Please visit the new BioModels platform to access the latest content. This website is no longer updated and will be retired on 31 May 2019.
BioModels Database logo

BioModels Database


Heldt2018 - Proliferation-quiescence decision in response to DNA damage

August 2018, model of the month by Lu Li
Original model: BIOMD0000000700


Under mild DNA damage, mammalian cells can escape the cell cycle to enter a quiescence state – a mechanism to avoid propagation of the damaged DNA to later generations. However, the molecular network underlying this proliferation-quiescence decision and the time window for cells to make such decision are unclear.

Heldt et al. combined mathematical modelling with live single-cell imaging to address these questions (1). Their model is build on key molecular components in two major cell-cycle checkpoints: the restriction point (RP), which determines whether cells commit to the cell cycle, and the transition from G1 to S phase, which indicates the start of DNA replication. These two transition points are equipped with their own molecular switches, formed by feedback loops of molecular interactions. They check whether certain requirements are fulfilled. If so, cell cycle would progress further from that point.


Figure 1

Figure 1. Computational model of RP and G1/S transition. (A) Model diagram of key components in RP and G1/S transition. (B) Deterministic simulation results of model shown in A. (C) p21 expression levels in stochastic simulations (n=30) of model depicted in A, and in live single-cell imaging of cycling hTert-RPE1 cells (n=29). Figure taken and adapted from [1].

More specifically, as shown in Fig.1A, RP network consists of the mutual inhibition of retinoblastoma protein (Rb) and E2F transcription factors. When mitogen level reaches beyond the threshold, the RP-switch will promote the passage through G1. Similarly, G1/S transition is controlled by a mutual inhibition, in this case, cycline-dependent kinase inhibitor (p21) and replication complexes (aRC). In addition, G1/S transition network contains the positive feedback from DNA damage to p21 expression. If there is a substantial DNA damage, cells will be prevented from entering S phase. Deterministic simulations of both networks show characteristic protein activities at specific cell-cycle phases (Fig1B). The activation pattern of p21 is also confirmed by stochastic simulations and live single cell imaging, showing its attenuation at S phase (Fig.1C).

Figure 2

Figure 2. p21 affects RP switch. (A) Model diagram of a sub-network underlying RP. (B) Bifurcation diagram showing the transformation of continuous growth factor level into discrete E2F activities. (C) Bifurcation diagram showing p21 level controls proliferation-quiescence decision making at a high growth factor level. (B,C) Solid lines: stable steady states; dashed lines: unstable steady states. (D) Stochastic simulations of the full model in Fig.1A, showing simulation instances initiated with similar p21 levels, displays distinct cell fates. Marginal histograms depicting the number of additional stochastic simulations, with various initial activities of p21, result in quiescence (grey) or proliferation (green) states. Figure taken and adapted from [1].

Numerical simulations of a sub-network, underlying RP (Fig.2B), show that, without p21, inhibitions between Rb and E2F are sufficient to form a bistable switch, converting the continuous mitogen activity into a discrete all-or-none commitment to proliferation (Fig. 2C). Although cells can enter the quiescence state with a low level of mitogen, the threshold of which for inducing proliferation is modulated by the expression of p21 that is linked to inherited DNA damage from mother cells (Fig2D,E). Heldt et al. explains that this is due to p21’s inhibitory role on cycling dependent kinase 2 (Cdk2), which hijacks the feedback loop between Rb and E2F. Hence, p21, in the RP network, integrates accumulated DNA damage with mitogen stimulation, to make the proliferation-quiescence decision.

The timing of DNA damage seems to play a vital role on the efficiency of p21 induced cell-cycle arrest. Both numerical simulations of the full-network model (Fig.1A), and live single-cell imaging, show that the latter DNA damage is induced, the harder it is to stop proliferation (Fig.3B and C). Interestingly, this model predicts a complete loose of sensitivity to DNA damage at S phase, due to the degradation of p21 by ubiquitin ligase such as, Skp2 and Cdt2. In fact, deleting Skp2 or Cdt2 in cycling cells in in vitro experiments, shows enhanced sensitivity to DNA damage in G1 phase, although to different extends (Fig.3 D,E,F and G).

In summary, the above results suggest a time window in G1 phase when cells become most sensitive to DNA damage. However, for many cancer cells, their loss of control on proliferation are not due to the incorrect timing, but defects in underlying control mechanisms (2,3). Therefore, Heldt et al. looked into which molecular components are having potentials to compromise the quality control in the cell cycle.

Figure 3

Figure 3. p21 degradation controls the system’s response to DNA damage. (A) Model diagram of a sub-network underlying G1/S transition. (B,D,F) Deterministic simulations of full model in Fig.1A, showing the effect of timing and level of DNA damage on cell cycle, in unperturbed (B), Skp2 depleted (D) and Cdt2 depleted system (F). (C,E,G) Experimental data showing the percentage of cells entering S phase or arrest upon DMSO (control) or CPT (DNA-damage inducer) treatment, in unperturbed (C), Skp2 depleted (E) or Cdt2 depleted (G) cells. Figure taken and adapted from [1].

Figure 4

Figure 4. Key components in proliferation-quiescence decision making. Deterministic simulations of full model in Fig.1A showing the combined efforts from growth factor and DNA damage on cell cycle in unperturbed (A), Rb removed (B) or p21 removed (C) system. Figure taken and adapted from [1].

Based on the insight provided by the computational modelling that, in normal cells, the proliferation is decided by both the level of mitogen and DNA damage, therefore molecules that rely these signals to the network, such as Rb and p21, could also make the system vulnerable. As shown in Fig.4B, removing Rb from the model, increases the threshold of DNA damage to induce cell-cycle arrest, and intriguingly, allows the system to proliferate even without the mitogen stimulation, as observed in published experiments (4,5). On the other hand, deleting p21 from the network, insulates the system from DNA damage, and guarantees proliferation by moderate levels of mitogen (Fig.4C).


In conclusion, Heldt et al. primarily applied computational modelling approaches to dissect the molecular interactions underlying RP and G1/S transition. Their findings suggest a sensitive decision-making window in the G1 phase to potentially arrest cell cycle. This is achieved by integrating both mitogen activity and DNA damage via key molecules such as Rb and p21. The authors also propose the degradation of p21 as a potential mechanism for the tolerance of intrinsic DNA damage during the S phase.


  1. Heldt, Frank S., Barr, A. R., Cooper, S., Bakal, C., & Novák,B. (2018). A comprehensive model for the proliferation-quiescence decision in response to endogenous DNA damage in human cells. Proc Natl Acad Sci USA , 115:2532-2537. doi: 10.1073/pnas.1715345115
  2. Sherr CJ. (1996). Cancer cell cycles. Science , 274:1672–1677.
  3. Malumbres, M. & Barbacid, M. (2001). To cycle or not to cycle: A critical decision in cancer Nat Rev Cancer , 1:222–231. doi: 10.1038/35106065
  4. Jan-Hermen Dannenberg, Agnes van Rossum, Leontine Schuijff, and Hein te Riele. (2000). Ablation of the retinoblastoma gene family deregulates G(1) control causing immortalization and increased cell turnover under growth-restricting conditions. Genes Dev , 14:3051–3064.
  5. Julien Sage, George J. Mulligan, Laura D. Attardi, Abigail Miller, SiQi Chen, Bart Williams, Elias Theodorou, and Tyler Jacks (2000). Targeted disruption of the three Rb-related genes leads to loss of G(1) control and immortalization. Genes Dev , 14:3037–3050.