Activated oncogenes induce premature cellular senescence, a permanent state of proliferative arrest in primary rodent and human fibroblasts. Recent studies suggest that generation of reactive oxygen species (ROS) is involved in oncogenic Ras-induced premature senescence. However, the signaling mechanism controlling this oxidant-mediated irreversible growth arrest is not fully understood. Here, we show that through the Ras/MEK pathway, Ras oncogene up-regulated the expression of superoxide-generating oxidases, Nox1 in rat REF52 cells and Nox4 in primary human lung TIG-3 cells, leading to an increase in intracellular level of ROS. Ablation of Nox1 and Nox4 by small interfering RNAs (siRNAs) blocked the RasV12 senescent phenotype including β-galactosidase activity, growth arrest and accumulation of tumor suppressors such as p53 and p16Ink4a. This suggests that Nox-generated ROS transduce senescence signals by activating the p53 and p16Ink4a pathway. Furthermore, Nox1 and Nox4 siRNAs inhibited both Ras-induced DNA damage response and p38MAPK activation, whereas overexpression of Nox1 and Nox4 alone was able to induce senescence. The involvement of Nox1 in Ras-induced senescence was also confirmed with embryonic fibroblasts derived from Nox1 knockout mice. Together, these findings suggest that Nox1- and Nox4-generated ROS play an important role in Ras-induced premature senescence, which may involve DNA damage response and p38MAPK signaling pathways.
The Ras gene family encodes small GTP-binding proteins, and activation mutation of Ras keeps it in the active GTP-bound state, constitutively activating downstream effectors (Downward 1996). Oncogenic Ras transforms immortal rodent cell lines by promoting cell proliferation and survival (Downward 1996). In contrast, oncogenic Ras causes a permanent growth arrest in normal primary cells, which is referred to as premature senescence (Narita & Lowe 2005). Oncogene-induced senescence is considered to function in vivo as a barrier to tumorigenesis (Narita & Lowe 2005). For instance, premalignant lung adenomas were positive for K-RasV12 oncogene-induced senescence in mouse models of cancer, whereas malignant adenocarcinomas were negative (Collado et al. 2005). Ras-induced senescence involves activation of the Raf/MEK/ERK pathway, leading to the accumulation of tumor suppressors such as p53, p21, p19Arf and p16Ink4a (Lin et al. 1998; Wang et al. 2002; Collado et al. 2005). It has been shown that activated Ras increases intracellular levels of reactive oxygen species (ROS) in primary cells, which in turn triggers senescence, suggesting that ROS serve as critical signaling molecules in oncogenic Ras-induced premature senescence (Lee et al. 1999; Chen et al. 2004). However, the key components of the ROS-generating molecular machinery in this setting have not been fully elucidated.
The Nox family proteins (Nox1–5 and Duox1–2), which are homologous to the catalytic subunit (gp91phox) of phagocytic NADPH oxidase, catalyze superoxide generation and are major sources for cellular ROS (Lambeth 2007). Nox homologues are transmembrane proteins that contain well-conserved catalytic domain and display differing regulatory modes depending on their regulatory components. Although the bactericidal activity was initially identified for Nox2 (gp91phox), Nox isoforms have been shown to play important roles in signaling pathways involved in various physiological processes including growth, immune response, inflammation and apoptosis (Lambeth et al. 2007).
With respect to senescence, Nox1 mediated fibroblast senescence that restricts fibrosis during wound healing process (Jun & Lau 2010). Overexpression of Nox4 in both mouse fibroblast cells (Geiszt et al. 2000) and normal breast epithelial cells (Graham et al. 2010) developed signs of cellular senescence. Nox1 and Nox4 were involved in polyphenol-induced senescence in vascular endothelial cells (Schilder et al. 2009). Although these observations implicate a biological role of Nox1- and Nox4-derived ROS in promoting senescence, it remained unknown whether oncogene-induced premature senescence depends on ROS generation by Nox isozymes. In this regard, a recent report (Weyemi et al. 2012) is noteworthy for demonstrating the importance of Nox4 in H-RasV12-induced DNA damage responses and subsequent senescence in human thyrocytes. However, the contribution of Nox1 to oncogene-induced premature senescence in primary cells has not yet been established. This led us to investigate whether Nox1 is involved in oncogenic Ras-induced premature senescence. Our data suggest that Nox1 cooperates with oncogenic Ras in inducing senescence in primary rodent fibroblast, as in the case of Nox4 in primary human embryonic lung fibroblasts. Furthermore, augmented ROS production by these Nox enzymes appears to promote senescence by activating the p53 and p16Ink4a pathways through both DNA damage response and p38MAPK signaling pathways. Our findings provide a further mechanistic insight into redox regulation of oncogenic Ras-induced premature senescence in primary cells.
Ras induces senescence in REF52 and TIG-3 cells via Nox1 and Nox4
To gain insight into NADPH oxidase function in oncogenic Ras-induced senescence, we introduced an activated form of Ras (H-RasV12) into both rat REF52 cell line and primary human embryonic lung fibroblasts (TIG-3) using retroviral vector infection. The REF52 cell line is immortalized but similar to primary cells under a tight growth control (Serrano et al. 1997). As described before (Serrano et al. 1997), H-RasV12-infected cells displayed a flat, enlarged morphology and had arrested growth by 7 days after infection, which was accompanied by an increase in the activity of a senescence-associated β-galactosidase (SA-β-gal; Fig. 1A). Furthermore, oncogenic Ras elevated the expression of senescence-related cell cycle regulatory proteins: p53, p21 and p19Arf in REF52 cells and p53, p21 and p16Ink4a in TIG-3 cells (Fig. S1 in Supporting Information). When H-RasV12-transduced cells were treated with diphenylene iodonium (DPI), a flavoprotein-dependent oxidase inhibitor and apocynin, an antioxidant, the number of SA-β-gal-positive cells was decreased (Fig. 1B). In contrast, treatment of cells with rotenone, a mitochondrial oxidase inhibitor had little or no effect (Fig. S2 in Supporting Information). The results suggest that the Nox family enzymes are specifically involved in Ras-induced senescence in REF52 and TIG-3 cells.
We then examined the expression of Nox family mRNAs in REF52 cells and TIG-3 cells. RT-PCR (Fig. 1C) and immunoblotting (Fig. 1D) analyses showed that Nox1 and Nox4 were expressed in REF52 cells, and that H-RasV12 transfection up-regulated Nox1 but not Nox4 in REF52 cells. The results suggest that Nox1, rather than Nox4, is involved in Ras-dependent induction of SA-β-gal activity in REF52 cells. In contrast, Nox4 was predominantly expressed in TIG-3 cells, and H-RasV12 transduction enhanced Nox4 expression (Fig. 1C,D). A MEK inhibitor PD 98059 reduced the Ras-induced expression of Nox1 in REF52 cells and that of Nox4 in TIG-3 cells, indicating the MEK/ERK pathway's role in Ras-induced expression of the Nox isoforms (Fig. 1D). Thus, it seems that Ras-dependent senescence correlated with Ras-induced Nox1 in REF52 cells and with Ras-induced Nox4 in TIG-3 cells.
To further assess the involvement of Ras-induced Nox1 and Nox4 in senescence, siRNAs targeting to Nox1 and Nox4 were transfected into the cells with concomitant H-RasV12 expression, and the transduced cells were subjected to the analyses of cell growth and SA-β-gal activity. Although REF52 cells transduced with H-RasV12 underwent growth arrest, transfection of Nox1 siRNA interfered with the growth suppression (Fig. 2A). Because Nox4 is expressed at the basal level in REF52 cells, we knockdowned both Nox1 and Nox4 to examine whether depletion of basal Nox4 has an additive effect to the suppression of senescence. However, co-transfection of siNox4 did not influence the effect of siNox1 (Fig. S7 in Supporting Information). We infer that Nox1 newly induced by H-RasV12 contributes to the Ras-induced senescence.
On the other hand, depletion of Nox4 by Nox4 siRNA removed Ras-induced quiescence of TIG-3 cells (Fig. 2A). Nox1 siRNA and Nox4 siRNA consistently abrogated both the accumulation of SA-β-gal (Fig. 2B) and up-regulation of p53, p21 or p16Ink4a (Fig. 2C) in H-RasV12-transduced cells. Immunoblotting analysis demonstrated that the expression levels of Nox1 and Nox4 proteins were decreased by overexpression of siRNAs, confirming the efficient elimination of their mRNAs (Fig. 2D). Thus, oncogenic Ras-induced premature senescence is, at least in part, mediated by Nox1 in REF52 cells and Nox4 in TIG-3 cells. In additional experiments, we demonstrated that H-RasV12-induced Nox4 expression was responsible for Ras-induced senescence in primary human embryonic lung IMR90 cells, as in the case of TIG-3 cells (Fig. S3 in Supporting Information).
Nox1 and Nox4 mediate Ras-induced ROS generation
We then addressed whether Nox1 and Nox4 are the sources of Ras-induced increase in ROS. Luminol assays showed that the level of superoxide in both REF52 cells and TIG-3 cells was increased 4 days after H-RasV12 infection, whereas treatment of cells with DPI and apocynin significantly reduced the ROS level (Fig. 3A). Similarly, depletion of Nox1 and Nox4 by siRNAs inhibited Ras-induced ROS production in REF52 and TIG-3 cells, respectively (Fig. 3B). Given that Nox1 was up-regulated by activated Ras in REF52 cells (Fig. 1C,D), the data suggest that Nox1 contributes to elevated production of superoxide in H-RasV12-transduced REF52 cells. Similarly, Nox4 is suggested to be involved in the H-Ras-Val12-induced ROS generation in TIG-3 cells. When TIG-3 cells were exposed to a low, nontoxic level of H2O2, senescence was induced (Fig. S4 in Supporting Information), as shown in the increased SA-β-gal activity and up-regulation of growth suppressive genes. This supports the view that Ras leads to senescence in a Nox-derived ROS-dependent manner.
Nox1 and Nox4 mediate both Ras-induced DNA damage responses and p38MAPK activation
DNA damages such as double-strand breaks and guanine oxidation are in part caused by intracellular ROS and contribute to oncogene-induced senescence (Mallette et al. 2007; Maynard et al. 2009). To know whether a DNA damage takes place in response to Nox4 activation in H-RasV12-expressing TIG-3 cells, the level of phosphorylated form of histone H2A.X (γ-H2A.X) as a biomarker for DNA double-strand breaks (Gorgoulis et al. 2005; Nuciforo et al. 2007) was examined. As expected, H-RasV12 increased the level of γ-H2A.X, whereas deficient generation of Nox4-derived ROS due to siRNA repressed the stimulatory effect of Ras on H2A.X phosphorylation (Fig. 4A). Notably, induction of DNA lesions detected by H2A.X phosphorylation was also found to be mediated by Nox1 in H-RasV12-transduced REF52 cells (Fig. 4A). The data suggest that both Nox1- and Nox4-dependent ROS production participate in DNA damage responses in different settings of oncogenic Ras-induced senescence.
Because p38MAPK is activated during the onset of Ras-induced senescence (Lin et al. 1998; Haq et al. 2002; Bulavin et al. 2003) and is involved in oxidative stress-induced senescence (Iwasa et al. 2003), we wondered whether Nox1- and Nox4-derived ROS also transmit senescence signals via p38MAPK upon Ras activation. To test this possibility, the phosphorylation state of p38MAPK was examined in H-RasV12-transduced REF52 and TIG-3 cells. As expected, introducing activated Ras increased phosphorylation of p38MAPK (Fig. 4B). On the contrary, knockdown of Nox1 and Nox4 by siRNAs in H-RasV12-expressing cells significantly suppressed phosphorylation of p38MAPK (Fig. 4B), indicating that Ras activates p38MAPK through Nox1 and Nox4. Thus, beside DNA damage responses, p38MAPK activation via Nox1 or Nox4 appears to be involved in Ras-induced senescence.
Overexpression of Nox1 and Nox4 induces senescence
To determine whether activation of Nox1 or Nox4 per se leads to senescence, Nox1 and Nox4 were overexpressed in REF52 and TIG-3 cells via retroviral gene transduction (Fig. 5A). The level of intracellular ROS was increased upon overexpression of Nox1 and Nox4 (Fig. S5 in Supporting Information). Nox1- and Nox4-overexpressing cells ceased to proliferate and accumulated SA-β-gal during 6–7 days postinfection, whereas vector control displayed little or no senescence biomarker and growth inhibition (Fig. 5B). Parallel to this and like H-RasV12, Nox1 up-regulated the expression of p19Arf in REF52 cells, and Nox4 caused accumulation of p16Ink4a in TIG-3 cells (Fig. 5C). Moreover, a marked increase in the level of phosphorylated p38MAPK was detected in Nox1- or Nox4-expressing cells (Fig. 5C). These results suggest that overexpression of Nox1 or Nox4 by itself leads to the activation of p38MAPK, triggering p19Arf and p16Ink4a tumor suppressor pathways and cellular senescence. The results are consistent with the notion that these Nox enzymes act as downstream effectors for Ras in the induction of premature senescence.
Suppression of Ras-induced senescence in mouse embryo fibroblasts isolated from Nox1KO mice
To further assess the key role of Nox1 in Ras-induced senescence, primary mouse embryo fibroblasts (MEF) derived from wild-type (wt) and Nox1-null (Nox1KO) mouse embryos were infected with retroviruses carrying H-Ras-V12. Ablation of Nox1 expression was confirmed as described in Fig. S6A,B. Oncogenic Ras expression did not induce SA-β-gal staining in MEF-Nox1KO cells, and the induction of p53 and p19Arf was substantially reduced compared with that in WT MEF cells (Fig. 6A,B). DCFH-DA assays showed that the production of ROS was significantly increased by RasV12 in WT MEF cells (Figs 6C, S6C in Supporting Information). Furthermore, the level of ROS generation was decreased in RasV12-transfected MEF-Nox1KO cells as compared with that in RasV12-transfected WT MEF cells. Ras-induced ROS generation was not completely prevented by knockdown of Nox1, which may be attributed to Ras-induced up-regulation of another source of ROS in MEF cells. Thus, the results suggest that Nox1, at least in part, contributes to Ras-induced senescence in primary MEF.
Oncogenic Ras-induced activation of the MEK/ERK pathway is known to lead to premature senescence in primary fibroblasts (Narita & Lowe 2005). Recent studies reported that an increase in intracellular ROS represents a critical signal for Ras-induced senescence in primary cells (Lee et al. 1999; Chen et al. 2004). However, the nature of oxidases involved in this process and its signaling mechanism had not been fully elucidated. Here, we demonstrated that oncogenic Ras up-regulates the expression of Nox4 in human primary lung fibroblasts, and that a subsequent increase in ROS generation activates tumor suppressors such as p53 and p16Ink4a, mediating oncogenic Ras-induced premature senescence. This is consistent with a recent report that expression of oncogenic Ras leads to senescence through Nox4 in human thyrocyte (Weyemi et al. 2012). However, the engagement of Nox family enzymes with oncogenic Ras-driven senescence does not seem to be restricted to Nox4. Our study showed that Nox1 is also involved in the senescence-triggering effect of activated Ras in rodent primary fibroblasts in a manner similar to Nox4 (see model in Fig. 7). Up-regulation of Nox1 and Nox4 by H-RasV12 is mediated by the ERK pathway in REF52 cells and TIG-3 cells, respectively, which is in agreement with the notion that oncogenic Ras-triggered senescence depends on the Ras/Raf/MEK signaling pathway (Narita & Lowe 2005). The results fit well with previous observations that moderate, nontoxic levels of exogenous H2O2 can induce a senescent phenotype (Bladier et al. 1997) and that levels of ROS are elevated in fully senescent cells (Takahashi et al. 2006).
Despite the fact that both Nox1 and Nox4 exist in REF52 cells, Nox1 but not Nox4 was up-regulated by H-RasV12. In contrast, Nox4 was predominant and up-regulated by H-RasV12 in TIG-3 and IMR90 cells. Although the reason for the distinct response to Ras activation between Nox1 and Nox4 expressions is presently unknown, the transcriptional control of Nox genes may vary depending on the cellular context and the type of Nox isoforms.
Another finding in this study is that both DNA damage and p38MAPK activation appear to participate in a senescence signaling network controlled by the Ras-Nox1 or Ras-Nox4 axes. H-RasV12-driven accumulation of Nox4-derived ROS damaged DNA and consequently caused senescence in thyrocytes (Weyemi et al. 2012). As such, we observed that in H-RasV12-expressed REF-52 and TIG-3 cells, inactivation of Nox1 and Nox4 abrogated phosphorylation of histone (H2A.X), a marker of DNA double-strand breaks (Gorgoulis et al. 2005; Nuciforo et al. 2007). In addition, our data showed that besides generating oxidative DNA damage, Nox1 and Nox4 mediated H-RasV12-induced p38MAPK phosphorylation (Figs 4B, 5C). Thus, it is possible that not only the DNA damage response pathway but also the p38MAPK pathway contribute to H-RasV12-triggered, Nox-dependent senescence (Fig. 7). In terms of MAPK activation by ROS, p38MAPK has been suggested to be activated through direct phosphorylation by redox-sensitive apoptosis signal-regulating kinase1 (Ask1; Noguchi et al. 2008). Therefore, it is possible that Ask1 mediates Nox1- or Nox4-induced phosphorylation of p38MAPK. As MAPK activation is also sustained by oxidative inactivation of the redox-sensitive cysteine residues of MAPK phosphatases (Traore et al. 2008), an alternative scenario is that a p38MAPK-specific phosphatase(s) controls Nox-mediated activation of p38MAPK. Studies are currently underway to test these possibilities.
Our previous study showed that oncogenic Ras activation up-regulates Nox1 expression through the MEK-ERK pathway, and that increased Nox1-generated ROS are required for maintaining Ras oncogene-transformed phenotype of the immortal rat kidney cell line NRK (Mitsushita et al. 2004; Adachi et al. 2008; Komatsu et al. 2008; Shinohara et al. 2010). In contrast, the present work suggests that Nox1-derived ROS participate in oncogenic Ras-induced senescence of REF52 cells. These apparent conflicting observations may be explained by the fact that tumor suppressors were frequently lost in immortal cells: p16Ink4a and p53 were deleted in immortalized rodent cell lines such as Balb/c3T3 cells (Harvey & Levine 1991; Quelle et al. 1995). In support of this, Ras failed to induce a decreased rate of proliferation in primary fibroblasts devoid of p53 (Serrano et al. 1996). However, it has been reported that Nox1 is pro-proliferative in many cell types. The present study, therefore, does not rule out the possibility that the observed effects of Nox1 on senescent phenotype are due to supraphysiological overexpression of Nox1. Further study is needed to clarify this issue.
In vitro studies on oncogenic Ras-induced senescence using ectopic expression of activated Ras in primary fibroblasts had been fueled by studies on mouse models of oncogenic Ras-driven tumorigenesis (Collado et al. 2005). However, recent technical advances in mouse modeling for cancer argued that Ras-induced senescence is dependent on the cellular context, as well as the expression level of activated Ras (Frese & Tuveson 2007). Despite the controversy over the physiological relevance of oncogene-induced senescence, accumulating evidence clearly suggests that senescence is not merely an artifact of in vitro cell culture, and that the incidence of cellular senescence is associated with preneoplastic lesions. For example, oncogenic mutation in B-Raf was responsible for senescence in human congenital nevi, a benign form of melanocyte hyperproliferation (Michaloglou et al. 2005). Oncogene-induced senescence has even been proposed to restrict tumor progression by blocking proliferation of damaged or stressed cells (Narita & Lowe 2005). Thus, it would be of particular interest to determine what role, if any, Nox-derived ROS-dependent senescence plays during the course of cancer development.
In summary, our study suggests that Nox1- and Nox4-generated ROS participate in oncogenic Ras-induced premature senescence, which may require biphasic signaling through the DNA damage response and p38MAPK pathways.
Cell culture, preparation of MEFs and reagents
TIG-3 cells were purchased from Health Science Research Resources Bank (Tokyo, Japan), and REF52 cells were provided by Dr M. Nakamura (Tokyo Medical and Dental University).
Generation and characterization of Nox1KO (Nox1−/Y) mice followed previously described methods (Matsuno et al. 2005). MEF cells were prepared from day 13 embryos derived from crosses between Nox1KO mice as described previously (Serrano et al. 1997) and were used between P2 and P6. Four independent wild-type and Nox1KO mouse embryos were used. DPI was obtained from Calbiochem and apocynin from Sigma-Aldrich.
Retroviral gene transfer
pBabe H-RasV12 was a gift from Dr. S. W. Lowe (Cold Spring Harbor Laboratory). pBabe (mock) or pBabe H-RasV12 was stably transfected into PT67 packaging cells (Clone Tech), and the stable clones were isolated. For infection, the first culture supernatants from the virus producing PT67 cells were inoculated into target fibroblasts for 12 h and the infection process was repeated with the second supernatant according to the manufacturer's protocol. The infected cells were then selected with puromycin (0.5 μg/mL for REF52; 1 μg/mL for TIG-3) for 2–3 days and reseeded for various assays. pBabe vectors carrying human Nox1 and Nox4 were constructed and transfected into HEK293 or 293T cells together with VSVenv and Gag/Pol using the calcium phosphate precipitation method. The culture supernatants were harvested 48 h later and inoculated into target cells as described above. For disruption of Nox1 and Nox4, target fibroblasts were first transfected with pSilencer vectors carrying Nox1 siRNA, Nox4 siRNA or scrambled siRNA as described before (Mitsushita et al. 2004; Yamaura et al. 2009) and 24 h later infected with pBabe H-RasV12 retroviral vectors.
SA-β-gal staining was performed as described previously (Serrano et al. 1997).
Cells were plated into 24-well plates and transfected with pSilencer vectors carrying Nox1 siRNA, Nox4 siRNA or scrambled siRNA 24 h before pBabe H-RasV12 or control virus infection. After selection with puromycin for 2 days, cells were reseeded (day 0) and counted at the indicated time points.
Total RNAs were extracted from cells, and PCR was performed by using specific primers for Noxs as described previously (Mitsushita et al. 2004).
Cells were lysed in RIPA buffer and subjected to immunoblotting as described (Shinohara et al. 2010). The following antibodies were used: rabbit anti-p53 and rabbit anti-Nox4 from Santa Cruz Biotechnology, rabbit anti-phospho-p38 (Thr180/Tyr182), mouse anti-p21 and rabbit anti-Nox1 from Sigma, rabbit anti-p16, rabbit anti-H2A.X, rabbit anti-γ-H2A.X and rabbit anti-Ras from Cell Signaling, and chicken anti-p19Arf from Gene Tex. Rabbit anti-Nox1 antibodies were also generated (Komatsu et al. 2008).
Measurement of ROS production
Cell suspensions were incubated with 200 μM luminol and 1 unit horseradish peroxidase for 20 min at room temperature as described (Komatsu et al. 2008). Luminescence was quantified by a Luminometer Lumat LB9507 (Berthold). Alternatively, MEF cells were loaded with 10 μm 2′,7′-dichlorodihydrofluorescein diacetate (DCFH-DA; Molecular Probes) for 30 min at 37 °C. Images were obtained, and fluorescence intensity of approximately 100 random cells was quantified as described (Yamaura et al. 2009).
Differences or correlations between two groups were assessed by Student's t-test with a P value < 0.05 being considered significant.
We thank Drs. S. Lowe and K. Kojima for providing pBabeRasV12 and technical advice in virus production, respectively. We are grateful to F. Ushiyama for assistance with manuscript preparation. This work was supported by a Grant on Cancer Research in Applied Areas from the Ministry of Science and Culture of Japan (T. Kamata).