Thirteen samples were obtained from patients with MPMs, including subtypes of 6 epithelioids, 6 biphasics and 1 sarcomatoid, with written informed consent at Toronto General Hospital (Toronto, Canada). For TMA analysis, a total of 53 MPM were obtained from patients who underwent surgery at Hokkaido University Hospital and its affiliated hospital between February 1990 and April 2012. Patient's clinical information was extracted from the medical records. The protocol was approved by the appropriate institutional review board of Hokkaido University (no. 012-0136). Detailed information about demography, clinical characteristics and pathological stage are summarized in Table I. Postoperative pathological staging evaluation was demonstrated only in curative operative cases (n=21), showing stage I disease in 1 case, stage II disease in 5 cases, and stage III disease in 15 cases. In all, 1 patient had T1 disease, 7 patients had T2 disease and 13 patients had T3 disease. A total of 9 patients had N0 disease, 5 patients had N1 disease and 7 patients had N2 disease. Histological classification of tumors and stage were performed according to the Union for International Cancer Control (UICC) pathological tumor/node/metastasis (pTNM) classification criteria (14).

The human malignant mesothelioma cell lines used in the present study were as follows: NCI-H28, -H226, -H2052 and -H2452 were purchased from the ATCC (Manassas, VA, USA). All cancer cells were grown in monolayers in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS). A human adult normal mesothelial cell line (MES-F) was purchased from Zimbio Inc. (Research Triangle Park, NC, USA) and was grown in mesothelial cell growth medium (Zimbio). All cell lines were maintained at 37°C in atmospheres of humidified air with 5% CO₂.

Tumor samples from MPM patients with surgery were excised and stored at −80°C. QIAzol lysis reagent (Qiagen, Valencia, CA, USA) and one 5-mm stainless steel Bead (Qiagen) were added before homogenizing with a TissueLyser Adapter Set (Qiagen) for 2 min at 20 Hz (15,16). Total RNA was then purified using a miRNeasy Mini kit (Qiagen). The amount and purity were measured using a spectrophotometer (NanoDrop; Thermo Fisher Scientific, Wilmington, DE, USA).

cDNA was synthesized from 1 μg total RNA using QuantiTect^® Reverse Transcription kit (Qiagen). The primers were designed as follows: for KIF11, forward primer, 5′-acagcctgagctgttaatgatg-3′ and reverse, 5′-gatggctcttgacttagaggttc-3′; for KIF23, forward primer, 5′-tggttcctacattcagaaatgaga-3′ and reverse, 5′-cgttctgatcaggttgaaagagta-3′; for actin, beta (ACTB), forward primer, 5′-gaaatcgtgcgtgacattaa-3′ and reverse, 5′-aaggaaggctggaagagtg-3′; Quantitative RT-PCR analysis was performed using the LightCycler480^® SYBR-Green I Master Ready-to-use hot start reaction mix and LightCycler480^® system (Roche, South San Francisco, CA, USA). The thermal cycler conditions were as follows: 5 min at 95°C for denaturation, 45 cycles at 95°C for 10 sec, 56°C for 20 sec, and 72°C for 10 sec for PCR amplification, and 1 min at 65°C for melting. The threshold cycle value was defined as the value obtained in the PCR cycle when the fluorescence signal increased above the background threshold. PCR reactions were carried out in duplicates.

All short interference RNA (siRNA) oligonucleotide sequences for KIF11 and KIF23 siRNAs were purchased from Qiagen for the present study. AllStar Negative Control siRNA^® (Qiagen) were used as the negative control (NC-siRNA). The final concentration of 10 nM of siRNAs was incubated with HiPerFect^® transfection reagent (Qiagen) according to the manufacturer's instructions. The CellTiter96^® AQueous One Solution Cell Proliferation Assay (Promega, Madison, WI, USA) was used for the evaluation of the number of viable cells according to the manufacturer's instructions, using a microplate spectrophotometer (μQuant; BioTek Instruments, Inc., Winooski, VT, USA). Each experiment was performed in triplicates.

Archival slides for all the cases were reviewed to select three representative areas for each sample by an experienced pathologist (K.H.). TMA blocks were then constructed using a manual tissue microarrayer (JF-4; Sakura Finetek Japan Co., Ltd., Tokyo, Japan) with a 1.0-mm diameter needle. The finalized array blocks were sliced into 4-μm-thick sections and mounted on glass slides. To check the histopathological diagnosis and adequacy of tissue sampling, a section from each TMA was stained with regular hematoxylin and eosin (H&E), as well as calretinin immunohistochemistry (IHC), and examined by the same pathologist (K.H.). For KIF23 immunostaining, deparaffinization, antigen retrieval and IHC were performed on paraffin-embedded TMA sections and on whole sections using an automated IHC platform (Bond Max™; Leica Microsystems, Newcastle, UK). Antigen retrieval was performed in the condition of pH 6.0 for 20 min. The detection kit used was the Bond Polymer Refine Detection (DS9800; Leica Microsystems), incubation with post-primary for 20 min, 3 times at room temperature (RT). Anti-KIF23 monoclonal antibody [MKLP-1 (H-110): sc-22793; Santa Cruz Biotechnology, Santa Cruz, CA, USA; 1/100] was diluted using mixed antibody diluent (Dako:X0909 Protein Block Serum-Free: Dako:S0809 antibody diluent: Nichirei 10% normal goat serum 1:1:1). A polymer-based detection system (Refine cat. #DS9800) was used with 3′,3-diaminobenzidine (DAB) as the chromogen. The positive controls included a sample of testis and the negative control was performed on all cases by the rabbit primary antibody (Dako:X0903, rabbit immunoglobulin fraction (normal)]. KIF11 immunostaining were performed using an automated IHC platform (Autostainer plus; Dako, Glostrup, Denmark). Antigen retrieval was performed at pH 9.0 for 20 min. The detection kit used was the EnVision™+ Dual Link (K4063; Dako), incubation with post-primary for 60 min at RT. Anti-KIF11 polyclonal antibody (GTX109054; GeneTex, Inc., Irvine, CA, USA; 1:1,500) was diluted using mixed antibody diluent (Dako S2022 antibody diluent). A polymer-based detection system (EnVision™+ Dual Link #K4063) was used with 3′,3-diaminobenzidine (DAB) as the chromogen. The positive controls included a sample of testis, and normal lung and normal pleural samples were used as negative controls. Slides were dehydrated and placed on coverslips.

Digital images of IHC-stained TMA slides including 53 MPM cases were obtained using a whole slide scanner (ScanScope CS, Aperio ePathology; Leica Microsystems Inc., Toronto, ON, Canada). Annotation of tumor regions on whole slides was performed blinded to clinical follow-up data using Aperio's annotation software (ImageScope Viewing software, positive pixel count ver. 9.1). KIF11 and KIF23 were quantified by IHC scoring, which summated the percentage of area stained at each intensity level multiplied by the weighted intensity (0, 1, 2 or 3) reported in other studies (17–19) (Fig. 3). Initially, the weighted intensity of staining was graded as follows; grade 0 (negative), 1+ (weak positive), 2+ (moderate positive), and 3+ (strong positive) according to Aperio's annotation software). According to the total amount of IHC scores, KIF11 and KIF23 expression were then finally divided into two groups each (the threshold leading to the lowest P-value in log-rank test): low-level KIF11 expression (KIF11-L, with an IHC score <0.7) and high-level KIF11 expression (KIF11-H, with an IHC score ≥0.7), and low-level KIF23 expression (KIF23-L, with an IHC score <0.4) and high-level KIF23 expression (KIF23-H, with an IHC score ≥0.4). We attempted to correlate clinicopathological variables such as age, gender, pathological TNM stage, and histological classification with expression levels of KIF11 and KIF23 protein as determined by TMA analysis. Immunoreactivity was assessed for association with clinicopathological variables using the Fisher's exact for variables. Kaplan-Meier method was used to generate survival curves, and survival differences were analyzed with the log-rank test, based on the status of KIF11 and KIF23 expression. Values of P<0.05 were considered statistically significant. All analyses were performed using StatView version 5.0 software (SAS Institute, Inc., Cary, NC, USA).

By screening for molecular targeted genes as described above, we identified KIF11 and KIF23 overexpression with quantitative RT-PCR experiments in a majority of MPM cases (Fig. 1A). We also confirmed a high expression of these genes using 4 human MPM cell lines and low expression in a normal human mesothelial cell line (MES-F) (Fig. 1B). Quantitative PCR analysis using a cDNA panel containing normal human tissues also showed that these genes expressed mainly in the testis and thymus among different normal human organs examined (data not shown), which provides further evidence supporting these genes as promising molecular targets.

To assess whether therapeutic candidate genes are essential for growth or survival of MPM cells, we transfected specific siRNAs against KIF11 (si-KIF11-#4, si-KIF11-#6 and si-KIF11-#8), KIF23 (si-KIF23-#1, si-KIF23-#8 and si-KIF23-#9) into human MPM cell lines H28 and H2052, using AllStar siRNA^® as the negative control (NC-siRNA). The mRNA levels of transfected cells with independent siRNAs targeting these genes were significantly decreased in comparison to cells transfected with control siRNAs 48 h after transfection (Fig. 2A). Next, to evaluate the relationship between cell proliferation and gene knockdown, we conducted a cell viability assay using the CellTiter 96^® AQueous One Solution Cell Proliferation Assay. After siRNA treatment, proliferation of H28 and H2052 cells was significantly suppressed compared with control groups at day 4 after transfection (Fig. 2B), suggesting that upregulation of these candidate genes are related to growth or survival of MPM cells. Typical microscopic images of the cells are shown in Fig. 2C. Compared to control cells or control siRNA-treated cells, KIF11 siRNA-treated mesothelioma cells more frequently induced cell cycle arrest for mitotic cells with monopolar spindles called ‘monoastral spindle’ (20,21). KIF23 siRNA-treated mesothelioma cells exhibited large cell bodies with two or more nuclei. Depletion of KIF23 induces the formation of ‘multinucleate cells’, likely because of a cytokinesis defect (22–24).

We categorized KIF11 and KIF23 expression on the TMA according to the IHC score described above. Positive staining of tumor cells by KIF11 and KIF23 generally showed a cytoplasmic pattern in cancer tissue, and no staining was observed in benign chronic or fibrous pleuritis and normal lung tissues (Fig. 3B). Of the 53 malignant mesothelioma cases examined, KIF11-H was observed in 23 cases (43.4%) and KIF23-H was observed in 27 cases (50.9%) (details are shown in Table II). Of those, 16 epithelioid type (47.1% of 34 cases), 6 biphasic type (46.2% of 13 cases), and 1 sarcomatoid type (20.0% of 5 cases) showed KIF11-H. Twenty-two epithelioid type (64.7% of 34 cases), 5 biphasic type (38.5% of 13 cases) expressed KIF23-H, but no KIF23-H case was found in sarcomatoid and desmoplastic type. We then tried to correlate KIF11 and KIF23 expression with various clinicopathological parameters. However, no significant association was noted between KIF11 and KIF23 expression and other clinicopathological variables in curative operative cases (data not shown). Analysis using the Kaplan-Meier method indicated significant associations between KIF23-H and MPM patients who received curative resection with KIF23-H had shorter overall 5-year survival than those with KIF23-L (P=0.0194, by a log-rank test; Fig. 3C). A similar trend was observed in KIF11 expression even though the difference was not statistically significant.