Eukaryotes ubiquitylate the replication factor PCNA (proliferating-cell nuclear antigen) so that it tolerates DNA damage. Although, in the last few years, the understanding of the evolutionarily conserved mechanism of ubiquitylation of PCNA, and its crucial role in DNA damage tolerance, has progressed impressively, little is known about the deubiquitylation of this sliding clamp in most organisms. In the present review, we will discuss potential molecular mechanisms regulating PCNA deubiquitylation in yeast.
The role of PCNA (proliferating-cell nuclear antigen) ubiquitylation in DNA damage tolerance
A variety of lesions block DNA replication fork progression during S-phase and this is intrinsically dangerous: it is thought that tolerance mechanisms exist because they prevent irreversible DNA replication fork collapse when the replisome encounters a (bulky) DNA lesion, so that the restrictive nature of the active site in processive DNA polymerases impedes adequate progression of the replicative machinery.
Mechanism(s) of tolerance are essential for cells to survive exposure to genotoxic agents that damage DNA (or to resist treatments with them). In all organisms, these mechanisms ensure that DNA can be replicated even when it is damaged. This is of critical importance for cells because DNA lesions at damaged sites slow or block the progression of DNA replication forks, increasing the risk for irreversible fork collapse. DNA lesions are circumvented by low stringency DNA polymerases that are able to bypass damaged sites in a so-called translesion DNA synthesis reaction. These DNA polymerases are called TLS (translesion synthesis) polymerases. TLS polymerases are not only damage tolerant enzymes but also mutagenic because they induce an error-prone process that causes damage-dependent mutations. Not surprisingly, in mammals, mutation of TLS polymerases may be associated with genomic instability and cancer. It is well established that in eukaryotes, covalent modifications of PCNA by ubiquitin regulate the choice of alternative pathways to bypass DNA lesions during S-phase [1–5].
The current understanding of the field is that cells ubiquitylate PCNA to allow the change from a processive or replicative DNA polymerase to another (TLS) polymerase to then replicate over-processive polymerase-blocking lesions in DNA. This implies that ubiquitylated PCNA signals damaged DNA. Experimental evidence suggests that the process of TLS requires the switch between polymerases in a step-wise-dependent manner: first, high-fidelity replicative polymerases are blocked when they encounter a given DNA lesion; secondly, probably as a consequence of the first step, PCNA is ubiquitylated; thirdly, the stalled replicative polymerase is replaced by a TLS polymerase. In contrast with replicative enzymes, TLS polymerases are low-fidelity DNA polymerases, non-processive enzymes that lack any proofreading activity and are capable of replicating over DNA lesions. Importantly, TLS-polymerase DNA lesion bypass is independent of lesion repair (the DNA damage is left behind once the replisome has passed the lesion, for a later repair independent of S-phase/DNA replication).
Why the need for PCNA deubiquitylation? It is a likely consequence of the high-affinity mechanism underlying translesion DNA synthesis. Mono-ubiquitylated PCNA not only signals damaged DNA but also is a powerful/specific binding site for TLS polymerases; consequently, cells need to deubiquitylate PCNA to prevent these non-processive, low-fidelity DNA polymerases from sampling DNA more frequently than strictly required. In fact, TLS polymerases may be constantly sampling chromatin during DNA replication, both in the presence or in the absence of ubiquitylated PCNA. However, mono-ubiquitylated PCNA may prolong (in time) this interaction .
Ubiquitylation of PCNA is induced by chemicals causing disruptive covalent modifications of DNA blocking replication and involving the accumulation of single-stranded DNA. In the yeast Saccharomyces cerevisiae, PCNA is ubiquitylated during S-phase in response to detection of DNA lesions caused by MMS (methyl methanesulfonate), HU (hydroxyurea), 4-NQO (4-nitroquinoline 1-oxide), UV light, H2O2 and ionizing radiation . In fact, it has been shown that PCNA is ubiquitylated in response to MMS and UV light in all eukaryotic organisms studied to date (reviewed in ). In particular, and in response to DNA damage, PCNA is ubiquitylated in budding yeast and fission yeast [8–10], Xenopus [11–13] and human cell lines [14,15]. Evidence also suggests that abasic lesions on DNA induce PCNA ubiquitylation because TLS DNA polymerases are required for proficient replication through abasic DNA lesions in yeast and mammalian cells [16,17].
Control of PCNA ubiquitylation in S. cerevisiae
The key regulatory role of covalent modifications of sliding clamp PCNA in the control of tolerance to DNA damage is now a solidly established model in S. cerevisiae probably conserved in all eukaryotes [2–5,18]. Interestingly, the PCNA homotrimer can be SUMOylated or ubiquitylated at the same lysine residue (Lys164). Addition of the SUMO (small ubiquitin-related modifier) residue to Lys164 is controlled in yeast by the SUMO ligase complex Ubc9–Siz1 (E2–E3). PCNA is mono-ubiquitylated at the same Lys164 residue by the E2–E3 complex Rad6–Rad18; thus a PCNA monomer can either be SUMOylated or ubiquitylated (at Lys164). Mono-ubiquitylated PCNA can be further polyubiquitylated by the E2–E3 complex Mms2–Ubc13–Rad5 whose activity is modulated by the complex Esc4–Slx4. Thus, with key regulatory implications, ubiquitylation of PCNA unavoidably occurs in a sequential manner. PCNA is first mono-ubiquitylated to enhance the affinity of Rev1–Rev3–Rev7 error-prone DNA polymerases that facilitate TLS and then eventually undergo polyubiquitylation to promote template switching, the error-free component of lesion bypass that involves sister-strand recombination ( and references therein). SUMOylated PCNA at Lys164 prevents homologous recombination during S-phase in the budding yeast . Finally, also in budding yeast, Lys127 of the sliding clamp monomer can be SUMOylated to inhibit sister chromatid cohesion by preventing Eco1 acetyltransferase binding to PCNA [5,20].
Does the ATR [ATM (ataxia telangiectasia mutated)/Rad3-related kinase]checkpoint pathway contribute to the regulation of PCNA ubiquitylation?
A checkpoint response ensures cell viability in the face of DNA damage. This surveillance mechanism delays, or even arrests, cell cycle progression, thus providing an additional time for cells to repair damaged DNA [21–23]. It is initiated by the kinase ATR that activates a signal transduction cascade that regulates repair responses, including transcription of the DNA damage response genes, activation of DNA repair processes and recruitment of proteins to sites of DNA damage. All major components, as well as major details in the regulation of the transduction pathway, are conserved in eukaryotes. In S. cerevisiae, the ATR homologue Mec1 plays a key role in the signalling cascade by phosphorylating downstream effector kinases Rad53 or Chk1 (checkpoint kinase 1) in response to lesions in DNA and to defects at DNA replication forks [24,25]. Studies in S. cerevisiae and Schizosaccharomyces pombe on a potential mutual dependence of the ATR checkpoint signalling and DNA damage tolerance mechanisms indicate that they are different responses to DNA damage [7,10], suggesting that the two pathways evolved independently. However, in Xenopus and human cells, the situation is more controversial and a regulatory role of the ATR-mediated checkpoint response or some of its components in the ubiquitylation of PCNA cannot be excluded [12,13,26–29].
UBPs (ubiquitin-processing proteases) in budding yeast
In S. cerevisiae, the E2–E3 ubiquitin ligase complex that ubiquitylates PCNA in response to DNA damage during S-phase, as mentioned earlier, is well characterized (reviewed in ). PCNA is ubiquitylated by the Rad6–Rad18 complex during S-phase and, then, it is actively deubiquitylated. However, the nature of the yeast enzyme (or enzymes) that deubiquitylates PCNA remains unknown. In budding yeast, there are 17 genes that code for different ubiquitin-specific proteases (therefore potential candidates), and some of them have been extensively studied, whereas the function of others remains uncharacterized [and so they are putative DUBs (deubiquitylating enzymes)] [30,31]. These are named UBPs (from UBP1 to UBP17). Among the characterized UBPs, the ubiquitin protease complex Bre5–Ubp3 has been circumstantially related to the cellular response to phleomycin-mediated DNA damage ; this is perhaps a consequence of its role in RNA polymerase II deubiquitylation . Ubp10/Dot4, which is a nuclear DUB, has been shown to regulate histone H2B deubiquitylation, helping cells to localize histone deacetylase Sir2 to the telomeres . Interestingly, Ubp9, which is a cytoplasmic DUB, and Ubp10 have been identified as in vitro checkpoint kinase Rad53 substrates , suggesting a potential regulatory framework in response to DNA damage. The other nuclear DUB, Ubp8, is a component of the SAGA (Spt–Ada–Gcn5–acetyltransferase) complex and plays a role in gene activation also through deubiquitylation of histone H2B . Despite these lines of evidence, the S. cerevisiae PCNA ubiquitin protease remains elusive, in part because neither genetic nor biochemical screening has been designed yet to identify potential candidates. However, a comparative analysis with characterized PIP (PCNA-interacting protein) domains may reveal key information.
PIP domains in yeast (S. cerevisiae) UBPs
Assuming that PCNA deubiquitylating enzyme must interact in vivo with the sliding clamp, what sequence analysis of yeast UBPs reveals is that there are no full canonical PIP domains conserved in S. cerevisiae UBPs. QXXL/I/MXXFF is a full consensus PCNA-binding site present in many proteins that interact with the processivity factor during DNA replication . S. cerevisiae UBPs lack such a consensus binding site. However, direct or indirect interaction between PCNA and a UBP regulating deubiquitylation of covalently modified PCNA must exist, because active PCNA deubiquitylation occurs as cells exit from DNA damage. Alternatively, it has been described as a PIP box variant that allows Eco1 to interact with PCNA. Eco1/Ctf7 is directly coupled with PCNA to trigger chromatid cohesion during replication via a conserved PIP box variant (QXXL/I) within the N-terminal PCNA-binding domain . A simple analysis of the primary amino acid sequences revealed that such a PIP plain variant is present a number of times in four out of 17 UBPs in yeast, namely Ubp3, Doa4/Ubp4, Ubp8 and Dot4/Ubp10 (Figure 1).
Amino acid sequence comparison of the catalytic domain of budding yeast UBPs
Is there (auto)regulatory control on PCNA deubiquitylase(s) in yeast?
In human cell lines, USP1 deubiquitylates PCNA constantly in the absence of DNA damage. USP1 has been identified as the DUB that deubiquitylates PCNA . On UV-light-induced DNA damage, USP1 is (auto)degraded so that PCNA becomes ubiquitylated [37,38]. PCNA ubiquitylation is required for mammalian cell survival after UV irradiation, HU and MMS . However, the persistence of ubiquitylated PCNA based on USP1 disappearance on UV irradiation is not observed when the progression of DNA replication forks is blocked with HU . USP1 is also involved in the deubiquitylation of the Fanconi's anaemia protein FANCD2 that, indeed, has a role in repair of DNA cross-links [41,42]. Of particular interest is that inactivation of murine USP1 results in genomic instability and a Fanconi's anaemia phenotype . An autocatalytic cleavage event in human USP1 in response to UV irradiation suggests a mechanism of self-control that eventually would allow net ubiquitylation of PCNA (as long as the DNA damage response is active). However, yeast DUBs lack the autoproteolytic domain that is observed in human and conserved in vertebrate USP1-like DUBs . Thus yeast UBPs are unlikely to be regulated in a similar manner.
Is there any mutagenesis safeguard mechanism to keep TLS polymerases in check?
Mutation of PCNA ubiquitin protease(s) should progressively result in increasing amounts of PCNA ubiquitylation. This is potentially damaging for the cell as an increase in mono-ubiquitylated PCNA would allow TLS polymerases to sample DNA more frequently than required, resulting in a net increase in mutation rates. If this is the case, it should be predicted that loss of PCNA DUB should result in a TLS polymerases-dependent mutator phenotype. Another direct effect of the net increase in the amounts of mono-ubiquitylated PCNA is that mutation of yeast PCNA ubiquitin protease(s) should break the relative balance in wild-type cells between TLS and strand-switching bypass of DNA lesions. The Rad6/Rad18 ubiquitylation pathway is naturally unbalanced towards the error-free branch, since it is estimated that strand switching accounts for more than 70% of the lesion bypassing events .
In undamaged human cells, deubiquitylation of PCNA by USP1 reveals the possibility of a mechanism against the mutagenic effect of damage-tolerant DNA polymerases . As mentioned before, in cycling cell lines, it has been suggested that USP1 continuously deubiquitylates the DNA replication processivity factor, PCNA, as a safeguard against error-prone TLS of DNA. When UV-light-mediated DNA damage is sensed, USP1 is down-regulated by an autocatalytic-induced two-step proteolysis process. Proteolysis of USP1 allows accumulation of mono-ubiquitylated PCNA .
The deubiquitylation of PCNA may be envisioned as a control process that provides a mechanism for the cell to counterbalance the potential deleterious effect of TLS polymerases. If this is the case, it should be predicted (or so our thinking goes) that mutation of the UBP enzyme controlling PCNA deubiquitylation should have an impact on the balance of the bypass of a given lesion by the error-free or the error-prone branches of covalent modifications of PCNA.
Model(s) for deubiquitylation of PCNA in budding yeast: two working hypotheses
We hypothesize simple alternative models that explain how a replicative complex switches back and forth to processive DNA polymerases when a given lesion on DNA is bypassed by using TLS polymerases. We propose two different scenarios. In the first model (Figure 2), an as yet uncharacterized UbpX deubiquitylating enzyme (a single DUB or a number of them) forms part of the replicative complex (thanks to its capacity to interact with PCNA). When a lesion is detected, PCNA is mono-ubiquitylated to regulate the interaction either with TLS (error-prone) polymerases that will result in lesion bypass or with the Mms2–Ubc13–Rad5 ubiquitin ligase complex with a similar, but error-free, purpose. Once the lesion is left behind, UbpX-dependent PCNA deubiquitylation will take place, thus preventing excessive action of these non-processive, low-fidelity DNA polymerases (TLS polymerases) or less-processive template switching, thus allowing the return to normal replicative DNA polymerases. In the second model, ubiquitylated PCNA will signal where the DNA lesion localizes. The mechanistic details of how the DNA lesion is bypassed will not differ from the first model; however, there will be no switch back to processive DNA polymerases and mono- and poly-ubiquitylated PCNAs will also help cells to attract repair enzymes for a later repair at the end of the replicative phase or even, perhaps, independently of it. A major difference between these alternative models lies in the time that lesion bypass occurs. In the first model, lesion bypass occurs immediately after the replisome finds the DNA lesion; in the second model, it may occur later during S-phase (even in G2) after 99% of the genome has been replicated. In fact, in mammals, two lines of evidence support the second model: first, ubiquitylated PCNA remains bound to chromatin well after the lesion has been removed ; secondly, Rev1 TLS polymerase is highly expressed at the G2/M transition . Several predictions may be inferred from these hypotheses: in particular, the second model leads to the conclusion that DNA replication will end near the PCNA ubiquitylation site (that is to say, where the DNA lesion is). Another prediction is that deubiquitylation of PCNA will take place late (or very late) during an extended S-phase. Future studies are required to address these issues.
A model for UbpX function in the regulation of PCNA deubiquitylation in budding yeast
Yeast as a model for understanding the control of PCNA deubiquitylation
Regulation of PCNA deubiquitylation may differ significantly from yeast to mammals and even among yeast models, as it has been shown that PCNA is ubiquitylated during unperturbed DNA replication in the fission yeast S. pombe , whereas, at least in some strain backgrounds of S. cerevisiae, PCNA becomes ubiquitylated only on DNA damage . Thus S. cerevisiae is unlikely to be the model that harbours a general/eukaryotic mode of control of PCNA deubiquitylation given the data known from human cells regarding the control of PCNA deubiquitylation. However, S. pombe may control deubiquitylation of the DNA replication processivity factor, PCNA, in a similar manner to multicellular eukaryotes, not only because PCNA appears to be ubiquitylated during S-phase but also because PCNA ubiquitylation further increases on MMS-induced DNA damage. This plausible scenario does not exclude the possibility that deubiquitylation of PCNA may be controlled by an evolutionarily conserved DUB, which, intriguingly, remains to be identified in budding yeast and fission yeast.
In Escherichia coli, it has been shown that TLS polymerases Pol II and Pol IV freely exchange with the replicative Pol III and form alternative replisomes, even before Pol III stalls at a given lesion, to slow down fork progression. And then, Pol III switches back with Pol II and Pol IV to produce a processive replisome and resumes rapid replication in vitro . Similarly, polymerase switching in eukaryotes may result in slow DNA replication (in MMS-mediated DNA damage) as a consequence of the increase in mono-ubiquitylated PCNA forms (TLS would sample DNA more frequently and prolong the interaction in MMS-perturbed S-phase). What would be the purpose of slowing down replication when the replisome encounters DNA lesions in wild-type cells? Usefully this TLS polymerases-dependent replication fork progression slowdown may give the cell additional time to repair its DNA by the BER (base excision repair) or NER (nucleotide excision repair).
From our point of view, three open questions remain to be answered: first, the nature of the yeast PCNA deubiquitylating enzyme (or enzymes); secondly, when lesion bypass occurs (prior to or after encountering the covalent modification of DNA that disrupts replicative fork progression); and thirdly, whether there is a switch that restores in vivo processive DNA replication resumption once the replisome has passed the lesion (on the understanding that the replisome contains both processive and non-processive DNA polymerases). A critical point (common to all three questions) is the analysis of the timing of PCNA deubiquitylation during the cell cycle.
Ubiquitin–Proteasome System, Dynamics and Targeting: 4th Intracellular Proteolysis Meeting, a Biochemical Society Focused Meeting held at Institut d'Estudis Catalans, Casa de Convalescència, Barcelona, Spain, 27–29 May 2009. Organized and Edited by Bernat Crosas (Institute of Molecular Biology of Barcelona, Spain), Rosa Farràs (Centro de Investigación Príncipe Felipe, Valencia, Spain), Gemma Marfany (University of Barcelona, Spain), Manuel Rodríguez (CIC bioGUNE, Derio, Spain) and Timothy Thomson (Institute of Molecular Biology of Barcelona, Spain)
Work in our laboratories was funded by the Spanish Science Ministry, Junta de Castilla y León and Health Institute Carlos III.