The ubiquitin-proteasome system (UPS) mediates targeted protein degradation. Notably, the UPS determines levels of key checkpoint proteins controlling apoptosis and proliferation by controlling protein half-life. Herein, we show that ovarian carcinoma manifests an overstressed UPS by comparison with normal tissues by accumulation of ubiquitinated proteins despite elevated proteasome levels. Elevated levels of total ubiquitinated proteins and 19S and 20S proteasome subunits are evident in both low-grade and high-grade ovarian carcinoma tissues relative to benign ovarian tumors and in ovarian carcinoma cell lines relative to immortalized surface epithelium. We find that ovarian carcinoma cell lines exhibit greater sensitivity to apoptosis in response to proteasome inhibitors than immortalized ovarian surface epithelial cells. This sensitivity correlates with increased cellular proliferation rate and UPS stress rather than absolute proteasome levels. Proteasomal inhibition in vitro induces cell cycle arrest and the accumulation of p21 and p27 and triggers apoptosis via activation of caspase-3. Furthermore, treatment with the licensed proteasome inhibitor PS-341 slows the growth of ES-2 ovarian carcinoma xenograft in immunodeficient mice. In sum, elevated proliferation and metabolic rate resulting from malignant transformation of the epithelium stresses the UPS and renders ovarian carcinoma more sensitive to apoptosis in response to proteasomal inhibition. (Cancer Res 2006; 66(7): 3754-63)
- ovarian cancer
- proteasome inhibitor
The ubiquitin-proteasome system (UPS) is responsible for >80% of the intracellular protein degradation in eukaryotes and has two components: the ubiquitin-conjugating system and the 26S proteasome ( 1). The UPS catabolic machinery is located in the 26S proteasome, which consists of a 20S catalytic core that degrades ubiquitinated protein, and two 19S regulatory subunits that regulate the entrance of the targeted substrates into the core. Tight regulation of UPS-mediated proteolysis is maintained to control half-lives of proteins involved in cell cycle regulation, transcriptional control, antigen processing, angiogenesis, and removal of incorrectly folded or damaged proteins ( 2).
Despite evidence that dysregulation of the catalytic processes mediated by the UPS system is associated with tumorigenesis, tumor progression, and drug resistance, the relationship between proteasome expression and cancer progression is still poorly understood. Indeed, whereas proteasome down-regulation has been described in solid tumors, abnormally high proteasome levels have been observed in hematopoietic tumor cells. In addition to the proteasome, ubiquitin-specific proteases may be overexpressed in several cancer and transformed cell lines ( 3).
Consistent with a greater requirement for proteasomal function in cancer cells and other rapidly proliferating cells, proteasomal inhibition causes increased cell death in lymphoma cell lines when compared with normal lymphoblasts ( 4). Further, rapidly proliferating cells, such as embryonic endothelial cells, are extremely sensitive to apoptosis induced by proteasomal inhibition ( 5). These observations suggest that rapidly proliferating cells, particularly cancer cells, have a greater requirement for proteasomal activity than their normal counterpart. The evidence linking the proliferative status of a particular cell with sensitivity to proteasomal inhibition indicates that the proteasome represents a promising target for cancer therapy. Currently, a proteasome inhibitor PS-341 (bortezomib, Velcade) is undergoing clinical testing for treatment of several tumors and has been Food and Drug Administration approved for treatment of multiple myeloma ( 6– 14).
Better treatments for ovarian cancer are urgently needed. Despite the ubiquitous nature of proteasome function and the potential of proteasome chemotherapy in a variety of tumors, it is currently unknown whether abnormalities in UPS system and elevated sensitivity to proteasomal inhibition are associated with epithelial ovarian cancer. Our findings herein show that ovarian cancer cells are under significant UPS stress and suggest that proteasome inhibitors may have utility in the treatment of ovarian cancer.
Materials and Methods
Animals. Six-week-old female immunodeficient BNX (beige nude xid) mice were obtained from National Cancer Institute-Frederick (Frederick, MD) and maintained in a pathogen-free animal facility at least 1 week before use. All animal studies were done in accordance with institutional guidelines.
Cell culture. IOSE-29 and IOSE-397 cell lines were kindly provided by Nelly Auesperg (University of British Columbia, Vancouver, British Columbia, Canada) and cultured in Medium 199 and MCDB105 (1:1) with 10% fetal bovine serum and 50 μg/mL gentamicin. IOSE-2Ap2 and IOSE-4p2 were kindly provided by Hidetaka Katabuchi (Kumamoto University School of Medicine, Kumamoto, Japan). OVCAR-3, ES-2, and SKOV-3 cell lines were obtained from American Type Culture Collection (Manassas, VA) and cultured according to their specifications. MOSEC cells were kindly provided by Dr. Katherine F. Roby (University of Kansas, Kansas City, KS; ref. 15). Unless otherwise specified, cell lines were maintained in DMEM supplemented with 10% FCS, 100 IU/mL penicillin, and 100 μg/mL streptomycin at 5% CO2.
Drugs. The proteasome inhibitors MG132 and epoxomicin were purchased from Affiniti Research Products Ltd. (Exeter, United Kingdom) and dissolved in DMSO. The proteasome inhibitor PS-341 (Millenium Pharmaceuticals, Inc., Cambridge, MA) mixed with mannitol was dissolved in 0.9% NaCl at the appropriate concentration before each drug injection.
Xenograft murine model. Mice were inoculated s.c. into the right flank with 1 × 107 ES-2 cells in 100 μL DMEM. When tumor was measurable, mice were randomly assigned into two groups receiving PS-341 or 0.9% saline. Treatment with PS-341 was given i.v. twice weekly via tail vein at 1 mg/kg PS-341. The control group received the vehicle alone at the same schedule. Caliper measurements of the longest perpendicular tumor diameters were done every 2 days to estimate the tumor volume (mean ± SE; mm3), using the following formula: 4π / 3 × (width / 2)2 × (length / 2), representing the three-dimensional volume of an ellipse. Animals were sacrificed when their tumors reached 2 cm.
Cell viability assay. Cell viability was determined by 2,3-bis[2-methoxy-4-nitro-5-sulfophenyl]-2H-tetrazolium-5-carboxanilide inner salt (XTT) assay (Roche Diagnostics GmbH, Mannheim, Germany). Cells seeded at the concentration of 1,000 per well in 100 μL medium in 96-well plate were treated with proteasome inhibitors at specified concentrations. After the indicated periods, cells were incubated according to the manufacturer's protocol with the XTT labeling mixture for 4 hours. Formazan dye was quantified using a spectrophotometric plate reader to measure the absorbance at 405 nm (ELISA reader 190; Molecular Devices, Sunnyvale, CA). All experiments were done in triplicate.
Rate of proliferation. Cell lines were seeded at a density of 5,000 cells per well (12-well plate) under the specified culturing conditions. At the indicated time points, cells were trypsinized and counted in a hemocytometer in the presence of trypan blue to exclude dead cells. Proliferation rate was determined from the steepest slope.
Antibodies and Western blot analysis. We used the following reagents for detection with standard Western blot analysis techniques at the concentration recommended by the manufacturer: anti-19S subunit Rpt4, HC9 anti-proteasomal subunit α3, rabbit polyclonal antibody to human 20S proteasome subunit β5, and rabbit polyclonal antibody to human 20S proteasome subunit β4 (Affiniti Research Products); anti-ubiquitin (clone 5-25; Signet, Dedham, MA); anti-p21WAF1 and anti-p27KIP1 (BD Biosciences, San Diego, CA); anti-β-actin (Sigma, St. Louis, MO); and peroxidase-linked anti-rabbit or anti-mouse IgG (Amersham, Piscataway, NJ).
Flow cytometry. Cell cycle status was analyzed with a FACSCalibur flow cytometer (Becton Dickinson, San Diego, CA) by measuring fluorescence from cells stained with propidium iodide (PI; Sigma). For Annexin V experiments, cells were treated for indicated amount of time, harvested, and immediately stained for Annexin V/FITC (BD PharMingen, San Diego, CA) and PI according to the protocol provided by the manufacturer.
Caspase-3 activity assay. Caspase-3-like activity assay was done with the colorimetric substrate Ac-Asp-Glu-Val-Asp-p-nitroaniline (Ac-DEVD-pNA) in accordance with the protocol supplied by the manufacturer (Chemicon International, Temecula, CA). To determine the specificity of the caspase-3 activity, the caspase-3-specific inhibitor Ac-DEVD-CMK was added to the assay mixture. To determine caspase-like activity, the pan-caspase inhibitor Z-VAD-CMK was incubated together with proteasome inhibitors and the cell viability was then tested. All the experiments were conducted in triplicates.
Terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling assay. Paraffin-embedded tissue sections were processed for terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling (TUNEL) using an established method to assay for cell death-associated DNA double-strand breaks ( 16).
Immunohistochemical analysis. Immunohistochemical analyses of human tissues were done with the approval of The Johns Hopkins University Institutional Review Board. Sections of a microarray (3-4 μm) of paraffin-embedded tissues were used for immunohistochemistry. The tissue microarray was deparaffinized in fresh xylenes and rehydrated through sequential graded ethanol steps. Antigen retrieval was done by citrate buffer incubation [18 mmol/L citric acid, 8 mmol/L sodium citrate (pH 6)] using a household vegetable steamer (Black & Decker, Shelton, CT) for 60 minutes. The microarray was incubated for 5 minutes with 3% H2O2, washed in 20 mmol/L Tris, 140 mmol/L NaCl, and 0.1% Tween 20 (pH 7.6), and incubated with the mouse monoclonal antibody to 20S proteasome subunit α4 HC6 (Affiniti Research Products) diluted 1:250 for 60 minutes at room temperature. The avidin-biotin-peroxidase complex method from DAKO (Glostrup, Denmark) was used to visualize antibody binding, and the tissues were counterstained with hematoxylin. For immunohistochemical analyses of mouse tissues, tumors from mice were excised, fixed in 10% neutral buffered formalin, and embedded in paraffin according to standard histologic procedures. For microvessel density (MVD) assay, 10-μm frozen sections were briefly fixed in acetone and immunohistochemically stained for mouse CD34 expression. MVD was determined by light microscopy. Neovascularized areas were identified by scanning tumor sections at low-power magnification (×40) and then counted at high-power magnification (×200). Five separate ×200 field were analyzed per each condition.
Proteasomal overexpression is associated with ovarian cancer cells in vivo and in vitro. It has been reported that proteasomes are expressed at abnormally high levels in various hematopoietic tumor cells, including leukemic cells ( 17, 18). We sought to determine whether the link between proteasomal overexpression and malignant transformation is restricted to hematologic tumors or it can be expanded to other tumor malignances, particularly ovarian carcinoma. Therefore, we examined whether increase in proteasomal levels could be documented in ovarian tumors compared with normal surface epithelium. The expression pattern of the 20S proteasome was assessed in specimens of benign adenofibroma, low-grade ovarian carcinomas, and high-grade ovarian carcinomas by immunocytochemical analysis. The results showed that cells of both low-grade and high-grade serous carcinomas are associated with more robust 20S proteasome (anti-α4) staining than the ovarian surface epithelial cells in serous adenofibroma ( Fig. 1A ). A semiquantitative analysis of the staining intensity is given in Fig. 1B. Although the 20S staining was slightly slower in low-grade compared with high-grade serous carcinoma, this difference did not reach statistical significance. Several other 20S proteasomal subunits were found to be similarly up-regulated in malignant versus benign tumors (data not shown), suggesting that this phenomenon is not restricted to the α3 subunit. Consistent with the immunocytochemical study, semiquantitative immunoblot analyses showed that both low-grade and high-grade serous carcinomas are associated with the significantly (P < 0.02) higher relative levels of 20S and 19S proteasomal subunits when compared with benign cystadenoma (Supplementary Fig. S1A). A semiquantitative analysis of the ratio between 26S proteasome and actin in vivo is given in Fig. 1C. Again, the proteasomal subunit levels were lower in low-grade versus high-grade serous carcinoma, but the difference was not statistically significant.
To establish a suitable in vitro model for investigating the biological significance of proteasomal levels in ovarian cancer, we tested if the trend of proteasomal overexpression occurred in ovarian cancer cell lines compared with immortalized, but not tumorigenic, surface epithelial cells (IOSE). A panel of IOSEs (IOSE-29, IOSE-397, IOSE-4p2, and IOSE-2Ap2) and ovarian cancer cell lines (OVCAR-3, SKOV-3, and ES-2) and a new murine model of epithelial ovarian cancer (MOSEC) was tested for the expression of 19S and 20S proteasome subunits. In line with the in vivo profile of proteasome expression, all the ovarian cancer cell lines tested show considerably higher levels of 20S and 19S proteasome subunits when compared with IOSE (Supplementary Fig. S1B). A semiquantitative analysis of the ratio between 26S proteasome and actin in IOSEs and ovarian cancer cell lines is given in Fig. 1D.
In vivo and in vitro evidence of UPS stress in ovarian cancer. The UPS system is responsible for the degradation of polyubiquitinated proteins within the cells ( 1). Because ovarian cancer cells express abnormally high levels of 26S proteasome in vivo and in vitro, we next asked whether alteration in the levels of polyubiquitinated proteins is associated with proteasomal overexpression. Immunoblot analysis of polyubiquitinated protein expression in serous cystadenoma (10 cases) and low-grade (8 cases) and high-grade (10 cases) serous carcinomas revealed that ovarian cancer cells of both low-grade and high-grade serous carcinomas are associated with more robust anti-ubiquitin staining than epithelial cells of serous cystadenoma ( Fig. 2A ). A semiquantitative analysis of the polyubiquitinated protein levels show that the actin-normalized levels of polyubiquitinated proteins are ∼2- to 3-fold higher in malignant tumors compared with cystadenoma ( Fig. 2B). Consistent with these in vivo findings, immunoblot analysis of IOSEs (IOSE-29, IOSE-397, IOSE-4p2, and IOSE-2Ap2) and ovarian cancer cell lines (OVCAR-3-3, MOSEC, SKOV-3, and ES-2) revealed that polyubiquitinated proteins are present at significantly higher levels (P < 0.02) in cancer cell lines versus their immortalized but nontumorigenic ovarian surface epithelial cell lines ( Fig. 2C). A representative example of immunoblot analysis of polyubiquitinated in IOSEs and ovarian cancer cell lines in given in Fig. 2D. Thus, despite the increase in proteasomal levels, ovarian cancer cells accumulate more polyubiquitinated proteins. Taken together, these results suggest that ovarian cancer cells are under greater UPS stress than their normal counterpart.
Proteasomal inhibition induces G2-M cell cycle arrest and caspase-mediated apoptosis in ovarian cancer cell lines. The higher demand for proteasomal activity in ovarian cancer cell lines versus their nontransformed counterpart may render the cells more sensitive to proteasomal inhibition. To test this hypothesis, we compared the effect of the proteasome inhibitors MG132 and epoxomicin on the panels of IOSEs (IOSE-29, IOSE-397, IOSE-4p2, and IOSE-2ap2) and ovarian cancer cell lines (OVCAR-3, SKOV-3, ES-2, and JH-514, which is derived from a low-grade serous carcinoma and the mouse ovarian cancer model MOSEC). Both MG132 and epoxomicin ( Fig. 3A, left and right, respectively ) diminished cell viability of ovarian cancer cell lines by 24 hours of treatment in a dose-dependent manner. In contrast, the IOSEs were resistant to high concentrations of both MG132 and epoxomicin ( Fig. 3A, left and right, respectively).
The UPS modulates levels of key cell cycle regulatory proteins whose dysregulation is expected to affect the cell cycle and viability ( 19, 20). To test whether the reduced cell viability of ovarian cancer cell lines is associated with cell cycle dysregulation, ES-2 cells were incubated for various times with the proteasome inhibitor MG-132 and the cell cycle status was analyzed by flow cytometry after staining with PI. After 24-hour incubation with MG132, the percentage of cells in G0-G1 phase was reduced by nearly half compared with control, with the treated cells accumulating in S and G2-M ( Fig. 3B, left and right, respectively). To investigate the fate of the cells accumulated in G2-M following proteasomal inhibition, we analyzed them by flow cytometric after staining with Annexin V. Annexin V protein specifically binds phosphatidylserine; this phospholipid is normally localized on the inner leaflet of the plasma membrane but flips to the outer leaflet during early apoptotic signaling ( 21). As shown in Supplementary Fig. S1C, treatment with 20 μmol/L MG132 for 8 hours results in an increase in Annexin V staining cells (4.8% in control versus 35.5% in MG132-treated). Similarly, the fraction of cells taking up PI, a feature of both advanced apoptosis and necrosis, is elevated from 6.5% in control to 22.4% in treated cells. These findings support our hypothesis that proteasomal inhibition triggers apoptosis in ovarian cancer cells.
Progression through the phases of cell cycle is regulated by periodic activation of several cyclin-dependent kinases (CDK) whose levels are controlled through ubiquitination and proteasomal degradation ( 20, 22, 23). In particular, levels of CDK inhibitors (CDKI) p21WAF1 and p27KIP1 have been shown previously to be dysregulated following proteasomal inhibition ( 24– 28). Western blot analysis of the levels of p21WAF1 and p27KIP1 in ovarian cancer cell lines (ES-2) following MG-132 treatment revealed an accumulation of both p21WAF1 and p27KIP1 starting at 16 hours after inhibition. These data suggest a role for p21WAF1 and p27KIP1 in blocking the G2-M transition of the cells after proteasomal inhibition. The induction of p21WAF1 and apoptosis on proteasomal inhibition suggested a possible role for p53 in apoptosis. To address this possibility, we explored the relative levels of the 19S and 20S proteasome subunits in serous carcinoma specimens with wild-type and mutant p53. A similar elevation in proteasomal subunit levels was observed between these serous carcinoma specimens with respect to serous cystadenoma (Supplementary Fig. S2A and B). Similarly, there was no significant difference in the levels of polyubiquitinated proteins in p53 wild-type versus mutant serous carcinoma specimens, although both were elevated with respect to serous cystadenoma specimens (Supplementary Fig. S2C and D). Finally, three ovarian cancer cell lines bearing wild-type p53 exhibited similar proliferation rates and sensitivity to proteasomal inhibition–induced apoptosis as three lines bearing null or mutated p53 (Supplementary Fig. S2E and F). Taken together, these findings suggest that cell death induced by proteasomal inhibition occurs independently of p53 mutation status.
Furthermore, cleavage products of both CDKIs p21 and p27 (19- and 22-kDa bands, respectively) were also detected in ES-2 cell lysate collected 24 hours after addition of the proteasome inhibitor MG132, suggesting caspase-3 activation as the apoptotic mechanism associated with the decrease in cell viability ( Fig. 3C; ref. 20).
To confirm that activation of caspase-3 was induced by proteasome inhibitors, subconfluent ES-2 cells were treated for up to 24 hours with 20 μmol/L MG132 or 1 μmol/L epoxomicin and the caspase-3 activity was evaluated by measuring the cleavage of the caspase-3-specific substrate Ac-DEVD-pNA. Proteolytic activity for the caspase-3-specific substrate was detected in ES-2 cell lysate starting at 16 hours after addition of proteasome inhibitors, increasing ∼10-fold by 24 hours of treatment ( Fig. 3D, left). To determine whether caspase-3 activation was entirely responsible for proteasome inhibitor–induced cell death, caspase-3-specific (Ac-DEVD-CMK) and pan-caspase (Z-VAD-CMK) inhibitors were tested. Whereas pan-caspase inhibition results in almost complete protection, caspase-3 inhibition provides only partial protection from cell death in the presence of MG132 or epoxomicin. This result indicated that although proteasomal inhibition caused cell death via activation of caspases there may be more than one effector caspase involved ( Fig. 3D, right).
Proteasomal sensitivity is dependent on proliferation rate and UPS stress. A correlation between proliferation rate and proteasome inhibitor–mediated apoptosis has been shown previously in rapidly proliferating leukemic and endothelial cells versus their quiescent counterparts ( 4, 5, 9, 26, 29). Thus, we hypothesized that the differential effect in terms of cell viability following proteasomal inhibition may be associated with the greater proliferation rates of ovarian cancer cell lines versus the IOSE counterpart. To test this hypothesis, we examined whether the previously characterized IOSEs and ovarian cancer cell lines proliferate at different rates. Consistent with their greater sensitivity to proteasome inhibitor–induced cell death, the proliferation rate of ovarian cancer cell lines was significantly (P < 0.02) higher than in IOSEs ( Fig. 4A ). To exclude that the difference in sensitivity was a cell type–dependent effect and to more directly link the proliferation rate with the proteasomal sensitivity, we lowered the proliferation rate of MOSEC and ES-2 ovarian cancer cell lines by two independent methods and tested sensitivity to proteasomal inhibition in relation to proliferation rate.
In the first set of experiments, the proliferation rate of MOSEC and ES-2 cell lines was compared under normal culture conditions (10% FCS) and after an extended period of serum starvation (0.1% FCS). Ovarian cancer cell lines cultured in presence of 10% FCS grow exponentially, whereas under serum starvation the level of cell proliferation is significantly reduced (P < 0.02; Fig. 4B, left). Normally cultured or starved MOSEC and ES-2 cells were subsequently subjected to proteasomal inhibition and the viability was determined compared with control untreated cells ( Fig. 4B, right). Whereas the proteasomal inhibition leads to significantly attenuated cell viability in the proliferating ES-2 and MOSEC cells in 10% FCS, serum starvation protects ES-2 and MOSEC cells from the cytotoxic effects of proteasome inhibitor ( Fig. 4B, right). This effect is dose dependent and is seen with epoxomicin treatment (data not shown).
For the second set of experiments, the proliferation rate of ES-2 cells was slowed by contact inhibition, and the sensitivity to proteasomal inhibition was tested. After 3 days in culture, ES-2 cells plated at high density were 100% confluent, whereas those plated at low density were subconfluent (≤50%; data not shown). Fluorescence-activated cell sorting analysis confirmed that the percentage of ES-2 cells within the G0-G1 phase of the cell cycle was 83% in confluent cultures compared with 54% of subconfluent ES-2 cells ( Fig. 4C, left and right, respectively). Under these conditions, reduction of the cell viability for subconfluent ES-2 cell is significantly (P < 0.02) higher than for confluent ES-2 cells in the presence of either 25 or 50 μmol/L MG132, whereas vehicle alone had no effect ( Fig. 4D).
Because a decreased proliferation rate of the ovarian cancer cells led to a significant decrease in the toxicity associated with proteasomal inhibition, we examined whether a decrease in proliferation rate of cells is similarly accompanied by a reduction in UPS stress. First, we examined the levels of 20S proteasome in ovarian cancer cells after either 4 or 15 days of serum starvation in comparison with cells in standard culture conditions (10% FCS). The results show that proteasome levels are maintained at high levels for up to 4 days of serum starvation, consistent with the known long half-life of the proteasome, but significantly decrease with prolonged serum starvation up to 15 days ( Fig. 5A ). The ES-2 ovarian cancer cells already show attenuated sensitivity to proteasomal inhibition by the third day of starvation, although proteasome levels are still high, suggesting that proteasome levels are not the primary determinant of sensitivity to proteasome inhibitor in ovarian cancer cell lines.
We then evaluated the overall levels of ubiquitinated proteins as a function of cell proliferation. The immunoblot analysis shows that the levels of ubiquitinated proteins decrease rapidly with serum starvation and better correlate with timing of decreased sensitivity to proteasomal inhibition ( Fig. 5B). This implies that the degree of UPS stress more directly relates to rate of proliferation and sensitivity to proteasomal inhibition than the absolute level of proteasome. A semiquantitative analysis of the ratio between 20S proteasome and β-actin in normally cultured and serum-starved ovarian cancer cell lines is given in Fig. 5C, whereas the semiquantification between polyubiquitinated proteins and β-actin is given in Fig. 5D.
In vivo activity of proteasome inhibitor PS-341 against an ovarian cancer xenograft. The proteasome inhibitor PS-341/Velcade has shown efficacy for the treatment of multiple myeloma as well as other hematologic and solid tumors ( 30, 31). Our observation that ovarian cancer cells, like multiple myeloma, exhibit significant UPS stress in vivo and enhanced sensitivity to proteasome inhibitor–induced apoptosis in vitro suggests the potential for therapeutic responses to treatment of ovarian cancer with proteasome inhibitors in vivo. Therefore, the ability of proteasome inhibitor PS-341 to inhibit ES-2 tumor growth was examined with a mouse xenograft model of ovarian cancer. BNX mice were inoculated s.c. in the right flank with 1 × 107 ES-2 cells and 93% animals developed a measurable tumor after within 7 days of inoculation. Mice were then randomly assigned to PS-341 treatment group (1 mg/kg; n = 15) or 0.9% normal saline-treated control groups (n = 15) and treated by i.v. injection twice weekly. The average tumor volume was 212 mm3 at the initiation of treatment with no significant difference in tumor volume between treatment and control groups at day 0. As shown in Fig. 6A , a significant reduction in tumor volume was observed in mice treated with PS-341 versus the control group by day 5 of treatment (P < 0.05). The difference became more significant by day 9 (P < 0.0001), whereupon some control animals were sacrificed due to their large tumor burden. A survival curve for mice in each group is shown in Fig. 6B. A log-rank test revealed a significant prolongation in overall survival in animals treated with PS-341 versus controls (P < 0.0001). Notably, by day 18 of treatment, all of the control mice were sacrificed due to their tumor burden, whereas only 15% of mice in the PS-341 treatment arm had to be euthanized. Our in vitro study clearly indicated that the viability of ovarian cancer cells is highly compromised following proteasomal inhibition due to the rapid onset of apoptosis. Further, to investigate the in vivo effect of PS-341, tumor sections harvested from control mice and mice treated with proteasome inhibitor were subjected to TUNEL assay and scored outside of necrotic regions. As shown in Fig. 6C, whereas the percentage of apoptotic cells in the control group is 7.2% (SE ± 1.3) and the percentage of apoptotic figures in PS-341 animals is 53.8% (SE ± 4.1). Previous studies have shown an antiangiogenic effect of proteasome inhibitors as a result of apoptotic effect on endothelial cells. To assess the relative contribution of the antiangiogenic effect of the reduction of tumor growth after PS-341 treatment, tumor sections harvested from control mice and mice treated with proteasome inhibitor were evaluated for MVD by immunohistochemical analysis for CD34 expression. As shown in Fig. 6D, we found a statistically significant decrease in terms of MVD in sections obtained from PS-341-treated versus control mice: 16.8 ± 5.3 blood vessels per high-power field (×200) in control versus 36.1 ± 7.7 in PS-341-treated mice (P = 0.05).
In this study, we show that up-regulation of proteasome subunit levels occurs in ovarian cancer in vivo and in vitro. However, despite the up-regulation of total proteasome levels, ovarian carcinogenesis is associated with increased accumulation of polyubiquitinated proteins. This suggests that the up-regulation of the proteasome levels is insufficient to meet the increased demand generated by the greater metabolic rate of ovarian cancer cells, and this situation is reflected in greater UPS stress. Consistent with this evidence of increased proteasomal stress in ovarian carcinoma in vivo and in vitro, ovarian cancer cell lines show increased sensitivity to proteasome inhibitor–induced apoptosis in vitro. Furthermore, treatment with a proteasome inhibitor slows the growth of a human ovarian cancer xenograft model in mice.
Alteration in proteasomal expression and activity has been linked previously to a variety of biological and pathologic conditions ( 32– 37). In this study, we show that the up-regulation of proteasome protein levels not only is a phenomenon confined to malignances of hematopoietic origin but also occurs in ovarian cancer. This suggests that, as in leukemic cells, ovarian cancer cells may have a greater requirement for proteasomal activity than their untransformed counterpart. We cannot rule out that the additional proteasomes are not fully functional perhaps because a component is not coordinately up-regulated and is therefore rate limiting. However, our analysis shows that all the proteasome subunits tested (consisting of α, β, and 19S subunits) are coordinately up-regulated.
A correlation between proliferation rate and proteasome inhibitor–mediated apoptosis has been found previously in rapidly proliferating leukemic and endothelial cells versus their quiescent counterpart ( 4, 5). Our study also extends these previous finding by showing that level of UPS stress and sensitivity to proteasomal inhibition both correlate with cell proliferation rate. Specifically, we show that a decrease in proliferation rate of ovarian cancer cell lines leads to a significant decrease of polyubiquitinated proteins levels and decreased toxicity of proteasomal inhibition without changes in the proteasome levels. Thus, cellular levels of UPS stress more directly relate to rate of proliferation and sensitivity to proteasomal inhibition than the absolute levels of proteasome subunits. The decrease in proliferation rate is accompanied by reduction in sensitivity to proteasomal inhibition, whereas the steady levels of 20S proteasome subunits are maintained at high levels for up to 4 days of serum starvation. However, with continued serum starvation, the proteasome levels decrease. This is consistent with the known long half-life of the proteasome ( 38). Thus, our findings show that the degree of UPS stress regulates the levels of proteasome expression in cells.
Proteasome inhibitors can induce apoptosis by targeting various classes of protein involved in the regulation of the cell cycle ( 19). Prior investigations have shown that proteasomal inhibition of several types of solid tumors blocks the cells at the G2-M phase of the cell cycle ( 24– 26, 39). In this study, we show treatment of ovarian cancer cell lines with proteasome inhibitors caused accumulation of p21WAF1 and p27KIP1, whose levels are regulated by the UPS, and accumulation of the cells in G2-M phase. The concomitant accumulation of the cells in G2-M of the cell cycle together with accumulation of CDKIs, such as p21WAF1 and p27KIP1, is actually consistent. Although p21WAF1 is typically associated with arrest in G1-S transition, a role for p21WAF1 at the G2-M-phase transition has also been proposed ( 25). This suggests that accumulation of p21WAF1 and p27KIP1 following proteasomal inhibition may cause a temporary stall of the cells in G1 followed by a block in G2-M. Further, studies have shown that pro-caspase-3 overexpression sensitizes ovarian cancer cells to proteasome inhibitor–induced apoptosis ( 40, 41). This may occur via Bik/NBK stabilization ( 42). Here, we show that although caspase-3 activation is certainly involved in proteasome inhibitor–mediated cell death, caspase-3 in not entirely responsible for the apoptotic process on proteasomal inhibition. This is consistent with a study of mediators of endoplasmic reticulum stress-induced cell death on proteasomal inhibition showing activation of caspase-12 and caspase-8 in addition to caspase-3 ( 43).
PS-341 represents a novel class of myeloma therapy, several studies have shown the efficacy of the proteasome inhibitor PS-341 in a variety of tumor cell lines, and its synergistic antitumor activity has been shown in combination with other chemotherapy ( 8, 44– 47). Our in vitro data suggest a differential sensitivity of highly proliferating ovarian cancer cells to proteasomal inhibition. Such selectivity suggests proteasome inhibitors as a candidate new chemotherapeutic agent for ovarian cancer treatment. The current study provides in vivo evidence suggesting that trials of PS-341 in patients with ovarian cancer are warranted perhaps in combination with other chemotherapeutic agents. Indeed, a phase I trial of bortezomib in recurrent ovarian or primary peritoneal cancer is currently in progress ( 14). Treatment of mice bearing palpable ovarian cancer xenograft with PS-341 profoundly retarded tumor growth. Consistent with the in vitro effects of the proteasome inhibitors, we observed a profound increase in the level of apoptosis in ES-2 xenograft of treated mice. Further, a ∼50% reduction in the MVD was observed within the ES-2 tumor of treated mice. This is also in line with the previously described antiangiogenic effects of proteasome inhibitors ( 46, 48– 50), although the mechanism driving these effects is poorly understood. These findings are particularly encouraging given our evidence of significant UPS stress in ovarian carcinoma specimens and numerous studies showing the importance of angiogenesis in development of ovarian cancer.
Grant support: HERA Foundation (M. Bazzaro and A. Santillan) and Department of Defense Ovarian Cancer Congressionally Directed Medical Research Program OC010017 (R.B.S. Roden).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Drs. Nelly Auesperg, Hidetaka Katabuchi, and Katherine F. Roby for the gift of cell lines and Millenium Pharmaceuticals for the gift of PS-341.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received July 1, 2005.
- Revision received December 29, 2005.
- Accepted January 20, 2006.
- ©2006 American Association for Cancer Research.