Peroxiredoxin II Restrains DNA Damage-induced Death in Cancer Cells by Positively Regulating JNK-dependent DNA Repair*

The 2-Cys peroxiredoxins (Prx) belong to a family of antioxidant enzymes that detoxify reactive oxygen and nitrogen species and are distributed throughout the intracellular and extracellular compartments. However, the presence and role of 2-Cys Prxs in the nucleus have not been studied. This study demonstrates that the PrxII located in the nucleus protects cancer cells from DNA damage-induced cell death. Although the two cytosolic 2-Cys Prxs, PrxI and PrxII, were found in the nucleus, only PrxII knockdown selectively and markedly increased cell death in the cancer cells treated with DNA-damaging agents. The increased death was completely reverted by the nuclearly targeted expression of PrxII in an activity-independent manner. Furthermore, the antioxidant butylated hydroxyanisole did not influence the etoposide-induced cell death. Mechanistically, the knockdown of Prx II expression impaired the DNA repair process by reducing the activation of the JNK/c-Jun pathway. These results suggest that PrxII is likely to be attributed to a tumor survival factor positively regulating JNK-dependent DNA repair with its inhibition possibly sensitizing cancer cells to chemotherapeutic agents.

Reactive oxygen species (ROS) 4 are produced by the partial reduction of oxygen as a by-product of metabolism and by exogenous insults, including anti-cancer drugs, ultraviolet light, and ␥-irradiation. High levels of endogenous ROS are cytotoxic because they can damage proteins, lipids, and DNA (1). Therefore, they have long been regarded as a possible cause of aging and many chronic diseases like cancer, arterio-sclerosis, and neurological disorders (2,3). In contrast, at relatively low levels, ROS are thought to be the second messenger that plays a key role in signaling events that induce proliferation, differentiation, and apoptosis (4,5). Accordingly, the cellular ROS should be tightly maintained at noncytotoxic levels and controlled in a regulated manner.
The mammalian cells have a variety of antioxidant systems, which maintain the cellular redox homeostasis and further protect cellular molecules against oxidative damage (6). Although glutathione (GSH) acts as a general redox buffer, several antioxidant enzymes function as local bystanders. Specifically, Cu,Zn-superoxide dismutase and Mn-superoxide dismutase reduce the superoxide anion (O 2 . ) to hydrogen peroxide (H 2 O 2 ) in cytosol and mitochondria, respectively. Catalase reduces H 2 O 2 to water in peroxisomes. The five isoforms of glutathione peroxidase reduce the peroxides, such as H 2 O 2 and lipid peroxide, to the corresponding alcohol in the presence of a redox recycling system (GSH, GSH reductase, and NADPH) in various compartments, i.e. extracellular, cytosol, and mitochondria (7). The six Prx enzymes reduce the peroxide, such as H 2 O 2 , lipid peroxides, and peroxynitrite, to the corresponding alcohol in the presence of another redox recycling system (thioredoxin, thioredoxin reductase, and NA-DPH) (8,9). The cellular distribution of Prxs is much more complex than that of other antioxidant enzymes. PrxI and PrxII are found mainly in the cytosol; however, PrxII has been shown to associate with the plasma membrane in the erythrocyte. Furthermore, PrxIII is a mitochondrial peroxide reductase. PrxIV is a secretory peroxidase. PrxV exhibits diverse distribution in the cytosol, peroxisome, and mitochondria. PrxVI, also called 1-Cys Prx, is a cytosolic protein. However, the presence and function of a nuclear antioxidant enzyme have not been determined. The 2-Cys Prx isoforms of Prx family enzymes, which are widely distributed in the tissues, have been shown to be overexpressed in various types of cancer cells and tumor tissues (10). Because the cancer cells were known to produce large amounts of ROS (11), it is readily appreciated that the overexpression of the antioxidant enzyme Prx could benefit cancer cell survival. For example, PrxI has been shown to suppress JNK activation and in turn cell death in the irradiated lung cancer cells (12). The reduction in the PrxII expression was shown to increase the radiation sensitivity of a head and neck cancer cell line (13). Mitochondrial PrxIII has been shown to sensitize cervical cancer cells to TNF-␣-induced cell death (14). However, it has become controversial in other studies, wherein Prx seems to suppress tumor progression. For example, PrxI was shown to interact with c-Myc via Myc Box II domain and inhibit c-Myc-induced cellular transformation (15). PrxI was also found to suppress Ras-or ErbB2-induced transformation by protecting phosphatase and tensin homolog from oxidative inactivation (16). Moreover, PrxI Ϫ/Ϫ mice developed age-dependent malignant cancers (17,18). In the case of PrxII, it is noteworthy that there have been no reports showing that PrxII interacts with proteins related to the tumorigenesis. The PrxII-deficient mice exhibited only splenomegaly phenotype due to the major antioxidant role of PrxII in erythrocyte but no cancer-related phenotype (19 -21). It is important to determine whether Prx is involved in tumorigenesis or cancer cell survival.
In this study, we found that the two isoforms of 2-Cys Prxs, PrxI and PrxII, are present in the nucleus and demonstrated that the nuclear PrxII prevents the cancer cell death induced by DNA-damaging agents, including topoisomerase inhibitors. More importantly, such protective function of PrxII is only effective in the cancer cells and independent of its peroxidatic activity. Our study also revealed that PrxII is required for the JNK-dependent DNA repair process in the nucleus. Thus, PrxII has turned out to be the first antioxidant enzyme involved in the DNA repair process via regulating JNK activation.
Immunocytochemistry-HeLa cells were cultured on glass cover slides for 1 day, washed twice with cold PBS, and fixed with pre-warmed 4% paraformaldehyde solution for 30 min. Cells were washed two times with PBS and permeabilized with 0.2% Triton X-100 for 15 min at room temperature. After permeabilization, the cells were then blocked with 2% BSA in PBS (blocking buffer) for 1 h and incubated at 4°C overnight with the indicated primary antibodies diluted in blocking buffer: anti-p65 (1:300), anti-PrxI (1:300), anti-PrxII (1: 300), anti-pJNK (1:100), and anti-␥H2AX (1:300). The cells were washed three times with blocking buffer and incubated with Alexa Fluor 568-conjugated secondary antibody. After 30 min, the coverslips were washed three times in a blocking buffer and mounted. The fluorescence images were taken by an LSM510 META confocal laser-scanning microscope (Zeiss). For ␥-H2AX staining, the number of foci was counted using Phoretix 2D evolution (Nonlinear Dynamics Ltd.).
Cell Death Assay-The stimulated cells were washed once with PBS, detached by incubating at 37°C for 2 min in 0.05% trypsin/EDTA, and then gently collected into the 5-ml FACS tubes, where the culture media and PBS wash were collected. The cells were then centrifuged for 3 min and once washed with cold PBS, and the final cell pellets were fixed with 70% ethanol in PBS and stained with propidium iodide (PI) for 5 min on ice. Cell death was assessed by FACS analysis of PIstained cells in a FACSCalibur system (BD Biosciences). Cell death was determined by the percentage of PI-positive cells. The cell death assays were performed in duplicate.
Measurement of Intracellular ROS-Intracellular ROS generation was assessed with an oxidation-sensitive fluorescent dye 5,6-chloromethyl-2Ј,7Ј-dichlorodihydrofluorescein diacetate (Molecular Probes) as described previously (23). The HeLa cells (3 ϫ 10 5 ) were cultured on 35-mm dishes and transfected with siRNAs for 24 h. The cells were then deprived of serum for 6 h and stimulated with 20 M etoposide in phenol red-free media for the indicated periods of time. After stimulation, the cells were quickly rinsed with Krebs-Ringer solution and incubated for 5 min with 5 M 5,6-chloromethyl-2Ј,7Ј-dichlorodihydrofluorescein diacetate. The 2Ј,7Ј-dichlorodihydrofluorescein fluorescence was collected for 10 s with an inverted Axiovert200 fluorescence microscope (Zeiss). The relative 2Ј,7Ј-dichlorodihydrofluorescein fluorescence was obtained by averaging the fluorescence intensities of the 60 -80 cells after background subtraction in each image using ImageQuant TM software (GE Healthcare). The presented data are representative of three experiments with the same result. Note that the detached round cells were omitted from quantification.
Retrovirus Production-The coding sequences for human PrxII wild-type (WT) and activity-dead cysteine double mutant (C51S/C172S) were inserted by PCR cloning into pShooter vectors (Invitrogen), pCMV/myc/cyto© and pCMV/ myc/nuc© to construct the plasmids encoding the Myctagged PrxII proteins targeted to cytosol and nucleus, respectively. The human PrxII sequence with a Myc tag and/or nuclear localization sequence was then subcloned into pQCXIX (Invitrogen) to produce the retroviral vectors, designated as pQ-PrxII-WT, pQ-PrxIIWT-Nuc, and pQ-PrxIIDN-Nuc. To avoid the ectopic expression being knocked down by PrxII siRNA, the three bases were substituted within the sequence targeted by the PrxII siRNA used in the study. The base substitution was carried out by site-directed mutagenesis using the following primers: sense, 5Ј-GCT GAC GTG ACC AGA CGC TTA TCC GAA GAT TAC GGC GTG CTG AAA ACA-3Ј; antisense, 5Ј-TGT TTT CAG CAC GCC GTA ATC TTC GGA TAA GCG TCT GGT CAC GTC AGC-3Ј, where the mutated bases are underlined. The final retroviral vector (20 g) along with the viral packaging vector set (gag-pol and VSVG-encoding plasmid, 20 g each) were transfected to 293T cells for 24 h by the calcium phosphate method. The cells were then washed twice in phosphate-buffered saline (PBS) and placed in fresh complete medium for virus collection. After 24 h, the culture supernatants containing the retroviruses were collected and filtered using a 0.450-m filter, and the aliquots were frozen at Ϫ70°C until used. For retroviral infection, the viral aliquots were thawed in warm water bath and mixed with 10 g/ml Polybrene.
Comet Assay-Neutral comet assays were performed using the Comet assay kit (4250-050-K, Trevigen) according to the manufacturer's protocol. Briefly, the cells were mixed with low melting agarose gel and spread on CometSlide. The cells were lysed by Lysis Solution (catalog no. 4250-050-01) on ice and then equilibrated with neutral electrophoresis buffer. The slides were subjected to electrophoresis and afterward immersed in DNA precipitation solution. DNA was stained with SYBR Green, and the fluorescence image was taken by fluorescence microscopy. The tail moments were measured and averaged from ϳ70 randomly selected cells using TriTek CometScore freeware program.
Statistics-Data were analyzed using Student's t test on SigmaPlot 8.0 software, and the p value was derived to assess statistical significance. A p Ͻ 0.05 was considered to be significant.

Prx I and II Are Located in the Nucleus-The six human
Prxs show broad subcellular distribution, including cytosol, mitochondria, and peroxisome (8). To examine whether any of the Prxs are localized in the nucleus, we used the HeLa cervical carcinoma cells, where the Prxs are highly expressed (24), and prepared the pure nuclear fraction using a hypotonic extraction buffer containing mild detergent (0.2% Nonidet P-40). The last nuclei pellet was enriched with a nuclear protein lamin B but without PrxIII and PrxIV proteins present in the mitochondria and endoplasmic reticulum, respectively ( Fig. 1A). This result indicates that several washes with detergent-containing extraction buffer completely eliminated the contamination of other organelles. Nonetheless, the two cytosolic 2-Cys Prxs, PrxI and PrxII, were apparently detected in the nuclei pellet (Fig. 1A). Their nuclear localization was further examined by immunostaining. First, the antisera against PrxI and PrxII were affinity-purified. Second, the specificities of the anti-PrxI and -PrxII antibodies were tested by blocking with the corresponding recombinant antigen proteins (data not shown). A significant amount of immunoreactive signal was detected in the nucleus, about 10% for PrxI and 5% for PrxII of total immunoreactive signals (Fig. 2, B and C). Given that both PrxI and PrxII do not have a featured nuclear localization signal, we hypothesized that their nuclear localization may be achieved by cytosol-nuclear shuttling. To test this, we blocked the shuttling with leptomycin B, an inhibitor of CRM-1-dependent nuclear export pathway (25). As a result, leptomycin B treatment induced the remarkable accumulation of PrxII in the nucleus, although there was less accumulation for PrxI (Fig. 2, B and C). NF-B p65, which is known to shuttle between cytosol and nucleus (26), was also accumulated by leptomycin B treatment. This result indicates that at least the nuclear localization of PrxII seems more dynamic than that of PrxI.
PrxII Protects Cancer Cells from DNA Damage-induced Cell Death-We studied the role of the nuclear PrxI and PrxII in the cell death induced by DNA-damaging agents, including topoisomerase inhibitors, a replication blocker, and a DNA intercalating agent. To do so, we knocked down expression of PrxI and PrxII using specific small interfering RNAs (siRNA). The siRNA transfection resulted in the reduction of PrxI and PrxII protein levels in HeLa cells by ϳ95% (Fig. 2A). Interestingly, the PrxII knockdown drastically enhanced the death of HeLa cancer cells in response to different DNA-damaging agents, such as etoposide, camptothecin, hydroxyurea, and doxorubicin (Fig. 2, B-E). In contrast, the PrxI knockdown had no influence on the cell death. Similarly, PrxII knockdown also enhanced cell death in HCT116 colon cancer cells and U2OS osteosarcoma cells treated with etoposide (Fig. 2,   F and G). Transfection of short hairpin RNA (shRNA) construct and other siRNA duplexes targeting PrxII similarly enhanced the etoposide-induced cell death in HeLa cells (supplemental Fig. S1, A and B), thus confirming the specific effect of PrxII knockdown. Throughout this study, we noticed that PrxII knockdown itself induces a certain amount of basal cell death (around 10% cell death after a 48-h PrxII knockdown) in the absence of DNA damage. The control experiment turned out that the extent of basal cell death is proportional to the duration of PrxII knockdown, suggesting that the PrxII knockdown somehow destabilized cancer cells. Albeit to a lesser extent, a shorter period (24-h transfection) of PrxII knockdown with apparently no basal cell death clearly enhanced the etoposide-induced cell death (supplemental Fig. S1C). Nonetheless, we kept 48 h of transfection of PrxII siRNA for experimental convenience. In addition, knockdown of neither a peroxisomal peroxidase catalase nor other cytosolic Prx isoform called PrxVI influenced the etoposide-induced cell death (supplemental Fig. S1D). These results indicate that PrxII specifically regulates DNA damage-induced cancer cell death. Next, to determine whether protective action of PrxII is applicable to normal cells responding to DNA damage, we used the primary fibroblast cells, such as MEFs and lung fibroblasts. Neither gene knock-out nor transient knockdown of PrxII did enhance cell death of primary fibroblasts in response to DNA damage (Fig. 2, H and I). These results indicate that PrxII is selectively involved in the survival of cancer cells against DNA damage stress.
Nuclearly Targeted PrxII Sufficiently Prevents DNA Damage-induced Death-To further confirm the specific action of PrxII in DNA damage-induced cancer cell death, we carried out a rescue experiment by exogenously expressing PrxII in PrxII siRNA-transfected cells. To avoid targeting of exogenously expressed PrxII by cognate siRNA, the three bases in the region corresponding to siRNA target sequence were substituted without causing amino acid change. Exogenous PrxII, which shows a size shift due to a Myc epitope tag and a nuclear localization signal sequence (NLS), was expressed in a level similar to the endogenous proteins (Fig. 3A), and the NLS-tagged version of PrxII was localized in the nucleus (Fig.  3B). Like endogenous PrxII, only some of the expressed wildtype PrxII (WT-Cyto) was detected in the nucleus. When these cells were treated with etoposide, add-back expression of nuclearly targeted PrxII (WT-NLS), as well as wild-type PrxII (WT-Cyto), rescued the HeLa cells from the augmented death by PrxII knockdown (Fig. 3C). Thus, the result indicates that the nuclear PrxII is sufficient for protecting cancer cells against DNA damage. Given that PrxII is a peroxidase enzyme reducing H 2 O 2 to water (23), we examined whether the peroxidase activity is required. Unexpectedly, both the wild-type and activity-dead mutant (C51S/C172S double mutant) rescued the HeLa cells from the augmented death by PrxII knockdown (Fig. 3D), suggesting that regardless of the peroxidase activity, the protective function of PrxII may rely on the protein-protein interaction.
DNA Damage-induced Cell Death Is Independent of ROS Generation-It has been controversial whether ROS production is involved in the DNA damage-induced cell death (27)(28)(29)(30). Therefore, we decided to determine whether the ROS production is increased by PrxII knockdown in etoposidetreated cells; if so, it is involved in the cell death. To determine the relationship between ROS and cell death, we first measured the intracellular ROS with an oxidation-sensitive fluorescent dye, 5,6-chloromethyl-2Ј,7Ј-dihydrochlorofluorescein diacetate. The intracellular ROS was barely detected in the etoposide-treated control HeLa cells (Fig. 4A). In contrast, the ROS level was markedly enhanced by PrxII knockdown, indicating that PrxII eliminates the intracellular ROS produced by etoposide treatment. In addition, the etoposide treatment did not induce the hyperoxidation of 2-Cys Prxs, a result of oxidative stress overwhelming cellular peroxidase capacity (31), although 100 M H 2 O 2 induced it (supplemental Fig. S2). This result suggests that the amount of ROS produced by etoposide is likely to be what PrxII can accommodate. We then utilized the BHA as a general antioxidant compound to examine whether the ROS increased by PrxII knockdown is involved in cell death. As expected, BHA prevented the increase of ROS level (Fig. 4B). However, the BHA pretreatment did not affect the etoposide-induced cell death both in the control and PrxII knockdown cells (Fig. 4C). Although ROS are produced by DNA damage, we concluded that they may not contribute to the cell death.

JNK-c-Jun Pathway Is Down-regulated by PrxII
Knockdown-Next, we studied the signaling mechanism underlying the protective role of PrxII in DNA damage-induced cell death. First, we tested whether the etoposide-induced cell death is caspase-dependent. A pan-caspase inhibitor, benzyloxycarbonyl-VAD-fluoromethyl ketone, did not block the etoposide-induced cell death both in the control and PrxII knockdown HeLa cells (Fig. 5A), suggesting that etoposide induces cell death in a caspase-independent manner. Second, we examined whether ATM, the master kinase that induces p53-dependent apoptosis in response to DNA double strand breaks (32,33), is involved. When the HeLa cells were pretreated with a specific ATM inhibitor KU55933, the ATM inhibition additively increased the etoposide-induced cell death in the PrxII knockdown cells (Fig. 5B). Furthermore, like parental HCT116 cells, the cell death of p53-deficient HCT116 cells induced by etoposide treatment was also augmented by PrxII knockdown compared with that of control cells (Fig. 5C). These results suggest that the PrxII-mediated protection against DNA damage-induced cell death may be independent of ATM and p53 function.
Because MAPKs are known to participate in cell cycle checkpoint, apoptosis, and DNA repair under cellular stresses such as ultraviolet and DNA damage (34,35), we explored the activation of MAPKs in etoposide-treated HeLa cancer cells. Interestingly, etoposide induced a persistent activation of all three MAPKs, among which JNK and ERK activations were apparently down-regulated by the PrxII knockdown (Fig. 6A). To dissect which kinase is involved in the cell death, we pretreated HeLa cells with ERK and JNK inhibitors. Clearly, the JNK inhibition enhanced the etoposide-induced cell death, whereas the ERK inhibition did not (Fig. 6B). Furthermore, the JNK inhibition synergistically augmented cell death in PrxII knockdown cells. Then, to ascertain how PrxII regulates JNK activation, we looked at the location of PrxII and JNK in the etoposide-treated cells by immunostaining. Interestingly, the etoposide treatment induced the nuclear accumulation of phosphorylated JNK for which an immunoreactive signal was completely blocked by a synthetic phospho-JNK antigenic peptide (Fig. 6C). As a control experiment, H 2 O 2 treatment also induced nuclear accumulation of phosphorylated JNK. Interestingly, the nuclear level of PrxII was increased by etoposide treatment in a timely manner ( Fig. 6D and supplemental Fig. S3). Moreover, the level of phosphorylated JNK in the nuclei of etoposide-treated cells was notably diminished by the PrxII knockdown (Fig. 6D), suggesting that PrxII may preserve the phosphorylated JNK in the nucleus. Collectively, these results indicate that JNK activation is essential for cell survival under DNA damage, and it is positively regulated by PrxII.
Next, we examined the phosphorylation and expression of c-Jun as a direct substrate of the activated JNK. The etoposide treatment in HeLa cancer cells markedly induced c-Jun expression and phosphorylation following the time course of JNK activation, which was reduced by PrxII knockdown (Fig.  6E). As expected from the result in Fig. 5B, the inhibition of JNK, not ERK, also reduced c-Jun expression in etoposidetreated control HeLa cells and more severely reduced it in the cell with PrxII knockdown (Fig. 6F). Furthermore, the PrxII knockdown also abolished the increase of c-Jun expression in other cancer cells, such as HCT116 and U2OS, that had ex-hibited the enhanced cell death by the PrxII knockdown in response to etoposide (Fig. 6G). These results collectively indicate that PrxII is an essential component of the JNK-dependent survival pathway in response to DNA damage.
PrxII Knockdown Results in a Defect in DNA Repair-Recent reports have shown that c-Jun and ATF play a protective role in DNA damage by inducing the genes involved in the DNA repair process (36,37). Therefore, we examined the influence of PrxII knockdown during DNA repair after etoposide treatment. The extent of DNA repair was estimated as the amount of DNA double strand breaks measured by a neutral Comet assay and ␥-H2AX focus formation assay. The DNA repair process was biphasic; therefore, the extent of damaged DNA was rapidly dropped by 40% within several hours after removal of etoposide and then slowly decreased close to the zero level (Fig. 7). As seen at zero time of recovery, the extent of DNA damage induced by etoposide was not affected by the PrxII knockdown. However, the DNA repair was retarded by the PrxII knockdown, as evident in a later phase (Fig. 7A). A similar effect was observed in the ␥-H2AX focus formation (Fig. 7B). These results suggest that downregulation of the JNK-c-Jun pathway by the PrxII knockdown affects completion of the DNA repair process. To support this, we performed a comet assay either in the presence or absence of the JNK inhibitor. The JNK inhibition resulted in the retarded DNA repair as much as was done by the PrxII knockdown (Fig. 7C). Similar results were obtained by the ␥-H2AX focus formation assay (Fig. 7D). Consequently, the dual blockade of JNK and PrxII was not an additive effect, indicating that they are in the same pathway.

DISCUSSION
Cancer has the highest mortality rate throughout the world. Thus, its eradication is the most crucial medical and scientific issue that must be resolved. Although radio-and chemotherapies are powerful treatments based on the unlimited hyperproliferative characteristics of cancer cells, the cytotoxic side effect to normal cells is debilitating to patients with cancer. To ensure a full recovery for a patient with cancer, alternative strategies have long been sought for combinatorial therapy. One promising example is anti-VEGF therapy that blocks the formation of angiogenic vasculature essential for tumor growth (38). However, recent studies have reported an adverse effect wherein the cancer cells that survived after short term anti-VEGF treatment became more aggressive and invasive (39,40). Undeniably, efforts must be continued to selectively and efficiently kill the cancerous cells.
In this study, we explored the role of nuclear PrxI and PrxII in DNA damage-induced cell death and made several important discoveries. 1) Nuclear PrxII specifically confers the resistance against DNA damage-induced death to cancer cells and not to normal cells. 2) DNA-damaging agents induce cancer cell death independently of ROS. 3) Nuclear PrxII positively regulates JNK activation in response to DNA damage in opposition to its negative role in TNF-␣-induced JNK activation (41). 4) PrxII-dependent regulation of JNK/c-Jun pathway is essential for completion, rather than initiation, of DNA repair. These findings implicate PrxII as a novel anti-cancer target that regulates DNA repair in cancer cells against DNAdamaging agents. The biological function of PrxI and PrxII related to tumorigenesis is somewhat complex. Knock-out mice studies have shown that the PrxII-deficient mice did not produce any phenotype related to tumorigenesis (21), whereas the deficiency of PrxI resulted in the development of malig-  MARCH 11, 2011 • VOLUME 286 • NUMBER 10 nant tumors (18) and increased DNA oxidation in various tissues of aged mice (17). Although the PrxI-deficient mice were produced by transposon or retroviral insertion, which could have been accompanied with additional mutations, the evidence suggests that PrxI can suppress the tumorigenesis. Some in vitro evidence provided mechanistic explanation for this tumor suppressor function such that PrxI inhibits c-Abl kinase and c-Myc by direct interaction (15,42). However, it is paradoxical that the expression of both PrxI and PrxII was increased in various tumor types (10). Moreover, decreasing PrxI or PrxII expression enhances radiosensitivity of cancer cells (43)(44)(45)(46). Collectively, this means that the function of cytosolic 2-Cys Prxs in tumorigenesis and tumor growth may differ. Our study provided new data that PrxII knockdown sensitizes cancer cells to cell death induced by chemotherapeutic agents through regulating DNA repair. Mechanistically, down-regulation of ERK and JNK activation by PrxII knockdown is a particularly salient result. Numerous studies have shown that both ERK1/2 and JNK are activated by DNA damage stresses. Etoposide-induced ERK activation was shown to mediate cell cycle arrest in an ATM-dependent but p53-independent manner (47). In the case of other stresses, such as hydroxyurea and ionizing radiation, ERK activation was required for S phase and G 2 /M checkpoint arrest, respectively (48,49). Moreover, Yan et al. reported that ionizing radiation-induced ERK1/2 activation required a direct interaction with BRCA1 for G 2 /M cell cycle checkpoint arrest (50). Thus, the evidence suggests that ERK activation by DNA damage may preferentially be involved in cell cycle arrest before cell death. However, the JNK activation pathway has a dual function of pro-and anti-apoptotic properties (51). In other words, the sustained JNK activation is predominantly associated with induction or enhancement of apoptosis, whereas transient JNK activation is involved in the cell survival response (52). In the DNA damage response, JNK activation was shown to promote drug resistance and cell survival by enhancing DNA repair (37). Furthermore, Hayakawa et al. (36) have also shown that the activated JNK leads to phosphorylation of downstream transcription factors, ATF2 and c-Jun, which induce the expression of components of DNA repair machinery. Our data shows that inhibition of JNK, not ERK, was linked to etoposide-induced cell death. Furthermore, the down-regulation of JNK activation by PrxII knockdown led to the reduced expression and phosphorylation of the c-Jun transcription factor, which in turn results in the defective repair of etoposide-induced double strand breaks. Although we have not analyzed the DNA repair machinery in etoposide-treated cells, our data clearly support that PrxII positively regulates JNK/c-Jun-dependent DNA repair in cancer cells.

Nuclear PrxII Modulating DNA Damage Response
The mechanism of how PrxII positively regulates JNK activation in DNA damage response remains questionable. Few studies have shown the interaction of Prx with the MAPKs. In lung cancer cells, PrxI interacted with a GSTpi-JNK complex and suppressed ionizing radiation-induced JNK activation and apoptosis (12). On the contrary, a yeast 2-Cys Prx called thioredoxin peroxidase 1 activated a homolog of mammalian p38/JNK MAPK, Sty1, via the peroxide-induced disulfide formation (53). We have failed to determine through our experiments the direct protein-protein interaction between PrxII and JNK. Nonetheless, because the nuclearly targeted PrxII rescued the cancer cells against etoposide-induced cell death and the level of phosphorylated JNK in the nucleus was reduced by the PrxII knockdown, it is most likely that PrxII contributes to maintain phosphorylated JNK in the nucleus.
In summary, we found that absence of nuclear PrxII sensitizes cancer cells to chemotherapeutic agents by impairing the JNK-dependent DNA repair. It was previously shown that a dominant negative mutant of PrxII also sensitizes cancer cells to TNF-␣-induced cell death (54). The former was ROS-independent, and the latter was ROS-dependent. Given that most cancer cells produce a large amount of ROS, the present find-ing leads us to propose that inhibiting PrxII can be a powerful means to kill the cancer cells by bursting death signaling in the cytoplasm and by impairing DNA repair in the nucleus.