Interactions of DNA Helicases with Damaged DNA: Possible Biological Consequences*

DNA helicases are a class of enzymes able to unwind the two complementary DNA strands in a reaction dependent on energy derived from nucleoside 59-triphosphate hydrolysis. Their action is required in fundamental cellular processes such as DNA replication, repair, recombination, and transcription. From the known crystal structures it appears that the enzymatic machinery of DNA helicases has been highly conserved; nevertheless their functions can be distinguished by co-factor utilization, substrate preference, directionality of unwinding, processivity, and interaction with other proteins. These aspects of DNA helicase biochemistry have recently been reviewed (1–4). Unrepaired DNA lesions can affect the outcome of DNA replication, repair, recombination, and transcription. Thus, because DNA helicases are among the first proteins that would encounter DNA damage during these processes, a complete understanding of the metabolism of damaged DNA requires a biochemical analysis of the interaction between helicases and lesions. Clearly, chemical or physical cross-links between opposite DNA strands (interstrand lesions) will represent a mechanical block to DNA helicase action. However, in vivo the functions of DNA helicases are also expected to be severely impaired by bulky, helix-distorting intrastrand DNA lesions like those produced by ultraviolet irradiation or by chemical agents such as cisplatin, acetylaminofluorene, and benzopyrene. (For a general description of DNA lesions, see Ref. 5.). Although helicases unwind duplex DNA, most known helicases associate and translocate unidirectionally along one strand of the DNA. Accordingly, except for interstrand lesions, the position of a lesion may differentially influence the action of a DNA helicase, depending on whether it is located on the DNA strand along which the enzyme translocates or on the opposite one. In this minireview we will focus mainly on in vitro studies aimed at investigating the effects of helix-distorting DNA lesions on the action of DNA helicases implicated in DNA repair, recombination, and replication; the possible biological consequences of these interactions will also be discussed. To facilitate organization of this review, helicases have been categorized depending on whether they function in repair, recombination, or replication. However, the reader should be aware that there is substantial overlap between these processes.

DNA helicases are a class of enzymes able to unwind the two complementary DNA strands in a reaction dependent on energy derived from nucleoside 5Ј-triphosphate hydrolysis. Their action is required in fundamental cellular processes such as DNA replication, repair, recombination, and transcription. From the known crystal structures it appears that the enzymatic machinery of DNA helicases has been highly conserved; nevertheless their functions can be distinguished by co-factor utilization, substrate preference, directionality of unwinding, processivity, and interaction with other proteins. These aspects of DNA helicase biochemistry have recently been reviewed (1)(2)(3)(4). Unrepaired DNA lesions can affect the outcome of DNA replication, repair, recombination, and transcription. Thus, because DNA helicases are among the first proteins that would encounter DNA damage during these processes, a complete understanding of the metabolism of damaged DNA requires a biochemical analysis of the interaction between helicases and lesions. Clearly, chemical or physical cross-links between opposite DNA strands (interstrand lesions) will represent a mechanical block to DNA helicase action. However, in vivo the functions of DNA helicases are also expected to be severely impaired by bulky, helix-distorting intrastrand DNA lesions like those produced by ultraviolet irradiation or by chemical agents such as cisplatin, acetylaminofluorene, and benzopyrene. (For a general description of DNA lesions, see Ref. 5.). Although helicases unwind duplex DNA, most known helicases associate and translocate unidirectionally along one strand of the DNA. Accordingly, except for interstrand lesions, the position of a lesion may differentially influence the action of a DNA helicase, depending on whether it is located on the DNA strand along which the enzyme translocates or on the opposite one.
In this minireview we will focus mainly on in vitro studies aimed at investigating the effects of helix-distorting DNA lesions on the action of DNA helicases implicated in DNA repair, recombination, and replication; the possible biological consequences of these interactions will also be discussed. To facilitate organization of this review, helicases have been categorized depending on whether they function in repair, recombination, or replication. However, the reader should be aware that there is substantial overlap between these processes.

Effects of DNA Lesions on DNA Helicases Involved in DNA Repair
To overcome the deleterious effects of DNA lesions, a variety of DNA repair systems is present in all living organisms (5). Nucleo-tide excision repair (NER) 1 is one of the most versatile systems and shows remarkable similarities in both prokaryotes and eukaryotes (for review see Refs. 6 and 7). Among the six proteins necessary for NER in Escherichia coli, the UvrA 2 B protein complex and the UvrD protein have been shown to possess DNA helicase activities. The 5Ј to 3Ј helicase activity of UvrA 2 B is thought to allow limited ATPdependent scanning of DNA to detect damaged bases. For example, UV-induced lesions have been shown to inhibit translocation of UvrA 2 B, suggesting that arrest of the helicase may be a signal for repair (8). Subsequently, a preincision complex is formed, and incision occurs on both sides of the lesion. UvrD then loads onto the free 3Ј end of the damaged oligonucleotide and displaces it using its intrinsic 3Ј to 5Ј helicase activity (6). The recent determination of the crystal structure of UvrB suggests that formation of the preincision complex requires unwinding of the DNA by UvrA 2 B (9). It was postulated that stalling of the UvrA 2 B complex upon encountering a lesion triggers dissociation of the complex. In another study (10), it was found that the presence of the 2-(acetylamino)fluorene lesion in either DNA strand stimulated the activity of UvrA 2 B, leading the authors to suggest that the strand-separating activity of UvrA 2 B may not play a major role in lesion recognition.
The function of UvrD helicase (helicase II) in E. coli NER is in the excision of damage-containing oligonucleotides. The helicase activity of UvrD was found to be only moderately sensitive to ultraviolet radiation damage and intrastrand cisplatin adducts, whereas it was exquisitively sensitive to intercalators that position in the major groove of DNA (11,12). However, in one study (11) but not in the other (12), UvrD helicase action was found to be affected by distamycin A, an intercalator binding to the minor grove of DNA. Generally speaking, it is possible that UV-induced lesions or other helix-distorting intrastrand adducts are better tolerated by DNA helicases than those produced by intercalating agents, which can either totally disrupt the continuity of what a helicase normally recognizes or increase the thermodynamic stability of the DNA duplex in such a way that it resists helicase action (13). In the same study (11), the activity of UvrD helicase on damaged DNA was compared with the product of the yeast Saccharomyces cerevisiae rad3 gene, a helicase that is absolutely required for yeast NER. The 5Ј to 3Ј helicase activity of the Rad3 protein was found to be profoundly inhibited by UV damage and intrastrand cisplatin adducts located on the strand along which the enzyme translocates, whereas lesions on the opposite strand had no effect. Blockage of Rad3 by a lesion results in the formation of an abnormally stable protein-DNA complex (14). These results indicate that the DNA helicase activity of Rad3 protein may have been adopted by the yeast NER machinery to locate and determine the strand specificity of DNA damage, thus establishing some functional relationship with E. coli UvrA 2 B. It has long been known that the recognitionincision step of NER in human cells requires ATP (15), and the recent reconstitution of NER with recombinant human proteins now provides an interesting tool to study the role of the different factors involved in this process (16). Eukaryotic TFIIH is a multisubunit protein complex involved in RNA polymerase II transcription and nucleotide excision repair, which contains two demonstrated DNA helicases, named Rad3 and Rad25 in yeast and XPB and XPD in mammals (17). Interestingly, TFIIH containing a mutant XPD with impaired DNA helicase activity was shown to be defective in incisions around the lesion (18). Consistent with the role of the yeast homologue of XPD, Rad3, these results support previous models that suggest a role for the DNA helicases of TFIIH in the identification of the damaged strand and the correct localization of the incisions (6,19). The strand-specific stalling of either the 5Ј to 3Ј XPD helicase or the 3Ј to 5Ј XPB helicase could result in the distortion of the DNA-TFIIH complex, leading to the identification of the damaged strand and to the formation of a fully opened preincision complex. A schematic representation of the putative roles of complexes containing prokaryotic UvrA 2 B and eukaryotic TFIIH in DNA damage recognition is presented in Fig. 1.
Calf thymus DNA helicase E is a moderately processive 3Ј to 5Ј helicase active on nicked DNA that has been proposed to play a role in DNA repair (20). In vitro, helicase E can displace a primer containing an internal GpG intrastrand cross-link, the major DNA lesion generated by the antitumor drug cisplatin (Pt-d(GpG)). If the lesion was placed on the strand to which it binds, helicase E was inhibited only if the adduct was in the single strand part of the template but not if it was within the annealed region, suggesting that, in the latter case, it did not represent an absolute stop to helicase progression (21). Subsequent work has shown that a protein which recognizes and binds cisplatin-damaged DNA, the high mobility group 1 protein (HMG-1), reduced DNA unwinding of a platinated substrate by helicase E (22). HMG-1 protein has also been reported to inhibit NER of cisplatin lesions catalyzed by either cell extract or purified proteins by shielding the lesions from repair proteins (23).

Effects of DNA Lesions on DNA Helicases Involved in DNA Recombination
DNA recombination assures exchange of information between DNA chains and contributes to the repair of DNA damage and to restart DNA replication (for review see Refs. 24 -26). Double strand DNA break repair by homologous recombination in E. coli is initiated by the RecBCD enzyme, which unwinds and simultaneously degrades DNA from a double-stranded DNA end (27). Early studies on RecBCD enzyme showed that the enzyme could translocate through DNA duplexes that contain UV adducts, although, as expected, progression of the enzyme was arrested by interstrand lesions (28). Of the three subunits, RecB is the sole one that possesses 3Ј to 5Ј helicase activity (29). The effect of a Pt-d(GpG) adduct, in the context of partially duplex synthetic oligonucleotides, on the helicase activity of RecB was examined in vitro. Both strand-separating and DNA-dependent ATPase activities of the RecB protein were inhibited by the adduct when located on the template strand (30). Although further studies are required, these results suggest that subunit interactions within RecBCD enzyme allow the enzyme to traverse regions of DNA damage whereas RecB alone is inhibited. Another protein that plays a key role in recombination in E. coli is RecA protein (24). RecA catalyzes DNA strand exchange in vitro and has been shown to possess ATP-dependent DNA unwinding activity that is limited to the unwinding of short duplexes (31). The capacity of RecA to unwind a damaged template identical to the one used for the RecB helicase was examined in the same study. Contrary to RecB, it was found that neither the strand-separating nor the ATPase activities of RecA were inhibited by the Pt-d(GpG) adduct (30). Previous work had shown that RecA protein was able to bypass UV pyrimidine dimers during strand exchange in vitro (32). Ku autoantigen is a DNA end binding protein that has been shown to act as a regulator of a DNA-dependent protein kinase (DNA-PK), which plays essential roles in DNA repair and recombination (for review see Ref. 33). Ku protein has been identified as human DNA helicase II (HDH II), an enzyme that preferentially unwinds partially duplex DNA proceeding in the 3Ј to 5Ј direction (34). It was found that intercalating agents such as actinomycin, daunorubicin, or nogalamycin severely inhibited the unwinding activity of the HDH II/Ku enzyme (35).

Effects of DNA Lesions on DNA Helicases Involved in DNA Replication
The familiar structure of a replication fork, a site where the two strands of a duplex DNA are separated to reveal the single strands of opposite polarity, is generated through the action of replicative DNA helicases (36). In eukaryotes, although many helicases have been characterized biochemically, their precise roles in vivo remain difficult to establish. Thus, despite the existence of several candidates, no eukaryotic DNA helicase has yet been unambiguously shown to be essential for the progression of the replication fork during chromosomal replication (37)(38)(39). Such is not the case for bacteria and related phages or for eukaryotic viruses, where replicative DNA helicases have been identified and characterized (40). T7 gene 4 protein acts both as a primase and a helicase. The gene 4 protein has been shown to translocate along single-stranded DNA in the 5Ј to 3Ј direction, and this movement was found to be arrested in vitro by bulky DNA adducts formed by the chemical carcinogen benzo[a]pyrene. The inhibitory effects of these adducts are strand-specific in that they blocked the DNA helicase activity of gene 4 protein only if they are located on the strand along which the helicase translocates. In addition, the data presented indicate that gene 4 protein was sequestered at the site of the adduct (41,42). In Herpes simplex virus type 1 (HSV-1) DNA replication, two helicases are required for viral origin-specific DNA replication (43). The first of the two helicases is the product of the UL9 gene, a 3Ј-5Ј helicase that together with the viral single-stranded DNA binding protein, ICP8, can specifically unwind the HSV-1 origins of replication (44). It was found that a Pt-d(GpG) adduct significantly reduced, but did not abolish, the helicase activity of the UL9 protein but only when it was present on the strand along which the protein translocated (45). However, addition of ICP8 greatly stimulated the capacity of the helicase to unwind platinated DNA. Furthermore, the stimulation appeared to be the result of the functional and physical interaction that is known to exist between UL9 and ICP8 and not because of a preferential binding of ICP8 at the site of the adduct. Results from a subsequent study showed that ICP8 stimulated the DNA helicase activity of UL9 protein by increasing its processivity, thus facilitating its translocation along DNA and through regions of secondary structure (46). Based on the finding of this study it is tempting to speculate that ICP8 enables the UL9 protein to bypass the cisplatin Pt-d(GpG) lesion by tethering it to the DNA substrate, thereby preventing its dissociation. The second HSV-1 replicative DNA helicase is the product of the UL5, UL8, and UL52 genes. This heterotrimeric 5Ј to 3Ј helicase is also endowed with DNA-primase activity and is responsible for concomitant DNA unwinding and primer synthesis at the viral replication fork (44).
Addition of ICP8 specifically stimulated unwinding of platinated DNA by the helicase-primase, but at variance to what was found for UL9, stimulation by ICP8 appeared to enable bypass of the cisplatin intrastrand cross-link by recruiting the enzyme to the DNA rather than by increasing its processivity (47). Thus, these studies (45,47) suggest that specific protein-protein interactions between a single-stranded DNA binding protein and two replicative DNA helicases allow substantial unwinding of substrates containing a bulky intrastrand DNA lesion.
In the case of simian virus 40 (SV40), the only viral protein required for viral replication is large T antigen. This protein has an intrinsic 3Ј to 5Ј DNA helicase activity and binds specifically to the origin, allowing it to initiate replication and functions as the replicative helicase at the fork (48). The impact of intrastrand DNA lesions on the unwinding activity of SV40 T antigen protein was recently examined in vitro (49). Using synthetic forklike substrates containing either single UV photoproducts or a specific 2-(acetylamino)fluorene adduct, it was shown that T antigen heli- In B only the interaction of the 5Ј to 3Ј helicase with a lesion is shown. This representation is derived from models previously presented in Refs. 6 and 17, respectively. See text for further details. Gray triangle, intrastrand DNA lesion; red oval, UvrA; blue oval, UvrB; small yellow and orange spheres on large green sphere, TFIII (both 3Ј to 5Ј (orange sphere) and 5Ј to 3Ј (yellow sphere) helicases are represented). case activity was not affected by the presence of the lesions on either the strand on which the enzyme translocates or on the opposite one. The capacity of T antigen to displace UV-irradiated DNA was also observed in a previous work (50). However, T antigen helicase was found to be blocked by DNA-intercalating drugs in vitro (51).

Tentative Conclusions from in Vitro Studies
In general, intrastrand DNA lesions seem to affect DNA helicase activity in vitro only when placed on the strand to which the protein translocates. Many of the DNA helicases mentioned here act as multimers (52). Replicative helicases in particular appear to exist as hexameric rings. A model for hexameric helicase action has been proposed where one strand of the DNA passes through the central channel of the helicase while the second strand is displaced outside the ring (53). This model is consistent with the observed strand specificity if one assumes that any bulky DNA lesion will have an inhibitory role when placed on the strand encircled by the helicase and no effect on the strand outside the ring, as recently suggested (54). However, in apparent contradiction of this view, the unwinding activity of proteins capable of forming oligomeric structures, such as SV40 T antigen or RecA, does not appear to be impaired in vitro by intrastrand DNA lesions on either strand (30,49). Perhaps the capacity of these proteins to oligomerize into large complexes can contribute to their ability to unwind past a lesion. HSV-1 UL9 helicase, whose action has been shown to be impeded but not blocked by a cisplatin lesion (45), may also belong to this category. Indeed, UL9 helicase action appears to be stoichiometric, requiring a DNA-dependent assembly of multimeric UL9 protein complex (43). The molecular mechanism(s) leading to DNA helicase inhibition by bulky DNA lesions have been investigated in some instances. It was found that S. cerevisiae Rad3 and T7 gene 4 helicases were sequestered on single-stranded DNA at the sites of damage, forming stable protein-DNA complexes (14,41), whereas calf thymus DNA helicase E was not (21). Whether or not these distinct in vitro characteristics are the reflection of different roles played by enzymes in vivo remains to be seen. HSV-1 ICP8 has been shown to interact with the HSV-1 UL9 helicase and with the UL8 subunit of the HSV-1 UL5/52/8 helicase-primase (43). ICP8 was found to specifically stimulate unwinding of platinated DNA substrates by both helicases, although the mechanisms of stimulation appear to be different for the two enzymes (45,47). The effect of ICP8 appeared to be the result of its interaction with the helicases and not a consequence of its preferential binding to the cisplatin adduct Pt-d(GpG). On the contrary, addition of the HMG-1 protein, which strongly binds the Pt-d(GpG) adducts, was found to inhibit the activity of calf thymus helicase E (22). These data point out how the interaction of helicases with damaged DNA can be modulated by other proteins.

Possible Biological Consequences of the Interaction of DNA
Helicases with Damaged DNA Many DNA helicases are active in the cell, and their wide range of functions probably explains why organisms have so many helicase genes (55). The molecular mechanism governing the interaction of a given DNA helicase with damaged DNA may be dictated, at least in part, by its cellular functions. For instance, DNA helicases implicated in the initial steps of NER have been proposed to check DNA for the presence of repairable DNA lesions and might be expected to be exquisitely sensitive to DNA damage whereas DNA helicases involved in the subsequent removal of the damaged DNA could be less affected. In vitro studies with UvrA 2 B, Rad3, or XPD helicases on one hand and UvrD helicase on the other hand roughly support this view. The sensitivity of various DNA helicases to DNA-damaging agents may also be influenced by their mechanism of translocation along DNA. For instance, it was recently shown that the RecBC DNA helicase is able to "step across" singlestranded DNA gaps, enabling it to unwind nicked or gapped duplexes (56). However, bulky intrastrand DNA lesions can still inhibit RecB activity (30).
Specific protein-protein interactions have been demonstrated for DNA helicases involved in DNA replication. Examples include: interactions of SV40 T antigen helicase with the eukaryotic singlestranded binding protein RP-A and DNA polymerase ␣ (57); HSV-1 helicase-primase with HSV-1 single-stranded binding protein ICP8 and HSV-1 DNA polymerase UL30/42 complex (43); T7 gene 4 helicase-primase with T7 single-stranded binding protein and T7 DNA polymerase (58); E. coli DnaB helicase with E. coli DNA polymerase III (59). In the latter case contact between DnaB and DNA polymerase increases the unwinding rate more than 10-fold (59) and imparts increased processivity to the leading strand polymerase of the DNA polymerase III holoenzyme dimer (60). When a DNA lesion is encountered during replication, the helicase will be at the forefront of the replication complex, and its initial interaction with the damage may influence the behavior of the DNA polymerase it contacts. Therefore, a stalled helicase could lead to the stalling of the whole replication complex whereas its capacity to unwind DNA past a lesion could promote replication bypass.
Current models dealing with replication of damaged DNA postulate the initial dissociation of replicative DNA polymerases when encountering sites of DNA damage. Replication can then restart beyond lesions, or DNA polymerases specialized in translesion synthesis (TLS) can temporarily replace replicative polymerases to synthesize across the lesion; replicative enzymes will then take over again (for review see Refs. 61-63, and references therein).
On the other hand, if interactions between components of the replisome are maintained, the bypass capacity of a helicase may drive some TLS by the replicative polymerases, possibly by increasing their dwell time at the lesion and/or by modifying the conformation of their active sites. Fig. 2 gives a simplified view of some of the possible molecular events following the interaction between a complex formed by a 5Ј to 3Ј DNA helicase, a replicative DNA polymerase, and a single-stranded binding protein with the site of DNA damage located at the replication fork. However, it should be stressed that the helicase-mediated bypass hypothesis remains purely speculative at the present time.
A further example of the potential importance of protein-protein interactions affecting a DNA helicase activity is given in the case of Werner's syndrome. The Werner's syndrome (WRN) gene encodes a polypeptide containing domains homologous to the RecQ family of DNA helicases, and mutations in these genes result in genomic instability and premature aging disorders (for review see Refs. 3 and 64). WRN protein (WRNp) has been purified and shown to possess a 3Ј to 5Ј helicase activity (65). In vitro, WRN helicase activity requires a 3Ј single-stranded tail, and its capacity to unwind small stretches of DNA is greatly stimulated by the singlestranded DNA binding protein RP-A (66,67). In addition to the capacity to unwind DNA, WRNp also possesses a 3Ј-to 5Ј-exonuclease activity (68,69). Several lines of evidence suggest that WRNp could play a role in DNA replication. (i) Werner cells exhibit a reduced rate of DNA replication and a prolonged S phase; (ii) a Xenopus laevis WRN homologue, FFA-1, has been shown to be present in replication foci (39); and (iii) WRNp copurifies with DNA replication complex and interacts with proliferating cellular antigen, RP-A, and topoisomerase I (67,70). Furthermore, a role for WRNp in DNA repair is also likely because (i) Werner's syndrome cells are sensitive to 4-nitroquinoline-1-oxide, an agent that produces bulky guanine and adenine adducts and oxidative DNA damage and (ii) WRNp interacts with p53 (71,72) and Ku autoantigen (73). However, in vitro WRNp shows no increased affinity for various types of intrastrand DNA damage, including adducts formed during 4-nitroquinoline-1-oxide treatment (74) but is potently inhibited by the structurally related minor groove binders distamycin A and neotropsin (75). Addition of RP-A did not alleviate this inhibition. Finally, the capacity of WRNp to partially suppress the hyper-recombination phenotype of mutants of its yeast homologue Sgs1 points to a role of WRNp in regulating homologous recombination (76). Thus, the roles of WRN protein in the cell appear to be multifaceted.
A recent finding may uncover a molecular link between the putative roles of WRNp in replication, repair, and recombination processes (77). In this study it is shown that WRNp, in the absence of proliferating cellular antigen, specifically interacts with the 32-kDa subunit of S. cerevisiae DNA polymerase ␦, a key DNA polymerase in replication and repair, and increases the rate of nucleotide incorporation. The authors suggest that one function of WRNp would be to play a role in replication reinitiation at forks blocked by DNA damage or unusual secondary structures that arise during replication/recombination processes and from which the normal replication machinery has dissociated, a role previously envisaged for E. coli RecQ (78).