Inhibition of TRIP1/S8/hSug1, a component of the human 19S proteasome, enhances mitotic apoptosis induced by spindle poisons

Mitotic spindle poisons (e.g., Taxol and vinblastine), used as chemotherapy drugs, inhibit mitotic spindle function, activate the mitotic spindle checkpoint, arrest cells in mitosis, and then cause cell death by mechanisms that are poorly understood. By expression cloning, we identified a truncated version of human TRIP1 (also known as S8, hSug1), an AAA (ATPases associated with diverse cellular activities) family ATPase subunit of the 19S proteasome regulatory complex, as an enhancer of spindle poison–mediated apoptosis. Stable expression of the truncated TRIP1/S8/hSug1 in HeLa cells [OP-TRIP1(88-406)] resulted in a decrease of measurable cellular proteasome activity, indicating that OP-TRIP1(88-406) had a dominant-negative effect on proteasome function. OP-TRIP1(88-406) revealed an increased apoptotic response after treatment with spindle poisons or with proteasome inhibitors. The increased apoptosis coincided with a significant decrease in expression of BubR1, a kinase required for activation and maintenance of the mitotic spindle checkpoint in response to treatment with spindle poisons. Small interfering RNA (siRNA)–mediated knockdown of TRIP1/S8/hSug1 resulted in a reduction of general proteasome activity and an increase in mitotic index. The siRNA treatment also caused increased cell death after spindle poison treatment. These results indicate that inhibition of TRIP1/S8/hSug1 function by expression of a truncated version of the protein or by siRNA-mediated suppression enhances cell death in response to spindle poison treatment. Current proteasome inhibitor drugs in trial as anticancer agents target elements of the 20S catalytic subcomplex. Our results suggest that targeting the ATPase subunits in 19S regulatory complex in the proteasome may enhance the antitumor effects of spindle poisons. [Mol Cancer Ther 2006;5(1):29–38]

Spindle poisons (e.g., Taxol and vinblastine) are commonly used chemotherapy drugs (1–3). In clinically relevant doses (e.g., 5–200 nmol/L in Taxol; ref. 4), they inhibit mitotic spindle function and activate the mitotic spindle checkpoint (5–7). The spindle checkpoint causes extended mitotic arrest through inhibition of a ubiquitin ligase complex called the anaphase-promoting complex/cyclosome and its activator Cdc20. In some cases, mitotic arrest results in cell death initiated during mitosis (mitotic apoptosis) or apoptosis observed after the cells exit mitosis abnormally without normal chromosome segregation (sometimes called adaptation or mitotic slippage; ref. 4). The signal transduction pathways by which spindle poisons and other mitotic inhibitors lead to cell death remain to be clarified (8). A few molecules have been identified to affect spindle poison–mediated cell killing. Mitotic apoptosis is observed on down-regulation of certain kinetochore components, such as Ndc80/Hec1 or Nuf2 by small interfering RNA (siRNA) or conditional promoter shutoff (9, 10). Taxol treatment activates p38 mitogen-activated protein kinase, and suppression of p38 by specific inhibitors suppresses Taxol-mediated cell death (11, 12). Spindle checkpoint components Bub1 and BubR1 kinases, if overexpressed, stimulate the apoptotic response (13). The hBubR1 protein is reduced during extended spindle poison–mediated mitotic arrest at least in part due to a proteasome-dependent degradation, and this reduction has been proposed to be part of the link between spindle checkpoint and induction of apoptosis (13). Postmitotic apoptosis is observed if the spindle checkpoint is compromised by repression of Mad2 or BubR1 with siRNA (14, 15) or expression of a dominant-negative form of the Cdc20 protein (16). Breast cancer cell lines SKBr3 and HCC-1433 and ovarian cancer cell lines A2780 and OVCAR have weakened spindle checkpoint function due to decreased expression of BubR1 and show elevated sensitivity to spindle poisons (17).

The proteasome, a large protease complex that degrades polyubiquitylated cellular proteins, has recently gained prominence as a potential target for cancer therapy (18–21). Proteasome inhibitors (e.g., bortezomib/Velcade, lactacystin, and MG132; refs. 18–21) are cytotoxic, but the precise mechanism of cell killing remains unclear. Bortezomib/Velcade has shown promise for a variety of cancers, including multiple myeloma (19). Regulation of proteolysis is crucial in cellular growth control for normal cells. Inappropriate accumulation or reduction of cell cycle regulators have been linked to oncogenesis (22), and regulated proteolysis plays a major role in maintaining normal levels of proteins. A large percentage of regulated proteolysis is carried out by ubiquitin-mediated targeting (23–26). The ubiquitin-mediated proteolysis system requires a set of enzymes: an ubiquitin-activating enzyme (E1), several ubiquitin-conjugating enzymes (E2), and a large variety of ubiquitin ligases (E3). These enzymes covalently attach multiple copies of the ubiquitin to the target. The resulting polyubiquitin chain on the target is recognized by the 26S proteasome. The 26S proteasome is a complex of two subcomplexes: a 19S regulatory complex and a 20S catalytic complex. Structural studies show that the barrel-shaped 20S catalytic complex is capped by the 19S regulatory complex(es) at one or both ends to form the 26S complex (18, 27). The 19S complex is believed to bind to, refold, and transfer the polyubiquitylated target protein into central cavity of the 20S catalytic core, where the target protein is degraded by the protease activity. Consistent with the chaperone-like activity required, the 19S regulatory complex contains six ATPase subunits. In yeast, conditional mutants in different proteasome subunits show a mitotic arrest phenotype (28). This implies that the proteasome acts as whole and each component is required for activity. It also suggests that mitosis is a particularly sensitive target when proteasome activity is compromised.

We set out to identify factors affecting spindle poison–mediated cell killing and developed a mammalian gene cloning protocol. One of the candidate plasmids, pSC3, encodes a portion of TRIP1/S8/hSug1, an ATPase subunit of the 19S proteasome. When stably integrated, expression of the truncated TRIP1/S8 showed no effect on normal cell growth but enhanced spindle poison–mediated cell death. Similarly, repression of endogenous TRIP1/S8 by siRNA also results in increased cell death in response to spindle poisons. Thus, TRIP1/S8 seems to be a promising target to enhance spindle poison–mediated cell killing and may define an additional class of targets to inhibit proteasome function.

We transfected COS7 cells (in two 15-cm plates, ∼70% confluent) with a human testis cDNA library (Clontech, Palo Alto, CA). Twenty-four hours later, the cells were extensively washed to remove untransfected DNA and dead cells and were treated with nocodazole (100 ng/mL). Rounded mitotic cells were shaken from the plate and collected every hour, up to hour 14, and transferred to new plates containing nocodazole (100 ng/mL). At 16 hours after the beginning of nocodazole treatment, the new plates were rinsed with PBS to remove any still rounded mitotic cells, and the cells that had altered their morphology and became attached in the presence of nocodazole were retained. DNA from the attached cells was recovered with DNAzol reagent used according to the manufacturer's instructions (Invitrogen, Carlsbad, CA). To avoid integration of the plasmids into the COS7 cell genome, we collected plasmid DNA no longer than 40 hours after transfection. Recovered DNA was transformed into UltraMax DH5 competent bacteria (Invitrogen). Plasmid recovered from the bacteria was used for an additional round of transfection of COS7 cells to enrich plasmids with positive activity. After the second round, we isolated individual plasmids from bacterial colonies and sequenced them.

We cultured and prepared cells as described previously (29). The fixed samples were analyzed with a Zeiss Axioplan IIi microscope equipped with a Hamamatsu Orca II camera and Metamorph imaging system (Universal Imaging Corp., Downingtown, PA). For live cell observation, a planapochromat ×60 (numerical aperture 1.4) objective (Nikon USA, Melville, NY) was used with a SenSys CCD camera (Photometrics Ltd., Tucson, AZ) connected to a Nikon Diaphot microscope and imaged with Metamorph software. We used Annexin V FLUOS (Roche Biochemicals, Indianapolis, IN) to stain phosphatidylserine exposed on the cell surface to identify apoptotic cells following the manufacturer's instructions. Propidium iodide (PI)–positive (necrotic) cells were not scored as apoptotic. We also used Live/Dead Cell Death Assay kit (Molecular Probes, Eugene, OR) to assess cell death using different markers.

The plasmid vector for the cDNA library, pEXP1 (Clontech), contains the cytomegalovirus promoter upstream of the cloning site for the library cDNA and an internal ribosome entry site that permits the cotranslation of a puromycin resistance gene. Thus, if integrants are established, nearly all puromycin-positive cells will stably express the gene of interest. We transfected HeLa cells with individual candidate plasmids and selected cells with puromycin (0.2–0.5 μg/mL) for 2 to 3 weeks. For each plasmid, some 200 to 500 surviving colonies were pooled and used for experiments to avoid clonal variation in expression of the integrant and to avoid indirect effects of stable cell generation, such as mutations within the parental genome induced by integration of the plasmid. For full-length TRIP1/S8 integrant generation, we used pCMS-EGFP vector (Clontech) and G418 (250 μg/mL) for selection.

We treated control HeLa cells and OP-TRIP1(88-406) cells with nocodazole (100 ng/mL) for 0, 16, or 24 hours. We suspended the cells by trypsinization and fixed them with 80% ethanol (−20°C) for at least 2 hours. The permeabilized cells were rehydrated for 5 to 10 minutes and resuspended in PBS, blocked with 20% boiled normal goat serum in PBS, and incubated with rabbit anti-p85 poly(ADP-ribose) polymerase (PARP) fragment antibody (1:200; Promega, Madison, WI) in 5% goat serum for 1.5 to 3 hours. After being rinsed with PBS twice, samples were incubated with secondary antibody (FITC anti-rabbit, 1:400) for at least 1 hour. Samples were then rinsed and treated with RNase (0.1 mg/mL), PI (50 μg/mL), and 0.1% Triton X-100 at room temperature for at least 3 hours. The samples were analyzed with a FACSCalibur flow cytometers, and the cell cycle profile was estimated by ModFit software with the aid of the Flow and Image Cytometry Laboratory (University of Oklahoma Health Sciences Center).

We plated ∼500 cells in 12-well plates, 1,000 cells in 60-mm plates, or 3,000 cells in 10-cm plates. The next day (day 0), we added drugs in different concentrations and incubated the plates at 37°C. On day 4, half of the medium was replaced and fresh drugs were added. The cells were fixed and stained on day 8 with 0.5% methylene blue in 50% ethanol for 20 minutes, rinsed with distilled water, and dried (30). Assays were repeated at least thrice and typical results are shown as pictures. Cell proliferation was quantified by imaging plates and then summing the intensities of stained colonies using Metamorph software.

Sample cells were directly extracted or harvested and frozen at −80°C until use. Extracts were prepared by vortexing cells in low-salt buffer [20 mmol/L Tris-HCl (pH 7.0), 5 mmol/L ATP, 1 mmol/L DTT, 0.1 mmol/L EDTA, and 20% glycerol supplemented with 400 nmol/L microcystin LR and protease inhibitor cocktail (Sigma, St. Louis, MO)] and were cleared by centrifugation at 15,000 × g for 15 minutes at 4°C. The sample protein concentration was adjusted to 150 μg/mL. Samples were incubated at 37°C for 30 or 60 minutes with the fluorogenic proteasome substrate III (SucLLVY-AMC; Calbiochem, La Jolla, CA) at 250 nm. Reactions were terminated by the addition of 100 μL ethanol to 50 μL of the reaction mixture. Fluorescence was measured in a 96-well plate with a plate reader (Tecan, Maennedorf, Switzerland; excitation, 360 nm; emission, 465 nm). Samples added MG132 (10 μmol/L) were used for negative controls.

We used chemically modified siRNA duplex (Stealth siRNA, Invitrogen) to knockdown human TRIP1/S8 targeting 5′-GCTCATCATACGGACTGTACCTTTA-3′. As a negative control, siRNA for green fluorescent protein was used. Transfection was done with Oligofectamine reagent (Invitrogen) following the manufacturer's instructions.

Cells were extracted either in radioimmunoprecipitation assay buffer [1% (w/w) NP40, 1% (w/v) sodium deoxycholate, 0.1% SDS, 1% Triton X-100, 10 mmol/L sodium phosphate (pH 7.2), 2 mmol/L EDTA, 150 mmol/L NaCl, 50 mmol/L NaF, and 0.2 mmol/L sodium vanadate] supplemented with 10 μmol/L MG132, 5 μg/mL protease inhibitor cocktail, and 400 nmol/L microcystin LR, or in NP40 lysis buffer [50 mmol/L Tris-HCl (pH 7.4), 1% (w/w) NP40, 250 mmol/L NaCl, 10 mmol/L NaF, and 5 mmol/L EDTA] supplemented with 10 μmol/L MG132, 5 μg/mL protease inhibitor cocktail, and 400 nmol/L microcystin LR and boiled with SDS loading buffer for 6 minutes. Protein concentration was measured with the BCA protein assay kit (Pierce, Rockford, IL) and protein amounts loaded on the gels were equalized. We used rabbit anti-BubR1 (gift from Dr. T. Stukenberg, University of Virginia, Charlottesville, VA), mouse anti–cyclin B (BD Transduction Laboratories, San Jose, CA), rabbit anti-PARP p85 fragment, mouse anti-Cdc27 (BD Transduction Laboratories), rabbit anti-TRIP1/proteasome S8 subunit (Calbiochem), mouse anti-Bub3 (BD Transduction Laboratories), rabbit anti-phosphorylated histone H3 (Upstate Biotechnology, Lake Placid, NY), and mouse anti-β-tubulin (Amersham Biosciences, Piscataway, NJ).

We noted that mitotic cells undergoing apoptosis exhibited increased adherence to the culture substratum compared with healthy mitotic cells. This allowed us to select for cells containing plasmids whose expression caused increased apoptosis in cells treated with spindle poisons. After screening 3 million cDNAs, we obtained 34 candidate plasmids, among which 7 cDNA fragments showed enhanced spindle poison–mediated cell killing when overexpressed in HeLa cells. For control purposes, we established 48 clones from the cDNA library without selection. None among the 48 colonies showed elevated sensitivity to spindle poisons.

One of the positive clones, pSC3, encoded a proteasome subunit TRIP1/S8/hSug1 (TRIP1/S8, NM_002805) with 87 amino acids truncated from its NH₂ terminus [TRIP1(48-406); Fig. 1A]. The TRIP1/S8 protein is an AAA (ATPases associated with a variety of cellular activities) family ATPase subunit in the 19S regulatory complex of the 26S proteasome (31–33). A subpopulation of TRIP1/S8 is found as a part of the APIS (AAA proteins independent of 20S) complex, which may play a role in transcriptional regulation independent of the full proteasome (34, 35).

To confirm the effect of the cDNA expression on spindle poison–mediated cell killing, we generated a HeLa-based cell line that stably expresses TRIP1(88-406) [OP-TRIP1(88-406)]. The messenger overexpression was verified by quantitative PCR (Supplementary Fig. S1).¹

In standard culture conditions, the estimated doubling time was 18.1 ± 0.7 hours for OP-TRIP1(88-406) and 18.1 ± 1.9 hours for control parental HeLa cells; thus, growth rates were comparable. Fluorescence-activated cell sorting analysis showed no apparent difference in the cell cycle profiles comparing cycling HeLa cells and OP-TRIP1(88-406) cells.² Annexin V labeling indicated no difference in spontaneous apoptotic death rate in standard culture conditions (<2.5%). However, in colony growth assays, OP-TRIP1(88-406) cells were significantly more sensitive than controls to the microtubule-depolymerizing drug nocodazole and the microtubule-stabilizing drug Taxol (Fig. 1B). OP-TRIP1(88-406) cells did not show heightened sensitivity to the topoisomerase II inhibitor VM26 (teniposide) 4′-dimethyl-epipodophyllotoxin or to the DNA-damaging drug bleomycin (data not shown).

The construct isolated with our screen encodes a protein fragment that lacks the NH₂-terminal 87 amino acids of TRIP1/S8. We speculated that it may function in a dominant-negative manner and interfere with normal proteasome activity. To test this hypothesis, we compared proteasome activities in extracts from control cells and OP-TRIP1(88-406) cells using the fluorogenic proteasome substrate (SucLLVY-AMC; Fig. 2A; refs. 36, 37). OP-TRIP1(88-406) cells showed a 34% reduction in proteasome activity compared with control HeLa cells in asynchronous culture and a 20% reduction in cells arrested with nocodazole. These results suggest that proteasome function is partially compromised in OP-TRIP1(88-406) cells. Proteasome activity is required at several stages of the cell cycle (e.g., to degrade mitotic cyclins and allow cell cycle progression). We reasoned that lower proteasome activity associated with expression of the TRIP1/S8 fragment might lead to increased sensitivity to challenge by sublethal concentrations of proteasome inhibitor in a growth assay. We treated OP-TRIP1(88-406) with MG132 or ALLN, drugs that inhibit proteasome function in assays in cultured cells (18). OP-TRIP1(88-406) cells showed reduced cell proliferation when treated with low concentrations of MG132 or ALLN (Fig. 2B). To distinguish if the truncated protein was functioning as a dominant negative, we generated a HeLa-based cell line that overexpressed the full-length TRIP1/S8 protein (Supplementary Fig. S2).¹ We observed an increase in proteasome activity in the extract, but the sensitivity to spindle poisons showed little difference from control. This supports that the truncated protein is dominant-negative form.

To study further the spindle poison sensitivity of OP-TRIP1(88-406) cells, we observed the cellular responses to spindle poison treatment over shorter time ranges by live cell microscopy. Used at moderate doses (25–200 ng/mL), nocodazole causes mitotic arrest of most cultured cells within one cell cycle without immediate cytotoxicity. Normally, we and other researchers use nocodazole at 100 ng/mL to accumulate living cells in M phase. We reasoned that direct microscopic observation of the effects of expression of TRIP1(88-406) would provide information about the mechanism of enhanced toxicity to nocodazole. We synchronized HeLa and OP-TRIP1(88-406) cells in early S phase with a double block in aphidicolin (an inhibitor of DNA polymerase), released the cells into medium containing nocodazole, and examined them by phase-contrast microscopy (Fig. 3A and B). Control HeLa cells entered mitosis ∼6 hours after release and remained arrested in mitosis. After 14 hours, a small proportion of control cells began to exhibit membrane blebbing, characteristic of apoptosis during mitotic arrest. OP-TRIP1(88-406) cells entered mitosis ∼6 to 8 hours after release from S-phase arrest and remained arrested in mitosis. However, a larger proportion of OP-TRIP1(88-406) cells became apoptotic soon after mitotic arrest. Apoptotic cells were counted and the apoptotic percentage in the total cells is plotted as the dark gray regions in Fig. 3A. Light gray regions represent percentage of normal mitotic cells. Frames from a time-lapse video of the apoptotic phenotype of OP-TRIP1(88-406) cells are shown in Fig. 3B, revealing conversion of the normal rounded mitotic cells in the first panel to the advanced apoptotic cells in the last panel. We also examined whether the membrane blebbing phenotype correlated with apoptosis. About 50% of the blebbing cells were Annexin V positive compared with <3% of cells with smooth membranes, suggesting that the membrane blebbing is an early sign of apoptosis, preceding cell surface exposure of phosphatidylserine detected by Annexin V. In all, these experiments indicate that expression of TRIP1(88-406) facilitates mitotic apoptosis under spindle poison challenge.

Both compromised spindle checkpoint function and elevated apoptosis in response to challenge with spindle poisons have been associated with deregulation of the expression of spindle checkpoint components (13–17, 38). We tested nocodazole-treated HeLa and OP-TRIP1(88-406) cells to compare whether mitotic and/or apoptotic marker proteins behave differently and whether the levels of expression of spindle checkpoint proteins are affected. We treated cells with nocodazole (100 ng/mL) for 4 hours, collected the arrested mitotic cells, incubated the mitotic cells further with nocodazole for the indicated amount of time up to 20 hours, and prepared samples for immunoblotting (Fig. 3C). Cyclin B levels remained high in both control and OP-TRIP1(88-406), suggesting that the cells remained arrested in mitosis with high Cdk1/cyclin B kinase activity. The PARP cleavage fragment, a marker of apoptosis generated by caspase-3 and caspase-7, appeared after 16 hours of nocodazole treatment in control HeLa cells. In contrast, in OP-TRIP1(88-406) cells, the PARP fragment was apparent at 4 hours and increased with time. Cdc27 is a component of the anaphase-promoting complex/cyclosome. In mitosis, Cdc27 is multiply phosphorylated and undergoes a large mobility shift on SDS-PAGE gels. Cdc27 is also a caspase-3-like protease target and is degraded during apoptosis (39). With time in nocodazole, the mitotic hyperphosphorylation of Cdc27 remained high in control HeLa cells, whereas in OP-TRIP1(88-406) both phosphorylation and total amount of Cdc27 protein decreased, consistent with increased apoptosis in OP-TRIP1(88-406) cells. Expression of the spindle checkpoint component BubR1 in OP-TRIP1(88-406) was of particular interest because the level of this protein has been proposed to serve as a link between spindle checkpoint and spindle poison–mediated cell killing (13). The BubR1 expression level was comparable with or slightly higher than that of control cells at 4 hours. However, with time, BubR1 levels diminished more rapidly in the OP-TRIP1(88-406) cells.

We interpreted the results above as indicating that the OP-TRIP1(88-406) cells initiate apoptosis directly from mitosis after spindle poison treatment. To verify this interpretation, we asked in which phases of the cell cycle phase PARP fragments, an apoptotic marker, were generated in response to spindle poison. We collected control and OP-TRIP1(88-406) cells with or without nocodazole treatment, labeled PARP fragment–positive cells by immunofluorescence, and monitored the cell cycle by fluorescence-activated cell sorting (Fig. 3D). Without nocodazole, the cell cycle profiles of control and OP-TRIP1(88-406) were indistinguishable, and a small number of apoptotic cells (PARP fragment–positive cells) were observed preferentially in the population with sub-G₁ DNA content in both cell lines. With nocodazole treatment, both cell types arrested in G₂-M. The degree of G₂-M arrest seemed to be higher in OP-TRIP1(88-406) cells (G₂-M: 59% after 16 hours in nocodazole) than in controls (G₂-M: 40% after 16 hours in nocodazole). Monitoring nocodazole-treated, PARP-positive cells, we observed two peaks in sub-G₁ and G₂-M in both cell lines, indicating a population of PARP fragment–positive cells has G₂-M DNA content (asterisks on right). The ratio of PARP-positive G₂-M cells is consistently higher in OP-TRIP1(88-406) cells than in controls as indicated in Fig. 3D (inset). Together, the results in Fig. 3 suggest that that cells expressing truncated TRIP1 are arrested in mitosis more readily than controls in response to spindle poisons, and cells expressing truncated TRIP1 are more prone to apoptosis during mitosis than are control cells.

The identification of a truncated proteasome subunit as a spindle poison–sensitizing factor and the observed decrease in proteasome activity in OP-TRIP1(88-406) (Fig. 2A) suggested that inhibition of proteasome activity might generally enhance the cytotoxicity induced by spindle poisons. To test this idea using an independent approach, we treated asynchronous HeLa cell cultures with nocodazole or Taxol for 16 hours with or without cotreatment with the proteasome inhibitor, MG132. The percentage of dead cells was assessed microscopically using the Live/Dead Cell Death Assay kit, which determines the fraction of cells with disrupted plasma membranes. A combination of MG132 with either nocodazole or Taxol showed higher cytotoxicity than any of the drugs alone. The combination of MG132 with nocodazole seemed to show synergistic effects, whereas the combination of MG132 and Taxol was additive (Fig. 4A). To examine the effects on a biochemical marker of apoptosis, we collected samples and monitored generation of the PARP cleavage fragment by immunoblot (Fig. 4B). When used alone, each drug induced low-level production of the PARP fragment. However, when cells were incubated in either nocodazole or Taxol, cotreatment with MG132 showed much higher generation of the PARP fragment.

Initiation of cell death during mitosis with perturbed proteasome function led us to investigate whether the modulatory effect of proteasome inhibition on the cytotoxic response to spindle poisons occurred during M phase or during another phase of the cell cycle. We tested whether proteasome inhibition would increase apoptosis in cells that were already in M phase. We treated HeLa cell cultures for 4 hours with nocodazole (100 ng/mL) or Taxol (1 μmol/L) and collected the arrested mitotic cells. We then treated each population for an additional 4 or 8 hours with spindle poisons in the presence or absence of MG132 (10 μmol/L). After this time, we collected samples and monitored PARP cleavage fragment production by immunoblotting (Fig. 4C). Surprisingly, treatment with MG132 did not substantially increase PARP fragment production in cells that were already arrested in M phase with both spindle poisons. Thus, sequential inactivation of the proteasome after mitotic arrest did not seem to enhance mitotic apoptosis. This finding suggests that the target of proteasome inhibition may function before M phase.

The studies above led us to investigate further the function of endogenous TRIP1/S8 and cellular responses to spindle poisons by siRNA-mediated inhibition of TRIP1/S8. We transfected siRNA into HeLa cells, and 40 hours later, we treated the cells with normal medium or with medium containing nocodazole (100 ng/mL) for a further 8 hours. We then observed the effects by phase-contrast microscopy (Fig. 5A). The results of these treatments were quantified in Fig. 5B. In the absence of nocodazole, cells treated with TRIP1/S8 siRNA were viable but exhibited a higher mitotic index and a higher percentage of Annexin V–positive (apoptotic) cells. The number of PI-positive cells (necrotic or terminal stage in apoptosis) was approximately equal to that of the control. Nocodazole treatment led to an increase in the number of mitotic cells and Annexin V–positive cells, the normal response to spindle poison. When TRIP1/S8 siRNA-treated cells were incubated in nocodazole, the number of cells that were Annexin V or PI positive increased, whereas the number of healthy mitotic cells was decreased. We interpret this reduction of healthy mitotic cells as the result of increased apoptosis initiated during mitotic arrest.

We prepared extracts from duplicate cultures and monitored the amount for TRIP1/S8, PARP fragment, Cdc27, phosphorylated histone H3, BubR1, and β-tubulin (Fig. 5C). The siRNA treatment produced a 50% to 70% reduction of TRIP1/S8 protein expression in both cycling cells and cells incubated with nocodazole. In siRNA-treated cells incubated in nocodazole, we detected enhanced production of the PARP fragment and diminished levels and phosphorylation of the Cdc27 protein. However, we did not observe significant differences in BubR1 expression.

To test whether TRIP1/S8 knockdown resulted in reduction of general proteasome activity, we transfected HeLa cells with TRIP1/S8 siRNA, prepared cell extracts, and measured proteasome activity (Fig. 5D). We observed a 60% reduction of proteasome activity in extracts with from TRIP1/S8 siRNA-treated cultures (Fig. 5D, gray column) compared with that in extracts from control siRNA transfections (black column). We also obtained the same reduction in proteasome activity using the 293T cell line (data not shown). Thus, repression of TRIP1/S8 leads to proteasome inhibition.

We did cell cycle analysis using the TRIP1 knockdown cells along with controls (Supplementary Fig. S3).¹ Although the siRNA transfection procedure produced higher numbers of apoptotic cells in the subG1 DNA content region, we observed that TRIP1 knockdown cells also showed a population of apoptotic cells with a G₂-M DNA content when treated with spindle poisons, particularly nocodazole. However, the level of G₂-M apoptotic cells in the siRNA-treated cell population was less than that obtained from expression of the truncated TRIP1 protein shown in Fig. 3D.

Mammalian cancer cells treated with clinically relevant doses of spindle poison eventually undergo apoptosis either directly from mitosis (mitotic apoptosis) or after an abnormal mitotic exit. Given the current routine use of spindle poisons (e.g., Taxol/paclitaxel and vinblastine) for cancer chemotherapy, identification of cellular factors that modulate the cell death response is critical in elucidating the mechanisms involved.

We report the identification of a cDNA fragment, pSC3, expressing a NH₂-terminal truncation of the 19S proteasome component TRIP1/S8 from an expression screen aimed at isolating enhancers of spindle poison–mediated cell killing. Expression of truncated TRIP1/S8 protein seems to work as a dominant negative for proteasome function. Decreased proteasome activity sensitizes HeLa cells to spindle poisons and proteasome inhibitors in growth assays. Microscopic analysis indicated that it caused increased mitotic apoptosis. siRNA-mediated knockdown of TRIP1/S8 also resulted in compromised proteasome function and enhanced cell death with spindle poisons. In contrast, overproduction of full-length TRIP1/S8 resulted in an increase in proteasome activity and in subtle resistance to spindle poisons (Supplementary Fig. S2).¹ These results suggest that TRIP1/S8 function is linked to proteasome activity and that proteasome activity plays critical role in spindle poison–mediated cell death. These findings also suggest that the TRIP1/S8 may be valid target for enhancing spindle poison–mediated cell killing in cancer therapy.

It seems that in OP-TRIP1(88-406) cells spindle poison–mediated cell death is initiated from mitotic arrest rather than after the exit of mitosis. The simultaneous finding of high cyclin B accumulation, high Cdc27 phosphorylation, and appearance of the PARP apoptosis marker fragment in OP-TRIP1(88-406) cells after only 4 hours of nocodazole treatment (Fig. 3C) is consistent with this interpretation. Cell cycle analysis of PARP fragment–positive cells also supports this view (Fig. 3D). Because BubR1 reduction seems to take place after initiation of apoptosis, it is unlikely to be the direct cause of apoptosis, although the reduction may accelerate apoptosis. This conclusion is supported by results from TRIP1/S8 siRNA-treated cells, in which BubR1 amount was not significantly affected when apoptosis occurred (Fig. 5B).

Our results are notable given recent interest in the proteasome and proteasome inhibitors in cancer chemotherapy (e.g., refs. 18–21). Our identification of a proteasome subunit in this screen suggests that the proteasome may be a promising target in multidrug strategies with spindle poisons. Consistent with this idea, we also show that simultaneous treatment of cultured cells with spindle poisons and a proteasome inhibitor caused enhanced cell death (Fig. 4A and B). Studies by others indicated that bortezomib (also known as Velcade or PS341), a proteasome inhibitor with clinical potential (18–21), enhances the cytotoxic activity of spindle poisons docetaxel (40) and Taxol (41) against tumor xenografts. The precise molecular mechanisms by which inhibition of the proteasome affects mitotic apoptosis and the reaction to spindle poisons remain uncertain. Based on our evidence, we suggest that cells with deficient proteasome activity may enter mitosis inadequately prepared perhaps through insufficient removal of a cell cycle inhibitor. This inappropriate state may trigger an apoptotic pathway. Alternatively, recent study indicates that the proteasome plays a key role in transcriptional regulation (42). Indeed, some evidence implicates TRIP1/S8 in direct regulation of transcription either within or apart from its role in the 19S proteasome (34, 35). Transcriptional errors may result in altered expression of proteins that play important roles in mitosis, and it may leave cells more prone to apoptosis during mitosis. This interpretation is consistent with our analysis that the window for enhancing mitotic apoptosis of spindle poisons with proteasome inhibitor is before not during M phase (Fig. 4C). Another but not mutually exclusive possibility is that proteasome inhibition alters the state of apoptotic machinery and makes cells prone to apoptosis. Further investigation will be required to test these possibilities.

Overall, our results suggest that inhibition of TRIP1/S8 and/or the proteasome alters cellular physiology and leaves cells prone to apoptosis when they are arrested in mitosis with spindle poisons. Our results suggest that targeting TRIP1/S8 or other components of the 19S proteasome may be useful in anticancer therapy either alone or in combination with spindle poisons. Known proteasome inhibitors [e.g., peptide boronate (bortezomib/Velcade), lactacystin, and peptide aldehyde (MG132 and MG115)] target catalytic residues for proteolysis in 20S proteasome catalytic core particle (18). TRIP1/S8, and possibly other ATPase subunits in 19S regulatory complex, may be useful additional targets for inhibition of proteasome activity.

We thank Drs. M. Kallio (VTT Biotechnology, Turku, Finland) and T. Stukenberg (University of Virginia) for discussion of this work, T. Stukenberg for providing BubR1 antibody, Dr. J. Henthorn (University of Oklahoma Health Sciences Center) for his assistance in fluorescence-activated cell sorting analysis, and L. Ahonen, J. Daum, J. Hudson, T. Jones, W. Martin, B. Pittman, T. Potapova, and V. Vorozhko for support in the laboratory.