Selenium Compounds Activate Early Barriers of Tumorigenesis*

Selenium chemoprevention by apoptosis has been well studied, but it is not clear whether selenium can activate early barriers of tumorigenesis, namely senescence and DNA damage response. To test this hypothesis, we treated normal and cancerous cells with a gradient concentration of sodium selenite, methylseleninic acid and methylselenocysteine for 48 h, followed by a recovery of 1–7 days. Here we show that selenium compounds at doses of ≤LD50 can induce cellular senescence, as evidenced by the expression of senescence-associated β-galactosidase and 5-bromo-2-deoxyuridine incorporation, in normal but not cancerous cells. In response to clastogens, the ataxia telangiectasia mutated (ATM) protein is rapidly activated, which in turn initiates a cascade of DNA damage response. We found that the ATM pathway is activated by the selenium compounds, and the kinase activity is required for the selenium-induced senescence response. Pretreatment of the MRC-5 non-cancerous cells with the antioxidant N-acetylcysteine or 2,2,6,6-tetramethylpiperidine-1-oxyl suppresses the selenium-induced ATM activation and senescence. Taken together, the results suggest a novel role of selenium in the activation of early tumorigenesis barriers specific in non-cancerous cells, whereby selenium induces an ATM-dependent senescence response that depends on reactive oxygen species.

Genome instability is a hallmark of carcinogenesis. Recent advances suggest that major barriers of human tumorigenesis at the early stage include DNA damage response and senescence (1)(2)(3), both of which involve ATM activation. Heritable mutations in ATM cause ataxia telangiectasia, a genomic instability syndrome characterized by cancer predisposition, neurodegeneration, and premature aging. In response to DNA damage, the ATM kinase is rapidly activated and mediates multiple downstream pathways, resulting in DNA damage checkpoint response and repair. ATM pathway activation in humans requires ATM phosphorylation at Ser-1981 (4). On the other hand, cellular senescence, a form of cell cycle withdrawal, can limit the proliferation of cells with persistent genomic instabil-ity. In the MRC-5 diploid fibroblasts expressing mos, the oncogene-induced senescence can be suppressed by inhibition of ATM, suggesting that ATM plays a major role linking pathways of senescence and DNA damage response during early tumorigenesis (2).
Selenium is an essential trace mineral widely distributed in inorganic forms in soil and in organic forms in certain foods. The Nutritional Prevention of Cancer Trial conducted in the United States concluded that daily selenium intake at a supranutritional level significantly decreases risks of cancer, and the prevention is most successful for prostate, lung, and colon cancers (5,6). Furthermore, a role of selenium in preventing cancer in patients with prostatic intraepithelial neoplasia has been inferred (7,8). However, the recent Selenium and Vitamin E Cancer Prevention Trial concluded that selenium supplementation alone or in combination with vitamin E does not prevent prostate cancer risks in the cohort of relatively healthy men (9). Differences in selenium formulation and the body selenium status prior to entering the trials may explain the seeming discrepancies between these two cohorts of clinical studies (5,6,9,10). Nonetheless, a consensus drawn from the two trials is that selenium supplementation prevents cancer risks only in men entering the trial with suboptimal levels of body selenium. Thus, it is necessary to elucidate the mechanism of tumorigenesis suppression offered by selenium.
Metabolites of selenium compounds have been shown to induce reactive oxygen species (ROS), 2 which in turn can induce oxidative modifications and breaks on DNA. Previous studies have focused on selenium-induced stress responses in various cultured cancer cells, from which it is suggested that much of the role of selenium in cancer prevention is attributable to ROS-induced apoptosis or cell cycle arrest in cancer cells (11)(12)(13). Consistent with this notion, it has been shown that selenium-induced apoptosis in cancer cells can be suppressed by antioxidants (14) and is p53-dependent (15). Furthermore, selenium can sensitize cancer cells to other apoptotic inducers, including TRAIL and doxorubicin (11,16).
Available evidence from the literature has not provided a full understanding of selenium in tumorigenesis or the linkage of selenium metabolites to genomic maintenance (17). It is known that selenium-induced oxidative stress or DNA damage can * This work was supported in part by an award from the General Research lead to apoptosis in some cancer cells (18); however, the roles of selenium in cellular senescence have not been studied. To investigate additional barriers of tumorigenesis elicited by selenium, we treated the MRC-5 normal lung fibroblasts, the CRL-1790 normal colon fibroblasts, and the PC-3 prostate and HCT 116 colon cancer cells with selenium compounds of organic and inorganic origins in a series of studies. We employed a number of senescence markers, including the measurement of senescence-associated ␤-galactosidase (SA-␤-galactosidase) expression and levels of the pulse-labeled 5-bromo-2-deoxyuridine (BrdUrd), together with investigating the ATM-mediated DNA damage response pathway, to explore a role of selenium in the activation of early tumorigenesis barriers. Our results indicate that selenium induces an ATM-dependent senescence response via redox regulation in non-cancerous but not in cancerous cells, suggesting a novel mechanism of selenium in counteracting tumorigenesis.
Senescence Assays-We detected SA-␤-galactosidase by using a senescence detection kit (MBL Co. Ltd., Woburn, MA) according to the manufacturer's instructions. Briefly, cells were seeded onto 24-well plates with a density of 5,000 cells/well and treated with the selenium compounds for 48 h, followed by a 7-day recovery. The cells were washed once in PBS, fixed at room temperature for 15 min, washed three times in PBS, and stained in the staining solution containing X-gal at a concentration of 1 mg/ml for 8 h. Cells were then overlaid with 70% glycerol and observed under a light microscope.
We performed a BrdUrd incorporation assay to indicate status of DNA replication. Cells were cultured on coverslips and incubated with Na 2 SeO 3 (1 M), MSeA (1 M) and MSeC (50 M) for 48 h followed by a 7-day recovery. Cells were pulselabeled with BrdUrd (10 M) for 1 h, followed by washing three times in PBS and fixation in 4% paraformaldehyde (in PBS) for 30 min. The fixed cells were then permeabilized in HCl (0.1 N) containing pepsin (100 g/ml) for 30 min at 37°C. DNA was denatured by HCl (1.5 N) for 15 min and then by sodium borate (0.1 M) for 5 min. After washing three times in PBS, the cells were incubated in an anti-BrdUrd antibody conjugated with fluorescein isothiocyanate for 1 h according to the manufacturer's instructions (BD Pharmingen). A drop of ProLong gold antifade reagent with 4Ј,6-diamidino-2-phenylindole (DAPI) (Invitrogen) was added to a slide, and then coverslips were mounted on the slides. We employed filter cube set 49 (excitation, 365 nm; filter, 395 nm; emission, 445 nm) and set 38 (excitation, 470 nm; filter, 495 nm; emission, 525 nm) for visualization of DAPI and fluorescein isothiocyanate, respectively, through a Zeiss AxioObserver 100 fluorescence microscope (Carl Zeiss, Oberkochen, Germany). All of the photomicrographs were taken using the same magnification scale (ϫ200) and exposure time (400 ms) within 10 min to avoid autofading of the fluorescence signal.
Immunofluorescence-Cells were seeded onto coverslips and incubated with Na . After washing, the cells were incubated with secondary antibodies conjugated with fluorescence dyes (Alexa 488 goat anti-rabbit lgG and Alexa 594 goat anti-mouse lgG, Invitrogen) for 1.5 h at room temperature. A drop of Pro-Long Gold antifade reagent containing DAPI (Invitrogen) was added to a slide, and the coverslips were mounted onto the slide. The slides were then placed under a Zeiss AxioObserver 100 fluorescence microscope (Carl Zeiss) for image acquisition. The fluorescence signals were visualized by using filter cube set 49 for DAPI (excitation 365 nm, filter 395 nm, and emission 445 nm), filter cube set 38 for GFP (excitation 470 nm, filter 495 nm, and emission 525 nm), and filter cube set 43 for DsRed (excitation 550 nm, filter 570 nm, and emission 605 nm). All of the photos were taken using the same magnification scale (ϫ630) and exposure time (100 ms for DAPI, 400 ms for GFP, and 800 ms for DsRed), followed by deconvolution according to the manufacturer's instruction. The ATM phosphorylated at Ser-1981 (pATM Ser-1981) and ␥H2AX focus-positive cells are defined as those containing at least five foci (19). Five pictures were randomly taken from each slide. All experiments were run in duplication and performed a minimum of three times.
Flow Cytometry Assay-Cells were cultured in 10-cm 2 dishes and treated with 1-2 M Na 2 SeO 3 , 1-2 M MSeA, and 50 -100 M MSeC for 48 h, followed by a 1-7-day recovery. The cell monolayers were rinsed with PBS, incubated in trypsin/EDTA, resuspended in ice-cold PBS, and fixed and stored in 70% ethanol at Ϫ20°C until analysis. Prior to the analysis, cells were stained in propidium iodide (20 g/ml) containing RNase. The DNA was then analyzed by a FACScalibur cytometer with the CELLQuest program (BD Biosciences). ModFit LT (version 3.0, Verity Software House, Topsham, ME) was applied for cell cycle analysis on overlaid histograms.
Cell Survival Assay-Cells were seeded at a concentration of 10,000 MRC-5 cells, 5,000 PC-3 cells, and 5,000 HCT 116 cells per 3.5-mm well, incubated with the selenium compounds at the aforementioned dosages for 48 h, followed by a 7-day recovery. On day 7, the cells were washed once with PBS and then trypsinized. Cells were transferred to the INCYTO C-Chip disposable hemacytometers (SKC Inc., Eighty Four, PA), and the cell number was counted under a light microscope.
Statistical Analysis-The data were analyzed by using the SAS version 9.0 software (SAS Institute Inc., Cary, NC). A twotailed Student's t test was applied to determine statistical signifi-cance between the treatments and the control. The linear regression was also computed to confirm the senescence response to gradient concentrations of selenium in Fig. 1 (p Ͻ 0.0001).

Senescence Is Induced by Selenium Compounds in MRC-5 and CRL-1790 Non-cancerous Cells but Not in PC-3 or HCT
116 Cancerous Cells-To determine whether or not selenium can counteract tumorigenesis through cancer barriers other than the well studied apoptosis, we first assessed senescence phenotypes after cellular exposure to the selenium compounds. To confirm the selenium-induced senescence, we measured the rate of BrdUrd incorporation, an indicator of DNA replication. The selenium-treated cells were pulse-labeled with BrdUrd, which can be incorporated into the newly synthesized DNA during S phase of the cell cycle. Compared with the MRC-5 cells without selenium treatment, cellular exposure to Na 2 SeO 3 , MSeA, and MSeC resulted in a 4-, 6.6-, and 2.3-fold reduction in BrdUrd incorporation, respectively ( Fig. 1D and supplemental Fig. 5). Thus, treatment of MRC-5 cells with the selenium compounds resulted in DNA replication suppression, a feature of cellular senescence. Taken together, selenium compounds induce cellular senescence in the noncancerous but not in the cancerous cells. We next performed survival assays to determine the cellular sensitivity to the selenium compounds and estimate their respective LD 50 values. Results from the cell proliferation analysis showed that PC-3 and HCT 116 cancerous cells are more resistant than MRC-5 cells to treatment with Na 2 SeO 3 ( Fig. 2A), MSeA (Fig. 2B), and MSeC (Fig. 2C) at day 7. When the two cancerous cell lines were treated with doses of selenium equivalent to their respective LD 50 range, there was no SA-␤galactosidase detected (Fig. 1, A-C, and supplemental Figs. 3 and 4). In contrast, treatment of MRC-5 cells  with selenium at doses comparable with their LD 50 resulted in significant SA-␤-galactosidase expression. Moreover, there was no increase in sub-G 1 cell population 3 days after cellular exposure to the selenium compounds (supplemental Table 1), suggesting that an apoptotic cell death pathway is not activated. The results further support the observation that the seleniuminduced senescence occurs specifically in non-cancerous cells; cancerous cells are deficient in selenium-induced senescence. H2AX phosphorylation at Ser-139 (known as ␥H2AX) is a marker of DNA breaks and is known as a substrate of several kinases, including the phosphatidylinositol 3-kinase family members ATM and the catalytic subunit of DNA-dependent protein kinase (DNA-PK cs ) (18,21,22). Thus, we determined whether the selenium compounds can induce ␥H2AX formation. Analysis of the immunofluorescent results indicated that treatment of MRC-5 cells with Na 2 SeO 3 (1-2 M; Fig. 3D), MSeA (1-2 M; Fig. 3E), and MSeC (50 -100 M; Fig. 3F) resulted in significant increases in the population of cells expressing ␥H2AX foci, the extent of which is comparable between days 3 and 7 post-treatment. To determine whether ATM kinase activity is involved in the increased ␥H2AX focus formation, we co-treated the MRC-5 cells with the selenium compounds and KU55933 (20). Immunofluorescent analyses of ␥H2AX foci showed that KU55933 marginally suppresses ␥H2AX focus formation in the selenium-treated MRC-5 cells (Fig. 3, H-J, and supplemental Fig. 7), suggesting that ATM is not the major kinase attributable to the selenium-induced ␥H2AX focus formation. We next co-treated the MRC-5 cells with the selenium compounds and NU 7026, a kinase inhibitor of DNA-PK cs . Compared with KU55933, treatment of the cells with NU 7026 resulted in a more robust suppression in the selenium-induced ␥H2AX focus formation (Fig. 3, H-J). Interestingly, the effect of KU55933 and NU 7026 on the inhibition of selenium-induced ␥H2AX focus formation appears to be additive. Thus, ATM is not the major kinase involved in the selenium-induced ␥H2AX focus formation in MRC-5 cells.

Selenium Treatment Results in ATM Ser-1981 Phosphorylation and ␥H2AX Formation in Non-cancerous Cells-ATM
Cancer cells are characterized by genomic instability and increased oxidative stress (23)(24)(25)(26). Thus, we assessed pATM Ser-1981 and ␥H2AX expression in PC-3 cells, which is negative in selenium-induced senescence (Fig. 1, A-C, and  supplemental Fig. 3). Compared with the non-cancerous MRC-5 cells, there were significantly greater PC-3 cell populations exhibiting intrinsic pATM Ser-1981 (58.4 versus 4.7%; Fig.   3, A versus K) and ␥H2AX (85.0 versus 33.3%; Fig. 3, D versus L) foci in the absence of selenium treatment. Although 7 days after recovery from treatment with Na 2 SeO 3 (1 M), MSeA (1 M), and MSeC (50 M), PC-3 cells showed an increase in cells containing pATM Ser-1981 foci, the extent of induction is much smaller (43% versus 14.6-fold) as compared with the non-cancerous MRC-5 cells. Noticeably, the PC-3 cells exhibit high levels of intrinsic ␥H2AX foci, and treatment of the cells with selenium did not further increase ␥H2AX expression (Fig.  3L). Thus, the PC-3 prostate cancer cells are predisposed to genomic instability, which may prevent the cells from responding to the selenium treatment for the activation of senescence and the ATM tumorsuppressing pathways.
A previous report indicates that senescent MRC-5 cells arrest in the G 1 phase of the cell cycle 7-10 days after cellular exposure to H 2 O 2 at a concentration of 500 mM (27). Thus, we measured MRC-5 cell cycle profiles 1, 3, and 7 days post-treatment of the selenium compounds. We found that 1 and 3 days after recovery from treatment with MSeA and MSeC resulted in significant increases in MRC-5 cells in the S and G 2 /M population, followed by cell cycle arrest in the G 1 phase at day 7 in cells treated with Na 2 SeO 3 , MSeA, or MSeC (supplemental Tables 2-4). The G 1 cell cycle arrest in MRC-5 cells after selenium treatment is consistent with the observation of the selenium-induced senescence phenotype.
Selenium-induced Senescence Requires ATM Kinase Activity-To determine whether ATM kinase activity is required for the selenium-induced senescence, we preincubated the MRC-5 cells with KU55933. Results from the SA-␤-galactosidase analysis demonstrated that inhibition of ATM kinase activity prevented senescence induction in MRC-5 cells exposed to Na 2 SeO 3 (1 M), MSeA (1 M), or MSeC (50 M) (Fig. 4). The ATM kinase inhibitor also prevented senescence induction in MRC-5 cells exposed to lower doses of the selenium compounds (data not shown). The results indicated that ATM kinase activity is required for the induction of senescence by the selenium compounds in MRC-5 cells.
Regulation of the Selenium-induced DNA Damage Response and Senescence by Oxidative Stress-It has been shown previously that MSeA and Na 2 SeO 3 can induce ROS formation in a number of prostate cancer cells (28). Thus, we determined the involvement of ROS in the selenium-induced DNA damage response and senescence. Treatment of MRC-5 cells with antioxidants, NAC (a H 2 O 2 scavenger) and Tempo (a superoxide dismutase mimic), significantly suppressed senescence in MRC-5 cells treated with the selenium compounds (Fig. 4). Interestingly, Tempo is more potent than NAC in the attenuation of senescence induced by Na 2 SeO 3 and MSeA. ROS contribute to the selenium-induced DNA damage response because the pATM Ser-1981 focus formation (Fig. 5, A-C) and ␥H2AX formation (Fig. 5, D-F) are attenuated by NAC treatment. Thus, selenium induces senescence and pATM Ser-1981 focus formation in MRC-5 cells in a manner dependent on reactive oxygen species.

DISCUSSION
The evidence for selenium being a chemoprevention agent includes that from geographic, animal, and epidemiological studies (5,9,10,29). However, recent human clinical trials reported mixed results on the role of selenium in chemoprevention, which may potentially be explained by the differences in body selenium status of the subjects entering the trials as well as the selenium dose and formulation between the studies (5,9,(31)(32)(33)(34). Whatever the reason, the molecular mechanism by which selenium mitigates tumorigenesis is largely unknown. An increasing body of recent evidence has suggested that selenium and its metabolites induce apoptotic pathways in cancer cells (15,18,36). Nonetheless, selenium could in principle exert its chemoprevention function at the onset of tumorigenesis by other mechanisms. In line with this notion, recent reports indicate that DNA damage response and senescence are barriers of carcinogenesis that function at the interface between the precancerous and the cancerous stages (1-3). The ATM kinase plays a pivotal role in the DNA damage-induced senescence (37, 38). Our findings suggest a new role of selenium in tumorigenesis; selenium induces an ATM-dependent senescence response in a manner dependent on ROS in non-cancerous but not in cancerous cells.
How do sublethal (ՅLD 50 ) doses of selenium compounds mount the ATM-dependent senescence response? One possibility is that selenium compounds induce ROS generation that impacts on genome stability. Ample evidence indicates that metabolites of selenium compounds at high doses can induce ROS formation and apoptosis in cancer cells (13,15,16,28). Our findings identify an ATM-dependent senescence response induced by doses of selenium well below LD 50 in non-cancerous cells, whereby this pathway is suppressed by antioxidant administration. Early experiments showed that doxorubicininduced ROS formation or H 2 O 2 treatment can activate the ATM pathway (39,40) and that astrocytes and hematopoietic stem cells isolated from Atm Ϫ/Ϫ mice exhibit increased oxidative stress, early onset of senescence, and/or suppressed self-renewal capacity (41,42). Although ATM is a prominent responder to DNA double strand breaks, this kinase can be activated by various forms of chromosome alterations (4,20,43). Importantly, our results suggest that the kinase activity of ATM mediates the senescence phenotype in the seleniuminduced MRC-5 cells. Because inhibition of ATM kinase attenuates but does not prevent selenium-induced ␥H2AX formation (Figs. 3 and 5), other kinases are capable of phosphorylating H2AX. Altogether, we propose that seleniuminduced oxidative stress activates the ATM pathway for the subsequent senescent response. Future studies are needed to elucidate the mechanism of ATM kinase activation by selenium-induced oxidative stress.
Of note, ROS is unlikely to be the only direct activator of the senescence and ATM pathways, as we do not observe a G 1 cell cycle arrest as robust as reported previously in senescent MRC-5 cells treated with H 2 O 2 (27). Rather, treatment of the cells with MSeA and MSeC at a dose of ϳLD 50 resulted in a minor S and G 2 /M arrest prior to senescence induction. It is conceivable that the ROS-induced DNA oxidation and the subsequent formation of DNA breaks in S phase may activate the ATM pathway for a checkpoint response at the early time point. ATM is known to function in the DNA damage checkpoint at G 1 , S, and G 2 /M and is activated by stalled or collapsed replication forks (20,40,43). We propose that, following ATM activation (day 1), the cells accumulate in the G 1 phase (day 3) and eventually senesce (day 7).
There are several forms of senescence. Replicative senescence is mainly caused by telomere attrition, whereas premature senescence can be triggered by damaged DNA and oxidative stress. ATM is involved in both forms of senescence (37, 38,44,45). The p53 protein is mutated in the majority of human malignant tumors and is required for induction of senescence by DNA replication stress (topoisomerase inhibitors) in an array of cancer cell lines (46). Interestingly, we found that the selenium-induced senescence is missing in both the p53-proficient HCT 116 and the p53-deficient PC-3 cancer cells, suggesting that the lack of selenium-induced senescence in the two cancerous cell lines is not attributable to p53 status. Another candidate target of selenium action on senescence is p21, which is a gatekeeper of the G 1 -S transition and is implicated in a ROS-dependent senescence response in normal human fibroblasts (47,48). In HCT 116 cells exposed to doxorubicin treatment, p21 expression is induced in both a p53-dependent and -independent manner (49). Further studies are therefore necessary to fully understand whether and how p53 and p21 regulate the ATM-and ROS-dependent senescence response after selenium exposure.
DNA replication stress and checkpoints are associated with oncogene-induced senescence (1-3). Genotoxic stress can induce persistent DNA damage, thus triggering a senescenceassociated secretory phenotype (SASP) and suppressing p53 in normal cells (30). SASP can change the tissue microenvironment in a p53-independent fashion, and cells with SASP can secrete various factors, such as interleukin-6 (35). ATM is required for interleukin-6 secretion that facilitates cell communications and bypasses senescence in damaged cells (35). Our MRC-5 cells developed a senescence phenotype 7 days posttreatment of selenium, suggesting that persistent DNA damage and oxidative stress exist. Of note, our data indicate that the selenium-induced senescence is not associated with a robust S-phase checkpoint, as opposed to conditions such as significant DNA double strand breaks that induce S-phase checkpoint. We propose that SASP may develop and lead to senes-cence by changing its cellular microenvironment after the selenium exposure.
In conclusion, we have provided the first evidence that selenium can mitigate tumorigenesis by mechanisms other than the well studied apoptotic pathway. The observation that selenium specifically induces senescence response in non-cancerous cells suggests a cost-effective scenario by which tumorigenesis can be stifled at the very beginning in individuals who consume selenium with a cancer prevention perspective. It is of future interest to elucidate the mechanism by which selenium activates an ATM and ROS-dependent senescence response, especially at the interphase between the precancerous and cancerous stages by using models for initiation, promotion, and progression in carcinogenesis.