Different Roles of Negative and Positive Components of the Circadian Clock in Oncogene-induced Neoplastic Transformation*

In mammals, circadian rhythms in physiological function are generated by a molecular oscillator driven by transcriptional-translational feedback loop consisting of negative and positive regulators. Disruption of this circadian clock machinery is thought to increase the risk of cancer development, but the potential contributions of each component of circadian clock to oncogenesis have been little explored. Here we reported that negative and positive transcriptional regulators of circadian feedback loop had different roles in oncogene-induced neoplastic transformation. Mouse embryonic fibroblasts prepared from animals deficient in negative circadian clock regulators, Period2 (Per2) or Cryptochrome1/2 (Cry1/2), were prone to transformation induced by co-expression of H-rasV12 and SV40 large T antigen (SV40LT). In contrast, mouse embryonic fibroblasts prepared from mice deficient in positive circadian clock regulators, Bmal1 or Clock, showed resistance to oncogene-induced transformation. In Per2 mutant and Cry1/2-null cells, the introduction of oncogenes induced expression of ATF4, a potent repressor of cell senescence-associated proteins p16INK4a and p19ARF. Elevated levels of ATF4 were sufficient to suppress expression of these proteins and drive oncogenic transformation. Conversely, in Bmal1-null and Clock mutant cells, the expression of ATF4 was not induced by oncogene introduction, which allowed constitutive expression of p16INK4a and p19ARF triggering cellular senescence. Although genetic ablation of either negative or positive transcriptional regulators of the circadian clock leads to disrupted rhythms in physiological functions, our findings define their different contributions to neoplastic cellular transformation.

In mammals, circadian rhythms in physiological function are generated by a molecular oscillator driven by transcriptionaltranslational feedback loop consisting of negative and positive regulators. Disruption of this circadian clock machinery is thought to increase the risk of cancer development, but the potential contributions of each component of circadian clock to oncogenesis have been little explored. Here we reported that negative and positive transcriptional regulators of circadian feedback loop had different roles in oncogene-induced neoplastic transformation. Mouse embryonic fibroblasts prepared from animals deficient in negative circadian clock regulators, Period2 (Per2) or Cryptochrome1/2 (Cry1/2), were prone to transformation induced by co-expression of H-ras V12 and SV40 large T antigen (SV40LT). In contrast, mouse embryonic fibroblasts prepared from mice deficient in positive circadian clock regulators, Bmal1 or Clock, showed resistance to oncogene-induced transformation. In Per2 mutant and Cry1/2-null cells, the introduction of oncogenes induced expression of ATF4, a potent repressor of cell senescence-associated proteins p16INK4a and p19ARF. Elevated levels of ATF4 were sufficient to suppress expression of these proteins and drive oncogenic transformation. Conversely, in Bmal1-null and Clock mutant cells, the expression of ATF4 was not induced by oncogene introduction, which allowed constitutive expression of p16INK4a and p19ARF triggering cellular senescence. Although genetic ablation of either negative or positive transcriptional regulators of the circadian clock leads to disrupted rhythms in physiological functions, our findings define their different contributions to neoplastic cellular transformation.
The circadian clock is a timekeeping system that allows organisms to adapt their physiological and behavioral functions to anticipatory changes in their environment. In mammals, the circadian clock system is hierarchically organized, consisting of a light-responsive central clock in the suprachiasmatic nuclei of the anterior hypothalamus and subsidiary clocks in other brain regions and peripheral tissues (1). The suprachiasmatic nuclei entrains and synchronizes subsidiary clocks with the environmental light-dark cycle, whereas peripheral clocks regulate tissue-specific functions in an anticipatory manner.
Central and peripheral clocks are both governed by interconnected transcriptional and translational feedback loops (2). The gene products of Bmal1 (also known as Arntl) and Clock form a heterodimer that activates the transcription of Period (Per) and Cryptochrome (Cry) genes. Once PER and CRY proteins have reached a critical concentration, they attenuate BMAL1/CLOCK-mediated transactivation. The alternating activation and suppression of the BMAL1/CLOCK-driven positive loop and PER/CRY-controlled negative loop result in a circadian oscillation in the molecular clock and also regulate 24-h variations in output physiology through the periodic activation/repression of clock-controlled genes (3,4).
Because the expression of up to 10% of genes has been suggested to be under the control of the circadian clock (5), it should not come as a surprise that disruptions in the circadian clock system lead to the onset of various diseases. In fact, several epidemiological analyses and laboratory animal studies also revealed a relationship between disruptions in circadian rhythms and cancer development. For example, human night shift workers are at an increased risk of developing breast, prostate, colon, and endometrial cancers as well as hepatocellular carcinoma and non-Hodgkin lymphoma (6 -7). These epidemiological findings are also supported by animal studies in which repetitive changes in the light-dark cycle facilitate the growth of implanted tumors (8,9). Furthermore, spontaneous as well as radiation-induced tumor development is enhanced in circadian gene-defective animals whose behavioral and physiological rhythms are abnormal phenotypes (10,11).
Although genetic ablation of either negative or positive transcriptional regulators of the circadian feedback loop leads to disruption of the rhythms in physiological functions (1,2), the potential contributions of each circadian clock regulator to oncogenesis have been little explored. In this study we found that negative and positive components of circadian feedback loop played different roles during oncogene-induced neoplastic transformation. Mouse embryonic fibroblasts (MEFs) 3 prepared from Per2 mutant (Per2 m/m ) or Cry1/2-null (Cry1/2 Ϫ/Ϫ ) mice were prone to oncogenic transformation induced by coexpression of H-ras V12 and SV40 large T antigen (SV40LT); however, these oncogene induced expression of cellular senescence-associated proteins in Bmal1-null (Bmal1 Ϫ/Ϫ ) or Clock mutant (Clk/Clk) cells, resulting in failure of their neoplastic transformation. Because the expression of ATF4, a potent repressor of cell senescence-associated proteins, was altered in oncogene-introduced clock gene-defective cells, we further focused on this gene to investigate the roles of negative and positive components of the circadian clock in the neoplastic transformation of mouse fibroblasts.

Experimental Procedures
Treatments of Animals and Cells-Per2 m/m , Bmal1 Ϫ/Ϫ , and Clock mutant (Clk/Clk) mice with an ICR background and wild type mice of the same strain were housed in a temperaturecontrolled (24 Ϯ 1°C) room under a 12-h light:12-h dark cycle. MEFs were prepared by standard techniques (12) from littermate embryos of Per2 m/m , Bmal1 Ϫ/Ϫ , Clk/Clk, or wild-type mice, and cells were maintained in Dulbecco's modified Eagle's medium (DMEM, Sigma) supplemented with 10% fetal bovine serum (FBS). Cry-null MEFs, defective in both Cry1 and Cry2, were kindly provided by Dr. Ueda (13). The preparation of MEFs from each genotype was conducted at least three times. Cells were expanded for three or four passages before the experiments. To achieve oncogenic transformation, MEFs were infected with 1 ϫ 10 6 colony-forming units/ml retroviral vectors expressing H-ras V12 and SV40LT (14). Plasmid vectors (pcDNA3.1; Invitrogen) expressing Per2, Bmal1, or Atf4 were also transfected into oncogene-introduced MEFs. Transgeneexpressing cells were selected with G418 (Wako, Osaka, Japan), and individual colonies were expanded and maintained in media containing 4 g/ml G418.
Male NOD-SCID mice were purchased from Charles River. They were inoculated with oncogenic-transformed MEFs (5 ϫ 10 6 cells/mouse). Tumor volumes were measured by caliper, and values were obtained by multiplying the square of the smallest diameter with the largest diameter. Animals were cared for in accordance with the guidelines established by the Animal Care and Use Committee of Kyushu University (Fukuoka, Japan).
Measurement of Locomotor Activity Rhythm-Mice were housed individually in breeding cages with food and water ad libitum. The cages were placed into an infrared ray area sensor, and locomotor activity was measured every 10 min. Locomotor activity was recorded under the light and dark cycles for 10 days and then continuously recorded under the constant dark condition for 10 days.
Anchorage-independent Cell Growth Assay-To evaluate the oncogenic transformation of cells, the ability of cells to grow in an anchorage-independent manner was assessed using the Quantitative 3D Cell Culture Colony Assay Kit (Nihon-ika Co. Ltd., Osaka, Japan). Cells were seeded in 10% FBS containing DMEM soft agar at a density of 1 ϫ 10 4 cells. The viability of cells in agar was determined by tetrazolium chloride on day 7 after seeding.
Construction of Vectors-Retroviral plasmid vectors expressing H-ras V12 (Addgene plasmid#9051) and SV40LT (Addgene plasmid#13970) were transfected into G3T-hi packaging cells. All of the infected cells were cultured in medium containing the appropriate antibiotics. The mouse Atf4 promoter region spanning bp Ϫ346 to ϩ170 (these numbers represent the distance in base pairs from the putative transcription start site, ϩ1) was amplified by PCR, and the product was ligated into the pGL3-Basic luciferase reporter vector (Promega). The E-box of the Atf4 promoter was mutated at Ϫ83 to Ϫ78 (CACGTG to GAGTCT) using a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). Expression vectors for mouse Per2, Cry1, Bmal1, Clock, and ATF4 were also constructed using complementary DNAs (cDNAs) obtained by RT-PCR derived from mouse hepatic RNA. Coding regions were ligated into the pcDNA3.1 vectors (Invitrogen).
Histochemical Analysis-Cells were fixed with 4% paraformaldehyde/PBS for 10 min. Senescence-associated ␤-galactosidase (␤-gal)-positive cells were detected using a senescence detection kit (BioVision, Inc. Palo Alto, CA). The percentage of ␤-gal-positive cells was determined by counting the number of positive cells in 3 randomly chosen fields at ϫ400 magnification.
Small Interfering RNA-Oncogene-transformed cells were transfected with siRNA against ATF4 (Santa Cruz Biotechnology, Inc. Dallas, TX). This siRNA is designed to prevent the expression of mouse ATF4 (15). Oncogene-introduced cells were transfected with the same amount of scrambled siRNA as the control and were then used in experiments 48 h after siRNA transfection.
RT-PCR Analysis-To quantify the mRNA levels of cellular senescence factors, cDNA was synthesized by reverse transcription using the ReverTra Ace quantitative real-time PCR kit (Toyobo, Osaka, Japan). Diluted cDNA samples were analyzed by PCR using the THUNDERBIRDSYBR qPCR Mix (Toyobo) and 7500 RT-PCR system (Applied Biosystems, Framingham, MA). To confirm the expression of H-ras V12 and SV40LT, diluted cDNA samples were also subjected to PCR as described above, and PCR products were run on agarose gel after staining with ethidium bromide. PCR primer sequences are described in Table 1.
Luciferase Reporter Assay-MEFs prepared from the wild type were seeded at a density of 1ϫ10 5 cells/well on 24-well culture plates. Cells were transfected 18 h later with 100 ng/well of reporter vectors and 1-2 g/well (total) of expression vectors. The pRL-TK vector (0.5 ng/well; Promega) was also cotransfected as an internal control reporter. Cells were then harvested, and cell lysates were analyzed using a dual-luciferase reporter assay system (Promega). The ratio of firefly (expressed from the reporter construct) to Renilla (expressed from pRL-TK) luciferase activities in each sample served as a measure of normalized luciferase activity.
Statistical and Data Analyses-To assess the endogenous period of each animal, locomotor counts were analyzed using ClockLab software (Actimetrics, Evanston, IL). The values presented are expressed as the means Ϯ S.E. The significance of differences among groups was analyzed by analysis of variance followed by Tukey-Kramer's post hoc tests. Equal variances were not formally tested. p Ͻ 0.05 was considered significant.

Different Oncogenic Phenotypes of Per2 m/m and Bmal1
Ϫ/Ϫ Cells-Spontaneous locomotor activity, which is an accurate measure of circadian activity, was monitored in wild-type, Per2 m/m , and Bmal1 Ϫ/Ϫ mice. Animals were initially maintained on a 12-h light/dark cycle (LD) for 10 days and were subsequently maintained under constant darkness (DD) (Fig.  1). The locomotor activity of wild-type mice increased during the dark phase, so that they entrained to the scheduled dark lighting cycle (Fig. 1A). Under constant darkness condition, wild-type mice also exhibited robust circadian rhythms of locomotor activity, and their free-running periods were 23.67 Ϯ 0.34 h (n ϭ 4, mean Ϯ S.E.). In contrast, neither Per2 m/m nor Bmal1 Ϫ/Ϫ mice failed to show entrained rhythms of locomotor activity even under LD cycle condition (Fig. 1, B and C). Furthermore, a spectral analysis of locomotor activity did not detect periodicity in the circadian range for Per2 m/m or Bmal1 Ϫ/Ϫ mice, suggesting that the apparent circadian rhythms are similarly disrupted even though the negative or positive components of the circadian clock are deleted.
To compare the roles of Per2 m/m and Bmal1 Ϫ/Ϫ in oncogenic transformation, we prepared MEFs from wild-type, Per2 m/m , and Bmal1 Ϫ/Ϫ mice and infected cells concomitantly with retrovirus vectors expressing H-ras V12 and SV40LT (Fig. 1D). The expression of mRNAs for these oncogenes was detected on day 3 after infection, and they were equally expressed in all types of cells (Fig. 1E). The concomitant introduction of H-ras V12 and SV40LT significantly enhanced the anchorage-independent growth of wild-type and Per2 m/m cells (p Ͻ 0.01, respectively, Fig. 1F), whereas the introduction of these oncogenes failed to enhance the growth of Bmal1 Ϫ/Ϫ cells. These findings suggested that the loss of Bmal1 resulted in the phenotype resisting oncogene-induced malignant transformation.
We also tested the tumorigenicity of oncogene-introduced cells using an implant model in mice. Equal numbers of wildtype, Per2 m/m , or Bmal1 Ϫ/Ϫ cells infected with the oncogenes (H-ras V12 and SV40LT) were inoculated into the flanks of NOD-SCID mice. Although mice inoculated with wild-type and Per2 m/m cells infected with oncogenes showed the significant growth of tumor masses (Fig. 1G), the average tumor volume in mice inoculated with Per2 m/m cells was significantly larger than that in mice inoculated with wild-type cells (p Ͻ 0.05 on days 17, 20, and 23, respectively). In contrast, mice inoculated with oncogene-introduced Bmal1 Ϫ/Ϫ cells showed no palpable tumor masses throughout the experimental period. Taken together, these results indicated that a deficiency in PER2 tended to enhance tumorigenicity, whereas the loss of BMAL1 resulted in the phenotype resisting oncogene-induced malignant transformation.
Oncogene Introduction Induces Cellular Senescence of Bmal1 Ϫ/Ϫ Cells-During oncogenic transformation, growth signals that override the normal cell-cycle mechanism are activated in cancer cells, resulting in abnormal proliferation. Failure of cells to override this mechanism often causes reversible cell-cycle arrest, apoptotic cell death, or cellular senescence (16). In contrast to reversible cell-cycle arrest, cellular senescence is defined by the irreversible loss of proliferative potential, acquisition of a characteristic morphology, and expression of specific biomarkers, such as ␤-gal (17,18). On day 14 after the infection with retrovirus vectors expressing H-ras V12 and SV40LT, the number of ␤-gal-positive Bmal1 Ϫ/Ϫ cells was significantly higher than those among wild-type and Per2 m/m cells (p Ͻ 0.01, Fig. 2A).
The tumor suppressor gene Cdkn2a acts as a canonical inducer of cellular senescence (18,19). Cdkn2a generates the different transcript variants, p16Ink4a and p19Arf (also known as p14ARF in humans), by using different first exons and alter-nate polyadenylation sites (20,21). The p16Ink4a variants encode structurally related protein isoforms that inhibit CDK4 kinase (19). The CDK4 inhibitor prevents the phosphorylation of pRB by disrupting the activity of the CDK4-cyclin D complex. On the other hand, p19ARF stabilizes p53 by sequestering MDM2 (22). The mRNA levels of p16Ink4a and p19Arf were significantly increased in oncogene-introduced Bmal1 Ϫ/Ϫ cells (p Ͻ 0.01, respectively; Fig. 2B), whereas the mRNA levels of both transcript variants of Cdkn2a in wild-type and Per2 m/m cells were slightly but significantly decreased by the introduction of oncogenes. Consistent with these results, marked increases in the protein levels of p16INK4a and p19ARF were also detected in oncogene-introduced Bmal1 Ϫ/Ϫ cells (Fig. 2C), whereas those protein levels were decreased in oncogene-introduced wild-type and Per2 m/m cells. The induction of cellular senescence factors in oncogene-introduced Bmal1 Ϫ/Ϫ cells was accompanied by the suppressed phosphorylation of pRB as well as increases in the protein levels of p53 (Fig. 2C). These results suggested that oncogene-introduced Bmal1 Ϫ/Ϫ cells failed to override the normal cell-cycle mechanism, resulting in cellular senescence.
SV40LT-transduced cells are considered to be insensitive to growth arrest by p16INK4a and p19ARF, because SV40LT inactivates p53 by protein-protein interaction (23). The results of an immunoprecipitation analysis revealed that the most amounts of p53 protein in oncogene-introduced wild-type and Per2 m/m cells were precipitated together with SV40LT (Fig.  3A). On the other hand, large amounts of p53 protein in Bmal1 Ϫ/Ϫ cells was unable to be precipitated together with SV40LT, suggesting that extensive expression of p53 in oncogene-introduced Bmal1 Ϫ/Ϫ cells overrides the binding capacity of SV40LT. In fact, the expression of the p53 target genes, p21 and Bax, was increased in oncogene-introduced Bmal1 Ϫ/Ϫ cells (Fig. 3B).
The transforming activity of SV40LT has also been attributed to its perturbation of pRb (24). Although large amounts of pRb were precipitated together with SV40LT, the protein was also detected in the supernatant fractions of all cell types (Fig.  3A). These results indicated that the growth inhibitory function of pRb still remained in oncogene-introduced wild-type, Per2 m/m , and Bmal1 Ϫ/Ϫ cells. The retroviral transfer of SV40LT into these cells appeared to partially suppress pRb function.
In an attempt to determine whether accumulation of p53 in oncogene-introduced Bmal1 Ϫ/Ϫ cells contributes to the induction of cellular senescence, oncogene-introduced wild-type and Bmal1 Ϫ/Ϫ cells were incubated in the presence and absence of the p53 inhibitor pifithrin-␣. The treatment with 30 M pifithrin-␣ had a negligible effect on the anchorage-independent growth of oncogene-introduced wild-type cells but significantly enhanced the growth of oncogene-introduced Bmal1 Ϫ/Ϫ cells (p Ͻ 0.01, Fig. 3C). Extensive expression of p53 in oncogene-introduced Bmal1 Ϫ/Ϫ cells appeared to override the binding capacity of SV40LT and allowed the induction of cellular senescence.
ATF4 Determines the Oncogenic Phenotypes of Per2 m/m and Bmal1 Ϫ/Ϫ Cells-A member of the cAMP-binding protein family, ATF4, promotes oncogene-induced transformation by suppressing the expression of p16INK4a and p19ARF (15). The protein levels of ATF4 were substantially increased in oncogene-introduced wild-type and Per2 m/m cells but not in Bmal1 Ϫ/Ϫ cells (Fig. 4A). The excessive expression of ATF4 in oncogene-introduced Per2 m/m cells was repressed by the transfection of functional "native" Per2-expressing vectors (Fig. 4B), whereas the transfection of oncogene-introduced Bmal1 Ϫ/Ϫ cells with Bmal1-expressing vectors caused an increase in ATF4 levels (Fig. 4C). The modulation of ATF4 protein levels by PER2 and BMAL1 occurred at the transcriptional level because the mRNA levels of Atf4 were also down-and up-regulated by PER2 and BMAL1, respectively (Fig. 4, B and C).
The transfection of oncogenic wild-type and Per2 m/m cells with siRNA against Atf4 decreased ATF4 protein levels as well as the phosphorylation of pRB (Fig. 4D). The down-regulation of ATF4 in both types of cells increased the protein levels of p16INK4a and p19ARF, but the effect was greater on p16INK4a than on p19ARF (Fig. 4D). Furthermore, anchorage-independent growth of oncogene-introduced wild-type and Per2 m/m cells was significantly suppressed by down-regulation of ATF4 (p Ͻ 0.01 for wild-type cells; p Ͻ 0.05 for Per2 m/m cells, Fig. 4E). On the other hand, the transfection of oncogene-introduced Bmal1 Ϫ/Ϫ cells with ATF4-expressing vectors decreased the protein levels of p16INK4a and p19ARF (Fig. 4F), whereas these protein levels in oncogene-introduced wild-type cells were not markedly affected by the transfection of ATF4-expressing vectors. The overexpression of ATF4 in Bmal1 Ϫ/Ϫ cells also significantly enhanced their anchorageindependent growth ability (p Ͻ 0.01, Fig. 4G). These results suggested that the tumorigenicity of clock gene-defective cells was dependent on the responsiveness of ATF4 to the introduction of oncogenes.

Different Oncogenic Phenotypes of Cry1/2 Ϫ/Ϫ and Clk/Clk
Cells-Next, we also investigated the tumorigenicity of other clock gene-defective cells. CRY proteins act as negative tran-  scriptional regulators of circadian feedback loops. Cry-null mice, defective in both Cry1 and Cry2 (Cry1/2 Ϫ/Ϫ ), accordingly show arrhythmic behavior, physiology, and metabolism (25). In addition, the CLOCK protein heterodimerizes with BMAL1 and acts as a positive transcriptional regulator in the circadian clock machinery. Clk/Clk mice also exhibit abnormal rhythms in physiology and behavior (26,27). The concomitant introduction of H-ras V12 and SV40LT significantly enhanced the anchorage-independent growth of Cry1/2 Ϫ/Ϫ cells (p Ͻ 0.01) but did not promote the growth of Clk/Clk cells (Fig. 5A). In contrast, on day 14 after the infection with retrovirus vectors expressing oncogenes, the number of ␤-gal-positive Clk/Clk cells was significantly higher than those among wild-type and Cry1/2 Ϫ/Ϫ cells (p Ͻ 0.01 respectively; Fig. 5B). The retroviral transfer of H-ras V12 and SV40LT into Clk/Clk cells had a negligible effect on ATF4 protein levels, whereas oncogene introduction enhanced the expression of p16INK4a, p19AFR, and p53 (Fig. 5C). These results also supported negative and positive regulators of circadian clock playing different roles in neoplastic transformation induced by the introduction of H-ras V12 and SV40LT.
Transcriptional Regulation of Atf4 by Products of Clock Genes-In the final set of experiments, we investigated the mechanisms by which circadian clock genes modulate ATF4 expression during oncogenic transformation. The mRNAs for Per2, Cry1, Cry2, Bmal1, and Clock were detected in normal wild-type cells (day 0). Their mRNA levels were elevated after the concomitant introduction of H-ras V12 and SV40LT; however, the time courses of mRNA elevations in oncogene-intro-duced cells were similar to those observed in cells that were infected with mock vectors (Fig. 6A). In contrast, the introduction of H-ras V12 and SV40LT into wild-type cells significantly increased the mRNA levels of Atf4 (p Ͻ 0.05).
The products of Bmal1 and Clock have been shown to positively regulate the expression of their target genes through an E-box (CACGTG) element (1,2). BMAL1/CLOCK-mediated transactivation is repressed by PERs or CRYs. An E-box motif has been detected upstream of the Atf4 gene in mice and all mammals examined, including rats, chimpanzees, and humans (28,29). The activity of the luciferase reporter of the mouse Atf4 promoter region containing the E-box (native Atf4::Luc) was enhanced when it was transfected into oncogene-transformed wild-type cells (Fig. 6B). The enhanced reporter activity of native Atf4::Luc was repressed by the co-transfection with Per2or Cry1-expressing vectors. In contrast, the co-transfection with Bmal1-or Clock-expressing vectors further enhanced the reporter activity of native Atf4::Luc in oncogene-transformed wild-type cells. Although the reporter activity of E-box-mutated Atf4 luciferase constructs (E-box mut. Atf4::Luc) was also enhanced in oncogene-transformed wild-type cells (Fig. 6B), enhanced reporter activity was not further modulated by the co-transfection with Per2-and Bmal1-expressing vectors. These results indicated that the oncogene-induced transformation of cells resulted in elevations in ATF4 levels without affecting clock gene expression, whereas components of the circadian clock appeared to influence the expression of Atf4 in oncogenic cells through E-box elements.

Discussion
The results of the present study suggest the opposite roles for the negative and positive components of the circadian clock in oncogene-induced transformation. As shown in Fig. 6C, when the circadian rhythms are disrupted due to a dysfunction in the transcriptional repressor PER or CRY, the expression of ATF4 is facilitated by oncogenic stimuli, and cells override normal cell cycle mechanism, and easily undergo malignant transformation. By contrast, when the circadian rhythms are disrupted due to a dysfunction in the transcriptional activator BMAL1 or CLOCK, ATF4 is unresponsive to oncogenic stimuli, and elevations in the expression of p16INK4a and p19ARF causes cellular senescence. Consequently, oncogene-introduced Bmal1-or Clock-defective cells fail to override the normal cell-cycle mechanism.  (n ϭ 4). C, possible mechanisms for the promotion of oncogenicity by disruptions in the molecular circadian clock. When the circadian clock is disrupted due a dysfunction in the transcriptional repressor PER or CRY, ATF4 is markedly expressed by oncogenic stimuli, and cells easily undergo malignant transformation. In contrast, when the circadian clock is disrupted due to a dysfunction in the transcriptional activator CLOCK or BMAL1, ATF4 is hardly expressed by oncogenic stimuli, and p16INK4a and p19ARF induce cellular senescence.
ATF4 is ubiquitously expressed throughout the body and is induced in response to various stress signals, including anoxia, hypoxia, endoplasmic reticulum stress, amino acid deprivation, and oxidative stress (29). The stress-induced expression of ATF4 causes adaptive responses in cells through regulating the expression of target genes involved in amino acid synthesis, differentiation, metastasis, angiogenesis, and drug resistance (30). The expression of ATF4 is also under the control of the circadian clock (28). BMAL1:CLOCK heterodimers activate the transcription of Atf4 through E-box enhancer elements, but this transactivation is repressed by PER and CRY. The reporter activity of native Atf4::Luc in oncogene-introduced cells was further enhanced by Bmal1-and Clock-expressing vectors. Therefore, the expression of ATF4 appeared to be diminished in Bmal1-or Clock-defective cells even when those cells were introduced with oncogenes. On the other hand, in the absence of negative circadian regulator PER and CRY, ATF4 seemed to be markedly expressed in response to oncogenic stimuli. Excessive expression of ATF4 is often observed in malignant tumors in humans and rodents (30). The highly expressed ATF4 is thought to facilitate tumor progression, because transcription of genes involved in tumor cell proliferation is modulated by ATF4 (31).
We previously reported that ATF4 promotes the oncogeneinduced neoplastic transformation of MEFs by suppressing the expression of p16INK4a and p19ARF (15). The introduction of H-ras V12 and SV40LT into wild-type MEFs induces the expression of the Atf4 gene along with that of p16INK4a and p19ARF. Elevations in ATF4 protein levels suppress the expression of these cellular senescence factors and drive oncogenic transformation. Therefore, elevated levels of ATF4 in oncogene-introduced wild-type and Per2 m/m cells may be sufficient to suppress the expression of p16INK4a and p19ARF. Conversely, low levels of ATF4 in Bmal1 Ϫ/Ϫ cells appear to allow the constitutive expression of these cellular senescence factors.
In this study we used retrovirus vectors expressing SV40LT as a transforming agent together with H-ras V12 . Because SV40LT inactivates p53 by protein-protein interactions and also perturbs the function of pRb (23,24), SV40LT-transduced cells are considered to be insensitive to growth arrest by p16INK4a and p19ARF. Inactivation of the growth-suppressive properties of p53 and pRb was previously shown to be essential for the immortalization of MEFs by SV40LT. The concomitant introduction of H-ras V12 and SV40LT induced elevations in p53 protein levels in Bmal1 Ϫ/Ϫ cells. p53 protein levels were markedly larger in oncogene-introduced Bmal1 Ϫ/Ϫ cells than in wild-type and Per2 m/m cells. Significant amounts of p53 protein in Bmal1 Ϫ/Ϫ cells was unable to be precipitated together with SV40LT. In addition, there were no significant mutations in the p53 gene or p16Ink4a gene among wild-type, Per2 m/m , and Bmal1 Ϫ/Ϫ cells (supplemental Fig. S1). Extensive expression of p53 in oncogene-introduced Bmal1 Ϫ/Ϫ cells overrode the binding capacity of SV40LT. Therefore, untrapped/free p53 protein may allow the induction of cellular senescence. On the other hand, in wild-type, Per2 m/m , and Bmal1 Ϫ/Ϫ cells, a large amount of pRb was unable to be precipitated together with SV40LT. The growth inhibitory function of pRb still remained in all cell types after the infection of oncogenes. The function of the remained pRb may be modulated by p16INK4a.
Although genetic ablation of either negative or positive components of the circadian clock leads to disrupted rhythm in physiological functions, our findings define the different contributions of each component of the circadian clock to neoplastic transformation. The fundamental mechanism of the mammalian circadian clock is conserved beyond the species (32). Similar mechanism may function in human cells during oncogenic transformation associated with the disruption of circadian clock machinery.
Author Contributions-C. K., S. K., and S. O. designed the study and wrote the paper. C. K. and S. K. performed and analyzed the experiments shown in Figs. 1-6. S. Shiromizu, N. Matsunaga, S. Shimba S., and S. Shibata provided technical assistance and contributed to the preparation of the figures. All authors reviewed the results and approved the final version of the manuscript.