Period 2 Mutation Accelerates Apc^Min/+ Tumorigenesis

Central and peripheral clocks generate self-sustained circadian rhythms of about 24 hours, which coordinate physiologic processes with the rhythmically changing environment (1-3). The master circadian pacemaker, in the suprachiasmatic nuclei of the brain, is synchronized by the daily light-dark cycle, which in turn synchronizes the organism and peripheral clocks in each cell (4). The molecular mechanisms of circadian oscillation in the suprachiasmatic nuclei and peripheral cells are based on a negative transcriptional-translational feedback loop generated by at least nine core clock genes (1). Individual cells within tissues and even cancer cells have circadian clocks composed of the same set of clock genes (2, 5). Circadian clockworks in peripheral tissues coordinate physiologic processes through tissue-specific clock-controlled genes, such as thymidylate synthase, p21, and Wee-1, which gate DNA replication and mitosis and other growth regulators, such as vascular endothelial cell growth factor (2, 5-9). The proliferation of intestinal epithelial cells and the rate of cell migration up the crypt-villus axis are gated throughout each day by the circadian clock (10, 11).

Disruption of circadian rhythms deregulates cell proliferation and increases cancer risk. Period 2 (Per2) is a core circadian clock gene, which also acts as a negative growth regulator and tumor suppressor (12, 13). Mice with homozygous mutation in the Per2 circadian clock gene (Per2^m/m) display altered circadian behavioral rhythms, deregulation of c-Myc and cyclin D RNA levels and daily rhythms, and spontaneous tissue hyperplasias and tumors. Following γ-radiation, these mice have a shortened survival, an abnormal DNA damage response, and increased incidence of lymphoid neoplasia, teratomas, and salivary hyperplasias (12). A recent screen of all mutations in several human breast cancers has identified mutations in the PER2 gene (14). Altered PER2 levels in breast cancers have been associated with altered methylation of the PER2 promoter (15).

Increasing evidence indicates that disruption of circadian rhythms significantly increases the incidence of epithelial cancers. Higher rates of breast, colorectal, and endometrial cancer have been found in shift workers with presumed circadian disruption (16-18). Shift work has recently been listed by the IARC as a probable carcinogen (19). Colorectal cancer is a major cause of cancer death in the United States (20). Mutation in the familial adenomatous polyposis coli (APC) gene and deregulated Wnt-APC-β-catenin signaling pathway contributes to the development of spontaneous and hereditary colorectal cancer. Heterozygous mutations in the germline APC gene causes familial adenomatous polyposis with colon, and sometimes small intestinal, polyps, and, eventually, cancer, following the inactivation of the remaining wild-type APC allele. Inactivation of both APC alleles often occurs in sporadic colorectal adenomas that progress to cancer (21-23). The Min (multiple intestinal neoplasia; Apc^Min/+) mouse, with a heterozygous mutation of the Apc gene, is a useful model for colorectal cancer (24-26). Loss of Apc function mimics the effect of constitutive Wnt stimulation, resulting in increased β-catenin levels, continuous β-catenin/T-cell factor (TCF) signaling, and increased β-catenin–dependent gene expression (e.g., cyclin D). This leads to excess intestinal epithelial cell proliferation and adenoma formation in the intestine and colon, in addition to extraintestinal proliferation abnormalities (26).

Because of our interests in how the circadian clock influences cellular proliferation and antiproliferative cytotoxic and targeted therapeutic drug responsiveness, we have investigated the effects of Per2 on colon cancer cell proliferation and intestinal and colonic neoplastic changes.

Because activation of the β-catenin pathway is key to the promotion of intestinal and colonic cell proliferation and tumorigenesis, we asked whether alteration in PER2 expression modulates β-catenin levels and the β-catenin signaling pathways in human colon cancer cell lines. We used small interfering RNA (siRNA) to down-regulate PER2 expression in HCT116 (wild-type APC, mutant β-catenin) and SW480 (mutant APC, wild-type β-catenin) colon cancer cell lines in vitro. In HCT116 cells, with lower endogenous β-catenin protein levels, down-regulation of PER2 increases β-catenin protein concentration (Fig. 1A and B). In SW480 cells, with high endogenous β-catenin levels, down-regulation of PER2 further elevates β-catenin protein (Fig. 1C and D). Following down-regulation of PER2, cyclin D protein, a β-catenin target gene associated with cell proliferation, increases in SW480 and HCT116 colon cancer cell lines (Fig. 1). This increase in β-catenin and cyclin D, which occurs with PER2 down-regulation, accelerates cell proliferation in both SW480 and HCT116 cell lines (Fig. 1E and F).

When β-catenin protein induction is prevented by its siRNA-mediated down-regulation, PER2 down-regulation no longer increases cyclin D protein in HCT116 (Fig. 2A and C) and SW480 (Fig. 2B and D) cells and this prevents the acceleration in cell proliferation (Fig. 2E and F). Partial down-regulation of β-catenin alone (e.g., 50% reduction), in each of these cell lines under these conditions, has no effect on proliferation compared with control siRNA–treated cultures (data not shown), similar to previous reports with β-catenin down-regulation (27).

Cyclin D expression can be induced by β-catenin/TCF–dependent signaling or through other β-catenin–independent pathways. The effect of PER2 down-regulation and PER2 overexpression on β-catenin/TCF transcriptional activation was investigated using the TOPFLASH luciferase assay. PER2 overexpression, compared with control vector used for overexpression, decreased β-catenin/TCF reporter activity, whereas PER2 down-regulation, compared with control vector used for down-regulation, increased β-catenin/TCF reporter activity (Fig. 3). Therefore, the effect of PER2 on cyclin D seems, at least in part, dependent on β-catenin transcriptional activation.

These effects of PER2 down-regulation on increasing both β-catenin and cell proliferation seem not to be restricted to colon cell lines because similar effects are also seen in HeLa cells (Supplementary Fig. S1).

Because down-regulation of Per2 expression in colon cancer cell lines increases β-catenin levels and β-catenin target gene expression, we asked whether Per2^m/m mice with a homozygous inactivation of Per2 would show a de novo increase in intestinal mucosa β-catenin protein at a time of day when wild-type PER2 protein levels are near their daily maximum (mid-activity/mid-dark phase in mice). Small intestinal mucosa β-catenin protein levels in 15-week-old Per2^m/m female mice are 2-fold higher compared with levels in small intestinal mucosa of age- and sex-matched wild-type mice (P < 0.05; Fig. 4A and B). This increase in small intestinal mucosa β-catenin in Per2^m/m mice is also associated with an increase in cyclin D protein (P < 0.05) compared with wild-type mucosa (Fig. 4A and B).

Because increased β-catenin expression is often associated with polyp formation, we examined Per2^m/m mice for intestinal polyps. Fifteen-week-old Per2^m/m mice show an increase in the numbers of colonic polyps compared with wild-type mice [Per2^m/m, 0.6 ± 0.16 (range, 0-2/mouse) versus 0 polyps in wild-type mice; P = 0.012; n = 18-32 mice/genotype], which occurs at a frequency of 31.3% (Per2^m/m, 10/32 versus wild-type 0/18 affected mice/total mice; P = 0.008; Fig. 4C). Colon polyp numbers and frequency are not increased in Per2^m/+ heterozygous mice. Small intestinal polyps were not seen in 15-week-old wild-type, Per2^m/+, or Per2^m/m mice.

Because alteration in Per2 modulates β-catenin and β-catenin target gene expression, cell proliferation, and polyp formation, we crossed Per2^m/m mice with Apc^Min/+ mice to determine whether the Per2 mutation would accelerate intestinal and colonic tumorigenesis in vivo. We compared the number and size of small intestinal polyps and number of colonic polyps in 10-week-old Apc^Min/+ mice with mice homozygous (Apc^Min/+Per2^m/m) and heterozygous (Apc^Min/+Per2^m/+) for the Per2 mutation. Compared with age-matched 10-week-old Apc^Min/+ mice, Apc^Min/+Per2^m/m mice have nearly twice the number of small intestinal polyps (83.9 ± 9.5 versus 46.7 ± 4.4; P < 0.001; Fig. 4A). This is the result of an increase in the number of small-sized (<1 mm) small intestinal polyps in Apc^Min/+Per2^m/m mice compared with Apc^Min/+ mice (51.7 ± 7.3 versus 21.3 ± 2.1; P < 0.001), whereas the number of larger-sized (>1 mm) small intestinal polyps failed to differ (32.3 ± 5.3 versus 25.4 ± 2.9; P = 0.122). Apc^Min/+Per2^m/m mice show a significant increase (2- to 3-fold) in total polyps in the second, third, and fourth segments of the small intestine (P < 0.01), whereas polyp numbers failed to differ in the proximal small intestine compared with Apc^Min/+ mice (Fig. 4B).

The number of total colonic polyps is also 2- to 3-fold higher in Apc^Min/+Per2^m/m mice compared with Apc^Min/+ mice (4.2 ± 1.3 versus 1.6 ± 0.3; P = 0.025; Fig. 4C). These effects require Per2^m/m homozygosity. Apc^Min/+Per2^m/+ mice that are heterozygous for the Per2 mutation fail to show differences in total small intestinal polyps (43.0 ± 5.5 versus 46.7 ± 4.4; P = 0.31; Fig. 4A) or colonic polyp numbers (2.7 ± 1.1 versus 1.6 ± 0.3; P = 0.14; Fig. 4C) compared with Apc^Min/+ mice. Wild-type, Per2^m/+, and Per2^m/m mice at this age do not show small intestinal polyps.

By histologic examination, the frequency of small intestinal microadenomas is higher in 10- to 15-week-old Apc^Min/+Per2^m/m mice than in Apc^Min/+ mice (88.9% versus 42.3%; P < 0.001). The majority of these microadenomas in Apc^Min/+ and Apc^Min/+Per2^m/m mice show low-grade dysplasia. Severe dysplasia/carcinoma in situ is infrequent in both Apc^Min/+ and Apc^Min/+Per2^m/m mice and did not differ significantly (2 of 26 versus 4 of 27 mice; P = 0.42).

The extraintestinal changes seen in the Apc^Min/+ mouse include progressive anemia, splenomegaly, and extramedullary hematopoiesis (EMH; ref. 26). At 10 weeks of age, Apc^Min/+ mice show lower hematocrits (46.3 ± 1.1% versus 54.6 ± 1.2%; P < 0.001) and larger spleens (higher spleen index, 0.532 ± 0.045% versus 0.375 ± 0.013%; P < 0.006) compared with wild-type mice (Supplementary Table S1). Ten-week-old Apc^Min/+Per2^m/m mice show acceleration of this anemia (40.6 ± 1.8% versus 46.3 ± 1.1%; P < 0.004) and greater splenomegaly (0.900 ± 0.119 versus 0.532 ± 0.045 spleen index; P < 0.002) compared with Apc^Min/+ mice. Hematocrits and the spleen index of Apc^Min/+Per2^m/+ mice did not differ from Apc^Min/+ mice Fig. 5.

We have previously shown that the increase in spleen weights in Apc^Min/+ mice is associated with increased splenic EMH, which progresses with increasing age and anemia (26). Apc^Min/+ mice at 10 weeks of age already have splenomegaly and EMH (Supplementary Table S1; Supplementary Fig. S2). Apc^Min/+Per2^m/m mice, compared with Apc^Min/+ mice, show greater splenic EMH by qualitative histologic scoring (2.97 ± 0.20 versus 1.97 ± 0.21 score; P < 0.001), consistent with the increase in spleen weight. This EHM is composed of subcapsular infiltration of cells of heterogeneous morphology consisting of clusters of myeloid and erythroid precursors at various stages of maturation, along with an increase in megakaryocyte numbers (26). These cellular infiltrates fail to stain with B-cell (PAX-5) or pan T-cell (CD3) markers (data not shown). Apc^Min/+Per2^m/m and Apc^Min/+ mice also show a decrease in content of mature, nonnucleated erythrocytes in the spleen. Apc^Min/+Per2^m/+ mice, heterozygous for the Per2 mutation, show similar EMH scores to Apc^Min/+ mice (2.10 ± 0.21 versus 1.97 ± 0.21 score; P = 0.28). These changes in spleen EMH and weights are not seen in Per2^m/m or Per2^m/+ mice.

With increasing age, the hematocrits of Apc^Min/+Per2^m/m mice remain persistently lower than age-matched Apc^Min/+ mice (Supplementary Fig. S3; P < 0.01). Age-dependent progression of anemia in Apc^Min/+Per2^m/+ mice fails to differ from that in Apc^Min/+ mice.

We show that PER2 circadian core clock gene modulates β-catenin signaling in colon cancer cell lines, and other cell types, in vitro. We find that down-regulation of PER2 in cell lines increases β-catenin protein levels and the β-catenin target protein cyclin D and accelerates cancer cell proliferation. PER2-mediated effects apparently require increased β-catenin expression because down-regulation of β-catenin prevents the PER2-associated increase in cyclin D and acceleration of cancer cell proliferation. The effects of PER2 on β-catenin also occur in vivo. Per2^m/m mice, with homozygous inactivation of Per2, show increased β-catenin and cyclin D protein levels in the mucosa of the small intestine and an increase in colonic polyp formation (number and frequency). Furthermore, we find that genetic inactivation of Per2 in vivo, by crossing Per2^m/m mice with Apc^Min/+ mice, accelerates intestinal and colonic tumorigenesis, as well as the extraintestinal manifestations, in the Apc^Min/+ mouse model. The effects of PER2 on β-catenin are not restricted to intestinal cell lines and therefore may be relevant to tumorigenesis in other tissues.

We do not yet know how PER2 modulates β-catenin protein levels. PER2 protein is, however, known to function as a transcription factor, which dimerizes with other clock proteins (CRY1 and CRY2) that then modulates transcription through interactions with BMAL1/CLOCK protein heterodimers at E box sites. Both PER2 and β-catenin protein turnover and nuclear translocation are affected by their phosphorylation and ubiquitination (28-30).

Per2 has tumor suppressor properties in mice (12). Decreased PER2 RNA levels have been reported in some human acute leukemias (31). Intratumoral injection of Per2 slows murine tumor growth in vivo (32). PER2 overexpression in cancer cell lines in vitro inhibits growth and causes cell cycle arrest, increased apoptosis, and decreased clonogenicity, supporting the negative growth-regulatory properties of Per2 (31, 33). We show accelerated cell growth following PER2 down-regulation in vitro and increased intestinal and colonic tumorigenesis following Per2 homozygous mutation in vivo, further supporting the role of Per2 as a tumor suppressor. We find in colon cancer lines, intestinal mucosa, and Apc^Min/+-mediated intestinal and colonic polyp formation that Per2 mediates these effects, at least in part, through modulation of β-catenin and β-catenin signaling. We do not yet know at what level or how Per2 and β-catenin pathways interact.

Per2 mutation can affect the expression of Per1, in addition to other clock genes (Cry1 and Bmal1). Period 1 (Per1) clock gene plays an analogous role in circadian time keeping to Per2, both acting in the negative feedback loop of the clock. Although homozygous Per1 mutation has, however, not been associated with a tumor-prone or proliferative tissue phenotype in mice, decreased PER1 RNA levels have been reported in some lung, breast, and endometrial cancers compared with adjacent normal tissues (31, 34, 35). One study of human colorectal cancer reports decreased PER1 RNA levels in high-grade colon cancers compared with patient-matched normal tissues (36). In vitro PER1, similar to PER2, can act as negative growth regulator (31, 34). The time of day of tissue sampling has not been qualified or examined in these clinical studies of clock gene expression. This may be critical because the expression of these clock genes in normal tissues, and even in cancers, varies rhythmically throughout each day. Therefore, specifying the time of day of tissue sampling is essential to defining alterations in clock gene expression in preclinical and clinical tumor samples.

We conclude that the Per2 gene product suppresses intestinal and colonic tumorigenesis, in part, by down-regulation of β-catenin and β-catenin signaling pathways. Thereby, the Per2 gene product may be an effective and novel target for colon cancer prevention and/or control.

Human cancer cell lines HCT116, SW480 (colon), HeLa (cervical), and HEK293 (kidney; American Type Culture Collection) were cultured in RPMI 1640 containing 10% fetal bovine serum at 37°C with 5% CO₂.

Cells were plated in six-well plates the day before to reach ∼30% confluency by the time of transfection. Cells were washed with PBS and incubated in 1.5 mL of fresh medium with 10% fetal bovine serum but without antibiotics. siRNA oligo (100 pmol) diluted in 250 μL Opti-MEM (Invitrogen) was mixed with 5 μL of Lipofectmine 2000 diluted in 250 μL Opti-MEM. The mixture was added drop wise to the cells after being incubated at room temperature for 15 to 30 min. The knocked down effect was examined 48 to 72 h later. The GFP control siRNA sequence is AAGGCAAGCUGACCCUGAAdTT. The sequence of human PER2 siRNA is GCGUUACCUCUGAGCACAUdTT. siRNA sequence against human β-catenin is GCUGGUGGAAUGCAAGCUUdTT.

Seventy-two hours after siRNA treatment, cells were trypsinized and replated in 96-well plates. Cell growth was measured daily by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay. Briefly, after removing the medium, 100 μL of serum-free medium containing 1 mg/mL MTT (Sigma) were added and incubated at 37°C for 3 h. Then, 100 μL DMSO was added to each well to dissolve the precipitate after removal of MTT/medium. Absorbance was read at 595 nm by microplate reader (Bio-Rad).

Cultured cells were washed with PBS and incubated in NP40 buffer [0.5% NP40, 150 mmol/L NaCl, 50 mmol/L Tris-Cl (pH 7.4) with protease inhibitors] to isolate total protein. Small intestinal mucosa was processed by mechanical homogenization in homogenization buffer [200 mmol/L Tris-HCl (pH 7.4), 20 mmol/L β-mercaptoethanol, 100 mmol/L NaF] with protease and phosphatase inhibitors to isolate total soluble protein. Proteins were then separated by standard SDS-PAGE using Mini-gel system (Bio-Rad) and transferred to nitrocellulose membranes by semidry transfer apparatus (Bio-Rad). Membranes were incubated in blocking buffer (5% fat-free dry milk in TBS-Tween 20 buffer) at room temperature for 1 h. Membranes were then incubated with optimal diluted primary antibodies (1:200-1:5,000) overnight at 4°C. Membranes were washed with TBS-Tween 20 and then incubated in horseradish peroxidase–conjugated secondary antibodies for 1 h at room temperature. After being washed with TBS-Tween 20 thrice, membranes were incubated in TLC chemiluminescence reagent (Bio-Rad) and exposed to X-ray film. Anti-PER2 antibody was from Alpha Diagnostic International and antibodies against β-catenin, cyclin D1, c-MYC, α-tubulin, and glyceraldehyde-3-phosphate dehydrogenase were from Santa Cruz Biotechnology and anti-actin antibody was from Sigma. Films are scanned, bands are quantitated by densitometry (ImageJ software), and samples are expressed as the ratio to control protein for each sample and normalized to a standardized sample on each gel.

Human HEK293 cells were used due to their high transfection rates for optimal signal intensity and assay. Cells were transfected with 0.5 μg of TOPFLASH-Luc and 0.1 μg of cytomegalovirus-β-galactosidase plasmid in combination of 0.5 μg of pcDNA, pcDNA-Per2, 100 pmol of control siRNA, or 100 pmol of PER2 siRNA. Forty-eight hours after transfection, cells were lysed by reporter assay buffer (Promega). Luciferase activities and β-galactosidase activities were determined by Luciferase Assay System and Beta-Glo Assay System (Promega), respectively, according to the manufacturer's protocol. The ratio of TOPFLASH to β-galactosidase activity represented β-catenin/TCF activity (37).

Apc^Min/+ mice in the C57BL/6J background (25) were maintained by breeding male mice heterozygous for the Min allele to wild-type C57BL/6J females and care was in accordance with institutional guidelines (38). Apc^Min/+ mice were maintained as heterozygotes, as homozygosity in the Min allele is embryonic lethal. Homozygous Per2^m/m mice (Per2^tm1/Brd/J; The Jackson Laboratory) in the C57BL/6 background were crossed to Apc^Min/+ mice to generate Apc^Min/+ mice with a homozygous mutation of the Per2 gene. The presence of the Min allele and the mutation of the Per2 gene were verified by PCR genotyping using Min- and Per2-specific primers, respectively (The Jackson Laboratory procedures) using genomic DNA from tail snips or ear punches as templates. Both strains were originally purchased from The Jackson Laboratory but were maintained as breeding colonies at the Mouse Core Facility of the Center of Colon Cancer Research of the University of South Carolina Center. The mice were housed four to five per cage, provided free access to food and water, and maintained on a 12-h light/12-h dark schedule. All animal experiments and procedures were approved by the University of South Carolina Center and Veterans Affairs Institutional Animal Care and Use Committee. Equal numbers of male and female 10-wk-old mice of six different genotypes (C57BL/6 wild-type, Per2^m/+, Per2^m/m, Apc^Min/+, Apc^Min/+Per2^m/+, and Apc^Min/+Per2^m/m) were humanely euthanized for tissue procurement (n = 20-39 mice/genotype). All genotypes were confirmed on sacrifice in the Veterans Affairs Laboratory. Complete blood parameters of peripheral blood from retroorbital sinus, before sacrifice, were determined using an automated blood cell analyzer (VetScan) at the Mouse Core Facility. Body and spleen weights were recorded and the spleen index was calculated (spleen weight/body weight × 100). For polyp enumeration, small intestines (divided into four segments) and colons (single segment) were removed, flushed with PBS, dissected longitudinally, flattened, and fixed in 10% buffered formalin overnight. Segments were stained with 0.5% methylene blue in saline. Intestinal and colonic polyps were enumerated under a dissecting microscope and small intestinal polyps were classified as small (≤1 mm) or large (>1 mm). Serial complete blood parameters were recorded every 2 wk in mice 15 to 23 wk of age (n = 16-27/genotype). For small intestinal mucosal protein studies, 15-wk-old wild-type and Per2^m/m female mice (n = 5/genotype) were euthanized in the middle of the activity/dark phase. Small intestines were harvested and rapidly flushed with PBS. The first two thirds of the intestine was opened longitudinally and mucosa was separated by mechanical dislodgement (glass slide) and stored at −80°C.

Spleens (10-wk-old mice) were fixed in buffered formalin and embedded in paraffin blocks. Sections (5 μm) were cut and histologically assessed on H&E-stained slides. Sections mounted on positively charged slides were stained with antibodies against CD3 pan T cell (Vector Laboratories) and PAX-5 pan B cell (Biocare Medical) and visualized as previously described (2, 5). Slides were examined in an Axioskop (Zeiss) compound microscope and representative areas were captured with a Zeiss Axiocam digital camera using Axiovision 4.5 software. Spleens were scored for EMH that we previously reported as characteristic of Apc^Min/+ mice (26). This subcapsular infiltrative process is characterized by cells that lack CD3 and PAX-5 staining, which are morphologically heterogeneous and consistent with hematopoietic precursors along with increased numbers of scattered megakaryocytes. EMH was scored by depth of infiltration from the splenic capsule (mild, moderate, severe) and extent (focal, diffuse) of involvement of the spleen (0 = none; 1 = mild, focal; 2 = moderate, focal; 3 = moderate, diffuse; 4 = severe, diffuse).

The distal third of the intestine was opened longitudinally and rolled into a Swiss roll, formalin fixed, and paraffin embedded and 5-μm sections were cut and histologically assessed on H&E-stained slides. Microadenomas were counted and dysplasia (low grade versus high grade) was graded.

For each numerical value, mean and SEs were calculated and graphed. Average and standard values were computed across gene mutations and manipulations. Mean comparisons across gene mutations/alterations were assessed through ANOVA, Kruskal-Wallis test, or t test among two genotypes. Repeated measures analysis of hematocrit levels at 10, 15, 18, and 21 wk of age were compared across gene mutations using Proc Mixed in SAS. Significant differences are set at α = 0.05. All statistical tests were two sided.

No potential conflicts of interest.

We thank Tia Davis for the assistance with animal breeding.