Bridging the gap: a family of novel DNA polymerases that replicate faulty DNA.

D replication can be likened to a car travelling down a long highway. Normally the road is paved smooth, but, as often occurs during travel, the road sometimes contains bumps and roadblocks. Such is the case for the replication machinery as it travels down DNA. While DNA damaged by natural and manmade agents normally is repaired by a multitude of repair pathways, some lesions inevitably escape repair and, as a result, are encountered by the DNA replication machinery. Many of the lesions cannot be bypassed by the replicative DNA polymerases and will lead to cell death if not overcome. Just how cells handle these road blocks has been a longstanding problem. Upon encountering a lesion, the replicative polymerase may dissociate from DNA, leaving a gap in the newly synthesized strand. Such a gap could be filled in by a recombinational mechanism (1) or by a ‘‘copy choice’’ type of DNA synthesis (2). Because both of these mechanisms use information from the undamaged sister duplex to fill in the gap, they are relatively error-free. However, replication of damaged DNA also can occur by synthesis across from the lesion in the template strand. In this case, a specialized replication complex inserts a random nucleotide across from the lesion and continues synthesis beyond the lesion. This process is usually mutagenic and is best exemplified by the Escherichia coli SOS system in which the complex of UmuC and UmuD9 proteins interacts with DNA polymerase III (PolIII) holoenzyme to promote damage bypass. Until recently, the prevailing notion for UmuC-UmuD9 action has been that it overrides the normally high fidelity of PolIII, enabling PolIII to insert a nucleotide across from the lesion and to replicate past the lesion (3). UmuC belongs to a protein family that includes E. coli DinB, Saccharomyces cerevisiae Rev1 and Rad30, and human hRad30 proteins. In this issue of PNAS, Gerlach et al. (4) identify another member of this family in humans, which they refer to as DinB1. Recent studies of members of the UmuC protein family have heralded the emergence of a new paradigm of damage bypass in which, rather than modifying the activity of the replicative DNA polymerase, these proteins themselves are DNA polymerases specific for bypassing lesions in a mutagenic or error-free manner (Table 1). The first hint of this new paradigm came from studies in S. cerevisiae, where genes belonging to the RAD6 epistasis group function in error-free and error-prone damage bypass. The REV1, REV3, and REV7 genes of this group are essential for mutagenic bypass, and RAD5 and RAD30 function in error-free bypass of UV-damaged DNA. Rev1 possesses an unusual deoxycytidyl transferase activity that specifically inserts a dCMP residue opposite a template abasic site (5). Rev1 also can insert a cytosine opposite template guanine, adenine, and uracil, but it does so with a much reduced efficiency (10–20%). The insertion of cytosine opposite the abasic site produces a terminus that can be extended by DNA polymerase z, comprised of the Rev3 and Rev7 subunits (6). Thus, mutagenic bypass of abasic sites in yeast could occur by the coordinated action of a transferase and an error-prone DNA polymerase (Fig. 1). Although the deoxycytidyl transferase activity of Rev1 is template specific, Rev1 is clearly distinct from classical DNA polymerases in that it inserts only a cytosine residue. The first member of this family of proteins shown to be a bona fide, classical DNA polymerase was S. cerevisiae RAD30encoded Polh (7). Unlike Rev1, Polh synthesizes DNA by incorporating all four nucleotides in a template-specific manner. Deletion of RAD30 results in moderate sensitivity to UV light, and deletion of both RAD5 and RAD30 causes a synergistic increase in UV sensitivity and in UV mutagenesis, implicating these genes in alternate error-free bypass pathways (8–10). By contrast to error-prone synthesis by the Rev proteins, Polh bypasses a thymine-thymine (T-T) dimer, a major UV photoproduct, in an error-free manner by inserting two A residues opposite the two Ts of the dimer (7) (Fig. 1). The sun-sensitive cancer-prone syndrome xeroderma pigmentosum (XP) can arise from a defect in nucleotide excision repair (NER) or from a defect in the replication of UV-damaged DNA. Cells from the variant form of XP (XP-V) have no defect in NER, but they are unable to replicate UVdamaged DNA (11–13). Additionally, XP-V cells are hypermutable with UV light, and they are less likely than normal cells to incorporate adenines opposite thymine photoproducts (14, 15). These observations, and the ability of the yeast enzyme to faithfully replicate dimer-containing DNA, prompted the proposal that mutations in human Polh would cause XP-V (7). The gene for the human RAD30 counterpart, hRAD30A, recently was identified by two different groups and shown to harbor nonsense or frameshift mutations in cell lines derived from XP-V patients (16, 17). Because the majority of such mutations in hRAD30A would produce severe truncations of the protein and result in the loss of function, hRAD30A is not essential for viability or for growth. Like its yeast counterpart, hRAD30A-encoded Polh bypasses a T-T dimer (17). In the absence of error-free replication by Polh, error-prone translesion synthesis by Polz would cause hypermutability and result in increased incidence of cancers that occurs in XP-V patients. Another human RAD30 homolog, hRAD30B, recently has been identified (18). The gene is located on chromosome 18q21.1, a region that frequently is deleted in many cancers. It remains to be seen whether mutations in this gene contribute to any of these cancers. The hRAD30B transcripts are highly expressed in testes and to a lesser extent in heart and pancreas. The human homolog of E. coli DINB, hDINB1, is located on chromosome 5q13.1. hDINB1 is expressed in a variety of tissues, and its expression is also highest in testes (4). The high level expression in testes may reflect a specific role of these genes in some aspect of spermatogenesis or spermiogenesis. In E. coli, mutagenic bypass of abasic sites recently has been reconstituted by using purified UmuC, UmuD9, activated RecA, b-sliding clamp, g-clamp loading complex, single-stranded DNA-binding protein