH values. Proteoliposomes contained 0.04 μg of purchased (A, C) or 0.4 μg of the affinity-purified (D, E) recombinant human iPLA₂γ per mg of phospholipids. (E) The H₂O₂ dose response for redox-stimulated FA-induced acidification rate is shown, yielding AC₅₀ of 0.2 μM H₂O₂, corresponding to 1.4 nmol H₂O₂ per nmol iPLA₂γ. (C, F, G) GC/MS quantification of free FAs cleaved during 60 min at 30°C as induced by 25 μM H₂O₂ (“H₂O₂”) in iPLA₂γ–proteoliposomes (C) with 0.04 μg of the purchased protein (Novus); or free FAs cleaved during 2 min at 37°C in INS-1E cells, ntg or silenced as ascribed (F, G), supplied with 25 mM glucose, as induced by 250 μM TBHP (F) or 75 nmol PA·l0⁻⁶ cells, corresponding to 0.2 pmol free PA (G). Indicated FAs were detected in the absence of H₂O₂ or TBHP (“no add.”), the presence of 25 μM H₂O₂ (“H₂O₂”), 250 μM TBHP (“TBHP”), or 150 μM PA with 5% BSA; or additional 1 μM (C) or 40 μM (F, G) r-BEL (“ +r-BEL”). Averages±s.d. of 3–7 (G) or 4–17 estimations are shown, recalculated relatively to the estimated total content of FA side chains (C) or the initial FA content (F, G). ANOVA: *p<0.1; **p<0.05; ***p<0.001. FA, fatty acid; GC/MS, gas chromatography/mass spectrometry; PA, palmitic acid.

Gas chromatography/mass spectrometry quantification of released FAs from iPLA₂γ-containing proteoliposomes after 60 min at 30°C had indicated a release of approximately 1 nmol⁻¹·min·(mg protein)⁻¹ (1.5% of phospholipid FAs) of free palmitic acid (PA), oleic, or linoleic acid, which was enhanced by H₂O₂, but not with r-BEL (Fig. 2C). GC/MS analysis of TBHP- or PA-induced (vide infra) FA release in ntgINS-1E cells 2 min after the addition of 25 mM glucose is summarized in Figure 2F and G. A substantial increase in free FAs occurred only on iPLA₂γ activation by 250 μM TBHP or 75 nmol PA·l0⁻⁶ cells, corresponding to 0.2 pmol free PA (other FAs increased, Fig. 2G) and was blocked by 40 μM r-BEL (Fig. 2F, G). Along with unsaturated FAs, palmitic (Fig. 2F) and stearic acid (Fig. 2F, G) also were cleaved, which is characteristic of PNPLA subfamily of phospholipases (38). Consequently, we can conclude that H₂O₂- (ex vivo TBHP-) activated or PA-induced (vide infra) iPLA₂γ cleaves FAs off mitochondrial phospholipids and provides them to UCP2 as the transport substrates, ensuring slight uncoupling of mitochondria.

Decrease of superoxide release into mitochondrial matrix on TBHP addition is absent in INS-1E cells silenced for UCP2 or iPLA₂γ

We further investigated whether the observed TBHP-induced uncoupling attenuates mitochondrial superoxide production. We employed MitoSOX Red fluorescence (10), measuring the in situ surplus superoxide release into the mitochondrial matrix as the rate J_m of fluorescence changes in regions of interests identical to the matrix (Supplementary Movie S1). We have evaluated (Fig. 3A, D) how J_m is altered after the immediate TBHP addition to ntgINS-1E cells pre-equilibrated with 25 mM glucose for 10 min (after 16-h preincubation with 3 mM glucose). In spite of the pro-oxidant nature of TBHP, J_m rates decreased to a minimum on TBHP addition (Fig. 3A, D). Such attenuation of mitochondrial superoxide production ceased in the presence of r-BEL, but not s-BEL (Fig. 3A, D), and was not observed in UCP2- (Fig. 3B, C) and iPLA₂γ-silenced INS-1E cells (Fig. 3E, F).

FIG. 3.

Prevention of matrix-released superoxide after TBHP (A–F) or PA (G, H) addition is absent in UCP2- or iPLA₂γ-silenced cells. ntgINS-E cells of the corresponding scrambled sequences: black symbols or bars; UCP2-silenced cells: dark red symbols or bars; iPLA₂γ-silenced cells: green symbols or bars. ANOVA (n=3–6): *p<0.05; **p<0.01; ***p<0.001. Rates of superoxide release into the mitochondrial matrix were assayed by confocal microscopy using MitoSOX Red. The magnitude of J_m in arbitrary units (a.u.) was determined as a slope of linearized fluorescence intensity traces versus time (exemplified in A, B, D, E). Cells were preincubated with 3 mM glucose for 16 h and glucose was raised to 25 mM at the beginning, while the assay was conducted with no other agent (triangles) or as indicated with 25 μM TBHP (diamonds), or with TBHP plus 5 μM r-BEL (squares, “ +rBEL”), or with TBHP plus 5 μM s-BEL (circles, “ +sBEL”). (C, F) Quantifications of J_m rates: The effects of r-BEL were quantified for J_m after TBHP induction of mitochondrial uncoupling (“TBHP”) or without TBHP (“No add.”). (G, H) Quantifications of J_m rates during β-oxidation of palmitate in INS-1E cells: The effects of r-BEL were quantified for J_m after PA induction of mitochondrial uncoupling (“PA”; 15 μM PA) or without PA (“No add.”).

We conclude that iPLA₂γ and UCP2 synergy provides a suppression of mitochondrial superoxide production by mild uncoupling initiated by the TBHP-mediated activation of iPLA₂γ. Consequently, we predict that in pancreatic β-cells the H₂O₂-activated iPLA₂γ/UCP2 synergy provides an antioxidant mechanism, enabling a feedback downregulation of oxidative stress.

iPLA₂γ activation during β-oxidation of palmitate in INS-1E cells

Mitochondrial iPLA₂γ was also activated in ntgINS-1E cells by a matrix ROS/H₂O₂ burst, resulting from the ongoing β-oxidation of the added PA (Fig. 2E). The electron-transfer flavoprotein:ubiquinone oxidoreductase forms superoxide during β-oxidation (41, 46, 50). Superoxide can be instantly transformed to H₂O₂ by SODs. We hypothesized that the resulting H₂O₂ directly activates iPLA₂γ. Our results with the reconstituted recombinant iPLA₂γ (Fig. 2A–C) and experiments described next confirmed this hypothesis. Thus, PA addition induced respiratory elevation (Fig. 4A, B) and ΔΨ_m decrease (Fig. 4C, D), indicating mild mitochondrial uncoupling. Respiratory elevations, due to PA (Δ_PAJ_O2), were more than two times higher in 25 mM glucose than in 3 mM glucose, and occurred even in the presence of s-BEL but not r-BEL (Fig. 4A, B), indicating the iPLA₂γ specificity. Importantly, the PA-induced uncoupling was virtually absent in cells silenced for either UCP2 (Fig. 4A, C) or iPLA₂γ (Fig. 4B, D).

FIG. 4.

iPLA₂γ activation during β-oxidation of palmitate in INS-1E cells. Color coding as in Figure 3. The effects of r-BEL and s-BEL were quantified for PA-induced acceleration of cell respiration, differences (Δ_PAJ_O2) in respiration before and after the addition of 75 nmol PA·l0⁻⁶ cells (A, B) in the presence of 3 mM glucose (“Glc 3”) and 25 mM glucose (“25”), and decreases (δ_PA) in IMM potential ΔΨ_m in the presence of 25 mM glucose (C, D). The ΔΨ_m decrease was assayed as a TMRE fluorescence increase (F_PA–F₀)/F₀ from the initial (F₀) to the elevated fluorescence after PA (F_PA) in relative units. Uncoupling of mitochondria by 2.5 μM FCCP prevented δ_PAΔΨ_m decreases, which then reached values <0.5 (data not shown). Both r-BEL and s-BEL were 40 μM. ANOVA (n=3–12): *p<0.05; **p<0.01; ***p<0.001. (E, F) The effects of palmitoyl-dl-carnitine (E) or dl-α-glycerolphosphate (F). The effects are shown (E) for 40 μM r-BEL or 50 μM etomoxir on respiratory responses to additions of 100 μM palmitoyl-dl-carnitine administered with 2 μl of Lipofectamine 2000 (at 3 mM glucose); (F) for 11 mM dimethylsuccinate (“DMS”) on respiratory responses to additions of 5 mM dl-α-glycerolphosphate at 3 mM glucose (“αGP”) or 25 mM glucose (“αGP25”). ANOVA (n=3–12): *p<0.05; **p<0.01; ns, nonsignificant differences. (G, H) The effects of antioxidants on β-oxidation of palmitate in INS-1E cells: respiration assays at 25 mM glucose such as in panel A (G) or MitoSOX J_m assay (H) were conducted in the presence of antioxidants as indicated: 0.5 μM SkQ1; 2 mM N-acetyl-l-cysteine; 7.5 μM ebselen; 200 μM Trolox; and 100 μM MnTBAP. “Catalase” represents catalase overexpression. ANOVA (n=6): *p<0.05; **p<0.01; ***p<0.001.

Altogether, these data suggest that FAs instantly cleaved off phospholipids by iPLA₂γ to determine the UCP2-mediated uncoupling, and that UCP2 requires such a specific nascent FA delivery (for UCP1 see Ref. 11). This is because no response to PA addition took place in the presence of r-BEL (but not s-BEL) in ntgINS-1E cells. Analogously, a negligible response was also observed in iPLA₂γ-silenced cells where UCP2 had to be intact. We have also demonstrated that PA can be substituted by palmitoyl-dl-carnitine administered to cells by Lipofectamine 2000 (Fig. 4E) but not by dl-α-glycerolphosphate (Fig. 4F). Both the β-oxidation inhibitor etomoxir and r-BEL inhibited palmitoyl-dl-carnitine-induced respiratory rise (Fig. 4E). The dl-α-glycerolphosphate-driven respiration was higher in UCP2-silenced cells and was inhibited by a membrane-permeable succinate analog dimethylsuccinate (Fig. 4F) but not by r-BEL (not shown).

A similar antioxidant effect in the mitochondrial matrix (Fig. 3G, H) as seen with TBHP activation of iPLA₂γ was found for iPLA₂γ activation by H₂O₂, resulting from the pro-oxidant PA β-oxidation (Fig. 5C). The superoxide release (J_m) rates declined to a minimum on the PA addition (Fig. 3G, H). Such attenuation of mitochondrial superoxide production ceased in the presence of r-BEL but not s-BEL (Fig. 3G, H), and this was not observed in UCP2- (Fig. 3G) and iPLA₂γ-silenced cells (Fig. 3H). We conclude that self-attenuation of oxidative stress at FA β-oxidation exists in INS-1E cells due to the UCP2/iPLA₂γ synergy.

FIG. 5.

Cytosolic ROS and extracellular FA monitoring in INS-1E cells. (A–D) Antioxidant effect of UCP2 and iPLA₂γ synergy activated by β-oxidation of palmitate. For color coding, see Figure 3. Rates of cytosolic oxidation J_c (ROS plus oxidized SH residues) were monitored by confocal microscopy using CellROX after additions of 15 μM PA. The J_c rates were taken after 15 min (A, B) or during the first 5 min (C, D). Both r-BEL and s-BEL were 50 μM. ANOVA (n=6–9):*p<0.05; **p<0.01; ***p<0.001. (E, F) TBHP-induced free FA release was detected by confocal microscopy in the extracellular space based on quenching of ADIFAB2 fluorescence. Pericellular FA concentration values (E) were estimated from spectra (exemplified in F) obtained in the regions of interest selected around the outer perimeter of GFP-positive cells before and after the addition of glucose to the final 25 mM and 12.5 μM TBHP, using a dissociation constant of 50 nM, which corresponds to an average value for plasma/serum FAs (cf. ADIFAB2 protocols, FFA Sciences, www.ffasciences.com). Where indicated, 200 nM SkQ1 was present (“SKQ”). GFP, green fluorescent protein; ROS, reactive oxygen species.

FA-induced uncoupling is prevented by antioxidants

The PA-induced iPLA₂γ/UCP2-dependent mitochondrial uncoupling in INS-1E cells at 25 mM glucose was prevented by the matrix-targeted antioxidant SkQ1, by ebselen but not by N-acetyl-l-cysteine (Fig. 4G). Unlike with TBHP, it was also blocked by Trolox and MnTBAP (Fig. 4G), reflecting superoxide and down-stream radical formation by β-oxidation. It was also blocked by catalase overexpression (Fig. 4G), which abolished the resulting attenuation of superoxide formation (Fig. 4H).

Concomitant redox changes in the cytosol of INS-1E cells

We have also studied redox changes in the cytosol of ntgINS-1E cells and confirmed that both PA β-oxidation (Fig. 5A–D) and, as expected, TBHP additions (not shown) oxidize cytosol, when rates of oxidation (J_c) were indicated by the cytosolic ROS fluorescent probe CellROX immediately after insults (Fig. 5C, D). However, when detected ∼15 min after PA addition, the cytosolic oxidation rates J_c were faster in UCP2- (Fig. 5A) and iPLA₂γ-silenced cells (Fig. 5B) when compared with ntgINS-1E cells. This difference ceased in the presence of r-BEL. We interpret that after the initial phase of β-oxidation, producing a substantial burst of superoxide converted thereafter to H₂O₂ penetrating to the cytosol (Fig. 5C, D), the resulting iPLA₂γ activation and the concomitant UCP2/iPLA₂γ antioxidant synergy later gradually reverses the established pro-oxidant state of the cytosol. The overall ROS release rates to the cytosol are therefore attenuated (Fig. 5A, B).

Mitochondrial iPLA₂γ-cleaved free FAs are accessible for plasma membrane

A confocal microscopy assay with the extracellular fluorescent free FA indicator ADIFAB2 demonstrated FA binding to the extracellular ADIFAB2, which ceased in iPLA₂γ-silenced cells (Fig. 5E, F). This demonstrates that FAs cleaved by iPLA₂γ in mitochondria can be exported to bind to the extracellular ADIFAB2 and therefore are also accessible to the entire cell cytosol and plasma membrane.

Role of UCP2/iPLA₂γ synergy in insulin release

Next, we set to analyze in detail the behavior of the glucose sensor. INS-1E cells with glucose lowered to 3 mM responded to the increasing glucose by elevating the respiration in a saturated manner till the final 25 mM glucose (49). ntgINS-1E cells responded in the same way (Fig. 6A). With the UCP2 silencing, beginning at 9 mM glucose, this relationship was shifted to higher respiration increases (Fig. 6A), indicating that mitochondria lacking UCP2 respond to a higher extent on GSIS. It is known that not only respiration but also ΔΨ_m increases after glucose addition (49), and this was observed in ntgINS-1E cells (Fig. 6C). Again, these ΔΨ_m elevations at GSIS were higher in UCP2-silenced cells (Fig. 6C). In contrast, both respiratory (Fig. 6B) and ΔΨ_m responses (Fig. 6D) to glucose in iPLA₂γ-silenced cells were identical to their corresponding ntg controls. We conclude that UCP2 presence in ntgINS-1E cells diminishes respiratory and ΔΨ_m responses to the added glucose. These responses are intimately inherent to the glucose-sensing function of pancreatic β-cells. Hence, the UCP2 presence sacrifices a small portion of the glucose-sensing capacity (function range) for benefits of antioxidant protection. The simplest reasoning predicts that UCP2 does it via a mild uncoupling of mitochondria at state-3 (i.e., at phosphorylating state). This is supported by the observed, slightly increased respiration on UCP2 silencing.

FIG. 6.

Changes after glucose addition to control, UCP2- or iPLA₂γ-silenced cells. For color and symbol coding, see Figure 1. Cells were preincubated for 16 h in a cultivation medium with 3 mM glucose and assayed in it. Responses to the additional glucose with its indicated final concentration are shown for: (A, B) Respiratory elevations (n=9–25; n=110 for 25 mM glucose); ANOVA: *p<0.05; **p<0.01; ***p<0.001; (C, D) IMM potential increases Δ_GlcΔΨ_m (n=3), taken as TMRE fluorescence decreases (F₀–F_Glucose)/F₀ from the initial (F₀) to the decreased fluorescence after glucose (F_Glucose) in relative units. Uncoupling of mitochondria by 2.5 μM FCCP prevented δ_TBHPΔΨ_m increases, which then reached values <0.1 (data not shown). ANOVA: *p<0.05; (E, F) cytosolic ATP levels in the absence or presence of TBHP (250 μM; dashed bars) assayed with no further addition (“Glc 3”) or after glucose adjustment to 25 mM (“Glc 25”). ATP levels were quantified after 60 min. ANOVA (n=4–11): **p<0.05.

However, the rates of glucose-stimulated insulin release in UCP2- or iPLA₂γ-silenced INS-1E cells were equal to ntgINS-1E controls (Fig. 7A). In the presence of TBHP addition, nonlinear (biphasic) insulin release was initially faster and attained higher magnitudes 10–60 min after glucose addition to UCP2- and iPLA₂γ-silenced INS-1E cells (Fig. 7B), demonstrating that unchanged and lower-magnitude insulin release rates in ntgINS-1E controls are depressed (45) in the presence of a pro-oxidant due to the iPLA₂γ/UCP2-initiated uncoupling, similarly as previously observed (48). Hence, a portion of maximum possible insulin-secreting capacity is sacrificed for benefits of antioxidant protection. But independently of TBHP presence, ATP cell levels were equal 60 min after glucose addition, to all studied cells, to ntgINS-1E, UCP2- (Fig. 6E) and iPLA₂γ-silenced cells (Fig. 6F). This implies that mild uncoupling is not able to substantially block OXPHOS, hence ATP synthesis in the long term. It should be noted, however, that TBHP alone accelerated insulin release in the absence of glucose, causing the redox-stimulated insulin secretion (RSIS) (cf. dashed fits in Fig. 7B vs. 7A).

FIG. 7.

Glucose-stimulated insulin release in the absence or presence of pro-oxidant insult. Changes in insulin release into Krebs-Ringer HEPES buffer were assayed: (A) in the absence or (B) presence of TBHP (25 μM); or (Ca, Cb, Cc) 15 μM and (D) 150 μM PA, all added together with glucose, during 60 min after raising glucose to 25 mM or without glucose addition (dashed fits and semi-filled symbols or gray circles or black circles for ntg INS-1E cells). Black symbols or fits: ntg INS-1E cells; dark red symbols or fits: UCP2-silenced cells; green symbols or fits: iPLA₂γ-silenced cells. (Ca, Cb, Cc) Yellow-filled symbols indicate data of GPR40-silenced cells; the gray-filled symbols represent measurements in the presence of GPR40 antagonist GW1100, added at the beginning. (E, F) Typical components extracted from (B) are shown in (E) and of (A, C) in (F), for GPR40-relevant fatty acid-stimulated insulin secretion (FASIS via GRPR), redox-stimulated insulin secretion (RSIS) and GSIS. Before the experiment, cells were routinely cultivated with 11 mM glucose, transferred to the Krebs-Ringer HEPES buffer, and incubated in it for 10 min before glucose addition. Rates of insulin release were 3.7±0.2, 4.2±0.5, or 3.6±0.4 ng·min⁻¹×10⁻⁶ cells in the absence of TBHP and 6.3±1.5, 9±3, or 12±5 ng·min⁻¹×10⁻⁶ cells during first 20 min in the presence of TBHP in ntg controls, UCP2-silenced or iPLA₂γ-silenced cells, respectively. (C) Rates of insulin release with 15 μM PA were 16±2, 2.5±0.1, or 2.4±0.1 ng·min⁻¹×10⁻⁶ cells in the absence of glucose and 22±8, 12±3, or 5.3±0.8 ng·min⁻¹×10⁻⁶ cells during first 20 min with glucose in ntg controls, UCP2-silenced or iPLA₂γ-silenced cells, respectively. GPR40, G-protein–coupled receptor 40; GSIS, glucose-stimulated insulin secretion.

Similar to TBHP, but to a much higher extent, externally added PA extremely elevated insulin release (also together with H₂O₂/ROS generated during its pro-oxidant β-oxidation). PA induced higher rates of insulin release and its nonlinear time course even in the absence of glucose in ntgINS-1E cells (Fig. 7Ca, black dashed fit). This was not altered in cells with silenced phospholipase D (Supplementary Fig. S4). Surprisingly, in UCP2- and iPLA₂γ-silenced INS-1E cells, PA promoted much less of the glucose-insensitive insulin release, regarding to an extent and also rates, for iPLA₂γ-silenced cells (Fig. 7Cb, Cc, dark red dashed or green dashed fit). At 15 μM PA (∼4 pmol×10⁻⁶ cells), glucose only slightly further increased the already fast insulin release in ntgINS-1E and also the relatively slower insulin release in iPLA₂γ-silenced cells (Fig. 7Cb, Cc dark red or green fit). In UCP2-silenced cells, 15 μM PA with glucose promoted an insulin release extent, attaining ∼50% of levels in ntgINS-1E cells after 60 min. It should be noted that at 15 μM PA, the majority of PA is bound to the albumin present and partitioned into membranes, whereas only ∼1.3 nM PA was free in our experiment, representing a concentration comparable to normal plasma levels. The time course was similar to that with the TBPH pro-oxidant insult (Fig. 7Cb vs. 7B). Similar results were obtained with 30 μM PA (not shown). To simulate acute lipotoxicity, we used a much higher PA (150 μM; exceeding 200 nM free PA). Here, the net GSIS was reduced down to 25% (Fig. 7D), indicating such an acute lipotoxicity that overcomes the stimulating effect of PA. As a result, the insulin release in the absence of PA was equal to the one with PA in UCP2- and iPLA₂γ-silenced cells.

Redox-activated iPLA₂γ supplies FAs for GPR40-stimulated insulin release

Surprisingly, the GPR40, previously indicated to act in insulin-stimulating responses (11), turned out to be a major player in excessive rates of PA–insulin release in ntgINS1-E cells (Fig. 7Ca). This conclusion stems from the finding that in GPR40-silenced cells (Fig. 7Ca, yellow symbols) and in ntgINS1-E cells subjected to the GPR40 antagonist GW1100 (Fig. 7Ca, gray symbols) the PA-induced insulin release was profoundly inhibited to the levels induced by TBHP (Fig. 7B) in both the absence and presence of glucose. In UCP2-silenced cells, where intermediate PA-induced insulin release rates versus ntg controls were assessed, GW1100 also slightly inhibited the PA-induced insulin release (Fig. 7Cb). In contrast, in iPLA₂γ-silenced INS-1E cells, GW1100 had no effect in the presence of glucose and its effect was very small in the absence of glucose (Fig. 7Cc).

Taken together, we interpret these results as follows: In the situation where, promoted by sole FA β-oxidation, ROS activate insulin release in the absence (the net RSIS; Fig. 7E, F) or presence of glucose (RSIS plus GSIS, Fig. 7E, F), the lack of iPLA₂γ evidently causes that no further FAs are cleaved, in contrast to the ongoing cleavage in ntg or UCP2-silenced cells. In ntg cells, it is not the added PA that stimulates GPR40, but FAs cleaved off phospholipids by iPLA₂γ. Lysophosphatidic acid is excluded from GPR40 stimulation and subsequent insulin secretion, since silencing of phospholipase D did not change the observed fatty acid–stimulated insulin secretion (FASIS; Supplementary Fig. S4).

Discussion

Chronic lipotoxicity is a situation of persistent oxidative stress in pancreatic β-cells originating from the long-term intake of excessive FAs and lipid peroxidation products, which leads to GSIS impairment (15, 19). At low doses, FAs, however, stimulate insulin release (Fig. 7C) (15). Studying immortalized but noncancer pancreatic β-cells, INS-1E cells (37), we demonstrated that a mild acute oxidative stress originates from the β-oxidation of FAs. This elevates superoxide in the mitochondrial matrix as well as H₂O₂ in the cytosol and simultaneously initiates a feedback downregulating mechanism that is capable of counteracting such a repeatable ROS burst given by postprandial physiological acute FA intake. Moreover, a different spectrum of FAs cleaved off mitochondrial phospholipids by iPLA₂γ can spread within the cell serving as another messenger (as revealed by the ADIFAB2 assay, Fig. 5E, F).

The antioxidant mechanism is provided by a synergy between H₂O₂-activated mitochondrial iPLA₂γ and UCP2. The latter provides mild uncoupling, which attenuates the otherwise excessive superoxide production, the one resulting from β-oxidation at an un-attenuated state. The iPLA₂γ/UCP2 antioxidant synergy is suited to provide a feedback downregulation of oxidative stress. It may counteract any pro-oxidant insult and shift of redox balance in the cell that comes over a certain threshold, that is, leading to H₂O₂ elevation sufficient for iPLA₂γ activation. Our model (Fig. 8A) is based on the finding that iPLA₂γ is directly and reversibly activated by H₂O₂ (ex vivo by TBHP). Consequently, on H₂O₂ burst, iPLA₂γ (or another isoform, if activated, but apparently not iPLA₂β) ensures an antioxidant function in concert with UCP2, which then attenuates mitochondrial superoxide formation by mild uncoupling (4, 18, 36) (Fig. 3). The consequent lowering of the inevitable H₂O₂ flow from mitochondria to the cell cytosol allows the suddenly released “spare” cytosolic antioxidant capacity to better cope with oxidative stress in the cell cytosol (24). This is the key principle behind how antioxidant protection from IMM is spread toward the cell (24).

FIG. 8.

Involvement of mitochondrial iPLA₂γ in antioxidant protection and amplification of FA-induced insulin release. (A) Antioxidant mechanism based on UCP2 and iPLA₂γ synergy that is also protective against lipotoxicity in pancreatic β-cells. Acute lipotoxicity leads to FA β-oxidation in mitochondria. Not only the higher substrate load for the respiratory chain (Complexes I, III, IV) contributes to higher superoxide release, but also the electron-transferring flavoprotein:ubiquinone oxido-reductase produces superoxide on β-oxidation (42). Superoxide can be transformed to H₂O₂ by superoxide dismutases, SOD2/MnSOD in the matrix, or SOD1/CuZnSOD in the mitochondrial intermembrane space and in the cytosol. Subsequently, H₂O₂ activates iPLA₂γ, which cleaves both saturated and unsaturated FAs off IMM phospholipids. Only these nascent FAs induce UCP2-mediated mild uncoupling, which inherently suppresses respiratory chain superoxide formation. Thus, iPLA₂γ can be also activated by external oxidative stress as well as by redox signaling pathways, elevating H₂O₂ levels in its vicinity. Ca_L, L-type Ca²⁺ channel; K_ATP, ATP–sensitive K⁺ channel; O₂^•−, superoxide; Rcarn, acylcarnitine; RCoA, acyl-CoA. (B) Involvement of iPLA₂γ in amplification of FA-induced GPR40 pathway of insulin secretion. After activation of iPLA₂γ by H₂O₂ generated due to β-oxidation of incoming palmitic acid, FAs released from mitochondrial membranes provide feedback amplification of GPR40 response, leading to insulin surplus secretion. Lysophospholipids may also be cleaved by specific mitochondrial phospholipase D (mtPLD) (14). Phosphatidic acid (PhosA) released from normal phospholipids by mtPLD might be further processed by phosphatidic acid preferring PLA1 (PAPLA1). All mentioned lipid metabolites are involved in the glycerol lipid/FA cycle signaling stimulating insulin secretion (44).

The substantial extent of respiration and insulin release (48, 54) are sacrificed and utilized for such antioxidant function. The respiration extent of the β-cell glucose sensor corresponds to ∼30% of state-3 respiration (Fig. 6A), whereas ∼15% is sacrificed for antioxidant mild uncoupling (Fig. 6A, differences between ntg and UCP2–silenced cells). Hence, till 50% of maximum sensor capacity is utilized for the antioxidant role. Deriving from elevations of insulin secreted after 40–60 min with TBHP in UCP2-silenced versus ntg cells (Fig. 7A, B), till 50% insulin surplus released in the absence of the antioxidant mechanism is also sacrificed for antioxidant protection. Nevertheless, the pro-oxidant impulse enhances insulin release rates in the absence of glucose (Fig. 7B, C) (31, 42). Details of insulin release at the glucose or FA intake should be further studied within the more complex system of the Langerhans islets possessing numerous autocrine and paracrine interrelationships. They evoke certain limitations of our in vitro system used.

Rather complex results were obtained for FASIS or FASIS combined with GSIS (Fig. 7Ca, E, F). At low (15 μM, corresponding to 1.3 nM free) PA, the overall FASIS+GSIS is lower in UCP2- and iPLA₂γ-silenced cells versus ntg cells, in spite of the tighter mitochondrial coupling versus ntg cells. The two components are involved in FASIS. The first one stems from the insulin release activated via the GPR40 pathway (11) (Fig. 7C). The second one represents the net RSIS and is solely manifested in iPLA₂γ-silenced cells without glucose. The GPR40 pathway of insulin secretion can be activated and is strongly amplified (Fig. 8B) by FAs cleaved off phospholipids by iPLA₂γ. Lysophospholipids remaining after the iPLA₂γ reaction could be hypothetically cleaved by specific mitochondrial phospholipase D (mtPLD) (14), but as demonstrated this does not affect FASIS (Supplementary Fig. S4). FAs are involved in the glycerol lipid/FA cycle signaling stimulating insulin secretion (44).

The iPLA₂γ is most likely initially activated in vivo by the pro-oxidant β-oxidation of FAs incoming from the blood. During FA β-oxidation, activation of iPLA₂γ by H₂O₂ burst arises from the electron-transfer flavoprotein:ubiquinone oxidoreductase (41, 46, 50) and SODs activities. The situation resembles a self-accelerating cycle, since more active iPLA₂γ provides more FAs for further β-oxidation, which further elevates ROS/H₂O₂. Besides iPLA₂γ, these surplus ROS/H₂O₂ themselves trigger insulin secretion and, in parallel, persistently cleaved FAs promote GPR40 activation. All these contributors add on the resulting overstimulation of insulin secretion (Fig. 7Ca), which is permanent (unlike with a single bolus of TBHP) until the added PA is metabolized. In its intermediate state, certain overstimulation is also possible with ablated UCP2 but logically not with ablated iPLA₂γ. We estimated that at 15 μM PA plus glucose, the insulin release induced during 60 min consists of ∼30% net GSIS, <10% net RSIS, and 60% GPR40 pathway (Fig. 7F). The activation of the GPR40 pathway vanished with ablated iPLA₂γ. Without UCP2, hence at a more coupled state without antioxidant protection, the net GSIS increases to >50% and RSIS is doubled (Fig. 7F). Similar 50% net GSIS existed with TBHP (Fig. 7E), inducing the net RSIS in the absence of glucose.

We also conclude that UCP2-mediated mild uncoupling is initiated only when free FAs are directly cleaved off the phospholipids at the UCP2 vicinity (7) by iPLA₂γ (for UCP1 see Ref. 12). In other words, only the nascent FAs are capable of initiating UCP2-mediated uncoupling and a simple PA addition to cells is insufficient (Fig. 4B). FAs are considered the UCP2 cycling/wobbling substrates/cofactors, respectively, essential for their uncoupling function (6, 7, 12, 18, 20, 53).

Demonstrating UCP2/iPLA₂γ antioxidant synergy and initiation of UCP2-mediated mild uncoupling by a nascent FA, we have resolved some contradictions that have been reported during the past decade for UCP2 role in pancreatic β-cells. Various reports of ROS-activated UCP2 function (1, 28) can be alternatively explained by the ROS-initiated iPLA₂γ activation providing FAs for UCP2. Previously, a mild uncoupling in mitochondria isolated from INS-1E cells was also linked to UCP2, while accounting for approximately 30% of H⁺ leak (1, 2). Unlike negative demonstrations (13), we now show that UCP2 is activated exclusively by nascent FAs. This ensures that the net glucose sensing in pancreatic β-cells is not substantially affected by mild uncoupling, unless acute lipotoxicity and/or excessive oxidative stress activate the nascent FA delivery toward UCP2 by iPLA₂γ. One would predict that the lack of either UCP2 or iPLA₂γ must result in an easier reach of the oxidative stress threshold from which a pathological outcome is initiated. These characteristics were reported for UCP2 ablation (29, 42, 48).

The unique feature of the iPLA₂γ protective role is represented by the fact that iPLA₂γ is not constitutively active. Hence, without redox signaling or excessive oxidative stress (such as resulting in acute lipotoxicity), iPLA₂γ does not contribute to mitochondrial uncoupling and to the diminished insulin release (Fig. 7A, B). In turn, when activated, its synergy with UCP2 leads to the antioxidant protection at the expense of insulin release. Unless redox activated, no other mitochondria-localized PLA₂ isoform can substitute iPLA₂γ in these roles (21, 25). We excluded redox activation of iPLA₂β, previously found in β-cell mitochondria (34).

In conclusion, our results elucidated iPLA₂γ- plus UCP2-dependent regulation of mitochondrial superoxide production and insulin release in pancreatic β-cells. We describe a mechanism of free nascent FA-mediated antioxidant function, due to the synergic action of iPLA₂γ and UCP2, counteracting mild sub-threshold lipotoxicity in pancreatic β-cells. The iPLA₂γ-cleaved FAs may also participate in the lipid-mediated signaling such as in amplification of FA-induced insulin release via the GPR40 pathway.

The revealed redox activation of iPLA₂γ and the requirement of nascent FAs for UCP2-mediated uncoupling ensure that the glucose sensing in pancreatic β-cells is not substantially affected by mild uncoupling, unless acute lipotoxicity and/or excessive oxidative stress activate the iPLA₂γ/UCP2 antioxidant synergy. This synergy developed as an inherent antioxidant, cytoprotective, and specifically anti-lipotoxic mechanism that is protective before reaching a critical threshold and provides defense against long-term dysbalanced conditions, such as lipotoxicity. When these pro-oxidant impulses (stimulating otherwise antioxidant protection) are compromised, the inevitable self-accelerating vicious cycle leads to permanent oxidative stress, cell dysbalance, and the progression of type-2 diabetes pathology (15, 23). This progression has to be further elucidated and extended to future translational research.

Materials and Methods

Chemicals

r-BEL, s-BEL, and Trolox were purchased from Cayman (Ann Arbor, MI); MnTBAP was from Merck Millipore (Billerica, MA). All other reagents were from Sigma (St. Louis, MO).

Cell cultivation

Rat insulinoma INS-1E cells were a kind gift from Prof. Maechler, University of Geneva (37). Cells were cultivated with 11 mM glucose in RPMI 1640 medium with 2 mM l-glutamine, 10 mM HEPES, 2 g·l⁻¹ sodium bicarbonate, 1 mM sodium pyruvate, 5% (v/v) fetal calf serum, 50 μM mercaptoethanol, 50 IU·ml⁻¹ penicillin, and 50 μg·ml⁻¹ streptomycin (49). Cells were preincubated for 16 h in cultivation medium containing 3 mM glucose. The desired final glucose concentration was adjusted just immediately before each measurement.

Cell transfections and silencing

A BLOCK-iT Pol II miR RNAi system (Life Technologies, Carlsbad, CA) has been employed to express miRNAs against UCP2. The two miRNA sequences, 5′-TACAGA GTCGTAGAGGCCAATGTTTTGGCCACTGACTGACAT TGGCCTACGACTCTGTA-3′ and 5′-ATTTCGGGCAAC ATTGGGAGAGTTTTGGCCACTGACTGACTCTCCCAA TTGCCCGAAAT-3′, were designed using the BLOCK-iT RNAi Designer and annealed into double-strand oligonucleotides, which were inserted into the linearized miRNA expression vector pcDNA6.2-GW/EmGFP-miR. Individual miRNAs were chained up into tandem constructs. The vector with UCP2-miRNA or control miRNA plasmids was cloned by Gateway BP/LR reaction into pLenti6.2/V5-DEST expression vector, and final constructs were validated by DNA sequence analysis. Lentiviral expression plasmids were co-transfected with the ViraPower™ Packaging Mix (Life Technologies) into the 293LTV cells using Lipofectamine 2000 (Life Technologies). The lentiviral stock was used to transduce INS-1E cells, followed by selection of a stably transduced cell line by blasticidin and verification by a green fluorescent protein (GFP) inherent cytosolic reporter.

For catalase overexpression (17), pZeoSV2(+) vector bearing human catalase sequence (gift from C. Glorieux and Prof. J. Verrax, Université Catholique de Louvain, Belgium) was transiently transfected into INS-1E cells using Lipofectamine 2000 (2.5 ml·mg⁻¹ DNA).

To silence iPLA₂γ or iPLA₂β, pGFP-V-RS vectors (Origene, Rockville, MD) as specific expression cassettes were used with inserted 29-mer shRNAs, 5′-CAAACACTGGC ACTCTTCAAGCAACCATGTCAAGAGCATGGTTGCT TGAAGAGTGCCAGTGTTTG-3′ or 5′-CAGTAGTGTCA CCAACTTGTTCTCAAACCTCAAGAGGGTTTGAGAAC AAGTTGGTGACACTACTG-3′, respectively. The same vectors with inserted scrambled shRNA expression cassettes served as negative controls. INS-1E cells were transiently transfected using Lipofectamine 2000 (2.5 ml·mg⁻¹ DNA) and verified by the GFP cytosolic reporter. For GPR40 (FFAR1) or phospholipase PLD6 silencing, we used Silencer Select siRNAs (Life Technologies), cat. No. 4390771, IDs s141419 and s141420; or IDs s142381 and s142382. They were used in comparison versus negative control siRNA (cat. number 4390843). RNA oligonucleotides (20+20 pmol of siGPR40 vs. 40 pmol of negative control) were used.

High-resolution respirometry and fluorescent assays for ΔΨ_m and ATP monitoring

Cells were trypsinized, counted by a Countess Automated Cell Counter (Life Technologies), and adjusted to 2×10⁶ cells/ml. Respiration was detected by an Oxygraph 2k (cells at 37°C, isolated mitochondria at 30°C; Oroboros, Innsbruck, Austria). Rates taken 3 min after sequential additions of glucose, TBHP, or PA (when applied) were corrected for KCN-insensitive respiration. Five micromolars of tetramethylrhodamine ethyl ester (TMRE; Life Technologies) added immediately before the assay were used for ΔΨ_m monitoring on an RF 5301 PC spectrofluorometer (Shimadzu, Tokyo, Japan) in a stirred 2 ml cuvette with excitation at 561 nm (3 nm slit) and emission at 579 nm (5 nm slit). Trifluoromethoxy carbonylcyanide phenylhydrazone (FCCP) was titrated to a maximum (respiration) or added twice (ΔΨ_m). ATP quantification was performed using ATP Assay bioluminescence kit HSII (Roche, Basel, Switzerland) and the spectrofluorometer at 557 nm (5 nm slit).

Confocal microscopy

A confocal inverted fluorescent microscope Leica TCS SP2 AOBS was employed, with a PL APO 100×/1.40–0.70 oil immersion objective (pinhole 1 Airy unit) and a sample chamber set to 37°C supplied with 5% CO₂. Cells were seeded on poly-l-lysine-coated 22 mm glass coverslips (Marienfeld, Lauda-Königshofen, Germany) at 0.25×10⁶ cells/coverslip 2 days before the experiment or at 0.4×10⁶ cells/coverslip 1 day before the transient transfection and 3 days before the experiment.

Confocal microscopy monitoring of cytosolic ROS and extracellular FAs

Cytosolic ROS monitoring was ensured by 10 μM CellROX (Life Technologies) after 30 min loading, while excited at 633 nm with a 1.2 mW HeNe laser and collecting emission between 650 and 700 nm. An ADIFAB2 FA indicator (FFA Sciences, San Diego, CA) was used to monitor extracellular free FAs, while excited with a 405 nm diode laser and collecting emission between 430 and 645 nm.

Confocal microscopy monitoring of superoxide released into mitochondrial matrix

A triphenylphosphonium-conjugated dihydroethidine, MitoSOX Red (Life Technologies), was used to monitor rates (J_m) of in situ surplus superoxide release into the mitochondrial matrix (10). The surplus represents the portion of superoxide not neutralized by the matrix Mn-SOD. Using rates, any variations in initial background fluorescence are eliminated and the method is feasible for semiquantification of matrix-released superoxide even at low or collapsed ΔΨ_m (10). Excitation was at 488 nm by a 20 mW Argon laser, with emission collected between 550 and 650 nm, while the GFP reporter was checked between 508 and 516 nm. Cells were loaded with 4 μM MitoSOX for 15 min. A series of confocal images were taken every 30 s for 20 min (Supplementary Movie S1). Regions of interest corresponding to mitochondria were selected using the Ellipse software (ViDiTo, Košice, Slovakia). Changes of integrated fluorescence intensity were quantified from plots of fluorescence in the selected areas versus time.

Assay for insulin release

Cells were seeded at 0.4×10⁶ cells/well in poly-l-lysine-coated 12-well plates 1 day before the transient transfection and 3 days before the experiment. Insulin release was assayed using Rat Insulin ELISA kit (U-E type; Shibayagi, Shibukawa, Japan) in Krebs-Ringer HEPES buffer containing 2 mM glutamine and 0.1% bovine serum albumin after 10 min of preincubation before glucose addition.

SDS-PAGE and Western blotting

Silencing was verified by SDS-PAGE followed by Western blotting on mitochondria isolated from cells and using primary antibodies against UCP2 (No. 662047; Merck Millipore) or in cell lysates for iPLA₂γ (PNPLA8 N-terminal., AP4706a; Abgent, San Diego, CA); otherwise, ATP-synthase β-subunit (ATPB, ab14730; Abcam, Cambridge, MA), TIM23 (611222; BD Biosciences, Franklin Lakes, NJ), β-actin (ab3280; Abcam), and horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit secondary antibody, respectively (Sigma) were used. Blots were visualized using an LAS-1000 Imager, and quantitative image analysis was performed with ImageJ software.

iPLA₂γ reconstitution

The recombinant human iPLA₂γ was reconstituted into liposomes by adopting published protocols (20). The enzyme was purchased (PNPLA8; Novus Biologicals, Littleton, CO) or was cell expressed from a vector encoding His-tagged enzyme (GeneCopoeia, Rockville, MD) and affinity purified using a nickel resin column (Qiagen, Germantown, MD). Ten microgram per milliliter of Novus protein was dissolved in 1% n-lauroylsarcosine, passed through a Sephadex G-25-300 column, and equilibrated with an assay medium plus 0.1% n-octylpentaoxyethyelene (Bachem, Bubendorf, Switzerland). One hundred microgram per milliliter of the affinity-purified protein had to be treated with 10 mM DTT and dialyzed against the assay medium, in order to reversibly modify the protein to its reduced form. A Sephadex G-25-300 column (0.5% n-octylpentaoxyethyelene) was used to remove DTT. Aliquots in μg·(mg lipid)⁻¹ (0.04 Novus protein; 0.4–0.8 affinity-purified protein) were mixed with phospholipids (Escherichia coli total lipid extract supplemented with 20% bovine heart cardiolipin; Avanti Polar Lipids, Alabaster, AL), to yield 28% cardiolipin, 46% phosphatidylethanolamine, 12% phosphatidylglycerol plus 14% other lipid, and 1 mg·(mg lipid)⁻¹ n-octylpentaoxyethyelene. Proteoliposomes were formed by detergent removal using Bio-Beads SM-2 (Bio-Rad, Hercules, CA). Both the intraliposomal and assay medium contained Na⁺ salts of TES (50 mM), SO₄²⁻ (80 mM), and EDTA (2 mM), pH 7.5. Sulfopropylquinolinium-based monitoring of FA release into the liposome interior or GC/MS analyses of cleaved FAs was performed.

Quantification of free FAs

Free FAs released in liposomes or cells were quantified using GC/MS (6890 gas chromatograph, 5973 mass spectrometer; Agilent Technologies, Palo Alto, CA). Reaction mixtures were extracted with acidic 2-propanol/n-heptane, treated with ether/diazomethane at room temperature for 20 min, and evaporated under argon. The methylated FAs were reconstituted in 100 μl of n-heptane and analyzed by GC/MS. FAs were identified/quantified in comparison with spectra of purified standards. FA total content in phospholipids was evaluated after hydrolysis by 0.5 M NaOH at 60°C for 30 min, and after acidification and extraction.

Abbreviations Used

BSA

bovine serum albumin

Ca_L

L-type Ca²⁺ channel

DTT

dithiothreitol

FA

fatty acid

FASIS

fatty acid-stimulated insulin secretion

FCCP

trifluoromethoxy carbonylcyanide phenylhydrazone

GC/MS

gas chromatography/mass spectrometry

GFP

green fluorescent protein

GPR40

G-protein–coupled receptor 40

GSIS

glucose-stimulated insulin secretion

IMM

inner mitochondrial membrane

K_ATP

ATP-sensitive K⁺ channel

miRNA

microRNA

MnTBAP

Mn(III)tetrakis(4-benzoic acid)porphyrin

mtPLD

mitochondrial phospholipase D

ntg

nontargeting

OXPHOS

oxidative phosphorylation

PA

palmitic acid

PAPLA1

phosphatidic acid preferring PLA1

PNPLAs

patatin-like phospholipase domain-containing lipases

r-BEL

r-bromoenol lactone

Rcarn

acylcarnitine

RCoA

acyl-CoA

ROS

reactive oxygen species

RSIS

redox-stimulated insulin secretion

s-BEL

s-bromoenol lactone

SOD

superoxide dismutase

TBHP

tert-butylhydroperoxide

TMRE

tetramethylrhodamine ethyl ester

UCP

uncoupling protein

Acknowledgments

This project was supported by grants of the Grant Agency of the Czech Republic (P303/11/P320 to J.J.; 15-15-02051S to M.J.; 13-02033 and 13-06666 to P.J.) and by the research project RVO67985823. The authors gratefully acknowledge the excellent technical assistance of Lenka Josková with cell cultivation, Jana Vaicová with immunoblotting and ELISA, and Jitka Smiková with plasmid isolations. The authors express their gratitude to Prof. Libor Vítek, M.D., Ph.D., from the First Faculty of Medicine, Charles University, Prague, for kindly providing them with GC/MS and to Prof. Vladimir Skulachev, from Moscow University, Russia, for providing them with SkQ1.

Author Disclosure Statement

No competing financial interests exist.