Multiple domains in Siz SUMO ligases contribute to substrate selectivity.

Saccharomyces cerevisiae contains two Siz/PIAS SUMO E3 ligases, Siz1 and Siz2/Nfi1, and one other known ligase, Mms21. Although ubiquitin ligases are highly substrate-specific, the degree to which SUMO ligases target distinct sets of substrates is unknown. Here we show that although Siz1 and Siz2 each have unique substrates in vivo, sumoylation of many substrates can be stimulated by either protein. Furthermore, in the absence of both Siz proteins, many of the same substrates are still sumoylated at low levels. Some of this residual sumoylation depends on MMS21. Siz1 targets its unique substrates through at least two distinct domains. Sumoylation of PCNA (proliferating cell nuclear antigen) and the splicing factor Prp45 requires part of the N-terminal region of Siz1, the ;PINIT' domain, whereas sumoylation of the bud neck-associated septin proteins Cdc3, Cdc11 and Shs1/Sep7 requires the C-terminal domain of Siz1, which is also sufficient for cell cycle-dependent localization of Siz1 to the bud neck. Remarkably, the non-sumoylated septins Cdc10 and Cdc12 also undergo Siz1-dependent sumoylation if they are fused to the short PsiKXE SUMO attachment-site sequence. Collectively, these results suggest that local concentration of the E3, rather than a single direct interaction with the substrate polypeptide, is the major factor in substrate selectivity by Siz proteins.

Sumoylation of the yeast Gcn5 protein.

Sumoylation, the process by which the ubiquitin-related SUMO protein is covalently attached to lysine side chains in other proteins, is involved in numerous processes in the eukaryotic cell, including transcriptional repression. In this study, we identify Gcn5, the histone-modifying subunit of the transcriptional regulatory complex SAGA, as a sumoylation substrate in yeast. In vitro, multiple sumoylation of recombinant Gcn5 alone or as a trimer with its interacting proteins Ada2 and Ada3 did not affect Gcn5's histone acetyltransferase (HAT) activity, suggesting that modification of Gcn5 with yeast SUMO (Smt3) may not directly regulate its HAT function. Through site-directed mutagenesis, the primary in vivo sumoylation site was identified as lysine-25, although an unsumoylatable K-to-R mutation of this residue led to no obvious in vivo effects. However, fusion of SUMO to the N-terminus of Gcn5 to mimic constitutive sumoylation resulted in defective growth on 3-aminotriazole media and reduced basal and activated transcription of the SAGA-dependent gene TRP3. Taken together with recent identification of multiple additional subunits of SAGA as sumoylated proteins in vivo, these data suggest that Gcn5 sumoylation may have an inhibitory role in transcriptional regulation.

Misregulation of 2 microm circle copy number in a SUMO pathway mutant.

Attachment of the ubiquitin-like protein SUMO to other proteins is an essential process in Saccharomyces cerevisiae. However, yeast mutants lacking the SUMO ligases Siz1 and Siz2/Nfi1 are viable, even though they show dramatically reduced levels of SUMO conjugation. This siz1Delta siz2Delta double mutant is cold sensitive and has an unusual phenotype in that it forms irregularly shaped colonies that contain sectors of wild-type-appearing cells as well as sectors of enlarged cells that are arrested in G(2)/M. We have found that these phenotypes result from misregulation of the copy number of the endogenous yeast plasmid, the 2 microm circle. siz1Delta siz2Delta mutants have up to 40-fold-higher levels of 2 microm than do wild-type strains. Furthermore, 2 microm is responsible for the siz1Delta siz2Delta mutant's obvious growth defects, as siz1Delta siz2Delta [cir(0)] strains, which lack 2 microm, are no longer heterogeneous and show growth characteristics similar to those of the wild type. Possible mechanisms for SUMO's effect on 2 microm are suggested by the finding that both Flp1 recombinase and Rep2, two of the four proteins encoded by 2 microm, are covalently modified by SUMO. Our data suggest that SUMO attachment negatively regulates Flp1 levels, which may partially account for the increased 2 microm copy number in the siz1Delta siz2Delta strain.