The Nuclear Pore Complex Protein Nup88 Is Overexpressed in Tumor Cells¹

Ovarian cancer is the sixth most common cancer among women in the United States and constitutes the fourth cause of death. Because of its asymptomatic early stage and the lack of an effective early-stage detection test, this cancer is the most lethal among all gynecological malignancies. Ovarian carcinogenesis involves multiple genetic changes, revealed as alterations in the expression of certain genes, particularly those related to cell cycle regulation, such as the tumor supressor genes p53, p16, and BRCA, and the oncogenes AKT2, ras, and c-myc. Understanding the mechanism of these changes could lead to using these genes as markers for the detection of ovarian tumors at an early stage.

In a previous report (1), we showed that the monoclonal antibody C6, raised against Candida albicans heat-shock mannoproteins, reacted specifically against a M_r 43,000-antigen present in human ovarian carcinoma cells but not in normal cells. These results suggested that the recognized antigen may be related to the process of malignant development. A cross-reaction between Candida krusei antigenic determinants and human tumor cells has also been described before by Yasumoto et al. (2). These authors reported that an antibody developed against Candida cytochrome c, specifically reacted with a protein of the cytoplasmic fraction of human lung cancer cells. However, the identity of the human proteins recognized by such antibodies remained to be determined.

Because the recognized antigens might play a role during the development of malignancy, we carried out the present study to identify the precise antigen detected by the monoclonal antibody C6. Using a cDNA library, we identified the protein recognized in mammalian cells as Nup88, a protein putatively involved in nuclear-cytoplasmic transport and located at the nuclear membrane (3, 4). In this report, we show that this gene is overexpressed in a large proportion of ovarian tumors and, thus, may serve potentially as a marker of malignancy.

A HeLa cDNA expression library cloned into phage λgt11 (Clontech, Palo Alto, CA) was immunoscreened using the monoclonal antibody C6 (1). The insertion of two positive recombinant phages was amplified by PCR, cloned into Bluescript, and sequenced. A 0.6-kb insert was subsequently labeled with [³²P]dCTP and used as a probe for a further screening of a HeLa cDNA library λNM1149/HeLa (a kind gift of Dr. Thomas Kahn, Deutsches Krebsforschungszentrum, Heidelberg, Germany). Three overlapping cDNA clones were identified and sequenced in both directions.

To produce antibodies against Nup88, the first 264 amino acids of the protein were expressed as a glutathione S-transferase fusion protein. SalI and XhoI restriction sites were introduced into the Nup88 cDNA by PCR. Primers containing restriction sites were as follows: 5-AGCAGTCGACCCAAGATGGCG-3 and 5-ATGCCACTACTCGAGCTTTGCCG-3. These primers were used to amplify a 784-bp fragment from the Nup88 open reading frame (nucleotides 511-1295) by PCR, and cloned into pGEX4T1 expression vector in SalI-XhoI sites (Amersham-Pharmacia Biotech, Uppsala, Sweden). The fusion protein GST-Nup88 was expressed in BL21 Escherichia coli and purified following the protocol described by Smith and Johnson (5). Polyclonal antiserum against Nup88 was obtained from rabbits injected with the recombinant protein.

SKOV3, OAW42, HeLa, SiHa, MCF7, BT20, HL60, and Jurkat cell lines were used. Human lymphocytes were used as a negative control. Ovarian tumor specimens were collected from patients undergoing surgery for ovarian cancer.

Proteins were separated by SDS PAGE, blotted onto nitrocellulose membranes, blocked, and incubated with Nup88 antiserum diluted 1:500 in PBS/10% nonfat milk. Bound antibody was visualized using the ECL method (Amersham-Pharmacia Biotech).

Cells grown on coverslips were washed twice in PBS, fixed, and permeabilized as described previously (6). Coverslips were incubated with anti-Nup88 serum diluted 1:1000 in PBS containing 1% BSA. After 5 washes with PBS, cells were incubated with antirabbit IgG antibody labeled with Rhodamine and observed with a Zeiss LSM510 confocal microscope.

Formalin-fixed, paraffin-embedded tumor sections were used for immunohistochemical analyses. We used a modification of the method as described previously (7). Briefly, the slides were deparaffinized, rehydrated in PBS, blocked, and incubated with rabbit anti-Nup88 polyclonal antibody diluted 1:1000. As a secondary antibody, we used a biotinylated antirabbit IgG antibody. Staining was performed with diaminobenzidine, and the slices were counterstained with hematoxylin and mounted.

RNA was extracted from cell lines or from normal human cervix using a RNA extraction kit (Qiagen, Hilden, Germany). Twenty χg of total RNA were electrophoresed and transferred overnight onto a nylon membrane. The 784-bp PCR product used for cloning into the pGEX vector was labeled with [³²P]dCTP by Random Priming (Amersham-Pharmacia Biotech) and used as a probe for Northern blot analysis. After exposure, blots were stripped and rehybridized with a [³²P]dCTP-labeled GAPDH³ probe. Hybridization signals were quantitated by a Phosphor Imager System (Molecular Dynamics, Freiburg, Germany)

It has been shown that a monoclonal antibody produced against a heat-shock mannoprotein of C. albicans recognized a protein in extracts of human ovarian tumors. We used this antibody to immunoscreen a human cDNA expression library. Two independent clones were detected containing fragments of 0.6 kbp and 0.2 kbp in length. A BLAST search of the sequence of these clones revealed 99% of identity with Nup88 (GenBank Y08612), which encodes for a novel nuclear pore component (4). Using the 600-bp fragment as a probe, we identified a 2.8-kbp cDNA clone in a human cDNA library, containing the complete coding sequence of Nup88 with additional 500 bp of the noncoding region.

The sequence corresponding to the first 264 amino acids of Nup88 was cloned into glutathione S-transferase vectors and fusion proteins were produced in Escherichia coli. The isolated M_r 56,000 fusion product was injected into rabbits and the antiserum was used for immunolocalization of the antigen. As shown in Fig. 1 for SiHa cells, most of the fluorescence was localized at the nuclear membrane, although a faint staining was also noticed in the cytoplasm. These results are in agreement with those published by Fornerod et al. (4), who used antibodies directed against the protein fragment comprised between amino acids 509 and 741.

To determine expression levels of Nup88 in normal and tumor cells, we carried out Northern blot analysis from several cell lines. A 784-bp fragment of Nup88 cDNA was radioactively labeled and used as a probe. In all of the cell lines investigated, a 2.5-kb-long transcript was detected. After hybridization, the blots were stripped and rehybridized with a radioactively labeled GAPDH probe to normalize values. We calculated the relative levels of expression by dividing the amount of Nup88 signal by that of GAPDH after quantitation by phosphoimaging. As shown in Fig. 2, the level of Nup88 mRNA expression in the tumor cell lines was always higher than in the normal cells.

These results prompted us to analyze Nup88 gene expression at the protein level. Proteins extracted from different cell lines were immunoblotted with Nup88 antiserum. In all of the samples analyzed a protein migrating at M_r 88,000 was detected, which corresponds with the molecular weight expected for Nup88 (4). As shown in Fig. 3 A, all of the tumor cell lines contained more Nup88 than the corresponding controls (lymphocytes and myometrium of healthy individuals).

To analyze whether the differences in gene expression found in the cell lines analyzed were also present in the primary human tumors, we prepared protein extracts from the ovarian tumors and performed Western blot analysis. Of a total of 21 tumors analyzed, 16 (76.19%) contained higher levels of Nup88 when compared with matching samples of healthy tissues of the same patient. The rest of the tumor samples yielded a heterogeneous pattern; some of them contained the same levels of Nup88 (19.05%), whereas for the rest (4.76%), the level in normal tissues was larger than in the corresponding tumor sample. As shown in Fig. 3 B, the three tumor samples used in our experiments revealed higher levels of Nup88 expression when compared with normal myometrium samples from the same patients. These results, therefore, suggest that increased Nup88 expression is a phenomenon associated not only with in vitro growth but also with human primary tumors.

To further substantiate this finding, we performed immunohistochemistry of several human ovarian tumor samples with Nup88 antiserum. Fig. 4 shows that a high level of Nup88 expression is observed in the malignant epithelial cells, whereas no, or only very low, expression is visible in the stroma cells. Interestingly, most of the staining is observed at the nuclear membrane and at the cytoplasm in the epithelial cells, whereas in the stroma cells, staining is mostly observed at the nuclear membrane.

It has been shown that a monoclonal antibody (C6), raised against C. albicans, reacted with human ovarian proteins (1). Now, we have identified the antigen reacting with this antibody in human cells as nucleoporin Nup88, a novel described protein related to nuclear transport and located at the nuclear membrane (3, 4). We performed a Western blot analysis with the C6 antibody and all of the samples shown in Fig. 3. Under the conditions used in our experiments, the C6 antibody recognizes Nup88 protein as well as our polyclonal antibody does (data not shown).

Our immunofluorescence analysis shows that Nup88 is located mainly at the nuclear membrane and, to a minor extent, at the cytoplasm. These results are in agreement with those of Fornerod et al. (4), who also demonstrated Nup88 to be localized to the nuclear membrane (4). In contrast, our immunohistochemistry studies revealed a predominantly cytoplasmic accumulation of Nup88 in tumor cells (Fig. 4). In rat cells that overexpressed Nup84 (the homologue to Nup88) a large accumulation in the cytoplasm has been demonstrated, which suggests that the residual protein was not degraded but accumulated, perhaps as structures, in the cytoplasm (8). This is further reinforced by experiments in which the amount of Nup88 was measured in nuclei and cytoplasm cell fractions. We observed that a variable amount of the protein was associated with the cytoplasm rather than with the nuclear membranes.⁴ Interestingly, it has been shown that in cells growing rapidly, large amounts of “annulate lamellae” are present in the cytoplasm. These structures resemble fragments of nuclear membranes and are considered to be remnants of the membrane that may be used for nuclear reconstitution during the following mitosis (9). Nevertheless, it seems that this is not a general feature, and some tumors have been shown to contain only small amounts of lamellae.

When we analyzed the protein extracts that were isolated from human ovarian tumors, we detected an apparent overexpression of Nup88. Although the total number of analyzed tumors (21 in these experiments) was low, it seems that Nup88 gene expression is increased in a large proportion of cases and may be, therefore, considered as a potential marker of malignancy. The reasons for this overexpression are unknown, but our RNA hybridization experiments show an increased amount of specific RNA, which suggests a transcriptional control of the gene. These data were obtained using established cell lines, and, thus, it is not possible to postulate the same control mechanism for Nup88 gene expression in human tumors.

Although the function of Nup88 continues to remain elusive, it has been found that the protein is associated with the central domain of CAN/Nup214, a nuclear pore complex component that has been suggested to be involved in nuclear protein import, nuclear mRNA export, and cell cycle regulation (10). CAN was first identified as an oncoprotein because of its relation with two types of leukemia, although its contribution to leukemogenesis is not known. CAN/Nup214 proto-oncogene is involved in two different chromosomal rearrangements related to acute myeloid leukemia and undifferentiated leukemia (11, 12). When these chromosome translocations occur, the binding region for Nup88 in CAN/Nup214 is lost (3). Whether the accumulation of Nup88 is due to the loss of its target partner (the CAN/Nup214-binding domain) remains to be studied; the status of CAN/Nup214 in the cell lines and tissues described in this report needs to be studied as well.

In addition, Nup88 maps to 17p13 (13), a region known to be involved in genetic changes in ovarian tumors. We can only speculate whether a relationship can be established between Nup88 and the putative oncogene located at 17p13.

Taken together, our results show that Nup88 is overexpressed in a series of tumor cell lines and in primary human ovarian tumors, when compared with the corresponding healthy tissue. The reasons and the mechanisms for this overexpression, and whether these mechanisms are linked to increased cell proliferation, are unknown, but the results shown in this communication suggest that Nup88 may be involved in the oncogenic activation of tumor cells and could be a marker of tumor growth.