Rhythmic profiles of cell cycle and circadian clock gene transcripts in mice: a possible association between two periodic systems

In mammals, the circadian time-keeping system shapes the rhythms of many physiological and behavioral functions. In rodents, a central circadian pacemaker located in the hypothalamic suprachiasmatic nucleus (SCN) is capable of sensing time cues from the external surroundings and internal body milieu. This major pacemaker is assumed to synchronize the peripheral secondary tissue's clocks (e.g. liver, heart, olfactory bulbs, pineal gland, etc.), utilizing neuronal and humoral pathways (Buijs and Kalsbeek, 2001; Gachon et al., 2004; Schibler and Sassone-Corsi, 2002). Consequently, physiology and behavior can be adjusted to the habitat conditions (Kalsbeek and Buijs, 2002). Virtually the same genes are involved in the molecular mechanism of the circadian clock in the SCN and peripheral tissues (Dibner et al., 2010; Mazzoccoli et al., 2012; Rana and Mahmood, 2010). Yet, the resulting phenotype of the clock tends to be tissue specific (Dibner et al., 2010; Rana and Mahmood, 2010; Yoo et al., 2004). The clock mechanisms are based on transcriptional–translational feedback loops of positive and negative effects. The CLOCK/BMAL1 protein complex constitutes the basic positive limb of the feedback loop, whereas the PER/CRY complex forms its negative limb. Other gene products were reported to shape the phenotype of the circadian clocks (Dardente and Cermakian, 2007; Dibner et al., 2010; Ko and Takahashi, 2006; Welsh et al., 2010; Yagita et al., 2001).

The cell division cycle is a time-dependent sequence of stages (G1, S, G2 and M) that are based on interlocking loops of positive and negative elements (resembling the circadian clock mechanism) (Alberts et al., 1994).

The overall rhythm pattern of the cell cycle is determined by the tissue's circadian clock and modified by developmental programming, illumination and metabolic (feeding) cycles (Borgs et al., 2009; Kondratov et al., 2007). In rodents, for example, regularly scheduled meal hours synchronize the timing of S-phase in the cornea, bone marrow, lymphoid system and intestine (Canaple et al., 2003; Li et al., 2009). The rhythm of the cell cycle stages in many tissues (bone marrow, lymphoid system, alimentary tract epithelium, skin, etc.) oscillates in a circadian but tissue-dependent manner (Abrahamsen et al., 1993; Bjarnason and Jordan, 2000; Canaple et al., 2003; Haus et al., 1983; Khapre et al., 2010; Kondratov et al., 2007; Matsuo et al., 2003; Potten et al., 2002; Smaaland et al., 2002). The different responsiveness of various cell types to ex-tissue signals further amplifies the diversity in the rhythmic pattern of the cell cycle between tissues (Edgar et al., 2001; Edmunds, 1992).

Cell proliferation timing is regulated via the expression of key cell cycle genes. Clock gene proteins (PER1/2 and CLOCK-BMAL complex) are assumed to regulate the cell proliferation timing through some cell cycle-related genes, such as Wee1, c-Myc and Ccnd1 (Fu et al., 2002; Gery et al., 2006; Khapre et al., 2010; Rana and Mahmood, 2010; Sahar and Sassone-Corsi, 2007). For example, extensive translation of PER1/2 leads to inhibition and even arrest of the cell cycle in colon cancer cells (Khapre et al., 2010). In contrast, inhibition of Per2 transcription in vivo almost doubled the daily amplitude of the tumor growth rhythm in breast cancer patients (Yang et al., 2009).

In this study, mRNA patterns of two clock genes (Clock and Per2) and three cell cycle (G1 and S)-related genes (Ccnd1, Ccne1 and Pcna) were monitored in two tissues of BALB/c inbred mice strain: esophagus, which is considered to have a relatively high proliferation index (Muskhelishvili et al., 2003; Scheving et al., 1979), and liver, with a relative quiescent status (Counts et al., 1996; Khapre et al., 2010; Michalopoulos and DeFrances, 1997). The choice of Clock and Per2 was based on studies that include these genes as regulators in the gating of the cell cycle stages (G1 to S transition) (Johnson, 2010; Lahti et al., 2012). Locomotor activity was monitored in order to establish the circadian organization of the SCN's master pacemaker (Davis and Viswanathan, 1996; Earnest et al., 1999; Reebs and Maillet, 2003).

BALB/c male mice, 2.5 months old, were housed in cages with food and water freely available. All mice were adapted to the experimental conditions for 3 weeks prior to the experiment. For the molecular studies, 16 mice were housed at a density of 3–4 per cage and for locomotor activity monitoring, a pair of mice was housed in a cage. The procedures followed the guidelines of Tel-Aviv University's Animal Care Committee.

Mice were exposed to a 12 h light:12 h dark cycle, with lights on at 06:00 h (‘normal’ conditions). Light intensity, at cage level, was 700 lx.

Every 6 h, starting at Zeitgeber time (number of hours and minutes past light onset) ZT 2 (08:00 h) and over a period of 24 h, 3–4 mice were killed by cervical dislocation. The esophagus and the right lobe of the liver were dissected from each mouse, rinsed with saline and immersed separately in RNAlater stabilization reagent (Ambion, Austin, TX, USA).

Total RNA was isolated from both tissues with the PARIS kit (Ambion) following the manufacturer's instructions. DNA and protein residues were removed by applying the RNase-Free DNase Set (Qiagen, Valencia, CA, USA), followed by the RNeasy MiniElute kit (Qiagen) according to the manufacturer's instructions.

Optical density in each sample was measured at 260, 230 and 280 nm wavelengths using a GeneQuantPro spectrophotometer (Amersham BioSciences AB, Uppsala, Sweden), in order to assess the RNA concentration (A₂₆₀) and the levels of DNA and protein contamination (A₂₆₀/A₂₃₀ and A₂₆₀/A₂₈₀, respectively).

The same amount of total RNA from each tissue of each mouse was reverse transcribed to single stranded cDNA using the Ommniscript kit with RNase inhibitor (Qiagen), according to the manufacturer's protocol. Briefly, the reaction (50 μl total volume) mix containing 10× RT buffer, deoxyribonucleotide triphosphate, random primers, MgCl₂, Omniscript RT enzyme, RNase inhibitor and RNA samples (including blanks) was incubated at 37°C for 60 min in a PTC-100 Thermal Cycler (MJ Research, Waltham, MA, USA).

In each of the tissues, mRNA levels of the target genes were assessed with specific TaqMan Assays-on-Demand Gene Expression (Applied Biosystems, Foster City, CA, USA) (Table 1). The transcript levels were determined by quantitative RT-PCR using ABI PRISM 7000 (Applied Biosystems). Table 1 details the primer sequences and amplicon length of each cDNA. Amplifications were performed in 20 μl volume reactions containing: 1 μg of cDNA, Universal TaqMan master mix (Applied Biosystems) and a primers/probe mix (Applied Biosystems) according to the manufacturer's instructions. The qRT-PCR reactions on each of the cDNA samples were run in triplicate. The 2^−ΔΔCt technique (Livak and Schmittgen, 2001) was applied in order to retrieve the relative mRNA level of the five genes. The expression level of each gene was normalized according to the reference gene Gapdh in the same animal along the 24 h time span.

Locomotor activity was monitored for the last week of the adaptation period. Activity data were collected every 15 s and recorded in 15 min time bins using electromagnetic sensors. In order to minimize the effects of external interruptions (e.g. water exchange, food supply and injections), the data recorded during the first 2 h following each known interruption were removed.

We use Two-dimensional Table Curve (TableCurve 2D, Jandel Scientific, San Diego, CA, USA) to assess the presence of periods equal to or longer than 12 h. The software derives the ‘basic’ characteristics for each of the rhythm components: period, acrophase (waveform-approximated peak time) and amplitude (half of the waveform-approximated extent of variation) with their standard errors (Weigl et al., 2004; Weigl et al., 2010). We refer in this study to periods of three domains: 24±4 h (circadian), and 18±2 and 13±3 h (semidian).

The fit of the expression patterns to a rhythmic function was evaluated by two statistical parameters that were obtained using TableCurve 2D: P-value of the rhythm and degrees of freedom (d.f.)-adjusted r² (d.f. r²). Two-sided unpaired t-tests with Welch correction were utilized to assess significant differences between the parameters of the different curves. Differences between amplitudes of <20% and between acrophases and periods of <2 h were not considered biologically meaningful. Differences of <10% in the actual rhythm fitness index (d.f. r²) was also regarded as biologically non-significant.

The rhythms of mRNA levels of two clock genes (Per2 and Clock) were monitored in the liver and esophagus and compared to that of locomotor activity, a behavioral marker of the SCN circadian pacemaker's temporal organization. In addition, this study assayed the daily transcript levels patterns of three cell cycle genes (Ccnd1, Ccne1 and Pcna) in these two peripheral tissues.

The parameters of the 3 day activity pattern of BALB/c mice are presented in Fig. 1. The circadian period best fitted the activity data with an acrophase at ZT 05:15.

The profiles of Per2 and Clock transcripts exhibited significant circadian rhythms (Table 2; Figs 2, 3). Yet, there were phenotypic differences: the amplitude of Per2 mRNA rhythm was 2-fold more robust than that of Clock mRNA (P=0.006), and ~7 h phase difference was found between the acrophases of these two transcripts. Similar temporal and amplitude relationships between these gene transcripts were documented previously (Liu et al., 2007).

In the case of the cell cycle genes, the rhythm of Pcna also demonstrated a significant circadian rhythm of 25:17 h. However, the Ccnd1 transcript level was of a shorter period (18:19 h). The acrophases of Pcna and Ccnd1 rhythms were shifted by ≥4:24 h (P≤0.003). No significant rhythm could be detected for Ccne1 mRNA.

In the esophagus, mRNA levels of all examined genes monitored over 24 h showed significant rhythms (Table 3; Figs 4, 5). The oscillations of Clock, Ccne1 and Pcna were similar in their temporal pattern: they all had ~17 h period and two acrophases around ZT 05:30 and ZT 22:30. Per2 mRNA showed a clear semi-dian period and the pattern of Ccnd1 was the only one to fit a circadian rhythm (21:17±0:28 h). The differences in the periods of the clock gene transcripts were highly significant (P<10⁻⁴).

The acrophase (around ZT 05:30) was common to the cell cycle- and clock-related transcripts rhythms (Table 3) and there were also amplitude differences: the largest fluctuation was displayed by Per2 transcript and the lowest by Pcna transcript (P=0.03).

Various studies have suggested that the circadian clock paces the timing of the cell cycle by synchronizing the expression of G1/S- and G2/M-related genes (Borgs et al., 2009; Khapre et al., 2010; Rana and Mahmood, 2010). At the same time, Oikonomou and Cross argued that the phases of the cell cycle stages could act as a time cue for the peripheral clock in those tissues (phase locking model) (Oikonomou and Cross, 2010). This study compared a possible interaction between the circadian clock and the passage from G1 to S stage in two peripheral tissues of different proliferation index, liver (low) and esophagus (high).

Under ‘normal’ conditions, the hierarchical model of the circadian time-keeping system anticipates that all peripheral clocks will be synchronized to the SCN circadian pacemaker and according to a ~24 h period (Dibner et al., 2010; Kowalska and Brown, 2007; Lowrey and Takahashi, 2011; Mohawk et al., 2012). The expression patterns of the positive and negative limb of the basic feedback loop (Clock and Per2 in our case) should be close to opposite in their phases in order for the clock to oscillate. The circadian profile of the locomotor activity predicts that the SCN pacemaker is responsible for the management of the circadian oscillations. The best-fitted liver mRNA rhythms of Per2 and Clock resemble the locomotor's activity circadian period. The circadian acrophases of Per2 and Clock transcripts were shifted by about 7.5 h; a similar timing alignment was reported in other tissues with circadian clocks, such as the lung (Liu et al., 2007).

In the esophagus, in contrast, Clock and Per2 transcripts exhibited non-circadian rhythms of diverse periods and amplitudes, and one similar acrophase (ZT 05:24, ZT 06:02). Several explanations can be proposed for the non-circadian phenotype: (i) the esophagus harbors neither a self-sustaining nor a SCN-paced circadian time device; (ii) the circadian phenotypes of other variables in the esophagus (but not the analyzed clock genes) are directly governed by the SCN and not by the tissue's clock; and (iii) genes other than Per2 (Per1 for example) function as the principal factors in the negative limb of the esophagus rhythm (Johnson, 2010). However, the relatively large amplitude of the Per2 mRNA rhythm does not favor this alternative. In addition, the non-circadian phenotype could also be the outcome of the differences in phase and amplitude between the various cell types in that tissue (epithelial, connective, muscular and nervous) (DeNardi and Riddell, 1991; Henrikson et al., 1997; Scheving and Gardner, 1998).

In the liver, the lack of significant oscillations in the transcript level of Ccne1 and the relatively low amplitude of the Pcna proliferation marker rhythm (0.8 cycles) seem to support the quiescent status of most of the hepatocytes (Counts et al., 1996; Khapre et al., 2010; Michalopoulos and DeFrances, 1997).

In the esophagus, however, the significant amplitude of the rhythm of both Ccne1 and Pcna transcripts affirms the relatively higher proliferation level of this tissue (Muskhelishvili et al., 2003; Scheving et al., 1979). According to Scheving and colleagues, most of the replicating cells are located in the sub-mucosal epidermis (Scheving et al., 1998).

At the protein level it was shown that the joint activity of cyclins D1 and E1 (in complexes with the cyclin-dependent kinases, CDKs) promotes the transcription of Pcna (Kondo et al., 2006). Several studies found the patterns of these complexes to be shaped by the cyclins' rhythms (Bjarnason and Jordan, 2000; Bjarnason et al., 1999; Morgan, 1995; Murray, 1992; Rensing et al., 2001). Thus, the timing of S stage is assumed to be determined by the G1 cyclins.

In the present study, the rhythm of Ccnd1 differed significantly from that of Pcna in both tissues. Thus, it seems probable that the phases of G1 and S stages are independently controlled. In the esophagus, for example, the highly significant circadian period of Ccnd1 might imply a synchrony in the G1 stage between the tissue cells. In the case of the short S stage, however, the shorter and less significant period of the Pcna transcript rhythm could point to a degree of desynchronization in the timing of this cell cycle stage. In addition, the fact that this study addressed the transcript patterns rather than the protein levels could also contribute to the different findings obtained here and previously (Kondo et al., 2006).

The coupling of the cell cycle stages to the tissue's clock device predicts that functions of the cell cycle stage will oscillate according to the local clock (Gérard and Goldbeter, 2012).

In the esophagus, the similarity of the rhythmic profiles of Ccne1 and Pcna transcripts (G1 to S-related genes) and the absence of circadian rhythm of the clock genes oppose the possibility that the cell cycle rhythm in this tissue is paced by a local time device. The synchrony of the mRNA rhythms of Ccne1, Pcna and Clock (a similar period and a common acrophase, ZT 05:00–06:00), suggests that the G1/S transition operates as a time cue of the tissue clock apparatus, through regulating the expression rhythm of the Clock gene. It should be noted that the temporal alignment on the protein level could differ from the current results, which are based on the transcript level.

The 5 h shift between the second acrophases of Per2 (ZT 18:03) and Clock (ZT 23:02) mRNA rhythms possibly implies a close relationship to the local clock device (esophagus). Thus, it seems that the rhythmic phenotypes of Per2 and Clock transcripts reflect a combination of the G1/S transition effect and the output of the local clock apparatus.

In tissues of low proliferation index, the rhythm of clock gene expression is not expected to be modified by the cell cycle's signals. The circadian oscillations of Per2 and Clock transcripts in the hepatic cells (locomotor-like period) suggest a SCN role in that tissue pace.

The expression rhythms of Pcna and Clock genes were similar in their circadian period and amplitude with about 12 h phase shift. Such a relationship questions whether the liver's circadian clock indirectly restrains the proliferation of the hepatocytes through pacing their timing pattern (Gérard and Goldbeter, 2012).

In conclusion, this study suggests the existence of two routes of interaction between the cell cycle and the clock time device in the esophagus and liver. The phase and amplitude of the G1 stage are governed by the circadian clock (either through Per2 or Clock). Yet, in dividing cells, the G1/S checkpoint seems to act as a time cue that can modulate the circadian clock output of the same tissue. The extent of this effect might be influenced by the proliferation index of that tissue and could explain the non-circadian oscillations of the clock genes' phenotypes.