Time‐dependent changes in proliferation, DNA damage and clock gene expression in hepatocellular carcinoma and healthy liver of a transgenic mouse model

Hepatocellular carcinoma (HCC) is highly resistant to anticancer therapy and novel therapeutic strategies are needed. Chronotherapy may become a promising approach because it may improve the efficacy of antimitotic radiation and chemotherapy by considering timing of treatment. To date little is known about time‐of‐day dependent changes of proliferation and DNA damage in HCC. Using transgenic c‐myc/transforming growth factor (TGFα) mice as HCC animal model, we immunohistochemically demonstrated Ki67 as marker for proliferation and γ‐H2AX as marker for DNA damage in HCC and surrounding healthy liver (HL). Core clock genes (Per1, Per2, Cry1, Cry2, Bmal 1, Rev‐erbα and Clock) were examined by qPCR. Data were obtained from samples collected ex vivo at four different time points and from organotypic slice cultures (OSC). Significant differences were found between HCC and HL. In HCC, the number of Ki67 immunoreactive cells showed two peaks (ex vivo: ZT06 middle of day and ZT18 middle of night; OSC: CT04 and CT16). In ex vivo samples, the number of γ‐H2AX positive cells in HCC peaked at ZT18 (middle of the night), while in OSC their number remained high during subjective day and night. In both HCC and HL, clock gene expression showed a time‐of‐day dependent expression ex vivo but no changes in OSC. The expression of Per2 and Cry1 was significantly lower in HCC than in HL. Our data support the concept of chronotherapy of HCC. OSC may become useful to test novel cancer therapies.

and tivantinib. Sorafenib was also applied in combination with radiation (RT-SOR). 4,5 However, these therapies have substantial side effects. The limited success of these therapies may in part be due to therapy in a phase in which the tumors are not particularly susceptible and determination of the optimal time point for therapies (chronotherapy) may improve their efficacy. 6 Ki67 and γ-H2AX are good markers to predict the response of HCC therapies. Ki67 is one of the most important cell proliferation markers which is increased during the tumor development. It is expressed in the S phase and G2/M phases of the cell cycle. Its expression changes during the day and is regulated by the circadian clock. [7][8][9] Because Ki67 is expressed only in proliferating cells, it is one of the most widely used proliferation markers in cancer cells. 10,11 γ-H2AX is a marker for DNA damage and repair. Upon DNA damage, DNA double-strand breaks (DSBs) are formed which are characteristic for cancer cells due to mutated and unchecked cell cycles. DNA-DSBs are always followed by the phosphorylation of H2AX histone and the formation of a new phosphorylated protein called γ-H2AX which starts the DNA repair process.
After DNA is repaired, γ-H2AX is dephosphorylated. 12,13 γ-H2AX can be used as a marker of radio-sensitivity of cancer and the normal surrounding tissues, their ability to recover from damage and the efficacy of the cellular repair process. This helps to control the dosage, the effectiveness and frequency of radiation therapy in localized target. 12 To evaluate any beneficial effect of chronotherapy, it is necessary to clarify whether cell proliferation and DNA repair mechanisms in HCC cells follow a diurnal pattern and whether this pattern differs from that in healthy liver (HL) tissue. These questions are addressed in the present study in an animal model for HCC, double transgenic c-myc/TGFα mice 14 by immunohistochemical demonstration of Ki67 and γ-H2AX.
Cell cycle and proliferation are closely intertwined with the molecular circadian clockwork and there is increasing evidence that cancer development and progression may be associated with dysfunction or mutation of this molecular clockwork. We have therefore investigated whether the circadian molecular clockwork is altered in HCC as compared to healthy liver tissue. The molecular circadian clockwork comprises clock genes which interact in positive and negative transcription-translation feedback loops. 15,16 Briefly, the expression of Per (Per1 and Per2) and Cry (Cry1 and Cry2) genes is activated by heterodimers of the transcription factors CLOCK/BMAL1 which act as the positive elements in the loop while dimers of PER/CRY form the negative loop. 16 The molecular circadian clockwork ticks in all nucleated cells and governs many physiological processes by controlling the expression of more than 3000, so-called clock-controlled genes.
Finally, we have addressed the question whether results obtained by ex vivo samples are compared to those obtained by in vitro samples such as organotypic slice cultures (OSC) to evaluate whether OSC of liver and HCC represent adequate models to test novel anticancer therapies. Previous studies have shown that OSC which maintain the three-dimensional structure of the tissue and a functional extracellular matrix maintain the circadian rhythms for several days. 17 Usage of OSC would allow much faster and more effective screening of any novel therapeutic strategy than experiments with whole animals.

| Experimental animals
The experiments described in our study were conducted according to accepted standards of humane animal care and were consistent with federal guidelines and Directive 2010/63/EU of the European Union. They were approved by the Regierungspräsidium Darmstadt (Gen. Nr. FU 1067). All experiments were performed with male c-myc/TGFα bitransgenic mice. The animals were generated by crossing homozygous metallothionein/TGFα and albumin/c-myc single transgenic mice in CD13B6CBA background in which hepatocarcinogenesis can be accelerated by zinc in the drinking water.
Food and water containing ZnCl 2 were supplied ad libitum. All animals were kept under normal light-dark (LD) cycle (12:12). The development and growth of HCCs was controlled by MRI as described recently. 18

| Ex vivo investigations
Twelve animals were used for immunohistochemical and 12 animals for real-time PCR analyses. All animals investigated had either single or multiple tumors ( Table 1). The animals were sacrificed at 4 different Zeitgeber time points: ZT00 (light on), ZT06, ZT12 (light off) and ZT18.
For immunohistochemical investigations, the animals (n = 3/ZT) were anesthetized by a mixture of ketamine (100 mg/kg body weight, Rotexmedica, Trittau, Germany) and xylazine ( The results indicate that the efficacy of antimitotic therapies depends on proper timing and that this time dependency should be evaluated further. investigations, the animals (n = 3/ZT) were decapitated at ZT00 (light on), ZT06, ZT12 (light off) and ZT18 and healthy liver tissue and tumors were excised separately, frozen rapidly in liquid nitrogen and stored at −80 C until further use. The experiments during the night were performed under dim red light.

| In vitro investigations of organotypic slice cultures
For in vitro investigations, six animals were sacrificed at 10:00 AM (ZT04) and healthy liver tissue and tumors were freshly excised under T A B L E 1 Number, size and volume of tumors in each mouse investigated ex vivo for qPCR, immunocytochemistry or in slice preparations (OSC) were put in six-well plates filled with 1 mL prewarmed culture medium modified after. 19 The medium consisted of DMEM, supplemented with 10% fetal bovine serum, 100 U/mL penicillin, 0.1 mg/mL strepto-

| Data acquisition
For the quantitative analysis of the number of cells which are positively stained with Ki67 or γ-H2AX, six representative images from each animal and each time point were taken using a confocal laser microscope (Olympus Fluo view SC20, Japan) at 20× objective. For each type of staining, the microscope settings were kept constant.  The results were regarded as significant at P < .05. showed a maximum at midday (ZT06) and second, smaller peak at midnight (ZT18) and a minimum in the morning (ZT00; Figure 1C). two-way ANOVA followed by Sidak's multiple comparisons test ( Figure 1C).
As a marker for DNA-DSBs repair, γ-H2AX immunoreactivity was investigated in the same ex vivo samples. The number of γ-H2AX immunoreactive cells was higher in HCC than in HL ( Figure 1B). In HCC, the number of γ-H2AX immunoreactive cells showed a peak at midnight (ZT18). At ZT18, the difference between HCC and HL was highly significant (P < .01, Figure 1D).
Ki67 and γ-H2AX immunoreactivities were also investigated in OSC of HL and HCC of c-myc/TGFα mice. The OSC were cultured for 24 hours and thereafter fixed at four time points (CT04, CT10, CT16 and CT22). The number of Ki67 immunoreactive cells was higher in HCC than in HL (Figure 2A). Two-way ANOVA showed that the difference between HCC and HL was significant at CT04 and CT16 (P < .05, Figure 2C). The number of γ-H2AX immunoreactive cells was higher in HCC than in the surrounding HL. The differences in the number of γ-H2AX immunoreactive cells between HL and HCC were significant at CT04 and CT10 (P < .05) and highly significant at CT22 (P < .01; Figure 2D).

| Investigations of Clock genes expression in HCC and surrounding HL
Expression of clock genes Per1, Per2, Cry1, Cry2 and Clock was investigated in HCC and HL using qPCR in both, ex vivo samples and OSC.
In addition, expression of Bmal 1 and Rev-erb α was analyzed in the ex vivo samples.
In ex vivo samples, the relative expression of Per1 showed a peak at ZT12 in HL which tended to be different from ZT00 (P = .058) and ZT06 (P = .07). A peak at ZT12 was also observed in HCC which was significantly different from the values at ZT00 and ZT06 (P < .01) and at ZT18 (P < .05; Figure 3A). The relative expression of Per1 did not differ significantly between HL and HCC at all time points investigated (P > .1, Figure 3A). The relative expression of Per2 in HL did not change significantly between day and night ( Figure 3B). The relative expression of Per2 was decreased in HCC as compared to the surrounding HL and this difference was highly significant at ZT12 (P < .001, Figure 3B).
The relative expression of Cry1 in HL showed a maximum at ZT00 which tended to be different from the values at ZT06 and ZT12 (P = .09). In contrast, the HCC showed a peak at ZT18 which was significantly different from the values at ZT12 (P < .01) and ZT06 (P < .05; Figure 3C). The relative expression of Cry1 was lower in HCC than in HL and the two-way ANOVA showed that this difference was significantly different at ZT00 (P < .001) and tended to be significant at ZT18 (P = .09, Figure 3C).
The relative expression of Cry2 in HL showed a peak at (ZT06) which was significantly different from the values at ZT00 (P < .05) and tended to be different from those at ZT18 (P = .06). The relative expression of Cry2 in HCC showed no significant changes during the day (P > .1, Figure 3D). The relative expression of Cry2 was lower in HCC than the HL although the difference between HCC and HL was not significant at all investigated time points as shown by two-way ANOVA (P > .1, Figure 3D).
The relative expression of Clock changed during the day in the HCC and the surrounding HL. The values at ZT12 tended to be different from those at ZT06 (P = .056) and ZT18 (P = .09) in the HL and at ZT18 (P = .08) ( Figure 3E) in the HCC. No significant changes were detected comparing the relative expression of Clock in HL and the HCC at the all investigated time points using the two-way ANOVA test (P > .1, Figure 3E).
The relative expression of Bmal 1 in HL and HCC revealed a peak at (ZT18) which significantly differed from the values at ZT12 (P < .05 and P < .01; respectively). In HCC, ZT18 also showed a significant difference from ZT06 (P < .01). The value at ZT00 was significantly different from ZT06 and ZT12 (P < .01, Figure 3F). The relative expression of Bmal did not differ significantly between HL and HCC at all time points investigated (P > .1, Figure 3F).
The relative expression of Rev-erb α in HL showed a maximum at ZT06 which was significantly different from the values at ZT00, ZT12 and ZT18 (P < .01, Figure 3G). A maximum at ZT06 was also observed in HCC but it did not differ significantly from the other ZTs. No significant differences were detected comparing the relative expression of Rev-erb α in HL and HCC at all time points investigated. In the OSC, the relative expression of Per1, Per2, Cry1, Cry2 and Clock showed a trend to daily variation in HCC and HL but the differences were not significant between day and night (P > .1, Figure 4). The relative expression was significantly higher in HCC than in HL for Per1 at CT04 (P < .05), for Cry2 at CT22 (P < .001) and for Clock at CT22 (P < .01; Figure 4, two-way ANOVA followed by Sidak's multiple comparisons test). In HCC, the number of Ki67 immunoreactive cells showed a maximum at midday (ZT06) and second, smaller peak at midnight (ZT18).

| DISCUSSION
These results are in agreement with a study by You et al 22 who showed that mammary tumors had two daily growth rate peaks, one minor at mid-sleep and one major peak at mid-activity. Two proliferation peaks were also observed in other fast-growing tumors. 9,23 In an early study, fast and slow growing hepatomas showed two mitotic activity peaks, one during the light phase and the other during dark phase. 24 It is well known that the expression of cell cycle regulators which either promote or inhibit cell proliferation are affected by the circadian clockwork. In mammary tumor, the expression of some known clock-controlled cell cycle genes which promote cell proliferation, such as CycD1 and C-Myc as well as cancer cell mitosis showed two peaks during the day, one at mid-day and the other at the midnight, 22,25 whereas only one peak was found in healthy tissue.
A highly relevant result of our ex vivo studies was that the difference in number of proliferating Ki67 immunoreactive cells between the HCC and the HL was significant at ZT06 (midday) and at ZT18 (midnight). Since it is well known that highly proliferating cells become more sensitive to DNA damage with cancer therapies, 26,27 we conclude that midday and midnight may be considered as optimal time points to apply antimitotic therapies to HCC with minimum side effects on the surrounding HL. This assumption now needs to be confirmed in further experiments. Since the two peaks occurred at midday and midnight, the findings in nocturnal species (mouse) might be easily transferred to diurnal species (eg, primates).
We found in human tumors of the urinary bladder, breast, lung and colon. 30,31 As shown by our study the number of γ-H2AX immunoreactive cells showed a trend to daily variation in the HCC and the surrounding HL. The mechanism behinds these changes remain to be elucidated.
One possibility is that ATM ! Chk2 signaling pathway which is mainly activated by double-strand breaks is subjected to the circadian rhythm of the clock genes. 32 Other publications also reported that cellular responses to DNA damage and repair process are influenced by the circadian rhythm. XPA, one of the DNA repair protein, was shown to be controlled by the circadian clock in the mouse brain, liver and skin. 33 Our data showed no changes in the expression of the Bmal 1, Rev-erbα and Clock gene in the HCC as compared to the surrounding HL. The same results were reported also for Bmal1 and Clock in the human HCC. 21,45 The reason for this is unclear and further studies are needed to clarify whether downregulation of some clock genes is associated with more advanced cancer stages. 46 The Per2 gene appears to be a functionally more relevant in the mammalian circadian clock than the Per1 gene. 47

Lower expression of
Per2 was shown to elicit more profound effects on the tumor growth, both in vitro and in vivo than lower expression of Per1. 25 In gastric cancer, the Per2 expression was reported to be a potential prognostic factor and lower expression of Per2 might help identify gastric cancer patients with a poor prognosis. Also, in chronic lymphocytic leukemia, the ratio between PER2 and CRY1 is suggested to be a prognostic marker that predicts the survival outcomes of patients. 43 In line with these results, OSC is a model which is made from primary tissue and maintains the three-dimensional structure as well as the extracellular matrix. 48 OSC of normal liver was shown to be viable in culture conditions for several days and to keep a robust circadian rhythm. 17,49 The present study with OSC which includes HCC and the surrounding HL showed that the number of Ki67 and γ-H2AX immunoreactive cells was much higher in the HCC than in the HL as was also observed in ex vivo samples. The number of Ki67 immunoreactive cells showed two peaks which occurred at CT04 and CT16 and thus slightly differed from the time points which were observed in the ex vivo samples (ZT06 and ZT18). Notably, also the expression pattern and amplitudes of the clock genes differed between the OSC and ex vivo samples. This difference may be due to the lack of entrainment signals which under in vivo conditions are transmitted derived from the master oscillator of the circadian system, the suprachiasmatic nucleus, to the periphery via neuronal pathways or the blood stream. It is well known that temperature can act as physical synchronizer and resetting cue for circadian peripheral oscillators. 50 The fact that the temperature was kept constant in our culture experiments may contribute to the differences between OSC and ex vivo samples, although previous studies using identical, constant culture conditions 17,49 have shown that OSC kept a robust circadian rhythm under constant temperature. The stress generated by the dissection process and the initiation of the culture may also contribute to these observed differences. The results suggest that OSC may be helpful to establish therapeutic strategies, but it remains to be established whether are suited to determine the optimal time points of antimitotic therapies.
In conclusion, our study with an experimental mouse model for hepatocellular carcinoma showed significant differences in proliferation rate as well as DNA damage and repair mechanisms between the HCC and the HL. The observation that the proliferation rate in the HCC showed two distinct peaks indicates that the efficacy of antimitotic therapies depends on the timing. Future studies in oncology should consider this time dependency and determine the optimal time point(s) for anticancer therapy for each tumor entity. Since γ-H2AX expression was higher in the HCC than in the HL, it can be used as a marker to determine HCC sensitivity to the antimitotic treatment.
Since expressions of Per2 and Cry1 were significantly lower and had different daily variation patterns in the HCC and the HL, these two clock genes might be closely linked to development and growth of the HCC. Overall, OSC may become a suitable model to develop and test anticancer strategies; however, future studies are needed to prove whether they could substitute for whole animal studies with regard to determination of the optimal time points for therapy.