|Home | About | Journals | Submit | Contact Us | Français|
In this issue, De Piccoli et al. (2012) show that, contrary to current models of DNA replication checkpoint function, replication proteins remain associated with each other and with replicating DNA when replication is stressed in checkpoint-deficient cells.
In eukaryotic cells, DNA replication slows down significantly when cells are exposed to DNA damage. In 1980 it was observed that, in contrast to normal human cells, cells lacking the serine/threonine kinase ATM (Ataxia telangiectasia mutated) continue to replicate their DNA quickly in the presence of DNA damaging agents (Painter and Young, 1980). Following this work, studies in a variety of organisms identified a set of genes, similar to ATM, that together constitute a signal transduction cascade, called the replication checkpoint, that recognizes stalled replication forks. Later work suggested that the checkpoint slows S phase primarily by inhibiting the firing of late origins of replication, rather than by slowing the rates of individual replication forks (Branzei and Foiani, 2009). Checkpoint mutants cannot restart replication successfully after DNA damage or other replication stress is removed (Desany et al., 1998); this is due, at least in part, to unrepaired breaks that persist at stressed replication forks in these mutants (Branzei and Foiani, 2009; Lopes et al., 2001) . In an effort to understand how the checkpoint prevented these breaks from occurring, researchers used chromatin immunoprecipitation (ChIP) to examine replication proteins at forks in checkpoint mutants (Cobb et al., 2003; Cobb et al., 2005; Katou et al., 2003). Results from these studies have generated a long-standing model in which the checkpoint signaling cascade is required to maintain the association of replication-fork proteins with replicating DNA and/or each other in order to prevent such breaks from occurring (Branzei and Foiani, 2009); however, the mechanism by which the checkpoint regulates this is not clear. In this issue of Molecular Cell, De Piccoli et al. (2012) present evidence that, contrary to previous models, DNA replication complexes remain both intact and associated with the replication fork upon replication stress in checkpoint-deficient Saccharomyces cerevisiae cells. Moreover, they provide data suggesting that the checkpoint might, in fact, slow replication in conditions of replication stress not only by inhibiting late-firing origins but also by slowing the rate of each fork.
The eukaryotic DNA replication machinery (or “replisome”) contains over 100 proteins, including a helicase (MCM2-7) that unwinds the template DNA and polymerases that copy it. DNA polymerase progression can be inhibited in several ways including by alkylating agents, such as methyl methanesulfonate (MMS) which chemically alter bases on the DNA template, or by drugs such as hydroxyurea (HU), which deplete dNTPs. In wildtype cells, the MCM helicase is functionally coupled to the polymerases: if the polymerases stop polymerizing, as they do in HU, the helicase stops as well (Katou et al., 2003). This could happen in three ways: (1) the helicase may not be able to move through chromatinized DNA in vivo without the polymerases pushing it; (2) the helicase may move by itself, with the energy of ATP hydrolysis, but stop because it is physically tethered to the rest of the replisome; (3) it may be that when polymerases stop, it is sensed by the checkpoint and the checkpoint inhibits the helicase. De Piccoli et al. show that, in HU, most replication forks stop at the same place in wildtype or checkpoint-deficient cells, suggesting that both the helicase and polymerases stop moving when dNTPs are depleted, and that (1) or (2) is true even in checkpoint-deficient cells. However, they observed that replication forks from the very earliest-firing origins, which begin replication when more dNTPs are available, travel further in checkpoint-deficient cells than in wildtype, suggesting that, as in (3), the checkpoint also directly inhibits fork progression under replication stress.
In wildtype cells, the replication checkpoint is activated when the sensor kinase Mec1 (ATR) is recruited to stalled forks, probably through an interaction with ssDNA revealed at the stalled fork (Branzei and Foiani, 2009). Once Mec1 is bound, it promotes activation of the effector kinase Rad53. Both Mec1 and Rad53 are required for cells to survive even short treatments with HU or MMS, which suggests that these kinases are required for replisomes to resume replication; indeed, EM shows that rad53 mutants accumulate up to 800 bases of ssDNA in HU, compared to 320 bases in wildtype cells (Sogo et al., 2002). Several models for how the replisome becomes compromised in mec1 and rad53 mutants in HU have been proposed (Figure 1). First, it has been suggested that dissociation of the MCM helicase from the polymerase allows the helicase to unwind DNA without associated replication (Cobb et al., 2005). De Piccoli et al. rule out this model by using IPs to show that the MCMs, and all other replisome components examined, remain associated with the rest of the replisome even in mec1 or rad53 mutants in HU. It has also been suggested that ssDNA is created when the entire replisome translocates along the template DNA, unwinding without polymerizing. Contrary to this model, De Piccoli et al. show by ChIP-seq that, as in wildtype cells, the MCMs and polymerases are both present on DNA around origins of replication in mec1 and rad53 mutants. To rule out the possibility that the entire replisome falls off chromatin (Branzei and Foiani, 2009), the authors additionally show by biochemical methods that these replisomes are associated with chromatin.
If mec1 and rad53 replisomes remain intact at the site of DNA synthesis in HU, why can't they resume DNA synthesis when the HU is washed out? It is likely that Mec1 and/or Rad53 phosphorylation of replisome proteins may be required to keep the replisome in a replication-competent state. For example, deletion of the nuclease EXO1 completely rescues the sensitivity of rad53 mutants to the S phase damaging agent MMS (Segurado and Diffley, 2008), suggesting that Rad53 inhibition of Exo1 promotes survival during replication stress. Additional unknown Rad53 and Mec1 targets are likely to be important for survival of HU treatment (Segurado and Diffley, 2008); De Piccoli et al. identify the helicase holoenzyme component Psf1 as a potential Mec1 target and it will be interesting to see whether Mec1-mediated phosphorylation of Psf1 affects replication fork stability during HU treatment.
The new data in De Piccoli et al. is at odds with ChIP data (Cobb et al., 2003; Cobb et al., 2005; Katou et al., 2003) showing that replisome components are lost from replicating DNA in mec1 and rad53 mutants, and changes how we may explain the ssDNA gaps and DNA breaks observed previously. Perhaps the ChIP data can be reconciled in light of the observation by De Piccoli et al. that replisomes from the earliest origins in the genome travel further in mec1 and rad53 mutants than in wildtype cells, which could dilute the ChIP signal. This observation might explain the previous EM and 2D gel data as well (Sogo et al., 2002; Lopes et al., 2001): if the polymerases from these replisomes were able to translocate in the absence of DNA synthesis, the newly-synthesized DNA might contain significant gaps. This might result in the ssDNA observed in EM, which could be subject to breakage to produce the 2D gel structures that have been observed. Alternatively, even though the replisome seems to be intact at most, if not all forks, in the absence of checkpoint proteins the DNA may form aberrant structures that are processed by nucleases such as Exo1 to yield ssDNA (Branzei and Foiani, 2009). Further study will be necessary to determine how DNA synthesis and processing are coupled to replisome movement in checkpoint-deficient cells.