News & Views | Published:

Molecular biology

The tug of DNA repair

Nature volume 505, pages 298299 (16 January 2014) | Download Citation

The transcription enzyme RNA polymerase stalls at DNA lesions, hindering their repair. Accessory factors dislodge the enzyme by pushing it forwards, but a study finds that pulling it backwards may also be effective. See Article p.372

DNA damage caused by genotoxic agents, from solar ultraviolet light to free radicals generated during cellular metabolism, is unavoidable. Unless repaired, such damage may directly threaten cell survival or lead to mutations and disease. All organisms therefore rely on repair systems to maintain DNA integrity, with many components of those systems highly conserved throughout evolution. In this issue, Epshtein et al.1 (page 372) describe how two enzymes — an RNA polymerase and UvrD, a helicase — cooperate to target a damaged site for repair.

RNA polymerase (RNAP) carries out the essential job of transcribing DNA into RNA. When RNAP runs into a lesion in the transcribed (template) strand, it backtracks along the DNA, but still occludes the damaged site (Fig. 1a, b), an effect that might be expected to hinder DNA repair. Unexpected observations2 that, in vivo, damage is preferentially repaired in transcribed rather than non-transcribed regions led to the discovery of transcription-coupled repair (TCR), a branch of the nucleotide-excision repair (NER) pathway that removes many lesions, including ultraviolet-induced thymine dimers3.

Figure 1: Alternative routes for transcription-coupled repair in bacteria.
Figure 1

a, b, In a transcription complex, the two DNA strands separate to form a bubble; the non-template strand (purple) is exposed on the surface of RNA polymerase (RNAP); the β flap forms part of the RNA exit channel. When RNAP encounters a DNA lesion (orange star), it backtracks, threading the nascent RNA into the secondary channel and blocking access to the lesion. c, Forward translocation of RNAP induced by sliding of the Mfd protein (yellow arrow) on the double-stranded DNA releases RNAP, allowing repair and generating naked, intact DNA. d, Epshtein et al.1 report that backward translocation (green arrow) induced by UvrD, which binds to the non-template strand and the β flap, exposes the lesion; the transcription factor NusA may stabilize the UvrD–β-flap interaction. Following repair, the arrested RNAP must be reactivated to position the 3′ RNA end in the active site, or removed (for example, by Mfd).

NER of naked DNA in bacteria involves a complex of Uvr proteins: UvrA and UvrB recognize the lesion; UvrC excises the damaged segment; and UvrD helicase displaces it. The resulting gap is then enzymatically repaired by two enzymes, DNA polymerase and a ligase. In TCR, RNAP takes over the damage-recognition task, but has to be removed to allow the Uvr complex to access the lesion.

In a broadly accepted model for TCR, a protein known as Mfd pushes the stalled RNAP forwards (Fig. 1c), releasing it from the DNA4, and recruits UvrA2, thereby directly coupling transcription to repair. Intriguingly, bacterial strains lacking Mfd are quite resistant to damage caused by ultraviolet light2, implying that alternative means of repair exist. Epshtein et al. describe a TCR pathway of opposite polarity to the conventional mechanism, wherein UvrD pulls RNAP even further backwards, exposing the lesion for repair (Fig. 1d).

The role of UvrD in NER is well established — it acts late in the pathway, after RNAP has dissociated from the DNA and the UvrABC proteins have processed the lesion. However, Epshtein and colleagues observed that, in the bacterium Escherichia coli, UvrD interacts with RNAP as frequently as the general transcription factors NusA and NusG. This finding prompted the authors to re-examine the role of UvrD in TCR.

Their analysis convincingly demonstrates that UvrD forms a binary complex with RNAP and induces it to backtrack both in vivo and in an in vitro system. The researchers also found that UvrD relieves an RNAP-imposed block to UvrABC excision of a thymine dimer in a minimal NER system reconstituted in vitro. Cells lacking UvrD are highly sensitive to several genotoxic agents that can induce NER, but Epshtein et al. observed that this effect is reversed when the two known anti-backtracking processes are inhibited so that RNAP can retreat even in the absence of UvrD. Moreover, the authors found that, in vitro, UvrD activities require an energy source (ATP nucleotides), consistent with the enzyme's function as a motor protein.

UvrD translocates on a single DNA strand to displace DNA-bound proteins, an activity thought to be mechanistically distinct from its helicase activity, which separates (melts) the two DNA strands5. The melted non-template DNA strand is a target for several transcriptional regulators. Could UvrD slide on the non-template DNA to push on RNAP?

Consistent with this model, Epshtein et al. showed that UvrD interacts with the non-template strand. They then sought to identify amino-acid residues of UvrD and RNAP that make contacts between the two molecules, using a powerful experimental approach6 that allows the construction of a high-confidence three-dimensional map of the complex, even though a structure of the complex has not been determined. This analysis identified the β-subunit flap domain of RNAP as a contact site for UvrD. The mapping results place UvrD at the upstream end of the transcription bubble, the structure that forms when part of the DNA helix is unwound by RNAP (Fig. 1). This is a logical location for pushing the RNAP backwards, and may explain the involvement of NusA — which also interacts with the β flap — in NER7. Indeed, Epshtein and colleagues show that NusA cooperates with UvrD in vitro and may favour backtracking in vivo.

UvrD was first implicated in DNA repair more than 40 years ago (see ref. 8, for example), but even the most basic questions about its role remain matters of debate. It is not known whether UvrD functions as a helicase or a translocase to carry out its diverse functions, or as a monomer or a dimer. The current study does not answer any of these questions, and instead poses a few more; for example, does UvrD bind only to an RNAP stalled at a lesion, and how does it compete with other regulators that bind to the non-template DNA? It also calls for textbooks to be revised: rather than (or in addition to) displacing the damaged DNA segment, UvrD displaces the lesion-stalled RNAP. This conceptual shift will undoubtedly reignite interest in UvrD, which should, in turn, help to address the outstanding questions.

The UvrD-dependent TCR pathway seems to be dominant in E. coli, but is it universally conserved? It relies on backtracking, a mechanism widely used by cellular RNAPs to regulate gene expression9. Yeast pol II, the RNAP that transcribes most yeast genes, is structurally similar to the bacterial enzyme, retreats at lesions and associates with Ssl2, a homologue of human XPB helicase, which has the same translocation polarity as UvrD and is essential for NER. Mutations in XPB are linked to xeroderma pigmentosum, a human disorder associated with a high risk of developing skin cancer.

But, unlike UvrD, XPB is thought to melt and load the DNA into RNAP. It also binds at the 'wrong' place, contacting the downstream DNA and a transcription factor (TFIIE) that occupies the upstream part of the would-be bubble in the transcription complex when it is poised to start synthesis10. TFIIE subsequently dissociates from RNAP but is replaced by a NusG-like factor, which forms a processivity clamp11 (a structure that enables uninterrupted RNA synthesis in all domains of life). NusG homologues bind to the non-template strand and would be expected to exclude XPB (or UvrD). Could the clamp be lost at a lesion? Dissecting the interplay between these and other regulators will require the powerful mapping approaches that helped Epshtein and colleagues to establish UvrD as a key factor for TCR.


  1. 1.

    et al. Nature 505, 372–377 (2014).

  2. 2.

    & Science 260, 53–58 (1993).

  3. 3.

    , & Prog. Mol. Biol. Transl. Sci. 110, 25–40 (2012).

  4. 4.

    & Curr. Opin. Microbiol. 7, 120–125 (2004).

  5. 5.

    Annu. Rev. Biophys. 39, 367–385 (2010).

  6. 6.

    J. Struct. Biol. 173, 530–540 (2011).

  7. 7.

    et al. Proc. Natl Acad. Sci. USA 107, 15517–15522 (2010).

  8. 8.

    , & Mol. Gen. Genet. 101, 227–244 (1968).

  9. 9.

    Cell 149, 1438–1445 (2012).

  10. 10.

    , & Nature Struct. Mol. Biol. 19, 788–796 (2012).

  11. 11.

    J. Mol. Biol. 417, 13–27 (2012).

Download references

Author information


  1. Irina Artsimovitch is in the Department of Microbiology and the Center for RNA Biology, Ohio State University, Columbus, Ohio 43210, USA.

    • Irina Artsimovitch


  1. Search for Irina Artsimovitch in:

Corresponding author

Correspondence to Irina Artsimovitch.

About this article

Publication history



Further reading


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.

Newsletter Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing