DNA replication

Three ways you could copy a double helix

Before 1958 there were three competing ideas. In conservative replication, the original double helix stays intact and a completely new copy is built beside it. In semiconservative replication, the two strands separate and each becomes half of a new double helix, so every daughter molecule has one old strand and one new one. In dispersive replication, both daughter molecules are patchworks of old and new pieces along each strand.

Conservative copying feels natural, because when you photocopy a page the original stays whole. But the base-pairing rules (A with T, G with C) mean each strand already carries all the information needed to rebuild the other. The question was which model cells actually use.

The Meselson–Stahl experiment

Meselson and Stahl grew E. coli for many generations in a medium whose only nitrogen was the heavy isotope ¹⁵N. Nitrogen is part of every DNA base, so all the bacteria's DNA became slightly heavier than normal. Then they moved the cells into a medium containing ordinary, lighter ¹⁴N and took samples after each round of division.

To tell heavy DNA from light DNA, they spun it at high speed in a solution of caesium chloride (CsCl). During the spin the salt forms a density gradient, denser at the bottom of the tube than at the top, and each DNA molecule settles as a band at the level that matches its own density.

1. 1Before the switch: one band, at the heavy (¹⁵N/¹⁵N) position.
2. 2After one generation in ¹⁴N: one band, halfway between heavy and light. That rules out conservative replication, which would have given one heavy band and one light band.
3. 3After two generations: two bands of equal size, one at the intermediate position and one at the light (¹⁴N/¹⁴N) position. That rules out dispersive replication, which predicts a single band drifting steadily lighter.

Only the semiconservative model predicts both results. After one generation every molecule is a hybrid: one old heavy strand plus one new light strand. In the next generation each hybrid separates; its heavy strand gets a new light partner (hybrid again) and its light strand does too (fully light).

Opening the helix: helicase, SSB and topoisomerase

Replication begins at specific sequences called origins of replication. E. coli has a single origin on its circular chromosome. Eukaryotic chromosomes are far longer, so each has many origins; a human cell may use tens of thousands across its genome. From each origin, copying runs in both directions, creating two Y-shaped replication forks.

1. 1Helicase moves along the DNA and breaks the hydrogen bonds between base pairs, separating the two strands.
2. 2Single-strand binding proteins coat the exposed strands so they don't snap back together before they're copied.
3. 3Unwinding a twisted rope makes the part ahead of your hands twist even tighter. The same happens to DNA: as helicase unwinds, the helix ahead of the fork becomes overwound, or supercoiled. Topoisomerase relieves this strain by cutting the DNA, letting it untwist, and resealing it. In bacteria the main enzyme doing this job is called DNA gyrase.

The golden rule: polymerase only builds 5′→3′

Each DNA strand has a direction. One end has a free phosphate on the 5′ carbon of its sugar (the 5′ end), and the other has a free –OH group on the 3′ carbon (the 3′ end). The two strands of a double helix run in opposite directions, which is what antiparallel means.

DNA polymerase builds a new strand by reading the template and adding matching nucleotides. It can only attach a new nucleotide to an existing 3′-OH group. So a new strand always grows in the 5′→3′ direction, and the polymerase reads its template 3′→5′. It can never work the other way.

There's a second catch: DNA polymerase can't start a strand from nothing. It needs a short stretch that's already paired to the template, with a 3′ end it can extend. That starter is a primer, a short piece of RNA (roughly 5–10 nucleotides in bacteria) made by an enzyme called primase. Unlike DNA polymerase, primase can start from scratch.

Leading and lagging strands

Put the golden rule together with antiparallel strands and you get a problem. The fork opens in one direction only. On one template, 5′→3′ synthesis points the same way the fork is moving, so the polymerase can follow helicase and build one long, continuous strand. That's the leading strand, and it needs just one primer.

On the other template, 5′→3′ synthesis points away from the fork. The polymerase can only build backwards, away from where the DNA is opening. So as the fork opens up a fresh stretch, primase lays down a new primer near the fork and polymerase extends it back until it meets the previous piece. This strand is made in short pieces called Okazaki fragments, and it's called the lagging strand. Okazaki fragments are about 1,000–2,000 nucleotides long in E. coli and only about 100–200 in eukaryotes.

1. 1Primase makes an RNA primer on the lagging-strand template near the fork.
2. 2The main replicative polymerase (DNA polymerase III in E. coli) extends the primer 5′→3′, away from the fork, making an Okazaki fragment. A ring-shaped sliding clamp keeps the polymerase attached to the DNA as it works.
3. 3Another polymerase (DNA polymerase I in E. coli) removes the RNA primer of the fragment ahead and fills the gap with DNA.
4. 4DNA ligase seals the last nick in the backbone by forming a phosphodiester bond, joining the fragments into one continuous strand.

Keeping mistakes rare

DNA polymerase is fast: about 1,000 nucleotides per second in E. coli and about 50–100 per second in eukaryotic cells such as yours. It occasionally inserts the wrong base, but three layers of checking catch almost all of these errors.

• Selection: the polymerase's active site fits a correctly paired base much better than a mismatched one.
• Proofreading: if a wrong base is added, the polymerase backs up, cuts it out and tries again.
• Mismatch repair: after replication, other enzymes scan the new DNA for mismatches, cut out a stretch of the new strand and resynthesise it. In E. coli they recognise the new strand because it hasn't been chemically tagged with methyl groups yet.

Together these bring the final error rate down to roughly one wrong base per billion (10⁹) copied, or even fewer. Estimates vary between sources and organisms, but the point stands: the rare errors that remain are one source of the mutations you'll meet in a later topic.

The trouble with chromosome ends

Linear eukaryotic chromosomes have a problem bacteria with circular chromosomes don't. At the very tip of the lagging strand, once the last RNA primer is removed, there's no upstream 3′-OH for polymerase to extend into the gap. So each round of replication leaves the chromosome slightly shorter. This is the end-replication problem.

The ends are protected by telomeres, long runs of a repeated sequence (TTAGGG in humans) that contain no genes, so losing a little of them does no harm at first. In cells that must keep dividing, such as germ cells and many stem cells, an enzyme called telomerase extends the telomeres using an RNA template that it carries inside itself. Elizabeth Blackburn, Carol Greider and Jack Szostak shared the 2009 Nobel Prize for this discovery.

Key terms

Semiconservative replication

Each new double helix has one parental strand and one newly made strand.

Helicase

Enzyme that unwinds DNA at the fork by breaking hydrogen bonds between base pairs.

Topoisomerase

Enzyme that relieves supercoiling ahead of the fork by cutting and resealing DNA.

Primase / primer

The enzyme that makes a short RNA starter / the starter itself, whose 3′ end DNA polymerase extends.

DNA polymerase

Enzyme that adds nucleotides to a free 3′-OH, so new DNA grows only 5′→3′.

Leading / lagging strand

The new strand made continuously toward the fork / the one made in pieces away from it.

Okazaki fragment

A short stretch of lagging-strand DNA, each started from its own primer.

DNA ligase

Enzyme that seals nicks in the sugar–phosphate backbone, joining Okazaki fragments.

Telomere / telomerase

Repeated DNA at chromosome ends / the enzyme that extends it using its own RNA template.

Check yourself

Try answering in your head before you open each answer.

• 1.A drug inactivates DNA ligase in bacteria. Which new strand, leading or lagging, would be affected most, and what would you find if you extracted and sized the newly made DNA?Show answerHide

  The lagging strand. Its Okazaki fragments could still be made but not joined, so the new DNA would include many short pieces of roughly 1,000–2,000 nucleotides. The leading strand is mostly one long piece and needs ligase far less.

• 2.Suppose Meselson and Stahl had seen only one band, at the intermediate position, after both one and two generations. Which model would that support, and why?Show answerHide

  Dispersive replication (or at least not semiconservative). If every molecule is a mix of old and new pieces, they all have similar densities, giving one band. Semiconservative replication instead predicts a separate light band appearing in the second generation.

• 3.Why does the lagging strand need many primers while the leading strand needs only one? Use the direction of DNA polymerase in your answer.Show answerHide

  Polymerase can only add to a 3′-OH, so it builds 5′→3′. On the leading strand that direction follows the opening fork, so one primer lets polymerase keep going. On the lagging strand that direction points away from the fork, so each time more template is exposed a new primer must be laid down and extended backwards.