REVIEWS THE EXPANDING POLYMERASE UNIVERSE Myron F. Goodman and Brigette Tippin Over the past year, the number of known prokaryotic and eukaryotic DNA polymerases has exploded. Many of these newly discovered enzymes copy aberrant bases in the DNA template over which ‘respectable’ polymerases fear to tread. The next step is to unravel their functions, which are thought to range from error-prone copying of DNA lesions, somatic hypermutation and avoidance of skin cancer, to restarting stalled replication forks and repairing double- stranded DNA breaks. DCMP TRANSFERASE In the beginning there was just one DNA polymerase — have been established for some members, including the ′ The DNA template-directed Escherichia coli DNA polymerase I (pol I), discovered by E. coli UmuD 2C complex (now known as pol V), the reaction catalysed by the yeast Arthur Kornberg and colleagues1,2 in 1956. Thirteen yeast Rev1 protein and human DNA polymerase eta (pol Rev1 protein, where C is years later, Paula de Lucia and John Cairns, at Stony η/Rad30). However, the functions of the remaining favoured for incorporation opposite an abasic template Brook, New York, isolated an E. coli mutant, polA (its members of the UmuC/DinB/Rev1p/Rad30 polymerase site, and to a much lesser extent designation being a play on de Lucia’s first name, as pro- superfamily are less certain. opposite a normal G site. posed to Cairns by Julian Gross) that seemed to have less A feature common to many of these polymerases is than 1% of the normal pol I activity3. From this strain, a their tendency to copy undamaged DNA with remark- PROOFREADING 4 Excision of a misincorporated new DNA-polymerizing enzyme, pol II, was isolated . ably poor fidelity, whether or not they are involved in nucleotide at a growing 3′- The polA strain was much more sensitive to ultravi- translesion synthesis. As its name suggests, translesion primer end by a 3′ exonuclease olet (UV) radiation than wild-type cells, suggesting that synthesis is the unimpaired copying of aberrant bases associated with the polymerase. pol I might be involved in DNA repair in addition to (see below) at which other cellular polymerases stall. chromosomal replication. Shortly after, using this same With undamaged DNA, these low-fidelity polymerases T–T 6–4 PHOTOPRODUCT 5 A form of damage occurring polA strain, Thomas Kornberg and Malcolm Gefter , incorporate an incorrect nucleotide once every 10–12 when DNA is exposed to UV and Friedrich Bonhoeffer, Heinz Schaller and 100–1,000 bases on average (TABLE 1). For compari- radiation, in which a covalent colleagues6, independently discovered DNA polymerase son, normal polymerases that do not PROOFREAD misin- bond is formed between the 6 III (pol III). Isolation of a conditionally lethal tempera- corporate nucleotides in the range of once every 104–106 and 4 pyrimidine ring 7 13 positions, coupling adjacent ture-sensitive pol III mutant showed that this enzyme bases . Examples of low-fidelity polymerases include 6 thymines on the same DNA is required for replicating the E. coli chromosome .In E. coli pol V,which preferentially misincorporates G strand. contrast, pol II remained an enigma until last year, opposite a 3′ T of a T–T 6–4 PHOTOPRODUCT; E. coli DNA when it was shown8 to be pivotal in restarting replica- polymerase IV (pol IV/DinB), which adds a nucleotide tion in UV-irradiated cells. onto the end of a misaligned primer; Rev1p, which Last year also saw the identification of a new class of incorporates C opposite a non-coding ABASIC LESION; and DNA polymerases — the UmuC/DinB/Rev1p/Rad30 human DNA polymerase iota (pol ι/Rad30B), which University of Southern superfamily (TABLE 1) — on the basis of five conserved favours misincorporation of G opposite T on undam- California, Department of sequence motifs present in all of these proteins (FIG. 1). aged DNA. All of these events lead to mutation. There is Biological Sciences and The yeast Rev1 protein had been shown to contain DNA- also the remarkable case of pol η, which copies pyrimi- Chemistry, Stauffer Hall of template-dependent DCMP TRANSFERASE activity nearly three dine T–T DIMERS accurately, resulting in mutation avoid- Science 172, Los Angeles, 9 California 90089-1340, USA. years earlier , but it was not until 1999 that the other fam- ance at this type of DNA damage (FIG. 2). e-mail: ily members were isolated and shown to be capable of The number of DNA polymerases has now grown [email protected] replicating DNA using all four bases. Biological functions from 3 to 5 in E.coli, and from 5 to at least 14 and count- NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 1 | NOVEMBER 2000 | 101 © 2000 Macmillan Magazines Ltd REVIEWS ABASIC LESION ing in eukaryotes (TABLE 1). Indeed, in a ‘back to the that there is a special class of polymerases to copy dam- A common form of DNA future’ moment during a recent conversation with Bob aged DNA templates. Instead, proteins such as UmuC damage in which a base is lost Lehman, Arthur Kornberg remarked, “In 1955, who and UmuD′ (the mutagenically active form of UmuD; from a strand of DNA, would have imagined that there could be five DNA poly- BOX 1) were thought to reduce the fidelity of E. coli pol III, spontaneously or by the action of DNA repair enzymes such as merases in E. coli?”.So what were the events that led to enabling a blocked REPLICATION FORK to carry out error- apurinic endonucleases or the discovery of these polymerases, and what do we now prone translesion synthesis21. We now know that this is uracil glycosylase, while leaving know of their biochemical functions and cellular proper- not the case — indeed, the cellular function of several the phosphodiester bond intact. ties? And why are there so many of them in eukaryotic new, errant DNA polymerases is translesion DNA syn- 22–25 (TABLE 1) T–T DIMER cells? Whereas prokaryotic cells have just one choice — thesis . Further studies using these ‘sloppier A form of damage occurring replicate damaged DNA or die — eukaryotic cells can, in copier’ DNA polymerases are now revealing a rich bio- when DNA is exposed to UV principle, use programmed cell death (apoptosis) as an chemical tapestry. For example, one member of this fam- radiation, in which two covalent ′ 26–28 ‘escape hatch’ to avoid a potential catastrophe. ily — the E. coli UmuD 2C polymerase (pol V) — bonds are formed between both does not act alone, but requires three further proteins to the 5 and 6 positions of the 12,28 pyrimidine ring on adjacent A growing family catalyse translesion synthesis . Moreover, yeast Rev1p thymines located on the same Genetic studies in Saccharomyces cerevisiae and E. coli requires DNA polymerase zeta (pol ζ) to copy past abasic DNA strand. have been instrumental in defining groups of proteins sites9 and T–T 6–4 photoproducts29. Other family mem- required for mutagenesis. For example, yeast lacking the bers probably also use accessory proteins. NUCLEOTIDE-EXCISION REPAIR (REF.14) (REFS 15,16) (REF.17) The main pathway for removal REV3 , REV7 or REV1 genes of UV-damaged bases. show significantly decreased spontaneous and UV- The E. coli pol V mutasome induced mutation rates. In E. coli, SOS mutagenesis A good place to start any tour of the new DNA poly- REPLICATION FORK (BOX 1) requires the umuC and umuD genes18. merases is with the E. coli pol V mutasome. The DNA- Site in double-stranded DNA at In 1968, Dean Rupp and Paul Howard-Flanders19 damage-inducible SOS response in E. coli was discov- which the template strands are separated, allowing a newly observed discontinuities (daughter-strand gaps) in ered more than 25 years ago (BOX 1). Many of the 30 or formed copy of the DNA to be DNA synthesized in an excision-defective strain of E. more SOS-regulated genes are involved in repairing synthesized, with the fork coli after UV irradiation. Because these strains cannot DNA damage30,31, but two genes, umuC and umuD,are moving in the direction of 20 18,32–34 carry out NUCLEOTIDE-EXCISION REPAIR , Rupp and Howard- instead required for SOS-induced mutagenesis . leading strand synthesis. Flanders suggested that a single pyrimidine dimer is Although SOS mutation rates are typically 100-fold 31 DISTRIBUTIVE POLYMERASE enough to kill the cell, presumably by blocking DNA higher than spontaneous rates , increased mutagenesis A polymerase that dissociates replication. But although it may be advantageous to cannot occur unless UmuD is first converted (by cleav- from the primer–template DNA copy a variety of template lesions as an alternative to cell age) to the mutagenically active UmuD′ protein in a after incorporating one (or at death, there is no such thing as a free lunch. The associ- reaction that depends on another SOS-induced protein, most a few) nucleotides. ated cost of survival is an increased number of muta- RecA (REF. 35). The UmuC and UmuD′ proteins then 36,37 ′ tions, targeted at the lesion sites. In E. coli, this is referred interact to form a tight complex ,UmuD2C (pol to as UV-induced SOS error-prone repair (BOX 1). V)27,28, with intrinsic DNA polymerase activity. Because both E. coli and yeast were known to have Working alone on an undamaged primed DNA tem- three DNA polymerases, there was no reason to suspect plate, pol V is a poor DISTRIBUTIVE POLYMERASE26,27.However, pol V cannot copy damaged DNA by itself — it requires I II III IV V RecA, single-stranded DNA binding protein (SSB) and β γ 12,27 β PROCESSIVITY CLAMP γ a E. coli pol IV/DinB H2N COOH / complex (where is the and the CLAMP-LOADER component of the replicative pol III holoenzyme).