|Home | About | Journals | Submit | Contact Us | Français|
Sister chromatid separation depends on the activity of separase, which in turn requires the proteolysis of its inhibitor, securin. It has been speculated that securin also supports the activation of separase. In this study, we found that PTTG1 was the major securin isoform expressed in most normal and cancer cell lines. Remarkably, a highly homologous isoform called PTTG2 was unable to interact with separase. Using chimeras between PTTG1 and PTTG2 and other approaches, we pinpointed a single amino acid that accounted for the loss of securin function in PTTG2. Mutation of the homologous position in PTTG1 (H134) switched PTTG1 from an inhibitor into an activator of separase. In agreement with this, PTTG1 lacking H134 was able to trigger premature sister chromatid separation. Conversely, introduction of H134 into PTTG2 is sufficient to allow it to bind separase. These data demonstrate that while the motif containing H134 has a strong affinity for separase and is involved in inhibiting it, another domain(s) is involved in activating separase and has a weaker affinity for it. Although PTTG2 lacks securin function, its differences from PTTG1 provide evidence of independent inhibitory and activating functions of PTTG1 on separase.
Sister chromatid cohesion is established during DNA replication and maintained throughout the G2 phase and early mitosis by cohesin complexes (1). Separase is responsible for cleaving the cohesin in the centromeric region, allowing sister chromatid separation at the onset of anaphase (2). Timely activation of separase is controlled by binding to an inhibitor called securin. At the end of mitosis, securin is degraded by the anaphase-promoting complex or cyclosome (APC/C) proteasome pathway, thereby allowing separase to fulfill its function (3). Mammalian securin was initially isolated from rat pituitary tumor cells as pituitary tumor-transforming gene 1 (PTTG1) (4) and was subsequently demonstrated to function as a securin (5). Human PTTG1 is located on chromosome 5q35.1 (6). A homologous intronless gene, PTTG2, was found to be located at 4p12 (7). A third intronless isoform, PTTG3, is located at 8q13.1 (8) and should be a pseudogene (9).
Given that securin is a relatively unconserved protein, it is not surprising that its mode of binding to separase differs considerably in different organisms. In budding yeast, securin prevents the interaction between the N-terminal domain and the C-terminal domain of separase, inhibiting both protease activity and access to substrates (10). In Drosophila, separase (SSE) is a much smaller protein and securin (PIM) binds to the N terminus of SSE, as well as to THR, which may functionally correspond to the N terminus of separase in other organisms (11). Biochemical studies of human securin suggest that it inhibits separase by preventing the access of substrates to the active site of separase (12).
Evidence from different organisms indicates that securin not only may inhibit separase, but is also required for proper separase activation. In budding yeast, securin mutants display retarded anaphase entry and synthetic lethality with separase mutations (13). In fission yeast, the loss of securin produces the same block to chromosome segregation as the loss of separase (14). In these systems, securin is implicated in the nuclear targeting of separase (10, 15, 16). Although securin is less conserved in higher eukaryotes, a role of securin in activating separase has also been suggested. Similar to separase mutants, securin mutants in Drosophila fail to initiate anaphase (17). Human HCT116 cells lacking PTTG1 lose chromosomes at a high frequency due to defective anaphase (18). Deletion of just one copy of the ESP1 gene is lethal to mouse embryos lacking PTTG1, indicating that deletion of PTTG1 reduces separase activity (19). As the nuclear envelope is dismantled during mitosis in higher eukaryotes, how securin promotes separase activity in organisms other than yeast remains unknown.
In this study, we showed that PTTG2 does not interact with separase. Using this as a starting point, we then discovered a single residue in PTTG1 that is indispensable for inhibition of separase without affecting the activation of separase. These results for the first time demonstrated separable inhibitory and activating functions of PTTG1.
HeLa (cervical carcinoma), Hep3B (hepatocellular carcinoma), HepG2 (hepatoblastoma), and IMR90 (normal fibroblasts) were obtained from the American Type Culture Collection (Manassas, VA, USA). The HeLa cell line used in this study was a clone that expressed the tTA tetracycline repressor chimera (20). HeLa cells expressing an APC/C biosensor were as described previously (21). NP361, NP460, NP550 (immortalized normal nasopharyngeal epithelial), and C666-1 (nasopharyngeal carcinoma) cells were gifts from George Tsao (University of Hong Kong). CNE2, HONE1, and HNE1 (nasopharyngeal carcinoma) cells were obtained from AoE NPC Research Tissue Bank (Hong Kong). LO2 and MIHA (immortalized normal liver) cells were gifts from Irene Ng (University of Hong Kong).
Cells were propagated in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (vol/vol) calf serum (Life Technologies, Carlsbad, CA, USA) (for HeLa cells) or fetal bovine serum (Life Technologies) (for Hep3B, HepG2, HONE1, LO2, and MIHA cells) and 50 U/ml penicillin-streptomycin (Life Technologies) in a humidified incubator at 37°C in 5% CO2. IMR90 cells were cultured in Eagle minimum essential medium (Life Technologies) supplemented with 2× essential and nonessential amino acids and 15% fetal bovine serum. NP361, NP460, and NP550 cells were cultured in Epilife plus defined keratinocyte serum-free medium (DKSFM) (1:1) with 0.75% penicillin-streptomycin. Supplements were also added according to the manufacturer's instructions. C666-1 cells were grown in RPMI 1640 medium (Life Technologies) supplemented with 10% fetal bovine serum. CNE2 and HNE1 cells were propagated in DMEM supplemented with 5% (vol/vol) calf serum and 5% (vol/vol) fetal bovine serum.
Stable inducible cell lines expressing PTTG1 (or PTTG2) were produced by transfection of enhanced green fluorescent protein (EGFP)-tagged PTTG1 (or PTTG2) in pUHD-P1/PUR into HeLa cells. Two days after transfection, the cells were selected in medium supplemented with puromycin and doxycycline. A mixed population was isolated after about 2 weeks of selection. The expression of EGFP-tagged proteins was analyzed after the cells were cultured in medium in the presence or absence of doxycycline for 24 h. The established cell lines were propagated in the presence of doxycycline but without puromycin. HeLa cells that stably expressed histone H2B-GFP (22) were used for live-cell imaging.
Unless otherwise stated, cells were treated with the following reagents at the indicated final concentrations: blasticidin (5 μg/ml), cycloheximide (10 μg/ml), doxycycline (1 μg/ml), hydroxyurea (1.5 mM), MG132 (10 μM), nocodazole (NOC) (0.1 μg/ml), puromycin (1 μg/ml), RO3306 (Alexis) (10 μM), and thymidine (2 mM).
Cells were transfected with plasmids and small interfering RNAs (siRNAs) (10 nM unless otherwise stated), using a calcium phosphate precipitation method (23) and Lipofectamine RNAiMAX (Life Technologies), respectively. Cell extracts were prepared as described previously (24). Cell cycle synchronization by double-thymidine block or NOC block was described previously (25). Mitotic chromosome spread was performed as described previously (26).
PTTG1 in pEF-plink2 and PTTG1DΔ (with mutations in both the destruction box [D box] and the KEN box) in pEF-plink2 were generous gifts from Michael Brandeis (Hebrew University of Jerusalem, Jerusalem, Israel). The coding region of PTTG1 (or the DΔ mutant) was amplified by PCR with 5′-AGAATTCATGGCTACTCTGATCTATGT-3′ and 5′-AGAATTCTTAAATATCTATGTCACAGC-3′; the PCR product was cut with EcoRI and ligated into pUHD-P1 (27) to generate FLAG-PTTG1 (or DΔ) in pUHD-P1.
PTTG2 cDNA (IMAGE clone ID 7262134) was amplified by PCR with 5′-AGAATTCATGGCTACTCTGATCTATGT-3′ and 5′-CCGAATTCTCAACATCCAGGGTCGACAGA-3′; the PCR product was cut with EcoRI and ligated into pUHD-P1 to obtain FLAG-PTTG2 in pUHD-P1.
Site-directed mutagenesis was carried out with a QuikChange site-directed mutagenesis kit (Stratagene, La Jolla, CA, USA) using the following primers. PTTG1H134R, 5′-AAGAGCGTCAGATTGCGCACCTCCC-3′ and 5′-AATCTGACGCTCTTCAGGCAGGTCA-3′; PTTG1H134A, 5′-AAGAGGCCCAGATTGCGCACCTCCC-3′ and 5′-AATCTGGGCCTCTTCAGGCAGGTCA-3′; PTTG1R153G, 5′-CGAGGAGGGAGAGCTTGAAAAGCTG-3′ and 5′-AGCTCTCCCTCCTCGTCAAGGATCA-3′; DΔ version of PTTG2 (changing the D-box motif RKAL to AKAA), 5′-AGCTACCGCAAAGGCTGCGGGCACTGT-3′ and 5′-CAGTGCCCGCAGCCTTTGCGGTAGCTT-3′; PTTG2R134H, 5′-TGAAGAGCACCAGATTGCACACCTC-3′ and 5′-AATCTGGTGCTCTTCAGGCAGGTCA-3′; PTTG2G153R, 5′-AGGAGAGAGAGCTTGAAAAGCTGTT-3′ and 5′-CTCTCTCTCCTCATCAAGGATCATG-3′.
PTTG1 was amplified by PCR with a forward primer in the vector and 5′-CGAATTCAAACAGCTTTTCAAGCTC-3′; the PCR product was cut with EcoRI-BamHI and ligated into pUHD-P1 to generate FLAG-PTTG1CΔ159 in pUHD-P1. FLAG-PTTG1CΔ123 in pUHD-P1 was obtained by removing the XbaI fragment from FLAG-PTTG1 in pUHD-P1 (the product contained an extra 20-amino-acid [aa] artifact at the C terminus). FLAG-PTTG1CΔ174 in pUHD-P1 was constructed by ligating the NcoI fragment from FLAG-PTTG1 in pUHD-P1 into pUHD-P1 (the product contained an extra 32-aa artifact at the C terminus). DΔ versions of the C-terminally truncated PTTG1 mutants were generated exactly the same way by using PTTG1DΔ as the parental clone.
The FLAG-PTTG1 (or -PTTG2) was cut with EcoR I and ligated into pGEX-6P-1 (GE Healthcare, United Kingdom) to create GST-PTTG1 (or -PTTG2) in pGEX-6P-1. The EcoRI fragment of FLAG-PTTG1H134R was ligated into pGEX-6P-1 to create GST-PTTG1H134R.
PTTG1 (or PTTG2) was released with EcoRI from FLAG-PTTG1 (or -PTTG2) in pUHD-P1 and put into FLAG-3C-EGFP/PUR (a gift from Ken Ma, Hong Kong University of Science and Technology) to obtain FLAG-3C-PTTG1 (or -PTTG2)-EGFP in FLAG-3C-EGFP/PUR. The FLAG-PTTG1 (or -PTTG2) was cleaved with EcoRI and subcloned into EcoRI-cut pcDNA3.1(−) (Life Technologies) to produce PTTG1 (or PTTG2) in pcDNA3.1(−).
The HindIII fragment from FLAG-PTTG1 (or -PTTG2) in pUHD-P1 was put into HindIII-cut FLAG-PTTG2 (or -PTTG1) in pUHD-P1 to generate FLAG-PTTG2CΔ86-PTTG1NΔ87 (or FLAG-PTTG1CΔ86-PTTG2NΔ87) in pUHD-P1. Similarly, the PvuII fragment from FLAG-PTTG1 (or -PTTG2) was put into PvuII-cut FLAG-PTTG2 (or -PTTG1) in pUHD-P1 to generate FLAG-PTTG2CΔ160-PTTG1NΔ161 (or FLAG-PTTG1CΔ160-PTTG2NΔ161) in pUHD-P1.
ESP1-V5 in pEF6-V5-H6-TOPO was a gift from David Pellman (Harvard University, Cambridge, MA, USA). The cDNA was amplified by PCR with 5′-AGCCATGGATTCCAGCAAGAAGAAGCTG-3′ and 5′-TCGAATTCTTACCGCAGAGAGACAGGCA-3′; the PCR product was then cut with NcoI-EcoRI and ligated into pUHD-P1 to create FLAG-ESP1NΔ1537 in pUHD-P1. The resulting construct was cut with NheI-EcoRI and ligated into pcDNA3.1(−) (Life Technologies) to create ESP1NΔ1537 in pcDNA3.1(−). A C2029A mutation was introduced into this construct using 5′-TGCTGTTTGGCGCTAGCAGTGCGGC-3′ and 5′-GCCGCACTGCTAGCGCCAAACAGCA-3′. The XbaI fragment from ESP1 was put into XbaI-cut pUHD-P1 to create ESP1NΔ1150 in pUHD-P1. The PvuI-SphI fragment from FLAG-ESP1NΔ1537+C2029A was ligated into PvuI-SphI-cut ESP1NΔ1150 in pUHD-P1. The XbaI fragment released from the resulting construct was religated to XbaI-cut ESP1 in pEF6-V5-H6-Topo to generate catalytically inactive ESP1C2029A.
PTTG1 short hairpin RNA (shRNA) was created by annealing the primers 5′-TTTGCCAGGCACCCGTGTGGTTGTTCAAGAGACAACCACACGGGTGCCTGGTTTTT-3′ and 5′-CTAGAAAAACCAGGCACCCGTGTGGTTGTCTCTTGAACAACCACACGGGTGCCTGG-3′ into BbsI- and XbaI-cut mU6pro (28) (a gift from David Turner, University of Michigan, Ann Arbor, MI, USA). PTTG1 resistant to the shRNA was created by introducing silent mutations using the oligonucleotide 5′-GAGAACCCGGGACTAGGGTGGTTGC-3′ and its antisense oligonucleotide. Stealth siRNA targeting PTTG1 (HSS167058) and control siRNA were obtained from Life Technologies; siRNA against ESP1 (GCAGGUUCUGUUCUUGCUU) was obtained from RiboBio (Guangzhou, China).
The setup and conditions of time-lapse microscopy of living cells were as previously described (29).
Flow cytometry analysis after propidium iodide staining was performed as described previously (30).
Antibodies against cyclin B1 (22), β-actin (31), FLAG tag (32), and phospho-histone H3S10 (21) were obtained from sources described previously. Monoclonal antibodies to securin (sc-56207; Santa Cruz Biotechnology, Santa Cruz, CA, USA), ESP1 (ab16170; Abcam, Cambridge, United Kingdom), and V5 tag (R960-25; Life Technologies) were obtained from the indicated manufacturers. Immunoblotting was performed as described previously (24). Where indicated, signals on immunoblots were quantified using the Odyssey infrared imaging system (Li-Cor Biosciences) according to the manufacturer's instructions.
Immunoprecipitation was performed using 500 μg of cell lysates. The lysates were diluted to 250 μl with wash buffer (25 mM Tris-HCl, pH 7.4, 2.5 mM NaF, 130 mM NaCl, 2.5 mM EDTA, 2.5 mM EGTA, 0.05% Nonidet P-40, 2 μg/ml aprotinin, 15 μg/ml benzamidine, 1 μg/ml leupeptin, 10 μg/ml soy bean trypsin inhibitor, and 1 mM phenylmethylsulfonyl fluoride [PMSF]) and added to 15 μl of protein A-Sepharose or protein G-Sepharose prebound with 3 μl of antiserum or 1 μg of FLAG monoclonal antibody, respectively.
For other immunoprecipitation experiments, 500 μg of cell lysates was diluted to 250 μl with wash buffer and added to 10 μl of M2 beads (Sepharose beads coated with anti-FLAG monoclonal antibodies; Sigma-Aldrich, St. Louis, MO, USA). After incubation at 4°C for 120 min with end-to-end rotation, the supernatant was removed, and the beads were washed three times with wash buffer (250 μl each time). The FLAG-tagged proteins were then eluted by incubation with 100 μg/ml of FLAG peptides (Sigma-Aldrich) in 50 μl of wash buffer at 4°C for 90 min. The supernatant was dissolved in 10 μl SDS sample buffer, and 15 μl was applied to SDS-PAGE. Total cell lysates (10 μg) were applied to indicate the input.
For immunofluorescence microscopy, cells grown on poly-l-lysine-treated coverslips were fixed with freshly made 3% formaldehyde and 2% sucrose in phosphate-buffered saline (PBS) at 25°C for 5 min. The cells were then washed three times with PBS for 5 min each time, blocked and permeabilized with 3% bovine serum albumin (BSA) and 0.4% Triton X-100 in PBS at 25°C for 30 min, and washed three times with wash buffer (2% BSA and 0.2% Triton X-100 in PBS) for 5 min each time. The cells were incubated with antibodies against FLAG at 4°C for 16 h. After being washed three times with wash buffer, the cells were incubated with Alexa Fluor–594 goat anti-rabbit IgG secondary antibodies (Life Technologies) for 2 h at 25°C. After being washed three times with wash buffer, the coverslips were mounted with 2% (wt/vol) N-propylgallate in glycerol.
The autocleavage assay was performed as described previously (33) with minor modifications. Briefly, the ESP1-V5-transfected and NOC-arrested HeLa cells were lysed in lysis buffer (20 mM Tris-HCl, pH 7.7, 100 mM NaCl, 1 mM NaF, 20 mM β-glycerophosphate, 5 mM MgCl2, 0.1% Triton X-100, and 1 μM microcystin-LR). The supernatant was incubated with 15 μl of protein A-Sepharose and protein G-Sepharose mixture prebound with 1 μg V5 antibody at 4°C for 90 min. The beads were washed three times with wash buffer. For cleavage assays, 5 μl of beads from the immunoprecipitation was incubated with 1 μg of glutathione S-transferase (GST)–PTTG1 or –PTTG1H134R in 10 μl of cleavage buffer (30 mM HEPES-KOH, pH 7.7, 30% glycerol, 25 mM NaF, 25 mM KCl, 5 mM MgCl2, 1.5 mM ATP, and 1 mM EGTA) at 37°C for 1 h.
Human PTTG1 and PTTG2 are highly homologous, sharing >80% amino acid sequence identity (Fig. 1A). The major difference is at the C terminus, which is due to a frameshift caused by a base deletion in PTTG2.
We analyzed the expression of PTTG1 and PTTG2 in HeLa cells using a monoclonal antibody raised against full-length PTTG1. PTTG1 and PTTG2 are recognized equally by the antibody (indicated by experiments when equal amounts of FLAG-tagged standards were used [our unpublished data]). As indicated by the nontagged standards, PTTG1 and PTTG2 could be distinguished by their gel mobility (Fig. 1B). Although endogenous PTTG1 was readily detectable, no PTTG2 could be detected in HeLa cells. Similar results were obtained with other cancer cell lines, immortalized epithelial cells, or normal fibroblasts (see Fig. 4F) or when cells were trapped at different phases of the cell cycle (Fig. 1C). These data indicate that PTTG1 is the major isoform of securin expressed in most cell lines.
Given that PTTG2 is normally not expressed, we next investigated if ectopic expression of PTTG2 affects the cell cycle. First, we generated cell lines that expressed FLAG- and EGFP-tagged PTTG2 under the control of doxycycline. Even when expressed to a level higher than the endogenous PTTG1, PTTG2 did not significantly affect the cell cycle profiles in growing cells or cells treated with various cell cycle blockers (Fig. 1C).
To detect more detailed cell cycle changes, PTTG2-expressing cells were synchronized using a double-thymidine block method. Flow cytometry analysis indicated that PTTG2-expressing cells progressed from early S through G2 to mitosis at a rate similar to that of control cells (Fig. 1D). In agreement with this, reduction of mitotic markers, including cyclin B1 and phosphohistone H3S10, occurred at similar times in control and PTTG2-expressing cells (Fig. 1E).
Since PTTG1 is a key player in mitosis, we next analyzed the effects of PTTG2 on mitosis using time-lapse microscopy (Fig. 2A). While the duration from DNA condensation to anaphase was ~40 min in control cells, it was extended to ~70 min in PTTG1-expressing cells. Most of the extension was due to a delay from metaphase to anaphase (Fig. 2B). Mitosis in typical control and PTTG1-expressing cells is shown in Videos S1 and S2 in the supplemental material, respectively. In marked contrast, mitotic progression was not affected in PTTG2-expressing cells (Fig. 2B).
PTTG1 is targeted to degradation by ubiquitination at the end of mitosis (3). To ensure that the lack of function of PTTG2 was not due to untimely degradation, we also generated a PTTG2 mutant that lacked the D box (DΔ). We confirmed that both PTTG1DΔ and PTTG2DΔ were more stable than the respective wild-type proteins during mitotic exit (Fig. 2D). Unlike the overexpression of PTTG1, which only delayed anaphase (Fig. 2B), PTTG1DΔ interfered with sister chromatid separation and prevented anaphase onset (see below). This resulted in an accumulation of tetraploids with 4N or larger DNA contents (Fig. 2C). In contrast, PTTG2DΔ-expressing cells displayed a DNA profile identical to that of control cells. Similar results were obtained using H1299 cells, indicating that the effects of PTTG1DΔ were not merely specific to HeLa cells (Fig. 2C).
Taken together, these data indicate that the isoform PTTG2 does not affect mitosis even when expressed at levels considerably higher than those of the endogenous PTTG1.
We next addressed whether PTTG2 could interact with separase (ESP1). When PTTG2 (tagged with both FLAG and EGFP) was isolated from lysates, no ESP1 was coimmunoprecipitated (Fig. 3A). In contrast, ESP1 was able to interact robustly with FLAG-tagged PTTG1 (Fig. 3B). A key characteristic of activated human ESP1 is its autocleavage activity (12, 34, 35). Notably, both full-length ESP1 and the cleaved product (ESP1Δ) were able to bind PTTG1 (Fig. 3B). PTTG2 also did not interact with ESP1 during mitosis (Fig. 3C).
To demarcate the ESP1-interacting region in PTTG1, we constructed several C-terminal-truncation mutants (Fig. 3D). ESP1 binding was only slightly reduced when PTTG1 was deleted up to residue 174 (CΔ174) (Fig. 3E), with further impairment in binding in CΔ159 (Fig. 3F). A more extensive deletion (CΔ123), however, abolished binding to ESP1 (Fig. 3G). Note that due to the stabilization of ESP1 by PTTG1 (see below), the expression of ESP1 was higher in the presence of interacting PTTG1 constructs. Nevertheless, the large difference in binding affinity was unlikely to be explained by the ESP1 level alone.
Consistent with the binding results, expression of DΔ versions of CΔ174 and CΔ159 increased the number of cells with 4N and 8N DNA contents, suggesting interference with mitosis similar to that of PTTG1DΔ (Fig. 3H). In accordance with the lack of ESP1 binding, the cell cycle profile was not affected by CΔ123DΔ.
Collectively, these results indicate that, unlike PTTG1, PTTG2 does not physically associate with ESP1.
As there was good correlation between the expression of ESP1 and PTTG1 that could bind ESP1, we next characterized their relationship in more detail. Downregulation of endogenous PTTG1 with siRNA resulted in a concomitant reduction in endogenous ESP1 (Fig. 4A). The result was reproducible using two shRNAs against different regions of PTTG1 (Fig. 4B), supporting the specificity of the effects. Conversely, overexpression of PTTG1 increased the abundance of cotransfected ESP1 (Fig. 4C). Expression of PTTG1 but not PTTG2 in stable cell lines also increased the level of endogenous ESP1 (Fig. 4D), further supporting the idea that PTTG1 influenced the abundance of ESP1. In the reciprocal experiment, overexpression of ESP1 enhanced the expression of cotransfected PTTG1 (Fig. 4C). Finally, depletion of ESP1 with siRNA significantly reduced the abundance of endogenous PTTG1 (Fig. 4E).
We also examined the endogenous ESP1 and PTTG1 in other normal and cancer cell lines. In particular, we analyzed a number of cell lines from nasopharyngeal carcinoma versus immortalized normal nasopharyngeal epithelial cells and liver cancer cells versus immortalized normal liver epithelial cells, as well as a normal fibroblast strain (Fig. 4F). Cancer cell lines typically contained higher concentrations of ESP1 than cells derived from normal tissues. Importantly, the expression levels of PTTG1 and ESP1 were closely related in all cell lines. Similar results were obtained when only mitotic cells were analyzed, indicating that the relative expression levels of PTTG1 and ESP1 in different cell lines were not due to differences in cell cycle distributions (Fig. 4G).
Collectively, these experiments demonstrate a tight relationship between ESP1 and PTTG1. Together with the fact that ESP1 is not increased by PTTG2 and is only increased by PTTG1 constructs that can bind ESP1, these results establish that the binding of PTTG1 to ESP1 stabilizes the complex.
The above-mentioned deletion studies suggested that the C-terminal region of PTTG1 is involved in ESP1 binding (Fig. 3D). A shortcoming of deletion analysis is the likelihood of major alterations of secondary structures. Given the high homology between PTTG1 and PTTG2, we generated domain-swapping chimeras between PTTG1 and PTTG2 (Fig. 5A), which are less likely to give rise to major structural defects. The interaction of the chimeras with ESP1 was analyzed with coimmunoprecipitation (Fig. 5B). As before, blank vector served as a negative control and wild-type PTTG1 served as a positive control. These experiments revealed that binding was disrupted whenever a chimera contained residues 86 to 161 from PTTG2, indicating that the corresponding region in PTTG1 was critical for ESP1 binding.
Since ESP1 binding was lost in PTTG1CΔ123 (Fig. 3D), we examined the importance of the region between residues 123 and 161. Remarkably, only 2 residues are different between PTTG1 and PTTG2 within the region. While PTTG1 contains H134 and R153, the corresponding positions in PTTG2 are R134 and G153 (Fig. 5A). To test their importance, H134 and R153 were mutated to the corresponding sequence in PTTG2 either individually or together (Fig. 5C). Overall, PTTG1H134R associated with less ESP1 than wild-type PTTG1. More significantly, however, PTTG1H134R induced the cleavage of ESP1. In the absence of PTTG1, ESP1 was present mainly in the uncleaved, inactive form. In marked contrast, ESP1 was present mainly in the cleaved form in the presence of PTTG1H134R, suggesting that PTTG1H134R actively promotes ESP1 activity (Fig. 5C, compare lanes 1 and 4). The change in the ratio of cleaved versus full-length ESP1 was not simply caused by the stabilization of ESP1 (similar results were obtained when more ESP1 was transfected in the control than in the PTTG1H134R-cotransfected cells [data not shown]). Although ESP1 cleavage was apparent in total lysates, it was more prominent in the ESP1 binding to PTTG1H134R. ESP1 cleavage was also induced by the H134R-plus-R153G double mutant but not by R153G alone. These results indicate that mutation of H134 reduced ESP1 binding and stimulated ESP1 proteolytic activity.
To ensure that the effects of PTTG1 constructs on coexpressed ESP1 were not simply due to different transfection efficiencies, they were turned on or off using an inducible system after transfection (Fig. 6A). These experiments confirmed that while PTTG1 (but not the negative control, PTTG2) mainly stabilized full-length ESP1, PTTG1H134R increased the cleaved ESP1Δ. The double mutant containing both H134R and R153G behaved identically to the single H134R mutant (Fig. 6B). By using the CDK1 inhibitor RO3306 (Fig. 6C), thymidine (Fig. 6D), and NOC (Fig. 6C and andD),D), we also showed that PTTG1H134R induced ESP1 cleavage independently of the cell cycle.
PTTG1H134R was designed by changing H134 to the equivalent residue in PTTG2. Whether R134 is actually required for activating ESP1 was explored by mutating H134 to a different residue. Figure 6E shows that ESP1 cleavage was also induced by PTTG1H134A. These results indicate that PTTG1H134R was a loss-of-function mutation and that H134 is required to prevent ESP1 activation.
To obtain more direct evidence that PTTG1H134R could facilitate ESP1 activation, we incubated ESP1 immunoprecipitates with bacterially expressed PTTG1 in vitro (Fig. 6F). Although it was difficult to preserve uncleaved ESP1 following immunoprecipitation, we found that GST-PTTG1H134R was able to induce more efficient cleavage of ESP1 than GST-PTTG1.
To further support the idea that PTTG1 normally inhibits ESP1 autocleavage, we generated a catalytically dead version of ESP1 (C2029A). As expected, ESP1C2029A abolished autocleavage activity (Fig. 7A). In contrast to ESP1, ESP1C2029A was only marginally stabilized when coexpressed with PTTG1 or PTTG1H134R (Fig. 7B, lanes 1 to 3), confirming that the normal increase in ESP1 caused by PTTG1 was due to inhibition of autocleavage. Collectively, these data suggest that mutation of H134 attenuates the ability of PTTG1 to inhibit ESP1.
Given that PTTG1H134R can promote the activation of ESP1, we predicted that PTTG1 should contain an ESP1-activating element that is absent in PTTG2. To test this hypothesis, we generated several chimeras between PTTG1 and PTTG2 that contain the H134R mutation (Fig. 7C). Substitution of the N-terminal region of PTTG2 into PTTG1H134R did not affect the promotion of ESP1 cleavage. In contrast, substitution of the C-terminal region of PTTG2 into PTTG1H134R impaired ESP1 cleavage (see quantification of the ESP1Δ/ESP1 ratios in Fig. 7C), suggesting that activation of ESP1 may in part involve the C terminus of PTTG1.
The ESP1 binding of various PTTG1-PTTG2 chimeras was also examined using catalytically inactive ESP1C2029A to circumvent the complication of autocleavage (Fig. 7B). Compared to PTTG1, the affinity of PTTG1H134R for ESP1C2029A was significantly reduced. Although substitution of the C terminus appears to further reduce binding, this was relatively minor in comparison to the contribution from H134.
Collectively, these results suggest that the H134-containing region has a strong affinity for ESP1 and is involved in inhibiting ESP1. On the other hand, the C terminus of PTTG1 is involved in activating ESP1 and has a relatively weak binding affinity for ESP1. However, the C terminus of PTTG1 is clearly not the only region involved in activating ESP1, because putting only the C terminus of PTTG1 into PTTG2 did not allow the chimera to activate ESP1 (Fig. 5B).
Given that PTTG1H134R can promote ESP1 activation, its effects on mitosis were next examined with live-cell imaging. We used a HeLa cell line that expressed both histone H2B-GFP and an APC/C biosensor (21). The time from DNA condensation to anaphase, when the APC/C biosensor was degraded, was ~40 min in control cells (Fig. 8A; see Video S3 in the supplemental material). Overexpression of ESP1 dramatically extended the mitosis (Fig. 8B). Although metaphase plates could form, anaphase onset was inhibited, together with signs of premature loss of sister chromatid adhesion as chromosomes became more scattered at later time points. The continued presence of the APC/C biosensor indicated that the spindle assembly checkpoint remained active during the mitotic trap (Fig. 8A; see Video S4 in the supplemental material). These results indicate that when ectopically expressed, some ESP1 escaped normal control and caused premature sister chromatid separation. Notably, PTTG1H134R induced mitotic defects similar to those of ESP1 overexpression (Fig. 8B; see Video S5 in the supplemental material). As a control, wild-type PTTG1 exerted only a minor effect on mitosis (Fig. 2A; see Video S2 in the supplemental material).
Mitotic defects induced by PTTG1H134R were further investigated by its effect on cell cycle distribution. During normal mitosis, endogenous PTTG1 was destroyed by the APC/C-proteasome pathway, resulting in ESP1 activation and normal mitotic exit (Fig. 8C). However, the stable PTTG1DΔ prevented ESP1 activation (Fig. 8D) and promoted an accumulation of tetraploids (Fig. 8E). Although PTTG1H134R+DΔ also promoted tetraploidization (Fig. 8E), it was likely caused by premature ESP1 activation (Fig. 8D). To verify this hypothesis, mitotic chromosomes were examined with mitotic spread. While most control mitotic cells contained paired sister chromatids, a high percentage of ESP1- and PTTG1H134R-expressing cells contained separated sister chromatids (Fig. 8F).
Collectively, these data indicate that PTTG1H134R can induce precocious sister chromatid separation similarly to ectopic expression of ESP1 and further support the idea that PTTG1H134R could activate ESP1.
Given that H134 of PTTG1 is critical for inhibition of ESP1, we next tested if H134 is sufficient to turn PTTG2 into a functional securin. R134 and G153 of PTTG2 were mutated to the corresponding sequences in PTTG1 either individually or together. While PTTG2 did not associate with ESP1, PTTG2R134H was able to bind ESP1, albeit with a slightly lower affinity than PTTG1 (Fig. 9A). In contrast, PTTG2G153R remained incapable of binding ESP1.
We found that while PTTG2 was evenly distributed between the nucleus and cytosol, PTTG1 was more concentrated in the nucleus (Fig. 9B). It is possible that while PTTG2 could shuttle in and out of the nucleus in equilibrium, PTTG1 was retained inside the nucleus after forming a complex with ESP1. In agreement with this, PTTG2R134H was also more concentrated in the nucleus (Fig. 9B).
To validate the gain of securin function of PTTG2R134H in cells, we generated a DΔ version of PTTG2R134H. As before, expression of PTTG1DΔ but not PTTG2DΔ promoted the accumulation of tetraploids (Fig. 9C). PTTG2R134H+DΔ, however, behaved similarly to PTTG1DΔ, suggesting that the R134H mutation was sufficient to allow PTTG2 to inhibit sister chromatid separation. In support of this, live-cell imaging revealed that both PTTG1DΔ and PTTG2R134H+DΔ prevented timely anaphase onset (Fig. 9D). Although cells entered mitosis normally, they were frequently unable to undergo anaphase, returning to interphase without proper chromosome segregation or cytokinesis (examples are shown in Fig. 9E and Videos S6 and S7 in the supplemental material). The APC/C biosensor was destroyed normally, indicating that switching off the spindle assembly checkpoint and APC/C activation were not affected by PTTG2R134H+DΔ.
Taken together, these results indicate that a single-amino-acid change in PTTG2 to create H134 is sufficient to convert an inactive isoform into one that can bind ESP1 and prevent anaphase.
PTTG1 was initially identified as an oncogene in a rat pituitary tumor (4). Human PTTG1 was subsequently found to be overexpressed in pituitary and many other cancers, often associated with a poor prognosis (36). PTTG1 is also one the 17 key signature genes associated with metastasis (37). Our results also indicate that PTTG1 is expressed at a higher level in hepatocellular carcinoma and nasopharyngeal carcinoma cell lines than in normal epithelial cells from the same tissues (Fig. 4F). In contrast, the expression of PTTG2 appeared to be extremely low, if present at all (Fig. 1B and and4F).4F). Since PTTG2 is an intronless gene, it is possible that it is derived from a retroposition (38). Nevertheless, low levels of PTTG2 mRNA have been detected in some normal tissues and some cancer cell lines by reverse transcription (RT)-PCR (7, 8), but the expression of PTTG2 protein has not been reported.
PTTG1 can induce transformation in both mouse fibroblasts (39) and human embryonic kidney cells (40). Its oncogenic potential is probably based on the uncoupling of mitotic exit from ESP1 activation, resulting in aberrant sister chromatid separation and aneuploidy (41–43). Accordingly, anaphase was delayed by PTTG1 overexpression (Fig. 2A and andB),B), which was accentuated when the more stable PTTG1DΔ was used (Fig. 9D). The inhibition of sister chromatid separation led to cytokinesis failure and tetraploidization (Fig. 2C and and9D).9D). In contrast, PTTG2DΔ did not affect the timing of cell cycle progression (Fig. 1D), anaphase onset (Fig. 9D), or tetraploidization (Fig. 2C). All these data suggest that PTTG2 probably does not have the same oncogenic potential as PTTG1.
Nevertheless, it is conceivable that PTTG2 possesses functions other than acting as a securin. PTTG1 has been implicated in multiple processes, including DNA damage repair, signal transduction, and transcription. For example, PTTG1 can bind Ku, the regulatory subunit of the DNA-PK in the nonhomologous end-joining repair pathway (44). PTTG1 can also associate with p53 and block its binding to DNA (45). There is a school of thought that suggests the transcription factor function of PTTG1 contributes to its oncogenic activities (reviewed in reference 36). The transactivation domain of PTTG1 is believed to be in the acidic C-terminal region. Two PXXP motifs in this region, which are predicted SH3-interacting domains, are important to the transforming activity of PTTG1. Although both of these PXXP motifs are conserved in PTTG2, a recent study suggested that PTTG2 does not contain transactivation activity (46). Furthermore, PTTG1 enhances the invasive properties of breast cancer cells by inducing epithelial-to-mesenchymal transition and expansion of the cancer stem cell population (47). However, studies with shRNA against PTTG2 suggest that downregulation of PTTG2 stimulates epithelial-to-mesenchymal transition and apoptosis (46). Finally, there is also a possibility that PTTG2 possesses other functions not attributed to PTTG1. For example, a genome-wide RNAi screen in human cells identified PTTG2 as a regulator of the circadian clock (48).
PTTG1 affects ESP1 in several ways. (i) PTTG1 stabilizes ESP1. As demonstrated with approaches including siRNAs (Fig. 4A and andE)E) and overexpression of PTTG1 or ESP1 (Fig. 4C), PTTG1 was also reciprocally stabilized by ESP1. This probably explains the good correlation between the expression levels of PTTG1 and ESP1 in different cell lines (Fig. 4F). A simple explanation may be the inhibition of autocleavage of ESP1. As PTTG1 itself is believed to be an unfolded protein (49), the structure induced by binding to ESP1 may prevent PTTG1 from degradation by machineries that target unfolded proteins. (ii) PTTG1 inhibits ESP1 activity. Hence, nondegradable PTTG1DΔ prevented ESP1 activation during mitotic exit (Fig. 8D), and the effects of PTTG1DΔ on mitosis (Fig. 9E) were similar to those when ESP1 was deleted (19). (iii) PTTG1 activates ESP1. This was apparent in the presence of PTTG1H134R, which lost the inhibitory activity but retains the activating activity.
Our studies indicated that H134 is critically important for PTTG1 to inhibit ESP1. In support of this, mutation of H134 (to either Arg or Ala) attenuated the ability to inhibit ESP1 (Fig. 6A and andE).E). The affinity for ESP1 was also reduced considerably (Fig. 5C). This was shown more clearly when ESP1 autocleavage was abolished with a mutation that disrupted its catalytic activity (Fig. 7B). Accordingly, PTTG2, which contains R134, was unable to interact with ESP1 (Fig. 3A and andB).B). Unlike PTTG2, however, PTTG1H134R did not simply fail to interact with ESP1, because it was able to promote ESP1 activation (Fig. 5C and and6A).6A). Our conclusion is that the motif containing H134 has a strong affinity for ESP1 and is involved in inhibiting ESP1. A separate region(s), possibly involving the C terminus of PTTG1, is involved in activating ESP1 and has a weaker affinity for ESP1.
In support of this model, both deletion of the C terminus (Fig. 3D) and its replacement with PTTG2's C terminus (Fig. 7B) reduced binding to ESP1. Although the presence of PTTG2's C terminus significantly reduced ESP1 activation, it was not completely abolished (Fig. 7C), suggesting a complex picture of ESP1 activation in PTTG1. A rigorous test of the idea was the mutation of PTTG2 back to a PTTG1-like protein. Indeed, a single-amino-acid change in PTTG2 to create H134 is sufficient to promote ESP1 binding (Fig. 9A) and to prevent anaphase onset (Fig. 9D). However, although PTTG2R134H could inhibit ESP1, whether it was able to activate ESP1 is unclear. The possibility that the activation of ESP1 by PTTG1 involves only a transient interaction suggests that many ESP1 molecules could be activated by a few PTTG1 molecules. This may explain why overexpression of ESP1 alone could induce premature sister chromatid separation (Fig. 8B and andF)F) and act as an oncogene (50).
Given that PTTG1 may be involved in activating ESP1, an interesting question is whether sister chromatid separation can occur in the absence of PTTG1. The question is, of course, relevant to whether PTTG1 is essential for mitosis. In particular, our data indicate that PTTG2 should not be able to compensate for PTTG1. In fact, unlike the essential function of securin in fission yeast and Drosophila, PTTG1-deficient mice are both viable and fertile (51). Similarly, although disruption of PTTG1 in HCT116 cells results in aberrant anaphase, most of the cells eventually are able to divide and remain viable (18). This was attributed to a backup mechanism in which separase is inhibited by CDK1-dependent phosphorylation and binding to cyclin B1-CDK1 (52, 53). While this explains how separase can be inhibited before anaphase onset, it does not explain how separase is activated in the absence of PTTG1.
In fission yeast securin (Cut2p), a region containing 127DIE129 is involved in binding to separase by acting as a pseudosubstrate (54). The fact that human securin inhibits separase by preventing access of substrates to separase's active site is also consistent with securin acting as a pseudosubstrate (12). Whether the PTTG1 domain containing H134, identified in this study as essential for binding to separase, acts as a pseudosubstrate is not immediately obvious but is an intriguing possibility to explore.
The structure of the securin-separase complex has so far evaded elucidation. Part of the problem is that PTTG1 is a natively unfolded protein (49). Nevertheless, protonless nuclear magnetic resonance (NMR) analysis indicates that PTTG1 contains some transient local structural elements (55). In particular, the fragment D150-F159 appears to be important for separase association. Our data were consistent with this, because while PTTG1CΔ159 could still associate with ESP1, further deletion abolished the interaction (Fig. 3E to toG).G). The finding of the importance of H134 further refines the interacting region and should help future structural analysis. We further noted that H134 is also within the region that binds PBF (56), encoded by a potential proto-oncogene that is upregulated in several cancers (reviewed in reference 36).
In conclusion, these results illustrate an interesting example of the loss of function of an isoform that probably is not expressed in vivo. Despite the similarity between the two isoforms, PTTG2 is unable to interact with ESP1. The differences between the isoforms led us to pinpoint a residue that is critical for the inhibition of ESP1 by PTTG1 that is distinct from the region required for ESP1 activation.
We thank Ken Ma for the APC/C reporter cell line, as well as help with preparing lysates from different cell lines, and Wing Yu Man for technical assistance.
This work was supported in part by Research Grants Council grants AOE-MG/M-08/06 and HKU7/CRG/09.
We declare that we have no conflict of interest.
Published ahead of print 24 June 2013
Supplemental material for this article may be found at http://dx.doi.org/10.1128/MCB.00057-13.