Mitotic spindle length is a crucial determinant for accurate chromosome segregation. Short spindles facilitate in establishing bipolar connections of sister kinetochores while longer spindles inhibit this process . In this work we have convincingly shown that cohesion mutant chl1, when challenged with 0.2 M HU, developed significantly longer spindles than the wild-type cells under similar conditions. Since Chl1p does not have an S-phase checkpoint role nor any kinetochore related defect, we can conclude that decreased cohesion between sister chromatids in chl1 cells offers lesser resistance to pulling forces on sister kinetochores by spindle microtubules. This alters the balance of forces on the mitotic spindle leading to its extension. We have also found that the chl1 null mutant is defective in the retention of Scc1p at centromeres and that sister centromeres lose cohesion during both S- and G2 phases of the cell cycle. Therefore, apart from establishing it, Chl1p is also required to maintain cohesion at centromeres after S-phase in these cells.
Reduced association of the cohesin complex with chromatin could either be due to inefficient loading in the G1 phase, or defective cohesion establishment during S-phase, or due to both these defects. Petronczki and co-workers  have shown that, in the absence of Chl1p in G1, the loss in SCC was much lesser than when the protein was absent in S-phase. Thus, the authors document a major requirement of Chl1p in S-phase for SCC establishment, although their experiment did not map SCC loss specifically to S- and/or G2 phase(s). It is, however, entirely possible that Chl1p is required in G1 as well to help in the efficient loading of the cohesin complex. In such a case, reduction in cohesin association with chromosomes in the absence of Chl1p could be modest. Therefore, enough cohesin could still get loaded to prevent significant cohesion loss in S- and G2 phases, provided Chl1p is expressed in these phases. In the second scenario, cohesin loading could be normal in the G1 phase. However, defective establishment of cohesion without Chl1p in S-phase could lead to unstable association of the cohesin complex with sister chromatids. This could result in the dissociation of cohesin from chromosomes during S- and/or G2 phases of the cell cycle. A combination of both these defects (defective loading and establishment) would show reduced chromatin association in all the three phases (G1, S and G2) of the cell cycle of chl1 cells. Experiments are in progress to differentiate between these possibilities by analyzing the cell cycle-dependent association of the cohesin complex with chromosomes, in the presence or absence of Chl1p. Since the chl1 mutant does not suffer from any detectable loss in cell viability and grows like the wild-type under normal conditions of growth [33, 37], it can be concluded that retention of as little as one-fourth cohesion at centromeres is sufficient to promote bi-orientation of chromosomes and preserve cell viability under normal conditions. We did, however, observe about 50% killing in chl1 cells after 3.5 hours of HU treatment. The loss in viability could, in part, be due to the inability of mutant cells to repair DNA breaks induced by HU in the absence of Chl1p. It has been shown that if SCC is compromised, there can be defects in the bi-orientation of sister kinetochores due to structural considerations and possible dislodging of the chromosome from the spindle . A greater fraction of chl1 cells had non-localized (Y+G, G+G and G) kinetochores as compared to the wild-type cells after HU treatment (Table 2). It is possible that SCC-related defects in this mutant gain prominence under prolonged arrest in S-phase. Thus, non-localized kinetochores in mutant cells could reflect precociously separated mono-oriented sister kinetochores (Y+G) and kinetochores dislodged from the spindle (G+G and G) due to bi-orientation defects that manifest when cells stay arrested for long periods of time in S-phase. Another cohesion mutant, ctf4, behaved similarly to chl1 in that its cells elongated their spindles relative to the wild-type when arrested in S-phase by HU. The role of SCC in spindle length maintenance in S-phase arrested cells was further confirmed by a temperature-sensitive mutant scc1-73, having a defective cohesin subunit, displayed extensive spindle elongation at both 32°C and 35°C, temperatures at which it should be respectively partially and completely defective in the maintenance of cohesin at chromosomes.
Loss of pericentromeric cohesion also led to considerable increase in spindle lengths and inter-kinetochore distances after three hours of S-phase arrest by HU. Although both chl1 and pericentromeric mutants elongated their spindles upon HU treatment, chl1 cells were more sensitive than the wild-type for growth towards this drug. This could be due to the additional DNA repair function of Chl1p, which may be separable from its SCC function. Indeed, observations of Ogiwara and co-workers  have shown that the repair of MMS-induced DNA damage by Chl1p does not require SCC.
It has been reported earlier that scc1/mcd1 mutant, having an intact S-phase checkpoint, does not elongate spindles at its non-permissive temperature when treated with HU for 2.5 hours [45, 65]. In these studies, cells were taken to have extended spindles only when the spindle lengths were above 3 μm. Our data agrees with these results in that less than 20% of cohesion mutants had their cells with spindles longer than 3 μm under HU treatment (For example, Figures 2B, 5B, 6A, B and 7). Nevertheless, within this ≤ 3 μm window, there was a significant increase (p ≤ 0.001) in spindle lengths of cohesion mutants relative to the wild-type during S-phase arrest. Surana and co-workers  have shown that in the absence of an active S-phase checkpoint pathway in the mec1 mutant, microtubule associated proteins Cin8 and Stu2, implicated in spindle elongation, accumulate to high levels during S-phase arrest. Increase in the levels of these two proteins leads to unrestrained spindle elongation with precocious and unequal segregation of chromosomes in mec1 cells. In our experiments, the S-phase checkpoint pathway was active. Consequently, Cin8 and Stu2 would be present at their normal low levels and not participate in undue spindle elongation. The increase in spindle lengths due to defective cohesion in our experiments was, therefore, less extensive as compared to that observed in mec1 cells , but nevertheless significant.
Thus, in the present study we have shown that in the absence of Chl1p, the maintenance of SCC is affected both in S- and G2 phases. Further, the chl1 mutation neither affects the functioning of the S-phase replication checkpoint pathway, nor does it lead to any kinetochore related defect. Still, this mutation causes spindle elongation when cells are treated with HU. Our observations for the first time clearly implicate the role of SCC and of pericentromeric cohesion in spindle length regulation and undue stretching of sister centromeres in S-phase arrested cells. Since Chl1p has human homologues, like the BRCA1-binding protein BACH1 implicated in tumor suppression, the characterization of Chl1p in yeast should help to shed light on the functions of its human homologues.