DNA replication

DNA replication

How a cell copies its DNA before it divides. Each strand is the template for a new one, two forks run from every origin, and the ends of our chromosomes need an enzyme of their own.

The eukaryotic replication fork
A replication fork in section, moving to the right. The parental double helix comes in from the right into the CMG helicase, a ring drawn cut open lengthwise so the two parental strands can be seen parting inside it. Behind CMG, Pol ε copies the leading-strand template, and the new leading strand leaves to the left along the bottom as a double helix, through PCNA. The lagging-strand template leaves CMG past primase, at the front of Pol α–primase, which lies along CMG’s top, and runs left above it; RPA coats it where it is single-stranded. Along it lie Okazaki fragments, each begun by a short RNA primer; PCNA with Pol δ, then FEN1 and DNA ligase I, work on them in turn. Step 1 of 6: CMG unwinds, Pol ε copies. CMG moves 3′ to 5′ along the leading-strand template and pulls the parental strands apart; Pol ε extends the leading strand without a break, and RPA coats the exposed lagging-strand template.3′5′3′5′5′3′

Step 1 of 6CMG unwinds, Pol ε copies. CMG moves 3′ to 5′ along the leading-strand template and pulls the parental strands apart; Pol ε extends the leading strand without a break, and RPA coats the exposed lagging-strand template.

Point at a name or a part to light it.

Both new strands grow 5′ to 3′. The leading strand is made continuously, towards the fork; the lagging strand is made away from it, in Okazaki fragments of 100–200 nt, each begun by an RNA primer that is later removed. Each daughter duplex keeps one parental strand: replication is semi-conservative.

Simplified: the fork moves on 84 bp over these steps; in the cell it moves on about a fragment’s length for each Okazaki fragment; the fragment drawn 168 nt long (a whole number of turns), about the 165 nt measured in yeast; proteins without a measured size drawn as lobes the size of the smallest sphere that could hold their mass (real proteins are a little larger); their shapes and the places of their subunits approximate; Pol α’s core, flexible until it takes the primer, drawn in one place beside the complex; CMG, PCNA and ligase I cut open to show the DNA inside them; CMG’s channel drawn to fit the DNA it holds, not measured; both arms drawn straight and in one plane; the leading-strand duplex drawn in line with the parental duplex, where it leaves Pol ε and PCNA at about a right angle to it; the single-stranded template along the arm drawn at the 0.34 nm per nucleotide it has once paired, elsewhere at 0.6 nm; the RNA–DNA primer drawn as B-DNA; the DNA drawn straight where FEN1 bends it by about 100°; the DNA FEN1 holds drawn without its 1-nt 3′ flap; the flap drawn 5 nt long so it can be seen, where most that FEN1 cuts are 1 nt; the DNA drawn as plain B-DNA where ligase I partly unwinds it; ligase I shown replacing Pol δ and FEN1 on PCNA, which can hold more than one of them at once; RFC, topoisomerases, nucleosomes and other replisome proteins left out; the N-terminal regions of POLA1, POLE2 and ligase I left out.

32 sources
  1. Burgers PMJ, Kunkel TA. Eukaryotic DNA replication fork. Annu Rev Biochem 2017;86:417–38. Roles, primer lengths, flaps, nick translation, ligation
  2. UniProt: MCM2–7 (P49736, P25205, P33991, P33992, Q14566, P33993), CDC45 (O75419), GINS1–4 (Q14691, Q9Y248, Q9BRX5, Q9BRT9). Masses
  3. UniProt: POLE, POLE2–4 (Q07864, P56282, Q9NRF9, Q9NR33). Masses, leading strand
  4. UniProt: POLA1, POLA2, PRIM1, PRIM2 (P09884, Q14181, P49642, P49643). Masses, primer synthesis; PRIM2’s interdomain linker 253–270, its C-terminal part moving away with the primer’s 5′ end
  5. UniProt: RPA1–3 (P27694, P15927, P35244). Masses
  6. UniProt: PCNA (P12004), RFC1 (P35251). Homotrimer, loading by RFC
  7. UniProt: POLD1–4 (P28340, P49005, Q15054, Q9HCU8). Masses, lagging strand
  8. UniProt: FEN1 (P39748). Mass, 5′ flap cleavage
  9. UniProt: LIG1 (P18858). Mass, ATP-dependent nick sealing, residues 1–270 predicted disordered
  10. Remus D et al. Cell 2009;139:719–30. Mcm2–7 double hexamer 150 × 230 Å in four ring-like tiers
  11. Choi JM et al. Genes Dev 2007;21:1316–21. Human GINS: 55 × 106 × 61 Å
  12. Yuan Z et al. Nat Struct Mol Biol 2016;23:217–24. CMG structure: GINS and Cdc45 scaffold the N tier (PDB 3JC5, 3JC6)
  13. Abid Ali F et al. Nat Commun 2016;7:10708. Cdc45 and GINS on Mcm2, Mcm3 and Mcm5
  14. Georgescu R et al. PNAS 2017;114:E697–706. CMG at a fork: the N tier leads, the duplex enters its channel, Pol ε below CMG, Pol α–primase on top (PDB 5U8S)
  15. Yuan Z et al. Nat Commun 2020;11:688. DNA unwinding by CMG: strands part at the bottom of the zinc-finger collar (about 30 Å high); the lagging strand diverted, perhaps out between the zinc fingers of Mcm3 and Mcm5
  16. Yuan Z et al. Nat Commun 2020;11:3156. Pol ε holoenzyme on CMG: about 100 Å (16 nt) of single strand to the active site, POLE3–4 bridging the two Pol modules, POLE2 guiding the strand
  17. Asturias FJ et al. Nat Struct Mol Biol 2006;13:35–43. Pol ε: globular catalytic domain and an extended part
  18. Jones ML, Aria V, Baris Y, Yeeles JTP. Mol Cell 2023;83:2911–24. Pol α–primase binds the leading edge of CMG through PRIM2 and extends from the N-tier face, atop Mcm3 and Mcm5, toward the C tier beside Mcm3 and GINS; PRIM1 above the exit for the lagging-strand template, which runs alongside it
  19. Baranovskiy AG et al. Nucleic Acids Res 2022;50:6264–70. An elongated platform of PRIM1, PRIM2’s N-terminal domain, POLA1’s C-terminal domain and POLA2; PRIM2’s C-terminal domain on the primer’s 5′ end while Pol α’s core extends its 3′ end; linkers PRIM2 253–270 and POLA1 1250–1267
  20. Baranovskiy AG et al. Nucleic Acids Res 2014;42:14013–21. Human Pol α catalytic core: residues 335–1257 crystallised, 338–1244 ordered
  21. Kim C, Paulus BF, Wold MS. Biochemistry 1994;33:14197–206. RPA binds about 30 nt
  22. Krishna TSR et al. Cell 1994;79:1233–43. PCNA, a closed ring round DNA
  23. Dieckman LM et al. Subcell Biochem 2012;62:281–99. PCNA 8 nm across, 3.5 nm hole, 3 nm thick
  24. Lancey C et al. Nat Commun 2020;11:1109. Human Pol δ on PCNA: the catalytic core on PCNA, the regulatory subunits to the side; FEN1 on a free PCNA protomer
  25. Tsutakawa SE et al. Cell 2011;145:198–211. FEN1 structures and flap cleavage: a 3′ flap of one nucleotide held in a pocket, the DNA bent by about 100° at the junction
  26. Pascal JM et al. Nature 2004;432:473–8. DNA ligase I encircles nicked DNA
  27. Smith DJ, Whitehouse I. Nature 2012;483:434–8. Okazaki fragments in yeast the length of the nucleosome repeat, about 165 nt
  28. Alberts B et al. Molecular Biology of the Cell, 4th ed. Garland Science 2002, chapter 5, DNA Replication Mechanisms. Okazaki fragments 100–200 nt long in eukaryotes, RNA primers about 10 nt
  29. Erickson HP. Biol Proced Online 2009;11:32–51. Minimal radius of a protein of given mass
  30. Wang JC. Helical repeat of DNA in solution. Proc Natl Acad Sci USA 1979;76:200–3 (10.4 ± 0.1 bp per turn); Rhodes D, Klug A. Helical periodicity of DNA determined by enzyme digestion. Nature 1980;286:573–8 (10.6 ± 0.1). B-DNA in solution: about 10.5 bp and 3.57 nm per turn
  31. Arnott S, Hukins DWL. Biochem Biophys Res Commun 1972;47:1504–9 (fibre); Dickerson RE et al. Science 1982;216:475–85 (crystal). B-DNA about 2 nm across, 0.34 nm rise per base pair
  32. Drew HR et al. Proc Natl Acad Sci USA 1981;78:2179–83, PDB 1BNA, on which the strands measure about 145° apart across the minor groove, so it spans about 0.4 of a turn

At a glance

  1. Replication is semi-conservative: each new double helix keeps one parental strand and has one new one, as Meselson and Stahl showed in 1958.
  2. It starts at origins: one on the circular chromosome of E. coli, tens of thousands along ours. Two forks move away from each origin.
  3. DNA polymerases add nucleotides only to a 3′ end, and those that copy chromosomes cannot start a chain, so they need a primer. At each fork the leading strand is made in one piece, the lagging strand in Okazaki fragments, each begun by an RNA primer that is later removed.
  4. In our cells the CMG helicase unwinds, Pol α–primase lays primers, Pol ε makes most of the leading strand and Pol δ most of the lagging strand, held on by the PCNA clamp; FEN1 removes the primers and DNA ligase I seals the nicks.
  5. Proofreading and mismatch repair leave about one error in a billion nucleotides copied. Telomerase rebuilds chromosome ends, which would otherwise shorten with every division.

Each strand is a template

Watson and Crick saw at once that base pairing suggested a way of copying: each strand could serve as the template for a new partner. In 1958 Matthew Meselson and Franklin Stahl tested how the strands are shared out. They grew E. coli for many generations on heavy nitrogen, ¹⁵N, so that all its DNA was dense, moved it to normal nitrogen, ¹⁴N, and spun its DNA in a caesium chloride gradient, in which DNA settles according to its density.

ModelAfter one division in ¹⁴NAfter two
Conservative: the parental helix stays whole and the copy is all newa heavy band and a light banda heavy band and a light band
Semi-conservative: each new helix keeps one parental strandone band, between heavy and lightan intermediate band and a light band
Dispersive: old and new pieces mixed along both strandsone intermediate bandone band, between intermediate and light

They saw one intermediate band after one division, then an intermediate and a light band: replication is semi-conservative.

What a DNA polymerase needs

  • A template. The polymerase reads the template strand 3′ to 5′ and pairs each incoming nucleotide with it.
  • A primer. No DNA polymerase that copies a chromosome can start a new chain; it can only add to the 3′-OH of a chain already paired with the template. A primase makes short RNA primers for it.
  • The four deoxynucleoside triphosphates, dATP, dGTP, dCTP and dTTP. The polymerase joins each to the 3′-OH and releases pyrophosphate, whose breakdown drives the reaction forwards. So DNA grows 5′ to 3′ only (Nucleic acids).

The polymerase also checks its work. A wrong pair leaves the new end sitting badly in the active site, and a second site, a 3′ to 5′ exonuclease, cuts the wrong nucleotide off: proofreading. A polymerase alone puts in about one wrong nucleotide in 100,000; proofreading improves this about a hundredfold, and mismatch repair, working just behind the fork, corrects most of what is left (DNA repair).

copied at each fork of a bacterium, about
1,000base pairs a second
copied at each fork in our cells, about
50base pairs a second
in an Okazaki fragment of our cells
100 to 200nucleotides
copied wrongly in the end, about
1 in a billionnucleotides

Origins and forks

E. coli has one origin on its circular chromosome of 4.6 million base pairs, a stretch of about 245 base pairs rich in A–T pairs, which come apart more easily. An initiator protein, DnaA, binds it and opens the helix; a loader, DnaC, then puts a ring of the DnaB helicase onto each separated strand, and two replication forks move off in opposite directions round the circle until they meet on the far side. Each fork moves on by about 1,000 base pairs a second, so the whole chromosome is copied in about 40 minutes.

Our chromosomes are far longer and our forks slower, so copying the largest chromosome from a single origin would take weeks. Each chromosome has many origins instead, tens of thousands in all, and S phase ends when the forks from neighbouring origins have met.

An origin must fire only once in each cell cycle, or stretches of the genome would be copied twice. In G1, the origin recognition complex (ORC), helped by Cdc6 and Cdt1, loads two inactive rings of the MCM2–7 helicase onto each origin: the origin is licensed. In S phase, two kinases, CDK and DDK, set off the loading of Cdc45 and GINS onto each ring, making the active CMG helicase, and the two rings move apart as the helicases of two forks. New rings cannot be loaded until the next G1, so no origin fires twice.

At the fork

The polymerases of the two new strands face opposite ways. The strands of the parental helix are antiparallel and DNA is made only 5′ to 3′, so one new strand, the leading strand, grows towards the fork in one piece from a single primer. The other, the lagging strand, grows away from the fork in short pieces, Okazaki fragments, each begun by its own RNA primer as the fork opens more template. The figure follows one fragment in our cells:

  1. 1

    CMG unwinds, Pol ε copies

    CMG, a ring of the six MCM proteins with Cdc45 and GINS, moves 3′ to 5′ along the leading-strand template and pulls the parental strands apart. Pol ε extends the leading strand without a break, and RPA coats the single-stranded lagging-strand template.
  2. 2

    A primer is laid

    Pol α–primase makes an RNA primer of about 10 nucleotides on the lagging-strand template, and Pol α extends it with about 20 to 30 nucleotides of DNA.
  3. 3

    PCNA and Pol δ take over

    A clamp loader, RFC, closes the ring of PCNA round the primer's 3′ end, and Pol δ, held on by PCNA, replaces Pol α, which cannot proofread.
  4. 4

    Pol δ makes the fragment

    Pol δ extends the Okazaki fragment away from the fork until it reaches the 5′ end of the previous fragment.
  5. 5

    The old primer is removed

    Pol δ runs on into the previous fragment and lifts its 5′ end into short flaps, mostly a single nucleotide, and FEN1 cuts each one off until all of the primer's RNA is gone.
  6. 6

    The nick is sealed

    DNA ligase I joins the two fragments into one strand, using ATP.

Unwinding twists the helix ahead of the fork ever tighter. Topoisomerases relieve the strain: topoisomerase I cuts one strand and lets it swivel, and topoisomerase II cuts both strands, passes another stretch of the double helix through the gap and seals it again. In bacteria, a topoisomerase II called DNA gyrase removes the twisting ahead of the forks, and topoisomerase IV separates the two new chromosomes, which end up linked like the rings of a chain.

Our DNA is also wrapped round nucleosomes, which are taken apart just ahead of the fork. Behind it, the old histones are shared out between the two new helices, and newly made histones fill the gaps.

JobE. coliOur cells
UnwindingDnaB helicase, a ring round the lagging-strand templateCMG helicase, a ring round the leading-strand template
Holding single strands apartsingle-strand binding proteinreplication protein A (RPA)
Laying primersprimase (DnaG)Pol α–primase
Making the new strandsDNA polymerase III, on both strandsmostly Pol ε on the leading strand and Pol δ on the lagging strand
Keeping the polymerase onthe β clampPCNA
Removing primersDNA polymerase I, whose 5′ to 3′ exonuclease removes the RNA as it fills the gapFEN1, cutting the flaps Pol δ lifts
Sealing the nicksDNA ligase, using NAD⁺DNA ligase I, using ATP
Okazaki fragments1,000 to 2,000 nucleotides100 to 200 nucleotides

Our cells have other DNA polymerases too. Pol γ copies the small circular DNA of mitochondria, and Pol β fills the one-nucleotide gaps of base excision repair (DNA repair).

The ends of our chromosomes

A linear chromosome has a problem at each end. Once the last primer of the lagging strand near the tip is removed, no DNA can be made to fill its place, since there is no 3′ end beyond it to extend, so every round of replication would leave the chromosome a little shorter. Most bacteria avoid this by having circular chromosomes.

Our chromosomes end in telomeres, a short sequence, TTAGGG, repeated hundreds to thousands of times and bound by a set of proteins, shelterin, that stops the ends being treated as broken DNA. Telomerase lengthens them. It is a reverse transcriptase that carries its own RNA template and adds TTAGGG repeats to the 3′ end of the chromosome; primase and DNA polymerase then fill in the other strand (RNA replication. Reverse transcription). Elizabeth Blackburn, Carol Greider and Jack Szostak were awarded the 2009 Nobel Prize in Physiology or Medicine for discovering how telomeres and telomerase protect chromosomes.

Telomerase is active in germ cells and stem cells, but not in most body cells, whose telomeres shorten a little with each division; once they are very short, a cell stops dividing. Many cancer cells have high telomerase activity, which helps them divide without limit.

Check yourselfAnswer, then say how sure you were: a lucky guess still counts as a weak point.

1Bacteria grown for many generations on heavy nitrogen (¹⁵N) are moved to normal nitrogen (¹⁴N) and allowed to divide once. What does their DNA show when it is spun in a density gradient?

2Why is the lagging strand made in short pieces?

3In Escherichia coli, which enzyme removes the RNA primers of Okazaki fragments and fills the gaps with DNA?

4What does telomerase use as the template for the repeats it adds to a chromosome's end?

5Ciprofloxacin kills bacteria by acting on which enzymes?

Linked topicsThe concepts this topic shares, and the topics they tie it to.

See on the map

Folate and vitamin B12

Thymidine, one of DNA’s building blocks, is made with folate’s help.

  • Vitamins: clinical applicationsPharmacologyNot written yetVitamin B12 and folic acid as treatments.
  • Haematopoietic drugsPharmacologyNot written yetHydroxocobalamin and folic acid restore normal blood formation.
  • SulfonamidesPharmacologySulfonamides stop bacteria making folate, which we take from food; trimethoprim blocks a later step.
  • Antitumor chemotherapeuticalsPharmacologyNot written yetMethotrexate blocks dihydrofolate reductase.
  • Macrocytic and megaloblastic anemiasInternal medicineA lack of B12 or folate gives large red cells, from megaloblasts in the marrow; lack of B12 can also damage nerves.
  • Rheumatoid arthritisInternal medicineFolic acid is given with methotrexate to cut its side effects.

DNA replication

The replication fork and its enzymes.

  • DNA replicationBiochemistryNot written yetThe same process, in more chemical detail.
  • DNA repairBiopolymers: structure, synthesis and self-assemblyHow copying errors and damage are mended.
  • QuinolonesPharmacologyBlock bacterial topoisomerases: DNA gyrase and topoisomerase IV.
  • Antitumor chemotherapeuticalsPharmacologyNot written yetMany cytotoxic drugs damage DNA or stop it being copied, so they hit dividing cells hardest.
  • Antiviral drugsPharmacologyNot written yetAciclovir, a nucleoside analogue, stops herpesvirus DNA polymerase.