• Open Access

Genomic profiling of malignant pleural mesothelioma with array-based comparative genomic hybridization shows frequent non-random chromosomal alteration regions including JUN amplification on 1p32

Authors


To whom correspondence should be addressed. E-mail: ysekido@aichi-cc.jp

Abstract

Genome-wide array-based comparative genomic hybridization analysis of malignant pleural mesotheliomas (MPM) was carried out to identify regions that display DNA copy number alterations. Seventeen primary tumors and nine cell lines derived from 22 individuals were studied, some of them originating from the same patients. Regions of genomic aberrations observed in >20% of individuals were 1q, 5p, 7p, 8q24 and 20p with gains, and 1p36.33, 1p36.1, 1p21.3, 3p21.3, 4q22, 4q34-qter, 6q25, 9p21.3, 10p, 13q33.2, 14q32.13, 18q and 22q with losses. Two regions at 1p32.1 and 11q22 showed a high copy gain. The 1p32.1 region contained a protooncogene, JUN, and we further demonstrated overexpression of JUN with real-time polymerase chain reaction analysis. As MPM cell lines did not overexpress JUN, our findings suggested that induction of JUN expression was involved in the development of MPM cells in vivo, which also might result in gene amplification in a subset of MPM. Meanwhile, the most frequent alteration was the 9p21.3 deletion, which includes the p16INK4a/p14ARF locus. With polymerase chain reaction analysis, we determined the extent of the homozygous deletion regions of the p16INK4a/p14ARF locus in MPM cell lines, which indicated that the deletion regions varied among cell lines. Our results with array comparative genomic hybridization analysis provide new insights into the genetic background of MPM, and also give some clues to develop a new molecular target therapy for MPM. (Cancer Sci 2007; 98: 438–446)

Abbreviations
BAC

bacterial artificial chromosome

CGH

comparative genomic hybridization

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

MPM

malignant pleural mesothelioma

PAC

P-1 derived artificial chromosome

PCR

polymerase chain reaction

RT-PCR

reverse transcription–polymerase chain reaction

STR

short tandem repeat

TSG

tumor suppressor gene.

MPM, a highly lethal neoplasm of the serosal lining of the pleural cavity, is thought to develop from superficial mesothelial cells.(1) In up to 80% of patients, MPM occurs within about 30 years of exposure to asbestos.(2–4) The incidence of MPM is expected to increase dramatically over the next few decades. It has been estimated that 250 000 people will die of MPM in Europe in the next three decades, and 2500–3000 new cases are diagnosed each year in the USA.(5,6) In Japan, a recent report has shown that there will be approximately 100 000 deaths due to MPM in the next 40 years using an age-cohort model.(7) Survival of patients with MPM is very poor, with a median survival of 7–11 months after diagnosis, especially in advanced-stage patients, regardless of a recent advancement in chemotherapeutic modalities that combines cisplatin and antifolate.(8–10)

The long latency period between asbestos exposure and tumor development implies that multiple, and likely diverse, genetic changes are required for the malignant transformation of mesothelial cells. Many studies have been conducted to determine underlying key genetic and epigenetic events responsible for the development of MPM, some of which may be directly caused by asbestos fibers. Traditional karyotype analysis using primary samples or cell lines uncovered multiple non-random chromosomal abnormalities that are frequently detected in most human MPM specimens, which include chromosomes 1p, 3p, 6q, 9p and 22q.(11–18) Subsequent studies of such common regions with allele loss, which indicate the sites of TSG, have identified the target genes of MPM, including p16INK4a/p14ARF on chromosome 9p21 and NF2 at 22q. The p16INK4a/p14ARF gene, one of the most frequently mutated TSG of human malignancies, has been shown to be inactivated in ∼90% of MPM, with most cases being targeted by homozygous deletion.(19,20) The NF2 gene at the 22q12 locus, which is responsible for a familiar cancer syndrome of neurofibromatosis type II, has also been shown to be inactivated in 40–50% of MPM, mainly with nonsense mutation or homozygous deletion.(21,22) In contrast, the p53 gene, another of the most frequently mutated TSG in human malignancies, is only occasionally mutated in MPM, with approximately 25% of MPM specimens being inactivated.(23,24) Meanwhile, MPM does not show frequent mutation of known protooncogenes including KRAS, NRAS and EGFR.(25–28) Thus, it has been suggested that there are other yet unidentified TSG or protooncogenes responsible for the development of MPM. Recently, a CGH technique introduced to search for additional genes that are potentially involved in MPM biology has identified other regions with alterations, including 1q, 4q, 5p, 6p, 7p, 8p, 8q, 10p13-pter, 13q, 14q, 15q, 17p12-pter, 17q and 20.(29–34)

In the present study, we carried out array CGH analysis with 17 resected MPM samples (from 16 patients) and nine MPM cell lines from a total of 22 individuals. We confirmed the same chromosomal alterations as described before in the literature and further identified new regions such as 8q24 and 13q33.2. We also identified high copy gain at 1p32, which includes the JUN protooncogene. The present study provides new insights into the genetic background of MPM, and also gives some clues to developing a new molecular target therapy for MPM.

Materials and Methods

Cell lines and tumor specimens.  Twelve MPM cell lines and one non-malignant mesothelial cell line (MeT-5A) were used. ACC-MESO-1, ACC-MESO-4, Y-MESO-8A, Y-MESO-8D, Y-MESO-9 and Y-MESO-12 were established in our laboratory,(28) whereas NCI-H28 (CRL-5820), NCI-H2052 (CRL-5915), NCI-H2373 (CRL-5943), MSTO-211H (CRL-2081) and MeT-5 A (CRL-9444) were purchased from the American Type Culture Collection (Rockville, MD, USA). NCI-H290 and NCI-H513 were gifts from Dr Adi F. Gazdar. All MPM cell lines were maintained in RPMI-1640 medium (Sigma-Aldrich, Irvine, UK) supplemented with 10% fetal bovine serum (Invitrogen, Carlsbad, CA, USA) and 1× antibiotic–antimycotic (Invitrogen) at 37°C in a humidified incubator with 5% CO2. MeT-5 A was cultured according to the instructions of the American Type Culture Collection. Nineteen MPM samples from 18 Japanese patients were obtained at Aichi Cancer Center Hospital, Nagoya University Hospital, Japanese Red Cross Nagoya First Hospital, Nagoya Second Red Cross Hospital and Kasugai City Hospital (KD332, KD355, KD471, KD476, KD905, KD977, KD1032, KD1033, KD1038, KD1039, KD1041, KD1042, KD1043, KD1044, KD1045, KD1046, KD1048, KD1049 and KD1050; of these, KD1039 and KD1041 originated from the same patient at surgery and autopsy, respectively). MPM samples and clinical data were collected after obtaining appropriate institutional review board approval and written informed consent from all patients. To confirm that there was no cross-contamination of clinical samples and cell lines, the uniqueness or identity of MPM tissues and established cell lines were evaluated by analysis of STR polymorphisms using the AmpFLSTR Identifier Kit (Applied Biosystems, Foster City, CA, USA), including the 16 STR loci D8S1179, D21S11, D7S820, CSF1PO, D3S1358, TH01, D13S317, D16S539, D2S1338, D19S433, vWA, TPOX, D18S51, Amelogenin, D5S818, and FGA. Primary tumors and cell lines used in the present study are summarized in Table 1.

Table 1. Summary of malignant pleural mesotheliomas analyzed with array comparative genomic hybridization (CGH)
KD numberSexSubtypeAsbestos exposureCell linep16INK4a/p14ARFNF2§JUN
  • KD Number indicates primary tumors available for array CGH analysis. Two primary tumors were obtained from the same patient at surgical resection (KD1039) and autopsy (KD1041).

  • p16INK4a/p14ARF status was indicated as follows: HL, high-level loss; L, loss; HD, homozygous deletion (detected in cell lines).

  • §

    +, No point mutation was detected with PCR sequencing analysis of exons 1–17 covering the entire open reading frame of NF2, and homozygous deletion was not detected in the corresponding cell line; (+), no point mutation was detected in exons 1–17, but homozygous deletion was not determined due to possible contamination of non-cancerous DNA; [+], undetectable point mutation for exons 2, 5, 7, 8, 9, 10, 11 and 12. Data of p16INK4a/p14ARF and/or NF2 of Y-MESO-8A, Y-MESO-8D, ACC-MESO-1, ACC-MESO-4, NCI-H28, H2052 and MSTO-211H referred to Sekido et al. and Usami et al.(21,28) Amp, amplification.

332MaleEpithelioid+ HL(+)No Amp
355MaleEpithelioid  (+)No Amp
471MaleEpithelioidUnknown HL(+)No Amp
476MaleBiphasicY-MESO-8 A, -8DHD+No Amp
905MaleEpithelioidUnknown HLdel(533–537)No Amp
977MaleEpithelioidUnknown  (+)No Amp
1032MaleBiphasic+  (+)No Amp
1033MaleEpithelioid+  (+)Amp
1038MaleEpithelioid+  (+)No Amp
1039MaleDuciduoid+  (+)Amp
1041MaleDuciduoid+ L(+)Amp
1043FemaleEpithelioid+  del(468–479)No Amp
1044MaleEpithelioid L(+)No Amp
1045MaleEpithelioid L(+)No Amp
1046MaleBiphasic+ L(+)No Amp
1048MaleEpithelioid+Y-MESO-9HDdel(527–528)No Amp
1049MaleEpithelioid+  (+)No Amp
FemaleEpithelioidY-MESO-12HD+No Amp
FemaleEpithelioidACC-MESO-1HDQ389XNo Amp
MaleEpithelioid+ACC-MESO-4HD+No Amp
MaleUnknownUnknownNCI-H28HD[+]No Amp
MaleUnknownUnknownNCI-H2052HDR341XNo Amp
UnknownUnknownUnknownMSTO-211HHD[+]No Amp

Preparation of DNA and RNA.  Genomic DNA was extracted using a standard phenol–chloroform method.(35) Normal DNA was prepared from peripheral blood of healthy male donors and non-cancerous lung tissue of the patients. Total RNA was prepared using TRIzol reagent (Invitrogen) according to the manufacturer's protocol. DNase treatment was carried out on columns during RNA purification using an RNase-Free DNase Set (Qiagen, Germantown, MD, USA). Random-primed, first-strand cDNA was synthesized from 2 µg total RNA using Superscript II according to the manufacturer's instructions (Invitrogen).

Genome-wide array-based CGH.  A genome-wide scanning array with 2304 BAC and PAC clones covering the whole human genome at a resolution of roughly 1.3 Mb was used as described previously.(36) In brief, clones were isolated from bacterial cultures containing the requisite antibiotics and extracted using a Plasmid Mini-kit (Qiagen). The location of all clones used for the array CGH was confirmed by standard fluorescence in situ hybridization analysis. BAC and PAC clones were amplified using degenerate oligonucleotide-primed PCR and spotted on glass slides. DNA preparation from cells, labeling, hybridization and scanning analysis were carried out as described previously(37) with minor modifications.(36,38,39) The data obtained were processed to detect chromosomal imbalances as described.(40)

Southern blot analysis.  Genomic DNA from patient samples (7 µg) was digested with EcoRI restriction enzyme, electrophoresed, and transferred to Hybond N+ (Amersham Biosciences, Piscataway, NJ, USA). Hybridization and washing were carried out using standard techniques.(35) The DNA probes were made by RT-PCR using normal lung cDNA. RT-PCR of JUN and β-actin were carried out using the primer sets: C-jun-S1, 5′-GACCTTATGGCTACAGTAACCC-3′ (sense) and C-jun-AS1, 5′-CTGCTCATCTGTCACGTTCT-3′ (antisense); and B-Actin-S, 5′-CTGTGGCATCCACGAAACTA-3′ (sense) and B-Actin-AS, 5′-AGGAAAGACACCCACCTTGA-3′ (antisense).

Quantitative real-time PCR.  The reaction mixture for real-time PCR using first-strand cDNA contained TaqMan universal PCR Master Mix (Applied Biosystems) and 200 nM of each primer, JUN (Hs 00277190_s1; Applied Biosystems) and FOS (Hs 00170630_m1). All real-time PCR assays were done in MicroAmp optical 96-well reaction plates on an ABI PRISM 7900 Sequence Detector System (Applied Biosystems) according to the manufacturer's instructions. For normalization between samples, PCR amplification of GAPDH (Hs 00266705_g1; Applied Biosystems) was included for each sample at each run. Fluorescence measurements and melting curve analyses were carried out using SDS 2.1 software (Applied Biosystems). The relative quantification of gene expression was computed using the comparative threshold cycle method with a mathematical formula described previously, and results are shown as a fold induction of mRNA.(41) We classified them into high-level expresser of JUN or FOS (defined as >0.15 of JUN or FOS mRNA expression relative to GAPDH mRNA expression), middle-level expresser (defined as >0.025 but <0.15), and low-level expresser (defined as <0.025).

Deletion mapping of 9p21.  Information on 16 microsatellite markers and one sequence-tagged site marker at 9p21 was searched, and their sequences were obtained from the Human Genome Database (GDB) and the Ensembl Genome Browser. Three primer sets for exons 1, 2 and 3 of p16INK4a were as described previously,(28) and the primer set of exon 1β of p14ARF was p14ARF-F, 5′-CACCTCTGGTGCAAAGGGC-3′ (sense) and p14ARF-R, 5′-CCTAGCCTGGGCTAGAGACG-3′ (antisense).

Mutation analysis of NF2.  Mutation analysis of NF2 was carried out by direct sequencing after PCR amplification of genomic DNA. Seventeen primer sets covering the entire coding region of NF2 were described previously.(28)

Results

Genomic profiles and data analysis of MPM.  Array CGH analysis was carried out using genomic DNA samples extracted from 19 MPM primary tumors and nine MPM cell lines (ACC-MESO-1, ACC-MESO-4, Y-MESO-8A, Y-MESO-8D, Y-MESO-9, Y-MESO-12, NCI-H28, NCI-H290 and MSTO-211H). Among 19 primary tumors, we did not detect any significant genomic alterations in two tumors, which was probably due to much contamination of genomic DNA from non-malignant cells, and we excluded these tumors for further analysis. Of the 26 MPM analyzed successfully, there were paired samples from the same individuals: the Y-MESO-8 A and Y-MESO-8D cell lines were established from the KD476 primary tumor, Y-MESO-9 was established from KD1048, and the other two primary tumors (KD1039 and KD1041) were obtained from the same patient at surgical resection and autopsy, respectively. Thus a total of 22 individual MPM were studied (Table 1). All of the clones on chromosome X were analyzed separately because of sex mismatching. Copy number changes were detected at high-resolution for genomes as a whole for primary tumor samples as well as cell lines. We defined regions of gain or amplification as log2 ratio > +0.2, and regions suggestive of heterozygous loss or deletion as log2 ratio < −0.2. Figure 1 shows representative data of the entire genomic profiles of three MPM primary tumors and one cell line from different individuals, with some shared altered regions being detected. For example, KD1033 (Fig. 1a) and KD1041 (Fig. 1b) showed shared regions including gain of 1p32.1, 5p, 8q, 11q22.1 and 20p and loss of 13q12 and 21q22. Figure 2 is a summary of chromosome imbalance detected in 17 MPM samples (black lines) and nine cell lines (red lines). Regions of high-level gain or amplification (defined as log2 ratio > +1.0) and those of homozygous loss or deletion (defined as log2 ratio < −1.0) are presented by thick lines. A summary of frequent chromosomal regions of gain and loss, and those of high-level copy gain or amplification, or homozygous loss or deletion detected in 17 MPM samples and nine cell lines is presented in Table 2. We also found that paired samples shared many chromosomal imbalances, although there were several different regions of gains and losses, or regions with relatively weak signals especially in the primary samples. The weak signals were thought to be due to contamination of non-malignant cell DNA (data not shown). Recurrent chromosomal imbalances found in at least six samples consisted of gain on chromosomes 1q (eight tumors/seven individuals), 5p (12/11), 7p (9/8), 8q24 (9/9), 20p (6/6) and loss on chromosomes 1p36.33 (13/13), 1p36.1 (7/7), 1p21.3 (7/6), 3p21.3 (10/8), 4q22 (7/6), 4q34-qter (6/6), 6q25 (7/6), 9p21.3 (16/16), 10p (6/5), 13q33.2 (11/9), 14q32.13 (13/11), 18q (7/6) and 22q (10/8).

Figure 1.

Array comparative genomic hybridization profile of malignant pleural mesothelioma from three primary tumors and one cell line. Log2 ratios are plotted for all clones based on chromosome position, with vertical dotted lines showing separation of the chromosome. Clones are ordered from chromosomes 1–22 and X within each chromosome on the basis of the Sanger Center Mapping Position, July 2004 version. (a) KD1033 sample shows chromosomal gain of 1p32.1-p32.3, 2p16, 3p22.2-pter, 3p12, 4q12, 5p, 6pter-q14.1, 8q, 9p, 10p, 11q22.1-q22.3, 11q23.3-qter, 14, 17p12-pter and 20p11.21-p12, and loss of 1p36.13-pter, 1q32-q42, 2q37.1-qter, 3q11-q13.31, 4q34.3-qter, 6q14.3-q21, 6q25-qter, 7q35-qter, 9q34.12-qter, 13q12.11-q13.3, 13q34, 16q23-qter, 17q11.2-qter, 18p, 18q12.2-qter and 21qcen-q22.2. (b) KD1041 primary sample shows chromosomal gain of 1p36.13-p36.32, 1p32.1, 5p, 6p22-pter, 6p12-p21.1, 8, 11p15.2-p15.3, 11q22.1, 20, 22q12-q13.2 and X, and loss of 3p21.31, 4q, 5q35.1-qter, 9p21.3, 11q23-qter, 13q12, 13q33.2, 15q22.3-qter, 16p13.2, 16q11-q12.2 and 21q22. (c) KD471 primary sample shows chromosomal gain of 1p22.2-p31.1, 1q, 2, 4p15-pter, 5p, 5q33.1-qter, 7, 8p21.1-pter, 8q, 9q, 12q24, 19 and 20, and loss of 1p36.31-p36.33, 1p36.13, 1p12-p22.1, 3p14.3-p21.31, 6q14-q25.1, 8p12-p21.1, 9p21.2-pter, 10, 11q12.1, 13, 15, 17p and 18q. (d) Y-MESO-12 cell line shows chromosomal gain of 5, 7pter-q21.3, 8q21-qter, 11qcen-q14.3, 15q11, 16, 19q13.2 and 20, and loss of 1p21-p31.1, 2p11, 4q22.1, 9p21.3, 11p12, 19p13.11 and 22.

Figure 2.

Summary of chromosome imbalance detected in 17 malignant pleural mesothelioma patients (black lines) and nine cell lines (red lines). Regions of loss and gain are shown by vertical lines on the left (loss) and right (gain) sides of each ideogram. Regions of high-level amplification are presented by thick lines.

Table 2. Chromosomal regions with frequent imbalances or high copy gain or loss detected in malignant pleural mesothelioma
AlterationChromosomal regionNo. patients (n = 17)No. cell lines (n = 9)No. individuals (n = 22)GeneBAC/PAC
  • Representative genes are listed at each region when bacterial artificial chromosome (BAC) and P-1-derived artificial chromosome (PAC) clones of continuously ordered gain or loss of maximum overlapped clones were less than 10, when known protooncogenes or tumor suppressor genes shown to be involved in human malignancies were located, or when only a few genes were located in this region.

  • A representative BAC/PAC clone was listed when continuously ordered gain or loss of maximum overlapped region was less than 10 clones, and the clone at the mid-point of the overlapped region was chosen.

  • §

    § High copy gain or loss was observed.

Gain
 1p32.1§ 30 2JUNRP11-63G10
1q 44 7  
5p 8411CDH10RP11-116O11
7p 54 8  
8q24 45 9MYCRP1-80K22
11q22.1§ 20 2IAPRP11-864G5
20p 33 6  
Loss
 1p36.3312113KITRP11-181G12
1p36.1 43 7NM_018125RP11-473A10
1p21.3 25 6RPL5RP4-716F6
3p21.3 73 8PFKFB4RP5-1034C16
4q22 25 6TMSL3RP11-309H6
4q34-qter 33 6Q9P2F5RP11-739P1
6q25 34 6PLEKHG1RP11-291C6
9p21.3§ 7916p16INK4a/p14ARFRP11-149I2
10p 24 5  
13q33.2 74 9DAOARP11-166E2
14q32.13 8511CHGA/ITPK1RP11-862G15
18q 43 6MALT1RP11-4G8
22q 73 8NF2RP1-76B20

High-level gain at 1p32.1 includes JUN protooncogene amplification.  The array CGH analysis of 26 MPM revealed that 1p32.1 and 11q22.1 were two distinct regions with high-level gains, which were detected in at least two individual samples (Table 2). Interestingly, these high-level gains were observed simultaneously in the two individuals of KD1033 (Fig. 1a) and KD1041 (Fig. 1b). Another sample, KD1039, was also detected for 1p32.1 amplification (data not shown), and KD1039 and KD1041 were derived from the same patient, with the former at the initial surgical resection and the latter at autopsy. Whereas the KD1033 primary tumor showed a larger gain of five consecutive clones at 1p32.1 including the RP11-63G10 clone, KD1039 showed only a gain of the RP11-63G10 clone but not of the neighboring clones, and KD1041 showed only a gain of the two clones RP11-63G10 and RP11-363E22, with RP11-363E22 located toward the centromeric direction from RP11-63G10 1.9 MB apart (data not shown). Thus, the gain of RP11-63G10 seemed to be a very specific, common genetic event for these MPM, and this BAC clone was found to contain the protooncogene JUN (Table 2).

Because previous studies have suggested that asbestos fibers induce JUN expression in rat pleural mesothelial cells,(42) we studied the JUN status of MPM cells in further detail. We carried out Southern blot analysis with nine primary tumors and nine cell lines, and confirmed JUN high-level amplification in the three samples but not in the remaining 15 samples (Fig. 3a). To determine whether these MPM overexpress the transcripts of JUN, we carried out quantitative real-time PCR with 11 MPM samples available for RNA analysis together with seven MPM cell lines and one non-malignant mesothelial cell line, MeT-5 A. We found that KD1041, with high-level amplification of JUN, overexpressed mRNA of JUN (Fig. 3b). Interestingly, we noticed that there seemed to be three groups with distinct levels of JUN expression. We classified MPM into three groups according to the levels of JUN expression: high-level expresser (defined as >0.15) for three tumors (KD977, KD1041 and KD1044), middle-level expresser (defined as 0.025 < JUN < 0.15) for eight tumors (KD1032, KD1033, KD1045, KD1046, KD1048, KD1049, ACC-MESO-4 and H290), and low-level expresser (defined as <0.025) for seven tumors (KD471, KD476, ACC-MESO-1, Y-MESO-8A, Y-MESO-8D, H28 and MSTO-211H) and MeT-5 A. Among the seven MPM cell lines, ACC-MESO-4 and H290 were classified into middle-level expresser and the remaining five into low-level expressers. We also studied the FOS expression to determine whether JUN coexpresses with FOS in MPM cells (Fig. 3c). Most of the MPM cells classified into either high- or middle-level expresser of JUN simultaneously expressed FOS equal or greater than 0.025, and most expressers of both genes were primary tumors.

Figure 3.

JUN amplification at 1p32.1 and expression of JUN and FOS messages in malignant pleural mesothelioma. (a) Southern blot analysis of JUN. Each lane was loaded with 7 µg genomic DNA from MPM samples. Southern blot shows high-level amplification of JUN in KD1039 and KD1041 and low-level amplification in KD1033. (b,c) Diagrammatic presentation of quantitative real-time polymerase chain reaction data for (b) JUN and (c) FOS mRNA from 11 primary samples, seven MPM cell lines and MeT-5 A. The results were averages of at least three independent experiments with error bars showing standard deviations. MPM were classified into three groups of JUN status expression: high-level expresser (defined as >0.15) for three tumors (KD977, KD1041 and KD1044), middle-level expresser (defined as 0.025 < JUN < 0.15) for eight tumors (KD1032, KD1033, KD1045, KD1046, KD1048, KD1049, ACC-MESO-4 and H290), and low-level expresser (defined as <0.025) for the remaining seven tumors and MeT-5A. MPM were also classified into three groups according to FOS expression status: high-level expresser (defined as >0.15) for five tumors (KD977, KD1033, KD1044, KD1045 and KD1046), middle-level expresser (defined as 0.025 < FOS < 0.15) for four tumors (KD1032, KD1041, KD1048 and KD1049) and MeT-5 A, and low-level expresser (defined as <0.025) for the remaining nine tumors.

Alterations of p16INK,4a/p14ARF at 9p21.3 and NF2 at 22q12.2.  We found frequent deletions of RP11-149I2 located at 9p21.3 in seven MPM samples and nine MPM cell lines, with five samples (two primary tumors and three cell lines) showing high-level loss. This BAC clone included p16INK4a/p14ARF, which is one of the most frequently mutated TSG in human malignancies, and we showed previously that p16INK4a/p14ARF was deleted in all MPM cell lines studied.(28) To determine whether the 9p21 deletion region in MPM extends further beyond the p16INK4a/p14ARF gene locus, which may indicate another target TSG of MPM in this region, we further carried out PCR analysis using multiple primer sets for comparison with locations of BAC and PAC clones on 9p21. Besides the nine MPM cell lines, another three MPM cell lines (NCI-H290, NCI-H513 and NCI-H2373) were also studied. After we confirmed homozygous deletions of exons 1, 2 and 3 of the p16INK4a gene and exon 1β of the p14ARF gene in all 12 (100%) MPM cell lines except MSTO-211H, which showed a partial retention of the gene, we used 16 microsatellite markers and one sequence site-tagged marker for the analysis (Fig. 4). For the telomeric direction, the INF-α cluster of genes was homozygously deleted in two cell lines but not in the remaining 10. For the centromeric direction, two cell lines (NCI-H290 and H2052) showed a larger deletion with consecutive losses at markers including D9S259, suggesting that these two cell lines had at least 4 Mb homozygous deletion. Meanwhile, four cell lines (Y-MES0-8A, -8D, NCI-H28 and H513) had a smaller homozygous deletion that was limited within D9S1749 and D9S790, suggesting that the maximum deletion size was less than 482 kb.

Figure 4.

Homozygous deletion map of the 9p21 region in 12 malignant pleural mesothelioma cell lines. Results of polymerase chain reaction analysis for each locus are shown by open ovals (retention) and closed ovals (homozygous deletion). Locations of genes and markers are according to those of the GDB Human Genome Database and Ensembl Genome Browser. Top bar shows the sizes between the selected markers proportionally: W55904 – (570 kb) – D9S162–(1.71 Mb) – IFNA – (418 kb) – D9S1749 – (191 kb) – D9S1748–(11 kb) – D9S1752 – (280 kb) – D9S790 – (1.41Mb) – D9S2020 – (609 kb) – GCT16E06 – (233 kb) – D9S171 – (246 kb) – D9S1679–(678 kb) – D9S265–(282 kb) – D9S126 – (280 kb) – D9S259 – (2.75 Mb) – D9S104 – (1.15 Mb) – D9S248 – (898 kb) – D9S251.

Finally, we studied any point mutations of the NF2 gene in 17 primary tumors. After sequencing 17 exons covering the entire coding region of NF2, we found that three tumors had small deletions, all of which resulted in a frameshift mutation (Table 1). Because genomic DNA extracted from snap-frozen primary tumor tissues was used for the analysis, the existence of homozygous deletion was not determined due to possible contamination of non-cancerous DNA.

Discussion

In the present study, we analyzed 17 MPM primary tumors and nine MPM cell lines using array CGH and identified regions of genomic gain and loss. Regions of genomic aberrations observed in >20% of individuals were 1q, 5p, 7p, 8q24 and 20p with gains, and 1p36.33, 1p36.1, 1p21.3, 3p21.3, 4q22, 4q34-qter, 6q25, 9p21.3, 10p, 13q33.2, 14q32.13, 18q and 22q with losses. We confirmed the same chromosomal alterations as reported earlier by other groups and further identified high gain or amplification regions including 1p32, which harbors the JUN protooncogene. To our knowledge, our present study provides the first detailed array CGH data on chromosomal imbalances in MPM patient tumors and cell lines.

Traditional allelotyping and karyotype analyses revealed non-random chromosomal abnormalities including 1p, 3p, 4p15.1-p15.3, 4q25-q26, 4q33-q34, 6q, 9p, 14q11.1-q12, 14q23-q24 and 22q.(11–18,43,44) Subsequently, chromosomal CGH (also known as conventional CGH) has been carried out to detect more detailed abnormalities in MPM (Table 3). For example, Krismann et al. showed a total of 77 cases of MPM in the main histological subtypes (epithelioid type, sarcomatoid type and biphasic type) using chromosomal CGH.(34) They reviewed common gains at the chromosomal regions of 1q23/1q32, 7p14-p15, 8q22-q23 and 15q22-q25, and common losses at the chromosomal regions of 1p21, 3p21, 4p12-p13, 4q31-q32, 6q22, 9p21, 10p13-pter, 13q13-q14, 14q12-q24, 17p12-pter and 22q in all subtypes. In the present study with array CGH analysis, we also detected similar aberrations of multiple loci that have been found in previous studies.(29–35) These regions include gains of 1p32, 1q and 7p, and losses of 1p21, 9p21 and 22q. In addition to these regions, we have identified new regions such as 8q24 and 13q33.2, which had not been detected with chromosomal CGH analysis. The gain of 8q24 locus was detected by array CGH in nine cases (nine individuals) of these 26 samples. A single BAC, RP1-80K22, which includes the known protooncogene MYC, was located at the overlapped regions of 8q24 amplification. As a previous study showed a significant increase in signal strength of MYC in the mesothelioma tissues from an experimental animal model, compared with basal expression in non-neoplastic mesothelial cells, our findings also support the importance of MYC alteration in the development of MPM.(45)

Table 3. Chromosomal regions with frequent imbalances shown in malignant pleural mesotheliomas from previous reports using chromosomal comparative genomic hybridization (CGH), and the current study using genome-wide array-based CGH
AuthorsYearSamplesFrequent gainsFrequent losses
Kivipensas et al.1996115p, 6p, 8q, 15q, 17q, 201p, 8p, 14q, 22q
Bjorkqvist et al.1997271cen-qter4q31.1-qter, 6q22-q24, 9p21-pter, 13, 14q24-qter, 22q13
Bjorkqvist et al.1998347p, 15q4q, 6q, 14q
Balsara et al.1999245p1p12-p22, 6q25-qter, 9p21, 13q12-q14, 14q24-qter, 15q11.1-q15, 22q
Krismann et al.2002771q23/1q32, 7p14-p15, 8q22-q23, 15q22-q251p21, 3p21, 4p12-p13, 4q31-q32, 6q22, 9p21, 10p13-pter, 13q13-q14, 14q12-q24, 17p12-pter, 22q
Current study 261q, 5p, 7p, 8q24, 20p1p36.33, 1p36.1, 1p21.3, 3p21.3, 4q22, 4q34-qter, 6q25, 9p21.3, 10p, 13q33.2, 14q32.13, 18q, 22q

Previous reports of chromosomal CGH analysis of MPM samples identified the region of gain at 1p32, although a specific candidate target gene was not referred to in detail.(34,46) Using array CGH, we found that a single BAC clone, RP11-63G10, detected the region of gain at 1p32.1 in three tumors from two individuals. The RP11-63G10 clone was the only clone that showed overlapping at this region, and harbored only one known gene, the JUN protooncogene. Whereas KD1033 showed relatively wide-range amplification including five consecutive clones, KD1039 and KD1041 showed only RP11-63G10 amplification or with another neighbor clone for the latter (data not shown). It is noteworthy that KD1039 and KD1041 were from the same patient at surgical resection and autopsy, respectively, but the ranges of amplification of the JUN locus were slightly different. Furthermore, except for 1p32, these two samples also showed distinct regions of chromosomal alteration for each locus, including a gain at 13q34 for KD1039, and gains at 11p15.2 and 11q22.1 and a loss at 13q33.2 for KD1041 (data not shown). Although we confirmed the identity of these two samples with 16 STR repeats, it remains unclear whether the KD1041 cells originated from a subclonal cancer cell population that existed in the KD1039 tumor and acquired another chromosomal alteration during propagation.

JUN is a transcription factor and functions through homodimerization or heterodimerization with FOS to form the transcription factor AP-1, which can bind to the promoter region of intermediate genes involved in cell division and other cell functions.(47) Heintz et al. reported that both crocidolite and chrysotile asbestos caused increases in the expression of JUN and FOS in rat pleural mesothelial cells.(42) They demonstrated that, in contrast to phorbol 12-myristate 13-ester, which induced rapid and transient increases in JUN and FOS mRNA, asbestos caused 2–5-fold increases in JUN and FOS mRNA dose-dependently, which persisted for at least 24 h in mesothelial cells. They concluded that by activating the early response gene pathway, asbestos may induce chronic cell proliferation that subsequently contributes to carcinogenesis in lung and pleura. Thus, our findings of JUN amplification and overexpression detected in MPM tumors is very intriguing, and we also found that three tumors with JUN amplification were from patients with high-grade asbestos exposure. Interestingly, five of seven MPM cell lines were classified into low-level expressers of JUN, compared with three high-level and six middle-level expressers of the 11 primary tumors. This finding suggests that primary MPM tumor cells are continuously exposed to some stress to induce JUN transcription, and that JUN transcription is not necessarily induced in the established MPM cell line and MeT-5 A cells under usual tissue culture conditions, which may also indicate that the levels detected in MPM cell culture are of baseline JUN expression. Meanwhile, the analysis of FOS expression revealed that it was expressed simultaneously with JUN in most MPM cases, with high levels of expression of both genes detected mainly in the primary tumors, but not in cell cultures. These findings suggest the possibility that some surgical manipulations cause artificial induction of some genes, including early response genes,(48) which leads to the observation of predominant expression of these genes in the primary tumors. Nevertheless, because gene amplification of JUN was indeed identified in three MPM tumors, we think that there were some strong and persistent factors for JUN activation during the development of the MPM tumor cells.

JUN has been shown to be induced by other factors such as hypoxia. A recent immunohistochemical analysis detected expression of hypoxia-inducible factor 1α at focal regions in most MPM tumors but not in mesothelial cells, suggesting that hypoxic stress exists in primary MPM tumors.(49) Although the mechanisms and causes of amplification of genes such as MYC family members remain poorly understood, amplification of several other genes has been implicated as being induced by carcinogens and other stresses, such as amplification of the dihydrofolate reductase gene via methotrexate treatment.(50) Thus, we speculate that the chronic induction of JUN expression might have been induced by multiple stimuli, most importantly by asbestos fibers at the initial stage and possibly by hypoxia and other unidentified factors continuously, and that this might result in gene amplification of JUN in a subset of MPM cells during long latency.

Using array CGH, we found a region of loss at 9p21 in 16 tumors (16 individuals) that was covered by a single BAC clone, RP11-149I2, which included the p16INK4a/p14ARF gene. It is well known that p16INK4a/p14ARF is one of the most frequently deleted genes in many types of human cancers. Previous studies by other groups identified frequent alteration of p16INK4a/p14ARF in most MPM, and we have also shown that p16INK4a/p14ARF was deleted in all 10 MPM cell lines studied.(28) Although studies with simple PCR techniques reported homozygous deletion of p16INK4a/p14ARF at a relatively lower frequency in MPM tissues than in cell lines, which may be due to contamination of a significant amount of normal stromal cells, we detected frequent deletion at 9p21.3 in seven MPM samples with array CGH. Furthermore, we determined the approximate lengths of deletion regions in 12 MPM cell lines, compared with the locations of DNA markers and BAC or PAC clones. We found that several cell lines showed a relatively small deletion with a maximum deletion size of 482 kb, whereas others showed at least a 4-Mb deletion size. Our findings of the p16INK4a/p14ARF deletion in MPM seem consistent with other reports that the sizes of homozygous deletions vary individually in any given tissue type of malignancy.(51–53) Although it is very clear that p16INK4a/p14ARF is the most important target TSG at the 9p21.3 region, other genes in this homozygous deletion region should also be studied to determine whether any of them play a role in the development of MPM.

Finally, the loss of 3p21.3 locus was detected by array CGH in 10 cases (eight individuals) of the 26 samples. One of the well-known TSG located at this region is RASSF1A, which is frequently inactivated by promoter hypermethylation in various types of human malignancies. The frequent hypermethylation of RASSF1A was also reported in MPM, which suggests that RASSF1A is a strong target TSG at 3p21 during the development of MPM.(54) Meanwhile, we also identified a homozygous deletion including CTNNB1 (β-catenin) at 3p22.1 in the NCI-H28 cell line, and further demonstrated that the exogenously transfected CTNNB1 gene inhibited the growth of NCI-H28 cells.(55,56) Thus, because several genes have been suggested as candidate TSG at the 3p21-22 region for various malignancies including MPM, further detailed analysis may be warranted to clarify the most important target TSG in this region for MPM.

To summarize, we subjected MPM samples to array CGH analysis and found genomic regions altered recurrently in MPM, including 1p32 JUN protooncogene amplification. Array CGH analysis can thus be expected to provide new insights into the genetic background of MPM and to offer some clues to developing a new molecular target therapy for this highly aggressive fatal tumor.

Acknowledgments

This work was supported by a Grant-in-Aid for Scientific Research from the Japan Society for the Promotion of Science. We would like to thank Dr Adi F. Gazdar for the cell lines, and Dr Yutaka Kondo, Dr Hirotaka Osada, Dr Masashi Kondo and Dr Toshimichi Yamamoto for helpful comments and special encouragement.

Ancillary