Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Cell. Author manuscript; available in PMC 2010 April 3.
Published in final edited form as:
PMCID: PMC2673135

Separase is recruited to mitotic chromosomes to dissolve sister chromatid cohesion in a DNA-dependent manner


Sister chromatid separation is triggered by the separase-catalyzed cleavage of cohesin. This process is temporally controlled by cell-cycle-dependent factors, but its biochemical mechanism and spatial regulation remain poorly understood. We report that cohesin cleavage by human separase requires DNA in a sequence-nonspecific manner. Separase binds to DNA in vitro, but its proteolytic activity, measured by its auto-cleavage, is not stimulated by DNA. Instead, biochemical characterizations suggest that DNA mediates cohesin cleavage by bridging the interaction between separase and cohesin. In human cells, a fraction of separase localizes to the mitotic chromosome. The importance of the chromosomal DNA in cohesin cleavage is further demonstrated by the observation that the cleavage of the chromosome-associated cohesins is sensitive to nuclease treatment. Our observations explain why chromosome-associated cohesins are specifically cleaved by separase and the soluble cohesins are left intact in anaphase.

Keywords: mitosis, chromosome dynamics, sister chromatid cohesion, separase, cohesin


Stable cohesion between sister chromatids before anaphase and their timely separation during anaphase are critical for chromosome inheritance. Sister chromatid cohesion is mediated by the cohesin complex, consisting of four core subunits: an α-kleisin subunit (SCC1/MCD1/RAD21), SCC3 (known as SA1 or SA2 in vertebrates), SMC1 and SMC3, (Huang et al., 2005a; Nasmyth, 2005). Although the cohesin complex is conserved from yeast to humans, the regulation of sister chromatid cohesion is more complex in metazoans. In vertebrates, sister chromatid cohesion is released in two steps via two distinct mechanisms. The first step occurs in prophase, which involves phosphorylation of SA1/2 (Hauf et al., 2005) and dissociates most of cohesins from chromosome arms but not the heterochromatin or centromeric regions. The second step occurs in anaphase, when separase cleaves SCC1 and initiates the final separation of sister chromatids (Hauf et al., 2001; Uhlmann et al., 2000).

In vertebrates, separase cleaves SCC1 exclusively in anaphase. This cleavage does not occur during the rest of the cell cycle because separase is inhibited by two independent and sometimes redundant mechanisms: binding by securin (Zou et al., 1999) and phosphorylation of serine 1126 (S1126) of separase (Stemmann et al., 2001). Moreover, the bulk of separase is excluded from the nucleus in interphase, avoiding the possibility of direct contact with cohesin (Sun et al., 2006). However, additional regulation is strongly suggested by the lack of expected phenotype in cells and animals missing securin and/or S1126 phosphorylation (Huang et al., 2005b; Jallepalli et al., 2001; Mei et al., 2001; Pfleghaar et al., 2005; Wang et al., 2001).

In addition to temporal regulation, cohesin cleavage is regulated spatially after separase activation. In budding yeast, cohesins are mostly chromosome bound and near completely cleaved in anaphase. Nonetheless, chromosome-associated cohesins are a preferred substrate for separase. Polo-like kinase (Plk1) phosphorylates the chromosome-associated cohesin on the SCC1 subunit, which stimulates cohesin cleavage two-three fold over that of the unphosphorylated soluble cohesins (Alexandru et al., 2001; Hornig and Uhlmann, 2004). In vertebrates, only a small fraction of cohesin, which is thought to be the chromosome-associated pool, is cleaved in anaphase (Waizenegger et al., 2000). The stimulation of cohesin cleavage by Plk1 is moderate (Hauf et al., 2005). Furthermore, it is unclear whether the chromosome-associated cohesins are phosphorylated by Plk1. In fact, phosphorylation of SA2 by Plk1 causes the prophase removal of cohesin from chromosomes (Hauf et al., 2005; Waizenegger et al., 2000). The remaining chromosome-associated cohesins are protected from Plk1 by Shugoshin and PP2A (Kitajima et al., 2005; Kitajima et al., 2006; McGuinness et al., 2005; Tang et al., 2006). Therefore, the mechanism that limits cohesin cleavage to the chromosome-associated pool in vertebrate cells remains unclear.

In order to better understand the spatiotemporal regulation of cohesin cleavage by separase in vertebrates, we searched for a new biochemical activity that would regulate cohesin cleavage by separase in the absence of securin, the phosphorylation inhibitions and the nuclear exclusion. This study led to the discovery that cohesin cleavage is dependent on the presence of DNA. The protease activity of separase per se does not exhibit this dependency, as evidenced by the ability of separase to cleave itself in the absence of DNA. Instead, our results indicate that DNA facilitates the cleavage reaction by bridging separase and cohesin. We propose that chromosomal DNA functions as this molecular bridge in vivo. In support of this model, we detected separase on mitotic chromosomes by immunofluorescent microscopy and cellular fractionation. We also demonstrated that chromosome-associated mitotic cohesin is no longer cleaved when chromosomal DNA is removed by nuclease digestion. Our findings help to explain why only chromosome-associated cohesins are cleaved by separase in anaphase.


Polynucleotide is required for SCC1 cleavage by separase in vitro

The cleavage of cohesin SCC1 by separase was previously reconstituted in vitro using partially purified enzyme and substrate (Fan et al., 2006; Stemmann et al., 2001). In order to identify additional regulators of cohesin cleavage, we tested whether the crude extract prepared from the securin−/− HCT116 cells (Jallepalli et al., 2001) could inhibit cohesin cleavage in this assay. The separase S1126A mutant (separase-PM2/4), which is resistant to phosphorylation inhibition, was used as the enzyme. In this system, all known mechanisms of separase regulation were absent. Remarkably, as little as 2 μl of this extract completely blocked SCC1 cleavage (Figure 1A, lane 4). This result suggested that SCC1 cleavage by separase was inhibited by an unknown activity. We performed biochemical purifications to identify this activity (Figure 1B). The activity was eluted from a gel filtration column with an apparent molecular weight of about 15 KDa, which overlapped with RNase A (Figure 1C). Furthermore, the activity was resistant to heat (Figure 1D) and sensitive to RNasin (Figure 1E). This prompted us to investigate whether it was actually a ribonuclease. Indeed, both RNase A and RNase T1 inhibited SCC1 cleavage (Figure S1). Finally, the inhibition was not caused by the hydrolyzed ribonucleotides because NTPs up to 100 μM did not affect the cleavage reaction (data not shown). Therefore, we concluded that a ribonuclease activity was responsible for the inhibition of cohesin cleavage.

Figure 1
The in vitro cohesin cleavage assay is inhibited by ribonuclease activity

The fact that the reaction was sensitive to various RNases suggested that an RNA component might be required for SCC1 cleavage in our assay. Further analyses revealed that the separase preparation was contaminated with a noticeable amount of RNA (Figure S2). These RNA molecules originated from the Xenopus egg extract, which was used to degrade securin, because they were not detected prior to the extract treatment. The Xenopus egg extract is known to contain a large amount of maternal mRNA. We directly analyzed whether RNA or other polynucleotide facilitates SCC1 cleavage. We pretreated the recombinant separase and the cohesin complex with RNase A to destroy the contaminating RNA and supplemented the reaction with various polynucleotides. No SCC1 cleavage was detected when the RNA-free separase and cohesin were incubated together (Figure 2A). Addition of either RNA or a generic plasmid DNA restored the cleavage. We performed cohesin cleavage in the presence of approximately 2, 20 and 200 nM separase to quantitatively determine the effect of DNA and RNA on SCC1 cleavage. In the presence of RNA, 20 nM separase (Figure 2B, lane 3) cleaved more SCC1 than 200 nM separase did in the absence of RNA (lane 5). In the presence of DNA, 2 nM separase (Figure 2C, lane 7) cleaved more SCC1 than 200 nM separase did in the absence of DNA (lane 8). Remarkably, even at the highest concentration, separase failed to produce any significant amount of cleaved SCC1 fragments in the absence of DNA or RNA. Taken together, these results indicate that the cleavage of SCC1 by separase requires DNA or RNA.

Figure 2
Polynucleotide is required for the cleavage of cohesin by separase

Separase is a DNA-binding protein

The contamination of the separase preparations by RNA suggested that separase might physically bind polynucleotide. We first attempted an electrophoretic mobility shift assay and found that a substantial amount of DNA (300 bps) was trapped in the wells in an apparent separase-dependent manner (data not shown). However, because separase failed to migrate into regular or blue native gels, we could not perform a super shift experiment to confirm that the mobility shift was specifically caused by separase. Subsequently, we adopted a pull-down assay using DNA-cellulose beads. Separase associated with the DNA beads but not the beads pretreated with DNase, indicating that separase is a DNA-binding protein (Figure 3A). The interaction was detected in buffers containing up to 100 mM NaCl (Figure S3A). Interestingly, the cohesin cleavage reaction exhibited a similar, if not identical, response to the concentrations of NaCl (Figure S3C and S3D), suggesting that the DNA binding activity of separase may be important for its ability to cleave cohesin. Furthermore, the separase-DNA binding was not regulated by securin or S1126 phosphorylation (Figure S4).

Figure 3
Separase and cohesin bind to DNA in a sequence-nonspecific manner

We next estimated the affinity of the separase-DNA interaction in the cohesin cleavage buffer. We used the securin-bound separase because they are easily purified. Two independent experiments revealed that the Kd of the separase-DNA interaction ranges from 21 to 24 nM (Figure 3B and 3C). In both cases, the Kd was well below the physiological concentration of separase, which we estimated to be higher than 180 and 250 nM in HeLa and 293T cells, respectively (Figure 3B). These results suggest that separase has the capacity to interact with DNA in vivo.

Because both poly (A) RNA and a generic plasmid were able to facilitate cohesin cleavage, we suspected that separase associated with DNA in a sequence-nonspecific manner. Indeed, heparin, a negatively charged polymer that interacts with many established sequence-nonspecific polynucleotide-binding proteins, competitively inhibited the binding of separase to DNA beads (Figure 3D, lane 2). Similarly, DNA fragments from 25 to 4000 bps competitively reduced the amount of separase associated with the DNA beads and concomitantly increased the amount of separase left in the supernatant (lanes 3–7). Longer fragments competed more efficiently than the shorter ones, suggesting that separase favors longer DNA.

Using this competition assay, we also characterized a physical binding between the cohesin complex and DNA. As shown in Figure 3E, the purified cohesin complex associated with the DNA beads but not the DNase-treated beads. This binding was also competitively inhibited by free DNA and heparin. Direct binding of cohesin to linear DNA through the SMC1/3 heterodimer has been reported previously (Akhmedov et al., 1999; Hirano and Hirano, 2006). Our experiments confirmed these observations and showed that DNA as short as 25 bps competitively inhibited the binding.

Auto-cleavage of separase does not require DNA

One possible explanation for the stimulatory effect of DNA is that the binding of DNA to separase may directly stimulate its proteolytic activity. Because vertebrate separase also cleaves itself upon activation in mitosis (Chestukhin et al., 2003; Waizenegger et al., 2002; Zou et al., 2002), we used this auto-cleavage as a reporter for the enzymatic activity of separase. We transiently expressed myc6-tagged separase in 293T cells and the separase-securin complex was subsequently purified on anti-myc beads. Separase was mostly in the full-length form because of the inhibition by the associated securin (Figure 4, lane 1). To remove the associated securin, the beads were incubated with a conventional Xenopus mitotic extract supplemented with a low concentration of nondegradable cyclin B-Δ90 (low Δ90). This extract degraded securin without introducing the inhibitory phosphorylation. Auto-cleavage of separase was evident from the decrease of the full-length separase and the increase of the cleaved form (lane 3, upper panel). In parallel, we performed this set of experiments in an RNA-free extract, which was generated by pre-treating the same Xenopus extract with RNase A. The RNase treatment was not detrimental to securin degradation, as indicated by the disappearance of securin (middle panel, lanes 2 and 4). The auto-cleavage of separase in the RNA-free extract was indistinguishable from that in the untreated extract, indicating that the proteolytic activity of separase is not dependent on RNA or presumably DNA. To rule out the possibility that the auto-cleavage was due to the presence of any residual RNA, we also measured SCC1 cleavage by the same separase. The cohesin cleavage reaction was less efficient when separase was supplied on the beads. Nonetheless, the separase prepared from untreated extract cleaved a noticeable amount of SCC1 (lane 3), while the separase prepared from the RNase-treated extract was unable to cleave SCC1 (lane 2) unless DNA was supplemented (lane 4). Therefore, DNA or RNA does not stimulate the proteolytic activity of separase per se.

Figure 4
The proteolytic activity of separase is independent of polynucleotide

Using this assay, we also reexamined whether securin and/or S1126 phosphorylation regulate the DNA-mediated cohesin cleavage. To remove securin but retain S1126 phosphorylation, we incubated the separase beads in a Xenopus egg extract, supplemented with a high-concentration of cyclin B Δ90 (high Δ90). As shown in Figure 4, lower panel, the high Δ90 extract inhibited the wild type separase in the presence of RNA (in the absence of RNase A, lane 7) or DNA (lane 8), but not the separase-PM2/4 mutant (lanes 11 and 12). Furthermore, pre-incubation with recombinant securin also inhibited the cleavage of cohesin in the presence of DNA (lanes 5, 9, and 13). Therefore, the DNA-mediated cohesin cleavage by separase is regulated by the two previously characterized in vivo inhibitory mechanisms.

DNA stimulates cohesin cleavage in a length-dependent manner

To shed light on the mechanism, we further characterized the stimulatory effect of DNA fragments with different lengths. DNA ranging from 25 to 4000 bps competed the binding of separase and cohesin to DNA-beads (Figure 3D and E), indicating that they all bind to separase and cohesin. We therefore investigated whether these fragments were also able to mediate cohesin cleavage. As shown in Figure 5A, DNA fragments shorter than about 300-bp gradually became less efficient. The 104-bp fragment still partially facilitated SCC1 cleavage. In the presence of the 65-bp fragment, only a trace amount of cleaved SCC1 fragment was detected (lane 8). No cleavage of cohesin was detected in the presence of the 25-bp fragment (lane 9), even at much higher concentration (see below). These observations indicate that the stimulatory effect of DNA is dependent on its length. The fact that the 25-bp fragment bound to separase or cohesin but failed to facilitate cohesin cleavage indicates that DNA binding to separase or cohesin alone is not sufficient to stimulate cohesin cleavage.

Figure 5
In vitro cohesin cleavage requires DNA of a minimum length and at an optimal concentration

Excess DNA negatively affects SCC1 cleavage

We also investigated the effect of DNA at various concentrations on cohesin cleavage. To this end, DNA (600 bps) was added in the standard cohesin cleavage assay at the final concentrations ranging from 0 to 800 ng/μl. As shown in Figure 5B and 5C, more cohesin was cleaved when the DNA concentration increased to 100 ng/μl. However, further increases resulted in a gradual decrease of SCC1 cleavage. Similar observations were also made with poly (A) and heparin (Figure S6). Finally, we compared the effects of DNA concentration using DNA fragments of 25, 600, and 4000 bps. The cleavage reaction responded similarly, if not identically, to both 600-bp and 4000-bp DNA (Figure 5D). Remarkably, in both cases, the optimal concentration was 100 ng/μl. The 25-bp fragment failed to mediate cohesin cleavage at any concentrations (Figure 5D), confirming our previous conclusion that this short DNA, although binds to separase and cohesin, is ineffective at mediating cohesin cleavage. Because the responses of cohesin cleavage to increasing concentrations of DNA did not exhibit a typical saturation curve, it is unlikely that binding of separase or cohesin to DNA alone is responsible for the stimulating effect.

A fraction of separase localizes to mitotic chromosomes

In vertebrates, only the small fraction of cohesin is cleaved by separase in anaphase. Based on the experiments above, we suspected that the underlying mechanism of this selective proteolysis is that chromosomal DNA is required for cohesin cleavage in vivo. Such a model predicts that separase localizes to chromosomes, at least briefly during the onset of anaphase. Consistent with this prediction, previous reports demonstrated a localization of separase to metaphase and/or anaphase chromosomes in budding yeast (Hornig and Uhlmann, 2004) and C. elegans (Bembenek et al., 2007). However, stable chromosome association of separase was not reported in other studies (Chestukhin et al., 2003; Sun et al., 2006). These negative results were seemingly accepted without further scrutiny because, as an enzyme, separase was expected to interact with its chromosome substrates transiently.

The finding that separase is a DNA binding protein in vitro prompted us to study this issue more carefully. Because of the lack of an appropriate antibody to directly analyze the localization of endogenous separase in immunofluorescent microscopy, we used a previously characterized 293T cell line that stably expresses a V5-tagged separase (Chestukhin et al., 2003). In the presence of 1.2 μM ponasterone A, separase-V5 was expressed at an overall level comparable with endogenous separase (Figure S7A). However, immunofluorescent staining revealed that individual cells exhibited a wide range of expression levels and separase-V5 was detected in about 60% of the cells (Figure S7B). When mitotic chromosome spreads were prepared from these cells, separase-V5 was detected along the entire length of the chromosomes in about 10% of the spreads (Figure 6A). Notably, separase was not detected on any interphase chromosomes, presumably due to its exclusion from the interphase nucleus (Sun et al., 2006). Under the same conditions, no signal was detected on the chromosomes of uninduced cells. Chromosome-associated separase was detected on tightly paired long prophase chromosomes, paired metaphase chromosomes, and unpaired anaphase chromosomes, suggesting that the localization occurs as early as in prophase or prometaphase and is maintained even after sister chromatid separation. This is in agreement with the finding that the separase-DNA interaction is not regulated by securin or S1126 phosphorylation (Figure S4). The reason for detecting the staining in only 10% of cells might lie in the heterogeneous expression of separase. We suspected that chromosomal separase was detected by this method only in the cells that overexpressed separase. Indeed, when we increased the expression of separase-V5 by adding 5 μM ponasterone A, we detected strong separase staining in up to about 75% of the chromosome spreads. We also performed similar analysis using a 293T cells stably overexpressing myc-separase. Again, about 90% of the mitotic chromosome spreads exhibited similar separase staining (Figure S7C). Therefore we concluded that overexpressed separase localizes to mitotic chromosomes along their entire lengths.

Figure 6
Separase localizes to chromosomes in mitosis

To confirm the results from immunofluorescent microscopy and to analyze the localization of endogenous separase, we also performed cellular fractionation in HeLa cells (Mendez and Stillman, 2000). We repeatedly detected more separase in the chromosome fraction (P3) from nocodazole-arrested mitotic cells than that from thymidine-arrested interphase cells (Figure 6B). We quantified the signals and normalized the amount of chromatin-bound separase to that of chromatin-bound Topo IIα. A significant 30-fold increase was detected in mitosis over interphase cells (Figure 6D). Taken together, these results indicate that a fraction of separase is recruited to chromosomes in mitosis.

Chromosomal DNA is required for the cleavage of chromosome-associated mitotic cohesin

The dependence of cohesin cleavage on DNA and the localization of separase to mitotic chromosomes suggest that chromosomal DNA may be required to mediate the cleavage of mitotic cohesin in vivo. We first investigated whether chromosome-associated mitotic cohesins are a better substrate for separase than soluble mitotic cohesins when both forms are mixed together. The soluble cohesins contained a myc6-tagged SCC1 so that they can be differentiated from the untagged chromosome-associated cohesins. As shown in Figure 7A, chromosome-associated cohesin was efficiently cleaved by separase in the absence of any additional DNA. On the other hand, soluble cohesin was not cleaved unless DNA was added. The background cleavage of soluble cohesin in the absence of added DNA was mostly like due to a minute DNA contamination in this sample. When mixed together, chromosome-associated cohesins were mostly cleaved whereas soluble cohesins remained largely intact. There was a small but reproducible increase of the cleaved products in the reaction containing mixed forms of cohesins, compared with the reaction containing chromosome-associated cohesin alone. This increase might be the result of the binding of a small amount of the soluble cohesins to chromosomal DNA during the reaction. These results indicate that chromosome-associated cohesin is indeed preferentially targeted by separase.

Figure 7
Chromosomal DNA is required for the cleavage of chromosome-associated cohesins

Finally, we performed the cohesin cleavage assay after removal of chromosomal DNA by a micrococcal nuclease digestion. In the absence of chromosomal DNA, the released chromosome-associated mitotic cohesins were not cleaved by separase (Figure 7B). Notably, adding back DNA after inactivating the nuclease rescued the cohesin cleavage reaction. Taken together, these results indicate that chromosomal DNA plays an indispensable role in mediating cohesin cleavage by separase in human cells.


The final separation of sister chromatids is triggered by the separase-catalyzed cleavage of cohesin. In addition, cohesin is also implicated in transcriptional regulation in S. pombe (Gullerova and Proudfoot, 2008) and metazoans (Parelho et al., 2008; Rollins et al., 1999; Rubio et al., 2008; Stedman et al., 2008; Wendt et al., 2008). This has led to the proposal that a critical pool of cohesin must be left intact so that its function can be quickly restored in the following G1 (Wendt et al., 2008). Coincidentally, only a small fraction of cohesin is cleaved in the anaphase of both S. pombe (Fousteri and Lehmann, 2000) and human cells (Waizenegger et al., 2000). In this study, we report that, in human cells, chromosomal DNA is required for cohesin cleavage and this requirement limits the proteolysis to the chromosome-associated pool (Figure 7C).

Early studies in various systems had suggested separase as the sole factor required for cohesin cleavage. Our findings indicate that the cleavage of cohesin by separase is more complex than previously appreciated and it highlights the importance of biochemical reconstitution in elucidating complex biochemical reactions. The requirement of a DNA rather than a protein factor may be the reason why it was not identified in previous studies. The added complexity of this reaction opens the possibility of additional regulatory mechanisms of cohesin cleavage and sister chromatid separation.

It is unlikely that separase might be a more active enzyme when associated with DNA, because the auto-cleavage of separase occurs in a DNA-independent manner (Figure 4). Instead, we propose that DNA function as a molecular bridge between separase and cohesin. In this bridging model, separase and cohesin must bind to the same DNA molecule to produce a cleavage event. This model is supported by the following observations. First, the 25-bp DNA bound to separase and cohesin (Figure 3D), but failed to stimulate cohesin cleavage, even at high concentrations (Figure 5D). If the cleavage were determined simply by the binding of separase or cohesin to DNA, we would have detected some cleavage. This observation, on the other hand, is compatible with the bridging model, which requires DNA of adequate length to accommodate both separase and cohesin. Second, if affinity binding were sufficient to stimulate cohesin cleavage, excess DNA would have no detrimental effect on the reaction. The response of cohesin cleavage to increasing concentrations of DNA would resemble a typical saturation curve. However, amounts of cleaved cohesin decreased when DNA concentrations increased above the optimum (Figure 5B and S6). This negative effect can be explained by the bridging model because excess DNA increases the likelihood of separase and cohesin binding to distinct DNA molecules and thus decreases the efficiency of cohesin cleavage.

The bridging model suggests that both separase and cohesin interact with DNA. Cohesin is an established chromosome-binding protein. It binds to chromosomes via two mechanisms. The hinge domain of the SMC1/3 heterodimer can physically interact with chromosomal DNA in vitro (Hirano and Hirano, 2006). This mode of binding is proposed to be further converted to a much more stable concatenation with chromosomes (Haering et al., 2008). The bridging model does not specify the mode of the interaction. When soluble cohesin and DNA are used in the in vitro cohesin cleavage reaction, the cohesin-DNA interaction is most likely mediated by the hinge domain. When chromosome-associated cohesins are used, the cohesin-chromosome interaction is most likely mediated by the concatenation. In both assays, the reaction occurs in a DNA-dependent manner.

We found that separase binds to DNA with a high affinity in a sequence-nonspecific manner. This disassociation constant is well below the physiological concentration of separase in HeLa and 293T cells. Furthermore, separase binds DNA fragments as short as 25 bps, as demonstrated by the competition assay. However, it binds longer DNA with higher affinity. It is possible that a cooperative binding may be involved, through which multiple separase proteins may bind to the same DNA molecule with higher affinity. Alternatively, separase may have multiple DNA binding domains. It is unclear which domain(s) on separase is (are) responsible for binding to DNA and chromosomes. We attempted to identify the DNA-binding domains on separase using a structural and mutational analysis, only to find out that this approach is impractical because of the large size of separase (about 2200 amino acid residues) and the lack of informative deletions that retain their DNA binding activity. Future structural biological studies will shed light on the DNA binding domain(s) of separase. Although we could not directly demonstrate the role of the DNA-binding activity of separase in cohesin cleavage, indirect evidence does suggest that this activity may be required. Both the cohesin cleavage and separase-DNA binding assays exhibit a similar sensitivity to the 150 mM or above NaCl in the buffers (Figure S3A and S3C). The sensitivity was the same even when we used chromosome-associated cohesins (Figure S3D), which concatenate with chromosomes and are expected to remain associated with chromosomes in buffers containing up to 300 mM cations (Ivanov and Nasmyth, 2005). It is possible that the salt sensitivity of the separase-DNA interaction leads to the salt sensitivity of cohesin cleavage.

Chromosome-associated separase was reported in budding yeast (Hornig and Uhlmann, 2004) and C. elegans (Bembenek et al., 2007). In human cells, we also reproducibly detected a fraction of separase on mitotic chromosomes (Figure 6, S7). Although the separase-V5 and myc-separase reporters were overexpressed, the results using these reporters were consistent with cellular fractionation, which analyzed endogenous separase. Interestingly, instead of being focused on the centromeric region where mitotic cohesins reside, separase localizes to the entire length of mitotic chromosomes. This pattern may be caused by the overexpression of the tagged separase. Alternatively, separase binds to DNA in a sequence-nonspecific manner. The indiscriminative binding of separase to the entire chromosomes may be necessary for the cleavage of strayed cohesins erroneously left on the chromosome arms. Indeed, the cohesins remaining on chromosome arms have to be resolved in a separase-dependent manner (Nakajima et al., 2007). Interestingly, although RNA is able to replace DNA to stimulate cohesin cleavage in vitro, it may not do so in vivo. It is possible that RNA binding proteins prevent the binding of cellular RNA to cohesin and/or separase. The unprotected RNA may be degraded by the cellular RNase activity, which we identified as the inhibitory activity to cohesin cleavage.

It was reported that phosphorylation of SCC1 by Plk1 stimulates cohesin cleavage by up to two-to-three-fold (Alexandru et al., 2001; Hauf et al., 2005; Hornig and Uhlmann, 2004). This raises the question whether the stimulation observed here is a biochemical artifact, in which the binding of cohesin to DNA mimics these phosphorylations. We consider this unlikely, based on the following three reasons. First, it is unclear whether the chromosome-associated mitotic cohesins are phosphorylated in vertebrate cells. In fact, chromosome-associated cohesins are protected from Plk1 by Shugoshin and PP2A. Second, even if these cohesins were phosphorylated at SCC1, they are still cleaved in a chromosomal-DNA-dependent manner (Figure 7B), indicating that the stimulation by phosphorylation and the dependence on chromosomal DNA are two different regulations. Third, the stimulation by DNA, which is at least 100-fold (Figure 2D), is much more pronounced than the two-to-three-fold stimulation facilitated by the phosphorylation of SCC1. The observation that chromosomal DNA is critical for cohesin cleavage in human cells is different from what was reported in budding yeast, where chromatin-associated cohesins were cleaved even after nuclease treatment (Hornig and Uhlmann, 2004). It is possible that yeast uses a different strategy, which involves Plk1, to facilitate the cleavage of chromosome-associated cohesins. In human cells, the requirement of chromosomal DNA is the major mechanism that restricts proteolysis to the chromosome-associate cohesins.


Cell culture, cell cycle synchronization, transfection, and antibodies

HeLa and 293T cells were grown in DMEM, whereas securin−/− HCT116 cells were cultured in McCoy’s 5A. Both media were supplemented with 10% FBS. HeLa cells were synchronized at G1/S and c-metaphase by double thymidine block and thymidine-nocodazole arrest, respectively. Transfection of 293T cells was performed according to a calcium-phosphate protocol. Antibodies to the C-terminus of separase (NB 100-439, Novus Biologicals, CO), V5 (46-0705, Invitrogen, CA), phospho-histone-H3-Ser10 (sc-8656-R, Santa Cruz, CA), Topo-IIα (sc-13058, Santa Cruz, CA), GAPDH (sc-25778, Santa Cruz, CA), Myc (sc-40, Santa Cruz, CA), SMC1 (A300-055A Bethyl, TX) and SA2 (ab4463, Abcam, MA) were commercially available. Antibodies to securin and separase were described previously (Zou et al., 1999). A polyclonal antibody to the C-terminus of SCC1 (CEPYSDIIATPGPRFH) was custom produced by Genemed Synthesis, CA and affinity-purified.

In vitro cohesin cleavage assay

Separase was prepared as described (Stemmann et al., 2001) and its concentration was estimated on a Coomassie-blue-stained SDS-PAGE gel. Unless noted otherwise, the cohesin complex was purified from nocodazole-arrested HeLa and/or 293T cells (Fan et al., 2006). Briefly, the cohesin complex was precipitated by 45% ammonium sulfate at 4°C, resuspended in the QA buffer (20 mM Tris (pH7.6) and 100 mM NaCl), and eluted from a HiTrap Q column at about 250 mM NaCl. After buffer change with SA buffer (20 mM HEPES (pH7.6) and 100 mM NaCl), the collected fractions were absorbed on a HiTrap SP column and eluted at 150 mM NaCl. Finally, the cohesin complex was moved into the cohesin cleavage buffer (30 mM HEPES/KOH pH7.7, 30% glycerol, 25 mM NaF, 25 mM KCl, 5 mM MgCl2, 1 mM EGTA) on a PD-10 column and its concentration was estimated on SDS-PAGE. The cleavage reaction was assembled in a 10 μl volume containing about 50 nM separase and 200 nM cohesin in the cohesin cleavage buffer. The mixture was incubated at 37°C for 30 minutes before the reaction was terminated with SDS-PAGE loading buffer. The integrity of separase and cohesin were determined by immunoblot.

To prepare the extract from the securin−/− HCT116 cells, asynchronous log phase cells were resuspended in the cohesin cleavage buffer. In the absence of any detergent, we resuspended the cells in the cohesin cleavage buffer and broke the cells by nitrogen cavitation in a 45 ml Parr Cell Disruption Bomb, pressurized to 1000 psi for 15 minutes at 4°C. The S100 was prepared and used directly in the cohesin cleavage assay.

DNA cellulose beads pull-down assay

Cellulose beads conjugated with calf thymus DNA (27-5581-02, Amersham) were digested with EcoR I and BamH I. This effectively reduced the average length of the conjugated DNA to about 2 kbps. Before being used, the beads were washed three times to remove any free DNA fragments. The control beads were made from the same DNA beads but pretreated with excess DNase. In a typical binding assay, 1 μl of DNA beads was used in a 10 μl binding mixture in the cohesin cleavage buffer. The concentrations of separase and cohesin were the same as in the cohesin cleavage reaction. The binding was performed at room temperature for 30 minutes and the beads were washed three times with the cohesin cleavage buffer before being analyzed by immunoblot.

Supplementary Material


We thank Lydia Chiu and Julie Covino for technical support, Hongtao Yu for helpful discussions, and Eric Olson for critical reading and comments of the manuscript. We also thank Hongtao Yu, David Pellman and Stephen Taylor for generously providing the myc6-SCC1, separase-V5, and myc-separase stable cell lines, respectively. H.Z. is the Kenneth G. and Elaine A. Langone Scholar supported by the Damon Runyon Cancer Foundation (DRS-#35-03). This work is also supported by a Research Scholar Grant from the American Cancer Society (RSG-04-171-01-CCG), a research grant from the Welch Foundation (I-1594), and a NIH R01 grant (R01GM081466-02) to H.Z..


Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.


  • Akhmedov AT, Gross B, Jessberger R. Mammalian SMC3 C-terminal and coiled-coil protein domains specifically bind palindromic DNA, do not block DNA ends, and prevent DNA bending. J Biol Chem. 1999;274:38216–38224. [PubMed]
  • Alexandru G, Uhlmann F, Mechtler K, Poupart MA, Nasmyth K. Phosphorylation of the cohesin subunit Scc1 by Polo/Cdc5 kinase regulates sister chromatid separation in yeast. Cell. 2001;105:459–472. [PubMed]
  • Bembenek JN, Richie CT, Squirrell JM, Campbell JM, Eliceiri KW, Poteryaev D, Spang A, Golden A, White JG. Cortical granule exocytosis in C. elegans is regulated by cell cycle components including separase. Development (Cambridge, England) 2007;134:3837–3848. [PubMed]
  • Chestukhin A, Pfeffer C, Milligan S, DeCaprio JA, Pellman D. Processing, localization, and requirement of human separase for normal anaphase progression. Proc Natl Acad Sci U S A. 2003;100:4574–4579. [PubMed]
  • Fan HY, Sun QY, Zou H. Regulation of Separase in Meiosis: Separase is Activated at the Metaphase I-II Transition in Xenopus Oocytes during Meiosis. Cell Cycle. 2006;5:198–204. [PubMed]
  • Fousteri MI, Lehmann AR. A novel SMC protein complex in Schizosaccharomyces pombe contains the Rad18 DNA repair protein. EMBO J. 2000;19:1691–1702. [PubMed]
  • Gullerova M, Proudfoot NJ. Cohesin complex promotes transcriptional termination between convergent genes in S. pombe. Cell. 2008;132:983–995. [PubMed]
  • Haering CH, Farcas AM, Arumugam P, Metson J, Nasmyth K. The cohesin ring concatenates sister DNA molecules. Nature. 2008;454:297–301. [PubMed]
  • Hauf S, Roitinger E, Koch B, Dittrich CM, Mechtler K, Peters JM. Dissociation of Cohesin from Chromosome Arms and Loss of Arm Cohesion during Early Mitosis Depends on Phosphorylation of SA2. PLoS Biol. 2005;3:1–14. [PMC free article] [PubMed]
  • Hauf S, Waizenegger IC, Peters JM. Cohesin cleavage by separase required for anaphase and cytokinesis in human cells. Science. 2001;293:1320–1323. [PubMed]
  • Hirano M, Hirano T. Opening Closed Arms: Long-Distance Activation of SMC ATPase by Hinge-DNA Interactions. Mol Cell. 2006;21:175–186. [PubMed]
  • Holland AJ, Taylor SS. Cyclin-B1-mediated inhibition of excess separase is required for timely chromosome disjunction. J Cell Sci. 2006;119:3325–3336. [PubMed]
  • Hornig NC, Uhlmann F. Preferential cleavage of chromatin-bound cohesin after targeted phosphorylation by Polo-like kinase. EMBO J. 2004;23:3144–3153. [PubMed]
  • Huang CE, Milutinovich M, Koshland D. Rings, bracelet or snaps: fashionable alternatives for Smc complexes. Philos Trans R Soc Lond B Biol Sci. 2005a;360:537–542. [PMC free article] [PubMed]
  • Huang X, Hatcher R, York JP, Zhang P. Securin and Separase Phosphorylation Act Redundantly to Maintain Sister Chromatid Cohesion in Mammalian Cells. Mol Biol Cell 2005b [PMC free article] [PubMed]
  • Ivanov D, Nasmyth K. A topological interaction between cohesin rings and a circular minichromosome. Cell. 2005;122:849–860. [PubMed]
  • Jallepalli PV, Waizenegger IC, Bunz F, Langer S, Speicher MR, Peters JM, Kinzler KW, Vogelstein B, Lengauer C. Securin is required for chromosomal stability in human cells. Cell. 2001;105:445–457. [PubMed]
  • Kitajima TS, Hauf S, Ohsugi M, Yamamoto T, Watanabe Y. Human Bub1 Defines the Persistent Cohesion Site along the Mitotic Chromosome by Affecting Shugoshin Localization. Curr Biol. 2005;15:353–359. [PubMed]
  • Kitajima TS, Sakuno T, Ishiguro K, Iemura S, Natsume T, Kawashima SA, Watanabe Y. Shugoshin collaborates with protein phosphatase 2A to protect cohesin. Nature. 2006;441:46–52. [PubMed]
  • McGuinness BE, Hirota T, Kudo NR, Peters JM, Nasmyth K. Shugoshin prevents dissociation of cohesin from centromeres during mitosis in vertebrate cells. PLoS Biol. 2005;3:1–17. [PMC free article] [PubMed]
  • Mei J, Huang X, Zhang P. Securin is not required for cellular viability, but is required for normal growth of mouse embryonic fibroblasts. Curr Biol. 2001;11:1197–1201. [PubMed]
  • Mendez J, Stillman B. Chromatin association of human origin recognition complex, cdc6, and minichromosome maintenance proteins during the cell cycle: assembly of prereplication complexes in late mitosis. Mol Cell Biol. 2000;20:8602–8612. [PMC free article] [PubMed]
  • Nakajima M, Kumada K, Hatakeyama K, Noda T, Peters JM, Hirota T. The complete removal of cohesin from chromosome arms depends on separase. J Cell Sci. 2007;120:4188–4196. [PubMed]
  • Nasmyth K. How might cohesin hold sister chromatids together? Philos Trans R Soc Lond B Biol Sci. 2005;360:483–496. [PMC free article] [PubMed]
  • Parelho V, Hadjur S, Spivakov M, Leleu M, Sauer S, Gregson HC, Jarmuz A, Canzonetta C, Webster Z, Nesterova T, et al. Cohesins functionally associate with CTCF on mammalian chromosome arms. Cell. 2008;132:422–433. [PubMed]
  • Pfleghaar K, Heubes S, Cox J, Stemmann O, Speicher MR. Securin is not required for chromosomal stability in human cells. PLoS Biol. 2005;3:e416. [PubMed]
  • Rollins RA, Morcillo P, Dorsett D. Nipped-B, a Drosophila homologue of chromosomal adherins, participates in activation by remote enhancers in the cut and Ultrabithorax genes. Genetics. 1999;152:577–593. [PubMed]
  • Rubio ED, Reiss DJ, Welcsh PL, Disteche CM, Filippova GN, Baliga NS, Aebersold R, Ranish JA, Krumm A. CTCF physically links cohesin to chromatin. Proc Natl Acad Sci U S A. 2008;105:8309–8314. [PubMed]
  • Stedman W, Kang H, Lin S, Kissil JL, Bartolomei MS, Lieberman PM. Cohesins localize with CTCF at the KSHV latency control region and at cellular c-myc and H19/Igf2 insulators. EMBO J. 2008;27:654–666. [PubMed]
  • Stemmann O, Zou H, Gerber SA, Gygi SP, Kirschner MW. Dual inhibition of sister chromatid separation at metaphase. Cell. 2001;107:715–726. [PubMed]
  • Sun Y, Yu H, Zou H. Nuclear Exclusion of Separase Prevents Cohesin Cleavage in Interphase Cells. Cell Cycle. 2006;5:2537–2542. [PubMed]
  • Tang Z, Shu H, Qi W, Mahmood NA, Mumby MC, Yu H. PP2A is required for centromeric localization of Sgo1 and proper chromosome segregation. Dev Cell. 2006;10:575–585. [PubMed]
  • Uhlmann F, Wernic D, Poupart MA, Koonin EV, Nasmyth K. Cleavage of cohesin by the CD clan protease separin triggers anaphase in yeast. Cell. 2000;103:375–386. [PubMed]
  • Waizenegger I, Gimenez-Abian JF, Wernic D, Peters JM. Regulation of human separase by securin binding and autocleavage. Curr Biol. 2002;12:1368–1378. [PubMed]
  • Waizenegger IC, Hauf S, Meinke A, Peters JM. Two distinct pathways remove mammalian cohesin from chromosome arms in prophase and from centromeres in anaphase. Cell. 2000;103:399–410. [PubMed]
  • Wang Z, Yu R, Melmed S. Mice lacking pituitary tumor transforming gene show testicular and splenic hypoplasia, thymic hyperplasia, thrombocytopenia, aberrant cell cycle progression, and premature centromere division. Mol Endocrinol. 2001;15:1870–1879. [PubMed]
  • Wendt KS, Yoshida K, Itoh T, Bando M, Koch B, Schirghuber E, Tsutsumi S, Nagae G, Ishihara K, Mishiro T, et al. Cohesin mediates transcriptional insulation by CCCTC-binding factor. Nature. 2008;451:796–801. [PubMed]
  • Zou H, McGarry TJ, Bernal T, Kirschner MW. Identification of a vertebrate sister-chromatid separation inhibitor involved in transformation and tumorigenesis. Science. 1999;285:418–422. [PubMed]
  • Zou H, Stemmann O, Anderson JS, Mann M, Kirschner MW. Anaphase specific auto-cleavage of separase. FEBS Letters. 2002;528:246–250. [PubMed]