Mutation of the gene encoding the circadian clock component PERIOD2 in oncogenic cells confers chemoresistance by up-regulating the Aldh3a1 gene

Disruption of circadian rhythms has been implicated in an increased risk for cancer development. The Period2 (Per2) gene encodes one of the major components of the mammalian circadian clock, which plays a key role in controlling the circadian rhythms in physiology and behavior. PER2 has also been reported to suppress the malignant transformation of cells, but its role in the regulation of cancer susceptibility to chemotherapeutic drugs remains unclear. In this study, we found that oncogene-transformed embryonic fibroblasts prepared from Per2-mutant (Per2m/m) mice, which are susceptible to both spontaneous and radiation-induced tumorigenesis, were resistant against common chemotherapeutic drugs and that this resistance is associated with up-regulation of the aldehyde dehydrogenase 3a1 (Aldh3a1) gene. Co-expression of the oncogenes H-rasV12 and SV40 large T-antigen induced malignant transformation of both WT and Per2m/m cells, but the cytotoxic effects of the chemotherapeutic agents methotrexate, gemcitabine, etoposide, vincristine, and oxaliplatin were significantly alleviated in the oncogene-transformed Per2m/m cells. Although introduction of the two oncogenes increased the expression of Aldh3a1 in both WT and Per2m/m cells, the ALDH3A1 protein levels in the Per2m/m cells were ∼7-fold higher than in WT cells. The elevated ALDH3A1 levels in the oncogene-transformed Per2m/m cells were sufficient to prevent chemotherapeutic drug–induced accumulation of reactive oxygen species. Consequently, shRNA-mediated suppression of Aldh3a1 expression relieved the chemoresistance of the Per2m/m cells. These results suggest a role for mutated PER2 in the development of multiple drug resistance and may inform therapeutic strategies for cancer management.

The rotation of the Earth with a period length of about 24 h has led to the evolution of an endogenous timing system within a large number of species, the circadian clock, which allows organisms to adapt their physiological and behavioral functions to anticipatory changes in their environment. In mammals, circadian rhythms in physiological functions are generated by a molecular oscillator driven by a transcriptional-translational feedback loop consisting of negative and positive regulators (1). The gene products of Clock and Bmal1 (also known as Arntl) form a heterodimer that acts as a positive transcription factor to activate the transcription of the Period (Per) and cryptochrome (Cry) genes. Once the PER and CRY proteins have reached a critical concentration, they act as negative transcription factors to attenuate CLOCK/BMAL1-mediated transactivation (2, 3). Rev-erb␣ (known as Nrd1d1) is also activated by CLOCK/ BMAL1 and transrepressed by PER and CRY, resulting in circadian oscillation in the expression of Rev-erb␣ (4). In turn, REV-ERB␣ periodically represses Bmal1 transcription, thereby interconnecting the positive and negative loops (5). Like the mechanism of Rev-erb␣ transcription, clock genes, which comprise the core oscillation loop, control rhythmic RNA and protein abundance (6 -8) and also allow organisms to synchronize their physiological and behavioral functions to anticipatory changes in their environment.
Because the expression of up to 10% of genes has been suggested to be under the control of the circadian clock (9), 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 have revealed a relationship between disruptions in circadian rhythms and cancer development. For example, long-term shift workers are at an increased risk of developing breast, prostate, colon, and endometrial cancers, as well as non-Hodgkin lymphoma (10 -12). These epidemiological findings are supported by animal studies in which repetitive changes in the light-dark cycle are found to facilitate the growth of implanted tumors (13,14). Furthermore, genetic ablation of the circadian clock gene also enhances the tumorigenesis in the laboratory animals. PER2 is an essential component of mammalian circadian clock (15). Mice with a mutated Per2 gene (Per2 m/m ) are predisposed to spontaneous as well as radiation-induced tumor development (16). We also demonstrated previously that embryonic fibroblasts prepared from Per2 m/m mice were susceptible to transformation induced by the co-expression of H-ras V12 and SV40 large T-antigen (SV40LT), and the onco-gene-transformed Per2 m/m cells have a high tumor formation potential (17). However, the role of the Per2 gene in the regulation of cellular chemosensitivity remains unclear.
In this study, we found that the cytotoxic effects of common chemotherapeutic drugs were diminished in oncogene-transformed Per2 m/m cells. Expression of the aldehyde dehydrogenase 3a1 (Aldh3a1) gene in Per2 m/m cells was remarkably increased by the introduction of oncogenes, and potent elevation of its enzymatic activity attenuated the cytotoxicity of chemotherapeutic drugs. Collectively, the results of the present study suggest a role for PER2 in the development of multiple drug resistance and offer new insights into therapeutic strategies for the treatment of cancers.

Oncogene-transformed Per2 m/m cells resist the cytotoxicity of chemotherapeutic drugs
We previously reported the preparation of oncogene-transformed WT and Per2 m/m cells that were infected concomitantly with retrovirus vectors expressing H-ras V12 and SV40LT (17). The expression of mRNAs for these oncogenes was detected on day 3 after infection, and they were equally expressed in both types of cells (17). The concomitant introduction of H-ras V12 and SV40LT significantly enhanced the anchorage-independent growth of WT and Per2 m/m cells (17). Therefore, we used these cells to investigate the role of the Per2 gene in the regulation of susceptibility of cells to chemotherapeutic drugs, methotrexate (MTX), 3 gemcitabine (GEM), etoposide (VP-16), oxaliplatin (L-OHP), and vincristine (VCR).
The viability of oncogene-transformed WT cells was dosedependently decreased by treatment with all five types of chemotherapeutic drugs (Fig. 1A). The IC 50 values of WT cells to MTX, GEM, VP-16, VCR, and L-OHP were 0.12, 0.27, 26.02, 0.39, and 4.32 M, respectively. Similar dose-dependent decreases in viability were observed when oncogene-transformed Per2 m/m cells were treated with MTX, GEM, VP-16, VCR, and L-OHP, but the cytotoxic effect of all five chemotherapeutic drugs on Per2 m/m cells was attenuated as compared with those on WT cells. We prepared oncogene-transformed cells three times. In every preparation, the chemosensitivity of Per2 m/m cells was lower than that of WT cells.
p53 acts as a universal sensor of genotoxic stress and plays a critical role in chemotherapeutic drug-induced apoptotic cell death (18,19). However, SV40LT-transduced cells are immortalized by inactivation of p53 through protein-protein interaction (20). After treatment with chemotherapeutic drugs, p53 protein was accumulated in the nuclear fraction of oncogeneintroduced WT and Per2 m/m cells. The results of an immunoprecipitation analysis revealed that the greatest amounts of p53 protein in WT and Per2 m/m cells were precipitated together with SV40LT (Fig. 1B), suggesting that p53 is unlikely to be involved in the chemotherapeutic drug-induced decrease in the viability of cells that were infected with H-ras V12 and SV40LT. In fact, treatment of oncogene-transformed WT and Per2 m/m cells with 30 M pifithrin-␣, an inhibitor of p53-mediated transcription, was also unable to modulate their chemosensitivity (Fig. 1C). Because 30 M pifithrin-␣ is sufficient to enhance the chemosensitivity of several types of cancer cell lines (21)(22)(23), SV40LT seemed to inactivate p53 in oncogenetransformed WT and Per2 m/m cells.
The sensitivity of cells to chemotherapeutic drugs is also thought to be dependent on cell-cycle phase, but comparison of flow cytometry histograms from oncogene-transformed WT and Per2 m/m cells revealed no significant difference in the cellcycle distribution between the genotypes (Fig. 1D). These results suggest that apoptotic process and cell-cycle phase are unlikely to contribute to the diminished chemosensitivity of Per2 m/m cells.
Elevated expression of several ABC transporters is often associated with multidrug resistance (24,25). However, the levels of P-glycoprotein (P-gp), multidrug resistanceassociated protein-2 (MRP2), and breast cancer-resistant protein (BCRP) in oncogene-transformed Per2 m/m cells were comparable with those expressed in WT cells (Fig. 1E, left). Intracellular accumulation of MTX, GEM, VP-16, VCR, and L-OHP was also not significantly different between WT and Per2 m/m cells (Fig. 1E, right panels), suggesting that the tolerance of the oncogene-transformed Per2 m/m cells to the chemotherapeutic drugs is not due to the function of efflux transporters.

Up-regulation of Aldh3a1 in oncogene-transformed Per2 m/m cells
To investigate the underlying mechanism of the chemoresistance of oncogene-transformed Per2 m/m cells, we carried out microarray analysis to identify the gene regulating the susceptibility of Per2 m/m cells to chemotherapeutic drugs. After concomitant introduction of H-ras V12 and SV40LT, 1,427 genes were induced or repressed in WT cells ( Fig. 2A). Similarly, expression of 1,687 genes in Per2 m/m cells were altered by oncogenic stimuli. A total of 73 genes in oncogene-transformed Per2 m/m cells were differentially expressed as compared with those in WT cells (Table S1). Of these differentially expressed genes, the expression of 10 genes in oncogene-transformed Per2 m/m cells was higher than that in WT cells, whereas the expression of 63 genes was lower in oncogene-transformed Per2 m/m cells. Among the differentially regulated genes, we identified Aldh3a1 as the up-regulated gene in oncogene-transformed Per2 m/m cells with the greatest -fold change (Fig. 2B). Elevated expression of Aldh3a1 mRNA in oncogene-transformed Per2 m/m cells was also confirmed by RT-qPCR analysis (Fig. 2C). The results of Western blot analysis revealed that ALDH3A1 levels in Per2 m/m cells were ϳ7-fold higher than in WT cells (Fig. 2D). High ALDH activity is often detected in cells with stemlike properties (26). Our previous study demonstrated that oncogene-transformed Per2 m/m cells have potent tumor formation ability (17). Indeed, the expression levels of known cancer stemness markers (Kruppel-like factor 4 (Klf4); POU domain, class 5, transcription factor 1 (Pou5f1); and c-Myc) in oncogene-transformed Per2 m/m cells were significantly higher than those in WT cells (p Ͻ 0.01, respectively; Fig.  2E), confirming that oncogene-transformed Per2 m/m cells have stemlike properties.

Elevated ALDH3A1 attenuates the cytotoxic effects of chemotherapeutic drugs in oncogene-transformed Per2 m/m cells through the prevention of H 2 O 2 accumulation
High ALDH activity in cancer cells is often relevant to their resistance against chemotherapy (27). Therefore, we further focused on this enzyme and investigated its role in the regulation of cellular chemosensitivity.
The expressions of antioxidant degradation enzymes, catalase, GSH peroxidase, and superoxide dismutase-3 (SOD3), were not significantly different between WT and Per2 m/m cells (Fig. 3A). Furthermore, N-acetylcysteine (NAC), an antioxidant precursor to GSH, had a negligible effect on the chemosensitivity of oncogene-transformed Per2 m/m cells (Fig. 3B). In contrast to these observations, treatment of oncogene-transformed Per2 m/m cells with 30 M CB29, a selective ALDH3A1 inhibitor,
To further investigate the role of ALDH3A1 in the regulation of cellular chemosensitivity, we prepared oncogene-transformed Per2 m/m cells with down-regulated expression of ALDH3A1. Infection of oncogene-transformed Per2 m/m cells with retrovirus vectors expressing shRNA against Aldh3a1 caused a reduction of its protein levels (Fig. 3D). Down-regulation of ALDH3A1 in oncogene-transformed Per2 m/m cells also restored their susceptibility to the chemotherapeutic drugs, with their susceptibilities becoming similar to those observed in oncogene-transformed WT cells (Fig. 3E). These results suggest that elevated expression of ALDH3A1 in oncogene-trans-formed Per2 m/m cells attenuates the cytotoxicity of chemotherapeutic drugs.
Chemotherapeutic drug-induced DNA damage ultimately causes cell death via enhanced production of reactive oxygen species (ROS) (28). High ALDH activity protects cells from the cytotoxic effect of chemotherapeutic drugs through degradation of ROS (29). After treatment with 1 M MTX, 50 M VP-16, or 5 M VCR, H 2 O 2 accumulated in both oncogenetransformed WT and Per2 m/m cells (Fig. 3F), but the accumulation of H 2 O 2 in oncogene-transformed Per2 m/m cells was lower than that in WT cells. Down-regulation of ALDH3A1 in oncogene-transformed Per2 m/m cells restored the chemotherapeutic drug-induced ROS accumulation (Fig. 3F).

Role of PERIOD2 in development of chemoresistance Histone modifications in Aldh3a1 gene of oncogenetransformed Per2 m/m cells
A large number of genes was induced or repressed by concomitant introduction of H-ras V12 and SV40LT ( Fig. 2A). Alterations of gene expression in oncogenic cells are often associated with epigenetic modifications (30). To investigate the underlying mechanism of the up-regulation of Aldh3a1 gene in oncogene-transformed Per2 m/m cells, we assessed the DNA methylation status of Aldh3a1. The upstream region of the mouse Aldh3a1 gene was retrieved by using an on-line genome browser hosted by the University of California, Santa Cruz (http://genome.ucsc.edu/index.html) 4 (49). Although there were no significant CpG islands within the 5,000 bp up-and downstream from the transcription start site of the mouse Aldh3a1 gene, several 5Ј-CCGG-3Ј sequences were located in the up-and downstream regions of the mouse Aldh3a1 gene (Fig. 4A). Because the methylation status of 5Ј-CCGG-3Ј sequences around the transcriptional start site of the human ALDH3A1 is relevant to its expression levels (31), we investigated the methylation status of these sites in the mice by a methylation-sensitive amplification polymorphism method using isoschizomers, HpaII and MspI. However, no significant difference in the DNA CCGG methylation was detected between oncogene-introduced WT and Per2 m/m cells (Fig. 4B). The methylation status of 5Ј-CCGG-3Ј sequences within Aldh3a1 in oncogene-transformed Per2 m/m cells was also not significantly different from those in mock-transformed cells,

Role of PERIOD2 in development of chemoresistance
confirming that changes in the status of promoter DNA methylation are unlikely to be associated with the elevation of Aldh3a1 expression in oncogene-introduced Per2 m/m cells.
The acetylation and methylation status of specific lysine residues on histone H3 is also involved in the regulation of the expression for numerous genes (32). Next, we investigated whether histone modifications were induced in the Aldh3a1 gene after the introduction of oncogenes. To achieve this, ChIP analysis was performed on WT and Per2 m/m cells to identify the enrichment of an active histone mark (histone H3 lysine 9 acetylation (H3K9Ac)) and repressive histone mark (trimethylation of histone H3 lysine 27 (H3K27me3)) in the upstream and downstream regions of the transcriptional start site of Aldh3a1 gene. The presence of H3K9Ac and H3K27me3 in the promoter region of Aldh3a1 was confirmed by qPCR analysis, which revealed that proximal to the transcriptional start site of Aldh3a1 gene in oncogene-introduced Per2 m/m cells was enrichment of the active histone mark H3K9Ac and depletion of the repressive mark H3K27me3 (Fig. 4C). Although a similar pattern of histone modifications was also observed in oncogene-transformed WT cells, the active mark H3K9Ac was significantly enriched in oncogene-introduced Per2 m/m cells (p Ͻ 0.01). Because H3K9 acetylation is implicated in chromatin remodeling to promote efficient gene transcription, the histone

Alleviation of HDAC1 and HDAC2 recruitment to Aldh3a1 gene in oncogene-transformed Per2 m/m cells
Because the acetylation state of histone H3K9 was different between oncogene-transformed WT and Per2 m/m cells, we investigated whether PER2 interacted with histone modification enzymes that are associated with acetylation of H3K9. In the nuclear fraction of oncogene-transformed WT cells, PER2 was co-immunoprecipitated with HDAC1 and HDAC2, which are known as PER2-associated enzymes (33, 34) exhibiting deacetylation activity toward H3K9 (Fig. 4D). Per2 m/m cells harbor a deletion of 87 amino acids from the PAS (PER-ARNT-SIM) domain of the PER2 protein (15). Mutated PER2 protein (missing residues 348 -434) exhibits reduced translocation into the nucleus and instead accumulates in the cytoplasm (35). As reported previously, the level of the mutated PER2 protein remained lower in the nuclear fraction of oncogene-transformed Per2 m/m cells (Fig. 4D, Input). Although obvious amounts of HDAC1 and HDAC2 were presented in nuclear fraction of both oncogene-transformed WT and Per2 m/m cells (Fig. 4D, Input), the amounts of immunoprecipitated HDAC1 and HDCA2 in the nuclear fraction of oncogene-transformed Per2 m/m cells were lower than those in WT cells (Fig. 4D, IP:  PER2). Furthermore, the binding amounts of both HDACs around the transcriptional start site of the Aldh3a1 gene were significantly decreased in oncogene-transformed Per2 m/m cells (Fig. 4E), although recruiting p300/CBP-associated factor (PCAF) was comparable between WT and Per2 m/m cells. Taken together, these data support a model in which PER2 acts as a repressor of oncogene-induced Aldh3a1 expression through recruitment of HDACs to the promoter region. Dysfunction of PER2 appears to allow H3K9 acetylation, therefore leading to enhancement of oncogene-induced expression of ALDH3A1 (Fig. 5).

Discussion
Recent accumulating evidence has established a significant role of circadian genes in the regulation of cell-cycle progression and DNA damage response (36,37). In mammals, dysfunction of the circadian machinery has been implicated in carcinogenesis as well as its recurrence (38). Although Per2 has a critical role in controlling the malignancy of cancers (17,34), our results also showed a mechanism regulating the resistance of oncogene-transformed Per2 m/m cells against the cytotoxicity of chemotherapeutic drugs. The development of chemoresistance was associated with up-regulation of the Aldh3a1 gene, accompanied by histone H3 modification.
Negative correlation between the expression levels of PER2 and tumor malignancies has been reported for breast cancers  (34). Down-regulation of PER2 promotes the malignancy of human breast cancer cell lines by enhancing invasion and activating expression of epithelial-mesenchymal transition genes. The retrospective analysis of patients with breast cancer also indicates that low expression of PER2 is associated with poor prognosis. The role of clock genes in controlling the sensitivity of cancer cells to chemotherapeutic drugs has been demonstrated by the overexpression of Bmal1 (39), loss of Cry1/2 (40), and down-regulation of Per2 (41). The modulation of chemosensitivity of cancer cells by clock genes is closely related to their ability to regulate cell-cycle progression and apoptosis. The products of several clock genes and/or clock-controlled genes regulate the expression of genes responsible for cell-cycle progression and DNA damage response (36,37). However, our present results indicated that the chemosensitivity of oncogene-transformed Per2 m/m cells was unlikely to be dependent on alteration of the apoptotic process and cell-cycle phase. p53 acts as a universal sensor of genotoxic stress and partially mediates ROS-induced cell death (42). However, the results of an immunoprecipitation analysis revealed that the greatest amounts of p53 protein in Per2 m/m cells as well as WT cells were precipitated together with SV40LT, even when cells were treated with chemotherapeutic drugs. Furthermore, the distribution of the cell-cycle phase of oncogene-introduced Per2 m/m cells was not significantly different from that of WT cells. Therefore, the underlying mechanism of the development of chemoresistance of oncogene-transformed Per2 m/m cells appeared to be distinct from findings in previous reports describing clock gene-deficient cells (39 -41).

Role of PERIOD2 in development of chemoresistance
The ALDH3 family includes enzymes able to oxidize medium-chain aliphatic and aromatic aldehydes (43). These enzymes also have noncatalytic activities, including antioxidant function and some structural roles. ALDH3A1 is highly expressed in the stomach, lung, keratinocytes, and cornea, but poorly detected in normal liver (43). High ALDH3A1 expression and activity have been correlated with cell proliferation, resistance against aldehydes derived from lipid peroxidation, and resistance against the cytotoxic effects of drugs; therefore, the activity is also used as a marker for cancer stemlike cells (44,45). In fact, the expression levels of cancer stemness markers Klf4, Pou5f1, and c-Myc were significantly increased in oncogene-introduced Per2 m/m cells. Our previous study also demonstrated that oncogene-transformed Per2 m/m cells show high tumor formation and increased anchorage-independent growth activity (17). Therefore, these malignant phenotypes may also be due to high ALDH3A1 expression. Chemotherapeutic drug-induced DNA damage ultimately causes cell death via enhancing the production of ROS (28). Due to antioxidant function (29), elevated expression of ALDH3A1 in oncogene-transformed Per2 m/m cells appeared to contribute to their resistance against chemotherapeutic drugs through preventing ROS accumulation. This notion was also supported by the present finding that down-regulation of ALDH3A1 in oncogene-transformed Per2 m/m cells restored the chemotherapeutic drug-induced accumulation of H 2 O 2 . The expressions of other H 2 O 2 degradation enzyme were not significantly different between WT and Per2 m/m cells. Therefore, enhanced expression of ALDH3A1 appeared to be involved in the development of chemoresistance of oncogene-transformed Per2 m/m cells.
The expression of rat ALDH3 is induced by polycyclic aromatic hydrocarbons, chlorinated compounds, or the activation of the aryl carbon receptor (46). Peroxisome proliferatoractivated receptor ␥ also negatively regulates the expression of human ALDH3A1 (47). Therefore, the orphan receptor agonist is able to suppress ALDH3A1 expression. Oncogene transformation of cells often causes irreversible changes in gene expression, leading to rapid proliferation and high invasiveness. Such alterations of gene expression in oncogene-transformed cells are thought to be associated with epigenetic modifications (30). Although methylation of CCGG sites around the transcriptional start site of Aldh3a1 was not significantly changed by the introduction of oncogenes, acetylation and trimethylation of histone H3 were modified in oncogene-transformed cells. As compared with WT cells, acetylation of H3K9 on Aldh3a1 gene was enriched in oncogene-transformed Per2 m/m cells. H3K9 acetylation is implicated in chromatin remodeling to promote efficient gene transcription, suggesting a potential underlying cause of enhanced Aldh3a1 expression in oncogene-transformed Per2 m/m cells. Actually, we prepared oncogene-transformed cells three times, and although the expression levels of ALDH3A1 protein were ϳ2-fold different in each preparation, the dehydrogenase levels in oncogene-transformed Per2 m/m cells were 7-12-fold higher than those in WT cells in every preparation.
PER2 is capable of interacting with several histone modification enzymes, such as HCAC1, HDAC2, SIN3-HDAC, EZH2, SUZ12, and SUV39H (33,34). In this study, we also observed the interaction of PER2 with HDAC1 and HDAC2 in nucleus of WT cells. HDACs catalyze H3K9Ac deacetylation, resulting in gene silencing (48). Mutated PER2 protein, which is produced in Per2 m/m cells, also interacted with HDACs in the cytoplasm, but the difficulties in nuclear translocation of the mutated PER2 protein appeared to prevent the recruitment of HDACs around the transcriptional start site of the Aldh3a1 gene. This may have allowed the enrichment of H3K9 acetylation, which accounted for the enhanced ALDH3A1 expression in oncogene-transformed Per2 m/m cells. However, we were unable to clarify how PER2 protein directs deacetylation activity of HDACs toward the Aldh3a1 gene. In addition to the Aldh3a1 gene, decreased deacetylation of H3K9 may also modify the expression of genes in oncogene-transformed Per2 m/m cells ( Fig. 2A), whose transcriptional activity is highly dependent on the acetylation state of H3K9. Further studies are required to investigate the role of PER2 in the regulation of histone modification during oncogene transformation.
Because the basal mechanism of the circadian clock is well-conserved in many mammalian species, PER2 is assumed to function in human cells in a manner similar to what we observed in murine cells. The present results suggest a newly discovered role for PER2 in the regulation of susceptibility of cancer cells to chemotherapeutic drugs and may contribute to the development of new strategies in the treatment of cancer.

Treatment of animals and cells
Per2 m/m mice in an ICR background and WT ICR mice were maintained under a standardized light and dark cycle. All animal experiments were reviewed and approved by the Animal Care and Use Committee of Kyushu University (Fukuoka, Japan). Fibroblasts were prepared from littermate embryos of WT or Per2 m/m mice using standard techniques. The cells were maintained in Dulbecco's modified Eagle's medium (DMEM; Sigma-Aldrich) supplemented with 10% fetal bovine serum and 0.5% penicillin-streptomycin. For oncogene transformation, embryonic cells were infected with 1 ϫ 10 6 cfu/ml of retroviral vectors (pBABE-puro retroviral vector, RTV-001-PURO, Cell Biolabs, Inc., San Diego, CA) expressing H-ras V12 and SV40LT (Clontech). Transgene-expressing cells were selected with 2 g/ml puromycin (Wako Chemical, Osaka, Japan). The preparation of oncogene-transformed cells of each genotype was performed three times.
To down-regulate ALDH3A1 expression, cells were infected with lentiviral vectors expressing shRNA against the mouse Aldh3a1 gene (pGFP-C-shAldh3a1 Lenti Vector; Origene Technologies, Inc., Rockville, MD). After infection, cells were maintained in DMEM. GFP-expressing cells were selected by sorting using FACS (BD Biosciences). Down-regulation of ALDH3A1 was confirmed by Western blotting.

Determination of cell viability
Cells were seeded at a density of 1 ϫ 10 3 cells/well in 200 l of DMEM in 96-well culture plates. After incubation for 24 h at 37°C, the cells were treated with MTX, GEM, VP-16, VCR, or L-OHP at the indicated concentrations. Cells were also treated concomitantly with 30 M pifithrin-␣ (Wako Chemical), 2 mM NAC (Sigma-Aldrich), or 30 M CB29 (Merck, Darmstadt, Germany). The viability of the cells was determined by an ATP luminescent assay using a Cell Titer-Glo Luminescent Cell Visibility Assay Kit (Promega, Madison, WI).

Immunoprecipitation
Nuclear fractions were prepared from cells after treatment with MTX, GEM, VP-16, VCR, L-OHP, or vehicle for 24 h. The nuclear fractions were diluted (ϳ1 mg/ml) in 25 mM Tris-HCl, pH 8.0, 137 mM NaCl, 2.7 mM KCl, and 1% Triton X-100 supplemented with protease inhibitor mixtures and were then subjected to immunoprecipitation with anti-SV40LT antibodies (sc-58665, Santa Cruz Biotechnology, Inc.). After centrifugation, the amounts of p53 and SV40LT in supernatants and immune complexes were assessed by Western blotting. We also assessed the amounts of p53 in nuclear fractions as input.

Cell-cycle analysis
Single-cell suspension was prepared, and cells were incubated with 0.05 mg/ml propidium iodide for specific DNA staining. The samples were analyzed on an EPICS Elite flow cytometer (Beckman Coulter, Inc.). The total number of cells analyzed from each sample was 10,000.

Role of PERIOD2 in development of chemoresistance
Shimadzu) was used for MS/MS. The multiple-reactionmonitoring transitions and the composition of the mobile phases are listed in Table 1. The amount of each drug was normalized to protein concentrations.

Microarray gene expression analysis
The quality of the extracted RNA was analyzed using an Agilent 2100 Bioanalyzer (Agilent Technologies). The complementary RNA was amplified and labeled using a Low Input Quick Amp Labeling Kit (Agilent). Labeled complementary RNA was hybridized to a 44-K Agilent 60-mer oligo-microarray (Whole Mouse Genome Microarray Kit version 2.0). Probe level data were processed using the robust multiarray analysis algorithm to obtain data at the expression level. This produced a gene expression matrix consisting of 39,427 probe sets and six samples (two groups with three replications). To identify upregulated or down-regulated genes, z-scores and ratios were calculated from the normalized signal intensities of each probe. The criteria for up-regulated genes were set at a z-score of 2.0 or more and a ratio of 3-fold or more. The criteria for down-regulated genes were set at a z-score of Ϫ2.0 or less and a ratio of 0.33 or less. The full data have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus (GEO) database (accession number GSE113242).

Quantitative RT-PCR analysis
mRNA was extracted from WT or Per2 m/m cells by using RNAiso (Takara Bio Inc., Otsu, Japan). The complementary DNA was synthesized by reverse transcription using the Rever-Tra Ace quantitative real-time PCR kit (Toyobo, Osaka, Japan). The real-time PCR assays were performed using THUNDER-BIRDSYBR qPCR mix (Toyobo) and the LightCycler 96 system (Roche Applied Science). The sequences of the PCR primers are shown in Table 2.

Measurement of H 2 O 2
Cells were seeded at a density of 4 ϫ 10 3 cells/well in 78 l of DMEM in 96-well culture plates. Cells were treated with 1 M MTX, 50 M VP-16, or 5 M VCR. The amount of hydrogen peroxide (H 2 O 2 ) in the cells was determined using a ROS-Glo H 2 O 2 assay kit (Promega), following the manufacturer's protocol. After treatment with the chemotherapeutic drugs, 20 l of H 2 O 2 substrate solution was added to the medium and further incubated for 3 h. After incubation, 100 l of detection solution was added to each well of cells and incubated for 20 min at room temperature. The intensity of luminescence was measured for the amount of H 2 O 2 in the cells. The H 2 O 2 amounts were normalized to the number of cells in each well.

DNA methylation analysis
The methylation status of DNA CCGG sites in the upstream and downstream regions of mouse Aldh3a1 gene was measured using an EpiJet DNA methylation analysis kit (Thermo Fisher Scientific) according to the manufacturer's protocol. Briefly, genomic DNA (500 ng) was digested with HpaII and MspI, whichareisoschizomerswithdifferentsensitivitiestoCpGmethylation. When the internal CpG in the 5Ј-CCGG-3Ј tetranucleotide sequence is methylated, cleavage with HpaII is blocked, but cleavage with MspI is unaffected. After digestion, the DNA was subjected to real-time PCR analysis using primers shown in Table 3. The status of DNA methylation was calculated based on Ct value using the following formula.

Immunoprecipitation analysis
One milligram of protein from the nuclear and cytosolic fractions were treated using a cross-link immunoprecipitation kit (Thermo Fisher Scientific). Three hundred microliters of the lysate was precleared with control protein A/G-agarose and then incubated at 4°C for 12 h with protein A/G-agarosebinding anti-PER2 antibodies (1:500, sc-101105; Santa Cruz Biotechnology) or mouse IgG (1:100, sc-66931; Santa Cruz Biotechnology). After washing of the reactants multiple times, the immunoprecipitation lysates were denatured at 90°C for 30 min with 0.1% Triton X-100, 1% SDS, 15% glycerol, 0.25 M Tris, and 5% 2-mercaptoethanol, separated by SDS-PAGE, and transferred to a polyvinylidene difluoride membrane. The immunoprecipitated proteins were detected by Western blotting. Forward Forward 5Ј-AGAAGCCCCTGGCACTCTAT-3Ј Reverse 5Ј-GCAAAGTGGGCACAGTGATG-3Ј Table 3 Primer sets for PCR analysis of DNA methylation The numbers indicate the distance from the putative transcription start site (ϩ1).

Role of PERIOD2 in development of chemoresistance ChIP analysis
Cells were treated with 4% paraformaldehyde for 20 min to cross-link the chromatin, and 250 mM glycine was added as a reaction-stopping agent. Cross-linked chromatin was sonicated on ice, and the nuclear fractions were obtained by centrifugation at 10,000 ϫ g for 5 min. Acetylated or trimethylated lysine residues in histone H3 were immunoprecipitated using antibodies against H3K9Ac (1:500, ab10812; Abcam), H3K4Me3 (1:500; ab8898), H3K9Me3 (1:500; ab8580), H3K27Me3 (1:500; ab6002), PCAF (1:500; ab176316), HDAC1 (1:5000; ab7028), and HDAC2 (1:500; ab7029). DNA was purified using a DNA purification kit (Promega) and amplified by PCR for upstream and downstream regions of the mouse Aldh3a1. The PCR primer sequences are listed in Table 4. The quantitative reliability of PCR was evaluated by kinetic analysis of the amplified products to ensure that the signals were derived from only the exponential phase of amplification. As negative controls, ChIP was performed in the absence of antibody or in the presence of rabbit IgG.

Statistical analysis
Statistical significance for differences among the groups was analyzed by Student's t tests or analysis of variance followed by Tukey-Kramer test. p Ͻ 0.05 was considered significant.

Table 4 Primer sets for PCR analysis of chromatin immunoprecipitation
The numbers indicate the distance from the putative transcription start site (ϩ1).

Genes Primers
Mouse