Each time one of your cells divides, it copies about 6.4 billion base pairs in a few hours and leaves only about one uncorrected error per billion nucleotides. Watson and Crick saw how this could work as soon as they built their model: because the two strands are complementary, each strand holds all the information needed to rebuild the other. As they put it in 1953, "It has not escaped our notice that the specific pairing we have postulated immediately suggests a possible copying mechanism for the genetic material." The details turned out to be intricate, because the helix must be unwound without tangling, and the copying enzyme obeys a strict rule about which direction it can build. Understand that one directional rule and almost everything else in this topic follows from it.
Three hypotheses and the Meselson–Stahl experiment
Three models were proposed. In semiconservative replication, the parent strands separate and each serves as a template, so every new double helix has one old strand and one new strand. In conservative replication, the original helix stays intact and a brand-new copy is made. In dispersive replication, every strand ends up as a patchwork of old and new pieces. In 1958 Matthew Meselson and Franklin Stahl grew E. coli for many generations on a medium whose only nitrogen source was heavy nitrogen, . Every base contains nitrogen, so all the bacteria's DNA became heavy. They then moved the bacteria to ordinary medium and took samples after each round of replication. To tell heavy from light DNA they used cesium chloride density-gradient centrifugation: spun at very high speed, each DNA molecule settles at the level where its density matches the salt solution, so heavy, hybrid and light DNA form separate bands.
| Generations in 14N | Conservative predicts | Semiconservative predicts | Dispersive predicts | Observed |
|---|---|---|---|---|
| 0 | 1 heavy band | 1 heavy band | 1 heavy band | 1 heavy band |
| 1 | 1 heavy band + 1 light band | 1 intermediate (hybrid) band | 1 intermediate band | 1 intermediate band |
| 2 | 1 heavy band + 1 light band (3 times as much light) | 1 intermediate band + 1 light band | 1 single band between intermediate and light | 1 intermediate band + 1 light band |
Origins, bubbles and the replication fork
Replication begins at specific sequences called origins of replication, which tend to be rich in A–T pairs (they have only two hydrogen bonds each, so they separate more easily). The strands open into a replication bubble, and copying proceeds in both directions from the origin, so each bubble has two replication forks: Y-shaped regions where the parent helix is being unzipped. The circular chromosome of E. coli has a single origin; its two forks each travel at about 1000 nucleotides per second and meet on the far side of the circle. Eukaryotic chromosomes are far longer and their polymerases slower (about 50 nucleotides per second), so each chromosome has thousands of origins, and the bubbles grow until they merge.
| Protein | Job at the fork |
|---|---|
| Helicase | Unwinds the double helix by breaking the hydrogen bonds between base pairs; moves ahead of the fork. |
| Single-strand binding proteins (SSBs) | Coat the exposed single strands so they do not pair back up or fold on themselves, and protect them from nucleases. |
| Topoisomerase (DNA gyrase in bacteria) | Works ahead of the fork. Unwinding over-twists the DNA further along, like twisting a rope; topoisomerase cuts the backbone, lets the strain release, and rejoins it. |
| Primase | An RNA polymerase that builds a short RNA primer (about 5–10 nucleotides) on the template, giving DNA polymerase a free 3'-OH to start from. |
| DNA polymerase III | The main copying enzyme in prokaryotes. Adds DNA nucleotides to the 3' end of the primer or growing strand, so the new strand grows 5'→3' while the template is read 3'→5'. Also proofreads. |
| DNA polymerase I | Removes the RNA primers and replaces them with DNA nucleotides. |
| DNA ligase | Seals the last nick in the sugar-phosphate backbone by forming a phosphodiester bond, joining Okazaki fragments into a continuous strand. |
| DNA polymerase II | Mainly a repair enzyme; it can also restart replication that has stalled at damaged DNA. |
At each fork the two template strands point in opposite directions, but the fork moves only one way. On one template, DNA polymerase III can follow right behind the helicase and build continuously toward the fork; the new strand made this way is the leading strand, and it needs only one primer. On the other template, building 5'→3' means moving away from the fork. The polymerase makes a short stretch, then has to go back near the fork and start again from a new primer as more template is exposed. This strand, built in pieces, is the lagging strand, and the pieces are Okazaki fragments (about 1000–2000 nucleotides long in bacteria and 100–200 in eukaryotes), named after Reiji and Tsuneko Okazaki, who discovered them in the 1960s.
- Helicase exposes a new stretch of lagging-strand template.
- Primase lays down an RNA primer near the fork.
- DNA polymerase III extends the primer 5'→3', away from the fork, until it reaches the primer of the previous fragment.
- DNA polymerase I removes the old RNA primer and fills the gap with DNA.
- DNA ligase seals the remaining nick, joining the fragments into one strand.
Proofreading and repair
DNA polymerase inserts the wrong base roughly once in every 100 000 nucleotides. Without correction, every human cell division would add tens of thousands of mutations. Three layers of checking reduce the final error rate to about one in a billion, and they also fix damage that happens long after replication.
- Proofreading (during replication): DNA polymerases III and I check each base right after adding it. If it is mispaired, the enzyme backs up, removes it (its 3'→5' exonuclease activity) and tries again. This alone improves accuracy about 100-fold.
- Mismatch repair (just after replication): repair proteins scan the new DNA for mispaired bases that slipped past proofreading. In E. coli they can tell which strand is new because the parent strand already carries methyl tags. They cut out a section of the new strand around the error, and a polymerase and ligase rebuild it correctly.
- Excision repair (any time): damage such as UV-induced thymine dimers or chemically altered bases distorts the helix. A nuclease cuts out the damaged stretch, DNA polymerase (pol I in bacteria) fills the gap using the undamaged strand as a template, and ligase seals it. People with the inherited disorder xeroderma pigmentosum lack a working excision-repair enzyme and develop skin cancers after very little sun exposure.