Register for Mechanobiology 2016!

The chromosome replication cycle
John F. X. Diffley, Karim Labib

To ensure that the relatively large genomes of eukaryotic cells are efficiently and precisely duplicated in each cell cycle, DNA replication initiates from multiple replication origins distributed along the individual chromosomes. In budding yeast, replication origins are short, well-defined DNA sequences; in early metazoan development (e.g. Xenopus and Drosophila), regulated DNA replication can initiate from virtually any DNA sequence. The situation in somatic mammalian cells appears to lie somewhere between these two extremes. In most eukaryotic cell types, replication origins are activated continuously throughout S phase according to a temporal programme. For simplicity, an early-firing and a late-firing origin are considered in this poster, which summarises data from a variety of eukaryotic systems (for details, see Kelly and Brown, 2000).

Step 1. Replication origins are determined, at least in part, by the binding of the six subunit origin recognition complex (ORC), which has been conserved in evolution from yeast to humans. In budding yeast, ORC binds specifically to an essential, bipartite sequence element within replication origins. ORC binds ATP and this binding is required for DNA binding. ORC can also hydrolyse ATP; however, ATP hydrolysis is not required for DNA binding and is, in fact, inhibited upon origin binding. ORC remains bound to origins throughout the cell cycle in budding yeast, even through mitosis. In multicellular eukaryotes some ORC subunits remain bound throughout G1, S and G2 phases, although there is conflicting evidence as to whether ORC remains bound to chromosomes during mitosis.

Step 2. At the end of mitosis two proteins, Cdc6 and Cdt1, are recruited separately to ORC. Cdc6 and several of the ORC subunits are members of the AAA+ ATPases, which include clamp loader proteins such as RFC (see below). Cdt1 does not show any significant similarity to other proteins although it may contain a region of coiled-coil.

Step 3. Six closely related proteins, known as the MCM2-7 complex, are recruited to chromatin by ORC, Cdc6 and Cdt1 into a complex known as the prereplicative complex (pre-RC). The MCM2-7 assembly reaction is also known as `licensing'. The MCM2-7 complex is required throughout S phase for both initiation and elongation. A subcomplex of MCM4, 6 and 7 has DNA helicase activity, which suggests a role for the MCM complex in unwinding DNA at replication forks. In addition, other members of the MCM family are essential for elongation, indicating that the functional MCM complex — perhaps the replicative helicase — includes all six MCM proteins. In organisms from yeast to humans, it appears that roughly 20 copies of the MCM2-7 complex are loaded onto chromatin per origin (or molecule of ORC). The reason for this stoichiometry is presently unclear.

Step 3a. Factors determining the timing of origin firing are still only poorly understood; however, chromosomal context is a critical determinant. For example, origins near telomeres tend to be activated late in S phase and early-firing origins moved to a sub-telomeric location become late-firing. Evidence in yeast and mammalian cells indicates that the temporal programme of origin firing is established early in the cell cycle, well before DNA replication begins. Elegant experiments from Raghuraman and colleagues using a sub-telomeric, and, consequently, late-firing origin flanked by FLP recombinase sites showed that an origin excised prior to mitosis replicated early in the subsequent S phase ( Raghuraman et al., 1997). However, if excision occurred during G1 phase, after pre-RC assembly, the origin replicated late in the subsequent S phase.

Gilbert and colleagues have exploited a system in which isolated, intact nuclei from Chinese hamster ovary (CHO) cells replicate in extracts from Xenopus eggs to show that this temporal programme of origin firing in mammalian cells is established some time early in G1 phase. Intriguing correlations between replication timing and nuclear position have been seen in both yeast and CHO cells.

In addition, when very early G1 phase CHO nuclei are incubated in egg extracts, initiation occurs at random sites across the DHFR locus. However, when later G1-phase CHO nuclei are used, initiation occurs specifically at the well-defined DHFR origin. This has been used to identify the `origin decision point' that occurs after both pre-RC assembly and the temporal decision point (see Gilbert, 2001).

Step 4. Many eukaryotic cells enter quiescence from G1 phase. In budding yeast, pre-RCs formed at the end of mitosis are lost from origins as they enter quiescence (stationary phase). Quiescent mammalian cells are also unlicensed, although it is not clear whether such cells have lost pre-RCs or not assembled them in the first place.

Step 5. The pre-RC matures into a pre-initiation complex (pre-IC) by recruitment of additional factors including Cdc45 and Sld3, and perhaps other initiation components such as Dpb11 (the S. cerevisiae homologue of the S. pombe Cut5 protein and of human TopBP1). Cdc45 subsequently plays an essential role during elongation as well as initiation. It is not known whether Sld3 or Dpb11 also have essential roles in elongation. A number of other proteins have been implicated in the initiation of DNA replication around this stage including Dna43 (also known as MCM10) and Sld2 (also known as Drc1).

Step 6. Two conserved protein kinases play essential roles in triggering the initiation of DNA replication during S phase. One of these is a cyclin-dependent kinase (CDK). In unicellular fungi, such as the budding and fission yeasts, a single CDK (Cdc28 in S. cerevisiae and cdc2 in S. pombe) is primarily responsible for driving all of the key cell cycle transitions. In multicellular eukaryotes, the cell cycle is controlled by multiple CDKs. In all organisms examined, S phase is normally activated by a specific cyclin CDK-combination. In S. cerevisiae, Cdc28 acts with two B-type cyclins, CLB5 and 6, to drive S phase; in S. pombe, S phase is driven by cdc2 and the cig2 B-type cyclin. In metazoans, CDK2 acts as the S phase specific CDK together with cyclins A and E. These two cyclins are likely to have distinct roles in initiation. Although many initiation factors are phosphorylated in vivo by CDKs, the physiologically relevant substrates have thus far been difficult to divine.

The other essential S phase-promoting kinase is Cdc7. Like the CDKs, Cdc7 requires a non-catalytic subunit, Dbf4, for its activity. There are other parallels between Cdc7-Dbf4 and the CDKs: for example, in budding yeast, Dbf4 is unstable in early G1 phase because it is targeted for ubiquitin-mediated proteolysis by the anaphase promoting complex, similar to the B cyclins. Dbf4 interacts with replication origins in vivo and the ORC binding site is essential for this interaction. Biochemical experiments in yeast have suggested that Cdc6 and consequently MCM2-7 are not required for this interaction, however, experiments in Xenopus have suggested that the interaction requires MCM2-7, not ORC or Cdc6. Genetic and biochemical evidence suggests that the MCM2-7 complex is an important target of the kinase.

Chromatin immunoprecipitation experiments after formaldehyde crosslinking in budding yeast have indicated that pre-ICs assemble at early-firing origins during G1 phase, prior to Start (restriction point in mammalian cells), but do not assemble at late-firing origins until just before initiation during S phase. Chromatin-binding experiments without crosslinking in budding yeast and Xenopus have indicated that pre-ICs only assemble after CDK and Cdc7 activation. One model to reconcile these differences suggests that, prior to kinase activation, factors such as Cdc45 are associated only loosely with origins and, after kinase activation, adopt a `locked' conformation. After the action of these two kinases, origins are unwound and the heterotrimeric single-strand DNA binding protein RPA is recruited.

Step 6a. In addition to their role in triggering initiation, CDKs play a second critical role in regulating DNA replication: they inhibit the assembly of new pre-RCs. This neatly prevents re-initiation from origins that get activated during S phase. Two generalisations can be made about the mechanisms by which CDKs prevent pre-RC assembly. First, the block to pre-RC assembly is achieved through multiple, redundant mechanisms. For example, in budding yeast, Cdc6, MCM2-7 and ORC are all negatively regulated by CDKs. Second, different organisms use different strategies to effect the block. For example, in budding and fission yeasts, Cdc6 is an unstable protein that is targeted for ubiquitin-mediated proteolysis at the end of G1 phase whereas, in metazoans, Cdc6 is stable during S phase but is exported from the nucleus at the onset of S phase. In metazoans, Cdt1 is inhibited by the small protein, geminin. Geminin is targeted for ubiquitin-mediated proteolysis by the anaphase promoting complex/cyclosome (APC/C), which, in turn, is activated by CDKs. For further discussion see Diffley (Diffley, 2000).

Step 7. After early unwinding, the first DNA polymerase is recruited. Protein-protein interactions with Cdc45 and/or RPA may play a role in this recruitment. Because DNA polymerases are unable to initiate chain synthesis de novo, there is an absolute requirement for a `primase' to synthesise a short tract of ribonucleotides. The short (6-10 nucleotides), template-bound RNA can then act as a `primer' for DNA synthesis. This early synthesis is carried out by the four subunit DNA polymerase α-primase complex.

Step 7a. Because of the anti-parallel nature of DNA, the two parental strands are replicated by different mechanisms during the progression of the replication fork. As the replicative helicase (perhaps MCM2-7) moves forward through the DNA (blue arrows), the parental strand, which is 3′ to 5′ relative to the direction of unwinding, can be replicated continuously by a DNA polymerase synthesising 5′ to 3′ in the same direction as the helicase. This is known as the `leading strand'. On the other strand, however, replication is trickier because DNA polymerases cannot synthesise DNA in a 3′ to 5′ direction. To circumvent this problem, this strand is replicated discontinuously; as the helicase unwinds DNA, DNA polymerase α-primase synthesises an RNA (red)-DNA (black) hybrid on the newly unwound DNA in the opposite direction of the helicase. As the helicase proceeds further, a gap is exposed on the template strand. This again serves as a template for a short RNA-DNA hybrid molecule. Thus, the `lagging strand' is replicated by repeated synthesis of short `Okazaki fragments'. Co-ordination of lagging strand synthesis with DNA unwinding is critical as DNA polymerase α must repeatedly initiate Okazaki fragments at ∼150-200 bp intervals while the DNA is being unwound. In bacterial and viral systems this is accomplished by critical interactions between the primase and the helicase. In eukaryotes, interactions between DNA polymeraseα and either MCM2-7 or Cdc45 may be important in this co-ordination.

Step 8. Replication proceeds bidirectionally from most chromosomal replication origins. That is, replication forks are established that move in opposite directions away from the origin. The first Okazaki fragment on each strand serves as the primer for the assembly of the leading strand machinery for each replication fork. Continuous DNA synthesis in the leading strand requires a DNA polymerase that can synthesise long stretches of DNA in a processive manner (i.e. without falling off). DNA polymerase α is unsuitable for this job because it exhibits very low processivity (∼30 nucleotides per initiation). In addition, DNA polymerase α does not have the 3′ to 5′ exonuclease activity required for `proofreading' (i.e. removing misincorporated nucleotides) and is therefore more error-prone than other polymerases. Consequently, leading strand synthesis requires a `polymerase switch' in which a more processive, high fidelity DNA polymerase replaces DNA polymerase α. In fact, because of the poor processivity and low fidelity of DNA polymerase α it is highly likely that most of the DNA in Okazaki fragments is synthesised by polymerases other than DNA polymerase α. In SV40 DNA replication, two DNA polymerases are sufficient for complete, efficient DNA replication both in vitro and in vivo. These are DNA polymerase α and DNA polymerase δ. Strong evidence that a third DNA polymerase, DNA polymerase ϵ, is required for efficient chromosomal DNA replication has come from work in yeast and Xenopus. The specific roles of DNA polymerase δ and ϵ have not yet been completely resolved.

Highly processive DNA synthesis by these DNA polymerases requires an auxiliary factor called PCNA (proliferating cell nuclear antigen). PCNA is a homotrimer that forms a ring shaped structure. This ring encircles the template DNA and is thus topolgically linked to the DNA. PCNA binds to the DNA polymerase and acts as a `sliding clamp', preventing the polymerase from falling off the DNA. The PCNA ring must be loaded onto the DNA primer end. This is accomplished by RFC, known as the `clamp loader'. Like ORC, Cdc6 and MCM2-7, RFC is a member of the AAA+ superfamily; it binds to the 3′ end of the primer and uses ATP to open up the PCNA ring and close it around the template DNA.

Step 8a. The topology of the DNA molecule changes as it is unwound during DNA replication. Unwinding generates positive supercoils ahead of the replication fork and precatenanes in the newly replicated DNA behind the fork. Topoisomerase I and II can relax positive supercoils, whereas topoisomerase II can remove catenanes and precatenanes (for details, see Lucas et al., 2001).

Step 8b. DNA damage and stalled replication forks trigger a global genome integrity checkpoint. Central to this response are two groups of kinases. The first are large proteins with homology to phosphoinositide 3-kinases. In humans, these include ATM, ATR and DNA PK; in yeast, these are Mec1 and Tel1 (S. cerevisiae) or Rad3 and Tel1 (S. pombe). The other kinase is known as Chk2 (humans), Rad53 (S. cerevisiae) and cds1 (S. pombe). Activation of this checkpoint prevents the firing of late-firing replication origins and prevents entry into mitosis. In addition, this checkpoint plays a role in preventing replication fork breakdown (see Donaldson and Blow, 2001).

Step 9. As detailed above, the lagging strand is synthesised discontinuously as a series of RNA-DNA hybrid molecules. Maturation of these Okazaki fragments involves removal of the RNA primer (and perhaps some DNA) in a concerted reaction requiring the flap endonuclease Fen1 and the helicase/endonuclease Dna2 followed by ligation by ligase I (for details, see Bae et al., 2001).

Step 9a. Replication of the genomic DNA is only one component of chromosome replication. A number of other important processes are coupled to DNA replication. PCNA plays a central role in coupling many processes to the replication fork via direct protein-protein interactions. After replication, the two daughter DNA molecules remain tightly associated with each other in a process known as sister chromatid cohesion. Sister chromatid cohesion requires a number of factors including Scc1-4, Eco1, Smc1,3, DNA polymerase σ, Ctf8,18, Dcc1, Pds5 and Rfc2-5, and can happen only during DNA replication.

The assembly of chromatin is coupled to DNA replication by the chromatin assembly factor CAF1. In yeast, CAF1 is not essential and it is believed that other factors including ASF1 are involved in chromatin assembly. The high-fidelity of DNA replication is achieved, in part, by the coupling of post-replication repair to DNA replication. These include mismatch repair, recombination and trans-lesion DNA synthesis.

Telomere elongation and maintenance are coupled to DNA replication. Evidence from yeast has indicated that proteins that participate in lagging strand synthesis, such as DNA polymerase α, are required for telomere synthesis. Finally, in addition to these basic processes involved in efficient chromosome replication and segregation, accurate chromosome replication includes the accurate inheritance of programmes of gene expression. Active genes must remain active after being replicated and silent heterochromatic regions remain inactive. This implies that chromatin states are inherited `epigenetically' during DNA replication.

Step 9b. Termination occurs when two opposing replication forks meet and the nascent DNA from the two forks is ligated together. Relatively little is known about this step in eukaryotes. Unlike prokaryotes, specific DNA sequences do not appear to be required. Proteins that act ahead of the DNA polymerases must be displaced before replication is completed to allow the polymerases to replicate the last bits of sequence. Since the helicase as well as topoisomerase I are likely to act ahead of the polymerases, their displacement prior to the completion of replication may have implications for the final steps of DNA replication and decatenation of the daughter molecules.

Step 10. As described earlier, the two sister chromatids are held together by cohesion. This is likely to be important to allow post-replicative repair of double strand breaks that are caused by DNA damaging agents. Cohesion does not appear to occur continuously across the chromatids, but instead appears to occur at specific and numerous cohesion sites, including centromeres.

Step 11. After assembly of a bipolar mitotic spindle in metaphase, controlled proteolysis of cohesins allows the two sisters to be pulled to opposite poles. After chromosome segregation is completed, CDKs are inactivated in a complex reaction that is coupled to cytokinesis by the mitotic exit network (MEN), also known as the septation initiation network (SIN). This final inactivation of CDKs occurs, in part, by the activation of the anaphase promoting complex, an E3 ubiquitin ligase that targets cyclins for degradation via the proteasome (for details, see Bardin and Amon, 2001). In addition to targeting cyclins for degradation, the APC also targets replication factors for proteolysis; these include geminin, an inhibitor of Cdt1 in metazoans, and Dbf4, the Cdc7 regulatory subunit in yeast. Cells enter G1 prepared for another round of pre-RC assembly.