Biochemical Journal

Research article

Mitochondrial protection by the thioredoxin-2 and glutathione systems in an in vitro endothelial model of sepsis

Damon A. Lowes, Helen F. Galley


Oxidative stress and mitochondrial dysfunction are common features in patients with sepsis and organ failure. Within mitochondria, superoxide is converted into hydrogen peroxide by MnSOD (manganese-containing superoxide dismutase), which is then detoxified by either the mGSH (mitochondrial glutathione) system, using the enzymes mGPx-1 (mitochondrial glutathione peroxidase-1), GRD (glutathione reductase) and mGSH, or the TRX-2 (thioredoxin-2) system, which uses the enzymes PRX-3 (peroxiredoxin-3) and TRX-2R (thioredoxin reductase-2) and TRX-2. In the present paper we investigated the relative contribution of these two systems, using selective inhibitors, in relation to mitochondrial dysfunction in endothelial cells cultured with LPS (lipopolysaccharide) and PepG (peptidoglycan). Specific inhibition of both the TRX-2 and mGSH systems increased the intracellular total radical production (P<0.05) and reduced mitochondrial membrane potentials (P<0.05). Inhibition of the TRX-2 system, but not mGSH, resulted in lower ATP production (P<0.001) with high metabolic activity (P<0.001), low oxygen consumption (P<0.001) and increased lactate production (P<0.001) and caspase 3/7 activation (P<0.05). Collectively these results show that the TRX-2 system appears to have a more important role in preventing mitochondrial dysfunction than the mGSH system in endothelial cells under conditions that mimic a septic insult.

  • endothelium
  • inflammation
  • oxidative stress
  • peroxiredoxin
  • sepsis
  • thioredoxin


Sepsis is a severe infection causing dysregulation of inflammatory responses and may result in MODS (multiple organ dysfunction syndrome), which has a mortality rate of approximately 70% [1,2]. Oxidative stress and mitochondrial dysfunction are associated with sepsis and organ failure in animal models [37] and with a poorer outcome in patients with sepsis [811].

The impact of mitochondrial functional capacity and the mechanisms related to mitochondrial dysfunction during sepsis and MODS are still not fully understood. During sepsis, leucocytes and mitochondria produce elevated amounts of ROS (reactive oxygen species; reviewed in [12]). Enhanced mitochondrial generation of the superoxide anion radical is observed in non-surviving groups in various models of sepsis and may be a starting point of mitochondrial oxidative stress [4,13,14]. Improved outcomes in animal models of sepsis have been achieved by us and other groups by utilizing mitochondrial-targeted antioxidants (extensively reviewed [1517]). Superoxide does not readily cross the mitochondrial membrane and is catalysed into hydrogen peroxide within the mitochondria by MnSOD (manganese-containing superoxide dismutase). Hydrogen peroxide can diffuse out of the mitochondrion and is metabolized by catalase in the peroxisome, but is primarily removed within the mitochondrion by oxidation of reduced mGSH (mitochondrial glutathione) catalysed by mGPx-1 [mitochondrial GPx (glutathione peroxidase)-1] with recycling back to reduced glutathione catalysed by GRD (glutathione reductase) [18]. It is now known that oxidation of mitochondrial TRX (thioredoxin)-2 in the presence of PRX (peroxiredoxin)-3 with subsequent recycling via TRX-2R (TRX reductase-2) is also important [19]. This TRX-2 system has been reported to be more efficient at maintaining mitochondrial proteins in a reduced state compared with the mGSH system [20], although under conditions of low-level oxidative stress, both systems keep hydrogen peroxide levels in check and have definitive roles in cell signalling [20]. However, under conditions of more severe oxidative stress, the TRX-2 system may become more important [15].

The relative importance of the mGSH and TRX-2 systems and mitochondrial dysfunction during the cellular events of sepsis are not known. Endothelial cells recognize and respond to invading pathogens [21] and the loss of endothelial integrity contributes to the morbidity and mortality associated with sepsis. Increased mitochondrial generation of hydrogen peroxide and pro-inflammatory cytokine production in response to bacterial cell wall components and inflammatory mediators by endothelial cells has been well documented [22,23]. Therefore the aim of the present study was to examine the consequences of inhibiting either mGPx-1 with thiomalic acid or TRX-2R with the gold compound auranofin on oxidative stress and relative mitochondrial function in human endothelial cells cultured under conditions which mimic a mixed Gram-negative and Gram-positive septic insult.



Auranofin was obtained from Alexis Biochemicals. CDFH [5-(6)-carboxy-2′,7′-dichlorodihydrofluorescein] diacetate and the Molecular Probes™ ATP determination kit were from Invitrogen. Caspase-Glo® was from Promega. The protease/phosphatase inhibitor cocktail was from Roche. The Bradford reagent was from Bio-Rad Laboratories, and antibodies for Western blotting were from Abcam. All other reagents were from Sigma–Aldrich.

Cell culture and treatment

The HUVEC C (human umbilical vein endothelial cell) line was obtained from the A.T.C.C. and was used at passages 5–14. This in vitro model of sepsis has been described in detail previously [16]. For experimentation, cells were grown in 96-well plates or on coverslips (see below) in the presence of 2 μg/ml LPS (lipopolysaccharide; from Escherichia coli strain 0111:134) plus 20 μg/ml PepG (peptidoglycan G; from Staphylococcus aureus strain 6571), prepared as described in [24], plus either thiomalic acid or auranofin for up to 7 days to allow mitochondrial dysfunction to develop. LPS and PepG are TLR (Toll-like receptor) 4 and 2 agonists respectively, that play key roles in the innate immune system. We have shown previously that HUVECs increase their mitochondrial generation of hydrogen peroxide and pro-inflammatory cytokine production maximally in response to these concentrations of LPS/PepG [16]. Auranofin is a co-ordinated gold(I) compound that can react with selenol-containing residues and is a potent inhibitor of purified TRX-R (thioredoxin reductase) protein. Thiomalic acid is a glutathione mimetic that can bind to active-site SH groups. Evidence suggests that the mitochondrial isoforms of TRX-R and GPx are more sensitive to inhibition than the cytosolic isoforms. To determine the selective inhibition of mitochondrial isoforms of the enzymes by auranofin and thiomalic acid we exposed cells to the inhibitors at a range of concentrations in preliminary experiments, and measured enzyme activities of mitochondrial and cytosolic isoforms of TRX-R and GPx (see below). As a result of these preliminary experiments, 4 mM thiomalic acid or 2 μM auranofin were used in subsequent experiments. Some cells were treated with solvent only (untreated), solvent plus inhibitor only, solvent plus 50 μM hydrogen peroxide or 25 μM t-butyl hydroperoxide as positive controls. Cell viability was assessed using acid-phosphatase activity [25].

TRX-R and GPx activity

TRX-R and GPx activity was determined in cytosolic and mitochondrial fractions using Sigma® TRX-R assay and GPx activity assay kits respectively. Briefly, cytosolic or mitochondrial lysates were added to the appropriate assay buffers containing NADPH and enzyme mixes in a 1 ml cuvette. Reactions were started by the addition of 5,5′-dithiobis-(2-nitrobenzoic acid) or t-butyl hydroperoxide for TRX-R or GPx activity respectively. TRX-R activity was determined by following the increase in absorbance at 412 nm after a 2 min delay, and GPx activity was determined by following the increase in absorbance at 430 nm after a 15 s delay using a Helios beta single-beam spectrophotometer.

Cytokine production

Cells were grown to confluence in 96-well plates and treated as above for 24 h. The levels of interleukin (IL)-6 and IL-8 were measured in culture medium using a commercially available enzyme immunoassay kit.

Intracellular total radical production

Briefly, cells were grown to confluence in 96-well plates and loaded with 50 μM oxidation sensitive CDHF diacetate or 50 μM oxidation-insensitive CDF for 60min in the dark. Cells were then washed twice with PBS before being treated immediately as described above. The rate (positive) of total radical formation was determined fluorimetrically as a result of oxidation of CDHFD as a continuous recording until saturation was reached, at a temperature of 37°C at an excitation wavelength of 485 nm and emission wavelength 530 nm. MitoQ (Mitoquinone; a kind gift from Professor Robin Smith, Otago University, New Zealand; 1 μM) was used as a ROS scavenger to demonstrate the selectivity and reproducibility of CDHFD.

Mitochondria isolation

Mitochondria were isolated following treatment for analysis of mGSH concentration and TRX-2, PRX-3 and mGPx-1 protein expression. Cells were typsinized, pelleted and washed with ice-cold PBS before being resuspended in 5 vol. of homogenization buffer containing 0.25 M sucrose, 20 mM Hepes/KOH (pH 7.5), 10 mM KCl, 1.5 mM MgCl2, 1mM EDTA, 1mM EGTA, 1mM dithiothreitol and protease/phosphatase inhibitor cocktail for 10 min on ice. The cells were then homogenized with a Dounce borosilicate glass homogenizer before being centrifuged at 300 g for 10 min at 4°C to remove the cell debris and nuclei. Cytosolic and mitochondrial fractions were obtained by centrifugation of the cleared lysate at 4°C for 10 min at 13000 g. Protein concentrations of the cytosolic and mitochondrial fractions were determined using the Bradford protein assay procedure.

Mitochondrial and total reduced/oxidized glutathione ratio

The ratio of GSH/GSSG in isolated mitochondria and whole-cell lysates was used to indicate mGSH system activity (mitochondrial oxidative stress) and total glutathione metabolism. Isolated mitochondria or cell lysate was added to reaction buffer containing 0.1% Triton-X 100 and 0.1 M potassium phosphate, pH 6.5. GSH was measured by adding 20 μM mono-bromobimane and 0.3 mg/ml glutathione transferase and incubated at 37°C in the dark for 15 min. To measure GSSG, it was reduced to GSH in separate lysed mitochondria or whole-cell lysate samples with 2.22 mM diethylenetriaminepenta-acetic acid and 2 mM dithiothreitol in 1 M Hepes buffer (pH 8.5) for 30 min at 37°C and measuring GSH as before. Fluorescence was determined at room temperature (20°C; excitation 355 nm, emission 520 nm) and normalized to micrograms of mitochondrial protein.

Mitochondrial TRX-2, PRX-3 and mGPx-1 protein expression

Mitochondrial expression of the TRX-2 protein and the enzyme PRX-3 were measured using redox Western blotting techniques to determine their redox (oxidation) states [26,27]. This was achieved by modifying any reduced protein with small molecules that are capable of covalently modifying reduced active sites. For redox Western blotting of TRX-2, mitochondria were resuspended in 20 mM Tris/HCL buffer (pH 8.0), and incubated with 15 mM AMS (4-acetoamido-4′-maleimidylstilbene-2,2′-disulfonic acid) for 3 h at 37°C. For redox determination of PRX-3, mitochondria were incubated with 100 mM N-ethylmaleimide, 40 mM Hepes (pH 7.4), 50 mM NaCl, 1 mM EDTA, 1 mM EGTA and 10 μg/ml catalase for 15 min at room temperature before the addition of 5% (w/v) CHAPS. mGPx-1 protein expression was measured using standard Western blotting techniques without redox modification. All proteins were separated by SDS/PAGE (12–15% gel) and transferred on to Immobilon-P™ transfer membrane using standard techniques. Blocked membranes were probed with primary polyclonal antibodies using rabbit anti-(human PRX-3), sheep anti-(human TRX-2) and rabbit anti-(human GPx-1) (with appropriate horseradish peroxidase-linked secondary antibodies). HPRT (phosphoribosyltransferase hypoxanthine–guanine) was used as loading control. Bands were visualized by enhanced chemiluminescence and quantified using GeneTools™ (SynGene from Synoptics).

Measurement of caspase 3/7 activity

Apoptosis was assessed by measuring the activation of caspase 3 and caspase 7 using the Caspase-Glo® 3/7 assay. Briefly, cells were grown to confluence in 96-well plates and treated as described above for 7 days, then an equal volume of Caspase-Glo® 3/7 substrate solution was added to the culture medium and mixed gently for 30 s on a vibrating platform. The cells were then incubated for 1 h at room temperature in the dark. Luminescence was measured immediately using a fluorimeter set to detect luminescence.

Mitochondrial membrane potential

Mitochondrial membrane potential was analysed in intact cells using the fluorescent probe JC-1 (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbocyanine iodide). Briefly, following cell treatments in 96-well plates for 7 days, cells were washed twice with PBS and then incubated for 30 min with 10 μg/ml JC-1 at 37°C in the dark. Following incubation, cells were washed twice with PBS and the orange/red and green fluorescence was measured immediately at 37°C (excitation wavelength 490 nm and emission wavelength 590/520 nm). As a positive control, cells were incubated with the uncoupling agent rotenone (1 mM) for 4 h prior to JC-1 analysis. In intact mitochondria J-aggregates form and JC-1 fluoresces red. When the mitochondrial membrane potential falls the JC-1 assumes a monomeric form and fluoresces green (Supplementary Figure S1 at

Metabolic activity

Metabolic activity was analysed by measuring the rate of reduction of AlamarBlue™ in intact cells after treatment for 7 days. AlamarBlue™ is a novel redox indicator that exhibits both fluorescent and colorimetric changes in response to changes in metabolic activity (oxidative metabolism) [28]. Briefly, following cell treatments, 10 μl of AlamarBlue™ in 100 μl of culture medium was added to the wells. Fluorescence was measured every 15 min for 2 h at 37°C at an excitation wavelength of 530 nm and emission wavelength 590 nm.

ATP/ADP ratio

Cellular ATP/ADP ratio was determined after 7 day of treatment using the Molecular Probes™ ATP determination kit. Briefly, following cell treatments in 96-well plates, 1 × 106 cells were washed with ice-cold PBS before the addition of 50 μl of ice-cold lysis buffer [25 mM Hepes (pH 7.8), 5 mM MgCl2, 140 mM NaCl and 1× detergent (somatic cell ATP-releasing reagent)]. Immediately following lysis, 10 μl of cell lysate was added to 100 μl of ATP-determination buffer containing 1 mM dithiothreitol, 0.5 mM D-luciferin and 12 pg of firefly luciferase. To determine the ADP levels, samples were treated with ADP-converting solution, which contains ATP sulfurylase to remove endogenous ATP, and pyruvate kinase and phosphoenolpyruvate to convert ADP into ATP. ATP was then measured as above. Luminescence was read immediately.

Oxygen consumption

Cellular oxygen consumption was measured following 7 days of treatments, using a Clark-type oxygen electrode. Briefly, cells were grown on poly-L-lysine-coated coverslips and, following cell treatments, were added to a 7 ml oxygen electrode containing normal growth medium maintained at 37°C. Oxygen consumption was measured for 15 min.

Lactate production

Lactate was measured in culture medium after 7 days of treatment in 96-well plates using the Abcam® lactate fluorometric assay kit. Briefly, conditioned culture medium was removed and diluted in lactate assay buffer before being added to 96-well plates containing the lactate enzyme mix and probe. Appropriate calibration standards were also made and added to the plate. The reaction was allowed to proceed for 30 min in the dark before fluorescence was measured at an excitation wavelength of 535 nm and emission wavelength 590 nm.

Statistical analysis

Six separate independent experiments were undertaken using six different cultures on different days including Western blotting. No assumptions were made about the distribution of the data and non-parametric testing was performed using Kruskal–Wallis analysis of variance and Mann Whitney post hoc testing as appropriate. A P value of <0.05 was taken to be significant.


TRX-R and GPx activity

The concentrations of the inhibitors chosen for the present study were determined from initial experiments where maximal inhibition of the mitochondrial isoforms with no significant loss in cytosolic enzyme activity was seen. Our results showed that mitochondrial TRX-R and GPx enzyme activity decreased maximally at 2 μM and 4 mM respectively (P=0.004 and P=0.024 respectively; Figures 1A and 1C), with no change in cytosolic enzyme activity (Figures 1B and 1D). Higher concentrations decreased mitochondrial isoform enzyme activities a little further, but with loss of mitochondrial selectivity such that cytosolic enzyme activities were also decreased.

Figure 1 Thioredoxin reductase and glutathione peroxidase activity in mitochondria and cytosol

Inhibition of TRX-R and GPx in mitochondrial (panels A and C) and cytosolic (panels B and D) fractions from endothelial cells treated with solvent control (untreated), auranofin or thiomalic acid. Box and whisker plots show the median, interquartile and full ranges (n=6). P value shown is from a Kruskal–Wallis test. *Significantly different from untreated cells (Mann Whitney U test p<0.05).

Cell viability

Cell viability was 100% for all cell treatments, measured at 24 h or 7 days, with the exception of cells exposed to LPS/PepG plus auranofin for 7 days treatment only, where the viability decreased to a median (interquartile range) of 69.6 (63–79)% (P=0.0022).

Cytokine production

As expected, IL-6 and IL-8 concentrations were significantly increased in cells exposed to LPS/PepG compared with untreated cells (P<0.0001; Figure 2). There was no effect of auronofin or thiomalic acid on LPS/PepG-induced cytokine production (Figure 2).

Figure 2 IL-6 and IL-8 concentrations in cell culture supernatants from endothelial cells

The cells were treated with solvent control (untreated), 2 μg/ml LPS plus 20 μg/ml PepG with 2 μM auranofin or 4 mM thiomalic acid for 24 h. Box and whisker plots show median, interquartile and full range (n=6). The P value shown is from a Kruskal–Wallis test. *Significantly different from untreated cells (Mann Whitney U test P<0.05).

Total intracellular radical production

To assess whether inhibition of either PRX-3 or mGPx-1 under conditions of sepsis would induce a higher rate of intracellular oxidative stress, we incubated cells with LPS/PepG and with auranofin or thiomalic acid. The results showed that the rate of total intracellular radical formation in cells cultured under our conditions of sepsis was significantly greater than in the untreated cells (P=0.03; Figure 3). The rate was higher still in cells exposed to the inhibitors in addition to LPS and PepG (P=0.03; Figure 3). Radical production was also slightly higher in cells treated with the inhibitors alone (P=0.041; Figure 3).

Figure 3 Oxidative stress (total radical production) measured as the rate of change in fluorescence in endothelial cells

The cells were treated with solvent control (untreated), 2 μg/ml LPS plus 20 μg/ml PepG with and without 2 μM auranofin or 4 mM thiomalic acid or 50 μM hydrogen peroxide. Box and whisker plots show median, interquartile and full range (n=6). The P value shown is from a Kruskal–Wallis test. *, significantly different from untreated cells; +Significantly different from LPS and/or PepG alone (Mann Whitney U test p<0.05).

Mitochondrial thioredoxin and glutathione

To evaluate the inhibition of PRX-3 or mitchondrial GPx with auranofin or thiomalic acid respectively, on overall glutathione and TRX-2 systems, we measured intra-mitochondrial expression of TRX-2 protein and GSH concentrations following a 7 day septic insult. Mitochondrial TRX-2 protein levels were higher in all LPS- and PepG-treated cells than in untreated cells (P<0.001; Figure 4A). However, cells treated with thiomalic acid in addition to LPS and PepG had significantly greater TRX-2 expression than other cell treatments (P=0.01; Figure 4A). In comparison, the intra-mitochondrial GSH/GSSG ratio was lower in cells treated with LPS/PepG (P<0.001; Table 1) and lower still in cells also treated with auranofin (P<0.001). Cells treated with thiomalic acid, with or without LPS/PepG, had similar mitochondrial GSH/GSSG ratios to untreated cells. In addition, all cell treatments resulted in a significant decrease in the total GSH/GSSG ratio (P=0.0159; Table 1).

View this table:
Table 1 Ratio of reduced/oxidised glutathione (GSH/GSSG)

Median (interquartile range), n=6. P=0.0089 for total GSH/GSSG and P=<0.0001 for mitochondrial GSH/GSSH (Kruskal–Wallis test). *, significantly different from untreated cells (P<0.05); †significantly different from LPS/PepG alone (P<0.05).

Figure 4 Expression of key proteins in mitochondria

Mitochondria were extracted from endothelial cells treated with solvent control (untreated), 2 μg/ml LPS plus 20 μg/ml PepG with and without 2 μM auranofin or 4 mM thiomalic acid, or 50 μM hydrogen peroxide. HPRT was used a loading control. Densitometric analysis and representative Western blots of protein expression of TRX-2 (A), mGPx-1 (B) and PRX-3 (C) are shown. Box and whisker plots show median, interquartile and full range from six separate cultures (n=6). The P value shown is from a Kruskal–Wallis test. *Significantly different from untreated cells; +Significantly different from LPS and/or PepG alone (Mann Whitney U test P<0.05).

Expression of PRX-3 and mGPx-1

We determined the effect of inhibition of PRX-3 and mGPx-1 activity on subsequent mitochondrial protein expression of these enzymes after 7 days of treatment with the inhibitors plus LPS/PepG. In cells treated with either LPS or PepG, or with LPS/PepG and the inhibitor, there was higher PRX-3 expression, which was greatest in cells treated with auranofin (P<0.001; Figure 4C). Expression of mGPx-1 was higher in all cell treatments compared with untreated cells and was highest in cells treated with auranofin plus LPS/PepG (P<0.05; Figure 4B).

Caspase 3 and 7 activation

The effect of inhibition of PRX-3 and mitochondrial GPx on apoptosis was determined by measuring the activation of caspase 3 or 7 following 7 days of culture. There was no effect of LPS and PepG exposure when compared with the untreated cells (P=0.3; Figure 5A). However, when cells were treated with auranofin, which inhibits TRX-2R, both with and without LPS/PepG, caspase activation was higher than in cells without auranofin (P<0.03; Figure 5A).

Figure 5 Caspase 3 and 7 activation and mitochondrial membrane potential

Endothelial cells were treated with solvent control (untreated), 2 μg/ml LPS plus 20 μg/ml PepG with and without 2 μM auranofin or 4 mM thiomalic acid, or 50 μM hydrogen peroxide (or 1 mM rotenone as control, JC-1 only). Caspase 3 and 7 activation was measured by luminescence (A). Mitochondrial membrane potential was measured using JC-1 (C). Box and whisker plots show median, interquartile and full range (n=6). The P value shown is from a Kruskal–Wallis test. *Significantly different from untreated cells; +Significantly different from LPS and/or PepG alone (Mann Whitney U test P<0.05).

Mitochondrial membrane potential

Mitochondrial membrane potential was determined by the ratio of red/green JC-1 fluorescence after 7 days of treatments (Figure 5B). Cells exposed to LPS and PepG had a significantly lower JC-1 fluorescence ratio than untreated cells, indicating the loss of mitochondrial membrane potential (P<0.001; Figure 5B). In those cells treated with LPS/PepG plus auranofin, there was an even greater loss of mitochondrial membrane potential (P=0.01 and P<0.001 respectively; Figure 5B). This was not seen with thiomalic acid-treated cells.

Metabolic activity, oxygen consumption, ATP/ADP ratio and lactate production

Investigation of the functional capacity of mitochondria after 7 days of septic insult was achieved by measuring the metabolic activity as determined by the rate of reduction of AlamarBlue™ by mitochondrial respiratory complex activity, total oxygen consumption, ATP/ADP ratios and lactic acid production (Table 2). The metabolic activity in cells treated with auranofin plus LPS/PepG was significantly higher than all other cell treatments (P<0.001). Cells treated with LPS/PepG had significantly reduced oxygen consumption rates (P=0.028) and ATP/ADP ratios (P<0.001), and increased lactate formation (P<0.001). However, cells treated with LPS/PepG plus auranofin had a considerably lower oxygen consumption rate (P=0.02) and ATP/ADP ratio (P<0.01) and a large increase in lactate formation (P<0.001).

View this table:
Table 2 Mitochondrial metabolic activity, oxygen consumption, ATP/ADP and lactate

Values are median (interquartile range), n=6. For Kruskal–Wallis test P values, see the main text. *Significantly different from untreated cells (P<0.05); +Significantly different from LPS/PepG alone (P<0.05).


We have shown that inhibition of the mitochondrial TRX-2 or glutathione systems in our endothelial cell model of sepsis resulted in a higher total intracellular radical production and loss of mitochondrial membrane potential, with altered protein expression of various components related to these systems. Inhibition of the TRX-2 system resulted in higher mitochondrial metabolic activity, lower ATP/ADP ratios, lower oxygen consumption, increased lactate formation and evidence of caspase 3 and 7 activation. Inhibition of the mGSH system had little effect on mitochondrial function. This suggests that the TRX-2 system may be more important preserving mitochondrial function in endothelial cells under conditions of sepsis.

Sepsis is associated with increased pro-inflammatory cytokine production, oxidative stress, consumption of endogenous antioxidants and mitochondrial damage [38]. Tight control of cellular redox homoeostasis is essential for protection against oxidative damage and for maintenance of normal metabolism. Therefore mitochondrial generation of ROS is strictly regulated by mitochondrial antioxidant enzymes, including MnSOD, mGSH/mGPx-1 and the TRX-2/PRX-3 systems [19,30]. Although both systems remove mitochondrial hydrogen peroxide, redox potentials differ substantially between the two to facilitate the complex multiple signalling roles of ROS [31]. Auranofin is a co-ordinated gold(I) compound that reacts with selenol-containing residues and is a potent inhibitor of purified TRX-R protein. At the concentrations used in our in vitro study, we showed inhibition of mitochondrial TRX-R only, which resulted in no loss in cell viability in unstimulated cells (higher concentrations were required to inhibit the cytosolic TRX-R isoform and caused a 50% decrease in cell viability). This concentration effect could be due to auranofin's characteristics as it belongs to the phosphine class of compounds that can be selectively concentrated within mitochondria at low concentrations. In addition, the concentrations required to inhibit the isoforms of TRX-R is cell-type-specific, indicating the complex nature of the thioredoxin system (extensively reviewed in [32,33]). Thiomalic acid is a glutathione mimetic that can bind to active-site SH groups. We inhibited mGPx-1 exclusively at the concentrations used in the present study. It been shown previously that mitochondrial GPx is much more sensitive to thiomalic acid than the cytosolic isoform [34].

During sepsis, the basal rate of ROS production increases, specifically mitochondrial superoxide formation [4,13,14]. This is presumed to deplete antioxidants, increase free radical formation, and is associated with pro-inflammatory cytokine production and ultimately damage to mitochondria [311,16]. Antioxidant therapies directed at mitochondria are now being studied, and we and others have demonstrated beneficial effects on mitochondrial function both in vitro and in animal models of sepsis [16,17]. In endothelial cells treated with LPS/PepG plus inhibitors, the rate of total radical formation was enhanced above that of cells treated with LPS/PepG alone. However, this increase did not augment IL-6 or IL-8 production when compared with LPS/PepG-stimulated cells.

Mitochondrial oxidative stress was assessed by measuring intra-mitochondrial TRX-2 expression and the intra-mitochondrial GSH/GSSG ratio compared with the total GSH/GSSG ratio. After 24 h (results not shown) and at 7 days, we found that LPS/PepG treatments increased TRX-2 expression. A higher level of expression was seen in LPS/PepG-treated cells treated with thiomalic acid and is probably compensatory for mGPx-1 inhibition. In addition, LPS/PepG treatment resulted in a smaller decrease in both intra-mitochondrial and total GSH/GSSG ratios. However, auranofin caused an even lower intra-mitochondrial GSH/GSSG ratio without affecting total GSH/GSSG ratio, suggesting mitochondrial oxidative stress. We performed redox Western blotting by covalent modification of any remaining reduced (after experimental treatment) TRX-2 in the mitochondrial lysates to assess the redox form of TRX-2. However, we found only one strong band indicating no oxidized TRX-2 in any treatment group. This result is consistent with a previous study using bovine aortic endothelial cells where TRX-2 was present only in the reduced form even when these cells were exposed to extreme oxidative stress (12 mM hydrogen peroxide) [35].

PRXs may be damaged by excessive ROS production such that a condition termed hyper-oxidation and inactivation occurs [20]. Previous work has shown that endothelial mitochondrial PRX-3 is much less sensitive to oxidation and inactivation than cytosolic PRX isoforms [26]. In the present study we found only reduced PRX-3 in all cell treatments using redox Western blotting. This may be due to the effects of the enzyme sulfiredoxin, which is able to reactivate PRXs following oxidation. This enzyme is up-regulated during oxidative stress and sepsis and can migrate into mitochondria, reactivating oxidised PRX-3 [36]. We propose that sulfiredoxin expression may be higher in endothelial cells than other cell types, explaining the increased resistance of PRX-3 to oxidation.

ROS removal in oxidative stress and sepsis is controlled in part by the level of endogenous antioxidant defences. The rate of removal of hydrogen peroxide in vivo by GPx-1 has been presumed to be independent of GSH concentration. However, kinetic studies have now revealed that this is not the case [37]. It has been shown in vitro using purified enzyme that the total upper limit of hydrogen peroxide removal by GPx-1 is approximately 700 nM, although, under conditions of oxidative stress and sepsis, levels of intracellular hydrogen peroxide may reach 5 μM [38]. This would cause the pool of mGSH to fall rapidly, which is replaced very slowly by diffusion of cytosolic GSH into the mitochondrion (reviewed in [39]). The relative reactivity of PRX-3 with hydrogen peroxide is approximately the same as mGPx-1. However, PRXs have been shown to be expressed at higher levels than GPx/glutaredoxins [31,36] and TRX-2 contains a mitochondrial-targeting sequence and is actively transported into the mitochondrion as our present results suggest.

Mitochondrial dysfunction is thought to be linked to the development of sepsis-induced organ dysfunction [311,17,40]. Many studies have shown that derangements in cellular oxygen utilization result from damage to the electron transport chain and damage to the inner mitochondrial membrane that can result in impairment of ATP generation, increased lactic acid production and mitochondrial swelling. In the present paper we assessed the relative roles of the TRX-2 and mGSH systems on mitochondrial dysfunction in LPS/PepG-treated cells. Collapse of the mitochondrial membrane potential, which may further have an impact on the oxidative phosphorylation pathway for ATP generation, may trigger apoptotic cell death pathways. The loss of mitochondrial membrane potential with a low ATP/ADP ratio, reduced oxygen consumption and increased lactate production, but no change in metabolic activity was seen in LPS and/or PepG-treated cells and was similar to cells where mGPx was inhibited. However, inhibiting TRX-2R caused further loss of membrane potential, substantially reduced oxygen consumption, a greater increase in lactate production and a large decrease in the ATP/ADP ratio, accompanied by a subsequent 5-fold increase in metabolic activity. The increase in metabolic activity is most likely due to the low ATP concentration acting as a switch to produce more ATP. AlamarBlue™ is an accurate and sensitive indicator of mitochondrial function as it is an exceptional detector of reduction of all of the elements of the electron transport chain that are not affected by oxidants [41].

Changes to the intracellular redox state (oxidative stress) have important implications for cell survival. Oxidation of TRX-2 and/or mGSH increases the susceptibility of the cell to oxidants and promotes apoptosis through the intrinsic pathway [42,43]. The low ATP/ADP ratio and mitochondrial dysfunction seen upon inhibition of TRX-R2 in LPS/PepG-treated cells was accompanied by increased caspase 3 and/or 7 activation and a 30% reduction in cell viability. This shows that the TRX-2 system in endothelial cells is important to prevent cell death via adequate LPS/PepG-induced hydrogen peroxide removal.

In conclusion we have shown that, in our in vitro endothelial cell model of sepsis, the proteins of the TRX-2 system are better resistant to the effects of oxidative stress and appear to have a larger protective role against mitochondrial dysfunction induced by LPS/PepG exposure than the mGSH system. Endothelial cells have an important role in host defence and inflammation during sepsis. However, further studies will be necessary to evaluate the relative importance of the mitochondrial hydrogen peroxide-removal systems of other cell types during LPS/PepG insult.


Damon Lowes conceived the study, carried out the experiments and prepared the first draft of the paper. Helen Galley provided intellectual input to the study design and finalized the paper for publication prior to submission.

Abbreviations: CDFH, 5,(6)-carboxy-2,7′-dichlorofluorescein; GPx, glutathione peroxidase; HPRT, hypoxanthine–guanine phosphoribosyltransferase; HUVEC, human umbilical vein endothelial cell; IL, interleukin; JC-1, (5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbocyanine iodide; LPS, lipopolysaccharide; mGPx-1, mitochondrial GPx-1; mGSH, mitochondrial glutathione; MODS, multiple organ dysfunction syndrome; MnSOD, manganese-containing superoxide dismutase; PepG, peptidoglycan; PRX, peroxiredoxin; ROS, reactive oxygen species; TRX, thioredoxin; TRX-R, TRX reductase; TRX-2R, TRX reductase-2


View Abstract