Regulation of Protein Function by SUMO Modification

PCNASUMO and Srs2: a model SUMO substrate–effector pair

H.D. Ulrich


Attachment of the SUMO (small ubiquitin-related modifier) to the replication factor PCNA (proliferating-cell nuclear antigen) in the budding yeast has been shown to recruit a helicase, Srs2, to active replication forks, which in turn prevents unscheduled recombination events. In the present review, I will discuss how the interaction between SUMOylated PCNA and Srs2 serves as an example for a mechanism by which SUMO modulates the properties of its targets and mediates the activation of downstream effector proteins.

  • genome stability
  • DNA replication
  • proliferating-cell nuclear antigen (PCNA)
  • small ubiquitin-related modifier (SUMO)
  • Srs2
  • yeast


SUMO (small ubiquitin-related modifier) is a post-translational protein modifier of the family of ubiquitin-like proteins, common to all eukaryotes. Modification by SUMO has been implicated in the control of numerous biological processes, ranging from nucleocytoplasmic transport to the regulation of transcription, chromosome segregation and genome stability. These functions have been discussed in several excellent reviews [15]. Like other members of the ubiquitin family, SUMO is covalently attached to lysine residues of its target proteins by an enzymatic machinery comprising an activating enzyme [E1 (ubiquitin- or SUMO-activating enzyme)], a conjugating enzyme [E2 (ubiquitin- or SUMO-conjugating enzyme)] and a ligase [E3 (ubiquitin or SUMO ligase)] [6]. In contrast with the ubiquitin system, however, a single E2, Ubc9, is responsible for the modification of all targets. SUMO conjugation exhibits a marked preference for a short consensus motif, ΨKXE (where Ψ denotes a hydrophobic and X any amino acid), which is directly recognized by Ubc9 in certain structural contexts. The number of SUMO-specific E3s continues to grow as new enzymes are being discovered, but again appears to be more limited than in the ubiquitin system.

Ubiquitin-like modifiers exert their functions by altering the properties of their target proteins. In principle, attachment of SUMO can result in the creation or the obstruction of a binding site for another cellular factor. As in the ubiquitin system, where specific ubiquitin-binding domains act as receptors that recognize ubiquitin in the context of the modified target [7], a conserved SIM (SUMO interaction motif) has been identified by several independent approaches [810]. The motif consists of a short hydrophobic stretch of four amino acids (predominantly isoleucine, leucine and valine residues), often, but not always, followed or preceded by several acidic amino acids. The motif can align with the β-sheet on the surface of SUMO in a parallel or anti-parallel orientation. Hence, the basal affinity of a downstream effector for the target protein may selectively be enhanced by an interaction between a SIM in the effector protein and the SUMO moiety on the target.

How to match substrates and phenotypes within the SUMO system?

Despite this seemingly simple concept, however, understanding how SUMO alters the properties of its targets has proven rather difficult. Many of the practical problems fall into one of several categories: in some cases, inactivation of components of the SUMO conjugation machinery causes phenotypes that cannot be assigned to a biologically relevant target protein. For example, deletion of UBC9 is lethal in the budding yeast, Saccharomyces cerevisiae, and causes a severe growth defect in the fission yeast, Schizosaccharomyces pombe, and disruption of the gene in mice results in embryonic lethality [4,5]. Problems include cell-cycle and chromosome segregation defects, but lethality has not been attributed to a specific set of substrates. Similarly, temperature-sensitive ubc9 mutants and mutants of the SUMO ligase gene MMS21 exhibit replication problems and accumulate abnormal recombination intermediates, but again no relevant substrate responsible for this phenotype has been identified [11].

A second category concerns SUMOylation events with unknown biological consequences. This applies to many SUMO substrates identified by MS in proteomic approaches [1215]. Mechanistic studies can be complicated by difficulties in finding relevant SUMOylation sites, which may be due to redundancies or the use of non-consensus sites. Alternatively, removal of the attachment sites may not result in a measurable phenotype, either because the substrate is part of a multiprotein complex with several other modified subunits present, or simply because the relevant function has not been addressed.

Finally, there are cases of matched substrates and phenotypes for which no effector protein or molecular mechanism for transmission of the function has been identified. For many years, this situation has applied to SUMOylated transcription factors, whose modification was often found to result in transcriptional repression by an unknown mechanism. Only recently has it become clear that SUMOylation of some transcription factors can result in the recruitment of specific co-repressors such as histone deacetylases or the Daxx protein [2,3,16]. However, details about how the choice of a particular effector is made by the modified target protein remain to be elucidated.

A matching substrate–effector pair: PCNASUMO (SUMOylated proliferating-cell nuclear antigen) and Srs2 in S. cerevisiae

The work of several laboratories has elucidated the mechanism of SUMO function in a case that can now be viewed as exemplary for a substrate–effector pair. PCNA (encoded in budding yeast by POL30), the eukaryotic sliding clamp protein that ensures processive action of DNA polymerases during DNA replication, was identified as a SUMOylation substrate in S. cerevisiae in a non-biased MS-based approach [17]. Unlike many other SUMO substrates, the predominant site of modification, Lys-164, was found not to obey the consensus motif, ΨKXE, despite being highly conserved among eukaryotic PCNA sequences. Lys-127, which conforms to the consensus, is a minor site of attachment that is predominantly used when Lys-164 is mutated (Figure 1). The ligase responsible for the modification is Siz1 [17,18], and the modification appears to be removed by the isopeptidase Ulp1, as a temperature-sensitive mutant shows elevated levels of SUMOylated PCNA [18].

Figure 1 The budding yeast PCNA is modified by SUMO

The ring-like structure of PCNA is shown as a cartoon on which the two SUMO attachment sites, Lys-164 and Lys-127, are highlighted in their sequence context. SUMO is depicted as a black lollipop symbol.

PCNA SUMOylation coincides with S-phase, the principal period of PCNA activity. However, it is not required for successful genome duplication, because a PCNA mutant that lacks both SUMOylation sites, pol30(K127R/K164R), exhibits no obvious replication defect. In fact, the sole phenotype that could initially be ascribed to defective PCNA SUMOylation became apparent only in the context of a second modification, the ubiquitination of PCNA. Ubiquitin is attached to PCNA in response to DNA damage by the components of the RAD6 pathway, an ensemble of ubiquitin-conjugating enzymes and ubiquitin protein ligases [17]. Mutants in the RAD6 pathway were originally identified by means of their sensitivity to various genotoxic agents and characterized by problems with completing the replication of damaged DNA [19,20]. From the work of several groups, we now know that mono-ubiquitination of PCNA promotes the replicative bypass of DNA lesions by means of recruiting damage-tolerant DNA polymerases to stalled replication forks, whereas polyubiquitination is required for an alternative, error-free damage avoidance pathway that is mechanistically not well understood [17,18,2123]. Intriguingly, PCNA is ubiquitinated exclusively at the major site of SUMOylation, Lys-164, thus implying a competition between the two modifiers. Genetic analysis indeed showed that a PCNA mutant, pol30(K164R), which is devoid of ubiquitination but can still be modified by SUMO at Lys-127, is considerably more sensitive to DNA-damaging agents than the mutant pol30(K127R/K164R), to which neither ubiquitin nor SUMO can be attached [17]. Similarly, a deletion of the gene encoding the relevant ubiquitin ligase, RAD18, is much more sensitive to DNA damage than the double mutant rad18 siz1 [18]. Based on these observations, it was postulated that SUMO and ubiquitin act antagonistically on PCNA with respect to genome stability [17]. This notion, however, suffers from two inconsistencies: on the one hand, the detrimental effect of SUMO modification on damage-sensitivity is apparent only in situations where PCNA cannot be ubiquitinated anyway, i.e. in mutants of the RAD6 pathway or the ubiquitin acceptor site mutant of PCNA. On the other hand, abolishment of PCNA SUMOylation does not lead to a concomitant increase in its ubiquitination, and vice versa [24]. This reflects a relationship between ubiquitin and SUMO that is more complicated than a simple competition for Lys-164 of PCNA [25].

Genetic analysis of the phenotypes of siz1 and pol30(K127R/K164R) mutants has given insights into the biological consequences of PCNA SUMOylation [24,26]. It was noted that their effect on the DNA-damage-sensitivity of yeast strains deficient in PCNA ubiquitination resembled that of mutants in the gene SRS2, which had originally been isolated by their ability to suppress the phenotype of RAD6 pathway mutants [27]. Defects in SRS2 function lead to a general hyper-recombination phenotype, and suppression of the damage sensitivity of rad6 or rad18 mutants requires a functional homologous recombination system. Accordingly, it has been postulated that SRS2 contributes to the channelling of lesions away from homologous recombination and into RAD6-dependent damage bypass [28]. Consistent with this hypothesis, the Srs2 protein, a multifunctional 3′–5′ DNA helicase, was found to disassemble filaments of the recombination factor Rad51 on single-stranded DNA [29,30]. In genetic terms, the effect of srs2 on the RAD6 pathway mirrors that of SUMOylated PCNA, since suppression by pol30(K127R/K164R) or siz1 mutants likewise requires homologous recombination, and combinations of srs2 with mutations of the SUMO acceptor sites or the SUMO ligase do not result in any enhancement of the phenotype [24,26].

Evidence for a direct participation of PCNASUMO in the control of Srs2 function has come from the observation that the C-terminal domain of Srs2 directly interacts with PCNA in vivo and in vitro and that SUMO modification of the clamp at Lys-127 or Lys-164 enhances this interaction [24,26]. The extreme C-terminus of Srs2 mediates the preference for the SUMOylated forms of PCNA [26], and it turns out that the C-terminal six amino acids, EIIVID, indeed match the consensus of a SIM. Surprisingly, interaction of Srs2 with PCNA is enhanced by SUMOylation of either Lys-164 or Lys-127, even though Srs2 has no measurable affinity for free SUMO. This notion suggests that the association of Srs2 with the modified clamp functions in a modular fashion that is spatially flexible enough to accommodate the modifier in a number of distinct positions in relation to PCNA itself. The extended C-terminus of SUMO itself is likely to facilitate this non-discriminate mode of binding.

In the context of DNA replication, PCNASUMO recruits Srs2 to replication forks, where the helicase counteracts the association of the recombination factor Rad51 [24]. As a consequence, Rad51 levels at replication forks are elevated in srs2 and also in pol30(K127R/K164R). These observations have resulted in a working model that postulates a safeguard function for Srs2 in protecting replication intermediates from unscheduled recombination events [31]. This model is supported by the observation that a pol30(K127R/K164R) mutant, like an srs2 deletion, displays elevated levels of spontaneous crossovers during mitotic growth. In situations of replication fork stalling due to lesions in the DNA, ubiquitin-dependent damage bypass is therefore expected to be required for survival, as indicated by the strong sensitivity of RAD6 pathway mutants in the presence of PCNASUMO. If PCNA SUMOylation is also prevented, however, stalled forks can alternatively be resolved by Rad51-dependent homologous recombination. Although it does not reveal how the switch from SUMOylation to ubiquitination on PCNA might be co-ordinated in response to DNA damage, this scenario effectively explains how a defect in PCNA SUMOylation rescues the damage-sensitivity of a RAD6 pathway mutant deficient in ubiquitin-dependent damage bypass.

PCNASUMO and Srs2: paradigm or exception?

As outlined above, SUMOylation of PCNA during S-phase creates an additional binding site on the clamp that is selectively recognized by the SIM of a PCNA-interacting protein, Srs2. As a consequence, the helicase is recruited to the modified form of PCNA, where it exerts its biological function by counteracting the formation of recombinogenic filaments (Figure 2). The system of PCNA SUMOylation can therefore be viewed as a paradigm for the mechanism of SUMO function, as it exemplifies in a very well defined manner the relationship between a SUMO substrate, its cognate receptor protein and a relevant downstream effect.

Figure 2 The interaction between PCNASUMO and Srs2 exemplifies how SUMO can affect the properties of a target protein

Conjugation of SUMO to PCNA facilitates interaction with the helicase Srs2 through a SIM in the C-terminus of Srs2. The helicase counteracts the accumulation of the recombinogenic Rad51 filament (labelled ‘51’) at replication forks.

At the same time, however, these findings indicate that PCNASUMO in budding yeast may fulfil a specialized function that, unlike PCNAUbi, is not conserved among other eukaryotes. For example, SUMOylated PCNA has not been observed in fission yeast [32]. Consistent with the absence of this modification, deletion of S. pombe SRS2 does not suppress the damage-sensitivity of RAD6 pathway mutants, and the protein lacks a discernible SIM. Whether this indicates an alternative mechanism of preventing recombination at replication forks or a complete absence thereof is yet to be determined, but another helicase, the F-box-containing Fbh1, appears to be a promising candidate that may act as a functional orthologue of S. cerevisiae Srs2 [33,34]. Similarly, helicases with remote homology to Srs2 can be found in higher eukaryotes, but none of them has yet been shown to act in a manner comparable with the budding yeast protein. Surprisingly, however, PCNA SUMOylation was observed in a number of experimental systems based on higher eukaryotes, such as Xenopus laevis egg extracts and the DT40 chicken cell line [35,36]. Owing to the absence of Lys-127 in frog and chicken PCNAs, SUMOylation was found to be limited to Lys-164, but a biological function has not been assigned in either case.

Finally, SUMOylation of PCNA in the budding yeast, in particular at Lys-127, may be able to exert inhibitory functions in addition to mediating recruitment of the Srs2 helicase. This secondary attachment site, Lys-127, is situated in the conserved IDCL (inter-domain connector loop), which overlaps with the major interaction site on PCNA, used by virtually all known PCNA-binding proteins [37]. As a consequence, the SUMO-specific E2, Ubc9, competes with other PCNA interaction partners for binding, and SUMOylation at this site is likewise expected to prevent access of other factors to the IDCL site [38]. Accordingly, mutation of the SUMOylation sites on PCNA alleviates to some degree the phenotype of an eco1ctf7203 mutant that is due to a perturbed interaction of an essential protein, Eco1, with the IDCL of PCNA. Based on this observation, PCNA SUMOylation has been proposed to act as a general ‘off-switch’ that clears the clamp of interaction factors [38]. However, the effect becomes manifest only in a temperature-sensitive mutant with faulty PCNA interaction. Hence, it is unclear whether the cell actually uses PCNA SUMOylation as a physiological means to regulate its interactions under normal conditions, and the question arises of why this principle would not be more conserved among other organisms.

In conclusion, the interaction between PCNASUMO and Srs2 is very likely to represent a general principle by which SUMO affects the properties of its target proteins, despite its restriction to the budding yeast system. Identification of matching substrate–SIM pairs will clearly be needed in order to understand the action of SUMO in other biological contexts.


Work in our laboratory is funded by Cancer Research UK and the European Commission.


  • Regulation of Protein Function by SUMO Modification: A Biochemical Society Focussed Meeting held at Manchester Conference Centre, Manchester, U.K., 25–27 June 2007. Organized and Edited by R. Hay (Dundee, U.K.) and A. Sharrocks (Manchester, U.K.).

Abbreviations: E1, ubiquitin- or SUMO-activating enzyme; E2, ubiquitin- or SUMO-conjugating enzyme; E3, ubiquitin or SUMO ligase; IDCL, inter-domain connector loop; PCNA, proliferating-cell nuclear antigen; SUMO, small ubiquitin-related modifier; SIM, SUMO interaction motif


View Abstract