Ub‐family modifications at the replication fork: Regulating PCNA‐interacting components

A vast array of proteins is recruited to the replication fork in a dynamic and coordinated manner through physical interactions with Proliferating Cell Nuclear Antigen, PCNA. How this complex exchange of PCNA binding partners is choreographed to ensure proper replication origin licensing, DNA synthesis during normal replication or repair of DNA damage, chromatin assembly, DNA methylation, histone modification, and sister chromatid cohesion is only beginning to be appreciated. In this review, several roles of ubiquitin‐related modifications in the recruitment and turnover of PCNA‐interacting proteins at the replication fork are considered.


Introduction
Replication of the genome and packaging of that duplicated DNA into chromatin during S phase is achieved by a complex collection of proteins that are dynamically present at the replication fork. Proliferating Cell Nuclear Antigen, PCNA, plays a central role in coordinating the association of these replication factors during DNA replication as well as during recognition and repair of DNA damage. The association of proteins with the replication fork through interactions with PCNA is regulated by several mechanisms including: (1) differences in affinity of individual binding partners for PCNA which can result in displacement of one PCNA interacting factor by another, (2) the addition or removal of posttranslational modifications to PCNA which either promotes or prevents protein-protein interactions, (3) the unloading and replacement of PCNA by alternative clamp loading complexes, (4) ubiquitin-dependent proteolytic digestion of PCNA followed by replacement of PCNA, and (5) ubiquitination of interacting proteins bound to chromatin-associated PCNA which promotes their degradation. The temporal exchange of PCNA binding partners and the regulation of this process as well as the composition of alternative PCNA complexes on the leading and lagging strands of DNA are only partially appreciated. This review explores several ways in which interactions between PCNA and binding partners are choreographed and roles of ubiquitin related modifications in this dance.
PCNA interacts with numerous proteins to facilitate DNA replication and repair, epigenetic processes and sister chromatid cohesion. At the replication fork, PCNA forms ring-shaped trimers around DNA in which each monomer is arranged in a head to tail orientation within the ring (Fig. 1). During DNA replication and repair, PCNA is loaded onto DNA at the 3 0 ends of primer-template junctions by the clamp-loading complex, RFC [1]. Once loaded onto DNA, PCNA acts a sliding clamp that interacts with and enhances the processivity of the DNA polymerases Pole and Pold on the leading and lagging strands of DNA [2][3][4][5]. Many PCNA binding proteins interact with a hydrophobic pocket and the interdomain connecting loop on PCNA (Fig. 1A) through a conserved PCNA binding motif known as a PIP (PCNA-interacting protein) box [6] (Fig. 1B). PIP boxes fold into a hydrophobic domain that binds a hydrophobic pocket beneath the interdomain connecting loop on PCNA [7,8]. Other residues outside of the PIP box may also affect the strength of interactions with unmodified PCNA as well as with different modified forms of PCNA (see below). In addition to Pold and Pole, PCNA acts as a binding platform for several other proteins directly involved in DNA synthesis ranging from the PIP box-containing Flap endonuclease I, Fenl [9], and DNA ligase I [10,11] to several alternative polymerases and other proteins involved in DNA repair (see below). PCNA also facilitates the establishment of sister chromatid cohesion during S phase through recruiting the PIP box-containing acetyltransferase Eco1 to chromatin [12]. Competition between PIP box proteins for binding to PCNA likely contributes to complex formation and stability, although the relative affinity of most proteins containing PIP boxes for PCNA remains to be determined.

PCNA, epigenetic processes and links to ubiquitin
In mammals, several PCNA binding factors influence chromatin composition and epigenetic processes including the histone acetyltransferase p300 [13], the H4 K20 methyltransferase PR-SET7 that co-localizes with PCNA at replication foci in vivo [14] (PCNAdependent ubiquitination of PR-Set7 is described below), the maintenance DNA methyltransferase Dnmt1 [15,16], and the histone deacetylase HDAC1 [15,17,18]. PCNA is also bound by the chromatin assembly factor CAF-1 [19,20], which, in mammals, targets MBD1, a methyl CpG binding protein, and SETDB1, a H3 K9 methyltransferase, to replication foci [6,21]. In budding yeast, silencing defects associated with PCNA mutants have been linked to defects in CAF-1 and ASF1-dependent chromatin assembly pathways [22,23]. These PCNA mutants also have defects in interacting with the acetyltransferases SAS-I plus Rtt109 and display global hypoacetylation of histone H4 K16, H3 K9 and H3 K56 [24,25]. Such histone modification defects result in mislocalization of silencing proteins to off-target sites and inappropriate silent chromatin formation as well as disruption of silent chromatin at normally silenced loci. Analogous epigenetic defects may occur in mammals when certain protein interactions with PCNA are disrupted.
In mammals, the transmission of histone modification and DNA methylation states during DNA replication has been proposed to be facilitated by the Epigenetic Code Replication Machinery (ECREM) [26]. A central component of ECREM, UHRF1 (Ubiquitin-like PHD RING Finger1), is a multi-domain protein that exhibits ubiquitin ligase activity [27][28][29] and targets Dnmt1 to newly synthesized DNA by preferentially binding hemimethylated DNA [30][31][32][33] as well as di-and trimethylated forms of histone H3 K9 [28]. Both UHRF1 and Dnmt1 bind the H3 K9-specific methyltransferase G9a, co-localize with G9a in vivo and may also facilitate the transmission of these modifications upon DNA replication [34,35]. Like PCNA, UHRF1 directly interacts with Dnmt1, is required for the maintenance of DNA methylation patterns, and co-localizes with Dnmt1 and PCNA in vivo [32,36]. In addition, UHRF1 regulates Dnmt1's activity during the cell cycle via proteolytic degradation. Dnmt1's stability is modulated by reversible acetylation and ubiquitination events by a large complex that includes PCNA, UHRF1, HDAC1, the acetyltransferase Tip60 and the ubiquitin protease USP7/HAUSP along with Dnmt1 [37]. Significantly, recent studies on human gliomas indicate that disruption of interactions between UHRF1, Dnmt1 and PCNA contributes to oncogenesis by promoting global DNA hypomethylation and correlates with poor survival rates [38]. Several components of this larger complex may also form functional subcomplexes. For example, PCNA, Dnmt1 and HDAC1 form a complex in vitro and co-localize in vivo [15,17,18] (see also [39,40]). However, how interactions between such alternative complexes and PCNA are regulated during DNA replication to promote locus-specific chromatin modification patterns governing epigenetic processes is currently poorly defined.
Multiple identical or different proteins can theoretically interact with PCNA simultaneously through dynamic protein-protein interactions at the replication fork, as three interconnecting loop domains are present per PCNA trimer. Consistent with different partners being capable of binding simultaneously to different subunits of PCNA, the Sulfolobu solfataricus homolog of PCNA, P2, is a heterotrimer, with one subunit preferentially binding to DNA polymerase, a second to DNA ligase I and a third to Fen1. Together, these interactions with PCNA would constitute the predicted architecture of replication complex on the lagging strand during the synthesis and processing of Okazaki fragments [41]. Presumably, ligation of Okazaki fragments could somehow signal the release of the replication factors from the lagging strand and permit other factors to bind PCNA, including CAF-1 for nucleosome assembly and enzymes like Dnmt1 or HDAC1 for the propagation of epigenetic states. Alternatively, a model in which chromatin-bound PCNA exists in a back-to-back dimer of homotrimers [42] would permit simultaneous association of both replication factors and chromatin processing factors at the replication fork as well as provide a possible mechanism for targeting such factors to the leading strand where PCNA is continually bound by Pole. However, this model does account for how back-to-back homotrimers could be assembled onto DNA as RFC exhibits orientation-specific loading of PCNA onto DNA [1]. Significantly, in the autoimmune disease Lupus Erythematosus, autoantibodies against PCNA are often present. During disease progression, autoantibodies appear to be generated against entire multi-protein complexes containing PCNA [43]. Investigation of such autoimmune responses to PCNA may clarify the composition of complexes containing PCNA at the replication fork as well as how misregulation of protein-protein interactions with PCNA contributes to abnormal epigenetic processes during the initiation and progression of autoimmune diseases [44].

PCNA ubiquitination
The addition and removal of posttranslational modifications to PCNA affect protein-protein interactions with PCNA during responses to DNA damage. PCNA is differentially modified with ubiquitin-related moieties in response to different forms of DNA damage to promote DNA repair through several alternate repair pathways. When the DNA replication machinery becomes stalled upon encountering a lesion, activation of DNA damage responses that permit the continuation of DNA replication through an ''error prone pathway'' can occur. In this pathway, termed translesion DNA synthesis, alternative low fidelity DNA polymerases are utilized to replicate the damaged DNA template. Alternatively, a poorly defined ''error free'' pathway, possibly involving template switching, can be used in which homologous recombination repair proteins utilize the newly synthesized sister chromatid to repair the site of DNA damage. When the lesion reflects incomplete lagging strand synthesis, an additional, recently discovered pathway involving ubiquitination of PCNA becomes activated.
Ubiquitination of proteins occurs through three consecutive steps. The 8.6 kDa protein ubiquitin is initially activated by an E1 ubiquitin-activating enzyme in an ATP-dependent manner. Activated ubiquitin is then transferred from the E1 enzyme to an E2 ubiquitin-conjugating enzyme via a thioester intermediate which covalently links ubiquitin via a C terminal glycine residue to a lysine residue on the substrate targeted by an E3 ubiquitin ligase [45,46]. Monoubiquitination of the protein may alter proteinprotein interactions, and in the case of PCNA, occurs in response to specific forms of DNA damage. Alternatively, this modification may then be converted to different forms of polyubiquitin chains, depending on the type of linkages found between ubiquitin moieties. When polyubiquitin chains are created through ligation via e.g., K48 or K29 on the ubiquitin moieties, the protein will generally be marked for proteolysis via the 26S proteasome [47]. PCNA can be targeted for degradation in this manner and misregulation of turnover of PCNA leads to genome instability [48,49]. However, what protein-protein interactions at the replication fork are regulated by proteolysis of PCNA has yet to be explored in depth. K63 on ubiquitin is also used to form polyubiquitin chains on PCNA. As described below, K63 polyubiquitin chains do not trigger proteolysis of PCNA, but rather act as signals signifying the presence of specific forms of DNA damage.

PCNA K164 monoubiquitination promotes DNA repair by translesion synthesis
In the error prone repair pathway, when the replication fork stalls at the site of DNA damage, the major replicative polymerase, Pold or Pole, is thought to be exchanged for a translesion synthesis polymerase, e.g., Polf or Polg, which can then synthesize DNA through the site of the lesion. Once beyond the site of damage, Pold or Pole will once again resume the role of replicating DNA [50][51][52]. Whether translesion synthesis is relatively mutagenic is dependent on the nature of the lesion and the alternative polymerase utilized to replicate through that lesion ( [53] and references within).
Chromatin-associated PCNA is ubiquitinated at K164 in response to damage in the template DNA strand caused by DNA alkylating agents or UV irradiation [54] (Fig. 2). In budding yeast and mammals, Rad6p/Rad18p orthologs catalyze monoubiquitination of PCNA on K164 [54,55] and RNF8 is also important for monoubiquitination of PCNA upon exposure to UV irradiation or MNNG in mammalian cells [56]. In addition, CLR4 Cdt2 (see below) can directly monoubiquitinate PCNA to promote translesion DNA synthesis [57]. Both stalling of the replication fork and monoubiquitination of PCNA are perquisites for the exchange of Pold and Polg [50]. Polg specifically interacts with the monoubiquitinated form of PCNA through an ubiquitin binding domain, UBD, plus a PIP box [58][59][60]. Once synthesis through the lesion has been completed, Polg may be removed from the replication fork through a degradation mechanism involving CRL4 Cdt2 -dependent ubiquitination requiring chromatin-associated PCNA [61]. As monoubiquitinated PCNA is refractory for replacing Polg with Pold, deubiquitination of K164 is also thought to promote the reassociation of Pold with PCNA at the replication fork [50]. In mammals, deubiquitination of PCNA is mediated by the deubiquitinase USP1 in conjunction with its interacting partner UAF1 in the absence of DNA damage [62][63][64]. Monoubiquitinated PCNA accumulates upon DNA damage; UV irradiation leads to the downregulation of USP1 protein and transcripts, but how monoubiquitinated PCNA is stabilized in response to other damaging agents is not yet known [62][63][64][65].

PCNA K164 polyubiquitination promotes error free DNA repair
''Error free'' repair is triggered by polyubiquitination of PCNA and results in the use of homologous recombination repair proteins to repair DNA damage. This poorly defined pathway appears to promote DNA repair through the use of the newly synthesized sister chromatid as a template or for recombination [66][67][68]. During error free repair, which may involve template switching (Fig. 2), chromatin-associated PCNA is first monoubiquitinated at K164 in a Rad6/Rad18 dependent manner. This prerequisite modification is then targeted for polyubiquitination through K63 linkages by the ubiquitin conjugating complex Ubc13-Mms2 and Rad5 in Saccharomyces cerevisiae [54,69,70], the Rad8 ortholog in Schizosaccharomyces pombe [71], and HLTF and SHPRH in human cells [72][73][74][75]. Thus, poly-but not monoubiquitination of PCNA is lost in ubc13, mms2 and rad5 mutants, whereas both mono-and polyubiquitination is disrupted in rad6 and rad18 mutants. Rad5 also exhibits DNA helicase activity and, on model substrates, can mediate replication fork regression, consistent with a role in bypassing lesions during DNA replication via template switching [76]. What additional repair proteins specifically interact with polyubiquitinated PCNA to promote error free repair is not known. USP1 also negatively regulates polyubiquitination in mammals and therefore may participate in deubiquitination of PCNA during recovery after DNA repair by this pathway [77].

PCNA ubiquitination in response to replication stress on the lagging strand
Recently, a previously unrecognized DNA damage response to replication stress was uncovered in S. cerevisiae that results in ubiquitination of K107 on PCNA in the presence of defective Okazaki fragment processing in DNA ligase-deficient cells [78] (Fig. 2). Ubiquitination of PCNA upon depletion of DNA ligase I has also been observed in human cells [78], although the modified residue(s) on human PCNA has not yet been mapped. Mono-and polyubiquitination of K107 on PCNA requires Mms2p, Ubc4p and Rad5p. Monoubiquitination of PCNA is sufficient to result in checkpoint activation in DNA ligase-deficient cells [78], but how this ubiquitination event affects binding of interacting partners to PCNA has not yet been explored. Ubiquitination on K107 appears to be critical for cellular recognition of damage to the nascent DNA strand caused by a failure to ligate Okazaki fragments, as cdc9-1 pol30 K107R mutants are inviable [78]. Polyubiquitin chains in this pathway are created through K29 linkages [78], which are often associated with targeting proteins for degradation by the proteasome [47]. Thus, turnover of PCNA could be involved in recovery from responses to this form of DNA damage.

PCNA sumoylation
Sumoylation of PCNA also influences protein-protein interactions at the replication fork by creating or disrupting binding sites for partners of PCNA. Sumoylation of PCNA occurs primarily in S phase in a DNA damage-independent manner and may play a role in regulating normal DNA replication [54,79]. Sumoylation of PCNA is also involved in responses to DNA damage [53,54,66,80], and may play a negative regulatory function by preventing ubiquitination at the same residues on PCNA. In S. cerevisiae, cell cycle-regulated sumoylation of PCNA is the result of the combined action of the E2 Ubc9 and the E3 Siz1, which preferentially sumoylate chromatinbound PCNA [79], and the hydrolase Ulp1, which removes SUMO from PCNA [55,81,82]. Ubc9/Siz1 sumoylates PCNA at K164 and, to a lesser extent, K127 or K127 plus K164 [54,79,83,84] (Fig. 2). Ubc9/Siz1 also polysumoylates PCNA [54,55,80], although the significance of mono-versus polysumoylation is not known. Ubc9 can sumoylate K127 on PCNA in the absence of Siz1, although a second E3, Siz2, may also play a minor role in sumoylation under these conditions [79]. Sumoylation of PCNA is also conserved in vertebrates [84]. Interestingly, human p150, the PIP box-containing large subunit of CAF-1, binds SUMO2/3 in addition to PCNA and mediates localization of SUMO2/3 to replication foci [85]. Whether SUMO2/3 is targeted to PCNA or another factor at the replication fork via CAF-1 during chromatin assembly on newly synthesized DNA remains to be determined.
Binding of several proteins to PCNA is regulated by sumoylation. Sumoylation of PCNA promotes binding to the helicase Srs2p and suppresses homologous recombination during S phase [86,87].
Interactions between an alternative clamp-loading complex and PCNA at the replication fork also may be regulated by sumoylation. Elg1, the large subunit of an alternative clamp loading complex, binds both PCNA and the ubiquitin protease USP1-UAF1 [88,89], and functions in sister chromatid cohesion [90,91] plus the maintenance of genome stability [89,92,93]. Although Elg1 contains a PIP box and interacts with unmodified PCNA, Elg1 also contains three SUMO interacting domains, SIMs, and preferentially co-precipitates with sumoylated PCNA [94]. Sumoylation of PCNA on K127 is predicted to disrupt binding of PIP box-containing partners through interfering with access to the hydrophobic pocket on PCNA. Consistent with this model, sumoylation of PCNA represses binding of Eco1 in budding yeast and perturbs sister chromatid cohesion [12]. Further biochemical characterization is required to clarify how sumoylation impacts the composition of proteins bound to PCNA at the replication fork However, sumoylation and ubiquitination appear not to be required for PCNA to contribute to the propagation of epigenetic states. Using PCNA mutants that are mutated at K127 and K164, we have observed that the nonubiquitinated and non-sumoylated forms of PCNA can support the establishment and maintenance of silent chromatin in budding yeast (Fig. 3A), in contrast to other PCNA mutants with defects in CAF-1 (e.g., Fig. 3B) and ASF1-dependent chromatin assembly pathways [22,23] (see above).

PCNA, CRL4 cdt2 -dependent ubiquitination
Ubiquitination of proteins during interactions with PCNA can promote the turnover of protein complexes at the replication fork. In addition to being ubiquitinated directly, PCNA is required for CRL4 Cdt2 -dependent ubiquitination of several proteins whose activities are tightly regulated during the cell cycle via proteolysis [95]. CRL4 cdt2 is a member of the Cullin-Ring E3 ubiquitin Ligase family and consists of the scaffold protein Cul4A/B, an adaptor protein Ddb1 that bridges the complex's substrate recognition factor Cdt2 to Cul4, and a small Ring finger protein Rbx1/Roc1, which recruits a different E2 ubiquitin-conjugating enzyme, UBCH8, UBE2G1 or UBE2G2, to the complex, depending on the substrate being targeted [96]. CRL4 Cdt2 is unusual in that ubiquitination of its substrates during S phase and in response to DNA damage requires their association with chromatin-bound PCNA [95].

PCNA-dependent ubiquitination of PR-Set7 regulates H4 K20 methylation
PCNA participates in cell cycle-dependent regulation of H4 K20 methylation, which is important for maintaining silent chromatin, chromosome condensation and genome stability, through binding the histone methyltransferase PR-Set7 (also referred to as Set7, Set8 or Set9/KMT5A in S. pombe). PR-Set7 catalyzes monomethylation of H4 K20 [112][113][114]. H4 K20me1 levels are controlled through both proteolysis of PR-Set7 [14,110,115,116] and further modifications to this residue [117,118]. H4 K20me1 levels are very low during S phase and rise in late S and G2 phases, peak during mitosis, then fall again during G1 [116,117,119]. This reduction in H4 K20me1 levels is primarily due to conversion to H4 K20me2 within one to two cell cycles and, for a small percentage of H4, conversion to H4 K20me3 by the methyltransferases Su(var)4-20h1 and Su(var)4-20h2 [117,118,120,121]. Thus, defects in PR-Set7-dependent monomethylation of H4 K20 also result in defects in di-and trimethylation of H4 K20. Although H4 K20me1 has been linked to transcriptional repression, H4 K20me1 also tends to be associated with transcriptionally active regions and enriched in gene bodies, whereas H4 K20me3 is enriched in some promoters, repetitive elements pericentric heterochromatin, and in imprinting control regions [120,[122][123][124][125][126][127].
PR-Set7 is subject to chromatin-bound PCNA-dependent ubiquitination by CRL4 Cdt2 plus UBCH8, which then targets PR-Set7 for proteosome-dependent degradation in G1 and S phase as well as in response to DNA damage [96,[108][109][110]128]. The E3 ubiquitin ligase CRL1 Skp2 also play a minor role in directing (PCNA-independent) proteolysis of non-chromatin associated PR-Set7 during the G1/S phase transition [108,110,115]. Chemical inactivation of the proteasome permits visualization of co-localization of PR-Set7 and PCNA at replication foci and depletion of PR-Set7 leads to H4 K20 mono-and trimethylation defects, reduced rates of replication fork progression, impaired S phase progression and double stranded DNA breaks [129]. Expression of PR-Set7 mutants lacking a PIP box results in stabilization of PR-Set7, a decrease in H4 K20me1, an increase in H4 K20me3 and the accumulation of cells with <2C and >4C DNA content, consistent with chromosome missegregation and/or re-replication events having occurred Fig. 3. Ubiquitination and sumoylation of PCNA are not required for silencing. A. and B. Colony color assays [149] monitoring silencing at HMR::ADE2 in S. cerevisiae expressing POL30 or the indicated pol30 mutants [54,150]. Cells were plated onto rich media (YPD) and grown for 2 days at 30°C, then stored at 4°C for three or 4 days prior to image acquisition. Independent experiments are shown in (A) and (B) and the variation in pigment accumulation between experiments reflect differences in exposure conditions for images and time of storage at 4°C. Dark pink colonies indicate HMR is silenced and loss of silencing results in white, light pink or sectored colonies. pol30-8 contains RD61,63AA and exhibits silencing defects in a CAF-1-dependent pathway as well as defects in H3 K9ac, H3 K56ac, and H4 K16ac [22][23][24][25]. [108]. Depletion of PR-Set7 also results in fewer replication origins firing and an increase in the distance between active replication forks, implying that proper regulation of PR-Set7 expression via PCNA and CRL4 Cdt2 is critical for regulating origin licensing and/ or the timing of origin firing [129]. As PR-Set7 associates with mitotic chromosomes, PR-Set7 is likely targeted to chromatin by PCNA-independent mechanism(s) as well [110,116].
Misregulation of PCNA-dependent degradation of PR-Set7 also contributes to genome instability through altered chromosome condensation and DNA damage responses [109,[130][131][132]. PCNA plays a role in initiating responses to DNA damage through targeting PR-Set7 to sites of damage [110]. In turn, catalytically active PR-Set7 is required for the recruitment of a mediator of DNA damage checkpoint activation, 53BP1 (Crb2 in S. pombe), to damaged DNA, likely via 53BP1 binding to methylated H4 K20 [110,133,134]. Subsequent CRL4 Cdt2 -dependent degradation of PR-Set7 may then enable binding of other factors to PCNA during DNA repair, but how this is coordinated with ubiquitination of PCNA in response to DNA damage is unknown. Together, these findings are consistent with PR-Set7 functioning during DNA replication and repair in a manner in which rapid turnover of protein-protein interactions with PCNA is important for function.

PCNA-dependent ubiquitination of Cdt1 regulates licensing of replication origins
A major mechanism for preventing re-replication in metazoans involves cell cycle restriction of origin licensing and DNA replication through the regulated protolysis of Cdt1 [100,101,106,135,136]. Both Cdt1 and the replication factor Cdc6 are required to recruit the replicative helicase MCM2-7 to origins of replication during G1 and rapid degradation of Cdt1 restricts initiation of DNA replication to once per S phase [137,138]. Ubiquitination of Cdt1 by CRL4 Cdt2 targets Cdt1 for degradation, requires chromatin-bound PCNA and, thus, is triggered during DNA replication or repair [99,100,135,[139][140][141]. Similarly, PCNA-dependent ubiquitination of the CDK inhibitor p21 by CRL4 Cdt2 also regulates origin firing by promoting nuclear export of Cdc6 [99,100,[105][106][107]142]. Whether both Cdt1 and p21 can bind PCNA simultaneously for their ubiquitination and subsequent degradation is unknown.
Shortly after UV irradiation of mammalian cells, Cdt1 transiently accumulates at sites of DNA damage in its PIP box-and a PCNA-dependent manner in conjunction with the arrival of PCNA and Cdt2 [147]. Cdt2 similarly requires PCNA to accumulate transiently at sites of DNA damage, whereas PCNA remains localized to sites of DNA damage cells after degradation of Cdt1 [147]. Interestingly, ubiquitination of PCNA in response to DNA damage can inhibit proteolysis of Cdt1 under certain conditions [98]. These observations imply DNA damage-induced ubiquitination or sumoylation of PCNA prevents CRL4 Cdt2 -dependent degradation by inhibiting binding of Cdt1 to PCNA. Other substrates of CRL4 Cdt2 may be stabilized similarly upon modification of PCNA in the presence of DNA damage. Further studies are needed to clarify the temporal order of events governing PCNA-dependent protein-protein interactions and posttranslational modifications to PCNA during responses to DNA damage.

Perspectives
High fidelity genome duplication and efficient propagation of epigenetic states are governed by a collection of dynamic protein complexes containing PCNA at the replication fork. In recent years, great strides have been made in identifying PCNA-interacting factors and modifications to PCNA as well as several biological processes mediated by both. In the near future, identification of which proteins bind PCNA simultaneously to form transient functional complexes in vivo will be critical for the eventual understanding of how coordinated exchange of these complexes at the replication fork contributes to the regulation of DNA replication, DNA repair, epigenetic processes and sister chromatid cohesion. Application of alternative strategies for rapidly probing transient protein-protein interactions, including single molecule-based approaches, have the potential to enable visualization of how these interactions with PCNA occur dynamically. Moreover, an appreciation of how DNA damagedependent ubiquitination of PCNA influences interactions with its numerous binding partners is needed to clarify why DNA damage can influence a broad range of PCNA-dependent processes, including epigenetic gene regulation and the fidelity of chromosome segregation.