High levels of MDM2 are not correlated with the presence of wild-type p53 in human malignant mesothelioma cell lines.

Prior analysis of 20 human mesothelioma cell lines for p53 status revealed only two mutations and one p53 null cell line, although p53 expression was detected in most cell lines. In addition, mRNA and protein expression of the retinoblastoma gene product in human mesothelioma cell lines is similar to normal controls. We have tested for p53 induction after exposure to ionising radiation and demonstrate this induction and, to a lesser extent, p21(WAF1) induction, in both normal mesothelial cells and p53-positive mesothelioma cell lines. We postulated that high levels of MDM2 might alter p53 and retinoblastoma tumour-suppressor function in mesothelioma. However, Southern blot analysis for mdm2 indicated that no amplification had occurred in 18 mesothelioma cell lines tested. Steady-state mRNA and protein levels also did not indicate overexpression. These results indicate that high levels of MDM2 are not responsible for inactivating the functions of wild-type p53 or the retinoblastoma gene product during the pathogenesis of malignant mesothelioma.

Mesothelioma, a cancer involving disregulation of mesothelial cell growth, has been linked epidemiologically to asbestos fibre exposure (Wagner 1960;Wagner and Berry, 1969;Craighead and Mossman, 1982). The 20 -50 year duration of the latency period for this malignancy suggests that it is generated by a multistep process induced by fibre exposure and involving chromosome breakage, rearrangement and deletions (Barrett, 1991;Hei et al., 1992), aneuploidy  and mutations secondary to damage by reactive oxygen species (Mossman et al., 1986). At a molecular level, the pathogenesis of the disease is likely to involve modulation of tumour-suppressor pathways that are critical for normal cellular regulation.
Alterations in the p53 tumour-suppressor gene have been identified as the most frequent genetic events in a wide variety of human cancers (Hollstein 1991;Levine et al., 1991;Vogelstein and Kinzler, 1992). In addition, retinoblastoma (Rb) gene dysfunction has been implicated in carcinogenesis (Benedict et al., 1990). The importance of p53 and Rb tumour-suppressor function is shown by the evolution, in DNA tumour viruses, of proteins which disrupt the function of these molecules. Interaction of p53 and Rb with SV40 T antigen (Farmer et al., 1992;Mietz et al., 1992;Jiang et al., 1993;DeCaprio et al., 1988), adenovirus Ela and Elb (Yew and Berk, 1992;Whyte et al., 1988), and papillomavirus E6 and E7 (Mietz et al., 1992;Dyson et al., 1989) has been shown to block wild-type functions of these molecules. Indeed, in recent work showing the presence, in mesothelioma specimens, of SV40 T antigen (Carbone et al., 1994), the authors suggest that latent SV40 infection of mesothelial cells may contribute to development of mesothelioma.
In contrast, the cell lines studied in this report have been shown previously to be negative for SV40 T antigen in a study in which all cells were stained with the polyclonal antibody to SV40 T, Pab 416, as an isotype-matched negative control for the p53 antibodies, Pab 1801 al., 1992). That study established, by sequencing, in most cases, exons 2-11 of the p53 gene, that the cell lines with wild-type p53 genes (18/20), as well as two normal samples examined, expressed a detectable level of p53 protein. One hypothetical explanation for this observation involves overexpression of a cellular protein, which can inhibit the function of the p53 protein.
The mdm2 gene, in mouse, has been shown to produce several sets of proteins Barak et al., 1993) resulting from alternative splicing (Wu et al., 1993;, and/or alternative promoter usage (Barak et al., 1994). It has been shown that initiation of translation from the third and fourth initiation codons results in molecules unable to bind to p53 Haines et al., 1994). The conservation of a p53-responsive element leading to alternative transcripts has been demonstrated in the human mdm2 gene (Zauberman et al., 1995), but its protein products remain to be characterised.
It has been suggested that MDM2 species that bind to p53 may be involved in an autoregulatory feedback loop (Wu et al., 1993;Perry et al., 1993;Barak et al., 1994;Picksley and Lane, 1993). This feedback control would function by stimulation of mdm2 transcription by p53 from the p53dependent promoter in intron 1, as opposed to the constitutive, upstream promoter (Wu et al., 1993;Barak et al., 1994;Zauberman et al., 1995). Increased levels of MDM2 protein reduce p53 stimulation of the p53 response element in the mdm2 gene (Wu et al., 1993), and inhibit the ability of irradiated colorectal carcinoma (RKO) and osteosarcoma (OSA-Cl) cell lines to arrest in the GQ stage of the cell cycle .
Previous work from this laboratory has demonstrated that both Rb mRNA and protein are expressed in malignant mesothelioma cell lines and primary cultures of NHM cells (Van der Meeren et al., 1993). In addition, the presence of immunohistochemically detectable Rb protein in paraffinembedded tissue from human mesothelioma specimens and normal human mesothelium has been reported (Ramael et al., 1994). These observations suggest that a downstream inactivator of Rb, as well as p53, could be crucial in mesothelial carcinogenesis. MDM2 has now been shown to bind to Rb as well as p53 (Xiao, ZX et al., 1995). The function of Rb would be additionally compromised by the binding of MDM2 to the transcription factors E2F1/DPI with consequent enhancement of their activation functions (Martin et al., 1995). Thus, disruption of cellular homeostasis, secondary to overexpression of mdm2, could lead to disregulated cell growth by: (1) inactivation of p53 tumoursuppressor activity; (2) interference with E2F sequestration on Rb; and (3) activation of the S phase promoting transcriptional factors, E2F1/DP1. We, therefore, proposed that overexpression of the mdm2 gene product in malignant mesothelioma could result in a functional inhibition of both the p53 and the Rb tumoursuppressor pathways, leading to diminished control of the GI checkpoint, even in the presence of functional p53 and Rb. A previous immunohistochemical study of mesothelioma demonstrated detectable MDM2 in six out of ten p53positive tumour specimens (Segers et al., 1995). We examined the contribution of mdm2 gene amplification and/or protein overexpression to mesothelial carcinogenesis by evaluating mdm2 DNA content, as well as steady-state levels of mRNA and protein in 18 of the mesothelioma cell lines previously characterised for p53 (Metcalf et al., 1992) and comparing these values with those obtained from primary cultures of normal human mesothelial (NHM) cells.
Materials and methods Cell lines and culture conditions Human mesothelioma cell lines were cultured as described previously (LaVeck et al., 1988). Detailed information on the history and derivation of the M prefix cell lines has been reported previously (Metcalf et al., 1992). Primary cultures of NHM cells were obtained from patients with non-malignant disease and initiated as described (LaVeck et al., 1988). OSA-Cl, a sarcoma cell line lacking p53 mutation (Leach et al., 1993) but exhibiting a 30to 40-fold mdm 2 gene amplification as well as overexpression of both mdm2 mRNA and protein (Oliner et al., 1992), was generously supplied by Dr Bert Vogelstein. This cell line, as well as human bronchial fibroblastic (HBF) cell strains, was maintained in Hut medium (Roswell Park Memorial Institute 1640 medium containing 10% fetal bovine serum; BioFluids, Rockville, MD, USA). Normal human bronchial epithelial (NHBE) cells were grown in LHC9 medium (BioFluids Inc.) as described previously .
RNA isolation and Northern blot analysis RNA was isolated from logarithmically growing cells by the method of Chomczynski and Sacchi (1987) and subjected to Northern blot evaluation as described previously (Metcalf et al., 1992). Filters were prehybridised at 65°C in Hybrisol I (Oncor, Gaithersburg, MD, USA) for a minimum of 2 h and then hybridised overnight with a random-primed 232Plabelled 900 bp (-312 to approximately +600 relative to ATG) XhoI restriction fragment from the MDM C14-2 clone of the human mdm2 gene, kindly supplied by Dr Bert Vogelstein (Oliner 1992). The filters were washed once at room temperature with 2 x standard saline citrate (SSC), 1% sodium lauryl sulphate (SDS), and then at 65°C twice with 2 x SSC, 1% SDS, twice with 1 x SSC, 1% SDS, twice with 0.5 x SSC, 1% SDS, and once at room temperature with 0.1 x SSC. RNA loading and integrity were verified by either concurrent or sequential probing of the filters with a random-primed a32P-labelled 900 bp PstI restriction fragment of a clone of the rat glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene (Fort et al., 1985), generously provided by Dr Marc Piechaczyk. Several exposures of each membrane were scanned on a Molecular Dynamics laser densitometer (Sunnyvale, CA, USA). Volumes for each band of mdm2 RNA were calculated and normalised to volumes for the GAPDH band in the same lane on the same filter. These ratios were then compared with mdm2/GAPDH ratios for the VAMT-1 mesothelioma cell line. Since VAMT-1 cells are null for p53 expression (Metcalf et al., 1992), these cells were chosen to provide a constitutive MDM2 expression level (Zauberman et al., 1995) for comparison. Interestingly, the MDM2 expression in this cell line was equivalent to that in the normal HBF cell strain used for protein normalisation. All expression ratios are arbitrary numbers and have not been normalised to an absolute concentration standard.
Protein extraction and Western blot analysis Lysates were prepared, from logarithmically growing cells, as the supernatant fraction of a 30 min, 4°C centrifugation at 38 600 x g in a buffer containing 50 mM Tris, pH 7.4, 150 mM sodium chloride, 1% (v/v) Triton-X-100, 1% deoxycholic acid (v/v), 0.1% SDS, 0.1 mM dithiothreitol and 0.57 mM phenylmethylsulphonyl fluoride. Routinely, samples containing 100 ,ig of protein, determined using the BCA Reagent (Pierce, Rockford IL, USA), were separated by 7.5% Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to Immobilon-P (Millipore, Bedford, MA, USA) membranes, probed with the anti-human MDM2 AB-1 (Oncogene Science, Uniondale, NY, USA) and detected by chemiluminescence using the Renaissance (Dupont/NEN, Boston, MA, USA) reagents and protocol. This anti-MDM2 antibody recognises an epitope in the amino terminus between residues 26 and 150 and would detect the p53binding protein species Haines et al., 1994). The membranes were reprobed, without stripping, with an anti-Rb rabbit polyclonal IgG, RB(C-15) (Santa Cruz Biotech, Santa Cruz, CA, USA). Protein loading for OSA-Cl was reduced to 25 Mg. A series of dilutions for OSA-Cl lysate, 10-100 pg (data not shown), established the linear range for protein densitometry and demonstrated that the Rb signal per jug lysate protein was similar for all mesothelioma cells tested. Since the Rb protein product is expressed at similar levels iln mesothelioma cell lines and NHM cells (Van der Meeren ct al., 1993), Rb was used as the internal control for protein loading. MDM2/'Rb ratios were calculated as described for mRNA and normalised to those obtained for HBF cells without conversion to an absolute concentration stanidard.
DNA isolation aci;l Soltlhcr;i blot analysis DNA was extracted, digested with EcoRI, and analysed by Southern blotting (Maniatis ct al., 1982). Membranes were probed with an f3'P-labelled nmdnI2 cDNA fragment, as described for RNA analysis. DNA loading and integrity were verified by reprobing with a randonm primed "P-labelled 900 bp Sau3AI fragmenit of the human single copy gene, J,,, generously provided by Dr Philip Leder (Ravetch et al., 1981a). The 4.6 kb and/or 8.8 kb banids of muhni2 DNA were normalised to values for the 22 kb JIJ band in the same lane on the same filter. These ratios were then compared with mdl,n2 J1, DNA content in HBF cells as described for Northerni blot analysis.

Results
Although the mesothelioma cell lines investigated here have been characterised for p53 expression (Metcalf et alt., 1992), assays indicating biological function have not been reported. Therefore, we tested the ability of p53 to be induced by DNA damiiagc, as demonstrated in most cell types (Kuerbitz et al., 1992;Lu and Lane, 1993). Selected mesothelioma cell lines with a basal level of detectable, wild-type p53 (Metcalf et al., 1992) were compared with NHM and the p53 null cell line, VAMT-1, for induction of' nuclear p53 and p2l1Al in response to ionising radiation. Figure 1 presents the data, which are illustrated photographically in Figure 2 for NHM cells and the mesothelioma lines, M33K and VAMT-1. The M33K cell line is representative of'mesotheliomas that express wild-type p53 at a low, but detectable, level in the absence of T antigen, while VAMT-l cells are null for p53 expression. Both M33K and VAMT-I are wild-type in exons 2 -I by sequence analysis (Metcalf'et al., 1992). Cells were scored as positive for inductionl, when p53 or p21" I'AF' showed nuclear accumulation as defined by the DAPI staiin. In both NHM and M33K cells, ionising radiation induces nuclear accumulation of p53 and _ ""'. Mairked cells in NHM and M33K panels illustrate positive, nuclear accumulation of the induced protein. The lower marked cell in the M33K was scored positive for p53 and negative for p2 1AF' induction. The VAMT-1 cells illustrate the appearance of background staining for both proteins. Figure 1 (Table I).
None of the samples showed inmbn2 gene rearrangements. The ;nd,n2 gene content of mesothelioma cell lines relative to HBF, ranged from 1.0 to 1.7 (Table I), suggesting that the ,ndmn2 gene is not amplified in malignant mesothelioma cell lines.
Since overexpression of mRNA could lead to increased levels of protein expression, four primary cultures of NHM cells, one primary culture of HBF cells and 18 malignant mesothelioma cell lines (Table I, Figure 3) were evaluated for steady-state levels of mdm2 mRNA. All of these specimens revealed a single message of 5.5 kb, consistent with the size of mdm2 mRNA previously reported (Ladanyi et al., 1993;Oliner et al., 1992) (Figure 3). The presence of multiple mRNA species (Zauberman et al., 1995) was not detected in these cells. Analysis of the four NHM revealed a range of mdm2 mRNA levels with three samples from 1.4 to 3.6 as much mRNA as VAMT-1, and one outlier at a 10.5-fold excess (Table I). Nothing in the donor history or culture characteristics of this NHM culture explained this observation. Steady-state levels of MDM2 mRNA in the malignant mesothelioma cell lines ranged from 1.1 to 5.1, relative to VAMT-1 (Figure 3, Table I).   Wt (2-11) Wt (2-11) Wt (2-11) Wt (2-11) Wt (2-11) Wt (2-11) Null (2 -11) Wt (2-11) Wt (2-11) Wt (5-9) Wt (4-11) Wt (2-11) Wt (2-11) Mt (4-11) Mt (4-11) Wt (2-11) Wt (5-9) Wt (2-11) ND ND Finally, since MDM2 protein expression may be controlled at the post-translational level , the steady-state level of MDM2 protein was studied in the mesothelioma cell lines, five NHM isolates, one culture of HBF cells and two primary NHBE cell strains by Western blot analysis (Table I, Figure 4). A single band of 90 kDa (Figure 3), similar in size to that previously reported for the MDM2 protein product (Leach et al., 1993;Oliner et al., 1992;Barak and Oren, 1992;Momand et al., 1992; and consistent with the single mRNA species, was observed in all samples tested. The five NHM cell samples demonstrated a protein range of 1.2 to 9.8, relative to HBF (data not shown) with the highest value corresponding to the cell strain with the highest mRNA level. The 18 malignant mesothelioma cell lines exhibited protein levels ranging from 0.1 to 5.8, relative to HBF (Table I). In general, the level of MDM2 GAPDH Figure 3 Steady-state MDM2 mRNA levels in human mesothelial cells and mesothelioma cell lines. Total cellular RNA (20,ug was analysed by Northern blotting as described in Materials and methods. The membranes were hybridised either serially or in combination with probes for MDM2 and GAPDH. Ratios of mRNA levels are shown in Table I relative Table I.  (Table I). The samples shown in Figure 4 are representative of steady-state protein expression in mesothelioma cell lines (lanes 1 -5), NHM (lanes 6 and 7), NHBE (lane 8), HBF (lane 9) and OSA-C1 (lane 10). All of the mesothelioma cell lines demonstrate MDM2 protein levels within the range of normal human cells, except for the VAMT-1 cell line, which shows a 10-fold reduction in MDM2 protein level when compared with HBF. This observation might indicate the presence of p53-stimulated mdm2 transcription in the other cell lines (Zauberman et al., 1995). Reprobing the membranes with anti-Rb antibody revealed a band at approximately 115 kDa (Figure 4). In other experiments (data not shown), addition of phosphatase inhibitors before cell lysis revealed the multiply phosphorylated Rb species in mesothelioma and HBF. Semi-quantitative comparison of Rb protein in mesothelioma cell lines and primary cultures of normal cells revealed similar amounts of the Rb protein product per jug protein analysed, relative to HBF (1.06+0.36 s.d., n=18 and 1.32+0.5 s.d., n=8 respectively).

Discussion
It has been reported that loss of p53 tumour-suppressor activity in a variety of human cancers with low frequency of p53 mutation, has been associated with amplification of the mdm2 gene and/or increases in MDM2 mRNA and protein levels (Leach et al., 1993;Oliner et al., 1992;Barak and Oren, 1992;Momand et al., 1992). An extensive study of soft-tissue sarcomas (Cordon-Cardo et al., 1994) indicated a complex pattern of mdm2 overexpression that did not result solely from gene amplification and existed even in the presence of p53 mutations and overexpression. Current data indicate that overexpression of the MDM2 protein could lead not only to (1) inactivation of p53 tumour-suppressor activity, but also to (2) interference with E2F binding on Rb and (3) activation of the E2F/DP1 transcriptional factors Martin et al., 1995), resulting in the loss of G, checkpoint regulation by p53 and Rb. It has been demonstrated that Rb is expressed in malignant mesothelioma cell lines and tumours (Van der Meeren et al., 1993;Ramael et al., 1994) (Figure 4) and that p53 mutation is an infrequent event (Metcalf et al., 1992). Furthermore, exposure of NHM and mesothelioma lines expressing wildtype p53 to ionising radiation induces nuclear accumulation of p53 and p2lWAFl in five of six lines tested (Figures 1 and   2). These data indicate that the p53 observed by immunohistochemical staining (Metcalf et al., 1992) is responsive to DNA damage and able to activate the p21 WAFi gene transcriptionally in most mesothelioma cell lines. Therefore, the possibility that the presence of mdm2 overexpression and/or amplification might be able to compromise functions of p53 and Rb and inhibit G1 checkpoint regulation was examined in a series of mesothelioma lines previously characterised for p53 and Rb expression (Metcalf et al., 1992;Van der Meeren et al., 1993). The relatively narrow range of mdm2 gene content, relative to HBF (  , 1989). Therefore, the variation observed is not likely to represent biologically significant overexpression. Since it has been demonstrated that the human mdm2 gene contains a functional, p53-responsive promoter (Zauberman et al., 1995), a correlation between levels of wild-type p53 and MDM 2 expression in mesothelioma cell lines might be expected. Elevated expression of wild-type p53 protein in the M9K, M32K and DND cell lines has been inferred from results of immunocytochemistry (Metcalf et al., 1992). However, these cell lines do not show consistently elevated mdm2 mRNA levels ( Table I). The p53 null VAMT-1 cells show a low level of MDM2 protein but not of mRNA (Table  I). Furthermore, the M15K and JMN lines, which contain mutant p53 genes, do not express low levels of MDM2 mRNA as might be expected if wild type p53 transcriptional activation were important in maintaining the steady-state mRNA level of this gene. These data are in agreement with observations of others who report a lack of correlation between levels of MDM2 mRNA and wild-type p53 protein in a subset of human gliomas and the murine cell line, C127 (Reifenberger et al., 1993;Perry et al., 1993;Gudas et al., 1995). This lack of correlation may reflect the complex regulatory interactions, which are being delineated for the genes and gene products involved in GI checkpoint regulation.
While loss of function of tumour-suppressor gene products has been implicated in the pathogenesis of a wide variety of cancers, no alteration in expression of either the p53 or Rb gene product has been demonstrated in this laboratory in a high percentage (85% for p53 and 100% for Rb) of malignant mesothelioma cell lines (Metcalf 1992; Van der Meeren et al., 1993). Additionally, the results reported here indicate that disruption of the wild-type p53 and Rb tumour-suppressor pathways by high levels of MDM2 protein is not a major factor in the aetiology of malignant mesothelioma. Thus, it is possible that uncharacterised downstream factors in the p53 and/or Rb tumoursuppressor pathways could be altered during the pathogenesis of malignant mesothelioma. Recent studies have revealed an increasing number of cellular proteins and DNA-binding proteins that interact with p53. These include XPB and XPD, components of transcription repair complex TFIIH (Wang et al., 1996); CBF, a CCAAT binding factor; heat shock protein 70; replication protein A; SPI, a general transcription factor; TBP, a TATA binding protein; and WT1, the Wilms' tumour gene product (for review see Pietenpol and Vogelstein, 1993).
Rb also has been found to interact with cellular protooncogenes, cell cycle-related proteins and other transcriptional factors, such as c-myc, N-myc, ATF-2, cdc2 proteins (for review see Goodrich and Lee, 1993), cyclin D2 and cyclin-dependent kinase 4 (CDK4) (Ewen et al., 1993). In addition, alterations of proteins that modulate the activity of these Rb-binding proteins, such as p16INK4, which inhibits CDK4 phosphorylation of Rb (Ewen et al., 1993), could be cellular components that participate in the loss of cell cycle regulation. Recent data indicate that a large proportion of mesothelioma cell lines (Okamoto et al., 1994;Cheng et al., 1994), as well as primary mesothelioma tumours (Xiao, S et al., 1995;Cheng 1994), have homozygous deletions of the p16 gene. Further studies of p16INK4 in primary mesothelioma tumours will be necessary to evaluate the physiological significance of the loss of p16INK4. In addition, other targets in these tumour-suppressor pathways need to be evaluated to advance the understanding of the pathogenesis of mesothelioma.