DNA repair

An APE that proofreads

An enzyme involved in one aspect of DNA repair is now shown to have another function that improves the accuracy of repair. The new findings should help to improve therapies for HIV.

'Base excision repair' is one of the ways in which damaged DNA is fixed. The enzyme DNA polymerase β has an important role in this process, but it tends to make mistakes. In their paper on page 655 of this issue1, Chou and Cheng show how those mistakes are corrected. This in itself is striking. But their research also provides pointers to improving the effectiveness of HIV treatments by making them less toxic to the patient.

The job of base excision repair is to remove damaged bases from genomic DNA, as well as the sugar–phosphates that remain after the loss of bases2. The process is outlined in Fig. 1. The trouble is that polymerase β, which is responsible for inserting the correct nucleotide, is embarrassingly error prone, making on average one mistake per 4,000 base insertions3. Assuming that between 20,000 and 80,000 base modifications take place daily in our genome4, polymerase β would be expected to introduce around ten mutations into our DNA each day.

Figure 1: Base excision repair in human cells.
figure1

a, In this example, a uracil residue has arisen spontaneously from cytosine through deamination, and needs to be replaced by another cytosine. b, The uracil is excised by the enzyme DNA glycosylase to generate an apurinic or apyrimidinic (AP) site, which in turn (c) is cleaved at its 5′ end by the AP endonuclease APE1. The repair process can then take one of two routes. d, The resulting single-nucleotide gap is filled in by polymerase β, which incorporates the dCMP-nucleotide carrier of cytosine, and also removes the sugar–phosphate flap (shown hanging from the lower strand). e, To complete successful repair, the remaining nick is sealed by a DNA ligase enzyme (ligation). The alternative route f–h occurs when polymerase β errs, and the events identified by Chou and Cheng1 occur. f, dTMP, the carrier of thymine, has been erroneously inserted into the gap. g, It is recognized and removed by the 3′→5′ proofreader, APE1. h, Polymerase β then correctly fills the gap by incorporating a dCMP residue, after which ligation completes the process as normal.

Why this does not happen has long been a puzzle, but Chou and Cheng1 have come up with a plausible answer. They show that the human apurinic or apyrimidinic (AP) nuclease APE1, an enzyme long known to function upstream from polymerase β in base excision repair (Fig. 1), and to interact with it physically5, can also act as its proofreading exonuclease by removing wrongly inserted nucleotides from the repair patches. In other words, an enzyme that normally acts by cleaving the DNA in intact strands, by hydrolysing the phosphate groups joining two sugar residues, can also act as an exonuclease that 'nibbles' off nucleotides from DNA ends that do not terminate in correct base pairs.

DNA bases can be damaged by many agents4. Some damaged bases are lost from DNA spontaneously; others are removed enzymatically by DNA glycosylases (Fig. 1). In both cases the result is an AP site in DNA, the removal of which is initiated by AP endonucleases, known as APEs. These enzymes cleave the sugar–phosphate backbone at AP sites to generate a free 3′-OH terminus upstream from the site of base loss. The sugar–phosphate fragment that remains transiently attached to the 5′-terminus of the incised strand is then removed by polymerase β when the missing nucleotide is inserted6.

In cases where the polymerase inserts the right nucleotide — that is, one that forms the correct (Watson–Crick) base pair with the nucleotide in the other strand — another enzyme, DNA ligase, completes the process by sealing the remaining nick2. But polymerase β often inserts the wrong nucleotide3. In this case, the ligase cannot seal the remaining nick because the 3′-OH and the 5′-phosphate termini are misaligned.

In the bacterium Escherichia coli, the polymerase (pol I) that acts in base excision repair is different, and misinsertions do not pose a problem. That's because pol I has a 3′→5′ proofreading exonuclease activity that can remove the mispair and resynthesize the correct nucleotide. But polymerase β has no associated proofreading activity, so it cannot catalyse the removal and replacement of the wrong nucleotide.

Chou and Cheng1 now show that a nicked double-stranded DNA, with a mispair on the 3′ side of the strand discontinuity, is the best substrate for the 3′→5′ exonuclease activity of APE1. This activity decreased when a substrate with a single nucleotide gap was used, and diminished further with a bigger gap. It seems likely, then, that APE1 has a proofreading function in base excision repair.

The APE1 enzyme has been known for many years and has been studied in detail by several groups. So how did its proofreading ability escape detection until now? The most likely answer is simply that no one looked for it, as there was no reason to suspect that it had this second function. The 3′→5′ exonuclease action of APE1 had been identified, along with other activities7. But the exonuclease action was much weaker than the endonuclease activity, and it was thought unlikely to be biologically significant. Moreover, the enzyme was known to act upstream from polymerase β, where there is no need for an exonuclease, so there seemed to be no compelling reason to test it on a substrate that might arise through erroneous gap- filling by polymerase β.

Chou and Cheng's discovery1 was fortuitous, because they had originally tackled a different task. They set out to identify a detoxifying enzyme that removed deoxyribonucleoside drugs from the DNA of leukaemic cells. (Leukaemias should be particularly sensitive to this type of therapy, because the cells affected proliferate rapidly and incorporate much more of these toxic drugs into their DNA than do other human cells.) When the detoxifying enzyme turned out to be APE1, the link with base excision repair was established, and it was only a short step to ask what the physiological substrates of this potent 3′→5′ activity might be.

But the authors did not forget the link of APE1 to the processing of nucleotide analogues. They also show that this proofreading exonuclease efficiently removes AZT and D4T (chain terminators used to inhibit HIV replication) from the 3′-termini of nicked DNA substrates. This finding could be of therapeutic importance.

Dideoxynucleoside drugs and their analogues — such as ddC, ddI, AZT, D4T and T3C — are not usually incorporated into genomic DNA in large amounts. But they do get into our genome through the action of polymerase β, which can incorporate them into repair patches during base excision repair8, and through another polymerase (γ), which inserts them into mitochondrial DNA9. Both of these events should cause problems, as incorporating these substances into DNA generates strand breaks that cannot be joined. Thus, by removing them from genomic DNA, APE1 helps our cells to survive.

In contrast, because DNA synthesis by HIV takes place in the cytoplasm, where there is no APE1, chain terminators incorporated into the proviral DNA by the HIV enzyme reverse transcriptase are not removed. As a result, synthesis of full-length HIV DNA can be inhibited. So, by improving the efficiency of APE1-mediated removal of chain terminators from genomic and mitochondrial DNA, it should be possible to reduce the side effects of HIV therapy.

The APE1-catalysed removal of nucleoside analogues from DNA termini is also interesting from the structural point of view. All of these substances form standard Watson–Crick pairs with the nucleosides in the complementary strand. This implies that, rather than seeing the mispairs as such, APE1 may be detecting subtle distortions of the double helix at the nick, characterized by misalignments of the strand ends that make it impossible to rejoin them.

Gorillas that have learned to use sign language remain a rarity, even though they pop up on television from time to time. The newly discovered proofreading APEs, by contrast, are clearly commonplace and, in contrast to signing gorillas, likely to be of considerable clinical importance.

References

  1. 1

    Chou, K.-M & Cheng, Y.-C Nature 415, 655–659 (2002).

    CAS  Article  Google Scholar 

  2. 2

    Lindahl, T. Mutat. Res. 462, 129–135 (2000).

    CAS  Article  Google Scholar 

  3. 3

    Osheroff, W. P. et al. J. Biol. Chem. 274, 3642–3650 (1999).

    CAS  Article  Google Scholar 

  4. 4

    Lindahl, T. Nature 362, 709–715 (1993).

    CAS  Article  Google Scholar 

  5. 5

    Bennett, R. A. et al. Proc. Natl Acad. Sci. USA 94, 7166–7169 (1997).

    CAS  Article  Google Scholar 

  6. 6

    Sobol, R. W. et al. Nature 405, 807–810 (2000).

    CAS  Article  Google Scholar 

  7. 7

    Wilson, D. M. & Barsky, D. Mutat. Res. 485, 283–307 (2001).

    CAS  Article  Google Scholar 

  8. 8

    Wiebauer, K. & Jiricny, J. Proc. Natl Acad. Sci. USA 87, 5842–5845 (1990).

    CAS  Article  Google Scholar 

  9. 9

    Feng, J. Y. et al. J. Biol. Chem. 276, 23832–23837 (2001).

    CAS  Article  Google Scholar 

Download references

Author information

Affiliations

Authors

Corresponding author

Correspondence to Josef Jiricny.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Jiricny, J. An APE that proofreads. Nature 415, 593–594 (2002). https://doi.org/10.1038/415593a

Download citation

Further reading

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Search

Quick links

Sign up for the Nature Briefing newsletter for a daily update on COVID-19 science.
Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing