Here we examine how mutations of the Smc5–Smc6 complex affect recombination at centromeric sequences, kinetochore protein modification, spindle properties, and chromosome loss. 2D gel analysis provides physical evidence of increased levels of recombination intermediates at centromeric regions in smc6-56
mutant cells (). Consistent with this, live cell imaging shows that smc6-56
cells contain increased levels of CEN-Rad52 foci (). These two pieces of evidence suggest that Smc6 is required to regulate recombination at centromeric DNA and surrounding regions during growth. Since a sumoylation-defective rad52
mutant that generates fewer centromeric foci can partially suppress the nocodazole sensitivity of smc6-56
(), recombinational roles of Smc6 at centromeric regions likely affect centromere-related functions. These data also suggest that recombinational repair at centromere regions involves a subpathway that entails both sumoylated Rad52 and the Smc5–Smc6 complex. A requirement for both has also been found in double strand break repair in rDNA, though mms21-11
does not affect Rad52 sumoylation 
. These results provide the genetic bases for further examination of how the two collaborate in recombinational repair at both loci.
The notion that recombinational roles of Smc6 are relevant to centromere function is also consistent with the previous observation that removal of the recombination protein Mph1 suppresses the centromeric separation defect of smc6-56
. This defect was initially thought to be unrelated to recombination, as rad51Δ
did not suppress it 
. However, we found that rad51Δ
exhibited similar centromeric separation defects as smc6-56
(). The reason for this defect is unclear, but may be due to the deleterious effect of eliminating multiple recombination sub-pathways. Since rad51Δ
moderately suppresses the nocodazole sensitivity of smc6-56
cells (), the accumulation of recombination intermediates appears to be more deleterious than the lack of recombination.
smc6-56 and rad51Δ cells are defective in centromeric LacO array separation.
Regulation of recombination by Smc6 is not restricted to centromeric regions. Increased levels of recombination intermediates and Rad52 foci were also detected in smc6-56
cells at non-centromeric regions ( and ). Such a general role fits with the presence of the Smc5–Smc6 complex at many chromosomal arm regions 
. Although probing the physiological importance of such a role at non-centromeric regions is not the focus here, we speculate that other replication blockage sites likely require proper regulation of recombination by this complex. The effects on recombination by Smc6 during growth described here are reminiscent of those under replication stress caused by exogenous DNA damage 
. This suggests that preventing the accumulation of toxic recombination structures is a crucial function of the Smc5–Smc6 complex both during growth and under DNA damage conditions. We therefore propose that the Smc5–Smc6 complex responds in a similar manner to multiple situations that create additional burden on the replication machinery, whether drug-induced or intrinsic to the nature of the locus. Thus, its function is not restricted to exogenous causes of replicative stress, but is likely to be an important component of the replication program. In agreement with this notion, the Smc5–Smc6 complex prevents sequence loss at telomeres and break-induced replication during growth, both of which can result from a failure to proper regulate recombination during replication 
. As rad51Δ
reduced the nocodazole sensitivity of smc6-56
cells, but differently affected recombination intermediate levels on 2D gel, it appears that the Smc5–Smc6 complex can affect more than one recombination steps as previously proposed.
This work also suggests that the Smc5–Smc6 complex affects kinetochore protein function. Our survey of 64 kinetochore and spindle proteins revealed ten SUMO substrates, four of which were not previously known ( and ). Only Ndc10 and Bir1 sumoylation was decreased by mms21-11
, an allele lacking the SUMO E3 ligase domain of Mms21 ( and data not shown). In addition, mms21-11
reduced the spindle localization of Ndc10 (), consistent with the finding that non-sumoylatable ndc10
eliminates this localization 
. The less penetrating defect of mms21-11
is likely due to its partial effect on Ndc10 sumoylation. As the Siz SUMO ligases also contribute to Ndc10 and Bir1 sumoylation, Mms21 may collaborate with them, though an indirect effect can not be excluded 
. We do not expect that mms21-11
and non-sumoylatable ndc10
exhibit the same set of defects, because mms21-11
affects sumoylation of other proteins 
. Short and mis-oriented spindles seen in mms21-11
anaphase cells () were not observed in non-sumoylatable ndc10
, and are likely due to a combined defect in the sumoylation of multiple Mms21 substrates. As smc6-56
exhibited similar defects as mms21-11
in spindle morphology and Ndc10 and Bir1 sumoylation ( and ), the smc6-56
allele likely impairs the sumoylation function of the Smc5–Smc6 complex.
In summary, our results suggest that the Smc5–Smc6 complex affects both recombination and kinetochore protein function during growth. While this work does not delineate the molecular connections between these two effects, results here provide bases for further study of their interplay. As mutants of the Smc5–Smc6 complex in fission yeast also exhibit chromosomal loss and an increased sensitivity to a microtubule destabilization drug, the roles of this complex at centromeric regions may be conserved 
. It is highly plausible that the integration of recombinational regulation and sumoylation via the Smc5–Smc6 complex contributes to the maintenance of other genomic regions during normal growth and under genotoxic stress. Future work will be needed to elucidate the interplay as well as the biological influence of these dual functions in genome maintenance.