DNA repair

DNA repair

How a cell finds and mends the damage its DNA suffers every day, by cutting out the damaged stretch of one strand and copying the other or by rejoining broken ends, how damage stops the cell cycle, and what happens when repair fails.

Contents4 sections

At a glance

  1. DNA is damaged all the time: each cell loses thousands of bases a day to its own chemistry, and oxygen radicals, UV light, chemicals, radiation and copying errors add more. Damage left unrepaired can become a mutation when the DNA is next copied.
  2. Damage to one strand is cut out, and the gap filled by copying the other strand: base excision repair for single damaged bases, nucleotide excision repair for bulky lesions such as UV dimers, mismatch repair for copying errors. A few lesions are simply reversed.
  3. A break in both strands is joined directly by non-homologous end joining, which may lose a few nucleotides, or mended exactly by homologous recombination, which copies the sister chromatid after replication.
  4. Damage halts the cell cycle, through ATM and p53, until it is repaired; if it is too great, the cell dies by apoptosis.
  5. Inherited faults in repair cause xeroderma pigmentosum, Lynch syndrome and the BRCA cancers. Radiotherapy, many cancer drugs and PARP inhibitors work by damaging DNA or blocking its repair.

Damage every day

DNA is stable, but not inert. In each of our cells, about 5,000 purines a day fall off their sugars, and about 100 cytosines lose their amino group and become uracil. Oxygen radicals made by respiration oxidise bases; ultraviolet light joins neighbouring pyrimidines into dimers that distort the helix; chemicals, such as those of tobacco smoke, add bulky groups to bases; and X-rays and gamma rays break the backbone. DNA polymerase, too, leaves the odd mismatch. A damaged base often pairs wrongly when the DNA is next replicated, and the change is then in both strands of one daughter molecule, a mutation (Gene mutations).

DamageExample causeMain repair
A lost or altered basespontaneous loss of purines, cytosine turning into uracil, oxidationbase excision repair
A methyl group on the O6 of guaninealkylating chemicals and drugsreversed by MGMT
A bulky lesion that distorts the helixUV light, chemicals of tobacco smokenucleotide excision repair
A mismatch or a small loopcopying errorsmismatch repair
A break in one strandoxygen radicals, radiationsingle-strand break repair, led by PARP
A break in both strandsradiation, a replication fork running into a nickend joining, homologous recombination
A crosslink between the two strandssome cancer drugs, such as cisplatinthe Fanconi anaemia pathway

The choice of thymine for DNA is thought to come from this. Because DNA should hold no uracil, the uracil made from cytosine is recognised as damage and removed (Nucleic acids). Methylated cytosine, though, turns into thymine, a normal base, and the T opposite a G is repaired less well, so methylated CG sites are hotspots of mutation.

lost from the DNA of each cell every day, about
5,000purines
turned into uracil in each cell every day, about
100cytosines
in copied DNA, thanks to mismatch repair
100 to 1,000times fewer errors
carry a mutated TP53 gene, about
halfof cancers

Mending one strand

Most damage affects one strand, and the other still holds the right sequence. The damage is cut out, a DNA polymerase fills the gap by copying the intact strand, and DNA ligase seals the nick. Tomas Lindahl, Aziz Sancar and Paul Modrich were awarded the 2015 Nobel Prize in Chemistry for working out the three main ways of doing this. A few lesions are simply reversed instead: MGMT removes a methyl group from the O6 of guanine.

Base excision repair fixes single bases that are damaged but do not distort the helix:

  1. 1

    The base is removed

    A DNA glycosylase, one of several, each for its own kind of altered base, cuts the base from its sugar and leaves an AP site, a sugar with no base. A purine that falls off leaves the same gap.
  2. 2

    The backbone is cut

    AP endonuclease cuts the backbone at the empty site.
  3. 3

    The gap is filled

    A DNA polymerase removes the sugar and puts in the right nucleotide, copying the other strand.
  4. 4

    The nick is sealed

    DNA ligase joins the new nucleotide to the strand.
Base excision repair
A short stretch of DNA drawn flat, as a ladder, in blue: the template strand along the top, running 3′ to 5′ from left to right, and the strand being mended along the bottom, running 5′ to 3′; each nucleotide a sugar ring, a phosphate ball marked P and a base marked with its letter. First, in the middle of the lower strand, a U sits where a C was, opposite a G. Next, a pale green enzyme, the glycosylase, has come up under the strand and holds the U in its pocket, cut from its sugar: an empty sugar is left in the strand. Then a light green enzyme, AP endonuclease, has cut the backbone on the left of the empty sugar, and a small ball, a free 3′-OH end, marks the cut. Next, a mid-green enzyme, DNA polymerase, has taken the empty sugar and its phosphate off, and they drift away below; a new C, paired with the G, stands in their place, and a nick remains on its right. Last, a deeper green enzyme, DNA ligase, has sealed that nick, and the strand is whole again. Step 1 of 5: Damage. A cytosine in the lower strand has lost its amino group and become uracil, which DNA should not hold. It sits opposite a guanine.PCPTPAPTPCPAPGPGPAPTPAPGPTPU5′3′5′3′

Step 1 of 5Damage. A cytosine in the lower strand has lost its amino group and become uracil, which DNA should not hold. It sits opposite a guanine.

Point at a name or a part to light it.

A damaged base is cut out and replaced in four steps: a glycosylase removes the base, AP endonuclease cuts the backbone beside the empty sugar, a polymerase puts in the right nucleotide, and a ligase seals the strand. The undamaged strand is the template throughout.

Simplified: DNA drawn flat, as a ladder, not as a double helix, and the hydrogen bonds between the bases left out; not to scale: the nucleotides drawn large and the enzymes as plain shapes, far smaller than they are; the uracil shown already cut free in the glycosylase's pocket; first the enzyme flips it out of the helix; the enzymes shown one at a time, and proteins that help some of them, such as XRCC1 and PARP1, left out; short-patch repair shown, the commonest kind; in long-patch repair 2 to 10 nucleotides are replaced.

3 sources
  1. Alberts B, Johnson A, Lewis J, et al. Molecular Biology of the Cell, 4th edition. Garland Science, 2002. DNA Repair.
  2. Cooper GM. The Cell: A Molecular Approach, 2nd edition. Sinauer Associates, 2000. DNA Repair.
  3. Krokan HE, Bjørås M. Base excision repair. Cold Spring Harb Perspect Biol. 2013;5(4):a012583.

Breaks in one strand are mended in much the same way. PARP1 binds the break, builds chains of ADP-ribose on itself and nearby proteins, and so brings the repair proteins in.

Nucleotide excision repair deals with bulky lesions, such as UV dimers. It recognises the distortion of the helix rather than any one lesion, so it can mend almost any large change. TFIIH, the factor that opens DNA at promoters (Transcription), unwinds the DNA round the lesion; nucleases cut the damaged strand on both sides of it, XPF with ERCC1 on the 5′ side and XPG on the 3′ side; and a short stretch, 12 or 13 nucleotides in bacteria and about 30 in our cells, is peeled away. Lesions in the strand of a gene that is being read are found first: RNA polymerase II stalls at them, and the stalled polymerase calls in the repair proteins, transcription-coupled repair.

Mismatch repair corrects the errors that DNA polymerase leaves behind, a wrong base or a small loop of extra or missing nucleotides in the new strand. It makes copying about a hundred to a thousand times more accurate. Proteins like the bacterial MutS (in our cells MSH2 with a partner) find the mismatch; proteins like MutL (MLH1 with PMS2) join them; and a stretch of the new strand, including the error, is removed and made again. The repair must take out the wrong base, not the right one, so it must know which strand is new. In E. coli, the old strand carries methylated adenines, which the new one gains only after a delay; our cells are thought to use the nicks still present in the new strand.

Mending a break in both strands

A break in both strands is the most dangerous damage: a chromosome can lose its end or be joined to the wrong partner. There are two ways of mending it.

  • Non-homologous end joining. The Ku protein binds the two ends and draws in a kinase, DNA-PKcs, that holds them in line; nucleases such as Artemis trim them, and DNA ligase IV joins them. It works at any stage of the cell cycle and needs no template, but it often loses or adds a few nucleotides at the join. Our cells also use it on purpose, to join the gene segments of antibodies and T-cell receptors (Antigen receptor genes), which is why a faulty Artemis causes severe combined immunodeficiency with sensitivity to radiation.
  • Homologous recombination. After replication, in S and G2, the sister chromatid holds an identical copy. The broken ends are trimmed back to leave single-stranded 3′ tails, BRCA2 loads the RAD51 protein onto them, and the coated strand finds the matching sequence on the sister, pairs with it and copies it, so the break is mended without loss (Homologous recombination).

Stopping the cycle

Damage also stops the cell cycle until it is mended (Cell cycle). Double-strand breaks activate the kinase ATM, and single-stranded DNA at stalled replication forks the related ATR. Among their targets is p53, a transcription factor called the guardian of the genome: it switches on repair genes and an inhibitor of the cyclin-dependent kinases, which holds the cell before S phase, or, if the damage is too great, sets off apoptosis (Cell death). Mutations in TP53 are the commonest genetic change in human cancers, found in about half of them. An inherited faulty copy causes most cases of Li–Fraumeni syndrome, which is autosomal dominant and carries a high risk of several cancers from childhood on, among them breast cancer and sarcomas of bone and soft tissue.

When a replication fork meets damage that has not been mended, special polymerases can copy past it, often less accurately: translesion synthesis. One of them, polymerase η, copies across UV dimers, mostly correctly, and people who lack it have a form of xeroderma pigmentosum.

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

1A cytosine in DNA loses its amino group and becomes uracil. How is this repaired?

2A child who freckled heavily on sun-exposed skin by the age of two has a basal cell carcinoma at eight. Which process is most likely faulty?

3How does the mismatch repair of E. coli know which base of a mismatched pair to replace?

4Why is a double-strand break more likely to be mended without any change in sequence after DNA replication than before it?

5Why does olaparib kill ovarian cancer cells with BRCA1 or BRCA2 mutations while sparing most normal cells?

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

See on the map

DNA replication

How copying errors and damage are mended.

  • DNA replicationBiopolymers: structure, synthesis and self-assemblyThe replication fork and its enzymes.
  • DNA replicationBiochemistryNot written yetThe same process, in more chemical detail.
  • 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.