Aberrant Expression of HDMX Proteins in Tumor Cells Correlates with Wild-Type p53

Introduction

The Mdmx gene, the first described homologue of the Mdm2 proto-oncogene, was isolated from a murine cDNA expression library in a screen for p53-binding proteins (1) . The human homologue, Hdmx, was also cloned and was mapped to chromosome region 1q32 (2) . Northern blot analysis showed that the Hdmx gene is transcribed into two mRNAs of approximately 2.2 and 10 kb and encodes a protein of about M_r 80,000. The MDMX protein shares several functional domains with MDM2, in particular the NH₂-terminal p53-binding domain and the COOH-terminal zinc and RING finger motifs. The interactions of MDMX and MDM2 with p53 appear to be very similar, and they lead to inhibition of transcription activation by p53 (1 , 3) . However, in contrast to MDM2, MDMX does not trigger the degradation of p53 but actually prevents the MDM2-mediated degradation of p53 (4, 5, 6) . In tumors, overexpression of HDM2 is regarded as an alternative mechanism to inactivate the p53 tumor suppressor function. However, HDM2 was also reported to play a role in the process of tumorigenesis independent of p53 (reviewed in Ref. 7 ). In contrast to HDM2, little is known about a putative role of HDMX in carcinogenesis, although a recent report by Riemenschneider et al. (8) shows amplification of the Hdmx gene in a subset of malignant gliomas. Data on HDMX protein expression in tumors and tumor cell lines are not available at present. Therefore, we studied HDMX expression in a panel of human tumor cell lines and in normal cells. We found that the expression level of HDMX proteins is increased in a significant percentage of the tumor cell lines and that several cell lines express alternative HDMX proteins. Increased HDMX expression was also detected in the resected tumor material from which some of the cell lines had been derived, indicating that the increase is not necessarily a result of cell line establishment.

Materials and Methods

Cell Lines and Tumors.

All cells, unless stated otherwise, were maintained in DMEM supplemented with 10% fetal bovine serum and antibiotics. G401-6TG C6 cells (9) were grown in the same medium with the addition of 0.032 μm thymidine and 0.1 μm hypoxanthine. H1299 cells were grown in RPMI 1640 supplemented with 10% fetal bovine serum and antibiotics. Normal cervical epithelial cells (a gift from Ing. J. Eendenburg; Department of Pathology, Leiden University Medical Center) were maintained in keratinocyte-SFM (Life Technologies, Inc.) supplemented with 0.005 μg/ml human recombinant epidermal growth factor (Life Technologies, Inc.) and 0.05 mg/ml bovine pituitary extract (Life Technologies, Inc.). Dr. A. A. Schothorst (Department of Dermatology, Leiden University Medical Center) kindly provided normal melanocytes. Melanoma and ovary carcinoma cell lines generated in the Leiden University Medical Center Department of Clinical Oncology have been described previously (10, 11, 12) . Tumor material was obtained from the same department.

Immunoprecipitation, Gel Electrophoresis, and Western Blot Analysis.

Cells were lysed in Giordano buffer [250 mm NaCl, 0.1% Triton X-100, 5 mm EDTA, 50 mm Tris-HCl (pH 7.4)] or in IPB 0.14 [0.14 m NaCl, 0.5% NP40, 0.2% sodium deoxycholic acid, 20 mm tri ethanolamine (pH 7.8)]. Lysis buffers were freshly supplemented with protease inhibitors (2 μg/ml leupeptin, 1 μg/ml trypsin inhibitor, and 5 μg/ml phenylmethylsulfonyl fluoride), and protein concentrations were determined with Bradford reagent (Bio-Rad). Direct Western blot analysis was performed on 50 μg of whole cell extract, and immunoprecipitations were performed on 200 μg of whole cell extract using anti-MDMX rabbit polyclonal sera p55 or p56 raised against full-length recombinant HDMX protein. Immunoprecipitations and subsequent gel electrophoresis on SDS-polyacrylamide gels were performed essentially as described previously (6) .

After SDS-PAGE, proteins were electrophoretically transferred onto Immobilon-P/polyvinylidene difluoride membranes (Millipore). Membranes were first blocked in TBST 2 containing 10% nonfat dry milk (Nutricia) for 30 min at room temperature and subsequently incubated with the appropriate antibody diluted in TBST/5% milk for 1 h. After washing three times for 10 min in TBST, the membranes were incubated with the secondary antibody [either horseradish peroxidase-conjugated goat antimouse or goat antirabbit (Jackson Laboratories) diluted 1:10,000 in TBST/5% milk] for 30 min. Again, membranes were washed three times for 10 min in TBST, and proteins were visualized by the enhanced chemiluminescence protocol (Amersham). MDM2 proteins were detected with a mixture of SMP14 (Santa Cruz Biotechnology; 1:500) and 4B2 (Ref. 13 ; kindly provided by Dr. A. J. Levine; 1:5). HDMX proteins were detected with a mixture of 1:4 diluted supernatants of the hybridomas 6B1A, 11F4D, and 12G11G (6) . For the detection of p53, a 1:500 dilution of the DO-1 monoclonal antibody (Santa Cruz Biotechnology) was used. Actin was detected with C4 (ICN Biomedicals, Inc.). Quantification of the amount of HDMX was done with LumiAnalyst 3.0 software (Roche Molecular Biochemicals), and amounts of HDMX proteins in tumor cells were determined relative to those in normal cells. HDMX levels were considered normal for amounts between 0.5- and 2-fold of the level in normal cells. Levels more than 2 times the normal levels were considered to be increased, and levels lower than 0.5 times the normal levels were considered to be decreased. All experiments were repeated at least three times with comparable results.

Results and Discussion

Expression of HDMX Proteins Is Aberrant in a Significant Proportion of Tumor Cell Lines.

To study HDMX protein expression, we performed immunoprecipitations and/or Western blot analysis on lysates of tumor cell lines and compared the levels of HDMX in these cell lines with the levels in various normal cells. As shown in Fig. 1A ⇓ , of a panel of eight different tumor cell lines, three (U2OS, G401, and C33A) showed increased levels of HDMX as compared with normal skin fibroblasts (VH10) and human embryonic kidney cells (data not shown). All normal cells tested thus far express comparably low levels of HDMX, which is consistent with the finding that expression of the protein is low in most mouse tissues tested. 3 The levels of HDMX were highly increased in the C33A cells (a cervical carcinoma cell line).

Previous studies on both cervical and ovarian cancers showed frequent chromosomal aberrations on chromosome 1q (14, 15, 16) . Because the Hdmx gene has been mapped to chromosome 1q32, we set out to determine HDMX expression levels in another cervical carcinoma cell line, HeLa, and in a number of ovarian carcinoma cell lines. When levels of HDMX expression in these tumor cells were compared with those in normal cervical epithelial cells (Plav6), five of nine cell lines (COV644, COV413B, COV434, C33A, and HeLa) appeared to express increased levels of full-length HDMX (Fig. 1B) ⇓ . In addition to full-length HDMX migrating at approximately M_r 80,000, an alternative HDMX protein that resolves at approximately M_r 60,000 is detected in some cell lines (U2OS, C33A, COV644, COV434, and HeLa). This M_r 60,000 protein most likely originates from an alternative translation start at the second AUG because a product of the same size is detected in addition to full-length HDMX after transfection of the cDNA and after transfection of a construct designed to start translation at the second AUG. None of these products coimmunoprecipitate with p53 (data not shown), which is in line with the fact that they lack the NH₂-terminal p53-binding region. In Fig. 1, A and B ⇓ , expression of actin is analyzed to show comparable loading.

Besides cervical and ovarian carcinomas, malignant melanomas also commonly show chromosomal aberrations of chromosome 1 (17, 18, 19) . We examined a possible correlation between p53 status and HDMX expression by studying HDMX protein expression in normal melanocytes and in a panel of melanoma cell lines in which the p53 status had been determined in part (20) . In 4 of 13 melanoma cell lines, we detected increased levels of full-length HDMX (Fig. 2A) ⇓ . Direct comparison of the expression levels of full-length HDMX in the cervical carcinoma cell line C33A and the melanoma cell line 634 showed them to be equivalent (data not shown). In addition to full-length HDMX, some cell lines express alternative HDMX proteins resolving at M_r 47,000 (603), M_r 33,000 (634), or M_r 30,000 (normal melanocytes, FM6, FM3, IGR39, 530C1, and FM55p). The protein band of M_r 30,000 is not the previously described MDMX-S form (21) because our monoclonal antibodies do not recognize the NH₂-terminal part of HDMX that is encoded in the MDMX-S form. Also, it does not appear to be a tumor-specific form because it is detected in normal melanocytes.

Resected Tumor Samples Also Express Increased Levels of HDMX.

The possibility existed that levels of HDMX were increased as a result of tissue culturing and the establishment of the tumor cell lines, precluding the direct involvement of increased levels of HDMX in the process of tumorigenesis. Because samples of the original tumors used for the generation of some of the ovarian carcinoma and melanoma cell lines were available, we examined the levels of HDMX in these tumors. Again, expression of actin was analyzed as a control. In four of six tumors tested, the results were consistent with those of the cell lines. When HDMX expression was normal in the cell line, it was also normal to low in the tumor (Fig. 2B ⇓ , Lane 1, a melanoma metastasis from which cell line 8823 was derived; Lane 2, a primary melanoma from which cell line 453A0 was derived). When HDMX expression was found to be high in the cell line, it was also high in the original tumor (Fig. 2B ⇓ , Lane 3, a mucine tumor that gave rise to cell line COV644; Lane 8, a s.c. metastasis from a melanoma that gave rise to cell line 610III). In two cases (530C1 and COV434), where the cell line was derived from a metastasis, HDMX expression did not correlate with that in the metastasized tumor cells from which the cell lines were derived but with the expression in the primary tumor. The COV434 cell line, which expresses increased levels of HDMX, is derived from a metastasis (Lane 7, low levels of HDMX) of a germ cell tumor (Lane 6, high levels of HDMX). Conversely, the cell line 530C1 expresses low levels of HDMX, whereas the metastasis from which it is derived (Lane 5) expresses increased levels of HDMX. Here, the primary melanoma shows low HDMX levels (Lane 4). Perhaps, in these two cases, selection for a more adherent phenotype occurred during cell line establishment, resembling more the primary tumor cells than the metastasis. All in all, these results do indicate that during cell line establishment, selection for high levels of HDMX expression does not take place.

Aberrant Expression of HDMX Correlates with wt p53.

The expression levels of full-length HDMX were quantified, and the amount of protein relative to that in the normal cells was determined as described in “Materials and Methods.” In addition, the levels of p53 protein in the cell lines were also investigated, keeping the status (wt versus mutant) in mind. The results shown in Figs. 1 ⇓ and 2 ⇓ and those from some other cell lines that were studied for the expression of HDMX proteins are summarized in Table 1 ⇓ . When the amount of HDMX protein was more than 2 times higher than that in the normal cells, it is referred to as “high,” values between 0.5 and 2 times the amount in normal cells are referred to as “normal.” Lower values are referred to as “low.” All cell lines are arranged according to their p53 status. As can be seen, 8 of 12 cell lines with wt p53 show increased levels of full-length HDMX, and 3 cell lines more clearly express a shorter form of HDMX. Most of these cell lines show relatively high levels of wt p53, suggesting that HDMX expression can indeed affect p53 levels. In contrast, none of the cell lines lacking p53 and only two of the nine cell lines with mutant p53 have high levels of HDMX. Vice versa, 8 of the 10 cell lines with increased levels of HDMX and for which the p53 status is known express wt p53. This might indicate that overexpression of HDMX can function in the cell as an alternative mechanism to circumvent regulation of cell growth by p53. Because we and others have shown that HDMX can stabilize the HDM2 protein, the expression levels of the HDM2 protein in the cell lines were also investigated (Figs. 1 ⇓ and 2 ⇓ ). The results are summarized in Table 1 ⇓ . It shows that 9 of the 13 cell lines expressing increased levels of HDM2 contain enhanced levels of full-length HDMX, and 3 other lines (HepG2, 530C1, and 603) express an alternative HDMX protein. Conversely, 9 of the 13 cell lines that express increased levels of HDMX show relatively high levels of HDM2. Although the number of cell lines is low, and some exceptions have been found, these data do suggest some correlation between HDMX and HDM2 levels in the tumor cell lines.

In conclusion, we have shown here that in approximately 40% (13 of 31) of the tumor cell lines tested, levels of full-length HDMX are increased compared with that in normal cells. Expression of HDMX was found to be low to undetectable in 5 of these 31 cell lines. As far as the p53 status is known, these cell lines do not express wt p53. Moreover, a number of cell lines also express lower molecular weight HDMX proteins. All cell lines containing wt p53 express relatively high levels of HDMX and/or HDM2 or an aberrant form of HDMX. These results indicate a possible role for HDMX in tumorigenesis, complementary to that of HDM2, which is supported by the finding that HDMX is amplified in a subset of malignant gliomas that show neither p53 mutation nor Hdm2 amplification (8) . It has yet to be established whether deregulated HDMX expression actually plays a causal role in tumor formation. For the Hdm2 gene, differential splicing leading to synthesis of alternative HDM2 proteins was found to correlate with the malignancy of certain tumor types (22 , 23) . In human glioblastomas, the expression of alternatively spliced transcripts of Hdm2 was even reported to correlate with stabilization of wt p53 (24) . At present, experiments are under way to identify the mRNAs encoding alternative HDMX proteins. Although other mechanisms cannot be excluded, preliminary results suggest the presence of alternatively spliced Hdmx transcripts in several cell lines. Functional analyses of the HDMX proteins encoded by these transcripts might help to further elucidate the functions of these proteins in tumorigenesis.