Pancreatic β-cells detoxify H₂O₂ through the peroxiredoxin/thioredoxin antioxidant system

Abstract

Oxidative stress is thought to promote pancreatic β-cell dysfunction and contribute to both type 1 and type 2 diabetes. Reactive oxygen species (ROS), such as superoxide and hydrogen peroxide, are mediators of oxidative stress that arise largely from electron leakage during oxidative phosphorylation. Reports that β-cells express low levels of antioxidant enzymes, including catalase and GSH peroxidases, have supported a model in which β-cells are ill-equipped to detoxify ROS. This hypothesis seems at odds with the essential role of β-cells in the control of metabolic homeostasis and organismal survival through exquisite coupling of oxidative phosphorylation, a prominent ROS-producing pathway, to insulin secretion. Using glucose oxidase to deliver H₂O₂ continuously over time and Amplex Red to measure extracellular H₂O₂ concentration, we found here that β-cells can remove micromolar levels of this oxidant. This detoxification pathway utilizes the peroxiredoxin/thioredoxin antioxidant system, as selective chemical inhibition or siRNA-mediated depletion of thioredoxin reductase sensitized β-cells to continuously generated H₂O₂. In contrast, when delivered as a bolus, H₂O₂ induced the DNA damage response, depleted cellular energy stores, and decreased β-cell viability independently of thioredoxin reductase inhibition. These findings show that β-cells have the capacity to detoxify micromolar levels of H₂O₂ through a thioredoxin reductase–dependent mechanism and are not as sensitive to oxidative damage as previously thought.

Introduction

Although oxidative stress in pancreatic β-cells has been widely implicated in the pathogenesis of both type 1 and type 2 diabetes, reactive oxygen species (ROS),⁶ under physiological conditions, are important signaling molecules necessary for maintaining cellular homeostasis (1). If the balance between ROS production and detoxification is disturbed, then ROS can accumulate and cause protein, lipid, and DNA oxidation, interfering with protein function and leading to DNA damage and eventual cell death (2). Numerous studies have reported elevated markers of oxidative damage, including 8-hydroxy-deoxyguanine, hydroperoxides, and oxidation of DNA bases, in islets of diabetic rodents and patients (3–10).

ROS, including superoxide and hydrogen peroxide (H₂O₂), are primarily produced during mitochondrial oxidative phosphorylation (11, 12). In β-cells, oxidative phosphorylation is tightly coupled to insulin secretion, where more than 90% of the carbons in glucose are oxidized to CO₂ on substrate supply (13–15). This is because glucokinase, the rate-limiting enzyme for glucose metabolism in the β-cell, has a much lower affinity for glucose than other hexokinases, allowing the β-cell to adjust the rate of glucose metabolism and, thus, the rate of insulin secretion in response to small changes in blood glucose concentration (16). For these reasons, the rate of mitochondrial oxidation in the β-cell is directly proportional to the blood glucose concentration. It is widely accepted that, in type 2 diabetes, when blood glucose is chronically elevated, increased oxidative phosphorylation results in elevated ROS production in β-cells. In support of this idea, treatment of rat and human islets with high glucose for 72 h increases intracellular peroxides compared with islets incubated under low-glucose conditions (17).

Pancreatic β-cells are considered to be particularly vulnerable to oxidative stress because of relatively low expression of antioxidant enzymes, including catalase, superoxide dismutase, and GSH peroxidase, compared with levels in liver and kidney (18–21). Indeed, overexpression of superoxide dismutase, catalase, GSH peroxidase, or thioredoxin in mouse β-cells affords protection against oxidative damage induced by alloxan, the combination of xanthine oxidase and hypoxanthine, streptozotocin, or H₂O₂ provided as a bolus (22–26). However, it stands to reason that, evolutionarily, β-cells should be protected from oxidative stress because they release a hormone that is essential for survival and that is controlled by the primary ROS-producing pathway: mitochondrial oxidative metabolism. Further, β-cells have a limited capacity for self-renewal. Consistent with this view, recent studies have shown that peroxiredoxins, a class of antioxidants capable of reducing H₂O₂, lipid peroxides, and peroxynitrite, are expressed in β-cells and, when overexpressed, protect β-cells from oxidative stress (27–31). However, the roles of endogenously expressed antioxidants in protecting β-cells from H₂O₂-induced stress, and the mechanisms by which this occurs, are unclear.

It is common practice to use a single or repeated bolus of H₂O₂ when studying β-cell responses to the oxidant (32). Although the physiological relevance of this delivery method can be debated, when delivered as a one-time bolus, we found that H₂O₂ exhausts the antioxidant defenses of the cell, leading to accumulation of DNA damage, depletion of intracellular energy stores, and cell death. To more closely mimic oxidant accumulation in a native setting, we used glucose oxidase to continuously deliver H₂O₂ over time and demonstrate that pancreatic β-cells are capable of detoxifying micromolar levels of the oxidant when delivered in this manner. These findings suggest that, in response to a more physiologically relevant system of H₂O₂ delivery, β-cells possess an antioxidant defense pathway that has been masked by use of nonphysiological bolus delivery of oxidants. We further provide evidence to support thioredoxin reductase and peroxiredoxins as the mechanism by which β-cells maintain this robust antioxidant defense pathway that provides protection from oxidative damage.

Results

Pancreatic β-cells can detoxify H₂O₂ when delivered continuously

When examining the responses of β-cells to H₂O₂, it is common to use a single or repeated dose of diluted H₂O₂ added to the cell culture medium (32). However, when 100 μm H₂O₂ is added as a bolus to our medium conditions, more than 50% disappears after 5 min, and nearly the entire dose is gone within 30 min, regardless of the presence of cells (Fig. 1A). This is likely due to the presence of pyruvate, which rapidly reacts with H₂O₂ to produce acetic acid, water, and carbon dioxide (33). Because of these complicating factors, we sought to develop a culture system to study β-cell responses to H₂O₂ produced under more physiological conditions. To address this goal, glucose oxidase was used to continuously generate this oxidant in the medium. In contrast to bolus delivery, H₂O₂ generation in response to treatment with 20 milliunits/ml glucose oxidase increases in a time-dependent manner to levels that reach 50 μm following 30-min incubation in the absence of cells (Fig. 1B). Interestingly, H₂O₂ levels are decreased by 20–40% in the presence of the cells compared with medium alone, suggesting that β-cells may be capable of detoxifying this oxidant (Fig. 1B). Area under the curve calculations demonstrate that, over the course of the 30-min experiment, cells are exposed to a similar amount of H₂O₂, regardless of the method of delivery (Fig. 1C). In the absence of cells, more oxidant is present in the medium when it is delivered continuously by 20 milliunits/ml glucose oxidase than when it is delivered as a one-time 100 μm bolus (Fig. 1C).

Figure 1.

Temporal generation of H₂O₂ by glucose oxidase. A and B, time-dependent formation of H₂O₂ in RPMI tissue culture medium following bolus addition of 100 μm H₂O₂ (A) or produced by 20 milliunits/ml glucose oxidase (B) with or without INS 832/13 cells present at full confluency (50,000 in a 96-well plate). H₂O₂ levels were determined by horseradish peroxidase–catalyzed oxidation of Amplex Red to resorufin. C, area under the curve calculations of A and B. Results are the average ± S.E. of three independent experiments; *, p < 0.05; GO, glucose oxidase.

To explore this observation more thoroughly, rat INS 832/13 cells or human EndoC-βH1 cells were exposed for 4 h to H₂O₂ delivered as a bolus or continuously using glucose oxidase (Fig. 2). As expected, extracellular H₂O₂ concentrations are indistinguishable from untreated medium after 4 h of incubation under bolus conditions, regardless of the presence of cells (Fig. 2A). However, H₂O₂ levels after 4 h of glucose oxidase treatment are inversely related to cell density (Fig. 2, B and C), with the highest H₂O₂ accumulation occurring in the absence of cells and the lowest in the presence of cells at full confluency (50,000). INS 832/13 cell viability decreases in a concentration-dependent manner in response to H₂O₂ delivered as a bolus, and this effect is not modified by cell density (Fig. 2D). When supplied as a bolus at a concentration of 200 μm, H₂O₂ reduces INS 832/13 cell viability by 50% under all conditions examined. In contrast, when delivered continuously using glucose oxidase, H₂O₂ decreases cell viability in a concentration-dependent manner that is also dependent on the initial cell density (Fig. 2, E and F). The concentration-dependent actions of H₂O₂ on cell viability are not due to glucose depletion by glucose oxidase, as there is minimal consumption of glucose by this enzyme (Fig. 2G). With increasing cell density, the toxic actions of H₂O₂ are decreased, and this protective effect of higher cell density is associated with lower detectable levels of H₂O₂ following the 4-h incubation. These findings suggest that β-cells have the capacity to detoxify H₂O₂ when provided in a physiologically relevant, continuous manner; however, when supplied as a bolus, β-cells are unable to remove this oxidant.

Figure 2.

β-Cell detoxification of H₂O₂ when continuously generated by glucose oxidase. A–C, concentration-dependent accumulation of H₂O₂ in the medium following 4-h incubation after addition of H₂O₂ as a bolus (A) or continuously produced by glucose oxidase (B and C) in the presence of increasing density of INS 832/13 cells (A and B) or human EndoC-βH1 cells (C). Results are the average ± S.E. of three independent experiments; *, p < 0.05 (no cells compared with either 25,000 or 50,0000 cells). D–F, cell viability (determined by Neutral Red assay) 4 h after treatment with H₂O₂ bolus (D) or glucose oxidase (E and F) at the indicated concentrations in INS 832/13 cells (D and E) or human EndoC-βH1 cells (F). G, glucose concentration in INS 832/13 medium in the absence of cells measured 4 h after incubation with the indicated concentrations of glucose oxidase. Results are the average ± S.E. of three independent experiments: n.s., not significant; *, p < 0.05; GO, glucose oxidase.

Continuous delivery of H₂O₂ fails to deplete nucleotides in INS 832/13 cells

Consistent with previous findings, we observed that bolus administration of 100 μm H₂O₂ for 1 h depletes intracellular energy stores, as evidenced by the decrease in both ATP and NAD⁺ levels in INS 832/13 cells (Fig. 3, A and B). This depletion of nucleotides is due to overactivation of PARP-1, as selective PARP-1 inhibition protects against bolus H₂O₂–induced death (Fig. 3C), as observed previously (34). However, H₂O₂ generated continuously by 20 milliunits/ml glucose oxidase over 1 h does not deplete ATP and NAD⁺ levels (Fig. 3, A and B), and PARP-1 inhibition does not protect against cell death induced by 4-h glucose oxidase treatment, suggesting that PARP-1 is not activated in β-cells in response to continuous delivery of H₂O₂ (Fig. 3D).

Figure 3.

Effects of H₂O₂ treatment on nucleotide levels in rodent insulinoma cells. A and B, ATP (A) and NAD⁺ (B) levels were determined by HPLC in INS 832/13 cells after 1 h of treatment with H₂O₂ provided as a bolus at 100 μm or continuously using glucose oxidase (20 milliunits/ml). ATP and NAD⁺ levels are normalized to total protein. C and D, cell viability (determined by Neutral Red assay) of INS 832/13 cells treated with either H₂O₂ bolus (C) or glucose oxidase (D) at the indicated concentrations for 4 h with or without the PARP-1 inhibitor PJ-34 (5 μm). Results are the average ± S.E. of at least three independent experiments; *, p < 0.05.

Differential activation of signaling pathways after bolus addition and continuous H₂O₂ delivery

Bolus H₂O₂ addition is known to activate the DNA damage response and energy-sensing pathways in β-cells (35). As expected, bolus addition of 100 μm H₂O₂ causes DNA double-strand breaks, as indicated by the phosphorylation of histone variant H2AX (γH2AX), and activation of energy-sensing pathways, as indicated by the phosphorylation of AMP-activated kinase (Fig. 4, A and B). Peroxiredoxins are a ubiquitous class of antioxidant enzymes that, when in their reduced form, detoxify H₂O₂ by reducing it to H₂O (36). During the catalytic cycle of H₂O₂ reduction, the peroxiredoxin active-site cysteine is oxidized to sulfinic acid (PrxSOH) and then reduced by the action of thioredoxins. However, following bolus addition of 100 μm H₂O₂, the peroxiredoxin active-site thiol is oxidized to sulfonic acid (PrxSO₃), inhibiting the enzyme (Fig. 4, A and B). This result suggests that peroxiredoxins, which may normally provide antioxidant defenses to the cell, are exhausted by treatment with an H₂O₂ bolus. In contrast, exposure to H₂O₂ delivered continuously over 30 min at glucose oxidase concentrations of 30 milliunits/ml or lower fails to activate the DNA damage response and energy-sensing pathways and does not modify the peroxiredoxin active site (Fig. 4). It is only under high concentrations of glucose oxidase (40 milliunits/ml or greater) that continuous H₂O₂ delivery elicits similar responses to bolus addition (Fig. 4C), suggesting the existence of a threshold effect for H₂O₂ removal. Below this threshold level, β-cells can detoxify H₂O₂. However, when this threshold is surpassed, the defense is no longer capable of reducing H₂O₂, leading to hyperoxidation of peroxiredoxins (PrxSO₃), activation of the DNA damage response (γH2AX), and activation of low energy–sensing pathways (AMPK phosphorylation).

Figure 4.

Differential activation of signaling pathways by H₂O₂ supplied as a bolus or continuously. A, Western blot analysis was used to examine the phosphorylation of AMP kinase (P-AMPK, energy-sensing pathway) and of the histone variant H2AX (γH2AX, marker of DNA damage) and PrxSO₃ (inhibition of the enzyme) following treatment of INS 832/13 cells for 30 min with either H₂O₂ provided as a bolus (100 μm) or continuously by glucose oxidase (20 milliunits/ml). B, quantification of the posttranslational modifications shown in A. C, phosphorylation of AMPK, hyperoxidation of peroxiredoxin, and formation of γH2AX after treatment of INS 832/13 cells for 30 min with increasing concentrations of glucose oxidase. Also shown for comparison is the activation of these cascades in response to 100 μm H₂O₂ provided as a bolus. GAPDH levels were determined to assess protein loading. Results are representative of at least three independent experiments; *, p < 0.05; GO, glucose oxidase.

Expression of peroxiredoxin, thioredoxin, and thioredoxin reductase in β-cells

As introduced above, peroxiredoxins are antioxidant enzymes that function to reduce H₂O₂ to H₂O, resulting in their oxidation. Thioredoxin is used to return peroxiredoxins to their reduced form and restore catalytic activity. The oxidation status of thioredoxin is controlled by the activity of thioredoxin reductase, which reduces thioredoxin and allows the cycle to continue. NADPH closes the catalytic relay by serving as an electron donor for reduction of thioredoxin reductase. Although β-cells are thought to be ill-equipped with antioxidant enzymes such as GSH peroxidase and catalase (18–21), they readily express peroxiredoxin, thioredoxin, and thioredoxin reductase genes, as determined by qRT-PCR (Fig. 5). The primary thioredoxin and thioredoxin reductase isoforms expressed in both INS 832/13 cells and rat islets are the cytoplasmic forms: Txn1 and Txnrd1, respectively (Fig. 5). Interestingly, there are some differences in expression of the different peroxiredoxin isoforms between INS 832/13 cells and rat islets. In rat islets, the primary peroxiredoxins are Prdx1 and Prdx2, both cytoplasmic (Fig. 5B), whereas INS 832/13 cells express these two forms along with the endoplasmic reticulum–localized isoform Prdx4 and Prdx5, an isoform with uncertain localization (Fig. 5A) (2). Another notable difference is the overall level of expression of the peroxiredoxins and thioredoxin reductases, with both being expressed around 10-fold greater in rat islets than in INS 832/13 cells. Together, these data show that all components of the peroxiredoxin antioxidant cycle are present in β-cells.

Figure 5.

Antioxidant gene expression in the β-cell. Shown is absolute quantification of peroxiredoxin, thioredoxin, and thioredoxin reductase steady-state mRNA levels, as determined by qRT-PCR, in INS 832/13 cells (A) and rat islets (B). Results are the average ± S.E. of three independent experiments.

Inhibition or knockdown of thioredoxin reductase sensitizes β-cells to H₂O₂ delivered by glucose oxidase

Because a 100 μm H₂O₂ bolus hyperoxidizes peroxiredoxins whereas 20 milliunits/ml glucose oxidase does not, and because peroxiredoxin, thioredoxin, and thioredoxin reductase are readily expressed in β-cells, we hypothesized that the peroxiredoxin/thioredoxin antioxidant system may be the pathway that allows β-cells to detoxify H₂O₂. To test this hypothesis, we inhibited thioredoxin reductase with the selective inhibitor auranofin (AFN). Treatment with auranofin does not affect INS 832/13 cell viability after bolus administration of H₂O₂ (Fig. 6A). However, when H₂O₂ is delivered continuously by glucose oxidase, auranofin treatment results in a significant decrease in INS 832/13 and EndoC-βH1 cell as well as dispersed rat islet cell viability (Fig. 6, B–D). The addition of 10 units/ml catalase completely restores cell viability, suggesting that H₂O₂, and not D-glucono-1,5 lactone, another product of the glucose oxidase reaction, is the cause of cell death (Fig. 6B). Similarly, when Txnrd1 is knocked down more than 70% (Fig. 6G) using specific siRNAs, INS 832/13 cells become significantly more sensitive to H₂O₂ generated by glucose oxidase (Fig. 6F) but not to H₂O₂ delivered as a bolus (Fig. 6E).

Figure 6.

Inhibition or knockdown of thioredoxin reductase sensitizes β-cells to H₂O₂ delivered continuously. A–C and H, INS 832/13 (A, B, and H) or EndoC-βH1 (C) cell viability was determined by Neutral Red assay following 4-h treatment with H₂O₂ provided as a bolus addition or continuously generated using glucose oxidase or menadione at the indicated concentrations in the presence or absence of the thioredoxin reductase inhibitor AFN (0.1 μm for INS 832/13 and 5 μm for EndoC-βH1). 10 units/ml catalase from bovine liver was included as a control in B. Results are the average ± S.E. of at least three independent experiments; *, p < 0.05 (compared with no AFN control). D, the viability of dispersed rat islet cells was determined by SYTOX Green following 4-h exposure to 15 milliunits/ml glucose oxidase in the presence or absence of AFN (5 μm). E, F, and I, cell viability was determined following 4-h treatment with H₂O₂ provided as a bolus addition or continuously generated using either glucose oxidase or menadione at the specified concentrations in INS 832/13 cells with siRNA-mediated knockdown of Txnrd1 or with nontargeting siRNA (Negative ctrl). Results are the average ± S.E. of at least three independent experiments: *, p < 0.05 (negative control versus Txnrd1 siRNA 1; †, p < 0.05 (negative control versus Txnrd1 siRNA 2). G, knockdown efficiency of Txnrd1 was determined by relative qRT-PCR, and mRNA accumulation was normalized to Gapdh levels. Results are the average ± S.E. of at least three independent experiments; *, p < 0.05.

Because glucose oxidase delivers H₂O₂ to cells extracellularly, we sought a method to deliver H₂O₂ intracellularly to more closely mimic how the oxidant might be generated during oxidative phosphorylation. To this end, we used menadione, a redox cycler that generates superoxide, which is subsequently dismutated to H₂O₂, in the mitochondria (37). When thioredoxin reductase is either inhibited or depleted, INS 832/13 cells become significantly more sensitized to increasing concentrations of menadione (Fig. 6, H and I). These results suggest that thioredoxin reductase, which maintains the reduced, active pool of peroxiredoxins, is necessary for β-cell detoxification of continuously generated H₂O₂.

Thioredoxin reductase protects β-cells from H₂O₂-mediated damage

Reactive oxygen species such as H₂O₂ are known to cause DNA damage (2). However, such damage (assessed by both mean tail moment and formation of γH2AX) is only evident in INS 832/23 cells (Fig. 7, A and B) or rat islets (Fig. 7C) when they are treated with bolus H₂O₂. In contrast, treatment with glucose oxidase does not stimulate DNA damage or γH2AX formation. However, when β-cells are treated with both glucose oxidase and the thioredoxin reductase inhibitor, auranofin, they become susceptible to DNA damage to a similar extent as damage induced by H₂O₂ bolus addition (Fig. 7). These results suggest that the peroxiredoxin/thioredoxin system, supported by thioredoxin reductase, protects β-cells from H₂O₂-mediated DNA damage.

Figure 7.

Thioredoxin reductase protects β-cells from H₂O₂-mediated DNA damage. A, DNA damage was determined by comet assay following 30-min treatment of INS 832/13 cells with H₂O₂ provided as a bolus (100 μm) or continuously generated using glucose oxidase (20 milliunits/ml) in the presence or absence of the thioredoxin reductase inhibitor AFN (0.1 μm). B and C, Western blot analysis was used to examine the phosphorylation of histone variant H2AX (γH2AX, marker of DNA damage) following 30-min treatment of INS 832/13 cells (B) and dispersed rat islets (C) with H₂O₂ provided as a bolus or continuously by glucose oxidase in the presence or absence of the thioredoxin reductase inhibitor AFN at the indicated concentrations. GAPDH levels were determined to assess protein loading. Results are representative of three independent experiments; *, p < 0.05; GO, glucose oxidase.

Discussion

ROS, such as superoxide and hydrogen peroxide, are produced during normal cellular metabolism. Superoxide is formed by addition of a single electron to molecular oxygen, a process that is mediated largely by electron leak from the mitochondrial electron transport chain (2). H₂O₂ is rapidly produced by dismutation of superoxide and can function as a signaling molecule through the oxidation of specific cysteine residues of target proteins, which, in turn, can modify the activity of those proteins (1). In β-cells, this oxidant has been proposed to promote insulin secretion and cellular proliferation (38–40). Paradoxically, if H₂O₂ accumulates to an uncontrolled level, then hydroxyl radicals are formed through reaction of H₂O₂ with metal cations such as Fe²⁺ in the Fenton reaction, causing DNA damage and lipid and protein oxidation (2). To combat this oxidative stress, several antioxidant enzymes detoxify elevated levels of H₂O₂, including catalase, GSH peroxidases, and peroxiredoxins, which all function by reducing H₂O₂ to water (1).

Because of reportedly low expression of catalase and GSH peroxidase, which are considered a main line of defense against H₂O₂, pancreatic β-cells have been described as being vulnerable to oxidative damage (18–20). As a result, oxidative stress has been postulated as one of the primary causes of β-cell failure in both type 1 and type 2 diabetes. This hypothesis is supported by observations of elevated markers of oxidative stress in islets from rodent models of diabetes as well as in pancreatic biopsy samples from diabetic patients (3–10). Additionally, overexpression of antioxidant enzymes, including superoxide dismutase, catalase, GSH peroxidase, thioredoxin, and peroxiredoxin, in rodent β-cells affords protection against oxidative stress induced by alloxan, by combined treatment of hypoxanthine and xanthine oxidase, or by bolus H₂O₂ (22–24, 26–28, 31). As a result, it is widely propagated that β-cells are particularly sensitive to damage by ROS.

When re-examining this model of a vulnerable β-cell, there are a number of assumptions that seem at odds with the functional roles of this insulin-producing cell in controlling whole-body glucose metabolism. First, β-cells are essential for survival, being the only cells in the body capable of secreting insulin. Second, the pathway responsible for controlling glucose-stimulated insulin secretion is the same as a major pathway by which ROS are produced: mitochondrial oxidative phosphorylation. Third, β-cells have limited self-renewal capacity, meaning that if they are damaged in any way, then it is difficult to replenish the population (41). Evolutionarily, it would be beneficial to the organism for β-cells to possess mechanisms that afford protection from oxidative stress.

Oxidative phosphorylation in β-cells is exquisitely coupled to insulin secretion, as more than 90% of the carbons in glucose are oxidized to CO₂ upon supply of the hexose substrate (13–15). Because of this tight coupling, the rate of mitochondrial oxidation in β-cells, is proportional to blood glucose, allowing for ATP, which accumulates during oxidative metabolism, to inhibit ATP-sensitive potassium channels. This results in membrane depolarization, voltage-dependent influx of Ca²⁺, and release of insulin granules. This allows β-cells to adjust the rate of insulin secretion in response to changes in blood glucose concentration (16). Because mitochondrial oxidation of glucose regulates insulin secretion, and insulin production is essential for organismal survival, we hypothesized that β-cells possess a mechanism that protects them from oxidative damage.

To address this hypothesis, a more physiological delivery system for H₂O₂ was employed. Most studies have used a single or repeated bolus addition of H₂O₂, and the net effect is a rapid delivery of high levels of this oxidant over a very short period of time. For example, Fig. 1 shows that, in response to a single bolus of 100 μm H₂O₂, almost 80% of the oxidant is reduced within 10–15 min because of reaction with medium components, such as sodium pyruvate (33). In contrast, using glucose and glucose oxidase, it is possible to continuously deliver H₂O₂ to both rodent and human pancreatic β-cells and rodent islets in a time-dependent manner at controlled rates so that H₂O₂ levels reach 40–50 μm following a 30-min incubation. Although it is not surprising that, in response to a bolus addition of 100–200 μm H₂O₂, there is rapid loss in cellular ATP and INS 832/13 cell viability (35), we were surprised that β-cells remained viable over the 4-h incubation in the presence of H₂O₂ produced by glucose oxidase, given that H₂O₂ is generated to levels that exceed those produced by the bolus addition over the course of this experiment (based on exposure levels after 30 min, Fig. 1C).

Consistent with differences in viability in response to H₂O₂ delivered by these two methods, we observed differential activation of signaling pathways. Bolus delivery (100 μm, following a 30 min incubation) activates the energy-sensing pathway of AMP-activated protein kinase (AMPK), causes DNA damage, depletes intracellular stores of ATP and NAD⁺, and activates poly(ADP-ribose) polymerase (PARP)-dependent death. In contrast, H₂O₂ delivered by glucose oxidase at 20 milliunits/ml does not activate AMPK or deplete ATP or NAD⁺ levels. It is only when high levels of glucose oxidase (50 milliunits/ml) are used that H₂O₂ activates AMPK or induces DNA damage (as measured by γH2AX formation).

It is intriguing that treatment with 100 μm H₂O₂ as a bolus results in rapid hyperoxidation of the antioxidant peroxiredoxin active site whereas continuous delivery using less than 40 milliunits/ml glucose oxidase does not. In its oxidized form, peroxiredoxin cannot detoxify H₂O₂, and requires reduction by thioredoxin and thioredoxin reductase to be capable of additional antioxidant cycles. This finding suggests that H₂O₂, when delivered as a bolus, exhausts the endogenous pool of reduced active peroxiredoxin before it can be replenished by thioredoxin and thioredoxin reductase. Because catalase and GSH peroxidases are reported to be expressed at low levels in β-cells compared with other cell types (such as liver), and because H₂O₂ delivered as a bolus stimulates peroxiredoxin inhibition, the role of the peroxiredoxin, thioredoxin, and thioredoxin reductase antioxidant system in detoxifying H₂O₂ was examined. We show that all of the components of this system are readily detectable in β-cells by qRT-PCR and that either inhibition or knockdown of thioredoxin reductase renders β-cells significantly more sensitive to glucose oxidase than controls. The effects of H₂O₂ on INS 832/13 cells and islets with decreased expression or activity of thioredoxin reductase are similar to the effects of H₂O₂ delivered as a bolus.

Our results suggest that Txnrd1, a cytoplasmic thioredoxin reductase, is the primary mediator of thioredoxin reduction in β-cells because specific knockdown of this gene abolishes the ability of these cells to detoxify H₂O₂. Txnrd2 encodes a mitochondrial form of the enzyme, whereas Txnrd3 is thought to be mainly testis-specific. Quantification of these transcripts by qRT-PCR suggests that Txnrd1 is the primary gene expressed in INS 832/13 cells and rat islets, supporting the findings of our knockdown studies. Additional experiments are necessary to determine the relative roles of the thioredoxins (Txn1 and Txn2) and of the peroxiredoxins (Prdx1–Prdx6) in the reduction process.

Why do β-cells express peroxiredoxins but do not express other typical antioxidant enzymes, such as GSH peroxidase? GSH peroxidases have catalytic efficiencies for H₂O₂ of around 10⁸ m⁻¹ s⁻¹ (42) whereas peroxiredoxins have 1000-fold lower efficiencies of around 10⁵ m⁻¹ s⁻¹ (42). Additionally, peroxiredoxins can be inactivated through oxidation by H₂O₂ (43). However, despite being less efficient detoxifiers, peroxiredoxins have a very high affinity for the oxidant (K_m < 20 μm) (44), making them prime candidates for mediators of H₂O₂ signaling in addition to detoxification. Indeed, Prdx2 has been shown to participate in a “redox relay” for signaling by transferring oxidizing equivalents from H₂O₂ to target proteins (45, 46). Given the role of H₂O₂ in promoting glucose-stimulated insulin secretion (38, 39), it is possible that β-cells predominantly express peroxiredoxins to perform dual roles of H₂O₂ signaling and detoxification while suppressing other antioxidant enzymes that may counteract this dual function. In support of this hypothesis, Prdx2 has been shown to be required for insulin secretion in Caenorhabditis elegans (47), suggesting a putative signaling role.

Collectively, our studies suggest a model in which β-cells utilize peroxiredoxins rather than catalase or GSH peroxidase to detoxify H₂O₂ produced from superoxide generated during glucose metabolism (Fig. 8). The peroxiredoxin antioxidant system may allow β-cells to protect themselves against oxidative stress while also providing a signaling role necessary for glucose-stimulated insulin secretion. This model provides a potential explanation as to why β-cells do not express catalase and challenges the widely held view that β-cells are particularly sensitive to H₂O₂, suggesting that they may not be so vulnerable to reactive oxygen species after all.

Figure 8.

Model of the thioredoxin reductase-dependent antioxidant system in the β-cell. Peroxiredoxins (Prx) are a ubiquitous class of antioxidant enzymes that reduce H₂O₂ to H₂O and become oxidized. Prx then relies on thioredoxin (Trx) to reduce it and restore its catalytic activity. Thioredoxin reductases (TrxR) are critical for this cycle because they reduce the oxidized thioredoxin, allowing the catalytic cycle to continue. Our results suggest not only that β-cells can detoxify H₂O₂, contrary to what is commonly thought, but also that they maintain a robust antioxidant system driven by peroxiredoxins and thioredoxin reductase rather than more classical antioxidants, such as catalase (Cat). TCA, tricarboxylic acid; OXPHOS, oxidative phosphorylation.

Experimental procedures

Materials and animals

Male Sprague-Dawley rats (250 to 300 g) were purchased from Harlan (Indianapolis, IN). Rat insulinoma INS 832/13 cells were obtained from Chris Newgard (Duke University, Durham, NC) (48). Human insulinoma EndoC-βH1 cells were obtained from Raphael Scharfmann (Paris Descartes University, Paris, France) (49). Connaught Medical Research Laboratories 1066 medium, RPMI 1640 medium, Dulbecco's modified Eagle's medium, ι-glutamine, sodium pyruvate, HEPES, penicillin, streptomycin, β-mercaptoethanol, Amplex Red powder, and SYTOX Green nucleic acid stain were purchased from Thermo Fisher Scientific (Waltham, MA). Amplex Red and SYTOX Green were dissolved in DMSO at concentrations of 10 mm and 5 mm, respectively, prior to use. Fetal bovine serum was purchased from HyClone (Logan, UT). BSA fraction V was purchased from Roche Diagnostics. All other reagents were purchased from Sigma-Aldrich (St. Louis, MO). Glucose oxidase was solubilized in 0.05 m sodium acetate buffer at a concentration of 1 mg/ml and horseradish peroxidase (HRP, type VI) was dissolved in 10 mm phosphate buffer at a concentration of 100 units/ml prior to use.

Rat islet isolation, dispersion, and culture

Animal studies were approved by the Medical College of Wisconsin's Institutional Animal Care and use Committee. Islets were isolated and cultured from male Sprague-Dawley rats by collagenase digestion as described previously (50). Glucose is present at a concentration of 5.5 mm. Prior to experimentation, islets were dispersed into single cells by incubation in 0.48 mm EDTA in PBS followed by agitation in 1 mg/ml trypsin in Ca²⁺/Mg²⁺-free Hanks' balanced salt solution.

Cell culture and treatment

INS 832/13 and EndoC-βH1 cells were cultured as described previously (34, 49), with glucose concentrations of 11 and 5.5 mm, respectively, and maintained at 37 °C in an atmosphere of 95% air and 5% CO₂. β-Mercaptoethanol was omitted, and fetal bovine serum was used at a concentration of 2% in medium used for H₂O₂ determinations by Amplex Red Assay (all additional components were present). For glucose oxidase treatment, the enzyme was added to cell culture medium at the indicated concentrations once at the start of the experiment and was present throughout the experiment.

siRNA-mediated knockdown

A dicer-substrate siRNA against either rat Txnrd1 or a negative control siRNA, purchased from Integrated DNA Technologies (Skokie, IL), was reverse-transfected into INS 832/13 cells using Lipofectamine 2000 and Opti-MEM reduced serum medium (Thermo Fisher) at a final concentration of 100 nm. Sequences were as follows: Txnrd1 siRNA 1, 5′-GAG AAU GCU UAC GGG AAA UUC AUT G-3′; Txnrd1 siRNA 2, 5′-GCA UCA GCA GUG ACG AUC UUU UCT C-3′; negative control, 5′-CGU UAA UCG CGU AUA AUA CGC GUA T-3′. 24 h after transfection, medium was replaced, and cells were cultured for another 24 h before treatment. Knockdown efficiency was determined using relative quantification qRT-PCR.

H₂O₂ concentration determination

H₂O₂ levels in medium were assessed following peroxidase-catalyzed oxidation of Amplex Red to resorufin. At specific time points, medium samples were mixed with phosphate buffer (100 mm (pH 7.4)) containing 100 μm Amplex Red and 1 unit/ml HRP (type VI), and fluorescence was measured at excitation/emission of 535 nm/590 nm. H₂O₂ levels in medium were quantified using an H₂O₂ standard curve.

Glucose concentration determination

Glucose concentrations in INS 832/13 medium 4 h after addition of glucose oxidase were determined using the Amplex Red assay as described above with addition of 2 units/ml glucose oxidase at the end of the incubation. To remove medium H₂O₂, which would confound glucose measurements, 20 units/ml catalase from bovine liver (Sigma, C1345) was added during the 4-h glucose oxidase incubation. Glucose concentration was quantified using a glucose standard curve.

Cell viability and death assays

Cell viability was measured using the Neutral Red dye uptake assay as described previously (51). Cell death was determined by assessing the fluorescence of SYTOX Green nucleic acid stain (Thermo Fisher Scientific) as described previously (52).

Comet assay

DNA damage was measured using the comet assay (53) and quantified as the mean tail moment using the Comet Assay Software Project program for 30–50 cells/condition.

Nucleotide measurements

HPLC was used to quantify the cellular levels of ATP and NAD⁺ as reported previously (54, 55).

Western blot analysis

Equal amounts of protein from cell lysates were resolved by nonreducing SDS-PAGE and transferred to nitrocellulose membranes. Proteins were detected using primary antibodies: mouse anti-GAPDH (anti-glyceraldehyde 3-phosphate dehydrogenase, 1:20,000, Invitrogen), mouse anti-phospho-H2AX (Ser-139, γH2AX, 1:10,000, EMD Millipore, Billerica, MA), rabbit anti-phospho-AMPKα (Thr-172, AMP-activated protein kinase, 1:1000, Cell Signaling Technology, Danvers, MA), and rabbit anti-peroxiredoxin-SO₃ (1:2000, Abcam, Cambridge, MA). Detection was performed by enhanced chemiluminescence (56) using species-specific HRP-conjugated donkey anti-mouse or donkey anti-rabbit (1:20,000) secondary antibodies.

qRT-PCR

Total RNA was isolated from INS 832/13 cells using the RNeasy kit (Qiagen). First-strand synthesis was performed using oligo(dT)s and Thermo Scientific Maxima H Minus reverse transcriptase according to the manufacturer's instructions. Quantitative PCR was performed using SsoFast EvaGreen Supermix (Bio-Rad) and a Bio-Rad CFX96 real-time system. Primers specific for the rat genes were purchased from Integrated DNA Technologies and are listed in Table S1. For relative quantification, gene expression was normalized to Gapdh using the 2^−ΔΔCt method. For absolute quantification, mRNA amounts were determined using standard curves generated from purified PCR products (57).

Statistical analysis

Statistical comparisons were made between groups using either Student's t test or two-way ANOVA with Tukey post hoc test. The minimum level of significance was set at p < 0.05.

Author contributions

J. S. S., K. A. B., B. J. O., and J. A. C. conceptualization; J. S. S., K. A. B., and B. J. O. data curation; J. S. S. and K. A. B. software; J. S. S., K. A. B., B. J. O., and J. A. C. formal analysis; J. S. S., K. A. B., and J. A. C. supervision; J. S. S., K. A. B., B. J. O., and J. A. C. validation; J. S. S., K. A. B., B. J. O., and A. N. investigation; J. S. S., K. A. B., and J. A. C. visualization; J. S. S., K. A. B., B. J. O., A. N., and J. A. C. methodology; J. S. S. writing-original draft; J. S. S. and J. A. C. project administration; K. A. B. and J. A. C. funding acquisition; K. A. B., B. J. O., A. N., and J. A. C. writing-review and editing.