Malignant mesothelioma is a highly aggressive, asbestos-related cancer frequently marked by mutations of both NF2 and CDKN2A. We demonstrate that germline knockout of one allele of each of these genes causes accelerated onset and progression of asbestos-induced malignant mesothelioma compared with asbestos-exposed Nf2+/− or wild-type mice. Ascites from some Nf2+/−;Cdkn2a+/− mice exhibited large tumor spheroids, and tail vein injections of malignant mesothelioma cells established from these mice, but not from Nf2+/− or wild-type mice, produced numerous tumors in the lung, suggesting increased metastatic potential of tumor cells from Nf2+/−;Cdkn2a+/− mice. Intraperitoneal injections of malignant mesothelioma cells derived from Nf2+/−;Cdkn2a+/− mice into severe combined immunodeficient mice produced tumors that penetrated the diaphragm and pleural cavity and harbored increased cancer stem cells (CSC). Malignant mesothelioma cells from Nf2+/−;Cdkn2a+/− mice stained positively for CSC markers and formed CSC spheroids in vitro more efficiently than counterparts from wild-type mice. Moreover, tumor cells from Nf2+/−;Cdkn2a+/− mice showed elevated c-Met expression/activation, which was partly dependent on p53-mediated regulation of miR-34a and required for tumor migration/invasiveness and maintenance of the CSC population. Collectively, these studies demonstrate in vivo that inactivation of Nf2 and Cdkn2a cooperate to drive the development of highly aggressive malignant mesotheliomas characterized by enhanced tumor spreading capability and the presence of a CSC population associated with p53/miR-34a–dependent activation of c-Met. These findings suggest that cooperativity between losses of Nf2 and Cdkn2a plays a fundamental role in driving the highly aggressive tumorigenic phenotype considered to be a hallmark of malignant mesothelioma. Cancer Res; 74(4); 1261–71. ©2013 AACR.
Malignant mesothelioma is a highly aggressive, incurable cancer linked to asbestos exposure. Although often perceived as a locally invasive cancer, the exact cause of death is poorly understood, and postmortem studies have revealed widespread dissemination of tumor in more than 85% of cases, with involvement of almost all organs (1). Genetic studies of human malignant mesothelioma have uncovered frequent loss/mutation of the tumor suppressor genes NF2 and CDKN2A (2, 3). Inactivation of NF2, and the resulting absence of its protein product, Merlin, causes loss of contact inhibition and increased tumor cell proliferation and migration through interactions with a myriad of downstream effectors (4). In studies with mouse models, heterozygous germline inactivation of one copy of Nf2 has been shown to consistently accelerate asbestos-induced malignant mesothelioma onset, providing experimental evidence implicating Merlin loss as an important event in malignant mesothelioma tumorigenesis (5). CDKN2A encodes the tumor suppressors p16INK4A and p14ARF (p19Arf in mice), components of the Rb and p53 pathways, respectively, and both protein products bind to and regulate proteins involved in cell-cycle progression/checkpoints and apoptosis (6). In human malignant mesothelioma, homozygous deletions of the CDKN2A locus are often large and typically inactivate both p16INK4A and p14ARF, with a synergistic effect with regard to tumorigenesis (6). Using mouse models with targeted knockout of Cdkn2a exons 1α or 1β, inactivating p16Ink4a or p19Arf, respectively, heterozygous loss of either Cdkn2a gene product proved sufficient to hasten malignant mesothelioma onset following exposure to asbestos (7). Moreover, asbestos-exposed mice with heterozygous loss of both protein products, via deletion of the shared Cdkn2a exon 2, showed accelerated onset of malignant mesothelioma compared with similarly treated mice with loss of either product alone.
NF2 and CDKN2A are frequently co-inactivated in human malignant mesothelioma (8). To model the effect of co-inactivation of these genes in malignant mesothelioma, we crossed Nf2+/− and Cdkn2a (exon 2)+/− mice to create doubly heterozygous mice referred to hereafter as Nf2+/−;p16/p19+/− mice. We show that upon exposure to asbestos, Nf2+/−;p16/p19+/− mice develop malignant mesothelioma at a greatly accelerated rate compared with Nf2+/− and wild-type (WT) littermates, and that deficiency for both genes drives a highly aggressive form of malignant mesothelioma, with an increased cancer stem cell (CSC) population and higher metastatic potential than for malignant mesothelioma cells from Nf2+/− or wild-type mice. In addition, c-Met upregulation/activation, through a p53-miR-34a–dependent mechanism, is shown to contribute to the increased migratory/metastatic phenotype and CSC maintenance of malignant mesothelioma cells derived from Nf2+/−;p16/p19+/− mice.
The data presented offer compelling support for cooperativity between Nf2 and Cdkn2a in driving the development of highly aggressive malignant mesotheliomas marked by enhanced tumor spreading capability and the presence of CSCs. We further show that c-Met activation contributes to the metastatic potential and CSC phenotype exhibited by this novel mouse model of malignant mesothelioma. These findings provide strong in vivo genetic evidence that helps to explain the highly aggressive nature of malignant mesotheliomas, which often harbor alterations of both NF2 and CDKN2A.
Materials and Methods
Animals and asbestos treatments
p16/p19+/− (01XB2, FVB/N.129-Cdkn2atm1Rdp) mice (6), obtained from the Mouse Models of Human Cancers Consortium, were crossed to Nf2+/− mice to obtain all of the genotypes used herein. All mice were in a comparable FVB genetic background. Mice at 6 to 8 weeks of age were injected intraperitoneally every 3 weeks with 400 μg crocidolite (UICC, SPI Supplies; total, 3.2 mg/mouse; refs. 6 and 7). Mice were scored as having malignant mesothelioma based on histological evidence and/or if tumor cells exhibited a combination of 3 or more malignant mesothelioma markers, including mesothelin, as assessed by reverse transcriptase-PCR (RT-PCR) and/or immunohistochemistry (IHC). Studies were performed according to NIH's Guide for the Care and Use of Laboratory Animals. The protocol was approved by the Committee on the Ethics of Animal Experiments of the Fox Chase Cancer Center (protocol 00-26).
Experimental metastasis and intraperitoneal orthographic transplant assays
As an experimental assay for metastasis, 0.5 × 106 or 1.0 × 106 murine malignant mesothelioma cells were injected into the tail vein of recipient severe combined immunodeficient (SCID) mice. Each individual cell line was injected in triplicate into 3 independent SCID mice. After 4 weeks, mice were euthanized via CO2 asphyxiation, and the lungs were excised, fixed in buffered formalin, and embedded in paraffin. Lung sections were stained using hematoxylin and eosin (H&E) stain or IHC.
For orthographic transplant studies, 0.5 to 1.0 × 106 malignant mesothelioma cells were injected intraperitoneally into recipient SCID mice. After 4 weeks, mice were euthanized, and organs and tumors were harvested for histopathology. H&E-stained slides of all organs were analyzed to look for invasion and frank metastases in each recipient SCID mouse.
Paraffin blocks were cut into 5-μm sections and then placed on positively charged microscope slides. Sections were dewaxed in xylene and hydrated through a graded ethanol series. Heat-induced antigen retrieval was then performed in 10 mmol/L sodium citrate (pH 6.0) in a microwave for 10 minutes. Endogenous peroxidase activity was blocked by immersing slides in 3% H2O2 in PBS for 30 minutes. Slides were then incubated for 30 minutes with goat-derived blocking serum, followed by overnight incubation in a 1:250 dilution of total c-met rabbit polyclonal antibody (Santa Cruz) or a 1:50 dilution of phosphorylated c-met (phosphorylated tyrosine sites pY1230, pY1234, and pY1236) rabbit polyclonal antibody (Invitrogen) at 4°C. Slides were next incubated with goat anti-rabbit biotinylated secondary antibody for 30 minutes and then with streptavidin peroxidase for 30 minutes, both at room temperature. 3,3′-Diaminobenzidine (Sigma-Aldrich) substrate chromogen was applied for 4 minutes. Other slides were stained in a 1:100 dilution of Ki67 rat monoclonal antibody (Dako) following the IHC procedure described above. Slides were counterstained with hematoxylin and mounted with permount (Fisher). Additional slides were incubated with antibodies against pan-cytokeratin and cytokeratin 8 (Sigma) or mesothelin (Santa Cruz). Other sections were stained with antibodies against Sox2, Nanog, Oct 3/4, and β-catenin.
Primary cell cultures
Malignant mesothelioma cells were isolated from ascitic fluid and/or peritoneal lavage as described (6, 7). All primary cell cultures used for molecular analyses were from passages ≤6. PCR analysis was carried out on all cultures that expressed mesothelial markers, and immunoblotting was performed on a random set of cultures to validate the PCR results. Cells were maintained in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% FBS and penicillin/streptamycin.
Drug treatments of malignant mesothelioma cells
Malignant mesothelioma cells in log growth phase were treated with vehicle control or c-Met inhibitors SU-11274 (5 μmol/L) or SGX2943 (5 μmol/L Calbiochem) for 24 hours before harvesting for protein extraction, immunoblotting, scratch, or aldefluor-based assays. Malignant mesothelioma cells were also treated with vehicle control or 100 μmol/L etoposide (a gift of M. Murphy, Wistar Institute, Philadelphia, PA) for 48 hours and harvested for both RT-PCR and immunoblot analysis. Cells were nucleofected with siRNA against Tp53 or miR34a (Dharmacon) using Amaxa Nucleofection Kit R and program T20 of an Amaxa nucleofector machine. Cells were harvested 48 or 72 hours post-nucleofection, and RNA or protein was extracted using standard methods.
Immunoblots were prepared with 50 μg of protein/sample, as described (9). Antibodies against P-Met, Akt, P-Akt, and MAPK (1:1,000 dilution), from Cell Signaling, and against c-Met, P-Erk, and β-actin (1:1,000), from Santa Cruz, and p53 (1:500, NCL-p53-505, Novocastra) were used. Appropriate secondary antibodies (anti-rabbit-, anti-mouse-, and anti-goat-HRP; Santa Cruz) were used at a 1:2,000 dilution.
mRNA expression of murine Met, miR34a, and Ppib (peptidylprolyl isomerase D) were measured using TaqMan technology and an ABI Prism 7900 Sequence Detection System. Total RNA was extracted using Trizol (Invitrogen), according to the manufacturer's instructions. For each sample, first-strand cDNA was generated from total RNA using a High Capacity cDNA kit (Applied Biosystems), according to the manufacturer's instructions. Reactions were prepared in triplicate for each gene using TaqMan Gene Expression Master Mix and the following TaqMan Gene Expression Assays (Applied Biosystems): Met (Mm01156980_m1), miR34a (mmu-mir-34a), and Ppib (Mm00478295_m1). 96-well plates were loaded, and reactions were cycled using TaqMan universal cycling conditions. During thermal cycling, the threshold cycle (Ct) was determined for each sample by taking the average of 3 replicates. The average Ct value for the control gene Ppib was subtracted from the average Ct value for each target gene (Met and mIR34a) to normalize the amount of sample RNA added to the reaction. The comparative Ct method was used to determine the level of the target gene mRNA in malignant mesothelioma cell cultures from Nf2+/−;p16/p19+/− and wild-type mice relative to that in the normal sample (Applied Biosystems User Bulletin #2, October 2001): relative quantification = 2 − Ct, where Ct = average Ct (diseased) − average Ct (normal).
Malignant mesothelioma cell lines 87 and 129 from Nf2+/−;p16/p19+/− mice were seeded on a 6-well tissue culture plate and grown to confluence. Upon confluency, monolayers were scratched with a 200-μL pipette tip, and a marked area was photographed with light microscopy at ×40 magnification, using phase-contrast microscopy. Cells were treated with dimethyl sulfoxide or SU-11274 (2 μmol/L) for 24 hours, and then the same marked area was rephotographed.
Matrigel invasion assay
In vitro invasion of malignant mesothelioma cells from Nf2+/−;p16/p19+/− wild-type mice were measured using 24-well BioCoat Matrigel Invasion Chambers (Becton Dickinson). The lower compartment contained RPMI medium with 10% FBS as a chemoattractant. In the upper compartment, 2.5 × 104 cells per well were placed in serum-free medium and incubated for 22 hours at 37°C in a humidified incubator with 5% CO2. For similar experiments with a c-Met inhibitor, 5 μmol/L SU-11274 or vehicle control was included in media of both the top and bottom compartments of the invasion chambers; then malignant mesothelioma cells were seeded into the invasion chambers. Invading cells on the underside of the Matrigel membrane were fixed with 10% formaldehyde and stained using Diff-Quik Stain (Dade Behring). Cells were counted with a light microscope, and invasion was estimated as the average number of cells in 5 fields using a ×10 objective.
Tumor spheroid growth assay of CSCs
Malignant mesothelioma cells (5,000–10,000 cells/60-mm plate; Becton Dickinson Falcon) were seeded in serum-free DMEM/F12 media supplemented with B27 supplement, EGF (10 ng/mL), basic fibroblast growth factor (10 ng/mL), and penicillin/streptomycin. Spheroids were photographed using light microscopy after 1 to 2 days of culture. Malignant mesothelioma cell lines were also subjected to spheroid assays in the presence or absence of SU11279 (5 μmol/L) or SGX523 (5 μmol/L) for 1 to 2 days, and spheroid size/number was determined using light microscopy.
CD24 immunofluorescent staining/flow cytometry analysis
A total of 1 × 106 malignant mesothelioma cells were incubated with or without 10 μL of anti-CD24 FITC-labeled antibody (BD Biosciences) for 30 to 40 minutes in 100 μL of PBS. Cells were washed in PBS 3 times and then resuspended in 500 μL of PBS for flow cytometry analysis. The percentage of CD24-staining cells was determined using a FACscan Flow Cytometer (Becton Dickinson) and quantitated using FlowJo analysis software.
Nf2;p16/p19 deficiency accelerates tumor development in a mouse model of asbestos-induced malignant mesothelioma
To evaluate cooperation between loss of Nf2 and Cdkn2a, wild-type, Nf2+/−, and Nf2+/−;p16/p19+/− mice were chronically injected with asbestos intraperitoneally over a 24-week period. Nf2+/−;p16/p19+/− mice developed malignant mesothelioma at a greatly accelerated rate (median, 24 weeks) compared with Nf2+/− (38 weeks) and wild-type (45 weeks) mice (Fig. 1). Moreover, Nf2+/−;p16/p19+/− mice had an increased incidence of advanced disease compared with Nf2+/− and wild-type mice, with most tumors being sarcomatous and having a high percentage of Ki67-positive cells (Supplementary Fig. S1). Thus, the findings in Nf2+/−;p16/p19+/− mice mirror those in the human disease counterpart, where sarcomatoid features are associated with aggressive disease and poor prognosis (10).
Asbestos-exposed Nf2+/−;p16/p19+/− mice develop ascites harboring tumor spheroids
Upon sacrifice of tumor-bearing mice, we routinely collected ascites to establish cell lines. During this process, we noted the presence of tumor spheroids in effusions from some Nf2+/−;p16/p19+/− mice; and these oncospheroids stained positively for various malignant mesothelioma markers (Fig. 2A). Overall, spheroids were found in 22% of Nf2+/−;p16/p19+/− mice compared with 4% of Nf2+/− mice and 0% of wild-type mice (Fig. 2B).
Malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice are metastatic
The increased frequency of oncospheroids in ascites from Nf2+/−;p16/p19+/− mice could be a reflection of a more advanced form of malignancy. Indeed, in other tumor types, for example, glioblastoma, tumor spheroids have been correlated with metastatic disease (11–13). To address this possibility, we used an experimental metastasis assay to compare the metastatic potential of malignant mesothelioma cells from Nf2+/−;p16/p19+/−, wild-type, and Nf2+/− mice. Three cell lines per genotype were injected individually into the tail vein of SCID mice, and lung tissues were evaluated for tumor formation 4-week postinjection. All 9 SCID mice injected with malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice showed multiple tumors in the lungs, whereas only 3 of 9 mice (embolisms) from Nf2+/− mice and 0 of 9 from wild-type mice developed tumors (Fig. 3A). In a separate experiment, 3 randomly selected malignant mesothelioma cell lines derived from asbestos-exposed p16/p19+/− mice (7) were injected via the tail vein; 6 of 9 lines from p16/p19+/− mice formed tumors in the lung, whereas none of the malignant mesothelioma cell lines from wild-type mice seeded the lungs. Together, these data suggest that loss of Nf2 and p16/p19 each contributes to the seeding potential of malignant mesothelioma cells in an experimental model of metastasis, and that their collective inactivation promotes enhanced metastatic potential.
In a complementary assay, malignant mesothelioma cell lines derived from Nf2+/−;p16/p19+/− and wild-type mice were injected intraperitoneally into SCID mice. Although malignant mesothelioma cells from both Nf2+/−;p16/p19+/− and wild-type mice could form primary tumors, only malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice formed tumors that were able to migrate through the diaphragm and invade the lungs (Fig. 3B). These data, and those from the experimental metastasis assay, show that malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice have a enhanced metastatic potential.
Malignant mesotheliomas from Nf2+/−;p16/p19+/− show enrichment for CSCs
The ability of CSCs to form nonadherent spheroids in the presence of defined growth factors has been described in many human and murine cancers (14). Thus, we tested whether the increased oncospheroid formation seen in Nf2+/−;p16/p19+/− mice correlates with the number of CSCs. To address this, we cultured malignant mesothelioma cells from Nf2+/−;p16/p19+/− and wild-type mice under nonadherent conditions with defined growth factors. Malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice formed much larger spheroids than malignant mesothelioma cells from wild-type mice (Fig. 4A). Moreover, adherent malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice stained more intensely for the malignant mesothelioma–specific CSC marker CD24 (15) than did malignant mesothelioma cells from wild-type mice (Fig. 4B).
CSCs are resistant to standard chemotherapies and are thought to represent the tumor cell population that persists after treatment and causes recurrence (14, 16–21). Pemetrexed, in combination with cisplatin, is a standard chemotherapeutic agent used as a first line therapy for malignant mesothelioma (22). We hypothesized that the CSC population found in malignant mesotheliomas from Nf2+/−;p16/p19+/− mice would be resistant to pemetrexed, and thus the proportion of CSCs, labeled as aldefluor-positive (21, 23), would increase after treatment. As predicted, treatment of malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice with pemetrexed significantly increased the percentage of aldefluor-positive cells, consistent with the idea that standard chemotherapies target cancer cells lacking CSC properties (Fig. 4C).
Tumor spheroids and lung-seeding tumors derived from malignant mesothelioma cells of Nf2+/−;p16/p19+/− mice harbor subpopulations of CSCs
To determine if tumor spheroids found in ascites of Nf2+/−;p16/p19+/− mice contain CSCs, sections of spheroids were stained for various stem cell markers. The spheroids stained intensely positive for Nanog and Oct 3/4, two stem cell markers (24), confirming the presence of CSCs; moreover, lung-seeding tumors from malignant mesothelioma cells of Nf2+/−;p16/p19+/− mice stained positively for four different CSC markers: β-catenin, Oct 3/4, Sox2, and Nanog (Supplementary Fig. S2). These findings suggest that CSCs are enriched in malignant mesothelioma cell lines derived from Nf2+/−;p16/p19+/− mice and contribute to the enhanced metastatic potential of these cells.
c-Met is upregulated and activated in malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice
As an initial step to understand mechanistically why malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice show enhanced metastatic potential and increased CSCs, we performed mRNA expression analyses using microarrays to identify genes that are differentially expressed in malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice versus from wild-type mice. A total of 929 differentially expressed genes were found, including numerous genes that regulate oncogenesis, angiogenesis, metastasis, and chemoresistance. Of these genes, we focused on the c-Met gene, which encodes a known regulator of cell motility and a potential therapeutic target in malignant mesothelioma (9, 25, 26). c-Met was found to be upregulated in malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice compared with that seen in malignant mesothelioma cells from wild-type mice (Fig. 5A). Moreover, both total c-Met protein expression and phosphorylation/activation were elevated in malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice (Fig. 5B) and in primary tumors from Nf2+/−;p16/p19+/− mice versus from wild-type animals (Fig. 5C).
c-Met upregulation is p53 and miR-34a dependent in malignant mesotheliomas from Nf2+/−;p16/p19+/− mice
c-Met upregulation in cancer cells can occur via gene amplification at the genomic level or by changes at the transcriptional/translational level (26). Recent work has shown that in cancer cell lines, c-Met upregulation/activation occurs through inactivation of p53 and the p53-regulated miR-34a, which normally downregulates c-Met mRNA and translation (27). A pilot microarray analysis revealed alterations of the p53 pathway, including upregulation of Mdm2 in malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice when compared with malignant mesothelioma cells from wild-type littermates. Thus, we decided to assess whether the p53–miR-34a axis is altered in malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice. As shown in Fig. 6A, we found that miR-34a was downregulated in malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice compared with that in malignant mesothelioma cells from wild-type littermates. Furthermore, reexpression of miR-34a in Nf2+/−;p16/p19+/− cells was sufficient to downregulate c-Met protein levels (Fig. 6B), consistent with a role for miR-34a in c-Met regulation in our model of metastatic malignant mesothelioma.
To determine if c-Met upregulation is p53-dependent, cell lines from Nf2+/−;p16/p19+/− mice were treated with etoposide to activate p53, and c-Met mRNA and protein levels were evaluated. Treatment with etoposide resulted in downregulation of c-Met mRNA and protein in malignant mesothelioma cell lines from Nf2+/−;p16/p19+/− mice (Fig. 6C and D). Etoposide treatment caused an upregulation of p53 protein levels, coinciding with downregulation of c-Met expression (Fig. 6D). In a complementary experiment, malignant mesothelioma cells from wild-type mice were nucleofected with siRNA against Tp53, and c-Met protein levels were examined. miR-34a levels were significantly downregulated when Tp53 was knocked down (Fig. 6E), and knockdown of p53 with siRNA caused a modest upregulation of c-Met protein levels (Fig. 6F), consistent with a model in which p53 regulation of miR-34a controls c-Met expression and translation in malignant mesothelioma, a mechanism not previously reported in this cancer.
c-Met activation is required for malignant mesothelioma cell migration/invasion and maintenance of the CSC population
To determine if upregulation/activation of c-Met contributes to the metastatic potential of malignant mesothelioma cells, we next performed functional studies using the Met inhibitor SU11274. Malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice were grown to confluence, and scratch assays were performed in the presence and absence of SU11274 to determine if c-Met activity contributes to wound closure/migratory capacity. As shown in Supplementary Fig. S3A, SU11274 treatment for 24 hours inhibited wound closure/migration of malignant mesothelioma cells from these mice. Immunoblot analysis demonstrated that SU11274 treatment blocks phosphorylation of c-Met and ERK1/2 in both malignant mesothelioma cell lines tested, although AKT phosphorylation, a known substrate of c-Met, was downregulated in only one cell line (malignant mesothelioma 87; Supplementary Fig. S3A). Although c-Met activity seems to be required for the migratory potential of malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice, potentially through the MAPK pathway, migration is only one aspect of metastasis (28).
To evaluate the contribution of c-Met activity to invasiveness, malignant mesothelioma cells from wild-type and Nf2+/−;p16/p19+/− mice were seeded in Matrigel transwells and treated with SU11274 or placebo to evaluate relative invasive capacity (Supplementary Fig. S3B). All 3 malignant mesothelioma cell lines derived from Nf2+/−;p16/p19+/− mice were more invasive than malignant mesothelioma cells from wild-type mice, consistent with c-Met activation correlating with invasiveness. Concordantly, treatment with SU11274 significantly reduced the invasiveness of all 3 malignant mesothelioma cell lines from Nf2+/−;p16/p19+/− mice (Supplementary Fig. S3B). These data suggest that c-Met upregulation/activation provides both a promigratory and proinvasive capacity to malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice.
c-Met activation has been linked to CSC formation in glioblastoma cell lines (11–13). To determine if c-Met activation is associated with the enriched CSC population found in malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice, malignant mesothelioma cells grown in a spheroid assay were treated with SU11274 and SGX-523, and both inhibitors significantly reduced the size of spheroids formed without affecting cell viability (Fig. 7A and C). This suggests that upregulation/activation of c-Met is required for maintenance of the CSC population. Furthermore, inhibition of c-Met with the less promiscuous small molecule inhibitor SGX-523 (29) decreased the percentage of aldefluor-positive malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice in a concentration-dependent manner (Fig. 7B). From these studies, we propose a model in which upregulation of c-Met through a p53- and miR-34a–dependent mechanism contributes to metastatic potential and maintenance of a CSC population in malignant mesotheliomas from Nf2+/−;p16/p19+/− mice (Fig. 7D).
Our findings represent the first direct evidence that haploinsufficiency for Nf2 and Cdkn2A can cooperate to drive an accelerated, highly aggressive form of asbestos-induced malignant mesothelioma. Somatic mutations of NF2 and CDKN2A occur frequently in human malignant mesothelioma, often in combination, and our data provide compelling genetic proof for the importance of these tumor suppressor genes in mesothelial cell carcinogenesis. These studies of asbestos-induced malignant mesothelioma are consistent with findings by Jongsma and colleagues, who used conditional knockout (CKO) mice to explore the importance of Nf2 loss and genetic lesions affecting Rb and p53 pathways in malignant mesothelioma development (30). In their system, various combinations of Nf2, Ink4a, Arf, and Tp53 were homo- or heterozygously excised from the mesothelial lining of the thoracic cavity using adenoviral delivery of Cre-recombinase (30). Malignant mesothelioma developed at high incidence in Nf2F/F;p16/p19F/F CKO mice with a median survival of 30 weeks, similar to the time frame seen in our asbestos-exposed Nf2+/−;p16/p19+/− mice. Mice in which only one of these tumor suppressor genes was homozygously deleted (F/F) developed spontaneous malignant mesotheliomas at a much lower rate, with spontaneous “malignant mesothelioma–like” thoracic tumors being identified in 5 of 30 Nf2F/F mice and 0 of 17 p16/p19+/− mice (30), supporting the idea that malignant mesothelioma development requires the combined involvement of multiple tumor suppressor gene alterations (31). Similarly, our asbestos-treated Nf2+/− mice developed less aggressive malignant mesotheliomas, with a longer latency, than the Nf2+/−;p16/p19+/− counterparts. Although our asbestos carcinogenicity studies more closely mimic the situation in humans, the fact that malignant mesothelioma arises spontaneously in CKO mice made deficient for one or both alleles of 2 or more tumor suppressor genes (Nf2, Cdkn2a, Tp53; ref. 30), is highly supportive of a genetic basis for malignant mesothelioma causation. In fact, germline heterozygous mutation of yet another tumor suppressor gene, BAP1, was recently shown to predispose to a high incidence of malignant mesothelioma in certain high-risk cancer families (32), and somatic mutations of BAP1 are common in human malignant mesotheliomas (8, 32, 33). Given their frequent alteration in human malignant mesothelioma, CDKN2A, NF2, and BAP1 seem to be primary drivers in this malignancy (8). Our in vivo experiments demonstrate that loss of both Nf2 and p16/p19 accelerates the onset and progression of asbestos-induced malignant mesothelioma, providing further evidence that mutations of these genes play a fundamental role in malignant mesothelioma pathogenesis.
Malignant mesotheliomas observed in Nf2+/−;p16/p19+/− mice were more aggressive than those found in Nf2+/− and wild-type mice. In the CKO studies by Jongsma and colleagues, Nf2F/F;p16/p19F/F mice showed increased pleural invasion compared with Nf2F/F;p53F/Fmice, but upon p16Ink4a loss in the latter mice, the median survival was significantly reduced and all tumors were highly invasive, implying that Ink4a loss contributes greatly to the poor prognosis characteristic of malignant mesothelioma (30). In our study, asbestos-induced malignant mesotheliomas developed at an accelerated rate in Nf2+/−;p16/p19+/− mice when compared with similarly treated Nf2+/− and wild-type mice, and malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice exhibited a more invasive and metastatic phenotype than their counterparts from wild-type mice.
Ascitic effusions from some Nf2+/−;p16/p19+/− mice, but not Nf2+/− or wild-type mice, harbored oncospheroids, a feature to our knowledge not previously reported in mouse models of malignant mesothelioma. Such spheroids have been reported in patients with human malignant mesothelioma and found to be more chemoresistant than adherent malignant mesothelioma cells (16, 17). We hypothesize that spheroids promote metastasis of the primary tumor to other locations in the peritoneal cavity. Indeed, intraperitoneal injections of malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice, but not wild-type mice, produced tumors that could penetrate the diaphragm and metastasize to the pleural cavity. Furthermore, tail vein injections of malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice produced a markedly increased rate of tumor seeding in the lung when compared with malignant mesothelioma cells from Nf2+/− or wild-type mice. This increased lung colonization was associated with increased c-Met expression/activation, which was also required for enhanced in vitro invasion/migration of these cells. c-Met encodes the hepatocyte growth factor receptor, which plays a role in cell migration through interaction with its ligand, HGF/scatter factor. Thus, it is not surprising that c-Met activation in malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice correlated with increased invasion, migration, and metastatic potential. MET upregulation/activation is common in human malignant mesothelioma and has been targeted with small molecule inhibitors in preclinical models of malignant mesothelioma (25, 34, 35). In malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice, c-Met upregulation was miR-34a and p53 dependent, a mechanism that has not been previously reported in human malignant mesothelioma. p53 has been shown to negatively regulate MET expression, in part by transactivation of MET-targeting miR-34 (36), and regulation of MET by both p53 and p53 regulation of miR-34 has been observed in other cancer types (27).
Previous work has demonstrated epigenetic alterations in malignant mesothelioma, suggesting that such events could play an important role in contributing to the pathogenesis of this disease. In support of the clinical relevance of our mouse model, it is notable that expression of miR-34a was found to be downregulated or absent in all 17 human malignant mesothelioma specimens examined by Guled and colleagues (37), and more recently, miR-34a was shown to be hypermethylated in human malignant mesothelioma cell lines and tumors (38, 39). Moreover, epigenetic changes are common in malignant mesothelioma generally, and asbestos exposure has been linked to hypermethylation of cell cycle genes in this tumor type (40, 41). Thus, there may be a direct link between asbestos exposure and hypermethylation of miR-34a seen in human malignant mesothelioma. However, in our model, transcriptional regulation of miR-34a seems to occur via a p53-dependent pathway. The identification of miR-34a as a recurrent player in both human and murine malignant mesotheliomas, albeit by different mechanisms, provides integrated evidence for a significant role of this particular microRNA in malignant mesothelioma tumorigenesis.
In the absence of DNA damage, the level of p53 protein is low in malignant mesothelioma cells from Nf2+/−;p16Ink4a/p19Arf+/− mice (Fig. 6D). Although the p53 gene (TP53) is infrequently deleted or epigenetically silenced in murine and human malignant mesothelioma, p19Arf is downregulated or absent altogether in malignant mesothelioma (5). We hypothesize that although lowered levels of p53 protein are retained in most malignant mesotheliomas, the protein may not be fully functional as a transcriptional regulator via a yet-to-be-determined mechanism(s). Moreover, it is noteworthy that mutations of the TP53 do occur in about 20% of human malignant mesotheliomas, and that mutations of TP53 and p14ARF tend to be mutually exclusive in these tumors (5).
The presence of tumor spheroids in ascitic effusions from some Nf2+/−;p16/p19+/− mice could be because of an enrichment of a CSC population that, in turn, contributed to the production of spheroids. CSCs from various human cancers have been isolated by growing tumor cells on nonadherent dishes in defined media (14). We showed that spheroids and malignant mesothelioma cell cultures derived from asbestos-exposed Nf2+/−;p16/p19+/− mice stain positively for known CSC markers, including Nanog, Oct 3/4, β-catenin, and Sox2 (16, 24). Furthermore, tail vein injections of malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice produced lung metastasis that stained positively for CSC markers, suggesting that the CSC population in these malignant mesothelioma cells contributed to the enhanced tumor spreading potential observed in vivo. In addition, malignant mesothelioma cells from Nf2+/−;p16/p19+/− mice formed CSC-like spheroids in vitro more readily than malignant mesothelioma cells from wild-type mice and expressed higher levels of the malignant mesothelioma–specific CSC marker CD24 (15, 16). Notably, c-Met activation was also recently shown to contribute to the CSC proliferation in other human cancers (11–13). We further demonstrated that c-Met activity is required for maintenance of CSCs in malignant mesothelioma lines from Nf2+/−;p16/p19+/− mice. These findings have implications for MET activation in human malignant mesothelioma, as they suggest that greater efficacy may be achieved by combining standard chemotherapies with a MET inhibitor to eradicate both bulk non-CSC tumor cell populations and CSCs, respectively.
In conclusion, our study indicates that inactivation of Nf2 and Cdkn2a cooperate to drive the development of aggressive malignant mesotheliomas characterized by enhanced invasive/metastatic potential and CSC-like features. Germane to this, a recent postmortem study of more than 300 patients with malignant mesothelioma revealed that widespread dissemination to various organs is common, indicative of metastatic potential (1), as was the case with malignant mesothelioma tumors and cell lines derived from Nf2+/−;p16/p19+/− mice. In our mouse model, such metastatic potential seemed to be closely linked to c-Met upregulation through a p53- and miR-34a–dependent mechanism, an observation not previously demonstrated in this cancer. Importantly, our in vivo findings provide strong genetic evidence indicating that cooperativity between losses of Nf2 and Cdkn2a plays a central role in driving the highly aggressive tumorigenic phenotype known to be a hallmark of malignant mesothelioma. Moreover, based on the high tumor incidence and short latency observed in asbestos-exposed Nf2+/−;p16/p19+/− mice, as well as the metastatic potential of malignant mesothelioma cells derived from these mice, this model has significant relevance for preclinical studies of novel molecularly targeted therapies in a cancer for which there is currently no effective treatment.
Disclosure of Potential Conflicts of Interest
J.A. Pachter is VP/Head of Research in Verastem. No potential conflicts of interest were disclosed by the other authors.
Conception and design: C.W. Menges, Y. Kadariya, J.A. Pachter, J.R. Testa
Development of methodology: C.W. Menges, Y. Kadariya, D. Altomare, J. Talarchek, Y. Wu, V.N. Kolev
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.W. Menges, D. Altomare, J. Talarchek, E. Neumann-Domer, I.M. Shapiro, V.N. Kolev, J.A. Pachter, A.J. Klein-Szanto
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.W. Menges, Y. Kadariya, D. Altomare, E. Neumann-Domer, G.-H. Xiao, I.M. Shapiro, J.A. Pachter, A.J. Klein-Szanto, J.R. Testa
Writing, review, and/or revision of the manuscript: C.W. Menges, Y. Kadariya, J.A. Pachter, J.R. Testa
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D. Altomare, J. Talarchek, Y. Wu
Study supervision: J.A. Pachter, J.R. Testa
This work was supported by NCI grants CA148805 and CA-06927 (J.R. Testa), a grant from the Mesothelioma Applied Research Foundation (J.R. Testa and C. W. Menges), an appropriation from the Commonwealth of Pennsylvania, and a gift from the Local #14 Mesothelioma Fund of the International Association of Heat and Frost Insulators & Allied Workers in memory of Hank Vaughan and Alice Haas.
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.