Both Rep1 and Rep2 required simultaneous mutation of multiple lysine residues to effectively block their sumoylation. Based on two-hybrid assays, the three sites Semaxanib mutated in Rep13R each likely represent sites targeted for SUMO conjugation; however, it seems unlikely that the thirteen residues mutated in Rep213R each normally serve as a SUMO acceptor site. Proteins targeted for RAD001 mTOR inhibitor sumoylation often contain multiple SUMO acceptor sites, but in other studies, abolishing major sumoylation sites has been shown to lead to increased SUMO conjugation at less preferred sites and this may also be the case for Rep1 and Rep2. Despite the multiple mutations in Rep13R and Rep213R that abolished two-hybrid interaction with SUMO, faint slower-migrating Rep protein species were detected in western blot analysis, suggesting sumoylation was still occurring, albeit at a low level, on remaining lysine residues. However, since inheritance of 2 mm plasmids encoding Rep13R or Rep213R mutants was perturbed, the residual sumoylation was not sufficient for normal Rep protein function. Sumoylation can have diverse effects, altering the activity, interactions, localization or stability of a targeted protein. Our analyses demonstrated that Rep1 and Rep2 were not dependent on sumoylation for their interaction with each other, consistent with observations that bacterially-expressed Rep1 and Rep2 interact in vitro. However, the Rep13R and Rep213R mutants were defective for association with the plasmid STB partitioning locus. Even mutation of a single sumoylation site in Rep1, K305, modestly reduced Rep1-STB association. Since efficient partitioning of the 2 mm plasmid has been shown to depend on association of Rep1 and Rep2 with STB, loss of association of Rep1 and Rep2 lysine-to-arginine mutants with STB is likely to be the primary defect leading to loss of efficient plasmid partitioning. Consistent with the hypothesis that impaired sumoylation of Rep1 correlates with defective assembly of the Rep1-Rep2 complex at STB, when Rep1 was sumoylation-deficient, both Rep proteins lost their localization to discrete nuclear foci previously shown to contain clusters of the 2 mm plasmid. The uniform nuclear staining pattern observed for sumoylation-deficient Rep1 is reminiscent of that observed for wild-type Rep1 in rsc2D yeast, in which association of Rep1 with STB is also impaired, supporting Rep1 interaction with STB as critical for Rep1 sub-nuclear localization. Sumoylation has been shown to regulate the sub-nuclear localization of other proteins, a notable example being the SUMO-dependent localization of promyelocytic leukemia protein to PML nuclear bodies in mammalian cells. In this study, 2 mm plasmid foci were observed 
even when Rep1 and Rep2 were both sumoylationdeficient. Further investigation is needed to assess whether localization of plasmid clusters in their normal spindle poleproximal nuclear address is altered by changes in Rep1 and Rep2 sumoylation status. Our data suggest that Rep protein sumoylation may promote stable association of the Rep proteins with 2 mm plasmid DNA, an association reminiscent of sumoylation-dependent targeting of proteins to centromeres in yeast and in higher eukaryotes. In human cells, proteins conjugated with SUMO-2/3 are enriched at centromeres.