2-Oxoadenosine induces cytotoxicity through intracellular accumulation of 2-oxo-ATP and depletion of ATP but not via the p38 MAPK pathway

2-Oxoadenosine (2-oxo-Ado), an oxidized form of adenosine, is cytotoxic and induces growth arrest and cell death, which has potential as an anti-cancer drug. However, it is not well understood how 2-oxo-Ado exerts its cytotoxicity. We examined the effects of 2-oxo-Ado on non-tumour cells, namely immortalized mouse embryonic fibroblast lines, and investigated mechanisms by which 2-oxo-Ado exerts its cytotoxicity. We found that cell death induced by 2-oxo-Ado is classical caspase-dependent apoptosis, and requires its sequential intracellular phosphorylation catalysed by adenosine kinase (ADK) and adenylate kinase 2, resulting in intracellular accumulation of 2-oxo-ATP accompanied by accumulation of 2-oxo-Ado in RNA and depletion of ATP. Moreover, we showed that overexpression of MTH1, an oxidized purine nucleoside triphosphatase, prevents 2-oxo-Ado-induced cytotoxicity accompanied by suppression of accumulation of both intracellular 2-oxo-ATP and 2-oxo-Ado in RNA and recovery of ATP levels. We also found that 2-oxo-Ado activates the p38 MAPK pathway. However, siRNAs against Mkk3 and Mkk6, or treatment with several p38 MAPK inhibitors, except SB203580, did not prevent the cytotoxicity. SB203580 prevented intracellular phosphorylation of 2-oxo-Ado to 2-oxo-AMP, and an in vitro ADK assay revealed that SB203580 directly inhibits ADK activity, suggesting that some of the effects of SB203580 may depend on ADK inhibition.

1,2-Dihydro-2-oxoadenosine (2-oxoadenosine; 2-oxo-Ado), an oxidized form of adenosine and also known as 2-hydroxyadenosine, isoguanosine or crotonoside, has been reported as a naturally generated nucleoside analogue in Croton tiglium 1 and Diaulula sandiegensis 2 . 2-Oxo-Ado has not been identified in mammals, but some biological activities have been shown in mammals, such as effects on muscle contraction 3 and cAMP accumulation 4 . Moreover, 2-oxo-Ado has an anti-tumour activity, namely toxicity that causes growth arrest and cell death of tumour-derived cell lines and suppression of tumorigenesis in a xenograft mouse model 5 . However, the mechanisms by which 2-oxo-Ado causes such cytotoxicity have not been elucidated.
It has been reported that 2-chloroadenosine (2-Cl-Ado) has an anti-tumour activity that induces apoptosis through its intracellular phosphorylation to 2-Cl-ATP, initiated by adenosine kinase (ADK) 6,7 . Moreover, it has been shown that 8-chloroadenosine (8-Cl-Ado), which is also phosphorylated to the triphosphate form (8-Cl-ATP), induces cell death and activates the p38 mitogen-activated protein kinase (MAPK) pathway 8,9 . Many studies have reported that growth arrest and cell death can be induced via the p38 MAPK pathway 10,11 .
It is thus reasonable to assume that 2-oxo-Ado also induces cell death through its intracellular phosphorylation to 2-oxo-ATP, and its activation of the p38 MAPK pathway. We have shown that 2-oxo-ATP is efficiently hydrolysed to 2-oxo-AMP and pyrophosphate by an oxidized purine nucleoside triphosphatase, MTH1 12 , thereby raising the issue of whether the cytotoxicity caused by 2-oxo-Ado can be supressed by MTH1. In the present study, we first confirmed that 2-oxo-Ado cytotoxicity against mouse embryonic fibroblast (MEF) lines is dependent on ADK and adenylate kinase 2 (AK2). We also found accumulation of 2-oxo-ATP with significant depletion of ATP in the cellular nucleotide pool. Over-expression of human MTH1 in an Mth1-knockout MEF line efficiently supressed 2-oxo-Ado-induced cytotoxicity with significant reduction of 2-oxo-ATP levels in the nucleotide pool. Moreover, we found that 2-oxo-Ado activates the p38 MAPK pathway through MKK3/MKK6 activation. However, such activation is dispensable for 2-oxo-Ado to exert its cytotoxicity.

2-Oxoadenosine induces cell death in immortalized mouse embryonic fibroblasts.
We used a spontaneously immortalized mouse cell line (T9) derived from MEFs to evaluate the effects of 2-oxo-Ado on non-tumour cells. First, we exposed T9 cells to various concentrations of 2-oxo-Ado and Ado (20-120 µM) for 24 h and then assessed their viability using a tetrazolium salt (WST-8) assay (Fig. 1a). 2-Oxo-Ado significantly decreased the viability of T9 cells to less than 10% of the control level, in a dose-dependent manner up to 80 µM, whereas Ado had little effect even at 120 µM.
Next, we performed trypan blue exclusion assays at various time points after incubation of T9 cells with or without 2-oxo-Ado (50 and 100 µM) or Ado (100 µM). The number of live and dead cells at each time point relative to the initial number of total cells is shown in Fig. 1b,c. The number of live cells doubled within 24 h with or without 100 µM Ado, indicating that 100 µM Ado had no toxic effect on T9 cells in this assay. Conversely, the number of live cells cultured in the presence of 50 µM 2-oxo-Ado did not increase within 24 h. Moreover, the number of live cells was significantly decreased to 37.5% of the initial number and the dead cell number was significantly increased after incubation for 24 h with 100 µM 2-oxo-Ado, but not 50 µM 2-oxo-Ado. These results indicate that 2-oxo-Ado induces growth arrest and cell death in a dose-dependent manner. Based on these results, we concluded that 2-oxo-Ado has toxic effects on non-tumour cells by inducing growth arrest and cell death.
We also examined the effects of 2-oxo-Ado on normal human fibroblasts and tumour-derived cell lines ( Supplementary Fig. S1). WI38 cells, normal diploid human fibroblasts derived from lung tissue, were resistant to 2-oxo-Ado up to 120 µM with a mild growth suppression at 80-120 µM. U2OS cells, a human osteosarcoma cell line, were also resistant to 2-oxo-Ado, but their growth was completely suppressed at 120 µM. In contrast, MOLT4 cells, a human acute lymphoblastic leukaemia cell line, were quite sensitive to 2-oxo-Ado, exhibiting apparent cell death at 80-120 µM. These results indicate that the cytotoxicity of 2-oxo-Ado is dependent on the cell type. Thus, we continued its characterization using MEFs in this study.

Induction of MEF cell death requires intracellular phosphorylation of 2-oxoadenosine by adenosine kinase.
To examine whether 2-oxo-Ado is required to be phosphorylated to 2-oxo-ATP by ADK to exert its effects, we exposed T9 cells to various concentrations of 2-oxo-Ado in the presence or absence of 0.1 µM 5-iodotubercidin (Itu), an inhibitor of ADK 15 , and then performed trypan blue exclusion assays (Fig. 2a). In the presence of 0.1 µM Itu, exposure to increasing concentrations of 2-oxo-Ado (40 to 120 µM) only slightly reduced the numbers of live cells, indicating that Itu efficiently suppressed the cytotoxicity of 2-oxo-Ado. As shown in Fig. 2b, WST-8 assays also revealed that Itu completely suppressed the severe reduction of cell viability caused by increasing concentrations of 2-oxo-Ado (20-120 µM), although Itu itself only slightly reduced cell viability and the vehicle had no effect on the cytotoxicity of 2-oxo-Ado.
We next evaluated the effects of small interfering RNA (siRNA)-mediated knockdown of Adk or adenylate kinase 2 (Ak2) on the cytotoxicity of 2-oxo-Ado. Ak2 encodes one of the adenylate kinase isozymes responsible for the reversible transfer of phosphate groups among adenine nucleotides 16 and possibly also on 2-oxoadenine nucleotides. T9 cells were pre-treated with siRNAs for 48 h ( Supplementary Fig. S2), exposed to various concentrations of 2-oxo-Ado for 24 h, and then subjected to WST-8 assays (Fig. 2c). siRNA against Adk significantly suppressed the cytotoxicity of 2-oxo-Ado compared with the effect of negative control (NC) siRNA, although the suppression was slightly less efficient than that achieved by Itu. siRNA against Ak2 also significantly suppressed the cytotoxicity of 2-oxo-Ado, but less efficiently than Adk siRNA, probably reflecting the presence of multiple isozymes such as AK1 or AK3.
Taken together, these data indicate that the cytotoxicity of 2-oxo-Ado requires intracellular phosphorylation of 2-oxo-Ado to 2-oxo-AMP, 2-oxo-ADP and 2-oxo-ATP. To confirm intracellular phosphorylation of 2-oxo-Ado, we extracted intracellular nucleotides from T9 cells after a 6 h of exposure to 100 µM 2-oxo-Ado and 0.1 µM Itu, and subjected them to quantitative high-performance liquid chromatography (HPLC) (Fig. 2d, Supplementary Relative values (% of control) of dead (e) and live (f) cell numbers to the total cell number in the absence of 2-oxo-Ado and Z-VAD (control) are shown as the mean ± SD of three experiments. (g) Protein levels in T9 cells were determined by western blot analysis after incubation with 100 µM 2-oxo-Ado for various times. Eight micrograms of total protein was applied to each lane. (h) Quantification of protein levels in (g). Results are normalised against total protein levels determined by Ponceau S staining. Relative values (fold change) of protein levels to those at time 0 are shown as the mean of two experiments. (a-c,e,f) Results were tested with two-way ANOVA and post hoc Tukey's HSD test. ns, not significant; ****p < 0.0001.   S3). In control T9 cells without exposure to 2-oxo-Ado, approximately 9 nmol ATP per 1 × 10 6 cells was detected, but no 2-oxo-Ado, 2-oxo-AMP, 2-oxo-ADP, 2-oxo-ATP was detected. In the HPLC condition, AMP was not detected, while ADP was merged with an unidentified peak. Therefore, we could not measure the amount of ADP. After exposure to 100 µM 2-oxo-Ado, approximately 70-90 pmol 2-oxo-Ado, 1.3 nmol 2-oxo-AMP and 8 nmol 2-oxo-ATP per 1 × 10 6 cells were detected, while ATP levels were decreased to 46% of the level detected in control T9 cells. The area of peaks containing ADP was not altered after exposure to 2-oxo-Ado, suggesting that the ADP level was unaffected by exposure to 2-oxo-Ado. After exposure to 2-oxo-Ado in the presence of 0.1 µM Itu, only ATP (approximately 8 nmol per 1 × 10 6 cells) was detected, and 2-oxo-AMP, 2-oxo-ADP and 2-oxo-ATP were not detected. Similar to the ADK inhibitor, siRNA-mediated knockdown of Adk or Ak2 inhibited both intracellular accumulation of 2-oxo-ATP and the reduction of ATP (Fig. 2e). Neither inhibition of ADK nor knockdown of Adk or Ak2 altered intracellular concentrations of 2-oxo-Ado ( Supplementary Fig. S3). These results confirmed that 2-oxo-Ado was indeed phosphorylated to 2-oxo-ATP in T9 cells, which was dependent on both ADK and AK2.
Because the cytotoxicity of 2-oxo-Ado was partly dependent on AK2 without detectable accumulation of 2-oxo-ADP, intracellular 2-oxo-ATP is most likely to be responsible for the cytotoxicity of 2-oxo-Ado. 2-Oxo-ATP is known to be efficiently hydrolysed by MTH1 12 . Therefore, we assumed that increased levels of MTH1 may decrease intracellular levels of 2-oxo-ATP, thus suppressing 2-oxo-Ado cytotoxicity. To examine this Values (fold change) of indicated nucleotides relative to ATP in T5v cells in the absence of 2-oxo-Ado are shown as the mean ± SD of three experiments. The concentration of ATP in T5v cells in the absence of 2-oxo-Ado was 11.6 nmol/1 × 10 6 cells and that in T5MTH1 cells was 14.9 nmol/1 × 10 6 cells. Results were statistically analysed by two-way ANOVA for ATP, 2-oxo-AMP, 2-oxo-ATP, separately. (c) 2-Oxo-Ado residues relative to 1 × 10 6 guanosine residues in RNA of T5v and T5MTH1 cells was determined by LC-MS/ MS. RNAs were collected after incubation with the indicated concentrations of 2-oxo-Ado for various times. Results are shown as the mean ± SD of three experiments. Results were statistically analysed by three-way ANOVA. possibility, we used two cell lines, T5v and T5MTH1. Both are derived from a MEF line (T5) established from an Mth1-knockout mouse embryo 17 . T5v cells are deficient for MTH1 and carry an empty vector, while T5MTH1 cells express high levels of human MTH1 (hMTH1) protein (630 ng/mg total protein) 17 . As shown in Fig. 3a, T5v cells exhibited essentially the same sensitivity to increased concentrations of 2-oxo-Ado as T9 cells, probably because the endogenous level of mouse MTH1 in T9 cells is very low 17 . Conversely, T5MTH1 cells exhibited a significant increase in resistance to 2-oxo-Ado cytotoxicity.
To confirm the intracellular level of 2-oxo-ATP, we extracted intracellular nucleotides from T5v and T5MTH1 cells after exposure to 100 µM 2-oxo-Ado for 6 h, and then performed quantitative HPLC analysis (Fig. 3b). In T5v cells exposed to 2-oxo-Ado, 16.1 nmol 2-oxo-ATP per 1 × 10 6 cells was detected, which was 1.38-fold higher than the level of ATP in control T5v cells without exposure to 2-oxo-Ado. Furthermore, T5v cells exposed to 2-oxo-Ado contained approximately 20% of the ATP seen in control T5v cells. However, in T5MTH1 cells exposed to 2-oxo-Ado, 3.0 nmol 2-oxo-ATP and 9.6 nmol ATP per 1 × 10 6 cells were detected with a 2-fold increase in the 2-oxo-AMP level compared with T5v cells exposed to 2-oxo-Ado.
We next isolated RNA from T5v and T5MTH1 cells after exposure to 50 or 100 µM 2-oxo-Ado for 6-12 h, and determined the levels of 2-oxo-Ado in RNA by liquid chromatography-tandem mass spectrometry (LC-MS/ MS) analysis (Fig. 3c). In T5v cells, 2-oxo-Ado levels in RNA were increased after exposure to 2-oxo-Ado in time-and dose-dependent manners. About 1200 residues of 2-oxo-Ado per 1 × 10 6 guanosine residues in RNA were detected in T5v cells after 12 h of exposure to 100 µM 2-oxo-Ado, while only 114 residues of 2-oxo-Ado per 1 × 10 6 guanosine residues in RNA (about 10% of that in T5v cells) were detected in T5MTH1 cells after the exposure. It is noteworthy that basal levels of 2-oxo-Ado in untreated T5v and T5MTH1 cells were under the limit of detection (one residue of 2-oxo-Ado per 1 × 10 7 guanosine residues in RNA).
We thus concluded that the cytotoxicity of 2-oxo-Ado is dependent on intracellular formation of 2-oxo-ATP mediated by ADK and AK2, which results in remarkable accumulation of 2-oxo-Ado in RNA and depletion of ATP. In addition, hMTH1 prevents accumulation of 2-oxo-Ado in RNA as well as accumulation of 2-oxo-ATP and depletion of ATP, and thus suppresses the cytotoxicity.
2-Oxoadenosine activates the p38 MAPK pathway which is dispensable for its cytotoxicity. We next examined whether 2-oxo-Ado activates p38 MAPKs (α, β, γ, and δ) prior to cell death. We  (Fig. 4). Increased phosphorylation of these proteins was observed as early as 3 h after exposure, reaching maximum levels at 9-12 h after exposure, indicating that 2-oxo-Ado activates the p38 MAPK signalling pathway.
To examine whether p38 MAPK activation is essential for 2-oxo-Ado-induced cell death, we applied four different types of p38 MAPK inhibitors, SB203580, SB202190, VX745 and BIRB796, to T9 cells with or without exposure to 2-oxo-Ado. As shown in Supplementary Fig. S4, while the increased levels of p38 MAPK phosphorylation after exposure to 2-oxo-Ado for 9 h were not altered in the presence of any inhibitor, phosphorylation levels of MAPKAPK-2 were markedly decreased in the presence of all four inhibitors to below the basal levels observed without exposure to 2-oxo-Ado. These results confirmed that all four inhibitors efficiently inhibited p38 MAPK activity, but not its phosphorylation, thus preventing activation of p38 MAPK downstream signalling in T9 cells exposed to 2-oxo-Ado. However, among these inhibitors, only SB203580 significantly suppressed the cytotoxicity of 2-oxo-Ado, as revealed by trypan blue exclusion and WST-8 assays (Fig. 5a,b, Supplementary  Fig. S5). VX745 and BIRB796 had no effect on the viability of T9 cells exposed to increasing concentrations of 2-oxo-Ado (20-120 µM). While SB202190 slightly improved cell viability even without exposure to 2-oxo-Ado, there was no significant interaction between the two factors in two-way analysis of variance (ANOVA), indicating that SB202190 did not suppress the cytotoxicity of 2-oxo-Ado (Supplementary Fig. S5).
To clarify whether p38 MAPK activation is essential for 2-oxo-Ado-induced cell death, we next treated T9 cells with siRNAs against Mkk3 and Mkk6, with or without exposure to 2-oxo-Ado (Supplementary Fig. S6). siRNA against Mkk3, which encodes a major p38 MAPKK in T9 cells, increased expression of Mkk6 mRNA 22 . Thus, we simultaneously treated T9 cells with two siRNAs against Mkk3 and Mkk6, which achieved efficient These results clearly showed that exposure of T9 cells to 2-oxo-Ado activates the p38 MAPK signalling pathway. However, such activation is not necessary for T9 cells to undergo 2-oxo-Ado-induced cytotoxicity. It was also noteworthy that SB203580 suppressed 2-oxo-Ado-induced cytotoxicity and that this suppression must be mediated by an unknown function of this compound and not from inhibition of p38 MAPK activity.

SB203580 inhibits adenosine kinase activity and suppresses 2-oxo-Ado-cytotoxicity. Suppr
ession of 2-oxo-Ado-cytotoxicity by SB203580 independently of p38 MAPK inhibition may provide a clue to understand the mechanism by which 2-oxo-Ado exerts its cytotoxicity. Because 2-oxo-Ado must be converted to 2-oxo-ATP or cause depletion of ATP to induce cell death, we first evaluated whether SB203580 alters the levels of these nucleotides in T9 cells exposed to 100 µM 2-oxo-Ado for 6 h (Fig. 6a). When cells were incubated for 6 h in the presence of 100 µM 2-oxo-Ado with or without SB203580 (10 and 30 µM), SB203580 significantly suppressed both the reduction of ATP levels and the accumulation of 2-oxo-AMP and 2-oxo-ATP in a dose-dependent manner. While VX745 also significantly suppressed accumulation of 2-oxo-ATP compared with T9 cells exposed to 100 µM 2-oxo-Ado in the absence of the p38 MAPK inhibitor, the level of 2-oxo-ATP was significantly higher than that in T9 cells exposed to 2-oxo-Ado in the presence of SB203580 (10 and 30 µM). The reduction of ATP levels and the accumulation of 2-oxo-AMP in T9 cells induced by 2-oxo-Ado was not altered by VX745. Under the same conditions, intracellular 2-oxo-Ado levels were significantly increased in the presence of SB203580 in a dose-dependent manner, but not in the presence of VX745 (Fig. 6b). These results suggest that SB203580 inhibits conversion of 2-oxo-Ado to 2-oxo-AMP, which is catalysed by ADK.
These results suggest that SB203580 directly inhibits or indirectly alters ADK activity. To examine the direct effect of SB203580 on ADK activity, we performed in vitro enzyme assays with recombinant human ADK (rhADK) (Fig. 7). When 100 µM of 2-oxo-Ado was incubated in the presence of rhADK and ATP, 2-oxo-AMP-formation was significantly inhibited by SB203580 (IC 50 = 8.6 µM) and to a lesser extent by BIRB796 (IC 50 ≥ 100 µM) in a dose-dependent manner, but VX745 had little inhibitory effect on ADK up to 100 µM. When  Supplementary Fig. S8, essentially the same results were obtained with other adenosine analogues, AICAR and 2-chloroadenosine (2-Cl-Ado), as those obtained with 2-oxo-Ado. We thus concluded that SB203580 directly inhibits ADK activity.
2-Cl-Ado is also known to induce cell death that is dependent on ADK and 2-Cl-ATP formation 6 . Therefore, we examined whether SB203580 suppresses its cytotoxicity in T5v and T5MTH1 cells (Supplementary Fig. S9). T5v and T5MTH1 cells exhibited essentially the same sensitivity to increased concentrations of 2-Cl-Ado (20-120 µM), indicating that MTH1 may not hydrolyse 2-Cl-ATP. Itu at 0.1 µM efficiently suppressed 2-Cl-Ado cytotoxicity, confirming that ADK is required for toxicity in MEFs, while 10 µM SB203580, but not VX745, also significantly suppressed cytotoxicity although less efficiently than Itu.
We thus demonstrated that 2-oxo-Ado causes growth arrest and apoptosis dependent on ADK, AK2 and the formation of 2-oxo-ATP (triphosphate form) similarly to various modified adenosines, such as 8-Cl-Ado and 2-Cl-Ado, which are potent anti-cancer drugs.

Discussion
In the present study, we showed that exposure of immortalized MEFs to 2-oxo-Ado induces growth arrest and classical caspase-dependent apoptosis, accompanied by accumulation of intracellular 2-oxo-ATP, resulting in accumulation of 2-oxo-Ado in cellular RNA, and depletion of ATP, both of which are dependent on ADK and AK2. Overexpression of human MTH1, an oxidized purine nucleoside triphosphatase, efficiently suppressed 2-oxo-Ado-induced cytotoxicity with significantly reduced accumulation of both intracellular 2-oxo-ATP and 2-oxo-Ado in cellular RNA, as well as recovery of ATP levels. Furthermore, 2-oxo-Ado activated the p38 MAPK pathway dependent on MKK3/MKK6. However, such activation was not necessary for 2-oxo-Ado to exert its cytotoxicity.
Exposure of T9 cells to 2-oxo-Ado caused intracellular accumulation of 2-oxo-ATP to a level equivalent to the normal level of intracellular ATP and halved the ATP level as early as 6 h after exposure. Both of these effects were almost completely prevented by Itu, an ADK inhibitor (Fig. 2). The cytotoxicity induced by 2-oxo-Ado was almost completely suppressed by Itu, and partially by knockdown of Adk or Ak2. These results indicate that 2-oxo-Ado-induced cytotoxicity is mainly dependent on intracellular phosphorylation of 2-oxo-Ado but not on extracellular adenosine receptors, which are known to mediate various effects of adenosine analogues 24,25 . However, phosphorylation of 2-oxo-AMP and 2-oxo-ADP by adenylate kinase is reversible depending on their concentration. Thus we could not explain why the 2-oxo-ATP level was so highly increased without accumulation of 2-oxo-ADP. Considering that the intracellular level of accumulated 2-oxo-ATP was equivalent to the normal level of intracellular ATP, 2-oxo-ATP may be actively generated by the same mechanism as ATP, namely oxidative phosphorylation in mitochondria, and it has been reported that 8-Cl-ADP is phosphorylated to 8-Cl-ATP by mitochondrial ATP synthase 26 . Therefore, AK2 may supply 2-oxo-ADP as a substrate for 2-oxo-ATP synthesis, because AK2 among the adenylate kinase isozymes, is thought to supply ADP as a substrate in ATP synthesis in the mitochondrial intermembrane space [27][28][29] . The contribution of AK2 to both 2-oxo-Ado-induced cytotoxicity and intracellular accumulation of 2-oxo-ATP, as shown in Fig. 2c,e, supports this interpretation.
Significant switching of intracellular ATP for 2-oxo-ATP is considered to be responsible for 2-oxo-Ado-induced cytotoxicity. However, the mechanism of this process is still unclear, and it is essential to examine whether 2-oxo-ATP itself is cytotoxic. If 2-oxo-ATP itself is not cytotoxic, the depletion of ATP resulting from the process of generating 2-oxo-ATP is likely to be responsible for the cytotoxicity. If this is the case, it also indicates that 2-oxo-ATP cannot substitute for ATP. While a large amount of 2-oxo-ATP is generated through ATP-dependent reactions, such as phosphorylation by ADK and AK2, an equivalent amount of ATP must be consumed, which may be one of the main causes for the depletion of intracellular ATP levels. However, the reduction of intracellular ATP levels after exposure to 2-Cl-Ado is caused by not only consumption of ATP itself but also deamination of AMP 6 , and 8-Cl-ATP may inhibit mitochondrial ATP synthase via binding to the α tp β tp site of the enzyme 26 . It has been reported that 2-oxo-ATP reduces the amount of RNA synthesized in vitro and increases base substitutions 30 , suggesting that 2-oxo-ATP decreases transcription levels with increased production of abnormal proteins in vivo. In the present study, we demonstrated that exposure of T5v cells to 2-oxo-Ado significantly increased the 2-oxo-Ado level in cellular RNA, which was efficiently suppressed by overexpression of hMTH1 (Fig. 2e). Taken together, it is still possible that 2-oxo-ATP itself exerts some cytotoxicity, including disruption of RNA synthesis, mRNA translation, and energy supply.
To evaluate the influence of 2-oxo-ATP itself, we analysed the effects of hMTH1 on changes to intracellular nucleotides caused by 2-oxo-Ado, because hMTH1 hydrolyses 2-oxo-ATP, but does not prevent phosphorylation of 2-oxo-Ado to 2-oxo-ATP. We found that overexpression of hMTH1 in T5MTH1 cells significantly increased intracellular ATP levels and significantly reduced intracellular levels of 2-oxo-ATP as well as 2-oxo-Ado levels in cellular RNA (Fig. 3b,c). This result indicates that 2-oxo-ATP itself contributes to the reduction of intracellular ATP levels. However, there is the possibility that hMTH1 may increase intracellular ATP levels even if 2-oxo-ATP does not contribute to ATP depletion, because in the presence of pyrophosphates, which are produced during hydrolysis of 2-oxo-ATP by hMTH1, ATP is regenerated from ADP-ribose by NUDIX5 31 .
Recently, MTH1 inhibitors have been developed as anti-tumour reagents, because cancer cells express increased levels of MTH1 to survive the increased oxidative conditions that are prevalent in cancer 32,33 . However, knockdown of MTH1 or inhibition of MTH1 in cancer cells is not sufficient to introduce cancer cell death 34 . The toxicity of 2-oxo-Ado is not specific to tumour cells as shown in the present study. However, mice can tolerate intraperitoneal administration of 96 mg 2-oxo-Ado per kg per day 5 . We also showed that normal human fibroblasts are resistant to 2-oxo-Ado up to 120 µM, and that a tumour-derived cell line, MOLT4, is highly sensitive to 2-oxo-Ado resulting in cell death at doses of 80-120 µM, while U2OS cells exhibited only growth arrest, suggesting increased expression of MTH1. Here, we propose that a combination of 2-oxo-Ado and an MTH1 inhibitor may have synergistic anti-tumour effects. For such a combination therapy, a cancer-directed drug delivery system would be very important to suppress side effects.
Another possible mechanism of 2-oxo-Ado-induced cytotoxicity is via the p38 MAPK pathway. It has been reported that the p38 MAPK pathway is involved in cell death caused by 8-Cl-cAMP or 8-Cl-Ado 9 . In the present study, we showed that 2-oxo-Ado activates the p38 MAPK pathway through activation of MKK3/MKK6. However, we found that the p38 pathway is dispensable for 2-oxo-Ado-induced cytotoxicity which was demonstrated using inhibitors or siRNAs targeting components of p38 MAPK signalling cascades. Intriguingly, we found that the cytotoxicity caused by 2-oxo-Ado was efficiently suppressed by SB203580, which has been used as a specific inhibitor for p38 MAPKs through activation of MKK3/MKK6, suggesting that SB203580 may exert suppression of cytotoxicity by inhibiting another enzyme. Indeed, we demonstrated that SB203580 inhibits ADK activity in vitro using recombinant human ADK protein, while other p38 MAPK inhibitors had little effect on ADK.
SB203580 can effectively, but not fully, inhibit p38 MAPKs at concentrations lower than 1 µM and is often used at 10 to 30 µM to achieve efficient suppression 21 . SB203580 also influences cellular functions through mechanisms other than inhibition of p38 MAPK at 10 µM 35,36 . Taken together, these data and our present results suggest that there may be erroneous conclusions in publications in which SB203580 was employed to examine the involvement of p38 MAPKs. In particular, results using high concentrations of SB203580 or in vivo experiments in which serum concentrations of SB203580 were not measured should be interpreted carefully. Our finding that ADK is a novel target of SB203580 is helpful for the re-interpretation of the results obtained using SB203580, especially when observations are not consistent with results following suppression of p38 MAPK inhibition.

Methods
Cells. All cells used in this study are listed in the Supplementary Methods.

Methods.
Trypan blue exclusion assay. A total of 3 × 10 5 T9 cells in 2 ml of culture medium were added to each well of 6-well plates and incubated for 12-16 h. When Z-VAD was used, cells were incubated for 1 h with Z-VAD or the vehicle before 2-oxo-Ado treatment. The medium was then exchanged with 1 ml of fresh medium without antibiotics, and the cells were treated with the indicated reagents (nucleosides and inhibitors) for various times (0-24 h). When only live cells were counted, cells were washed with phosphate buffered saline (PBS) and then incubated with 200 µl PBS containing 0.0625% trypsin/0.005% EDTA (0.25 × trypsin/EDTA solution) for 5 min. Cell suspensions were mixed with 200 µl of fresh medium. When both live and dead cells were counted, cells harvested using the 0.25 × trypsin/EDTA solution were combined with the removed medium and PBS wash, which SCIeNtIfIC RepoRts | 7: 6528 | DOI:10.1038/s41598-017-06636-8 may contain floating dead cells, and then centrifuged. Cell pellets were resuspended in 200 µl of fresh medium. Then, 10 µl of the cell suspension was mixed with 10 µl of trypan blue solution (Sigma-Aldrich Japan). Live cells, which were not stained with trypan blue, and dead cells, which were stained with trypan blue, were counted by an automated cell-counting machine (TC20; BIO-RAD, Hercules, CA, USA).

WST-8 assay.
A total of 4 × 10 3 cells in 100 µl of culture medium were placed in each well of 96-well plates and incubated for 12-16 h. If needed, cells were treated with siRNA (see Materials and Methods: Gene knockdown) for 32-36 h and then replated as described above and incubated for 12-16 h for a total time of 48 h. The medium was changed to fresh culture medium without antibiotics, and cells were treated with the indicated reagents (nucleosides and inhibitors) for 24 h. Fifty microliters of 0.2 × WST-8 reagent (Cell Counting Kit-8; Dojindo, Kumamoto, Japan) diluted in medium was added to each well, and cells were incubated for 2 h. The optical density of each well was measured at 450 nm (OD450) by a microplate reader (infinite M200 pro, TECAN, Männedorf, Switzerland). The background was determined by the OD450 of wells containing medium and reagents but without cells.
Gene knockdown. All siRNAs used in this study were purchased from Sigma-Aldrich Japan: negative control (Mission_SIC-001), Adk (SASI_Mm01_00192948), Ak2 (SASI_Mm01_00189879), Mkk3 (SASI_ Mm01_00081897) and Mkk6 (SASI_Mm01_00127382). A total of 8 × 10 3 T9 cells in 500 µl of culture medium without antibiotics were added to each well of 24-well plates and incubated for 12-16 h. The siRNA solution was prepared at a concentration of 300 nM or 150 nM of each siRNA when two different siRNAs were used simultaneously in Opti-MEM ® I Reduced Serum Medium (Thermo Fisher Scientific) containing 1% Lipofectamine ® RNAiMAX Reagent (Thermo Fisher Scientific), and incubated for 10 min at room temperature. One hundred microliters of the siRNA solution was added to each well (containing 500 µl of medium) of the 24-well plates. The final concentration of siRNAs was 50 nM.

RNA extraction and real-time quantitative reverse transcription PCR (qRT-PCR). Procedures for
RNA extraction and qRT-PCR are described in the Supplementary Methods.

Protein extraction.
A total of 2 × 10 5 T9 cells in 2 ml of culture medium were added to each well of 6-well plates and incubated for 12-16 h. If needed, cells were treated with siRNA (see Materials and Methods: Gene knockdown) for 32-36 h, replated as described above, and incubated for 12-16 h for a total time of 48 h. The medium was then exchanged with 1 ml of fresh medium without antibiotics, and cells were treated with reagents (nucleosides and inhibitors) for various times. The cells were then washed twice with PBS and lysed in 2 × SDS sample buffer (0.125 M Tris-HCl pH 6.8, 4% SDS, 10% glycerol, 2% 2-mercaptoethanol, and 0.005% bromophenol blue) containing 1% phosphatase inhibitor cocktail (Nacalai Tesque, Kyoto, Japan). Lysates were mixed, heated at 95 °C for 5 min, sonicated and then heated again at 95 °C for 5 min. Protein concentrations were determined using an XL-Bradford kit (APRO Science, Tokushima, Japan).
Western blotting. Western blotting was performed as described in the Supplementary Methods. Intracellular nucleotide/nucleoside extraction. A total of 1.4 × 10 6 cells in 10 ml of culture medium were added to a 10-cm dish and incubated for 12-16 h. The medium was then exchanged with 5 ml of fresh medium without antibiotics, with or without 100 µM 2-oxo-Ado and the indicated inhibitors. The cells were incubated for 6 h, washed twice with ice-cold PBS, and lysed in 250 µl an ice-cold 10% trichloroacetic acid solution (Wako Pure Chemical Industries Ltd, Osaka, Japan) containing 0.2 mM 2′-deoxycoformycin (Santa Cruz Biotechnology, Dallas, TX, USA). Lysates were diluted to about 500 µl with ice-cold PBS, supplemented with 5 µl of 1 M 2,2,6, 6-tetramethylpiperidine 1-oxyl (TEMPO), and placed on ice. Before and after the preparation of cell lysates, the numbers of cells in additional control dishes that received no treatment (two dishes for each) were counted. Lysates were centrifuged at 17,800 × g for 10 min at 4 °C. Three hundred and sixty microliters of each supernatant were mixed with 40 µl of 3 M K 2 HPO 4 for neutralization and then stored at −80 °C. The volume of the remaining supernatants was measured to calculate the total volume. The concentration of nucleotides and nucleosides in stored solutions was determined by HPLC, and total amounts in original lysates were calculated.
HPLC. HPLC analysis was performed on an Alliance 2690 HPLC system (Waters, Milford, MA, USA) equipped with a 996 photodiode array (PDA) detector (Waters). A negative ion exchange column (TSKgel DEAE-2SW, Tosoh, Tokyo, Japan, Cat. no. 0007168) was used for separation of nucleotides, and a octadecylsilyl (ODS) column (Handy-ODS, Wako Pure Chemical Industries Ltd, Cat. no. 234-50013) was used for separation of nucleosides. The mobile phases are described in Supplementary Information. The amount of each nucleoside or nucleotide was calculated from the peak area under the curve (AUC) of the PDA absorbance at the wavelength of its maximum absorption: 292 nm for 2-oxo-Ado and its phosphate forms; 259 nm for Ado and its phosphate forms; 263 nm for 2-Cl-Ado and its phosphate forms; 268 nm for AICAR and its phosphate forms. Standard curves for ATP, 2-oxo-ATP and 2-oxo-Ado were used to determine their absolute amounts. 2-Oxo-ATP was partially dephosphorylated by alkaline phosphatase (Roche Applied Science, Penzberg, Germany), and the resulting four HPLC peaks corresponding to 2-oxo-Ado, 2-oxo-AMP, 2-oxo-ADP and 2-oxo-ATP were used to identify peaks of 2-oxo-AMP and 2-oxo-ADP. Quantities of 2-oxo-AMP and 2-oxo-ADP were calculated based on the maximum absorption coefficient of 2-oxo-ATP because these three nucleotides have the same base moiety and are considered to have the same absorption coefficient. Chromatograms of the reaction products of ADK with 2-Cl-Ado or AICAR showed novel peaks with the same absorption spectrums as those of their parent molecule, which were considered to correspond to each monophosphate form.