A NONO-gate times the cell cycle

When a cell divides, a series of tightly regulated events are sequentially passed through to end up with two healthy daughter cells. A dysregulation of these steps such as an escape from cell cycle control checkpoints may lead to tumor formation. In recent years, it became increasingly clear that our 24-h internal clock also contributes to proper timing of cell cycle progression and likely acts as a tumor suppressor (1). Although epidemiological studies argue that chronic living against one’s circadian clock increases the risk for cancer (2) and other diseases, the molecular mechanisms for circadian timing of the cell cycle are little understood. In PNAS, Kowalska et al. (3) uncover one such mechanism by demonstrating that the multifunctional nuclear protein NONO, together with circadian PERIOD proteins, controls the time-of-day–dependent expression of the key cell cycle checkpoint gene p16-Ink4A and thus the exit from G1 cell cycle phase. This study presents an animal model in which cell division is uncoupled from a functional circadian clock.

Circadian clocks are endogenous, molecular oscillators present in nearly every eukaryotic cell. In mammals, they control the 24-h timing of multiple processes in physiology, metabolism, and behavior predominantly by governing the rhythmic expression of ∼10% of all transcripts in a given tissue. Rhythm generation is accomplished by a molecular clockwork; briefly, the heterodimer CLOCK/BMAL1 activates the transcription of period (Per1, Per2, Per3) and cryptochrome (Cry1, Cry2) genes, whose protein products—after a delay of several hours—inhibit their own expression by suppressing CLOCK/BMAL1 activity. This so-called transcriptional–translational feedback loop, which involves many additional proteins (including NONO) and posttranscriptional steps, creates molecular ∼24-h oscillations of clock gene levels (reviewed in ref. 4). Importantly, in addition to clock genes, the molecular clockwork also drives the rhythmic expression of a large suite of genes responsible for various cellular processes including cell cycle, DNA repair, and genotoxic stress response (5).

Already ∼50 y ago, circadian rhythms in cell division were observed in unicellular organisms (6). They have been interpreted as a way to protect the cells from the DNA-damaging effects of the sunlight by restricting sensitive cell cycle stages to safer times of the 24-h cycle. Also in mammals, daily rhythms of cell division in proliferating tissues (e.g., oral mucosa, bone marrow) have been shown; however, investigating the molecular link to the circadian clock is not an easy task. First, many tissues harbor both nondividing and proliferating cells, making it difficult to analyze circadian cell division. Second, uncoupling the clock from the cell cycle without also disrupting either of the two mechanisms has been unsuccessful to date. In a pioneering study (7), Matsuo et al. demonstrated 10 y ago that the circadian clock gates the transition from cycle stages G2 to M in regenerating liver (with nearly all cells proliferating), likely by rhythmic expression of Wee-1, a kinase phosphorylating the key checkpoint component Cdc2. Thereby, G2-to-M cell cycle progression may be possible only at specific phases during the circadian cycle, i.e., when Wee-1 concentration is minimal. In addition to Wee-1, many other critical cell cycle components are targets of the circadian clock (reviewed in ref. 8). Yet it remains unclear whether such a regulation is required for normal timing of cell division. However, it is likely, because, in mouse models with disrupted circadian systems (i.e., with a destroyed circadian pacemaker in the brain or clock gene mutations), important processes such as regulation of apoptosis and DNA repair are dysregulated, which leads to accelerated tumor development and progression (8).

When they started their study, Kowalska et al. (3) did not expect to discover a regulator of cell cycle progression; rather, they aimed to characterize the role of NONO within the circadian clockwork, a protein they previously discovered as part of PERIOD-containing complexes (9). They created a gene-trap, loss-of-function mouse (NONO^gt), which—probably as a result of redundancy to SFPQ, another PERIOD-binding, NONO-family member (10)—showed only a subtle circadian phenotype. However, they noticed that primary fibroblasts from NONO^gt mice divided approximately twofold faster in culture than WT cells, an effect that could be rescued upon reintroducing NONO. This proliferation phenotype seems to primarily result from a reduction in cellular senescence rather than an acceleration of cell division; senescence-associated β-gal activity was lower in NONO^gt compared with WT cells. In addition, to be able to follow cell division over several generations, Kowalska et al. (3) permanently stained the cytoplasm of NONO^gt and WT cells and found that the percentage of dividing cells is much higher without NONO, suggesting NONO promotes senescence. Senescent cells lose their ability to divide by repressing genes that drive cell cycle entry. Factors such as NONO promoting cellular senescence might therefore inhibit the G1-to-S phase transition. Indeed, a twofold enrichment of cells in S phase was detected in NONO^gt primary fibroblasts, indicating that NONO may act as a regulator of G1 phase exit.

How does NONO control G1 phase exit on a molecular level? Cell cycle inhibitors like p16-Ink4A and p53 are key regulators of G1 arrest and cellular senescence. Indeed, systematic analysis revealed that p16-Ink4A likely is a direct target of NONO: although upstream regulators appeared unaffected in cells lacking NONO, expression of p16-Ink4A was low. Furthermore, NONO binds to the promoter region of p16-Ink4A and promotes its transcription. As NONO is not known to directly bind DNA, the mechanism of its coactivator function remains open. Surprisingly, although NONO itself is not rhythmic (3), NONO binding to the p16-Ink4A promoter is time-of-day–dependent, leading to oscillations in p16-Ink4A activation.

How can the nonrhythmic NONO confer rhythmicity to its target? From a previous work by the same authors (9), we know NONO forms complexes with the rhythmic PERIOD proteins, providing a possible mechanism. In fact, Kowalska et al. (3) find that PERIOD2 is also rhythmically present at the p16-Ink4A promoter and that NONO–PERIOD partners need each other to activate p16-Ink4A transcription. In both NONO^gt and Per1/Per2 double mutant mice, neither rhythmic p16-Ink4A promoter binding of the remaining partner nor rhythmic transcription was observed. As a consequence, circadian gating of the cell cycle (S phase) by the circadian clock depended on the presence of NONO in primary fibroblasts.

For two reasons, we believe this is a remarkable finding: (i) the NONO^gt mouse is the first animal model in which clock and cell cycle are still running but are uncoupled from each other, which allowed a direct mechanistic demonstration of circadian clock control on cell cycle events; and (ii) it supports an emerging new clock output concept—a protein that is not oscillating (NONO) can be “hijacked” by rhythmic components of the circadian system (PERIOD2) via protein–protein interaction to convey circadian information to the process it is controlling (cell division). Interestingly, PERIOD proteins, in this case, rather than repressing—their normal job within the circadian clockwork—help to activate transcription (Fig. 1).

Fig. 1.

NONO couples the circadian clock to the cell cycle. The multifunctional nuclear protein NONO interacts with circadian PERIOD proteins only at certain times of the day when PER levels are high. Together they activate (with the help of yet unknown factors) the expression of p16-Ink4A, a crucial regulator of G1 arrest and cellular senescence. When NONO is absent, cell division is uncoupled from the (still ticking) clock, leading to reduced senescence and a loss of circadian gating of the cell cycle S phase.

What are the physiological consequences of functionally disconnecting the circadian clock and the cell cycle? Are the NONO^gt mice more cancer-prone or

Although NONO itself is not rhythmic, NONO binding to the p16-INK4A promoter is time-of-day–dependent.

do they show aging-related phenotypes—both possible consequences of reduced cellular senescence? These are definitely very interesting lines of future research; Kowalska et al. (3), however, focus on dermal wound healing, in which cellular senescence has also been implicated to play an important role. In skin, proliferation and DNA damage response have been shown to be controlled by the circadian clock (11, 12). Wound healing is a complex process in which multiple cell types not only proliferate but also migrate and differentiate to reaccomplish skin integrity. For example, myofibroblasts first proliferate and secrete ECM, but eventually go into senescence and become matrix-degrading cells (13). In NONO^gt mice, wound healing is severely defective in many aspects: hyperproliferation of the fibroblast and the keratinocyte layer (paralleling the in vitro data), disorganization of granulation tissue, and, surprisingly, hardly any ECM production.

Is the loss of circadian cell cycle or the reduced senescence (or both) responsible for the malfunction in skin repair? Although this needs still to be investigated in detail, circadian regulation is definitely an attractive explanation because clock KO mice (Bmal1^−/− and Per1/Per2 double mutants) also show defective wound healing. Remarkably, these clock mutant mice show roughly “opposite” wound healing phenotypes possibly corresponding to the different roles of BMAL1 and PERIOD proteins within the circadian oscillator; Bmal1^−/− animals have reduced fibroblast numbers with excessive ECM, whereas Per1/Per2 double mutants show the opposite. Thus, the circadian clock might control the timing of organized (maybe alternating) cycles of cell division and ECM secretion, thereby governing the formation of complex tissue structures.

Chronobiology provides more and more examples how molecular ∼24-h rhythms translate into rhythmic physiology. Obeying the clock seems to be necessary for health. However, shift work, reduced sleep, irregular meal times, and jet-lag are pervasive in our modern “24/7” society. Knowledge about the underlying mechanisms of circadian clock-controlled processes may help to develop tailored treatment strategies that exploit the time dimension of cellular physiology. Cancer chronotherapy is one example; applying chemotherapy at specific times during the day aims to determine an optimal time window during which effectiveness and best tolerability of drugs is optimal (14). Although this is very attractive as a concept and also successful in some cases, we need a much better understanding of the underlying principles to define in which cases chronotherapy should become clinical routine. The study by Kowalska et al. (3) not only extends our knowledge in this respect but also introduces an invaluable model in which the importance of circadian timing of cell cycle events for tumor formation and aging can be mechanistically analyzed.