DNA is duplicated within a complex macromolecular machine. Insights into how replication begins and how this is coordinated with progression of DNA synthesis come from a diverse range of sources.
DNA is replicated by unzipping the double helix to expose the bases that act as a template for copying the genetic material. Both strands of DNA serve as templates, and thus one double helix becomes two. Conceptually, this is a simple reaction, but the devil — as so often — is in the detail: the process is mediated by a multitude of proteins and turns out to be mechanically complex. A trio of papers in this issue1,2,3 have made considerable headway in understanding the intricacies of replication.
One level of complexity in the replication reaction comes from the fact that DNA polymerase, the enzyme that synthesizes the new DNA, cannot begin a strand itself. Rather, it extends a short RNA ‘primer’ that is already bound to the template. This means that an additional enzyme, a primase, is required to generate the primer to start the polymerase reaction4,5.
A second complexity lies in the fact that the two strands of the DNA double helix are arranged antiparallel to one another; the opposite directions are termed 5′ to 3′ and 3′ to 5′ (from the positions of carbon atoms in the sugars that make up the DNA backbone). However, DNA polymerase can synthesize DNA in only one direction: 5′ to 3′ (Fig. 1). So the 3′ to 5′ template strand — the ‘leading’ strand — can readily be replicated, but how is the other, ‘lagging’ strand copied? This dilemma was resolved by the discovery that the lagging strand is replicated discontinuously. It is synthesized in short pieces, called Okazaki fragments, that are then joined together — essentially, the polymerase takes two ‘steps’ forward and then synthesizes one back. The difference in the synthesis of the two strands means that, in principle, the leading strand requires only a single priming event, whereas the lagging strand needs a new primer for each Okazaki fragment. And all these events must somehow be coordinated to produce two daughter strands at roughly the same rate.
One of the outstanding conundrums concerning replication is how the process deals with broken or damaged DNA, for example that generated by ultraviolet radiation, particularly at the leading strand. Replicating the damage could obviously be harmful for the daughter cells. So does the replication machinery just stall and wait for the damage to be fixed, or is there some way to sidestep the damage and continue without copying the damaged DNA?
Heller and Marians (page 557)1 addressed this issue by biochemically mimicking a blockage of the leading-strand template. Their findings indicate that a new priming event can occur downstream of the lesion, so that the portion of the leading strand encompassing the lesion is not copied and the resulting DNA will be single-stranded. This would seem to contradict the prevalent theory that leading-strand synthesis is continuous. However, precisely this phenomenon is observed in bacterial cells, where ultraviolet irradiation can lead to single-stranded gaps on both leading and lagging strands. The gap is filled in afterwards using the remaining undamaged strand as a template. So presumably the existence of a mechanism for re-priming the leading strand by generating a single-strand gap buys the cell additional time to deal with the damage without interrupting the essential process of DNA replication.
How is the primase delivered to the site where it needs to act? In bacteria, the primase interacts physically with the enzyme responsible for unzipping the double helix, the replicative helicase. Some bacterial viruses that have their own replication machinery take this a step further by having the primase covalently attached to the helicase. This union of the unzipping and priming activities provides a mechanism to couple progression of the replication ‘fork’ with depositing the primase on the lagging strand. Heller and Marians1 reveal that the helicase is also crucial in re-priming the leading strand when replication is restarted after DNA damage.
The coupling of helicase and primase presents a potential dilemma. When the primase is deposited on the lagging strand, it synthesizes RNA in the opposite direction to the progression of the fork (Fig. 1). What happens to the motion of the fork when this happens? Do the enzymes uncouple, does the helicase run ahead, or does the fork stall transiently?
Lee et al. (page 621)2 examine the priming process using a model DNA replication system derived from a bacterial virus called bacteriophage T7. This well-known system contains a covalently fused primase–helicase called gp4. The authors analyse the priming process using a single-molecule approach with a model DNA substrate and three purified proteins: gp4, the T7 DNA polymerase and an accessory factor, thioredoxin.
They find that the rate of leading-strand synthesis in the absence of lagging-strand synthesis is constant and uninterrupted. However, when ribonucleotides are added to the reaction, permitting the primase to act on the lagging-strand template, the progression of the leading strand pauses briefly at defined positions. These positions correlate well with the preferred priming sites of the gp4 primase, and each pause lasts about five seconds, in good agreement with the RNA synthesis rate of gp4 primase. However, in five seconds the leading-strand DNA polymerase can synthesize about 1,000 bases of DNA. So pausing the machinery prevents an uncoupling of the synthesis of the two strands. The stalling mechanism remains unknown, although Lee et al. speculate that the primase domain may regulate the DNA-unzipping activity of gp4. Although T7 gp4 has coupled primase and helicase domains in one protein, the physical interaction of primase and helicase seen in bacteria makes it likely that this pausing might occur in a broad range of organisms, even if it requires two proteins.
Does all priming have to be performed by a specialized primase? Some clues may come from the unusual replication process carried out by ‘selfish DNA’ elements such as bacteriophage. Some of these elements have their own DNA replication machinery (such as phage T7, discussed above); but others, such as bacteriophage M13, co-opt the host cell's proteins to ensure their own replication. In fact, M13 subverts the host gene-expression machinery to initiate replication of its genome. The bacteriophage uses the bacterial RNA polymerase — usually employed in copying DNA into RNA destined to make protein — to produce the RNA primer for its replication instead. Normally, RNA polymerase peels the RNA off DNA, acting like a wire stripper, resulting in free single-stranded RNA and re-forming double-stranded DNA. But, to serve as a primer for M13 replication, the RNA produced by the polymerase must remain paired to DNA.
Severinov and colleagues (page 617)3 have uncovered the basis of this altered behaviour. The key lies in the fact that the M13 template is a single-stranded DNA. RNA polymerase usually acts on double-stranded DNA, which enters and leaves the enzyme molecule by specific channels, with only a short region unwound in the centre of the enzyme (Fig. 2). As the two DNA strands bind back together, a lid-like structure peels the RNA off and directs it out through a separate exit channel. But, because M13 is single stranded, there is no return to double-stranded DNA and the RNA–DNA hybrid becomes hyper-extended. This seems to force the hybrid molecule to reposition in the downstream channel of the RNA polymerase, with RNA extending beyond the body of the enzyme ready to act as a primer for DNA synthesis (Fig. 2). The authors propose that this may be a general property of the replication of a number of selfish elements in bacteria. Notably, several plant and animal viruses are proposed to extrude single-stranded regions at their DNA replication start sites, so perhaps RNA polymerase primes replication in some of these situations too.
Thus, bacteria and their viruses employ a range of strategies to effect priming, to couple priming events to the coordinated progression of the replication fork, and to ensure that DNA damage need not be an absolute barrier to fork progression. The power of these reconstituted bacterial and bacteriophage systems is apparent in the exquisite mechanistic detail that can be gleaned from their analysis. Similarly detailed studies of the more complex primases found in higher organisms5 are eagerly anticipated.
Heller, R. C. & Marians, K. J. Nature 439, 557–562 (2006).
Lee, J. -B. et al. Nature 439, 621–624 (2006).
Zenkin, N., Naryshkina, T., Kuznedelov, K. & Severinov, K. Nature 439, 617–620 (2006).
Kornberg, A. & Baker, T. A. DNA Replication 2nd edn (Freeman, New York, 1992).
Frick, D. N. & Richardson, C. C. Annu. Rev. Biochem. 70, 39–80 (2001).