Identification of a Vertebrate Sister-Chromatid Separation Inhibitor Involved in Transformation and Tumorigenesis

See allHide authors and affiliations

Science  16 Jul 1999:
Vol. 285, Issue 5426, pp. 418-422
DOI: 10.1126/science.285.5426.418


A vertebrate securin (vSecurin) was identified on the basis of its biochemical analogy to the Pds1p protein of budding yeast and the Cut2p protein of fission yeast. The vSecurin protein bound to a vertebrate homolog of yeast separins Esp1p and Cut1p and was degraded by proteolysis mediated by an anaphase-promoting complex in a manner dependent on a destruction motif. Furthermore, expression of a stableXenopus securin mutant protein blocked sister-chromatid separation but did not block the embryonic cell cycle. The vSecurin proteins share extensive sequence similarity with each other but show no sequence similarity to either of their yeast counterparts. Human securin is identical to the product of the gene called pituitary tumor-transforming gene (PTTG), which is overexpressed in some tumors and exhibits transforming activity in NIH 3T3 cells. The oncogenic nature of increased expression of vSecurin may result from chromosome gain or loss, produced by errors in chromatid separation.

The metaphase to anaphase transition is the final discrete event in duplication and separation of the genetic material of a cell. Its timing is regulated by the activation of the anaphase-promoting complex (APC), which mediates selective proteolysis of various mitotic regulators (1–3). Experiments with Xenopus egg extracts indicated that a putative protein factor might exist whose degradation was required for the onset of sister-chromatid separation (4). Proteins with such an activity were subsequently found in both budding yeast and fission yeast, encoded by the genes PDS1 andCUT2, respectively (5–7). Both proteins are APC substrates and their degradation is required for chromatid separation (8, 9). Pds1p and Cut2p associate with the yeast separin proteins Esp1p and Cut1p, respectively (10, 11), and prevent the separins from promoting chromatid separation. Because of their similar cell cycle functions, Pds1p and Cut2p are also called anaphase inhibitors or securins (12).

The regulation of sister-chromatid separation in metazoans might be similar. Unfortunately, Pds1p and Cut2p show no sequence similarity to each other, and currently no sequence in the GenBank and EST databases shows any similarity to either of them. The COOH-terminus of a putative human separin homolog (hESP1), found by cDNA sequencing (13), has 28% identity with budding yeast Esp1p and 30% identity with fission yeast Cut1p (11). There is no similarity in the NH2-terminus. We therefore attempted to identify the human securin homolog through its expected association with the putative human separin.

Antibodies to hESP1 were raised to a 269–amino acid fragment at the COOH-terminal region of hESP1 (14). After affinity purification, the antibodies were covalently coupled to protein A beads (15). Using these antibodies, we looked for proteins that coimmunoprecipitated with hESP1 proteins that were present in extracts of cells in metaphase but not in extracts of cells in anaphase, the expected properties of a human securin homolog. These extracts were prepared by release of human HeLa S3 cells from nocodazole-induced metaphase arrest (16). At various times, extracts were prepared and immunoprecipitated with anti-hESP1 (17). Immunoprecipitated proteins were separated by SDS–polyacrylamide gel electrophoresis (PAGE) and detected by silver staining. Among the many proteins that immunoprecipitated, two had apparent molecular sizes of 28 kD (EAP1, for hESP1-associated protein 1) and 42 kD (EAP2, for hESP1-associated protein 2). Both proteins were present in constant amounts in extracts prepared at various times up to 90 min after removing nocodazole. Neither protein was present in extracts prepared 4 hours after release from metaphase arrest (Fig. 1A). As a temporal control for progress through M phase, amounts of cyclin B1 were monitored by protein immunoblot analysis. Like EAP1 and EAP2, cyclin B1 was stable up to 90 min after release but was not detected at the 4-hour time point (Fig. 1A), indicating that the APC-mediated proteolysis pathway became active between 90 min and 4 hours after release. The coincidence between the loss of association of EAP1 and EAP2 with hESP1 and the activation of APC suggests that EAP1 and EAP2 could be candidates for a human securin-like molecule.

Figure 1

Identification of an APC substrate associated with human EPS1. (A) Extracts were made from synchronized HeLa S3 cells. The time after release from nocodazole arrest is indicated in minutes. ESP1-associated proteins were coimmunoprecipitated with anti-hESP1. Beads were washed four times in lysis buffer and proteins were analyzed by SDS-PAGE. The amounts of cyclin B1 were determined by protein immunoblotting. (B) Immunoprecipitation with anti-hESP1 from extracts of HeLa S3 cells. As a control, preimmune antiserum (lane 1) was used in parallel with anti-hESP1 (lane 2). Both hESP1 and PTTG were detected by protein immunoblotting with respective antibodies. (C) Immunoprecipitations with antibody to the Myc tag (Santa Cruz) were performed in extracts prepared form 293T cells transfected with either Myc-tagged human PTTG (lane 2) or Myc-taggedXenopus p25 (lane 1). Protein immunoblotting was performed to detect hESP1. We were unable to immunoprecipitate hESP1 with antibody to full-length PTTG, presumably because the binding sites on PTTG for ESP1 and the antibody overlap. (D) The fractions from a Superdex 200 column (33) were analyzed by protein immunoblotting to detect hESP1 and PTTG. Most of the PTTG cofractionated with hESP1 at an apparent molecular size of ∼500 kD. Some PTTG was detected in fractions corresponding to ∼70 kD, the same elution position of recombinant PTTG protein (24). (E) Sequence alignment of vertebrate securins (25). The hSecurin h(Sec) sequence obtained in this study (14) is identical to that of the human PTTG (21). The sequences of mouse securin (mSec) and rat securin (rSec) were obtained from GenBank (accession numbers AF069051 andU73030, respectively) (xSec, Xenopus securin). Amino acid sequence alignment was obtained with MegaAlign (DNAStar) by the clustal method. Residues that are identical or conserved among all four proteins are shaded in black. The conserved D-box is boxed.

We had also isolated Xenopus APC substrates by small-pool expression cloning (18, 19). In this approach, small pools of cDNA clones from a Xenopus blastula library (20) were translated and labeled in vitro in rabbit reticulocyte lysates. Each pool was divided and incubated in mitotic extract or interphase extract. The cDNA clones corresponding to proteins that were proteolyzed in mitotic extract but not in interphase extract were isolated. We identified various cyclin Bs and geminin, an inhibitor of DNA replication degraded upon exit from mitosis (19). Two other proteins were identified in this screen: a 70-kD kinesin-like protein and a 25-kD protein (p25). The primary structure of Xenopus p25 shares extensive similarity with a human protein encoded by a gene called pituitary tumor-transforming gene (PTTG) (21–23). The open reading frame ofPTTG from a human fetal thymus cDNA library was isolated and antibody to full-length PTTG protein was prepared (14). The following evidence suggested that EAP1 might be identical to PTTG. First, in vitro–translated PTTG protein migrated with the same mobility as EAP1 on SDS-PAGE (24). Second, PTTG also associated with hESP1. Endogenous PTTG coimmunoprecipitated with antibody to hESP1 in HeLa cell extract as detected by the antibody to PTTG (Fig. 1B). Endogenous hESP1 also coimmunoprecipitated with antibody to the Myc tag in extracts prepared from 293T cells transiently expressing Myc-tagged PTTG (Fig. 1C). Furthermore, hESP1 and PTTG from HeLa cell extracts cofractionated on both anionic exchange (24) and gel filtration columns (Fig. 1D). Thus, PTTG appears to be a vertebrate APC substrate that is associated with a vertebrate separin until activation of the APC. Therefore, we tentatively named EAP1/PTTG as hSecurin andXenopus p25 as xSecurin.

Among the vertebrate securins, sequence similarity was observed throughout the entire sequence (Fig. 1E). A conserved motif [RKALG(T or N)VN] (25) matches the destruction box (D-box) [RX(A or V or T)LGXXXN] shared by many APC substrates (26). The vertebrate securins share no sequence similarity with their yeast counterparts. In fact, the frog securin displays unusual diversity from its mammalian homologs (about 30% identity); most other cell cycle proteins are more than 80% identical in sequence. Nonetheless, there are conserved sequence features shared by all securins. All securins contain clusters of acidic and basic domains. The NH2-terminal half of the proteins is rich in lysine residues surrounding the D-box. This is common for D-box–containing APC substrates, presumably because lysine is the residue that forms a covalent isopeptide linkage with ubiquitin.

To characterize the cell cycle function of vertebrate securins, we determined their abundance at various stages of the cell cycle. HeLa S3 cells were synchronized by release from a double-thymidine block, and extracts were prepared during the following 12 hours. Securin was detected by anti-hSecurin as two closely spaced bands. The amount of securin begins to accumulate at the onset of S phase and peaks at G2-M phases in parallel with cyclin B1. As expected, its level drops precipitously when APC is activated, indicated by the decline of cyclin B1 (Fig. 2A).

Figure 2

Degradation of the vertebrate homolog of Pds1p by APC-mediated proteolysis. (A) HeLa S3 cells were synchronized at the G1-S transition by a double-thymidine block. After release from arrest, extracts were prepared at various times up to 12 hours. The bottom two panels show the amounts of hSecurin and cyclin B1 analyzed by protein immunoblotting. The top graph indicates the percentage of cells in the G1, S, and G2-M phase of the cell cycle at the corresponding time points, as determined by FACS analysis. (B) Both xSecurin and xSecurindm protein were translated in vitro in the presence of [35S]methionine. A portion of the translation mixture (2 μl) was added into interphase or mitotic extracts (7 μl) supplemented with bovine ubiquitin (10 μg). The reaction was incubated at room temperature for the time (minutes) indicated below the autoradiograph. (C) In vitro–translated xSecurin (1.5 μl) was added to mitotic extract (10 μl) supplemented with ubiquitin (10 μg) and incubated for 15 min. As competitors, a purified Xenopus cyclin B1 NH2-terminal fragment (amino acids 1 to 102) and NH2-terminal fragments lacking the D-box (CycB1-dbΔ) (34) were added to the reaction mixture. The final concentrations of competitors are indicated above the autoradiograph. (D) The same as in (B), except that Xenopus cyclin B1 NH2-terminal fragments were labeled by iodination, and xSecurin or xSecurindm were used as competitors.

Genetic studies in yeast and biochemical experiments inXenopus egg extracts using yeast Pds1p and Cut2p suggested that Pds1p and Cut2p are ubiquitinated by the APC in a D-box–dependent manner (8, 27). To determine whether the putative D-box of the vertebrate securin is functional, we mutated the RKAL residues to AKAA (25) by site-directed mutagenesis. Mutated xSecurin (xSecurindm) was stable in mitotic extracts, confirming that the RKAL sequence is required for degradation (Fig. 2B). Similar observations were made with hSecurin (24). Furthermore, xSecurin was stabilized in the presence of an excess of an NH2-terminal fragment of cyclin B1 that contains a D-box. However, the same cyclin B1 fragment lacking a D-box motif did not affect the degradation of xSecurin (Fig. 2C). Conversely, an excess of wild-type xSecurin, but not xSecurindm, inhibited the degradation of cyclin B1 (Fig. 2D). Taken together, these results demonstrate that the abundance of xSecurin is regulated by APC-mediated proteolysis in a D-box–dependent manner.

In yeast, the securins (Pds1p and Cut2p) function as inhibitors of chromatid separation. We therefore tested the effects of xSecurindm on sister-chromatid separation inXenopus egg extracts (4, 28). In these experiments, we allowed extracts to go through one full cell cycle and observed chromatid separation at the following anaphase. Approximately 1 μl of purified xSecurindm protein (0.5 mg/ml) was added with Xenopus sperm nuclei and rhodamine tubulin to a 10-μl portion of a freshly prepared egg extract arrested at a metaphase-like stage (unfertilized egg extract). For comparison, equal amounts of bovine serum albumin were added to a separate sample of the same extract. The extracts were released into interphase by addition of calcium to allow DNA replication and then driven into mitosis by addition of unfertilized egg extract (2.5 μl) to allow the formation of the metaphase spindle. To prevent chromosome decondensation, which makes detection of the chromosomes difficult at late anaphase, we also added a nondegradable cyclin B1 lacking the NH2-terminal 90 amino acids (Δ90 cyclin B1) (final concentration, 20 μg/ml). In this system, the cell cycle is arrested at metaphase and can be released into anaphase by addition of calcium. Anaphase movement of chromosomes can be monitored by the positions of the mitotic chromosomes, which were detected by DNA-specific fluorescent dye. We did not observe any differences in interphase nucleus formation, chromosome condensation, or nuclear envelope breakdown between extracts to which xSecurindm had been added and control extracts. Within 10 min of the second addition of calcium, chromosomes in control extracts had begun to move to the spindle poles (5 out of 10 spindles). No movement was seen in extracts containing xSecurindm even after 30 min, whereas at the same time in control extracts, 48% (11 out of 23) of the spindles were at late anaphase and 43% (10 out of 23) were at telophase (Fig. 3A).

Figure 3

Inhibition of chromatid separation by Xenopus securin. Anaphase was induced in extracts in the presence of (A) and in the absence of (B) nondegradable Δ90 cyclin B1. Photographs show the Hoechst 33342–stained chromosomes (blue) and rhodamine-labeled mitotic spindle (red) at various times after metaphase release. The percentage of spindles at metaphase for each time point (10 to 25 total spindles) is indicated. The white bar in the lower right panel of (A) represents 10 μm.

In budding yeast, Pds1p has been suggested to be part of a checkpoint pathway that arrests the cell cycle at metaphase in the presence of DNA damage that occurs after G1 (7). However, this function of Pds1p may reflect the unique properties of the budding yeast S-phase DNA damage checkpoint. Other eukaryotes, such as fission yeast and vertebrates, arrest the cell cycle at the G2-M boundary in response to DNA damage occurring after G1 by inhibiting the activation of the CDC2 cyclin-dependent kinase.

To test whether xSecurin inhibits any aspect of the anaphase progression other than chromatid separation, we performed the above assay with xSecurindm without the addition of Δ90 cyclin B1 and tested for inhibition of spindle disassembly, chromosome decondensation, and nuclear membrane reformation. No chromatid separation was observed in extracts containing xSecurindmup to 15 min after the second addition of calcium (eight spindles). Between 15 and 20 min after the second calcium addition, extracts had begun to decondense chromosomes and disassemble spindles. After 25 to 30 min, interphase nuclei were detected in both extracts (Fig. 3B). Protein immunoblot analysis with antibodies to xSecurin and cyclin B1 revealed that cyclin B1 is degraded in the presence of xSecurindm (24). These data demonstrate that xSecurindm does not interfere with assembly or disassembly of the spindle, with condensation or decondensation of chromosomes, or with breakdown or reformation of the nuclear envelope and thus appears not to interfere with the cycle of CDC2 cyclin-dependent kinase activation and inactivation. Instead, xSecurin specifically inhibits chromatid separation in Xenopus egg extracts. It remains possible that the checkpoint pathway is absent in frog embryos and that the vertebrate somatic cells have a checkpoint mechanism involving securins.

The vertebrate securin proteins have been implicated in transformation and tumorigenesis. Overexpression of securins led to the transformation of NIH 3T3 cells, and resulting transformants induced tumors when implanted into nude mice (22–23). In addition, expression of hSecurin is high in all carcinoma cell lines that have been tested, and in one case, the levels of hSecurin expression correlate with the malignancy of disease (29). The finding that a vertebrate securin has tumorigenic activity is somewhat anticipated because chromosome missegregation has been predicted to be a major source of genetic instability with profound consequences for cancer (30). On the basis of its function reported here, the simplest explanation is that tumor formation is the result of aneuploidy caused by defects in the sister-chromatid separation. In yeast, aneuploidy often occurs in mutants defective in sister-chromatid separation (6, 31, 32). Chromosome missegregation could lead to increases in the dosage of proto-oncogenes or loss of heterozygosity of tumor suppressors. Alternatively, the tumorigenic activity could result from an unknown function (21).

Our results indicate that, despite the low level of conservation among the securins, the basic process of chromatid separation is conserved in all eukaryotes. Identification of human securin as an oncogene suggests that misregulation of chromatid separation may contribute to the generation of malignant tumors.

  • * To whom correspondence should be addressed. E-mail: marc{at}


View Abstract

Navigate This Article