news and views nature structural biology • volume 8 number 10 • october 2001 827 The number of DNA polymerases sky- rocketed recently with the discovery of a very large group of error prone DNA polymerases, the Y-superfamily 1 , capable of waltzing past DNA lesions that trip up normal replicative polymerases. Structures of two of these remarkable enzymes were recently reported. The crys- tal structures of the Saccharomyces cere- visiae RAD30A polymerase (Pol η) 2 and the Sulfolobus solfataricus Dbh poly- merase 3 provide the initial glimpses of Y-superfamily polymerases that restart replication at sites of mismatched or dam- aged bases 4–9 and that contribute to adap- tive mutagenesis in cells growing under adverse conditions 10,11 . Error prone poly- merases allow cells to cope with un- repaired DNA damage by enabling the completion of replication in the face of otherwise terminal roadblocks. These ‘sloppier copiers’ lack many of the virtues of other polymerases, including speed, fidelity and a firm grip (processivity) on the DNA template. However, the biologi- cal fitness of the error prone DNA poly- merases should be judged by a different standard. They are, after all, specialists in resolving messy situations that make respectable polymerases blush. It was unforeseen that cells would be equipped with such a vast number of polymerases to deal with replication mis- takes and chemical insults to DNA. The sheer number of lesion bypass polymeras- es, and in particular the existence of many orthologs in higher organisms, suggests that individual enzymes have highly spe- cialized cellular roles. Additional support for this proposal comes from the unique biochemical characteristics of various Y-superfamily polymerases, including their different efficiencies in bypassing particular lesions and their mutagenic propensities 12–18 . The broad phylogenetic distribution of the lesion bypass poly- merases underscores the strategic impor- tance of tolerating DNA damage by replication bypass as a means of survival when DNA repair cannot be completed in a timely manner. Although we do not fully understand the molecular logic behind the decision to either repair a DNA lesion or replicate past it, both options appear to be important. Human patients with a variant form of the inherited disorder xeroderma pigmentosum (XP-V) lack functional Pol η, resulting in the inability to bypass several types of UV-induced lesions in DNA 19 . Although cells from XP-V patients have apparently normal nucleotide excision repair activities that should repair these lesions, they are unable to replicate UV-damaged DNA. This loss of lesion bypass activity is associ- ated with a high incidence of sunlight- induced cancers in XP-V patients. Almost 40 years elapsed between the identification of the first DNA poly- merase 20 and the discovery of the error prone polymerases 12,19,21–23 , which lack the conserved sequence motifs characterizing the DNA polymerases that handle most genomic replication and repair 24,25 . Residues within these conserved motifs interact with bound substrates and have profound effects on the rate and fidelity of DNA synthesis by most polymerases 26–28 . Error prone DNA polymerases have a dif- ferent set of five sequence motifs (I–V) 6,7,13,29 , which are broadly conserved in more than 50 members of the Y-poly- merase superfamily 1 . Amino acid substi- tutions at several of these conserved positions eliminate or greatly diminish polymerase activity and the effects of many others remain to be tested. The crys- tal structures of Pol η 2 and Dbh 3 reveal the locations of the conserved motifs and sug- gest what their functions may be, provid- ing a Rosetta Stone for ciphering how DNA lesions are skirted and why these polymerases make so many mistakes. Like conventional DNA polymerases, Pol η and Dbh have a shape resembling a right hand with fingers, palm, and thumb subdomains 26,28 . The palms of Pol η and Dbh closely resemble the palm of A-fami- ly polymerases, such as Escherichia coli DNA polymerase I 30 and the bacterio- phage T7 DNA polymerase 31 . The palm comprises the floor of the polymerase active site and includes three highly con- The anatomy of infidelity Tom Ellenberger and Laura F. Silvian Recent crystal structures of error prone DNA polymerases that bypass damage in DNA templates provide counterexamples to high fidelity polymerases. Fig. 1 Comparison of error prone and high fidelity polymerases. There is no sequence similarity between the error prone DNA polymerases, represented here by Sulfolobus solfataricus Dbh 3 and Saccharomyces cerevisiae Pol η 2 , and highly accurate replicative polymerases like the bacterio- phage T7 DNA polymerase 31 . Nonetheless, all three polymerases have similar shapes, resembling a right hand composed of fingers (yellow), palm (blue) and thumb (red) subdomains. Dbh and Pol η have finger and thumb domains that are smaller than those of T7 DNA polymerase. This creates a shallow and unencumbered active site in the error prone polymerases. The surfaces of all three enzymes are shown from the vantage point of DNA exiting the polymerase. a, The S. solfataricus Dbh catalytic fragment (residues 1–205, PDB accession 1IM4). b, The S. cerevisiae pol η polymerase fragment (residues 1–509, PDB accession 1JIH) contains an additional C-terminal polymerase asso- ciated domain (green) that is proposed to contact DNA. c, The T7-DNA polymerase (PDB accession 1T7P) is composed of palm, fingers and thumb subdomains with additional proofreading exonu- clease (gray) and processivity domains (tan). The extended thumb of T7 DNA polymerase and its more expansive fingers create an enclosed substrate binding site that undoubtably contributes to the high fidelity of this polymerase. The DNA and nucleotide substrates have been omitted from the T7 structure in order to show the active site. a b c © 2001 Nature Publishing Group http://structbio.nature.com © 2001 Nature Publishing Group http://structbio.nature.com