Preparation for DNA Replication: The Key to a Successful S phase

Abstract

Successful genome duplication is required for cell proliferation and demands extraordinary precision and accuracy. The mechanisms by which cells enter, progress through, and exit S phase are intense areas of focus in the cell cycle and genome stability fields. Key molecular events in the G1 phase of the cell division cycle, especially origin licensing, are essential for pre-establishing conditions for efficient DNA replication during the subsequent S phase. If G1 events are poorly regulated or disordered, then DNA replication can be compromised leading to genome instability, a hallmark of tumorigenesis. Upon entry into S phase, coordinated origin firing and replication progression ensure complete, timely, and precise chromosome replication. Both G1 and S phase progression are controlled by master cell cycle protein kinases and ubiquitin ligases that govern the activity and abundance of DNA replication factors. In this short review, we describe current understanding and recent developments related to G1 progression and S phase entrance and exit with a particular focus on origin licensing regulation in vertebrates.

Introduction

The somatic eukaryotic cell cycle is a series of phases: G1, S, G2, and mitosis. Cells start the cell cycle in G1 (gap phase 1) progress through S (DNA replication), G2 (gap phase 2), and then divide in M (mitosis) [1]. Cells enter G1 either from the preceding mitosis and cytokinesis or from a quiescent state (also known as G0 phase) which is distinct from the replicative cell cycle. Cells transit between G0 and the replicative cycle in response to a variety of extracellular mitogenic signals [2, 3].

Genome stability, the faithful duplication and transmission of chromosomal DNA from one cell to its daughter cells, depends on accurate and efficient DNA replication. Events that prepare for DNA replication must be ordered and precise to avoid catastrophic consequences and genome instability which includes increased frequencies of base pair mutations, duplication or deletion of chromosomal loci, variation in chromosome number and structure, and changes in microsatellites (tandem nucleotide repeats scattered throughout the genome) [4]. Molecular events in G1 set the stage for DNA replication in S phase. Likewise, a collection of molecular processes governs orderly progression through S phase in preparation for G2 phase. Decades of intense investigation using a variety of experimental systems have identified key DNA replication proteins and their regulators in G1 and S. Nevertheless, important questions remain: How exactly do cells “know” when to transition from G1 into S phase

Origin licensing prepares cells for S phase by loading inactive replicative helicases onto DNA during G1. To achieve complete but precise genome duplication, licensing must start after mitosis and finish prior to the initiation of S phase when origin firing takes place. Origin firing involves helicase activation in S phase and establishing bidirectional replication forks, but once origins have fired, no new licensing is permitted [5, 6] (Figure 1A). The consequences of any new origin licensing in S phase are serious because re-replication creates extra copies of loci that can lead to genome instability (Figure 1B). On the other hand, too few licensed origins can also lead to genome instability associated with incomplete replication (Figure 1C).

Figure 1. Illustrations of normal and abnormal origin licensing outcomes.

A. Normal origin licensing and CDK2-dependent firing in G1 and S phase. B. Unscheduled licensing and firing leading to re-replication and genome instability. C. Underlicensing in G1 leading to replication stress and genome instability.

G1: Preparing DNA for S phase

A primary activity during G1 is preparing DNA for replication during S phase. Replication initiates at specific sites known as DNA replication origins, and efficient eukaryotic replication requires many active origins per chromosome [7, 8]. What defines an origin varies by organism: among eukaryotes some yeasts such as S. cerevisiae have origins that are (at least partly) defined by conserved DNA sequences [9–12]. While origins have been successfully mapped in other eukaryotes, their locations are not easily predicted from DNA sequences. Highly active replication origins in several vertebrates have recently been shown to have some common general features such as open chromatin, GC-rich sequences, G4 quadruplex structures, or proximity to active genes (reviewed in [12, 13]).

Certain histone modifications play an active role in regulating the chromatin landscapes relevant to origin licensing. In particular, monomethylation of histone H4 lysine 20 (H4K20me1) promotes origin licensing. The monomethyltransferase SETD8 (also known as PR-Set7) catalyzes H4K20me1. Depleting SETD8 causes defects in origin licensing, whereas inappropriate SETD8 expression in S phase promotes unregulated re-licensing and re-replication[14–16]. Subsequent tri-methylation of histone H4K20, which requires H4K20me1 as a substrate, ensures that late firing origins are activated during S phase. Furthermore, replication origins are active even in heterochromatin where lower accessibility likely makes licensing and origin activation challenging. Some proteins that are not required at origins in other genomic regions, such as ORCA [17] and shelterin [18], are important for licensing heterochromatin or telomeres. Considering the complexity of the eukaryotic genome, it is remarkable that enough origins are licensed in G1 to ensure the entire genome is usually replicated in S phase with few problems.

Origin licensing involves the sequential action of licensing factors to load the core of the replicative helicase, MCM, onto DNA (Figure 2, text box, and Table 1). Our understanding of the licensing mechanism has advanced significantly from biochemical reconstitutions using S. cerevisiae or X. laevis licensing proteins. Origin licensing starts with ORC (Origin Recognition Complex, a DNA binding complex of six subunits Orc1–6), which directly binds origins of replication. ORC works with two additional helicase loading factors, CDC6 (Cell Division Cycle 6) and CDT1 (CDC10-dependent transcript 1), to load at least two hexamers of the MCM replicative helicase (Mini-chromosome Maintenance Complex, six subunits MCM2–7) [19, 20]. Licensed origins have at least one loaded MCM double hexamer (= two MCM hexamers). MCM complexes form the core of the replicative helicases at replication forks but are not active during G1; they will be activated or “fired” in S phase. Once MCM double hexamers are loaded, the loading factors ORC, CDC6, and CDT1 are not required for MCM complexes to remain on DNA [21]. MCM complexes are stably DNA-bound (reviewed in [20]).

Figure 2. Origin licensing.

The concerted action of ORC, CDC6, and CDT1 load hexamers of MCM onto DNA during G1 phase. A loaded MCM double hexamer constitutes a licensed origin; See also the text box.

Table.

Glossary of Primary Terms: Vertebrate proteins and complexes in this mini-review*			Active Cell Cycle Phase
Abbreviation	Full Name	Function	G1	S	G2-M
APC/C	Anaphase promoting complex/Cyclosome	E3 ubiquitin ligase, targets many substrates from anaphase through G1; inactive in S and G2	✔		(✔)
ATM	Ataxia Telangiectasia Mutated	protein kinase activated by DNA damage, blocks S phase and M phase entry and stimulates DNA repair	✔	✔	✔
ATR	ATM-related	protein kinase active during replication stress, and by DNA damage (all phases), arrests/delays cell cycle progression and stimulates DNA repair	✔	✔	✔
CDC45	Cell Division Cycle 45	component of the CMG helicase		✔
CDC6	Cell Division Cycle 6	loads MCM, degraded in early G1, stabilized in late G1, cytoplasmic in S/G2	✔
CDC20	Cell Division Cycle 20	substrate coactivator for the APC/C ubiquitin E3 ligase in M phase			✔
CDH1	CDC20 Homolog 1	substrate coactivator for the APC/C ubiquitin E3 ligase in G1 phase	✔
CDK	Cyclin Dependent Kinase	protein kinase heterodimer of one cyclin with one catalytic subunit, different CDK in different phases	✔	✔	✔
CDT1	Cdc10-Dependent Transcript 1	loads MCM2–7, degraded in S	✔		✔
CDT2	Cdc10 Dependent Transcript 2	substrate adaptor for the CRL4CDT2 E3 ubiquitin ligase, activated by substrate binding to DNA-loaded PCNA		✔
CHK1	Checkpoint Kinase 1	protein kinase activated by ATR after replication stress or DNA damage	✔	✔	✔
CHK2	Checkpoint Kinase 2	protein kinase activated by ATM after DNA damage	✔		✔
CMG	Cdc45 + MCM2–7 + GINS	active form of the replicative DNA helicase		✔
CRL4	Cullin Ring Ligase 4	E3 ubiquitin ligase complex, recruited by CDT2 to target CDT1, SETD8, and p21 in S phase		✔
cyclin A	Cyclin A	subunit of cyclin A/CDK2 and cyclin A/CDK1 kinase		✔	✔
cyclin D	Cyclin D	activates CDK4 and CDK6 to phosphorylate Rb and contribute to Rb inactivation	✔
cyclin E	Cyclin E	subunit of cyclin E/CDK2 kinase, stabilizes CDC6, stimulates origin firing	✔	✔
DBF4	Dumbbell Forming Protein 4	subunit of DBF4/CDC7 kinase (DDK)		✔
DDK	DBF4 dependent kinase	protein kinase heterodimer of DBF4 and CDC7; phosphorylates MCM4 & MCM6		✔
E2F	Adenovirus E2 gene Factor	transcriptional regulator of genes encoding many key S phase proteins	✔	✔
EMI1	Early mitotic inhibitor 1	inhibits the APC/C ubiquitin E3 ligase during S & G2 phases		✔	✔
GINS	Go-Ichi-Ni-San	Heterotetramer component of the CMG helicase (SLD5, PSF1–3)		✔
GMNN	Geminin	inhibits re-replication by blocking CDT1-MCM binding		✔	✔
MCM	Mini-chromosome Maintenance	heterohexamer of subunits MCM2–7, loaded at origins in G1, component of CMG helicase	✔	✔
ORC	Origin Recognition Complex	heterohexamer of subunits ORC1–6, binds DNA and loads MCM	✔
p21	protein 21 kDa	inhibitor of CDK2 and CDK1	✔
p27	protein 27 kDa	inhibitor of CDK2 and CDK1	✔
p53	protein 53 kDa	transcription factor induced by DNA damage and other stresses	✔		✔
PCNA	Proliferating Cell Nuclear Antigen	DNA polymerase clamp/processivity factor at replication forks; coordinates multiple interactions		✔
Rb	Retinoblastoma protein	transcriptional repressor of E2F-regulated genes; phosphorylated by cyclin D/CDK4, cyclin D/CDK6, and cyclin E/CDK2	✔
SCF	SKP1-CUL1-F-box	E3 ubiquitin ligase complex, targets some CDK-phosphorylated substrates		✔	✔
SETD8	Su(var)3–9, enhancer-of-zeste, and trithorax	histone h4 lysine 20 monomethylase, promotes licensing (a.k.a. PR-SET7, SET8)	✔		✔
SKP2	Suppressor of Kinetochore Protein Mutant 2	F-box protein family member, substrate adapter for SCFSKP2	✔	✔

^*Additional firing and replication fork factors: TOPBP1, treslin/MTBP, MCM10, RECQL4, POLε, RPA, RFC

Origin licensing is promoted in G1 through a combination of transcriptional activation of the genes encoding licensing proteins and post-transcriptional control. CDC6, CDT1, and each subunit of ORC and MCM are the products of genes controlled by the E2F transcription factor family [22, 23]. In G1 phase, mitogen stimulation triggers production of cyclin D which binds and activates the CDK4 and CDK6 protein kinases. Cyclin D/CDK4–6 complexes de-repress E2F by mono-phosphorylating its inhibitor, Rb, at one of many individual phosphorylation sites [24]. Recent work has shown that the 14 principal sites of Rb mono-phosphorylation lead to different profiles of E2F-dependent transcription, indicating considerable complexity in the G1 transcriptional program [25]. In late G1 phase one of the many E2F-activated genes, CCNE1 which encodes the cyclin E protein, is sharply activated [24]. Cyclin E binds and activates the CDK2 protein kinase (cyclin-dependent protein kinase 2) which then hyper-phosphorylates and inactivates Rb in a positive feedback loop. As a result, late G1 cells synthesize considerably more of the licensing proteins than early G1 cells.

At least one of the origin licensing proteins, CDC6, also requires post-transcriptional activation during G1 phase. CDC6 is a substrate of the APC/C^CDH1 E3 ubiquitin ligase which is active from mitosis through G1. As a result, CDC6 protein levels remain low during G1 because it is ubiquitylated by APC/C^CDH1 and degraded by the proteasome. In late G1 phase however, CDC6 is no longer degraded because APC/C^CDH1 cannot bind and ubiquitylate CDC6 once it has been phosphorylated by cyclin E/CDK2 [26]. CDC6 is only one of the 14 polypeptides that form the core of the licensing system (MCM2–7, ORC1–6, CDT1, and CDC6), and each of these proteins is subject to a variety of post-translational modifications at some point in the cell cycle. Many of these modifications are associated with inactivating licensing outside of G1 phase (i.e. in the previous S, G2, and M phases), and thus must be reversed to activate licensing during G1. For example, CDT1 is hyperphosphorylated in mitosis, and those phosphorylation events are associated with inactivating CDT1 for licensing [27]. Similarly, ORC subunits bind chromatin poorly in mitosis as a result of mitotic CDK activity [28, 29]. Presumably these inactivating phosphorylations are reversed by phosphatases in G1, or the substrate proteins are degraded and resynthesized. In support of phosphatases as activators of licensing, the Rif1-PP1 phosphatase has recently been shown to protect the largest ORC subunit from premature phosphorylation and degradation during G1 phase [30].

Origin licensing checkpoint.

Origin licensing is restricted to G1 phase to avoid re-replication (described below). When has enough origin licensing taken place during G1 to ensure complete genome duplication? In other words, is there a mechanism that ensures sufficient genome-wide origin licensing prior to the activation of S phase? Such a mechanism would be an “origin licensing checkpoint,” and it would delay S phase entry while origin licensing is incomplete. It is clear that in budding yeast no such mechanism exists because null mutations in essential origin licensing proteins do not block cell cycle progression; such mutants enter mitosis with un-replicated DNA [31, 32]. On the other hand, evidence that a licensing checkpoint in mammalian cells delays the activation of the S phase kinase CDK2 when origin licensing is inhibited is described below.

By what criteria can we infer the existence of a mammalian origin licensing checkpoint? In 1988, Hartwell described cell cycle checkpoints as “control mechanisms enforcing dependency in the cell cycle.” Checkpoints [33] refer to signaling pathways that ensure proper completion of key cell cycle steps before the next cell cycle phase transition, but checkpoint regulators are not themselves essential to the processes they regulate. This important aspect of checkpoints is sometimes only evident when one identifies treatments or mutations that bypass the regulation to allow premature cell cycle progression. An analogy for a checkpoint is the roller coaster in Figure 3. Before the ride begins, safety monitors conduct a series of safety checks of the riders themselves and of the equipment. These monitors will not allow the ride to begin unless the rules have been satisfied. If the monitors are absent, then a ride could, in principle, proceed, albeit with added risk. Cell cycle checkpoints are evident when a component of the monitoring system is eliminated that then allows inappropriate or risky cell cycle progression.

Figure. 3. Progression through G1 and S phase represented as a roller-coaster track.

Major decision points in G1 and at the G1/S transition are indicated in yellow, and a cell cycle checkpoint at or just prior to the G1/S transition is marked in green.

Evidence for an origin licensing checkpoint is not simply less DNA replication when MCM loading is reduced. A true licensing checkpoint would operate during G1 phase and prevent initial S phase entry. Moreover, the checkpoint regulators must not themselves be essential DNA replication proteins, but rather proteins that restrain S phase entry (i.e. safety monitors). When the checkpoint regulators are depleted or mutated, then cells can proceed prematurely into S phase, albeit with a higher risk of incomplete replication. Indeed, in non-transformed mammalian cell lines, manipulations that inhibit origin licensing result in a longer G1 phase and low CDK2 activity [33–36]. Interestingly, the same manipulations in transformed cell lines do not affect G1 length or CDK2 activity, and cells instead proceed into S phase with considerably less than their typical amount of licensed origins. The subsequent S phase is therefore “underlicensed” (Figure 1C). In the most severe licensing depletions, cells experience replication failure and apoptosis [35, 37]. Perhaps the most compelling evidence for a true licensing checkpoint is the fact that removing p53, a central player in the response to cellular stresses that is absent or compromised in tumor derived cells, bypassed all of the checkpoint behaviors. Like transformed cells, non-transformed cells lacking p53 do not delay in G1 with low CDK2 activity even when licensing is impaired [34, 35], and, like their yeast orthologs, can even enter premature mitosis [38]. Thus, it is clear that p53-proficient cells can couple the activation of CDK2 and S phase with the status of origin licensing.

This checkpoint has long been under-investigated because it cannot be studied in either budding or fission yeast (there is no yeast p53 ortholog) or the more conveniently manipulated tumor cell lines. Multiple studies have documented the origin licensing checkpoint phenomenon [36, 39–45], though the molecular details of this pathway are still very poorly understood. For example, it is not known if p53 is activated by post-translational modification in the same ways it is activated during a DNA damage response or if basal p53 activity is sufficient to confer licensing checkpoint proficiency. There is little evidence that p53 accumulates after licensing inhibition, suggesting that the strong stabilizing signals that characterize responses to strong DNA damage signals are not part of the licensing checkpoint. Some licensing checkpoint experiments have detected increased p21 protein, and since p21 is the product of a p53-inducible gene, the increased p21 could be an indicator of p53 activation. On the other hand, high p21 is also an indirect consequence of G1 arrest, since p21 is degraded in S phase cells and stable in G1 [46]. We recently found that cells re-entering the cell cycle from quiescence have an impaired licensing checkpoint [34]. This natural checkpoint deficiency may provide a unique avenue to dissect the pathway connecting licensing to S phase entry. Considerable future work will need to determine what exactly cells are measuring about MCM loading, how that measurement is made, how the information is converted into CDK2 regulation, and the role of p53 in that process.

In contrast to the licensing checkpoint, a very well-studied checkpoint is the p53-dependent G1/S checkpoint. This network of surveillance proteins is activated by DNA damage to block S phase entry. DNA damage in G1 triggers ATM and/or ATR kinases depending on the nature of the damage. These kinases stimulate signaling cascades that block cell cycle progression and activate DNA repair (reviewed in [47, 48]). The DNA damage checkpoint does not prevent DNA from being damaged but rather responds to damage and blocks cells from progressing into S phase by inducing inhibition of CDK2 and allowing time for repair.

S phase entry

The large size of eukaryotic genomes presents challenges for timely and efficient DNA replication. Meeting this challenge requires that cells prepare many origins in G1 and activate many origins in S phase. To start S phase, replication origins are activated or “fired” through a series of sequential protein phosphorylation events. Interestingly, not all origins are fired at the same time but rather, are fired at different times as S phase proceeds. Heterochromatin loci can pose more of a challenge for origin firing and thus, are generally replicated later in S phase than the more accessible euchromatic regions. Presumably origins in euchromatin are activated before origins in heterochromatin because the heterochromatic landscape restricts access to firing factors [7].

The replication timing program ensures the temporal order of origin firing (early versus late origin firing) and may be influenced by origin licensing parameters that were established in the preceding G1 phase. High-resolution protein-DNA interaction mapping experiments in budding yeast detected more MCM loading at early-firing origins compared to late-firing origins. This extra MCM could compete for limiting concentrations of firing factors during S phase [49]. Similarly, more ORC has been detected at early versus late-replicating regions in human cells [50]. Interestingly however, this differential loading may not have translated to more origin licensing because no differences in MCM loading were detected in late-replicating heterochromatin versus early replicating euchromatin in human cells [8]. Thus, MCM loading differences do not fully account for establishing replication timing programs. Additional proteins such as the ORC-associated ORCA protein, which binds different origins at different times during G1 [51], may distinguish early versus late origins. Such proteins could either preferentially delay late origins or preferentially accelerate early origins, and such factors could be preloaded in G1 or act exclusively in S phase. Furthermore, topologically associated domains (TADs), three-dimensional chromosome conformations comprised of self-interacting regions of the genome, have also been suggested to influence replication timing. Discrete cis-regulatory units have recently been shown to shape the chromatin landscape and architecture to ensure properly timed replication [52].

Origin firing creates the replisome, or replication “machine”: a protein complex that is assembled at replication origins that were licensed in G1 (reviewed in [53–56]). The replisome is responsible for DNA replication in a bidirectional fashion from each origin that fires. Replication forks initiate where DNA is unwound by the replicative helicase. Helicase activation requires CDK2 with contributions from DDK (DBF4-dependent kinase). The replicative helicase is assembled from the MCM complexes that were loaded in G1 during origin licensing to which the CDC45 protein and GINS complex are added during origin firing. This assembly is the CMG (CDC45-MCM-GINS) helicase (reviewed in [53, 54]). GINS and CDC45 are recruited to MCM through TopBP1 and CDK-phosphorylated treslin, a TopBP1-binding protein homologous to the yeast Sld3 initiation protein. In budding yeast, phosphorylation of residues on the MCM2–7 helicase by DDK triggers helicase activation by recruiting Sld3 to mediates CDC45 loading [54, 56]. By extension, vertebrate DDK may function similarly and is required for S phase completion [55]. Origin firing also depends on RECQL4, the MCM10 protein, and DNA polymerase (Pol ɛ). Once the helicase is activated, the bifurcation of two opposing replication forks from the origin commences. Finally, PCNA, RPA, and RFC, among other proteins, are recruited to fully activate the replisome. These proteins are the products of E2F-regulated genes, and thus their availability in S phase requires their prior accumulation in G1 as part of the preparation process [22].

Since origin firing is dependent on both CDK and DDK activity, the activation of these protein kinases in S phase is essential for genome duplication. For DDK, activity depends on accumulation of the DBF4 subunit, a product of an E2F-regulated gene, which binds to the CDC7 catalytic subunit to form active DDK [22]. In the case of CDK2, full activation requires multiple molecular events. First, cyclin protein must accumulate because it is an essential activator of CDK2. In late G1 and early S phase, the cyclin most responsible for CDK2 activation is cyclin E which, as discussed above, is produced by E2F-dependent activation of the CCNE gene. CDK2 must also be phosphorylated at the activating T160 by CDK-activating kinase (CAK) and also lack phosphorylation at Wee1-dependent inhibitory sites T14 and Y15 [57].

Two primary CDK2 inhibitor proteins, p21 and p27, are abundant in G1 and must be degraded in late G1 for full CDK2 activation. The p27 CDK inhibitor is ubiquitylated by the SCF^SKP2 E3 ubiquitin ligase after p27 has been phosphorylated by CDK2 [58]. Thus, CDK2 stimulates its own activation by triggering the destruction of one of its inhibitors. The p21 CDK2 inhibitor is also an SCF^SKP2 substrate, and SCF targeting causes p21 levels to fall beginning in late G1. The largest drop in p21 levels occurs in early S phase and is dependent on the CRL4^CDT2 E3 ubiquitin ligase. Targeting by CRL4^CDT2 requires recruitment of both the E3 ubiquitin ligase and the p21 substrate to DNA-loaded PCNA at replication forks in a process termed replication-coupled destruction [46, 59–61]. The precise dynamics of CDK inhibitor degradation is a combination of the rate of ubiquitylation and the action of deubiquitinases that oppose ubiquitylation [62]. To date, the role of deubiquitinases that govern p21 and p27 degradation at G1/S are not fully known, though a recent study indicates an important contribution of USP11 for p21 control [63] or USP37 for p27 control [64].

Dynamics at G1/S.

Once a cell passes through the G1/S transition and has commenced origin firing, it is committed to full genome duplication and ultimately, cell division. This commitment is irreversible which necessitates that many molecular events switch robustly and in a coordinated fashion. Our understanding of precise inter-relationships at the G1/S transition has been advanced in recent years by single cell studies using live-cell imaging combined with fluorescent reporters of specific molecular events [65–68].

In addition to the many factors contributing to CDK2 activation described thus far, a master E3 ubiquitin ligase is integrated into the G1/S commitment. The anaphase promoting complex, also called the cyclosome or APC/C, is an E3 ubiquitin ligase that functions in G1 to keep mitotic CDKs inactive by targeting S phase and mitotic cyclins for destruction. APC/C also targets multiple proteins directly or indirectly involved in origin firing and activation, including CDC6 (discussed above), the DBF4 subunit of the DDK kinase, the SKP2 component of the SCF^SKP2 ubiquitin ligase, and the re-replication inhibitor protein Geminin (discussed below). Robust S phase entry requires that APC/C be active in G1 but inactive during S phase [66, 69, 70].

APC/C activity is dependent on one of two substrate coactivator subunits: CDC20 in M phase and CDH1 in G1 phase. In G1 phase, neither cyclins D nor E are substrates of APC/C, whereas the S phase and M phase cyclins A and B are actively degraded by APC/C. As CDK2 activity rises, CDK2 phosphorylates CDH1 to contribute to initial APC/C inactivation since phosphorylated CDH1 cannot bind APC/C well. CDH1 is subsequently degraded in S phase by SCF^{cyclin F}-dependent ubiquitination[70], an additional contribution to APC/C inactivation [68]. In addition, E2F-stimulated transcription drives production of the APC/C inhibitor protein EMI1 whose levels rise through the G1/S transition (reviewed in [71, 72]). EMI1 is subsequently stabilized by a variety of ubiquitination and phosphorylation events near the G1/S transition [73]. Once APC/C is inactivated, DNA replication (i.e. S phase) is destined to begin. Recent work has underscored APC/C inactivation as an indicator of passage through this “point of no return” regardless of stressors that could have blocked entry into S phase had they been encountered in G1 (e.g. DNA damage) [66]. Whether early S phase can start before APC/C is fully inactive however, is not known.

During S phase

As cells transition into S phase, CDK2 is activated first by cyclin E (early S phase) and then by cyclin A (mid-to-late S phase) to fire origins. Cyclin A accumulation is the result of both E2F-dependent transcription during preparation for S phase in G1 and the early S phase inactivation of APC/C which ubiquitylates cyclin A from early mitosis through the end of G1 phase. CDK2 activity rises from the overall increase in cyclin concentration and the full destruction of both p21 and p27 CDK inhibitors. In mid-to-late S phase, cyclin E is downregulated by a ubiquitylation pathway that, similar the CDK inhibitors, is stimulated by substrate phosphorylation. In the case of cyclin E, the ubiquitin E3 ligase is SCF^Fbw7 which targets phosphorylated cyclin E [74, 75]. By this point, cyclin A levels are high meaning that during S phase, CDK2 switches binding partners from cyclin E to cyclin A. Although not all CDK2 substrates can be phosphorylated by both cyclin E/CDK2 and cyclin A/CDK2, both forms of CDK2 share the substrates that are phosphorylated during origin firing. In mid-to-late S phase, cyclin A/CDK2 is thus responsible for origin firing later in S phase to ensure replication completion [76, 77].

Re-replication.

As mentioned earlier, origins are not fired at the same time in S phase. Throughout S phase, there are regions of the genome that have already been replicated and regions that remain un-replicated. Genome stability requires that no region be replicated more than once. Thus, mechanisms have evolved to ensure that no sequences become re-replicated in a single cell cycle. The consequences of re-replication are serious and include DNA breaks, replication stress, gene amplification, mutagenesis, chromosome instability, and oncogenesis [78–82]. Re-replication is blocked by preventing origin licensing once S phase starts ensuring that even the earliest regions to be replicated are not re-licensed. Vertebrate genomes are large and complex, and S phase is typically many hours, so no single molecular mechanism can prevent all re-licensing. Instead, multiple mechanisms operate in parallel, and all of them are required for full re-replication control [78, 81]. Even seemingly minor perturbations that increase the likelihood of re-replication have serious organismal consequences from the increased genome instability [80, 81, 83].

In vertebrate cells, early S phase is marked by the destruction of both the CDT1 MCM loading protein and the SETD8 histone monomethytransferase. Both of these proteins are targeted by CRL4^CDT2 as soon as origins fire because the trigger for CRL4^CDT2 targeting is the presence of active PCNA-containing replication forks [59]. CDT1 overproduction or failure to degrade either CDT1 or SETD8 can induce substantial re-replication [16, 84]. During S phase (once APC/C is inactive), the geminin protein is stabilized because it is no longer targeted by APC/C. Geminin binds tightly to CDT1 and blocks interaction with MCM, thus inhibiting CDT1’s licensing function [85–87]. This inhibition may be particularly important in G2 phase when CDT1 becomes stable again and re-accumulates both to perform an essential mitotic function and in preparation for the next G1 phase [88, 89]. Even minimal licensing in G2 would, by definition, occur on DNA that had been replicated already, and the high CDK activity in G2 cells could rapidly stimulate re-firing and re-replication at such sites [90, 91].

In S phase, cyclin A/CDK2 specifically inactivates some licensing proteins that are apparently poor substrates for cyclin E/CDK2 (and thus not inactivated in late G1 by cyclin E/CDK2). This specificity between cyclin E and cyclin A preserves licensing activity through the end of G1, but as cyclin A accumulates it inhibits licensing during S and G2 phases. Cyclin A/CDK2 phosphorylates a number of licensing proteins during S phase: (1) CDT1, creating a phosphodegron recognized by the SCF^SKP2 E3 ubiquitin ligase to contribute to CDT1 degradation in S phase [92, 93]; (2) CDC6 (near its nuclear localization signal), promoting CDC6 export to the cytoplasm [94]; and (3) ORC1 (the largest subunit of ORC), stimulating Orc1 degradation via SCF^SKP2 [94, 95]. At least one additional subunit of ORC, ORC2, is also phosphorylated by CDK2 and/or CDK1 which is activated by cyclin A in G2 and by cyclin B in mitosis. ORC2 phosphorylation reduces its affinity for chromatin [28, 29].

S phase progression is also under control of the intra-S phase checkpoint that responds to DNA damage, to slowed or stalled replication, and to the occasional replication stress that occurs even during a normal, unperturbed S phase. Various types of DNA damage activate the ATR and ATM pathways, which phosphorylate CHK1 and CHK2 kinases respectively. One of the downstream consequences of checkpoint activation is S and M phase CDK inhibition, which slows or stops the pace of new origin firing and blocks mitotic entry (reviewed in [96–98]). As a consequence of turning off the origin firing kinases CDK2 and DDK, the intra-S phase checkpoint blocks origin firing, avoiding replication of damaged DNA, protecting replication forks, and also stimulating DNA repair [48, 91, 97, 99]. Even in the absence of DNA damage, ATR controls the pace of origin firing [99, 100].

Interestingly, the intra-S checkpoint does not suppress all origin firing. In the absence of DNA damage, origins fire in clusters as indicated by single-molecule in vivo DNA labelling experiments that detect origin firing [101, 102]. The ATR-CHK1 pathway suppresses origin firing in clusters that have not yet initiated, but licensed origins within an active cluster are not suppressed. It is not yet known how the intra-S checkpoint pathway distinguishes origins that should be suppressed from those that should not.

Dormant origins.

DNA fiber analysis has been used to estimate the replicon size in somatic mammalian cells relative to the number of licensed origins per nucleus, and the number of licensed origins greatly exceeds the number that fire [8, 103]. The number of licensed origins can be reduced substantially by partially depleting MCM subunits or MCM loading factors with little impact on short-term overall proliferation [104, 105]. Why do cells license extra origins during G1 that typically remain dormant during S? Origin efficiency, the likelihood that a particular origin will fire in a given S phase, varies by location and can be influenced by cell type and by local gene expression [106]. It may be that a universal replication licensing program is established in G1, but that cell type or environmental signals that impact local chromatin structure restrict origin firing to a selected subset.

The strongest explanation for why cells license many dormant origins comes from analyzing cellular responses to replication stress and long-term impacts on genome stability [107]. Some replication forks stall because they encounter lesions and blocks during S phase. A suite of lesion bypass, replication fork restart, and DNA repair systems are available during S phase largely because the genes encoding these proteins are also E2F-regulated and therefore induced during G1 in preparation for S phase [47]. If a replication fork stalls and cannot be repaired, then the DNA ahead of the stalled fork must be replicated by a fork converging from the opposite direction. Licensing more origins in G1 ensures cells are prepared to fire these “backup” or dormant origins in S phase in the locations where problems arise. Much like a savings account stores money for unexpected expenses, dormant origins serve as a “savings account” for the genome. Dormant origins (extra origins licensed in G1) are activated or fired in S phase in response to problems encountered during replication [104, 108]. These dormant origins escape suppression by the intra-S checkpoint to generate rescue replication forks near a stalled replisome [81, 104, 105, 108, 109]. If dormant origins are not available, the consequences could be dire (Figure 1C): incomplete replication of the genome could lead to genome instability, hypersensitivity to replication stress, and possibly trigger tumorigenesis if left unchecked [110].

S phase exit

DNA replication should complete before the onset of mitosis. As described above, origin firing during S phase is regulated by the ATR-CHK1 pathway, which simultaneously prevents premature mitosis [100]. The ATR pathway is activated by ongoing replication and blocks not only mitosis itself, but also the transcriptional program that stimulates mitotic protein accumulation; this relationship has been proposed as an S-G2 checkpoint [111]. In addition to ATR, cyclin A/CDK1 is also important for regulating late S phase origin firing [76]. Because cyclin A is an APC/C substrate, and APC/C is inactive during S and G2 phase, cyclin A accumulates to high levels in late S phase to activate both CDK2 and CDK1. Premature activation of cyclin A/CDK1 caused aberrant replication earlier in S phase as a result of dysregulated origin firing. Furthermore, inhibiting CDK1 lengthened the time cells spent in S phase, as cells struggled with inefficient late origin firing at the end of S phase [76]. By these measurements, Cyclin A/CDK1 has a relatively unique role in controlling late S phase events, and high cyclin A/CDK1 activity is one of the few molecular markers of G2 phase.

Essentially all of the MCM that was originally loaded during origin licensing in G1 phase is unloaded by the end of S phase [112]. Most of the dormant origins will not fire, and presumably these MCM complexes are removed by the replication forks that transit those sites. What happens when a replication fork encounters a loaded and unfired MCM double hexamer is currently unknown. MCM complexes in active CMG complexes at replication forks are removed during replisome disassembly. Two proposed mechanisms have been reported: 1) association with an alternate Mcm2 subunit, MCM-BP, has been suggested to unload MCM2–7 from DNA during S phase [113, 114] (though MCM-BP has many partners and proposed roles), 2) polyubiquitylation of the Mcm7 subunit stimulates Mcm7 removal from the complex by a non-degradative mechanism promoting complete replisome disassembly [115–117] and reviewed in [118–120]. However, many questions remain about the mechanisms and order of events of MCM unloading at the end of S phase.

Conclusion

In summary, the events of G1 play critical roles that determine how well cells are prepared for DNA replication in the subsequent S phase. The sequence of molecular events in G1 and in S phase are currently being further elucidated by extensive use of quantitative high-content single-cell analysis and single-molecule studies. Generating and analyzing time-lapse images of cells with fluorescent cell cycle reporters has proven quite useful for gathering robust data to update our understanding of cell cycle dynamics at both the G1/S and S/G2 transitions [65, 67, 68]. We anticipate that such endeavors will continue to address the relationships among the critical pathways in G1 and S phase. The kinetics of origin licensing during G1 relative to E2F-depedent transcription, CDK activity, and APC/C inactivation are not yet known for normal vertebrate cells. Moreover, how genetic perturbations associated with oncogenic transformation impact licensing kinetics and the downstream consequences of such perturbations in S phase are also incompletely understood. In addition, the precise relationships among epigenetic regulators, licensing locations, and subsequent origin firing by CDK and DDK are still unclear and could be better understood with these newer techniques. In addition, we still don’t fully understand the relationship between replication licensing and the G1/S transition nor the relationship between replication completion and the onset of G2. Because genome stability depends on the efficiency of DNA replication, further studies of how cells enter and exit S phase are still crucial.

Text Box: Origin Licensing is the process of loading MCM hexamers onto DNA by ORC, Cdc6, and Cdt1; see Figure 2.

The molecular details of loading MCM complexes to form DNA-bound double hexamers have been largely determined with purified S. cerevisiae (budding yeast) proteins in vitro. MCM loading in other species is presumed to follow a similar sequence.

1:

• ORC binds DNA directly

• Cdc6 binds ORC

• Cdt1 binds a single MCM2–7 hexamer

• The Cdt1-MCM2–7 complex binds ORC-Cdc6

• ORC-Cdc6-Cdt1 load the MCM2–7 hexamer onto DNA; MCM encircles dsDNA

• MCM hydrolyzes ATP, and Cdc6 and Cdt1 are released

2:

• Cdc6 binds ORC

• A second Cdt1-MCM2–7 complex interacts with the loaded ORC-Cdc6-MCM. Alternatively, a second ORC may bind DNA on the opposite side of the loaded MCM.

• The second MCM2–7 is loaded with the two N-termini of the MCM complexes interacting

• ORC, Cdc6, and Cdt1 are released, leaving a loaded MCM double hexamer on ds DNA