Induction of excessive levels of reactive oxygen species (ROS) by small-molecule compounds has been considered a potentially effective therapeutic strategy against cancer cells, which are often subjected to chronic oxidative stress. However, to elucidate the mechanisms of action of bioactive compounds is generally a time-consuming process. We have recently identified NPD926, a small molecule that induces rapid cell death in cancer cells. Using a combination of two comprehensive and complementary approaches, proteomic profiling and affinity purification, together with the subsequent biochemical assays, we have elucidated the mechanism of action underlying NPD926-induced cell death: conjugation with glutathione mediated by GST, depletion of cellular glutathione and subsequent ROS generation. NPD926 preferentially induced effects in KRAS-transformed fibroblast cells, compared with their untransformed counterparts. Furthermore, NPD926 sensitized cells to inhibitors of system xc−, a cystine-glutamate antiporter considered to be a potential therapeutic target in cancers including cancer stem cells. These data show the effectiveness of a newly identified ROS inducer, which targets glutathione metabolism, in cancer treatment.
Reactive oxygen species (ROS) are natural by-products of cellular metabolism. Although certain ROS (e.g. H2O2) play important roles in cellular signal transduction and innate immune responses, other highly reactive species or excessive levels of ROS cause indiscriminate damage to cellular nucleic acids, proteins and lipids. Therefore, for redox homoeostasis and to limit cellular damage, cells control ROS levels by balancing ROS generation with ROS elimination by ROS-scavenging molecules, such as superoxide dismutases and glutathione peroxidase [1,2].
Approximately 20% of human tumours have activating point mutations in RAS oncogenes, with ~85% of these occurring in the KRAS gene . In clinical practice, the KRAS mutation is associated with poor prognosis, limited treatment options, and chemotherapy and radiotherapy resistance [4,5]. Therefore therapeutic methods for selectively killing oncogenic KRAS-harbouring cancer cells are required. Higher levels of ROS are generated in cancer cells harbouring oncogenic KRAS than in normal cells [6,7]. Increased expression of NADPH oxidase 1 (Nox1), an endogenous superoxide-generating molecule, and mitochondrial dysfunction have been identified as potential mechanisms underlying KRAS-induced increases in ROS levels [6,7]. As elevated ROS levels promote cancerous phenotypes, including excessive proliferation , genetic instability  and angiogenesis , such intrinsic oxidative stresses in cancer cells are often considered adverse events. In contrast, chronic oxidative stress also renders cancer cells highly vulnerable to chemical agents that further elevate ROS to levels that induce cell death [1,11]. There exist several lines of evidence that ROS-inducing compounds, such as lanperisone and erastin, selectively induce cell death in oncogenic KRAS-harbouring cells [12–14]. Whereas current molecular-targeted therapeutic strategies still face significant challenges because of acquired drug resistance and the genomic instability of cancer cells [15,16], targeting the unique biochemical alterations in cancer cells (such as ROS levels) suggests a feasible approach towards achieving therapeutic effectiveness . Therefore induction of excessive levels of ROS generation is now considered an effective strategy for selective killing of cancer cells [17–21].
Despite the effectiveness of ROS-inducing compounds in cancer treatment, to elucidate the precise mechanisms of action of bioactive compounds may generally be a challenging and time-consuming process. As elucidation of the mechanisms of action of bioactive compounds often provides investigators many insights into cellular functions and may render a therapeutic regimen more successful, various comprehensive analysis methods have been developed [22,23]. Profiling analysis based on induced phenotypic responses and affinity purification of the binding molecules are representative and complementary approaches to elucidate the mechanisms of action of bioactive compounds. Therefore we have previously developed and utilized (i) the ChemProteoBase profiling system on the basis of comparison of proteomic perturbation induced by uncharacterized compounds of interest with those by well-characterized compounds [24–26], and (ii) a method for universal coupling involving a photo-cross-linking reaction to prepare affinity matrices [27,28].
In the present study, during the screening for compounds with anti-cancer activity, we have identified NPD926, a small molecule that induces rapid cell death. By using a combination of proteomic profiling and affinity purification, we have elucidated the mechanism of action of NPD926, that is conjugation with GSH in a GST-dependent manner, cellular GSH depletion and subsequent acute ROS generation. Furthermore, we have found that NPD926 selectively induces cytotoxicity to a greater extent in oncogenic KRAS-transformed cells than in healthy control counterparts and sensitized cells to inhibitors of system xc−, a cystine/glutamate antiporter considered to be a potential therapeutic target in cancers.
MATERIALS AND METHODS
The human cell lines Jurkat, HL-60, K562, U937, HT-29, MCF-7, MDA-MB-231, DU-145, SK-MEL-28, MKN74, OSRC2 and PANC-1 were cultured in RPMI 1640 (Invitrogen) containing 10% FBS, penicillin G (50 units/ml) and streptomycin (50 μg/ml). The remaining human cell lines HeLa, PC-3, HepG2, Hep3B, A549, A431, HT-1080, DMS114, HCT-116, and MIA PaCa-2 were cultured in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen) containing 10% FBS, penicillin G (50 units/ml) and streptomycin (50 μg/ml). Mouse embryonic fibroblast NIH 3T3 cells and its transformant NIH 3T3/KRAS cells  were cultured in Eagle's minimum essential medium (EMEM) (Sigma–Aldrich), containing 10% FBS, glutamine (0.3 g/l), penicillin G (50 units/ml) and streptomycin (50 μg/ml). All cell lines were incubated at 37°C in a humidified atmosphere containing 5% CO2.
Cell viability and growth assays
Jurkat cells were seeded in 96-well plates, cultured overnight and treated with test compounds for the indicated time. Cell viability was measured using the Trypan Blue exclusion assay with Trypan Blue solution (Sigma–Aldrich). Each cell line was seeded in 96-well plates, cultured overnight and treated with test compounds. Subsequently, cell viability was measured using the water-soluble tetrazolium salt 8 (WST-8) assay, using Cell Count Reagent SF (Nacalai Tesque) according to the manufacturer's instructions, with measurement of absorbance at 450 nm on a microplate reader (Perkin Elmer). The compounds used for screening were obtained from a chemical library of the RIKEN Natural Products Depository (NPDepo) . N-acetylcysteine (NAC), SB202190, 1-chloro-2,4-dinitrobenzene (CDNB), erastin and sulfasalazine (SSZ) were purchased from Sigma–Aldrich. GSH was purchased from Wako Pure Chemical Industries.
Profiling by ChemProteoBase
ChemProteoBase profiling analysis was performed as described previously . Briefly, HeLa cells were treated with NPD926 for 18 h. Proteome analysis of the cell lysate was performed by using the 2D difference gel electrophoresis (2D-DIGE) system (GE Healthcare), and images of the gels were analysed with Progenesis SameSpots (Nonlinear Dynamics). Of more than 1000 spots detectable in each 2D gel, 296 variational spots were found to be common between gels of reference-compound-treated cells and were selected as described previously . Next, the volume of each spot was normalized by using the average of the corresponding control values from DMSO-treated HeLa cells. From the normalized volumes of the 296 spots, cosine similarity between compounds was calculated, and hierarchical clustering analysis was performed using Cluster 3.0 (clustering method; centroid linkage with the means of uncentred correlation). The predictive dendrogram was visualized using Java Treeview 1.1.3.
Cellular ROS levels were measured as described previously with minor modifications . Jurkat cells were plated in 24-well plates, cultured overnight and treated with test compounds. Next, cells were incubated in PBS containing 10 mM carboxy-2′,7′-dichlorodihydrofluorescein diacetate (carboxy-H2DCF-DA) (Life Technologies) for 30 min at 37°C, washed with PBS and immediately analysed using a Cytomics FC500 flow cytometer (Beckman Coulter).
Western blot analysis
Cells were lysed by sonication in extraction buffer (20 mM Tris/HCl, pH 7.4, 150 mM NaCl, 25 mM NaF, 1 mM Na3VO4, 1 mM EDTA, 2.5 mM EGTA, 1 mM PMSF and 0.1% Tween 20) supplemented with a protease-inhibitor cocktail (Roche). Samples were subjected to SDS/PAGE and transferred onto a PVDF membrane (Millipore) for Western blot analysis. Membranes were incubated with the primary antibodies against caspase-3 (sc-7148; Santa Cruz Biotechnology), p38 mitogen-activated protein kinase (MAPK; prepared as described previously ), phospho-p38 MAPK (#4511; Cell Signaling Technology), or GST Pi 1 (GSTP1) (#3369; Cell Signaling Technology), and horseradish peroxidase-labelled secondary antibodies. They were then visualized with ImageQuant LAS 4000 (GE Healthcare), using the SuperSignal West Pico Chemiluminescence Substrate (Pierce).
Identification of NPD926-binding proteins
Compound-immobilized affinity beads were prepared by a photo-cross-linking method as described previously . Jurkat cells were harvested, washed with PBS and lysed in sonication in binding buffer (10 mM Hepes, pH 7.4, 150 mM NaCl, 50 μM EDTA and 0.05% Surfactant P20) supplemented with a protease-inhibitor cocktail. The cell lysate containing 8 mg of protein was then incubated with compound-immobilized beads or control beads for 12 h at 4°C. The reacted beads were washed with binding buffer, and the bound proteins were eluted with SDS/PAGE sample buffer, separated by SDS/PAGE and visualized by staining with Coomassie Brilliant Blue (CBB). The bound proteins were identified by MALDI–TOF-MS and LC–MS/MS analyses as described previously .
Detection of NPD926–SG conjugate
NPD926–SG conjugate, formed by conjugation of NPD926 to GSH, was detected by a binding reaction both in vitro and in cells. NPD926 (10 μM) and GSH (5 mM) were incubated at 25°C for 20 h in the presence or absence of recombinant GSTP, GST Mu 2 (GSTM2) and GST Alpha 1 (GSTA1) (500 nM; Prospec-Tany TechnoGene). After incubation, the reaction mixtures were extracted with 1-butatnol, re-suspended in methanol and analysed by HPLC with UV and ESI–MS detectors. Jurkat cells were plated in 24-well plates, cultured for 1 h, treated with NPD926 (10 μM) and then lysed in the culture medium by sonication. After removal of the insoluble components by centrifugation, the cell lysate, including both intracellular and extracellular components, was extracted with butan-1-ol, resuspended in methanol and analysed by HPLC with UV and ESI–MS detectors.
Jurkat cells were plated in 12-well plates, cultured for 2 h and treated with test compounds. Subsequently, the concentration of total cellular glutathione was measured using the Glutathione Assay Kit (Cayman Chemical) according to the manufacturer's instructions, with measurement of absorbance at 405 nm on a microplate reader (Molecular Devices).
NPD926 induces rapid cell death
To identify small molecules that exhibit anti-cancer activity, we have screened compounds from a chemical library of the RIKEN NPDepo. During the course of screening, we have identified a compound named ‘NPD926’ that exhibited cytotoxicity against a wide variety of human cancer cell lines (Figure 1A and Table 1). Interestingly, NPD926 rapidly induced cell death in T-cell lymphoma Jurkat cells (Figure 1B), whereas well-characterized anti-cancer drugs, such as mitomycin C, camptothecin (CPT), daunomycin and etoposide, did not affect cell viability 3 h after treatment at their fully effective respective concentrations (results not shown), suggesting a characteristic mode of action of NPD926. Thus we have attempted to elucidate the mechanism of action of NPD926 and its effects on cancer cells.
NPD926 induces rapid cell death in Jurkat cells
|Cell line||Organ||IC50 (μM)|
|A549||Non-small-cell lung cancer||20|
|DMS114||Small-cell lung cancer||11|
|Cell line||Organ||IC50 (μM)|
|A549||Non-small-cell lung cancer||20|
|DMS114||Small-cell lung cancer||11|
First, we examined whether caspase-3, a key effector molecule of apoptosis, was involved in NPD926-induced cell death. NPD926 induced the activation of caspase-3 prior to cell death (Figure 1C). However, the pan-caspase inhibitor benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone (z-VAD-fmk) did not suppress cell death induced by treatment with NPD926 at high concentration, whereas z-VAD-fmk completely suppressed cell death induced by an apoptosis inducer CPT (Figure 1D). These data suggest that NPD926-induced cell death in Jurkat cells occurs primarily in a caspase-independent manner.
NPD926 induces cellular ROS generation
To predict the mechanism underlying NPD926-induced cell death, we have performed profiling analysis using ChemProteoBase, a comprehensive database of cellular proteomic variation induced by treatment with well-characterized bioactive compounds. The hierarchical cluster analysis with 42 standard compounds (Figure 2A) and the cosine similarity score (Table 2) showed that the proteomic variation induced by NPD926 shared similarity with those induced by CDNB, a substrate for GST, and methyl ester of ethacrynic acid (EA), an inhibitor of GST. Both CDNB and EA are known to induce cellular ROS accumulation [33,34]. Among the 296 protein spots, proteins significantly increased or decreased by treatment with NPD926 or CDNB in HeLa cells are listed in Table S1 (http://www.biochemj.org/bj/463/bj4630053add.htm). Both in NPD926-treated cells and in CDNB-treated cells, the expression levels of HSP70 (heat-shock 70 kDa protein 1A/1B; spots #934, #949 and #1975 in Table S1) and HSP27 (heat-shock protein β-1; spots #1754 and #1757 in Table S1) were increased significantly. Up-regulation of HSP70 and HSP27 is known to be induced in response to various cellular stress, including oxidative stress . On the basis of these proteomic analyses, we have hypothesized that NPD926 induces cellular ROS generation.
NPD926 induces cell death via ROS generation
|Ranking||Cosine similarity||Compound||Target of compound|
|2||0.62||Ethacrynic acid methyl ester||GSTs, GSH|
|10||0.41||Purvalanol A||Cyclin-dependent kinases|
|Ranking||Cosine similarity||Compound||Target of compound|
|2||0.62||Ethacrynic acid methyl ester||GSTs, GSH|
|10||0.41||Purvalanol A||Cyclin-dependent kinases|
Indeed, NPD926 rapidly induced ROS generation 1 h after treatment (Figure 2B). Furthermore, pre-treatment of cells with NAC, an ROS scavenger, moderately reduced NPD926-induced ROS generation and cell death (Figures 2C and 2D), indicating that ROS generation is involved in NPD926-induced rapid cell death. Since ROS generation often induces activation of MAPKs , we have examined the effect of NPD926 on MAPK activity and confirmed that NPD926 induced phosphorylation of p38 MAPK (Figure 2E). Furthermore, pre-treatment with SB201290, a specific inhibitor of p38 MAPK, moderately suppressed NPD926-induced cell death (Figure 2D). These data indicate that NPD926 induces cell death through ROS generation and, to an extent, subsequent p38 MAPK activation.
GSTP is a specific NPD926-binding protein
To elucidate the mechanism underlying NPD926-induced ROS generation, we have proceeded to identify any NPD926-binding molecules in the cell lysate. To this end, we have prepared the affinity matrix, where NPD926 or negative control compounds, NPD10714 and NPD1671 (Figure 3A), was immobilized using a photo-cross-linking method. NPD10714 and NPD1671 exhibited ~20-fold and more than 100-fold less cytotoxicity than NPD926 respectively (Figure 3A). Using these affinity matrices, we have performed a pull-down assay followed by identification of the co-precipitated proteins in the Jurkat cell lysate by MALDI–TOF-MS or LC–MS/MS analyses. Among the detectable NPD926-binding proteins, only one protein, which was identified as GSTP, a Pi-class isoenzyme of GST, was specific for NPD926 but not structurally related to less active compounds NPD10714 and NPD1671 (Figure 3B). Western blot analysis of the co-precipitates also showed the binding specificity of GSTP to NPD926 (Figure 3C). These data indicated the involvement of GSTP in NPD926-induced ROS generation.
NPD926 is a substrate for GST isoenzymes
NPD926 is a substrate for GST isoenzymes
We have examined whether NPD926 modulated the enzymatic activity of GSTP by using an in vitro enzymatic assay with CDNB as the substrate. However, NPD926 did not affect the enzymatic activity of GSTP (results not shown).
GST isoenzymes are known to catalyse the conjugation of GSH to various electrophilic compounds , and NPD926 possesses an electrophilic α-chloroacetamide moiety. In addition, ChemProteoBase profiling analysis showed that the effects of NPD926 shared similarity with those of CDNB, a GST substrate (Figure 2A and Table 2). Therefore we have hypothesized that this compound may be a substrate for GSTs. To confirm this hypothesis, we have performed an in vitro binding reaction between NPD926 and GSH, followed by detection of the NPD926–SG conjugate and unreacted NPD926 by HPLC. Addition of NPD926 to GSH resulted in the spontaneous formation of the NPD926–SG conjugate, and the formation of this NPD926–SG conjugate was enhanced in the presence of recombinant GSTP (Figure 3D), indicating that NPD926 is a substrate for GSTP. According to the ESI–MS spectra, the mol-ecular mass of the NPD926–SG conjugate formed by the reaction was ~750 Da and this compound did not contain any chlorine atoms, indicated by the finding that no isotope peak characteristic of chlorine atoms was observed (results not shown), whereas the UV spectrum of this product (λmax 227 nm and 259 nm) was quite similar to that of NPD926 (λmax 228 nm and 262 nm; results not shown). These spectral data suggest that the chemical structure of the NPD926–SG conjugate formed by the reaction is that shown in Figure 3(E). GSTs are a superfamily of enzymes that consist of a number of isoenzymes. Therefore, to check the isoenzyme selectivity, we have examined whether GST isoenzymes classified into Mu-class and Alpha-class, the two major classes other than Pi-class, also catalysed the conjugation of NPD926 to GSH and confirmed that NPD926 is a substrate for not only GSTP but also GSTM2 and GSTA1 (Figure 3D). These data indicate that NPD926 is a substrate for GST isoenzymes, although NPD926 also spontaneously conjugates with GSH.
Furthermore, the NPD926–SG conjugate was also generated by Jurkat cells treated with NPD926 (Figure 3F), suggesting that NPD926 is a substrate for GST isoenzymes in cells as well as in vitro. In addition to the NPD926–SG conjugate, a related product was detected (NPD926–SG metabolite in Figure 3F). The UV spectrum of this product (λmax 231 nm and 257 nm) was similar to that of NPD926, and its ESI–MS spectrum suggested that the molecular mass is ~621 Da (results not shown). On the basis of these spectral data, this product is speculated to be a metabolite of the NPD926–SG conjugate produced by hydrolysis of γ-glutamyl bond in the NPD926–SG conjugate.
NPD926 induces cellular GSH depletion
Certain GST substrates, such as CDNB, are known to deplete cellular GSH by conjugation with GSH . Therefore we have examined the effect of NPD926 on the cellular GSH concentration and confirmed that NPD926 also lowered the cellular GSH concentration (Figure 4A). Consistent with NPD926-induced GSH depletion, GSH pre-treatment rescued cells from NPD926-induced ROS generation and subsequent cell death (Figures 4B and 4C). Conversely, inhibition of GSH synthesis with L-buthionine sulfoximine (BSO), an inhibitor of γ-glutamylcysteine synthetase, potentiated NPD926-induced cell death (Figure 4D). In addition, erastin and SSZ, another type of GSH synthesis inhibitor that inhibits the cystine/glutamate antiporter system xc−, also potentiated NPD926-induced cell death (Figures 4E and 4F). These data indicate that depletion of GSH, the primary antioxidant in cells, is the predominant mechanism by which NPD926 causes severe ROS accumulation and subsequent cell death.
NPD926 induces cell death via cellular GSH depletion
NPD926 induces cell death in oncogenic KRAS-transformed cells
As described in the Introduction, oncogenic KRAS-harbouring cells show increased ROS levels and are therefore highly sensitive to ROS-inducing compounds. Therefore we have examined the sensitivity of oncogenic KRAS-transformed mouse embryonic fibroblast NIH 3T3 cells (NIH 3T3/KRAS cells) and their parent NIH 3T3 cells to NPD926. Compared with NIH 3T3 cells, NIH 3T3/KRAS cells were highly sensitive to NPD926 (Figure 5A). Consistent with their sensitivity to NPD926, the cellular ROS levels were considerably higher both before and after NPD926 treatment in NIH 3T3/KRAS cells than in NIH 3T3 cells (Figure 5B). NPD926 potentiated the effect of erastin in Jurkat cells (Figure 4E), and erastin is known to be selectively lethal to tumour cells harbouring oncogenic KRAS. Therefore we have investigated whether the combination treatment of NPD926 and erastin is also effective against NIH 3T3/KRAS cells and confirmed that both compounds co-operatively induced cell death in NIH 3T3/KRAS cells, but not in NIH 3T3 cells (Figure 5C). These data suggest that NPD926 is also effective, as a single agent or in combination with erastin, against oncogenic KRAS-transformed cells with high levels of ROS accumulation.
Preferential induction of cell death by NPD926 in oncogenic KRAS-transformed cells
In the present study, we have identified NPD926 characterized by the induction of rapid cell death in cancer cells. Although elucidation of the mechanisms of action of bioactive compounds promotes improved understanding of cellular functions and may lead to the development of more effective treatments, this process is sometimes a bottleneck. Therefore we have analysed the mechanism of action underlying NPD926-induced cell death using a combination of two comprehensive and complementary approaches: proteomic profiling and affinity purification. ChemProteoBase profiling analysis showed a marked similarity between NPD926 and CDNB, a GST substrate (Figure 2A and Table 2). However, this prediction could not exclude the possibility that NPD926 was a GST inhibitor, an HSP90 inhibitor or a proteasome inhibitor (Table 2). Therefore we have performed affinity purification as a complementary approach. It is noteworthy that only one specific NPD926-binding protein identified as GSTP was detected in the affinity-purification experiment (Figure 3B), although NPD926 possesses the α-chloroacetamide moiety, which is highly reactive with a nucleophilic thiol group of the cysteine residues in proteins. These two separate approaches strongly suggested the involvement of GST isoenzymes, including GSTP, and thereby successfully enabled us to elucidate the mechanism of action underlying NPD926-induced cell death, that is via GST-catalysed conjugation with GSH, GSH depletion and subsequent ROS generation. In general, the signal transduction pathways induced by ROS depend on cell lines. In the case of Jurkat cells, the activation of p38 MAPK is involved partially in NPD926-induced cell death (Figure 2D), whereas that of caspase-3 is not essential (Figure 1D).
Human cytosolic GST exists in the form of a large number of isoenzymes that are divided into six classes, Alpha, Mu, Omega, Pi, Theta and Zeta, on the basis of amino acid sequence similarities, substrate specificity and immunological cross-reactivity, whereas each possesses a certain degree of structural and functional redundancies [37,38]. Although NPD926 is a substrate for GST isoenzymes classified into three major classes, Alpha, Mu and Pi (Figure 3D), only GSTP was identified as a detectable specific NPD926-binding protein in Jurkat cells (Figure 3B). One plausible interpretation is that the expression level of GSTP may be higher than for the other isoenzymes. Although the actual contribution and expression level of each GST isoenzyme in Jurkat cells have not been determined, we have found that GSTP is one of the abundant intracellular proteins in HeLa cells and that the expression level of GSTP in Jurkat cells is considerably higher than that in HeLa cells (results not shown). It was reported that GSTP is the predominant isoenzyme (up to 2.7% of the total cytosolic proteins) in all but two of the 60 tumour cell lines used in the Drug Screening Program of the National Cancer Institute (NCI) .
To date, several compounds, including CDNB and β-phenylethyl isothiocyanate (PEITC), have been identified as substrates for human GSTs that induce ROS generation by depletion of cellular GSH as well as NPD926 [17,33,39]. As the substrate selectivity of GST isoenzymes is relatively low [37,38], these structurally diverse compounds may serve as GST substrates and induce ROS generation using the same mechanism. CDNB and PEITC were reported to be conjugated with GSH by GST, followed by expulsion from cells (probably by a multi-drug-resistance-associated protein), and eventual extracellular degradation by γ-glutamyltranspeptidase and dipeptidase [40,41]. In the case of NPD926, a putative metabolite, possibly generated by hydrolysis of γ-glutamyl bond in the NPD926–SG conjugate, was detected after treatment of the cells with NPD926 (Figure 3F), whereas this product could not be detected in the recombinant GST-catalysed reactions between NPD926 and GSH in vitro (Figure 3D). On the basis of these findings and the previous reports about CDNB and PEITC [40,41], we have speculated that the NPD926–SG conjugate may also be excreted from cells and hydrolysed by γ-glutamyltranspeptidase.
Treatment of the cells with 3–30 μM NPD926 or CDNB resulted in depletion of cellular GSH (Figure 4A), which was present at a concentration in the millimolar range. An explanation for this stoichiometric discrepancy is that these compounds can be highly concentrated in the cells. A previous study demonstrated that treatment of HL-60 cells with 5 μM [14C]PEITC for 3 h resulted in accumulation of PEITC to a concentration of 0.5 nmol/106 cells . The diameter of HL-60 cells was reported to be 12.4 μm , which means that the volume of one HL-60 cell is approximately 1 pL. On the basis of these reports, the concentration of PEITC in a cell is 100-fold higher than that in the medium, and the final concentration of PEITC would be 0.5 mM in HL-60 cells treated with 5 μM PEITC. Similarly, another study reported that the treatment of HeLa cells with micromolar concentrations of piperlongumine, another GSH-depleting agent, also results in an effective concentration greatly in excess of that of cellular GSH, based on cell number and size .
NPD10714, an NPD926 analogue without a chlorine atom at the α position of the acetamide moiety, showed weak cytotoxicity (Figure 3A). Because NPD10714 cannot covalently bind to GSH, this observation suggests that there is a submechanism by which NPD926 exhibits cytotoxicity. However, despite underlying a sub-mechanism, our data indicate that GST-catalysed conjugation with GSH and subsequent depletion of GSH are the primary mechanisms by which NPD926 induces rapid cell death via ROS generation.
NPD926 selectively killed oncogenic KRAS-transformed fibroblast cells to a greater degree than their healthy control counterparts (Figure 5A). Among 22 human cancer cell lines shown in Table 1, MDA-MB-231 (G13D), A549 (G12S), HCT-116 (G13D), PANC-1 (G12D) and MIA PaCa-2 (G12C) are known to have oncogenic KRAS mutations. Although PANC-1 cells were relatively sensitive to NPD926, this compound did not show selective lethality in KRAS-mutated cancer cell lines over KRAS wild-type counterparts (Table 1). In our cancer cell line panel, Jurkat cells, which have a wild-type KRAS gene, are most sensitive to NPD926 (Table 1). Although it is not clear why Jurkat cells are comparatively sensitive to NPD926, we have speculated that cell lines growing in suspension are generally sensitive to various kinds of compounds [44,45]. These data suggest that there are multiple determinants of sensitivity to NPD926 other than KRAS mutation when analysing sensitivity across diverse genetic contexts. However, by comparison of the sensitivity of NIH 3T3/KRAS cells with that of NIH 3T3 cells (Figure 5A), we have demonstrated that KRAS mutation is one of the determinants of sensitivity to NPD926 in an individual genetic context. As well as NPD926, several ROS inducers, such as lanperisone and erastin, are known to kill oncogenic KRAS-harbouring cells selectively [12,14], despite their ROS- detoxification systems [46,47], showing the effectiveness of ROS inducers in the treatment of oncogenic KRAS-harbouring cancers. Furthermore, the cellular ROS levels are also up-regulated by other oncogenes, including HRAS, MYC and BCR-ABL , resulting in sensitization of cancer cells to ROS inducers [17–21]. It was reported that PEITC, piperlongumine and parthenolide selectively kill oncogenically transformed cells, cancer cells and CD34+ acute myelogenous leukaemia cells respectively [17,18,48]. Mechanistically, these three compounds induce ROS generation via depletion of cellular GSH as well as NPD926 [17,18,48]. As piperlongumine and parthenolide possess the electrophilic α,β-unsaturated carbonyl group, they are considered to bind covalently to a thiol group of GSH [43,48]. Thus NPD926, PEITC, piperlongumine and parthenolide may at least partially share a common mechanism of action, that is conjugation with and depletion of GSH. It is notable that not all ROS inducers are sufficient for inducing cancer cell death and that some ROS inducers containing electrophilic sites tend to show selective toxicity for transformed cells [48,49]. Taken together, we have speculated that conjugation with and depletion of GSH may be a common mechanism underlying the selective killing of cancer cells and oncogenically transformed cells by the electrophilic ROS inducers.
NPD926 significantly potentiated the cytotoxic effect of two distinct types of GSH synthesis inhibitors: an inhibitor of γ-glutamylcysteine synthetase (BSO; Figure 4D) and inhibitors of system xc− (erastin  and SSZ ; Figures 4E, 4F and 5C). However, the combination effect of NPD926 and the same type of GSH-depleting agent, piperlongumine, was relatively less (results not shown). These data indicate the effectiveness of the combination treatment of a GSH synthesis inhibitor and a GSH-depleting agent due to co-operative inhibition of GSH metabolism. GSH serves as the primary intracellular antioxidant in itself and by the action of glutathione peroxidases and contributes to the protection of cancer cells from high levels of ROS in the tumour microenvironment. In cancer stem-like cells that express a variant isoform of CD44, the xCT subunit of system xc− (also known as SLC7A11) is stabilized and protects cells from ROS generation by enhancing GSH synthesis, thereby promoting tumour growth . It is known that xCT/SLC7A11 expression correlates with increased glutathione level and poor prognosis in patients with bladder cancer . In triple-negative breast cancer cells with a poor prognosis, the inhibition of xCT/SLC7A11 also attenuates tumour growth . These studies indicate that system xc− is the Achilles heel of certain types of cancers, including cancer stem cells. Recently, the clinically approved anti-inflammatory agent SSZ has been tested in clinical trials for cancer therapy. As a sensitizing agent to system xc− inhibitors such as SSZ, NPD926 may contribute to the development of novel cancer therapies.
We have demonstrated that NPD926, a newly identified ROS inducer, kills cancer cells via GSH depletion. Whereas some classical anti-tumour agents such as doxorubicin also induce ROS generation, our findings show another possible strategy for cancer treatment, that is ROS induction by targeting GSH metabolism.
We thank Kaori Honda, Tomomi Sekine, Hisae Kondo, Yasuko Hirata, Mizue Yuuki and Harumi Aono for the technical support; Dr Tamio Saito for providing NPD926 and its analogues; Dr Tatsuhiko Sudo for providing the anti-p38 MAPK antibody; and Dr Toshihiko Nogawa, Dr Takeshi Shimizu, Dr Nobumoto Watanabe and Dr Konstanty Wierzba for helpful suggestions.
This work was supported by the RIKEN Special Postdoctoral Researcher Program (to T.K.) and the JSPS KAKENHI [grant number 25750397] (to T.K.).
Coomassie Brilliant Blue
2D-difference gel electrophoresis
GST Alpha 1
GST Mu 2
mitogen-activated protein kinase
reactive oxygen species
water-soluble tetrazolium salt 8
Hiroyuki Osada and Tatsuro Kawamura conceived and designed the study. Tatsuro Kawamura and Makoto Muroi acquired the data. Tatsuro Kawamura, Yasumitsu Kondoh, Makoto Muroi, Makoto Kawatani and Hiroyuki Osada analysed and interpreted the data and wrote and reviewed the paper.