Inositol-requiring enzyme 1 alpha (IRE1α) is an endoplasmic reticulum (ER)-transmembrane endonuclease that is activated in response to ER stress as part of the unfolded protein response (UPR). Chronic activation of the UPR has been implicated in the pathogenesis of many common diseases including diabetes, cancer, and neurological pathologies such as Huntington's and Alzheimer's disease. 7-Hydroxy-4-methyl-2-oxo-2H-chromene-8-carbaldehyde (4µ8C) is widely used as a specific inhibitor of IRE1α ribonuclease activity (IC50 of 6.89 µM in cultured cells). However, in this paper, we demonstrate that 4µ8C acts as a potent reactive oxygen species (ROS) scavenger, both in a cell-free assay and in cultured cells, at concentrations lower than that widely used to inhibit IRE1α activity. In vitro we show that, 4µ8C effectively decreases xanthine/xanthine oxidase catalysed superoxide production with an IC50 of 0.2 µM whereas in cultured endothelial and clonal pancreatic β-cells, 4µ8C inhibits angiotensin II-induced ROS production with IC50 values of 1.92 and 0.29 µM, respectively. In light of this discovery, conclusions reached using 4µ8C as an inhibitor of IRE1α should be carefully evaluated. However, this unexpected off-target effect of 4µ8C may prove therapeutically advantageous for the treatment of pathologies that are thought to be caused by, or exacerbated by, both oxidative and ER stress such as endothelial dysfunction and/or diabetes.
The endoplasmic reticulum (ER) is the site for the synthesis and processing of secretory and membrane proteins. Perturbations in ER homeostasis that interfere with protein folding result in the activation of an adaptive response termed the unfolded protein response (UPR) [1,2]. Chronic activation of the UPR has been implicated in many human pathologies including infectious, neurodegenerative, autoimmune, and metabolic conditions [3–5].
The UPR is classically mediated by three ER-transmembrane proteins: PKR-like ER kinase (PERK), activating transcription factor 6 (ATF6), and inositol-requiring enzyme 1α (IRE1α). IRE1α senses perturbations in ER homeostasis via its luminal domain which results in a conformational change [6,7]. This, in turn, promotes oligomerisation and the activation of its cytoplasmic protein kinase and RNAse domain. Activation of IRE1α leads to the cleavage and subsequent ligation of the mRNA encoding the X-box binding protein-1 (XBP1), resulting in a frame shift and the synthesis of a truncated and transcriptionally active spliced form of XBP1 (XBP1s) . In an attempt to relieve ER stress, XBP1s enhances the transcription of genes important in facilitating protein folding including the ER chaperone glucose-regulated protein 78 [6,9]. However, chronic IRE1α activation can lead to programmed cell death through the activation of multiple signalling pathways (e.g. [6,10–12]).
A number of small molecule inhibitors of IRE1α have been identified and characterised including 7-hydroxy-4-methyl-2-oxo-2H-chromene-8-carbaldehyde (4µ8C) . 4µ8C is an aromatic aldehyde that binds to IRE1α's RNAse domain and inhibits its activity. This inhibitor has provided valuable insights into defining the mechanism of action of IRE1α in both cellular physiology and pathophysiology (e.g. [13–19]). 4µ8C has also been used in animal studies  although its effectiveness in vivo is uncertain. A recent study investigating the effect of 4µ8C on insulin secretion has highlighted the potential of this inhibitor to have off-target effects . Consistent with this, we demonstrate in this report that 4µ8C not only inhibits IRE1α but also acts as a potent scavenger of reactive oxygen species (ROS) both in cell-free assays and in cultured cells. In the light of this discovery conclusions reached using 4µ8C as a specific IRE1α, RNAse inhibitor should be carefully considered.
Mouse insulinoma 6 (MIN6) cells  were used between passages 25 and 35 at ∼80% confluence and cultured as previously described . Mouse microvascular cerebral endothelial cells, bEnd.3 , were used between passages 24–34 (ATCC CRL-2299) and cultured in Dulbecco's Modified Eagle Medium (Life Technology, Australia) supplemented with 10% foetal bovine serum (FBS; Bovogen, Australia), at 37°C and 5% CO2.
Quantification of superoxide in a cell-free enzyme system
The xanthine/xanthine oxidase cell-free assay coupled with 5 μmol/1 lucigenin-enhanced chemiluminescence  was used to assess the superoxide scavenging properties of 4µ8C. Briefly, to initiate the reaction, xanthine oxidase (50 mU/ml) was added to Krebs-Hepes solution (NaCl 99 mmol/l, KCl 4.7 mmol/l, KH2PO4 10 mmol/l, MgSO4 1.2 mmol/l, NaHCO3 25 mmol/l, glucose 11 mmol/l, CaCl2 2.5 mmol/l, and EDTA 0.026 mmol/l, pH 7.4) containing xanthine (100 μmol/l) and lucigenin (5 μmol/l). Where indicated, experiments were performed in the absence or presence of superoxide dismutase (SOD; 250 U/ml), 4µ8C (3 nM–30 µM), DMSO ((0.1%) vehicle for 4µ8C), or Tiron (0.3–3000 µM). Superoxide counts were measured using a BMG Clariostar plate reader (BMG Labtech, Melbourne, Australia). Background counts were then subtracted, and the superoxide level was expressed as relative luminescence units (RLU).
Quantification of cellular superoxide levels
Superoxide levels were measured using L-012 (100 µmol/l) enhanced chemiluminescence as recently described . Cells were plated on a 96-well Optiplate (PerkinElmer, Melbourne, Australia). Upon treatments, 20 µl of 1 mol/l L-012 (in Krebs-Hepes) was added per well in semi-darkness to give a final concentration of 100 µmol/l. After treatments, superoxide counts were measured using a BMG Clariostar plate reader over 90 min (BMG Labtech, Melbourne, Australia; 45 cycles, 3 s per well). The accumulated luminescence counts obtained were subtracted from the corresponding vehicle control and expressed as RLU.
Measurement of intracellular ROS generation
Intracellular ROS generation was measured using the cell-permeable ROS detector 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate (CM-H2DCFDA) (Thermo Scientific, Australia). Briefly, cells incubated in phenol-red-free growth media were loaded with CM-H2DCFDA (10 μmol/l). The oxidation of CM-H2DCFDA was captured kinetically (Ex/Em: 485/528 nm) at 37°C for 90 min. ROS production is expressed as accumulated relative fluorescence units.
Data are expressed as mean ± SEM, unless otherwise stated. Data were analysed by one-way ANOVA followed by Tukey's post hoc test for multiple comparison between means using Prism 6 (GraphPad Software, USA). Differences were considered statistically significant at P < 0.05.
4µ8C is a potent superoxide scavenger in a cell-free assay
While studying the role of angiotensin II (AngII), an inducer of ROS, in the development of β-cell dysfunction , we observed that the commonly used IRE1α inhibitor 4µ8C inhibited not only the activation of IRE1α but also PERK, indicating that 4µ8C may have off-target effects (unpublished observations). Many naturally occurring antioxidants are polyphenolic compounds that act as free radical scavengers . As 4µ8C  (Figure 1a) is a polyphenolic compound, it also has the potential to act as a free radical scavenger. To investigate this possibility, cell-free xanthine/xanthine oxidase assays were performed using the chemiluminescent probe lucigenin. As expected, the addition of xanthine oxidase to a solution containing xanthine and lucigenin led to a rapid increase in superoxide levels (Figure 1b), which was abolished in the presence of SOD, thus confirming that the reaction between xanthine and xanthine oxidase generates superoxide . Furthermore, the addition of 4µ8C alone (i.e. in the absence of xanthine oxidase) had no effect on relative luminescence (Figure 1b). Next, we measured superoxide levels in the presence or absence of either increasing concentrations of the IRE1α inhibitor 4µ8C or as a positive control, Tiron, a classical ROS scavenger . 4µ8C (3 mM to 30 µM) (Figure 1c) or Tiron (0.3 µM to 3 mM) (Figure 1d) reduced superoxide levels generated by the xanthine/xanthine oxidase cell-free assay in a concentration-dependent manner. Tiron inhibited xanthine/xanthine oxidase superoxide production with an IC50 of 7.3 × 10−6 M, whereas 4µ8C inhibited superoxide production with an IC50 of 2.0 × 10−7 M (Figure 1e). Therefore, 4µ8C is a more potent superoxide scavenger than Tiron in this cell-free enzyme assay.
The superoxide scavenging effects of 4µ8C.
4µ8C inhibits AngII-induced superoxide production in pancreatic β-cells and brain endothelial cells
AngII increases cellular superoxide production by activating the NADPH oxidases (NOX) via the angiotensin type 1 receptor (AT1R) . Therefore, to determine whether 4µ8C also acts as an antioxidant in cells, the mouse pancreatic β-cell line, MIN6, and the mouse cerebral endothelial cell line, bEnd3, were treated with AngII in the presence or absence of 4µ8C (30 µM) or, as control, the AT1R antagonist irbesartan (IRB). Superoxide levels were then measured using the luminol-based chemiluminescent probe, L-012. As anticipated, AngII caused a significant increase in superoxide production in both MIN6 (Figure 2a) and bEnd3 cells (Figure 2b) relative to control, and this was blocked by IRB (Figure 2a,b). Importantly, AngII-induced increase in superoxide in both cell lines was effectively inhibited by 30 µM 4µ8C (Figure 2a,b), a concentration that is commonly used to inhibit IRE1α's RNAse activity.
AngII-induced superoxide formation in cells is blocked by 4µ8C.
To formally demonstrate that AngII-generated superoxide production was through the activation of NOX rather than ER stress-induced IRE1α activation, MIN6 and bEnd.3 cells were treated with AngII in the presence or absence of two NOX inhibitors, diphenyleneiodonium (DPI) and apocynin  (Figure 2c,d). AngII treatment caused an increase in superoxide production, which was effectively inhibited by both of the NOX inhibitors. This provides evidence that AngII-induced superoxide was generated through the activation of NOX and not IRE1α activation.
4µ8C concentration-dependent inhibition of AngII-induced superoxide production in MIN6 cells and bEnd.3 cells
To investigate the potency of 4µ8C at inhibiting AngII-induced ROS production in cells, MIN6 and bEnd.3 cells were treated with AngII in the presence of increasing concentrations of 4µ8C and superoxide production was measured using L-012 (Figure 3). The addition of 4µ8C led to the concentration-dependent inhibition of AngII-generated superoxide with an IC50 of ∼1.92 µM in MIN6 cells (Figure 3a) and 0.293 µM in bEnd.3 cells (Figure 3b). As 4µ8C inhibits IRE1α in mammalian cultured cells with an IC50 of 6.89 µM , 4µ8C antioxidant activity is more potent than its ability to inhibit IRE1α RNAse activity.
Dose-dependent effects of 4µ8C on superoxide formation.
4µ8C inhibits AngII-induced ROS production in MIN6 and bEnd.3 cells
As further evidence that 4µ8C is an antioxidant in mammalian cells, MIN6 and bEnd.3 cells were treated with AngII in the presence and absence of 4µ8C and changes in intracellular ROS were detected using 5-(and-6)-chloromethyl-2′,7′-dichlorodihydrofluorescein diacetate (CM-H2DCFDA), a fluorogenic dye that reacts with hydroxyl, hydrogen peroxide, peroxynitrite, and, to a lesser extent, superoxide . Because hydrogen peroxide is a downstream product of superoxide, CM-H2DCFDA fluorescence is often used to implicate superoxide production . Using CM-H2DCFDA, we confirmed that AngII increases ROS production in both MIN6 cells (Figure 4a) and bEnd.3 cells (Figure 4b), and that this is inhibited by apocynin. Importantly, AngII stimulates that ROS production was also inhibited by 4µ8C (30 µM). Thus, 4µ8C acts as a ROS scavenger in cultured mammalian cells.
The intracellular free radicals scavenging effect of 4µ8C.
Cell-permeable, low-molecular weight inhibitors are used extensively to understand the role of specific enzymes in cellular physiology. Unfortunately, these compounds often lack specificity and display multiple off-target effects that can lead to the misinterpretation of data. Our data demonstrate that 4µ8C, at concentrations widely used to inhibit IRE1α activity, also acts as a potent ROS scavenger both in a cell-free assay and in cultured cells. Importantly, no other studies have shown that 4µ8C acts as an ROS scavenger.
The IC50 values for scavenging ROS by 4µ8C differed in the two cell types used in this study. Although this may reflect the relative ability of AngII to stimulate ROS in these cells, comparison of ROS production between different cell types is difficult due to differences in, for example, an ROS detector probe loading into the cell. Regardless, the effectiveness of the IRE1α inhibitor as a ROS scavenger is likely to be dependent on the amount of ROS being generated. In cells, such as phagocytes, that generate high concentration of ROS, the IC50 of the inhibitor against IRE1α RNAse activity may be greater than its IC50 for scavenging ROS. Interestingly, 4µ8C has been used in macrophages to provide evidence for the role of IRE1α in ROS-dependent killing of bacteria . Whether some or all of the effects of 4µ8C reported on ROS-dependent killing were mediated by the antioxidant properties of 4µ8C, as demonstrated in the report, is unclear but again highlights the importance of this study. Recently, Sato et al. examined the role of IRE1α on insulin secretion in pancreatic β-cells using 4µ8C. They discovered that 4µ8C inhibited insulin secretion even in cells lacking the IRE1α RNAse domain, demonstrating that 4µ8C is able to block insulin secretion independent of the IRE1α/XBP1 pathway . Given the results presented in this report, it is possible that the inhibitory effects of 4µ8C on insulin secretion are through the scavenging of ROS.
4µ8C ability to act as a potent ROS scavenger is likely due to its coumarin-type conjugated structure (Figure 1a), which can enable stabilisation of the free radical either by donation of a hydrogen atom or an electron . This stabilisation is a result of the numerous resonance forms that are possible once the 4µ8C radical is formed, i.e. the 4µ8C radical is resonance delocalised.
This off-target effect is a particular problem for researchers interested in delineating the role of IRE1 in ER stress responses, as oxidative stress can induce ER stress and conversely ER stress can promote oxidative stress [1,33]. Despite the potential problems of interpreting data using 4µ8C, off-target effects of low-molecular cell-permeable inhibitors can be advantageous in a clinical setting. For example, Gleevec (Imatinib), initially developed as an inhibitor of BCR-Abl, was later found to also inhibit the c-kit and platelet-derived growth factor receptor-A. This off-target effect proved advantageous in its use as a treatment of gastrointestinal stromal tumour . Given that we demonstrate that 4µ8C can also act as a potent antioxidant, this may prove therapeutically advantageous for the treatment of pathologies that are thought to be caused by, or exacerbated by, both oxidative and ER stress, e.g. endothelial dysfunction or diabetes [1,4,35–37]. Therefore, it is possible that 4µ8C, or a more pharmokinetically favourable structural analogue, may prove to have a therapeutic value against such diseases.
angiotensin type 1 receptor
mouse microvascular cerebral endothelial cells
inositol-requiring enzyme 1
mouse insulinoma 6
PKR-like ER kinase
protein kinase R
reactive oxygen species
relative luminescence units
unfolded protein response
S.M.H.C. helped in study design, performed the experiments, analysed the data and contributed to the writing of the manuscript. A.A.M and A.B. helped with the acquisition of data and provided scientific advice. M.P.L. contributed to discussions and reviewed/edited the manuscript. T.P.H. conceived and designed the experiments, contributed to the acquisition of data and data analysis, and wrote the manuscript.
This work was supported by a RMIT University research development fund awarded to T.P.H.
The Authors declare that there are no competing interest[AQ6] s associated with the manuscript.