Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Molecular profiling of malignant peritoneal mesothelioma identifies the ubiquitin–proteasome pathway as a therapeutic target in poor prognosis tumors


Malignant mesothelioma is an aggressive neoplastic proliferation derived from cells lining serosal membranes. The biological and clinical characteristics of epithelial type malignant mesothelioma are distinct from those of biphasic and sarcomatous type tumors. The goal of our study was to examine the molecular basis for this distinction. Microarray analysis confirmed that the molecular signatures of epithelial and biphasic histologic subtypes were distinct. Among the differentially expressed functional gene categories was the ubiquitin–proteasome pathway, which was upregulated in biphasic tumors. Cytotoxicity experiments indicated that 211H cells derived from biphasic tumors were synergistically sensitive to sequential combination regimens containing the proteasome inhibitor bortezomib and oxaliplatin. The mechanism of this synergistic response, which was not detected in cells of epithelial tumor origin, was apoptosis. Together, our results identify the ubiquitin–proteasome pathway as a biomarker of poor prognosis biphasic peritoneal mesothelioma tumors and suggest that proteasome inhibitors could increase the effectiveness of cytotoxic chemotherapy in this subset of patients.


Malignant mesothelioma is an aggressive neoplastic proliferation derived from cells lining serosal membranes. The annual incidence in the United States is approximately 2500 cases per year of which approximately 25% are peritoneal mesothelioma (Taub et al., 2000). Peritoneal mesothelioma is associated with documented asbestos exposure, but less frequently than pleural mesothelioma (15–30 vs 60–70%). Other potential predisposing factors include chronic peritonitis and prior radiation therapy (Gentiloni et al., 1997; Khalil et al., 2001). Peritoneal mesothelioma has a historical median survival of less than 1 year, but the clinical course is heterogeneous with several reports of long-term survival (Mohamed and Sugarbaker, 2002).

Similar to the pleural type, peritoneal mesothelioma histology may have a malignant spindle cell component (biphasic or sarcomatous subtypes) or be comprised solely of epithelial tumor cells (epithelial subtype). Tumors of biphasic/mixed or sarcomatous histologic subtypes tend to be locally aggressive and bulky, whereas the epithelial variant typically does not invade solid organs, but occasionally may infiltrate into the omentum. Pathologic parameters that identify prognostically distinct subsets have been studied in malignant pleural and peritoneal mesothelioma. Prior reports have identified localized mesothelioma (Kitagawa et al., 1996) and well-differentiated papillary mesothelioma (Goldblum and Hart, 1995) as mesothelioma subtypes associated with favorable prognosis. Once analysis is confined to diffuse malignant mesothelioma, epithelial subtype is associated with prolonged survival compared with biphasic and sarcomatous subtypes (Sebbag et al., 2000; Sugarbaker et al., 2003). It is unclear whether this clinical difference is attributable to aggressive growth patterns, frequent chemoresistance of mixed/sarcomatous tumors or to other biological factors.

Currently, malignant mesothelioma therapy is guided by clinical stage and patient factors rather than by tumor histologic or molecular features. Understanding the biological basis of mesothelioma clinical behavior may facilitate a personalized treatment approach that will identify poor prognostic parameters early, reducing the heterogeneity of clinical response. Previously, we and others identified molecular characteristics associated with histology and outcome, including loss of expression of P16 in biphasic tumors (Borczuk et al., 2005b; Lopez-Rios et al., 2006). In this study, we use global molecular profiling with DNA microarrays to identify pathways associated with histologic subtype of malignant peritoneal mesothelioma. We show that histology is associated with unique molecular signatures and we identify the ubiquitin–proteasome pathway as a potential therapeutic target in biphasic malignant mesothelioma.


The clinical and biological characteristics of epithelial type malignant mesothelioma are distinct from those of biphasic and sarcomatous type tumors. The goal of our study was to examine the molecular basis for this distinction. The study was performed using resected tumors acquired from patients undergoing laparotomy and omentectomy in preparation for entry into pilot multimodality protocols. There were no significant differences in the gender and age of the patients with epithelial subtype compared with biphasic subtype tumors (Table 1). As an initial step, we examined tumor DNA microarray profiles to determine if histology was associated with a unique molecular signature. We performed average linkage agglomerative hierarchical clustering within BRB array tools (developed by Richard Simon and Amy Peng Lam; using 16 microdissected peritoneal mesothelioma specimens and four normal peritoneal tissues. The tumors segregated into two clades associated with histological subtype (Figure 1), which were distinct from the nonmalignant tissues. One epithelial tumor (e6) clustered with the mixed subtype tumors. Pathological review of the ‘outlier’ specimen confirmed the histological diagnosis and there were no clinical parameters that otherwise distinguished this specimen. Other unsupervised clustering experiments have demonstrated that although tumors frequently cluster with tumors of similar morphology, additional biological and/or clinically associated variables may contribute to clustering patterns. To assess the stability of the clustering results in the background of experimental noise, we performed two tests of cluster reproducibility that measure the robustness index and the discrepancy index (McShane et al., 2002). These measures were 1 and 0, respectively, indicating that the clustering results were highly reproducible and robust. Together, the unsupervised clustering suggests that histological subtype is associated with distinct molecular signatures.

Table 1 Patient characteristics
Figure 1

Unsupervised average linkage hierarchical clustering dendrogram of 16 peritoneal mesothelioma specimens. Epithelial subtype (e) tumors cluster separately from biphasic subtype (b) tumors with a single exception (e6). We performed reproducibility analysis on the two major clusters identified in the dendrogram. The robustness index (R-index) measures the proportion of pairs of specimens within a cluster for which the members of the pair remain together in the reclustered perturbed data. The discrepancy index (D-index) measures the number of discrepancies (additions or omissions) comparing an original cluster to a best-matching cluster in the reclustered perturbed data. Using 100 permutations, the R-index was 1 and the D-index was 0, indicating that the clustering results were reproducible.

To identify genes associated with biphasic and epithelial histological subtypes, we performed supervised analysis using a t-test. For this analysis, we used the tumor specimens only. A total of 476 genes were differentially expressed between the epithelial and mixed subtypes, P<0.001 (Table 2 and Supplementary Table 1). A multivariate permutation test was computed based on 1000 random permutations, which determined that the probability of getting at least 476 genes significant by chance (at the 0.001 level) if there are no real differences between the classes is 0.001.

Table 2 Representative genes differentially expressed in biphasic and epithelial subtype peritoneal mesothelioma

To confirm the microarray results and to examine the relationship between gene expression and protein expression, we performed immunohistochemistry on a mesothelioma tissue microarray (TMA) using antibodies for representative proteins encoded by the classifiers (Figure 2). Similar to the gene expression pattern, P16 (cyclin-dependent kinase inhibitor 2A) immunostaining was lower in biphasic mesothelioma compared with epithelial subtype tumors. Conversely, Fascin (fascin homolog 1) immunostaining was increased in biphasic tumors (Supplementary Table 2). Interestingly, Fascin staining was intense in both the sarcomatous and epithelial components of biphasic tumors compared with lower immunoreactivity in the epithelial subtype tumors. This suggests that detection of fascin staining in epithelial subtype tumors may predict biological characteristics more closely related to biphasic tumors, similar to P16 loss in mesothelioma (Borczuk et al., 2005b).

Figure 2

TMA immunohistochemistry validation of gene expression data. (a) Epithelial mesothelioma is intensely and diffusely positive for p16, with nuclear and cytoplasmic immunoreactivity. (b) A biphasic mesothelioma is negative for p16. (c) Fascin immunoreactivity is weakly positive in this epithelial mesothelioma, and (d) diffusely and intensely positive in a biphasic mesothelioma (ad, original magnification × 150).

To identify biologically important differentially expressed genes and pathways, we examined the functional annotation of mesothelioma histologic subtype classifiers using The Pathway Comparison tool of BRB-Array Tools. This program provides a list of KEGG and BioCarta pathways that have more genes differentially expressed among the classes than would be expected by chance (Supplementary Table 3). Gene pathways included cell cycle regulation, transforming growth factor-β signaling pathway and several pathways associated with tumor immune response (i.e. complement pathway, T-cell receptor signaling pathway, interleukin (IL)-3- and IL-6-mediated signaling pathways). Importantly, the ubiquitin–proteasome pathway (ubiquitin-mediated proteolysis, proteasome) and genes associated with these pathways demonstrated significantly higher expression levels in biphasic tumors compared with epithelial subtype tumors. Similarly, these genes were expressed at higher levels compared with nonmalignant peritoneal cells (Supplementary Figure 1).

In eucaryotic cells, tight regulation of the ubiquitin–proteasome pathway is essential for normal cell cycling, function and survival; deregulation is frequently detected in neoplastic cells and is associated with chemoresistance, reduced apoptosis and enhanced proliferation (Voorhees et al., 2003). Specific inhibition of the pathway by the dipeptide boronic acid bortezomib, which reversibly binds the 26S proteasome, is cytotoxic to tumor cells in hematologic and solid tumors (Adams, 2004). In solid tumors, activity as a single agent is limited, but preclinical studies in melanoma, tumors of the pancreas and head and neck indicate that chemosensitivity to conventional regimens is enhanced by combination regimens incorporating bortezomib (Rajkumar et al., 2005).

Based upon the aggressive clinical behavior and frequent chemoresistance of mixed subtype malignant mesothelioma tumors and upon our gene expression results suggesting that the ubiquitin–proteasome pathway was dysregulated in these tumors, we hypothesized that treatment with bortezomib would enhance chemotherapy cytotoxicity in tumors with increased expression of genes involved in the proteasomal degradation pathway. As an initial step, we validated the microarray gene expression data for two ubiquitin–proteosomal pathway genes, UBE2D2 (ubiquitin-conjugating enzyme E2D 2) and PSMD8 (proteasome 26S subunit, non-ATPase, 8). Quantitative real-time polymerase chain reaction (PCR) confirmed that mRNA expression of UBE2D2 and PSMD8 was higher in biphasic tumors (P=0.004 and 0.0001, respectively; Figure 3, top panel).

Figure 3

Quantitative real-time PCR for UBE2D2 and PSMD8. Tumor RNA from malignant peritoneal mesothelioma specimens that were examined by microarray analysis (top panel) and mesothelioma cells (lower panel) were analysed by quantitative real-time PCR, normalized to β2 microglobulin. Expression of two representative genes in the ubiquitin-mediated proteolysis pathway was increased in biphasic subtype tumors (b) and in cells derived from a biphasic tumor (211H). Each bar represents the mean normalized copy number ±s.e.m. for experiments performed in duplicate.

We examined cytotoxicity by the 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2H-tetrazolium (MTS) assay in 211H (biphasic) and H28 (epithelial) malignant mesothelioma cells for oxaliplatin and bortezomib. mRNA expression of UBE2D2 and PSMD8 was higher in biphasic 211H cells compared with H28 cells (Figure 3, lower panel), suggesting that these cells could model drug effects associated with ubiquitin–proteasome pathway gene expression. Oxaliplatin (l-OHP) is a third-generation platinum compound with activity in mesothelioma cell lines and in the clinical treatment of patients with malignant mesothelioma, where it is generally used in combination with other cytotoxic agents such as gemcitabine (Fizazi et al., 2003; Schutte et al., 2003). The clinical efficacy is limited by chemoresistance, which we hypothesized would be diminished by bortezomib.

As single agents, the sensitivity to oxaliplatin was higher in epithelial cells as indicated by the IC50 at 24 h (IC50 H28: 2.91+0.5 μ M; 211H: 4.02+0.9 μ M) and the sensitivity to bortezomib was higher in biphasic derived cells (IC50 H28: 0.83+1.5 μ M; 211H: 0.39+0.3 μ M), but the differences were small and not statistically significant. Recent reports indicate that in solid tumors, the cytotoxicity of bortezomib is enhanced by combination regimens and suggest that the effects of sequential chemotherapy may be different from those of concomitant therapy (Ricotti et al., 2003; Nawrocki et al., 2004). We calculated the combination index (CI) for bortezomib regimens to determine the synergistic or antagonistic properties of various combinations (Figure 4). In biphasic derived 211H cells, concomitant therapy with bortezomib and oxaliplatin was antagonistic and both sequential regimens were synergistic, with the sequence of bortezomib preceding oxaliplatin demonstrating the greatest cytotoxicity. In H28 cells, all combination regimens were antagonistic. This result suggests that protein alterations by bortezomib-induced proteasomal inhibition in 211H cells enhances sensitivity and/or reduces resistance to oxaliplatin.

Figure 4

Sequential treatment with bortezomib (B) and oxaliplatin (L) is synergistic in biphasic 211H cells. Cytotoxicity was measured by the MTS assay. Cells were treated with drug for 24 h for the concomitant association of L and B (LB), or for 24 h followed by 24 h, for the sequential associations of L preceding B and of B preceding L (L → B and B → L). Experiments were performed in triplicate. Reported values are percentage of untreated cells. The CI of each drug association indicates additivity (CI=1), synergy (CI<1) or antagonism (CI>1).

The mechanism of synergistic cytotoxicity of bortezomib sequential combination regimens in biphasic 211H cells was apoptosis, which was dramatically increased in 211H cells treated sequentially with bortezomib followed by oxaliplatin (Figure 5). Examination of apoptosis pathways reported to be modulated by bortezomib (Adams, 2003) indicated that caspase-3 and caspase-9 were activated in the biphasic tumor-derived cells treated with proteasomal inhibitors (Supplementary Figure 2). We tested whether bortezomib-mediated apoptosis was related to functional changes of NF-κB. We did not detect changes in IκBα levels and moreover, nuclear extracts and electrophoretic mobility shift assay showed no detectable changes in NF-κB functionality (data not shown), suggesting that apoptosis was independent of NF-κB. There was rare apoptosis in epithelial-derived H28 cells; cell death was mediated by G1 arrest (data not shown) associated with p21/WAF1 protein, which demonstrated concentration-dependent induction in H28 cells treated with oxaliplatin.

Figure 5

Apoptosis is increased in 211H cells treated with bortezomib and oxaliplatin. Apoptosis was measured by quantitation of Annexin V labeling in cells counterstained with propidium iodide. Concomitant drug stimulation (BL) was 12 h; for sequential drug association, each drug treatment was 12 h at IC50.


Global molecular profiling indicates that the gene expression signatures of biphasic and epithelial subtype malignant peritoneal mesothelioma are distinct. Examination of differentially expressed pathways identified potential biological properties associated with tumor progression, response and clinical outcome in biphasic tumors. Similar approaches have been used by other researchers to examine molecular profiles of malignant mesothelioma of the pleura. Singhal et al. (2003) identified genes differentially expressed in tumors compared with normal tissue. Pass et al. (2004) and Gordon et al. (2003) have identified distinct signatures of prognosis in resected pleural mesothelioma, which interestingly were independent of histology subtype. More recently,Gordon et al. (2005) used unsupervised clustering to identify three groups of tumors, two of which were comprised primarily of epithelial and biphasic tumors, respectively. Together, these results suggest that molecular signatures distinguish these subtypes in both peritoneal and pleural malignant mesothelioma. The similarity of peritoneal and pleural malignant mesothelioma is further demonstrated by high correlation of gene expression of these two tumor types (AC Borczuk, CA Powell, H Pass, unpublished data).

The results of the unsupervised hierarchical clustering and of the class comparison analysis indicate that the signatures of biphasic tumors are distinct from epithelial subtype tumors and that this distinction is more pronounced than previously described in pleural mesothelioma tumors. It is possible that microdissection enrichment of the tumor specimens enhanced the gene signatures by reducing gene expression heterogeneity caused by peri-tumoral nonmalignant cells. Our analysis focused on differentially expressed pathways in order to identify genes that are most likely to contribute to important biological properties of mesothelioma. Among the differentially expressed pathways with increased expression in biphasic tumors were mitotic cell cycle, DNA replication, DNA repair, cell proliferation, SMAD protein heteromerization and the ubiquitin–proteasome pathway. It is difficult to determine which pathways are differentially expressed as a consequence of intrinsic biological differences and which are actually driving intrinsic biological differences. However, the ubiquitin–proteasome pathway is an attractive drug target that has clinically relevant implications, such as chemoresistance.

Differential expression of representative genes within the ubiquitin–proteasome pathway was confirmed by real-time PCR and differential response to specific 26S proteasome complex inhibition was established in vitro. Similar to preclinical observations in other solid tumors, combined therapy with bortezomib enhanced apoptosis and chemosensitivity to a conventional drug, oxaliplatin, in cells derived from a biphasic mesothelioma. Interestingly, synergy was detectable only when the drugs were given sequentially, with most effect detectable when bortezomib preceded oxaliplatin. Schedule-dependent synergy of chemotherapy has been reported in vitro and in clinical trials (Fahy et al., 2003; Leu et al., 2004; Robinson et al., 2004; Schenkein, 2005). The mechanism of synergy in our experiments was apoptosis and was NF-κB independent. The specific molecular mechanisms driving apoptosis in response to treatment with bortezomib are unclear. Additional studies are required to determine if these sequential effects detected in vitro are also operative in vivo.

Chemotherapy given systemically to patients with malignant mesothelioma has yielded disappointing results. The agents that produce response rates in 10–20% of patients include doxorubicin, detorubicin, epirubicin, carboplatin, mitomycin, cisplatin, cyclophosphamide and ifosfamide (Lee et al., 1996). More recently, combinations of gemcitabine/cisplatin (Nowak et al., 2002) and ralitrexed/oxaliplatin (Fizazi et al., 2003) for patients with pleural mesothelioma have shown objective response rates of up to 33%. A significant increment in chemotherapy effectiveness has been seen using pemetrexed in combination with carboplatin; a recent Phase III trial demonstrated a response rate of 42% and prolongation of median survival compared with cisplatin alone (Vogelzang et al., 2003). It is unclear whether subsititution of pemetrexed for oxaliplatin would increase cytotoxicity in our in vitro studies. Despite recent improvements in mesothelioma chemotherapy, chemoresistance remains a significant problem that is more frequent in biphasic subtype tumors.

The enhanced sensitivity of cells of biphasic origin to bortezomib is intriguing and suggests that biphasic tumors or mesothelioma tumors with increased expression of genes representing the ubiquitin–proteasome pathway may be particularly attractive targets for combination regimens that include proteasome pathway inhibitors. There are some limitations to this study. First, the small sample size may suggest that our results are not generalizable to other patients with malignant peritoneal mesothelioma. We were able to generate robust differentially expressed gene expression signatures from a relatively small sample size by enriching tumor content with microdissection and we confirmed differential expression of several targets using a larger cohort of tumors present on a TMA. Second, it is not clear that the in vitro drug sensitivity results will be applicable in vivo. Although in vitro conditions do not mimic in vivo tumor stromal interactions, they often predict clinical response as was the case of bortezomib in multiple myeloma (Adams et al., 1999). In conclusion, as there are no standard second-line regimens for malignant mesothelioma, we suggest that Phase II clinical trials of sequential regimens of bortezomib followed by conventional agents such as oxaliplatin and gemcitabine are appropriate for consideration in tumors with a biphasic-type molecular signature.

Materials and methods

Tissue specimens obtained at debulking were snap frozen and stored at −80°C in the Tumor Bank Facility of the Columbia University Cancer Center. Specimens were cryostat sectioned and mounted as 10-μm sections. Sections were fixed in 95% ethanol, stained with eosin, dehydrated and air dried.

DNA microarray analysis

As is our routine practice, tissue specimens were microdissected to enrich for tumor (Borczuk et al., 2003). As normal tissue contamination of clinical mesothelioma tumor samples is variable in terms of quantity and cell type, it is difficult to determine accurately the contribution of normal tissue to gene expression profiles generated from tumor homogenates. Although homogenized tumor specimens are adequate to study differences between tumor and nonmalignant tissues, the heterogeneous contamination of tumor tissue by surrounding normal tissue may confound studies directed towards other questions such as gene classifiers of histology. Uncoverslipped sections were examined microscopically at × 40 and neoplastic and stromal cells were microdissected with a 20-gauge needle. The microdissected areas of tumor were collected directly into guanidine thiocyanate for RNA extraction, using the RNeasy Mini Kit (Qiagen, Valencia, CA, USA) following the manufacturer's protocol. As a control for nonmalignant serosal tissue, we used four nonmalignant adult peritoneal specimens obtained at herniorraphy. Normal peritoneal cells were acquired using needle microdissection. Methods for RNA extraction, labeling and hybridization to the Affymetrix U95Av2 DNA array (Santa Clara, CA, USA) have been described previously (Borczuk et al., 2003). Probe level analysis and normalization to nonmalignant peritoneal tissue was performed using Robust Multi-array Algorithm (Irizarry et al., 2003). Normalized data were filtered using BRB-Array Tools to remove genes for which less than 25% of the expression data values had at least a 1.33-fold change in either direction from the gene's median value. A total of 5203 genes passed the filtering criteria. We used average linkage hierarchical clustering with Pearson correlation to cluster the samples and used the R (reproducibility) measure described by McShane et al. (2002) to evaluate the robustness of the clusters. The R measure is based on perturbing the expression data with Gaussian noise, reclustering and measuring the similarity of the new clusters to the original clusters. For each pair of samples in a cluster of the original data, the R measure is the proportion of the time they stay in the same cluster after perturbation and reclustering. We used the D (discrepancy) measure to evaluate the robustness of the clusters. The D measure is based on perturbing the expression data with Gaussian noise, reclustering and measuring the similarity of the new clusters to the original clusters. The variance used for the Gaussian noise was the median variance of log ratios (or log signals) over the set of samples, the median taken over the genes.


Peritoneal mesothelioma TMAs were constructed with cores from mesothelioma specimens that were acquired at surgical resection as described previously (Borczuk et al., 2003). TMA sections underwent immunohistochemical staining to detect two proteins found to be associated with tumor histologic subtype, for which antibodies were commercially available (Borczuk et al., 2003). Review of sections was performed using uniform criteria by a pathologist (ACB). Antibody staining was scored as negative (score 0), low positive with multifocal or diffuse faint staining (score 1), intermediate positive with multifocal or diffuse staining (score 2) or strong diffuse positive (score 3). For biphasic tumors, immunostaining results are presented for the sarcomatous regions of the tumor. The antibody sources were Dako (Carpenteria, CA, USA) for Fascin and Labvision (Fremont, CA, USA) for P16 (16P07).

Quantitative real-time PCR in tumors

Tumor RNA was isolated from frozen sections and from cell pellets with the RNeasy Kit and converted into cDNA using SuperScript III (Invitrogen, Carlsbad, CA, USA). Primers were as follows: UBE2D2 forward, 5′-IndexTermTTCCGAAGGAGCTACGTCTT; reverse, 5′-IndexTermTGAGAAACCACATACAAACCAAA; and PSMD8 forward 5′-IndexTermTCGAGAAGGCCTACGAGAAAA; reverse, 5′-IndexTermGATGACCTGTTTGGCCAGTT. PCR products were detected using the LightCycler System (Roche, Alameda, CA, USA), normalized to the value of the internal standard B2M.

Cell culture and cytotoxicity

Human-derived malignant mesothelioma cells MSTO-211H and NCI-H28 (CRL 2081 and 5820, American Type Culture Collection, Manassas, VA, USA) were seeded in 96-well plates at a density of 5000 cells/well in pentuplicate. The IC50 for bortezomib (B) (Millennium, Cambridge, MA, USA) and oxaliplatin (L) (Sanofi-Aventis, Bridgewater, NJ, USA) was determined by the MTS assay (Promega, Madison, WI, USA) in triplicate (Reile et al., 1990). Drug association activity was tested in cell lines as concomitant (IC50 L+IC50 B for 24 h) and as sequential exposure (IC50 L 24 h followed by IC50 B 24 h or IC50 B 24 h followed by IC50 L 24 h). The CI of each drug association was calculated by the following formula: As/AAss+Cs/CAss, where As and Cs are the IC50 at 24 h of drug A and C and AAss and CAss are the concentration of drug A and C as single agents able to reach the cell growth inhibition achieved by the association with drug A and C at 24 h for combination and 48 h for sequential treatment studies (Tallarida, 2001). In this analysis, synergy is defined as CI values less than 1.0, antagonism as CI values greater than 1.0 and additivity as CI values equal to 1.0.

Apoptosis analysis

Cells were seeded at a density of 250 000/well and treated with single drugs and concomitant drug stimulation for 12 h. For sequential drug treatment, each treatment at IC50 was 12 h. Apoptosis was quantified as reported previously (Kim et al., 2003). Briefly, phosphatidylserine translocation was measured using the Annexin V-PI Apoptosis Detection Kit (BD Biosciences, San Diego, CA, USA) for fluorescence labeling and the FACScan for measurement of the resulting fluorescence. Cells were counterstained with propidium iodide (Sigma, St Louis, MO, USA) to confirm early apoptotic events.

Western analysis

Cells were treated for 12 h with single drugs at three increasing concentrations (1/10 IC50, IC50 and 10 × IC50) and with concomitant drug stimulation at IC50 for 12 h. For sequential drug treatment, each treatment at IC50 was 6 h. Immunoblotting was performed as described previously (Borczuk et al., 2005a). Sources of antibody were as follows: Cell Signaling Technology (Beverly, MA, USA): cleaved caspase-3 (Asp175), cleaved caspase-9 (Asp330); Santa Cruz Biotechnology (Santa Cruz, CA, USA): WAF1-p21 (F-5); and Sigma: β-actin.

Accession codes




  1. Adams J . (2003). Cancer Treat Rev 29 (Suppl 1): 3–9.

  2. Adams J . (2004). Nat Rev Cancer 4: 349–360.

  3. Adams J, Palombella VJ, Sausville EA, Johnson J, Destree A, Lazarus DD et al. (1999). Cancer Res 59: 2615–2622.

  4. Borczuk AC, Gorenstein L, Walter KL, Assaad AA, Wang L, Powell CA . (2003). Am J Pathol 163: 1949–1960.

  5. Borczuk AC, Kim HK, Yegen HA, Friedman RA, Powell CA . (2005a). Am J Respir Crit Care Med 172: 729–737.

  6. Borczuk AC, Taub RN, Hesdorffer M, Hibshoosh H, Chabot JA, Keohan ML et al. (2005b). Clin Cancer Res 11: 3303–3308.

  7. Fahy BN, Schlieman MG, Virudachalam S, Bold RJ . (2003). J Surg Res 113: 88–95.

  8. Fizazi K, Doubre H, Le Chevalier T, Riviere A, Viala J, Daniel C et al. (2003). J Clin Oncol 21: 349–354.

  9. Gentiloni N, Febbraro S, Barone C, Lemmo G, Neri G, Zannoni G et al. (1997). J Clin Gastroenterol 24: 276–279.

  10. Goldblum J, Hart WR . (1995). Am J Surg Pathol 19: 1124–1137.

  11. Gordon GJ, Jensen RV, Hsiao L-L, Gullans SR, Blumenstock JE, Richards WG et al. (2003). J Natl Cancer Inst 95: 598–605.

  12. Gordon GJ, Rockwell GN, Jensen RV, Rheinwald JG, Glickman JN, Aronson JP et al. (2005). Am J Pathol 166: 1827–1840.

  13. Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP . (2003). Nucleic Acids Res 31: e15.

  14. Khalil N, Parekh TV, O'Connor R, Antman N, Kepron W, Yehaulaeshet T et al. (2001). Thorax 56: 907–915.

  15. Kim H, Xu G-L, Borczuk AC, Busch S, Filmus J, Capurro M et al. (2003). Am J Respir Cell Mol Biol 29: 694–701.

  16. Kitagawa Y, Wong F, Lo P, Elliott M, Verburgt LM, Hogg JC et al. (1996). Am J Respir Cell Mol Biol 15: 45–54.

  17. Lee JJ, Trizna Z, Hsu TC, Spitz MR, Hong WK . (1996). Cancer Epidemiol Biomarkers Prev 5: 191–197.

  18. Leu KM, Ostruszka LJ, Shewach D, Zalupski M, Sondak V, Biermann JS et al. (2004). J Clin Oncol 22: 1706–1712.

  19. Lopez-Rios F, Chuai S, Flores R, Shimizu S, Ohno T, Wakahara K et al. (2006). Cancer Res 66: 2970–2979.

  20. McShane LM, Radmacher MD, Freidlin B, Yu R, Li MC, Simon R . (2002). Bioinformatics 18: 1462–1469.

  21. Mohamed F, Sugarbaker PH . (2002). Curr Treat Options Oncol 3: 375–386.

  22. Nawrocki ST, Sweeney-Gotsch B, Takamori R, McConkey DJ . (2004). Mol Cancer Ther 3: 59–70.

  23. Nowak AK, Byrne MJ, Williamson R, Ryan G, Segal A, Fielding D et al. (2002). Br J Cancer 87: 491–496.

  24. Pass HI, Liu Z, Wali A, Bueno R, Land S, Lott D et al. (2004). Clin Cancer Res 10: 849–859.

  25. Rajkumar SV, Richardson PG, Hideshima T, Anderson KC . (2005). J Clin Oncol 23: 630–639.

  26. Reile H, Birnbock H, Bernhardt G, Spruss T, Schonenberger H . (1990). Anal Biochem 187: 262–267.

  27. Ricotti L, Tesei A, De Paola F, Ulivi P, Frassineti GL, Milandri C et al. (2003). Clin Cancer Res 9: 900–905.

  28. Robinson BW, Ostruszka L, Im MM, Shewach DS . (2004). Semin Oncol 31: 2–12.

  29. Schenkein DP . (2005). Clin Lung Cancer 7 (Suppl 2): S49–S55.

  30. Schutte W, Blankenburg T, Lauerwald K, Schreiber J, Bork I, Wollscgkaeger B et al. (2003). Clin Lung Cancer 4: 294–297.

  31. Sebbag G, Yan H, Shmookler BM, Chang D, Sugarbaker PH . (2000). Br J Surg 87: 1587–1593.

  32. Singhal S, Wiewrodt R, Malden LD, Amin KM, Matzie K, Friedberg J et al. (2003). Clin Cancer Res 9: 3080–3097.

  33. Sugarbaker PH, Welch LS, Mohamed F, Glehen O . (2003). Surg Oncol Clin N Am 12: 605–621, xi.

  34. Tallarida RJ . (2001). J Pharmacol Exp Ther 298: 865–872.

  35. Taub RN, Keohan ML, Chabot JC, Fountain KS, Plitsas M . (2000). Curr Treat Options Oncol 1: 303–312.

  36. Vogelzang NJ, Rusthoven JJ, Symanowski J, Denham C, Kaukel E, Ruffie P et al. (2003). J Clin Oncol 21: 2636–2644.

  37. Voorhees PM, Dees EC, O'Neil B, Orlowski RZ . (2003). Clin Cancer Res 9: 6316–6325.

Download references


This work was supported in part by the Columbia University Mesothelioma Center. Bortezomib was provided by Millennium.

Author information



Corresponding author

Correspondence to C A Powell.

Additional information

Grant support: American Cancer Society CRTG00058.

Supplementary Information accompanies the paper on the Oncogene website (

Supplementary information

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Borczuk, A., Cappellini, G., Kim, H. et al. Molecular profiling of malignant peritoneal mesothelioma identifies the ubiquitin–proteasome pathway as a therapeutic target in poor prognosis tumors. Oncogene 26, 610–617 (2007).

Download citation


  • mesothelioma
  • microarray analysis
  • bortezomib
  • drug therapy combination

Further reading


Quick links