Securin Is Not Required for Chromosomal Stability in Human Cells

Abnormalities of chromosome number are frequently observed in cancers. The mechanisms regulating chromosome segregation in human cells are therefore of great interest. Recently it has been reported that human cells without an hSecurin gene lose chromosomes at a high frequency. Here we show that, after hSecurin knockout through homologous recombination, chromosome losses are only a short, transient effect. After a few passages hSecurin−/− cells became chromosomally stable and executed mitoses normally. This was unexpected, as the securin loss resulted in a persisting reduction of the sister-separating protease separase and inefficient cleavage of the cohesin subunit Scc1. Our data demonstrate that securin is dispensable for chromosomal stability in human cells. We propose that human cells possess efficient mechanisms to compensate for the loss of genes involved in chromosome segregation.


Introduction
A number of factors are involved in ensuring that in dividing cells chromosomes are copied exactly once and then distributed correctly to daughter cells. Chromosome cohesion is established during chromosome replication in S-phase and is mediated by the multisubunit cohesin complex, which forms a giant ring structure possibly encircling sister chromatids [1]. Sister chromatid separation in anaphase depends on the removal of cohesin complexes from chromosomes [2]. In vertebrates removal of cohesin from chromosomes occurs in at least two steps. The ''prophase pathway'' removes the bulk of cohesin from chromosome arms during prophase and prometaphase [3,4]. By metaphase only minor amounts of cohesin remain on chromosomes, preferentially at centromeres [4]. Centromere-specific factors, such as shugoshin, protect the cohesion between sister centromeres from the prophase pathway [5,6]. At the metaphase-toanaphase transition, residual cohesion is dissolved by the large cysteine endopeptidase separase, which cleaves the socalled kleisin subunit of cohesin (Scc1/Rad21 in mitosis; Rec8 in meiosis). This cleavage allows sister chromatids to move apart [7,8] and is, in fact, essential for anaphase to occur [9].
For most of the cell cycle, separase activity is inhibited by binding of an inhibitory chaperone called securin [10][11][12] or by phosphorylation-dependent complex formation with Cdk1 [13,14]. Separase is eventually activated by proteolysis of securin or the cyclin B subunit of Cdk1, which in both cases is mediated by a ubiquitin protein ligase named anaphase promoting complex or cyclosome (APC/C) and its cofactor Cdc20 [15,16].
Thus, securin is a key substrate of the APC/C CDC20 pathway. Though conserved in function, securins from different phyla are highly divergent in sequence [17]. Earlier studies had already implicated securin in functional mechanisms related to cell-cycle control and tumorigenesis [18,19]. To further address securin's function, both copies of the gene encoding hSecurin were inactivated via homologous recombination in the karyotypically stable human colorectal cancer cell line HCT116 [20]. The results indicated that hSecurin is indeed needed for chromosomal stability in human cells, as hSecurindeficient cells exhibited high rates of chromosome missegregation, similar to those observed in many cancers. Furthermore, the data suggested that hSecurin, through its chaperone activity, plays a crucial role for the proper function of separase, especially for separase-dependent cleavage of the cohesin subunit Scc1 [20]. (Our group contributed to that paper some of the cytogenetic data using fixed cell suspensions provided by C. Lengauer's laboratory.) However, the important role of hSecurin elucidated in the Jallepalli et al. study [20] contrasts with the results of another investigation, which found mice lacking securin to be viable and apparently normal [21]. Furthermore, only 20% of mouse cells without securin exhibit gains or losses of chromosomes [22].
To resolve this discrepancy, we conducted further studies with the hSecurin À/À cell line. Here, we show that the initial missegregation phenotype was superseded by a regaining of chromosomal stability in only a few passages. The karyotype of chromosomally stable hSecurin À/À cells was indistinguishable from that of the parent cell line with intact securin. Surprisingly, the initially described biochemical defects caused by the lack of securin, i.e., significantly reduced levels of separase and inefficient cleavage of the cohesin subunit Scc1 [20], were still present.
This indicates that securin is not required for faithful chromosome segregation and that alternative mechanisms may compensate for the absence of securin and/or reduced separase levels.

Results
Analysis of Chromosomal Instability in the hSecurin À/À Cell Line at Different Passages In an initial step, metaphase spreads of the hSecurin À/À cell line were karyotyped by multiplex fluorescence in situ hybridization (M-FISH) at different passages ( Figure 1). For passages 2 and 3, we confirmed the loss of numerous chromosomes in the majority of analyzed cells ( Figure 1A and 1B). About 60% (12/20) of metaphase spreads showed losses of at least one chromosome. Surprisingly however, the high rate of chromosome losses in the hSecurin À/À cell line had almost vanished by passage 8 ( Figure 1C), when chromosome losses were noted in only 10% (2/20) of cells. By passage 12, we observed no chromosome losses ( Figure 1D). In the latter two analyses, merely one metaphase spread each had a gain of a single chromosome ( Figure 1C and 1D).
The entire experiment was repeated, and it again showed the same phenomenon, i.e., decreasing chromosome losses in the hSecurin À/À cell line with increasing passage numbers ( Figure 1E).
As expected, the parent cell line HCT116 was chromosomally stable and remained stable throughout all analyses (unpublished data).
When we karyotyped the hSecurin À/À cells at passage 12, we found that the karyotype of the cells was identical to the karyotype of the parent cell line HCT116 ( Figure 1F; a detailed karyotype description is given in Materials and Methods). Thus, the hSecurin À/À cells showed all known structural rearrangements from the parental cell line HCT116, which we had described before [20,23]. There were no additional, new changes, and none of the known aberrations was lost.
Interphase FISH confirmed that hSecurin À/À cells at passage 12 were indeed chromosomally stable, similarly to the parental cell line HCT116 ( Figure 2).

Confirmation that Chromosomally Stable Cell Lines Are Indeed hSecurin À/À Cells
In the next step we confirmed that chromosomally stable hSecurin À/À cells were indeed hSecurin deficient. We used PCR primer pairs spanning the second and third exons of the securin gene, as described previously [20] ( Figure 3A and 3B). In addition, we designed a primer set flanking exon 3 to specifically demonstrate that the hSecurin À/À cells lack exon 3 of the securin gene ( Figure 3A and 3C). Genomic PCR analyses with these primer sets using DNA extracted from chromosomally stable hSecurin À/À cells demonstrated that the cells had indeed a homozygous deletion of the exon 3 region of the hSecurin gene. Primers for exons 8 and 9 of p53 were used as an additional control ( Figure 3B).
Normal Execution of Anaphase in hSecurin À/À Cells Previously, it was reported that cells lacking hSecurin grew somewhat more slowly than wild-type cells [20]. In contrast, the growth pattern of chromosomally stable hSecurin À/À cells was indistinguishable from that of the HCT116 parent cell line (unpublished data). Therefore, we performed immunofluorescence experiments to examine the distribution of centromeres during mitosis with cells from passages 12 or higher. hSecurin þ/þ cells and chromosomally stable hSecurin À/À cells in various stages of mitosis were stained with the CREST (calcinosis-Raynaud's phenomenon-esophageal dismobilitysclerodactyly-telangiectasia syndrome of scleroderma) antibody, which recognizes kinetochore proteins ( Figure 4). In addition, we used cyclin B1 as a marker for mitotic stage. Anaphase cells were first identified by virtue of chromosome condensation and lack of cyclin B staining and then scored for unsegregated chromatids remaining at the metaphase plate. In previous experiments, about 30% of hSecurin À/À cells in anaphase still had paired sister chromatids left behind at the metaphase plate, when most of the other chromosomes had segregated to the poles [20]. When we analyzed a total of 75 cells in three separate experiments in each hSecurin þ/þ ( Figure 4A-4G) and hSecurin À/À ( Figure 4H-4N) cell line, we found no differences. In fact, the vast majority of analyzed anaphase cells in each cell line displayed a complete separation of sister chromatids and migration of centromeres to opposite poles ( Figure 4O).
Furthermore, as in the first paper [20], we did not observe premature sister chromatid separation when cells were exposed to microtubule poisons such as colchecine (unpublished data).
These results indicate that chromosomally stable hSecurin À/À cells execute anaphase normally, with complete sister chromatid separation.
We analyzed separase levels in chromosomally stable hSecurin À/À cells synchronized by nocodazole. Lysates from hSecurin þ/þ and hSecurin À/À cells were probed with antibodies to separase. For each cell line we detected both the full-length and the cleaved forms of separase ( Figure 5A). However, both the full-length and the cleaved forms of separase were consistently 3-to 4-fold weaker in the chromosomally stable hSecurin À/À cells.
We reconstituted the cleavage reaction, which dissociates Scc1 from the centromeric regions, by using immunoprecipitated separase complexes that were first incubated with Xenopus egg extracts as a source of mitotic APC/C CDC20 . In the case of HCT116 parent cells, incubation of activated separase with in vitro-translated 35 S-hScc1 resulted in typical cleavage fragments that were readily detectable after a 20-min incubation ( Figure 5B). Four times as much starting cell material was used to purify separase from hSecurin À/À cells as was used to purify the same amount of separase from wildtype cells ( Figure 5C). However, the separase from hSecurin À/À cells did not display any cleavage activity towards Scc1, even after a 90-min incubation ( Figure 5B).
Despite the absence of activity in vitro, separase autocleavage products ( Figure 5A) demonstrate the presence of at least some separase activity in hSecurin À/À cells, which, apparently, is sufficient to execute anaphase normally (see above).
Interestingly, immunoprecipitated separase from nocodazole-arrested cells showed a higher degree of self-cleavage in hSecurin À/À cells compared to that in wild-type cells. This suggests that separase might be partly deregulated in the hSecurin À/À cells ( Figure 5C).

Discussion
We report here that hSecurin À/À cells are capable of compensating securin loss and rapidly regain chromosomal stability within a few passages. Our findings were unexpected, as chromosomally stable hSecurin À/À cells continue to have the initially described biochemical defect, i.e., reduction of both the amount and the activity of separase [20; this study].
Our data may explain why mice lacking securin are viable and normal [21]. However, mouse cells without securin have little change in the level of separase [22]. This is in contrast to our observations in human cells, where absence of securin resulted in the aforementioned significant reduction of both separase and its cleavage product. Furthermore, about 20% of mouse embryonic stem cells without securin were aneuploid [22], while the percentage of aneuploid human hSecurin À/À cells was reduced to background levels similar to those observed in the chromosomally stable parent cell line HCT116. These data indicate that significant differences in separase regulation between human and mouse cells must exist.
The reconstitution of chromosomal stability and complete separation of sister chromatids in hSecurin À/À cells suggest that significantly lower than normal amounts of separase are sufficient for normal execution of anaphase. This is in agreement with separase at wild-type levels being able to efficiently remove from chromosomes even vastly increased amounts of cohesin [7].
In contrast to budding yeast, human cells lacking hSecurin still manage to arrest sister chromatid separation in the presence of spindle poisons [20; unpublished data]. Therefore, additional mechanisms that regulate the removal of cohesin in human cells must exist. One additional, securinindependent mechanism of separase inhibition involves phosphorylation by Cdk1 and subsequent binding of the kinase [13,22]. As securin and Cdk1 bind separase in a mutually exclusive manner [14], Cdk1 may be capable of compensating for the loss of securin. Indeed, mouse embryonal stem cells lacking both forms of separase regulation suffer from precocious sister chromatid separation under mitotic checkpoint arrest [22]. Another level of control is probably exerted by shugoshin, which prevents removal of centromere-specific cohesin before the onset of anaphase [5].
Our findings demonstrate that deletion of hSecurin has little or no effect on long-term chromosome segregation fidelity in human cells. In fact, our results even raise the possibility that the chromosomal instability (CIN) phenotype observed in early passages of the hSecurin À/À cells might have been caused by insults during the deletion process rather than by loss of securin per se. Alternatively, cells might upregulate other control pathways in response to loss of securin, thereby regaining chromosomal stability. Comparative gene expression profiling before and after loss of hSecurin might reveal compensatory changes in the expression of genes involved in anaphase regulation. We therefore compared the transcriptomes of the HCT116 parent and the hSecurin À/À cell line using the Affymetrix U133A chip. Indeed, significantly different expression levels were found for PLK2 (Polo-like kinase 2), RCC1 (regulator of chromosome condensation 1, also known as chromosome condensation 1 [CHC1]), and SMC6L1 (SMC6 structural maintenance of chromosomes 6-like 1) (unpublished data). However, future experiments will be required to determine the physiological significance of these findings and whether the above proteins might play a currently unknown role in the other two known regulations of sister chromatid separation.
In summary, we have shown that securin is not required for chromosomal stability in human cells. Our results affect current mathematical models of colorectal cancer investigating the role of genetic instability in tumorigenesis [24]. The crucial effect attributed to CIN is acceleration of the mutation rate. However, our data indicated that a CINcausing mutation may not reach fixation in a given cell compartment, which should therefore change existing assumptions on the evolution of CIN lesions and their growth rates.
Finally, implications of this study extend beyond mechanisms leading to chromosomal instability and affect possible strategies for cancer therapy. It has been suggested that, as stability pathways of tumor cells are defective, cancer cells may be more sensitive to stress-inducing agents and they should be especially susceptible to attack by instabilityinducing drugs [25]. However, our results suggest that targeting a particular pathway may not destroy a cell but rather activate alternative pathways. A search and detailed characterization of these alternative pathways will provide further clues to the nature of CIN in human cancers.
As an additional control we used primers for exons 8 and 9 within TP53. The 39 primer was CATGATTCAGAACCCTGGAG; the 59 primer was AGGACCTGATTTCCTTACTGC.
Analysis of sister chromatid separation. Cells were grown on coverslips for 24 h in McCoy's 5A medium (Gibco Invitrogen) plus 100 units/ml penicillin and 0.1 mg/ml streptomycin without FBS. Aphidicolin was added to a final concentration of 0.15 lg/ml in McCoy's 5A medium with 10% FBS. After 14 h, the medium was removed, and cells were washed four times with PBS and cultured for 8-10 h in McCoy's 5A medium plus 10% FBS to obtain cells in anaphase.
Cells were fixed in a 4% paraformaldehyde solution, washed three times with PBS/0.2% Tween, and permeabilized in 0.5% TritonX in PBS/Tween for 15 min.
Separase activity assay. HCT116 wild-type (four 75-ml dishes) and hSecurin À/À cells (12 75-ml dishes) were lysed as above, and separase was immunoprecipitated with a rabbit polyclonal antibody raised against the sequence GSDGEDSASGGKTPA of human separase. For each immunoprecipitation, 8 lg of antibodies was prebound to 30 ll of Protein G Sepharose 4 Fast Flow (Amersham Biosciences, Little Chalfont, United Kingdom). Separase activation in Xenopus extract and Scc1 cleavage assays were done as described elsewhere [13], except that the Scc1 cleavage reaction was performed in the presence of 1.3 lg/ll antigenic peptide. Amounts and self-cleavage of separase were analyzed by immunoblotting aliquots before and after incubation in the extract.