Regulation of Rad6/Rad18 Activity During DNA Damage Tolerance

INTRODUCTION

Replicative polymerases (pols) have very stringent polymerase and proofreading domains and, thus, cannot accommodate damaged template bases. Consequently, progression of replication forks is stalled when such offenses are encountered during S phase. Failure to restart replication often results in double-strand breaks (DSBs) that may lead to gross chromosomal rearrangements, cell-cycle arrest, and cell death. Therefore, it is often more advantageous to circumvent such replicative arrests and to postpone repair of the offending damage in order to complete the cell cycle and maintain cell survival (1, 21, 34, 64–66, 100, 103, 128). Such a process, referred to as DNA damage tolerance (DDT), may be carried out either by translesion DNA synthesis (TLS) or by template switching. In TLS, the replicative pol is replaced with a specialized TLS pol that is able to catalyze stable, yet often erroneous, nucleotide incorporation opposite damaged templates, allowing replication to proceed (64, 100, 103, 128). In template switching, enzyme-catalyzed regression of the replication fork creates a chicken-foot structure in which the original template strands are reannealed, and the stalled primer terminus utilizes the newly synthesized sister strand as a damage-free template. Subsequent reversal of the chicken-foot structure reconstitutes the replication fork at a point beyond the damage, bypassing the block in an error-free manner. Both pathways have been observed in organisms ranging from bacteria, such as Escherichia coli, to humans, and many components of these pathways are well conserved (64, 100, 103, 122, 128). A major difference between DDT in eukaryotes and prokaryotes, however, is the use of the ubiquitin conjugation system as a regulatory mechanism (34, 122).

Ubiquitin is a small protein that is present exclusively in eukaryotes and that may be covalently attached to a target protein through an enzyme cascade. The E1 ubiquitin-activating enzyme activates ubiquitin and transfers it to an E2 ubiquitin-conjugating enzyme (Ubc), which catalyzes formation of an isopeptide bond between ubiquitin and a lysine residue on a target protein. An E3 ubiquitin ligase can bind an E2 enzyme and a target protein simultaneously, mediating the specificity of this process (29). In response to sublethal doses of DNA-damaging agents that cause replication-blocking lesions, ubiquitin is conjugated to lysine residue 164 (K164) of a PCNA monomer. No such modification occurs in untreated cells during S phase. Furthermore, the number of ubiquitins conjugated to PCNA by distinct E2/E3 pairs orchestrates DDT: Conjugation of a single ubiquitin by the Rad6/Rad18 pair mediates the TLS pathway, whereas extension of this ubiquitin to a polyubiquitin chain by Ubc13-Mms2/Rad5 diverts DDT to the template switching pathway (47). Over the past decade or so, researchers have invested much effort into delineating how the conjugation of ubiquitin to PCNA is regulated and how this posttranslational modification (PTM) orchestrates DDT. In this review, we focus on the monoubiquitination of PCNA by Rad6/Rad18 and the regulation of this process. We begin with a brief overview of the eukaryotic replisome.

THE EUKARYOTIC REPLISOME

Replicative DNA pols alone are distributive and achieve their characteristic high processivity by anchoring to ring-shaped sliding clamps that encircle double-stranded DNA (dsDNA) and slide freely along it. The eukaryotic sliding clamp, proliferating cell nuclear antigen (PCNA) (Figure 1), has two structurally distinct faces and is a trimer of identical subunits. Each subunit consists of two independent domains joined by an interdomain connecting loop (IDCL). The C-terminal face of PCNA contains all C-termini and IDCLs and is a platform for interaction with the eukaryotic replicative pols, pol ε and pol δ (45). Specifically, the well-conserved PCNA-interacting peptide (PIP) box within replicative pols makes extensive contact with the IDCLs of PCNA (15). Orchestration of DNA replication also involves many accessory proteins, each of which has a unique role in enhancing the efficiency of this highly complex process. Together, PCNA, the replicative DNA pols, and these accessory proteins (discussed below) compose the replisome, which is responsible for the rapid and accurate replication of genomic DNA (Figure 2).

At each origin of replication, two replicative DNA helicases are loaded onto the DNA, one on each strand, and they utilize ATP to unwind dsDNA. In eukaryotes, the core of the replicative DNA helicase comprises the Mcm2–7 proteins, and it is loaded onto DNA in an inactive form. Subsequent activation occurs by the recruitment of the Cdc45 and GINS accessory proteins, forming the holo-helicase referred to as CMG (Cdc45, Mcm2–7, GINS). In contrast to most helicases, CMG translocates with a 3′–5′ polarity and, hence, tracks along the leading strand of the DNA (12, 80, 115). Upon unwinding of dsDNA by CMG, the single-stranded DNA (ssDNA) templates are bound by replication protein A (RPA), protecting them from cellular nucleases and preventing formation of alternative DNA structures (7, 131). The bifunctional DNA pol α/primase complex then synthesizes short, complementary RNA/DNA hybrid primers on the ssDNA templates for each Okazaki fragment and replication origin (33, 63, 97, 111, 142, 143). These hybrid primers are recognized by replication factor C (RFC), the clamp loader complex in eukaryotes. RFC utilizes ATP binding and hydrolysis to open PCNA rings and place them around the 3′ end of primer/template (P/T) junctions such that the C-terminal face of PCNA is oriented toward the terminal 3′ hydroxyl end of the hybrid primer, where DNA synthesis will initiate. Hence, the C-terminal face of PCNA is often referred to as its “front side” (45). For comparison, the highly conserved K164 residue is located on the opposite face, i.e., the “back side” of PCNA (Figure 1) (32). Replicative pols then bind to the front side of PCNA encircling DNA, completing formation of the replisome. Specifically, the CMG helicase selects and stabilizes DNA pol ε on the leading strand while DNA pol δ utilizes the hybrid primers synthesized every 100–250 nucleotides by the DNA pol α/primase complex on the lagging strand (38, 111) (Figure 2).

RAD6/RAD18 CONJUGATES UBIQUITIN TO PCNA ENCIRCLING A BLOCKED Primer/Template JUNCTION

During normal DNA replication, the activities of the helicase and the replicative pols are tightly coupled, such that the amount of RPA-coated ssDNA is minimal. When a damaged template base is encountered, leading and lagging strand synthesis and/or helicase and polymerase activities become out of sync, leading to the buildup of RPA-coated ssDNA downstream of the offending damage (17, 19, 89). This structure recruits Rad6/Rad18 to the blocked P/T junction, where the Rad6/Rad18 complex catalyzes the conjugation of a single ubiquitin moiety (a process referred to herein as monoubiquitination) to K164 of the resident PCNA. This response can also be elicited by agents that block progression of replication forks without modifying (damaging) DNA at all, suggesting that RPA-coated ssDNA, rather than the actual DNA lesion, is the signal for Rad6/Rad18-mediated monoubiquitination of PCNA (11, 19, 23, 37, 44, 47, 72, 85, 86, 88, 90, 96, 105, 129). In vitro, Rad6/Rad18 can conjugate ubiquitin to all three monomers of a PCNA ring loaded onto DNA (44, 90). A single report (56) suggests that such conjugation may also occur in vivo, but further experiments are needed to confirm this and to ascertain whether such conjugation is required for DDT within the cell. Although there are roughly 10 E2 enzymes and more than 60 E3 ligases in eukaryotes, Rad6/Rad18 is the predominant E2/E3 pair responsible for the monoubiquitination of PCNA during DDT (29, 51, 118, 140). As the following section describes, the innate properties of Rad6 and Rad18 are uniquely attuned such that the Rad6/Rad18 complex is specifically targeted to PCNA present at stalled P/T junctions.

RAD6 AND RAD18 ARE A TAILOR-MADE MATCH FOR DNA Damage Tolerance

Rad18 is a multidomain E3 ubiquitin ligase that functions as a homodimer. Dimerization occurs via an N-terminal RING domain. The Rad6-binding (R6B) domain is located at the C-terminus of Rad18 and interacts with the noncovalent ubiquitin interaction site on Rad6. This site is imperative for ubiquitin chain formation by Rad6, and its association with the R6B domain of Rad18 prevents this activity. The Rad18 RING domain also aids in Rad6 binding, independently of the R6B domain, and together these two domains form a stable interaction with Rad6. Dimerization of Rad18 is asymmetric, however, such that only one of the Rad6 binding sites is open. Thus, the Rad18 homodimer binds only a single Rad6. Whether the R6B and RING domains that bind to Rad6 are from the same or different Rad18 monomers remains unknown (4–6, 46, 50, 77, 90, 117).

Rad18 also contains distinct domains that bind DNA (the SAP domain), PCNA, and RPA (23, 54, 77, 90, 117). The PCNA-binding motif has been mapped to the N-terminal region of Rad18 and acts independently of the other domains in this region. No sequence within this region shows any similarity to the classic PIP box (data not shown), however, and the amino acid residues involved in PCNA binding have yet to be identified (90). Furthermore, the location(s) on PCNA to which Rad18 binds remains unknown. Some reports suggest that Rad18 may not require access to the front side of PCNA for monoubiquitination, as overexpression of proteins that also bind to this side with high affinity, such as p15^PAF and p21, does not impair PCNA monoubiquitination (99, 113). Further experiments are needed to identify the Rad18-binding site(s) on PCNA. The SAP domain of Rad18 has a high preference for ssDNA, particularly that present at forked DNA substrates, and is required for assembly of Rad18 at stalled replication forks following DNA damage. Thus, the current model suggests that the SAP and RPA-binding domains together recruit Rad18 to RPA-coated ssDNA that builds up downstream of replication-blocking lesions. Once localized, the PCNA-binding domain of Rad18 directs it to the sliding clamp present at the blocked P/T junction (5, 23, 85, 90, 121). Note that Rad18 also contains ATPase and zinc finger (ZnF) domains. The function of the ATPase domain is unknown, but it has no effect on the DNA binding behavior of Rad18 or the Rad6/Rad18 complex and is dispensable for the interaction of Rad18 with Rad6 (5, 6, 54). The function of the ZnF domain is unclear; conflicting reports suggest that this domain has a role in DNA binding, Rad18 dimerization, ubiquitin binding, and controlling the cellular location of Rad18 (4, 54, 81, 117, 123).

Rad6 is an E2 Ubc that is highly conserved in eukaryotes and that can form independent complexes with alternative E3 ligases (6). Rad6 alone has an intrinsic ability to form polyubiquitin chains, and this activity is inhibited by its association with Rad18, permitting the conjugation of only single ubiquitin moieties to PCNA. Rad6 is also incapable of binding any nucleic acid, so by associating with Rad18, Rad6 is specifically targeted to RPA-coated ssDNA resulting from DNA damage (4). Once localized, the PCNA-binding domain of Rad18 is surmised to direct the ubiquitin-conjugating activity of Rad6 to PCNA encircling a blocked P/T junction. As discussed in the next section, the expression level, assembly, and cellular location of the Rad6/Rad18 complex are stringently controlled to ensure that the complex is poised for action when needed.

THE SPATIOTEMPORAL REGULATION OF THE RAD6/RAD18 COMPLEX PRIMES IT FOR DNA Damage Tolerance

The expression levels of both Rad6 and Rad18 are cell cycle controlled. Interestingly, Rad18 expression decreases during S phase, whereas Rad6 expression gradually increases during S phase and peaks in late S/G₂ phase (72, 73, 75, 78). Upon exposure to agents that halt progression of replication forks, however, the expression levels of both proteins are upregulated (53, 72, 75, 78, 124). Such a response ensures that the supply of both proteins meets the demand for them, and this response may also both drive formation of the Rad6/Rad18 complex and control its cellular location.

During unperturbed DNA synthesis, Rad6 is present in both the cytoplasm and the nucleus, but very little of it is associated with DNA (73, 78). This is expected, as the DNA-binding activity of Rad6 depends entirely upon Rad18, which is downregulated in S phase. In response to DNA damage, Rad6 is redistributed exclusively to the nucleus, where it binds to chromatin and monoubiquitinates PCNA present at stalled replication forks (72, 73). Rad6 does not contain a nuclear localization signal, however (60), whereas Rad18 contains three (85), suggesting that the cellular location of Rad6 is dictated by Rad18 during DDT as follows. In the absence of DNA damage, low levels of Rad18 ensure that the nuclear concentration of the Rad6/Rad18 complex is low, allowing Rad6 to maintain its relationship with other E3 ligases; then, upon exposure to DNA-damaging agents, upregulation of Rad6 and Rad18 sequesters them into a complex that is directed to the nucleus, where the damaged chromatin resides. Future studies are needed to confirm this pathway.

PHOSPHORYLATION OF THE MAMMALIAN RAD6/RAD18 COMPLEX ENHANCES ITS ACTIVITY

Human Rad6 is phosphorylated at conserved serine residue 120 by cyclin-dependent kinase 9 (CDK9). This PTM increases the activity of Rad6 and is critical for its role in cell cycle progression, suggesting that this modification may also be vital for the role of Rad6/Rad18 in DDT. Indeed, knockdown of CDK9 reduces ubiquitination of PCNA (104, 108, 132). How regulation of this Rad6 PTM correlates with exposure to DNA damaging agents and DDT, however, is not known. Interestingly, the basal level of phosphorylated Rad18 present during unperturbed conditions is greatly enhanced in response to DNA damage by two kinases: Dbf4/Drf1-dependent Cdc7 kinase (DDK) and ATR/Chk1-dependent c-Jun N-terminal kinase (JNK) (9, 25). Identification of the latter kinase agrees with previous independent reports of a reduction in Rad18 foci at stalled replication forks (88) and of PCNA monoubiquitination (11) upon knockdown or inhibition of ATR (ataxia telangiectasia and Rad3-related), a DNA damage checkpoint kinase (14, 19, 23, 31, 86, 135).¹ Recent studies suggest that phosphorylation of Rad18 mediates its interaction with Polη.

In mammalian cells, Rad18 exists in complexes with Polη regardless of genotoxic stresses, owing to mutual binding domains within each protein. This interaction is modulated by DNA damage-dependent phosphorylation of the Polη-binding domain within Rad18. In the absence of DNA damage, a basal level of Rad18 is phosphorylated and associated with Polη. In response to UV irradiation, both DDK and JNK phosphorylate serine residues within the Polη-binding domain of Rad18. Each kinase is independent of the other, and both DDK and JNK promote the association of Rad18 with Polη. Interestingly, both kinases are essential for the cellular tolerance of UV-induced lesions and for the role of Rad18 in TLS (9, 25, 129, 139). Furthermore, Polη protein expression is induced upon exposure to UV irradiation (26, 69). Together, these distinct responses drive formation of the Rad18-Polη complex and the subsequent redistribution of cellular Polη to replication foci. Thus, by associating with Rad18, Polη may be chaperoned into the vicinity of stalled replication forks upstream of RPA-coated ssDNA. Once the complex is localized, however, Polη may actually direct Rad6/Rad18 to PCNA encircling stalled P/T junctions. This function is unique to Polη among TLS pols, distinct from the catalytic activity of Polη, and dependent upon the ability of Polη to bind PCNA (via its PIP box and flanking domains). Thus, by bridging Rad18 and PCNA, Polη stimulates PCNA monoubiquitination following UV irradiation (26). Combining both of these purported roles for the Rad18-Polη complex in TLS suggests a dynamic “horse-and-carriage” scenario in which (a) a basal level of Rad18 is phosphorylated and associated with Polη in the absence of DNA damage; (b) phosphorylation of Rad18 is greatly enhanced in response to UV irradiation, promoting its association with Polη and hitching the “carriage” to the “horse”; (c) functioning as the “horse,” Rad18 pulls the Polη “carriage” to RPA-coated ssDNA; and (d) once it is localized to the damaged chromatin, Polη becomes the “horse” and directs the Rad18 “carriage” to PCNA residing at blocked P/T junctions, promoting monoubiquitination of PCNA.

The aforementioned studies primarily utilized UV irradiation as the DNA-damaging agent, suggesting that the horse-and-carriage scenario described above is specific to UV-induced lesions. Indeed, Polη is responsible for the error-free bypass of cyclobutane pyrimidine dimers (CPDs), the major photoproduct of UV irradiation. Given the prominence of solar UV irradiation and the presence of at least three alternative TLS pols that can bypass CPDs erroneously, such a model may account for the highly efficient and error-free bypass of UV-induced CPDs within mammalian cells (137, 144). Rad6/Rad18, however, is responsible for the monoubiquitination of PCNA in response to all agents that elicit TLS, so this “horse-and-carriage” scenario may apply broadly to any circumstance in which progression of the replication fork is stalled. Several independent studies on benzo[a]pyrene diolepoxide (BPDE), suggest this may be the case. BPDE is a genotoxic metabolite of the chemical carcinogen benzo[a]pyrene that primarily forms guanine adducts that block the progression of replication forks. Similar to UV irradiation, these studies showed that exposure to BPDE induced JNK-dependent phosphorylation of Rad18 (9). In addition, treatment of DDK-depleted cells with BDPE specifically impaired the association of Polη with stalled replication forks (25), and in Polη-depleted cells, PCNA monoubiquitination by Rad6/Rad18 was decreased after BPDE treatment (26). One study also showed that (a) JNK activity depended upon the ATR kinase recruited to and activated by RPA-coated ssDNA and (b) replication fork arrest induced by either nucleotide depletion [via hydroxyurea (HU)] or UV irradiation promoted formation of Polη-Rad18 complexes in vivo as well as their association to chromatin (9). Taken together, these findings suggest that the horse-and-carriage scenario is not genotoxin specific; rather, it describes a generic response solicited by RPA-coated ssDNA generated during S phase (139). Thus, of all available TLS pols within mammalian cells, Polη may be the default choice at any DNA lesion.

The apparent disparity in both accuracy and efficiency among various TLS events within mammalian cells, raises many issues pertaining to how TLS pols are selected for bypass across a given DNA lesion. At one end of the spectrum are lesions induced by chemicals such as BDPE. Pol κ bypasses BPDE-induced lesions in an error-free manner in vitro and is responsible for their faithful replication in vivo (3, 70, 91, 116, 141). Polη, however, replicates past such lesions in an error-prone manner in vitro and, perhaps owing to the horse-and-carriage model, contributes to the mutagenic bypass of BPDE adducts in vivo (20, 59, 102, 141). At the other end of the spectrum are the two major UV-induced lesions that elicit TLS: the more-prominent CPD and the less-prominent pyrimidine (6-4) pyrimidone photoproducts [(6-4) photoproduct] (10). Polη accurately replicates across thymine-thymine (TT) CPDs in vitro and is responsible for the highly error-free (>90%) bypass of UV-induced CPDs in vivo (125, 137). In contrast, Polη has an eightfold preference for inserting an incorrect G residue opposite the 3′ T of a (6-4) TT photoproduct, yet TLS opposite these lesions is predominantly error-free (<2% mutation frequency), suggesting that another TLS pol must replicate past (6-4) photoproducts in vivo (52, 136). In the horse-and-carriage model, Polη binds to the front side of PCNA and is afforded ample opportunity for nucleotide incorporation. Does Polη simply insert a nucleotide across from a BPDE adduct, but not across from a (6-4) photoproduct? If so, how is this insertion (or lack thereof) ensured? How is the “correct” TLS pol selected for bypass of certain lesions but not others? The linked chaperoning activities of Rad18 and Polη during TLS raise these and many more intriguing questions.

THE ACTIVITY OF THE RAD6/RAD18 COMPLEX IS MEDIATED BY VARIOUS PROTEINS

Over the last 10 years or so, more and more seemingly unrelated proteins have been implicated in modulation of the activity of the Rad6/Rad18 complex in addition to their primary duties in other cellular pathways. For example, various proteins involved in DSB repair, such as BRCA1, NBS1, PTIP, and Claspin-Timeless, have been shown to interact with Rad18 and/or PCNA and may mediate recruitment of the Rad6/Rad18 complex to blocked replication forks, enhancing monoubiquitination of PCNA during DDT. Similar behavior was also observed for two unrelated E3 proteins involved in initiation of DNA replication (CRL4^CDT2) and inhibition of p53-mediated apoptosis (SIVA1) (40, 42, 107, 119, 133, 135). Certain proteins have garnered considerable attention, each of which is summarized below.

RAD6/RAD18-MEDIATED DNA DAMAGE TOLERANCE IS SELECTED FOR WHEN A REPLICATION-BLOCKING LESION IS ENCOUNTERED

At least three distinct mechanisms may rescue a stalled replication fork during S phase. The two Rad6/Rad18-dependent pathways are primarily elicited when a small DNA adduct is encountered. The third possibility, homologous recombination (HR), is a very complex process that utilizes the undamaged sister chromatid as a homologous template to resume DNA synthesis. HR is functional throughout the S and G₂ phases of the cell cycle and is primarily summoned for the repair of more drastic genomic offenses that cannot be replicated past, such as DNA strand breaks. HR can also rescue replication forks arrested at small DNA adducts, but it may be more harmful than good in such cases, as the process can trigger cell cycle arrest, and failed attempts can lead to gross chromosomal rearrangements. Indeed, eukaryotic cells have established a temporal hierarchy to avoid such disasters: The Rad6/Rad18-mediated pathways are elicited first at replication-blocking lesions, whereas HR serves as a so-called salvage pathway and is summoned only when the other pathways are malfunctioning (61, 79). In S. cerevisiae, conjugation of SUMO (small ubiquitin-like modifier) to PCNA proctors this hierarchy.

SUMO is a small posttranslational modifier in eukaryotes that is reversibly conjugated to a plethora of proteins involved in a broad range of cellular processes (30). During normal DNA replication within S. cerevisiae, SUMO is conjugated to K127 and/or K164 within PCNA, a process referred to herein as SUMOylation. K164 is highly conserved and is the predominant attachment site, whereas K127, which is less conserved, is a secondary site (47, 93). The interaction of DNA pols with PCNA is not affected by SUMOylation of either site (16). Similar to ubiquitination, only PCNA encircling DNA is SUMOylated in vivo (94). In complete contrast to ubiquitination, however, SUMOylation of PCNA (a) is upregulated in and limited to unperturbed S phase, (b) is not induced by sublethal levels of agents that elicit TLS, and (c) has a role in maintaining the integrity of the genome during normal DNA replication (47, 93, 94, 98, 130). This role is manifested through recruitment of the Srs2 helicase, which contains distinct domains that bind PCNA, SUMO, and Rad51 independently (2, 16, 93, 98). Rad51 is an imperative HR protein responsible for locating a homologous template within the sister chromatid and annealing it to the invading strand primer (61). Because of its tandem SUMO- and PCNA-binding domains, Srs2 preferentially interacts with SUMO-PCNA (2, 93). This behavior directs a portion of cellular Srs2 to SUMO-PCNA that is encircling genomic DNA (engaged in DNA replication), where Srs2 prohibits the initiation of HR during unperturbed S phase either by dissolving inappropriately placed Rad51 nucleoprotein filaments or by preventing their formation (16, 43, 62, 67, 93, 98, 126).

During normal DNA replication within S. cerevisiae, a significant portion of cellular PCNA is SUMOylated, predominantly at K164. Conjugation of SUMO to a given PCNA trimer is not refractory toward its subsequent ubiquitination, however, so removal of SUMO is not a prerequisite for ubiquitination (93, 94, 130). In fact, genetic studies in S. cerevisiae revealed that polyubiquitination-dependent template switching actually requires previous SUMOylation of PCNA (13), possibly to keep the HR machinery at bay in order to permit conjugation of ubiquitin to PCNA. Furthermore, the activity of Rad6/Rad18 toward a given PCNA trimer is directly enhanced by the presence of SUMO on the same clamp. The N-terminal half of Rad18 from S. cerevisiae contains both a PCNA-binding domain and a SUMO interaction motif (SIM) that mediates its noncovalent interaction with SUMO. This pairing accounts for preferential binding of SUMO-PCNA by Rad18 and dramatically improves the efficiency of ubiquitin conjugation to SUMO-PCNA compared to native PCNA. In vivo, this SIM-SUMO interaction enhances the recognition of PCNA engaged in DNA replication and contributes to the full activity of Rad6/Rad18 required for efficient ubiquitin-dependent DDT (95). Taken together, these findings suggest that SUMOylation of PCNA during normal DNA replication functions as a traffic cop in S. cerevisiae: SUMOylation of PCNA directs the recovery process toward Rad6/Rad18-mediated bypass in the event that a replication block is encountered by recruiting Srs2 to inhibit HR and/or by recruiting Rad6/Rad18 to promote ubiquitination of PCNA that is engaged in DNA replication (93, 95, 98).

SUMOylation of PCNA has only recently been observed in human cells and occurs at very low levels compared with those observed in S. cerevisiae (36, 82). Hence, studies analogous to those cited above are in their infancy. Nonetheless, some insight has been provided. The human Rad18 sequence lacks an obvious SIM, and the presence of SUMO on PCNA has no effect on the ability of the human Rad6/Rad18 complex to conjugate ubiquitin to PCNA in vitro (95). Thus, SUMO does not seem to promote ubiquitination of PCNA by recruiting the Rad6/Rad18 complex. However, a mechanism similar to that described above for Srs2 may exist in humans. The human protein PARI (PCNA-associated recombination inhibitor) contains a UvrD helicase-related domain, a PIP-box, a SIM, and a Rad51-binding domain. In cells, PARI is found at very low levels, and a recent study suggests that SUMO-PCNA concentrates PARI to replication forks, where it suppresses HR by disrupting Rad51 filaments. Interestingly, PARI is not an active helicase and, in contrast to Srs2, which can move along DNA and remove multiple Rad51 molecules within a single binding encounter with DNA (82), PARI must be present in amounts stoichiometric to Rad51 in order to efficiently inhibit HR. As mentioned above, however, SUMO-PCNA levels are very low in human cells, so this method of HR suppression may not be the predominant one. Indeed, other human helicases may suppress HR by inhibiting various stages of initiation independently of SUMO-PCNA (8, 35, 49).

CONCLUDING REMARKS

It was first observed in 2002 that Rad6/Rad18 conjugated ubiquitin to eukaryotic PCNA in response to DNA-damaging agents that cause replication-blocking lesions. In the ensuing years, substantial efforts have begun to unveil the complexity and intricacies of this process. Because it involves multiple levels of regulation for a wide variety of circumstances, this process has been optimized to operate mostly during S phase and only on PCNA trimers encircling damaged DNA. In the pursuit of answers, however, more questions have been raised, particularly those pertaining to the spatiotemporal control of this process. For instance, where in relation to the replication fork does Rad6/Rad18-mediated DDT occur? There are currently two models that address this question (please see the excellent review, Reference 134). In one, replicative DNA synthesis on the afflicted strand and progression of the replication fork do not resume until the offending damage is bypassed. Hence, Rad6/Rad18 must conjugate ubiquitin to PCNA rings present at or near a replication fork, where the replisome may or may not remain intact. In the contrasting model, the replisome progresses ahead and replicative DNA synthesis resumes downstream of the lesion, leaving behind a gap opposite the offending damage. In this case, Rad6/Rad18 and the ensuing DDT function behind the progressing replication fork (and the replisome) to fill in the gap. Given the ample experimental evidence for each, the two models may not be mutually exclusive. However, each model has many persisting questions. For instance, if DDT occurs at or near the replication fork, what is the composition of the blocked P/T junction? Do the replicative polymerases or other PCNA-binding proteins remain engaged? If so, how does Rad6/Rad18 gain access to the resident PCNA clamp? In contrast, if DDT occurs behind the replication fork, does the original PCNA ring stay behind with the gap? If so, how is it retained at the stalled P/T junction? Another pressing issue is the temporal correlation between the ubiquitination events (monoubiquitination versus polyubiquitination) and the DDT pathways (TLS versus template switching). The E2/E3 pairs involved in DDT act sequentially, suggesting that the DDT pathways also do so. Indeed, UV-induced lesions are bypassed predominantly by TLS in S. cerevisiae; template switching functions only as a backup (22, 96). Furthermore, a recent in vivo report suggests that the two Rad5-related proteins in human cells, HLTF and SHPRH, actually bind to Rad18 in a DNA damage-specific manner to enhance monoubiquitination of PCNA and recruitment of the appropriate TLS pol (68). Although instances in which template switching is solicited first cannot yet be ruled out, these reports suggest that the simpler yet potentially mutagenic TLS pathway is preferred over the faithful, more complex template switching pathway. But what dissuades Ubc13-Mms2/Rad5 from extending the single ubiquitin moieties on PCNA into polyubiquitin chains? Perhaps one of the many known ubiquitin-binding proteins involved in TLS? Is it a timing issue, i.e., does TLS occur prior to the arrival of Ubc13-Mms2/Rad5 to the stalled P/T junction? Furthermore, what ultimately signals the extension of the monoubiquitin and solicitation of the template switching pathway, and when does this occur? A failed TLS attempt? These questions, in addition to those raised throughout the text, underscore the need for additional studies. Because the answers will require scrutiny of more quantitative data, in vitro biochemical experiments will play a vital role in answering these questions.

SUMMARY POINTS

1. Translesion DNA synthesis (TLS) is one of two DNA damage tolerance (DDT) pathways that can replicate past DNA lesions that halt the progression of replication forks. In this pathway, the replicative polymerase is replaced with a specialized TLS polymerase that is able to catalyze stable, yet often erroneous, nucleotide incorporation opposite damaged templates so that replicative DNA synthesis can resume.

2. DDT in eukaryotes is regulated by the E2/E3-catalyzed attachment of ubiquitin to lysine residue 164 of the the sliding clamp, PCNA. An E2 ubiquitin-conjugating enzyme catalyzes the attachment of ubiquitin to the target protein while an E3 ubiquitin ligase binds to an E2 enzyme and the target protein simultaneously, mediating the specificy of the conjugation.

3. Conjugation of a single ubiquitin by the E2/R3 pair of Rad6/Rad18 mediates the TLS pathway of DDT, wherease extension of this ubiqutin to a polyubiquitin chain by the Ubc13-Mms2/Rad5 E2/E3 pair routes DDT to the other DDT pathway, template switching.

4. The buildup of RPA-coated ssDNA downstream of a replication blocking lesion recruits Rad6/Rad18 to the blocked P/T junction where the Ra6/Rad18 complex attaches a single ubiquitin to K164 of the resident PCNA. The innate properties of Rad6 and Rad18 are uniquely attuned such that the Rad6/Rad18 complex is specifically targeted to PCNA present at stalled P/T junctions.

5. Rad6 has an instrinsic ability to form polyubiquitin chains and does not contain a nuclear localization signal (NLS) or any domain that interacts with PCNA, RPA or DNA wherease Rad18 contains multiple NLS as well as distinct domains that bind DNA, PCNA, and RPA. Thus, by pairing with Rad18, Rad6 is specifically recruited to PCNA residing at a blocked P/T junction within the nucleus where Rad18 permits the conjugation of only single ubiquitin moeities by Rad6.

6. The expresion level, assembly, and cellular location of the Rad6/Rad18 complex are stringently controlled by the cell cycle such that the nuclear concentration of the complex is low during unpertrubred S-phase. Upon exposure to agents that halt the progression of replicaiton forks, upregulation of Rad6 and Rad18 sequesters them into a complex that is directed to the nuclease where the damaged chromatin resides.

7. The activity of the Rad6/Rad18 complex is regulated by posttranslational modifications to each protein, phosphorylation in particular, as well as by a multitude of accessory proteins that modulate the recruitment of Rad6/Rad18 to RPA-coated ssDNA by various mechanisms.

8. Eukaryotic cells have established a temporal hierarchy of DNA repair pathways when dealing with replication-blocking lesions; Rad6/Rad18- mediated TLS is elicited first, followed by Ubc13-Mms2/Rad5-mediated template switching, and homologous recombination serves as a salvage pathway that is summoned only when the other pathways are malfunctioing. In essence, this is achieved by promoting the monoubiquitination of PCNA by Rad6/Rad18 while simultaneously inhibiting the initiation of homologous recombination.