TIP30 Deficiency Increases Susceptibility to Tumorigenesis

HIV-1 Tat-interacting protein TIP30, identical to the putative metastasis suppressor CC3, has been implicated in the regulation of tumor cell growth and metastasis (1, 2, 3, 4). Recent experiments demonstrated that a reduced level of TIP30 mRNA has been found in a number of tumor cell lines that include variant-small cell lung carcinoma (V-SCLC), classic-SCLC, neuroblastoma, colon cancer, and melanoma cell lines (2, 3, 4, 5, 6). Ectopic expression of the TIP30/CC3 gene in a variant SCLC line lacking TIP30/CC3 expression led to a significant suppression of the metastatic potential of these cells in SCID-hu mice. In addition, metastases in melanoma-bearing mice are significantly reduced after i.v. treatment with the TIP30/CC3 gene-cationic liposome-DNA complex (6). Subsequent work has led to the proposal that the suppression of cell growth and metastasis by ectopic expression of TIP30 may result, in part, from a propensity of cells to undergo apoptosis (2, 3, 4).

TIP30 was believed to be a transcription cofactor because it was demonstrated to potentiate Tat-mediated transcription in cooperation with components of elongation factor P-TEFb in transient transfection assays and is essential for Tat-mediated transcription in cell-free transcription assays (1, 3). TIP30 displays a serine-threonine kinase activity that can phosphorylate the carboxyl terminal domain of RNA polymerase II in a Tat-dependent manner in vitro (3). The intrinsic kinase activity is essential for TIP30 to enhance Tat-activated transcription and to sensitize NIH3T3 and v-SCLC cells to apoptosis (3). Consistent with the role of TIP30 in the suppression of tumor growth and metastasis via transcription mechanisms, studies revealed that the ectopic expression of TIP30 in v-SCLC cells and other tumor cell lines up-regulates the expression of proapoptotic factors (3) and angiogenic inhibitors, and down-regulates expression of angiogenic stimulators (4). Therefore, TIP30 may function as a tumor suppressor or tumor modifier that controls expression of genes involved in tumor growth and metastasis.

Although TIP30 has several characteristic features of a tumor suppressor, tumor development as a result of TIP30 inactivation has not been established in vivo, and abnormal expression of TIP30 in human cancer has not been reported. In this study, the generation of mice lacking Tip30 has allowed us to demonstrate a role of TIP30 in the suppression of tumor development in vivo. In addition, we have identified missense mutations in the TIP30 gene in the clinical human hepatocellular carcinomas (HCCs), which resulted in aberrant expression of TIP30.

To isolate the mouse Tip30 gene, a mouse 129SvJ genomic library (Stratagene, La Jolla, CA) was screened with a mouse Tip30 cDNA probe. Fourteen overlapping clones contained a 49-kb genomic region that included five coding exons of the Tip30 gene locus. The HindIII-SacI 2.7-kb genomic fragment was replaced by LacZ in-frame and a phosphoglycerate kinase neo-cassette (Fig. 1 A). This replacement ablated the two exons that encode the NH₂-terminal portion of TIP30, except for 10 amino acids after the translation start site. The targeting vector (7) included a 5.2-kb upstream homologous region and a 6.5-kb downstream region. Mutated Tip30 cDNA were generated by using pRSET-TIP30 (1) as templates and a site-directed mutagenesis kit (Stratagene). Plasmids containing Flag tag at the NH₂ terminal of TIP30 were generated by cloning wild-type and mutant Tip30 coding regions between NdeI and BamHI pFlag-7 (8). pCIN4-flag-TIP30, pCIN4-flag-TIP30^R106H, and pCIN4-flag-TIP30^G134V expressing flag TIP30 proteins were generated by cloning Flag-tag TIP30 cDNA between EcoRI and BamHI sites of pCIN4 (1). Plasmids expressing green fluorescent protein-tagged wild-type TIP30 fusion protein (GFP-TIP30), GFP-TIP30^R106H, and GFP-TIP30^G134V fusion proteins were generated by cloning wild-type and mutant TIP30 coding regions between EcoRI and BamHI sites of pEGFP-C2 (Clontech Laboratories, Palo Alto, CA).

E14 embryotic stem (ES) cells were electroporated with the linearized targeting vector and selected with geneticin on embryonic fibroblast feeder cells as described previously (7). In total, 348 G418-resistant clones were screened by Southern blot analysis using the 5′ external probe, and 65 clones displayed evidence for the homologous recombination of the disrupted Tip30 gene. Ten ES clones were microinjected into blastocysts of C57BL6/J female mice. Germ-line chimeras were bred to C57BL6J mice to generate heterozygous mutant F1 mice. All of the animal experimentation was performed according to the NIH guidelines in the Rockefeller University Laboratory Animal Research Center and the University of Nebraska Medical Center.

For genotyping, the genomic DNA isolated from ES cells or mouse tails was subjected to Southern blot analysis with a 5′ or 3′ external probe for lacZ or used for PCR analysis (primer sequences available on request). Mouse embryotic fibroblasts (MEFs) were prepared from E14.5 embryos obtained by heterozygous crossings of mice that had been backcrossed 10 times with C57BL6/J mice.

Northern blot analysis was performed as described previously (7). The full-length TIP30 cDNA was used as a probe, and labeled with [α³²p]dCTP and a random primer DNA labeling system (Invitrogen, Carlsbad, CA).

Primary Tip30^+/+, Tip30^+/−, and Tip30^−/− MEFs in passage 2 were used to introduce the SV40 T-antigen expression vector by LipofectAMINE reagent (Invitrogen) and be immortalized after 20 passages. For growth curves, immortalized cells (10⁴ per well of 12-well plates) were seeded, and cells were stained with trypan blue and counted daily. For soft agar assays, immortalized MEFs (5 × 10³) were grown in DMEM supplemented with 10% fetal bovine serum and 0.53% agarose on top of a layer containing 0.4% agarose for 21 days. The clones were stained with crystal violet, and the clones with diameters >1 mm were counted.

Tissues were fixed with 10% buffered-formalin and embedded in paraffin blocks. Tissue sections were deparaffinized, rehydrated, and stained with H&E. For immunohistochemistry, rehydrated sections were incubated overnight at 4°C with affinity-purified antihuman TIP30 antibody. Staining was developed using VECTASTAIN Elite ABC kit (Vector Laboratories, Burlingame, CA).

Genomic DNA was prepared from paraffin tissue blocks with the EX-WAX DNA extraction kit for paraffin-embedded tissue according to the manufacturer’s instructions (Intergen Co., Purchase, NY). Exon 3 of the Tip30 gene was amplified with Vent polymerase (New England Biolab. Inc., Beverly, MA). The primers used for amplification were 5′-GGCCTCCCAGCCTGCTAC-3′ and 5′-GTACTGCCAAAATAGCTAG-3′. The primers for the nesting PCR were 5′-GCTACCCAAGCCAAGTAGTAATC-3′ and 5′-GCTAGGGTTCTCCAAAGTG-3′. The first reactions were carried out for 25 cycles of 94°C, 30 s; 60°C, 30; and 72°C, 30 s. The nesting PCR reactions were carried out for 32 cycles of 94°C, 30 s; 60°C, 30; and 72°C, 30 s. The PCR products were incubated with Taq polymerase for 10 min at 72°C for adding A to the ends, and purified with Qiagen PCR purification columns (Qiagen, Valencia, CA), then cloned to the TA vector pCR2.1 (Invitrogen). Ten clones for each sample were sequenced for both DNA strands by the Eppley core facility at the University of Nebraska Medical Center using M13 reversal and universal primers. At least three clones containing identical mutations were sequenced for each patient. The HepG2 cell line was purchased from the American Type Culture Collection (Manassas, VA). Experiments in this study have been approved by the Institutional Review Board for the protection of human subjects at the University of Nebraska Medical Center.

For Western blot analysis in Fig. 3 B, HepG2 cells were transfected with pCIN4-flag-TIP30 and pCIN4-flag-TIP30^G134V using SuperFect reagents (Qiagen). After 48-h transfection, cells were harvested and whole cell extracts were prepared as described previously (3). Whole cell extracts were incubated with anti-Flag M2-agarose (Sigma, St. Louis, MO) at 4°C overnight. After brief centrifugation, supernatants were removed for Western blots with anti-β-actin antibodies. Precipitated proteins were washed and subjected to Western blots with anti-TIP30 antibodies. Fluorescence microscopy was performed as described previously (9). Before being fixed with 4% buffered paraformaldehyde, cells were grown in DMEM supplemented with 10% fetal bovine serum in the presence or absence of 20 nm of Leptomycin B (LMB; Sigma) for 3 h.

HepG2 cells were cultured in 150-mm dishes and transfected with pCIN4-flag-TIP30 and pCIN4-flag-TIP30^G134V using FuGene 6 reagents (Roche Diagnostics Corp., Indianapolis, IN) according to the manufacturer’s instruction. To inhibit protein synthesis, cycloheximade (100 μg/ml) was added to the medium, and cells were then harvested at the indicated time points. Cell lysates were subjected to immunoprecipitation and Western blot analysis as described above. Alexa Fluor 680 goat antirabbit IgG and goat antimouse IgG (Molecular Probes) were used for quantitative Western blot analysis. Li-Cor system was used to visualize and quantitate TIP30 protein and β-actin protein.

To investigate the roles of TIP30 in both cultured cells and living animals, we generated mice carrying inactivated Tip30 gene by homologous recombination. ES cell clones with a disrupted Tip30 locus were used to obtain germ-line chimeras that, in turn, were used to generate heterozygous F1 mutant 129SvJ/C57BL6/J hybrid mice (Fig. 1,A). Crosses between mice heterozygous for the Tip30 gene led to the expected Mendelian ratios of live born progeny, thus indicating that the Tip30 gene is not essential for development. The loss of TIP30 expression was confirmed by Southern blot analyses of tail genomic DNAs, and by Northern blot analyses of liver and stomach total RNAs (Fig. 1, B and C). Histological examinations of the Tip30^−/− mice before the age of 1 year did not reveal obvious gross alterations or developmental abnormalities in organs other than the mammary glands.⁵ Both Tip30^−/− males and females were fertile and generated normal litter sizes.

To determine whether TIP30 is a tumor suppressor, we have kept a cohort of TIP30-deficient and wild-type animals (50% of C57BL6/J and 50% 129SvJ of genetic background) for long-term observation of the development of spontaneous malignancy. At the age of 18–20 months, 31 females and 16 males of the F2 littermates were sacrificed. Autopsy and histological analyses revealed tumor development in 9 of 18 (50%) TIP30-deficient female mice and 2 of 9 TIP30-deficient male mice. In contrast, none of the 13 female and 7 male wild-type mice displayed tumors (Table 1,A; wild-type mice versus mutant mice; P < 0.005; X² = 11.3). The spectrum of tumor types in Tip30-mutant mice (Table 1 B) is distinct from that seen in older wild-type mice described by others (10, 11, 12), as many of the tumors found in TIP30-deficient mice were carcinomas. Among those tumors, HCC showed the highest incidence, constituting 30% of the total tumors. The relatively high incidence of HCC indicates that TIP30 has a prevalent role in tumor suppression in hepatocytes. Interestingly, 1 Tip30^+/− mouse had two tumors, a lipoid cell tumor at the right ovary and a neuroblastoma that arose at the right adrenal gland, with multiple metastases to tissues that included spleen, thymus, and salivary glands. The metastases were also found in the neck, mesenterium, and mediastinum. In addition, 1 mouse developed a transitional cell carcinoma in the urinary bladder, and exhibited metastases along the epithelium that involved the ureters and kidneys, whereas 2 male mice developed sarcomas that metastasized to the liver, spleen, and pancreas. These data demonstrate that TIP30-deficient mice are prone to tumor development.

To reveal the basis for the susceptibility of Tip30-mutant mice to tumor development, we analyzed Tip30 wild-type and mutant MEFs. In an examination of the growth rates of MEFs, no significant difference was detected (data not shown). We then asked if TIP30-deficient MEFs may be more susceptible to transformation by SV40 large T antigen, which is able to inactivate p53 and retinoblastoma protein (reviewed in Ref. 13). These cells were immortalized by ectopic expression of the T antigen, and growth rates were measured. Immortalized Tip30^−/− and Tip30^+/− MEFs grew much faster than immortalized Tip30^+/+ MEFs (Fig. 2,A). Tip30^−/− and Tip30^+/− MEFs also grew in a more disorganized fashion thanTip30^+/+ MEFs (data not shown). Therefore, we assessed the ability of each type of MEF to grow in soft agar. Tip30^−/− and Tip30^+/− MEFs showed a marked increase in the number and size of colonies as compared with Tip30^+/+ MEFs (Fig. 2, B and C). Western blot analysis revealed similar levels of T-antigen expression in Tip30^+/+ and Tip30^−/− MEFs (Fig. 2 D), and slightly higher expression of T antigen in Tip30^+/− MEFs. These results demonstrate that a lack of TIP30 in cells enhances transformation by the SV40 T antigen.

To determine whether our studies in mice are relevant to human disease, we analyzed 24 surgical specimens of human HCC and compared the expression of TIP30 in cancerous cells with expression in adjacent benign hepatocytes on formalin-fixed, paraffin-embedded tissues with antihuman TIP30 antibody (1). Undetectable or significantly decreased TIP30 expression was found in cancerous cells in 8 specimens (33%) in comparison with the adjacent tissues. An example of the immunohistochemical analyses is shown (Fig. 3,A). These data indicate that abnormal expression of TIP30 is implicated in human HCC. We next sought to examine whether these HCC cells harbor mutations in the TIP30 gene. Because the SNP database has listed four single nucleotide polymorphisms (SNPs) in the TIP30 coding regions, we first compared the TIP30 cDNA sequences in National Center for Biotechnology Information databases (14) to identify bp changes other than these SNPs in the human TIP30 coding region. Initially, we examined the sequences of TIP30 cDNA clones to identify insertion, deletion, or bp changes that will result in amino acid substitution. Besides those reported SNPs in the TIP30 gene that were found in both normal and cancer cells, we only identify missense mutations in exon 3. We found that none of the normal cells in the database (0 of 75) harbored a single base-pair substitution in exon 3, and 24% of various types of cancer cells (21 of 52) harbored at least a single base-pair substitution in the same region. Interestingly, the amino acid sequences encoded by Tip30 exon 3 are conserved among human, mouse, and Caenorhabditis elegans TIP30 proteins, and this region is important for the nuclear localization of TIP30.⁶ Therefore, we focused our mutational study on Tip30 exon 3 for those liver cancer patients who have abnormal TIP30 expression in their cancer cells. Three of 8 cancer samples contained at least a single bp change in the Tip30 gene (Table 1 C). These bp changes were not found in normal cells according to the National Center for Biotechnology Information database. In addition, whereas analyzing the sequences of cloned exon 3 DNA, we found that some of clones (20–80%) from the same cancer sample did not have any bp change, and individual clones from 1 HCC even harbored different bp changes. Importantly, we noted that 2 patients even had the same mutations in their carcinoma cells. Therefore, it is very likely that these base changes are somatic mutations that occurred during cancer development.

To explore the functional consequences of these amino acid substitutions in TIP30, we first investigated whether these mutations affected expression of TIP30 in cells. We transiently transfected a HCC cell line, HepG2, with mammalian expression plasmids encoding wild-type TIP30 and a mutant form of TIP30 with substitutions in G134V (TIP30^G134V) identified from the HCC showing in Fig. 3,A. On Western blot analysis showed in Fig. 3,B, TIP30^G134V was less expressed compared with the level of the wild-type TIP30. We then investigated whether this mutation affects the stability of the TIP30 protein in HepG2 cells. The relative level of TIP30 was monitored after addition of cycloheximide, a protein synthesis inhibitor. As shown in Fig. 3, C and D, G134V substitution markedly reduces the half-life of TIP30. This result suggests that G134V substitution reduces the stability of TIP30, thereby resulting in a lower level of TIP30 in cells. To investigate whether mutations affect the cellular localization of TIP30, we transfected HepG2 cells with mammalian expression vectors encoding GFP-TIP30 and mutant fusion proteins (GFP-TIP30^R106H and GFP-TIP30^G134V). The GFP tag had no influence on the cellular distribution of TIP30, because GFP-TIP30 had a subcellular distribution similar to endogenous TIP30 in HeLa and MCF7 cells (data not shown). In cells expressing wild-type GFP-TIP30 (Fig. 3,E, panel 1), ∼90% of transfected cells contained GFP-TIP30 in the cytoplasm, and 10% of transfected cells expressed GFP-TIP30 in the nucleus. When cells were treated with LMB, a CRM1-dependent nuclear export inhibitor (15), nuclear localization of GFP-TIP30 (Fig. 3,E, panel 2) was increased to ∼30% of transfected cells, suggesting that TIP30 is exported to the cytoplasm from the nucleus. However, cells expressing GFP-TIP30^R106H showed an exclusively cytoplasmic staining (Fig. 3,E, panel 5), and LMB treatment did not increase the nuclear localization of GFP-TIP30^R106H (Fig. 3,E, panel 6). This indicates that the R106H change may abrogate nuclear localization of TIP30 through inhibition of nuclear import. In contrast, GFP-TIP30^G134V (Fig. 3,E, panel 3) was expressed in both the cytoplasm and nucleus similar to the expression of GFP alone, suggesting that this mutation may affect its nuclear export. However, LMB treatment did not increase accumulation of GFP-TIP30^G134V in the nucleus (Fig. 3 E, panel 4), suggesting that the mutation may affect the nuclear import of TIP30. Therefore, it is possible that the mutation alters the protein structure, and results in abnormal nuclear import and export of TIP30. Given that TIP30 acts as a transcription cofactor in the nucleus and predisposes cells to apoptosis (1, 2, 3, 4, 5), it is conceivable that these mutations may impair the ability of TIP30 to regulate gene expression and apoptosis. Therefore, it is possible that TIP30 deficiency as a result of mutations in the TIP30 gene may inhibit its function in the liver cells and contribute to the pathogenesis of human HCC.

Cellular gene products can function as tumor suppressors or tumor modifiers to regulate tumorigenesis (16, 17). In this study, we showed that TIP30-deficient mice spontaneously develop tumors in their second year of life. Some tumors were carcinomas such as HCC that rarely occurs in 129SvJ/C57BL6/J mice (18). In addition, 1 Tip30 mutant mouse even had two different types of tumor. These data clearly represent a significant enhancement of susceptibility to tumorigenesis in TIP30-deficient mice. The same incidences of tumors in Tip30^−/− and Tip30^+/− mice is similar to the observations described in previous studies with DMP1, an Arf transcription activator, and p27^kip1, a potent tumor suppressor, that are haploinsufficient for tumor suppression (10, 19). We found that TIP30 mRNA and protein were expressed in both primary and metastatic sites of the neuroblastoma and hemangiosarcoma arisen in Tip30^+/− mice by reverse transcription-PCR and immunohistochemical analyses (data not shown). Although we are not sure that the second allele in the tumors of Tip30^+/− mice encodes a normal TIP30, based on the similar tumor incidence and latency in Tip30^+/− and Tip30^−/− mice, we suggest that TIP30 is haploinsufficient for tumor suppression.

The role of TIP30 in tumorigenesis is also suggested by the observation of reduced expression of TIP30 in human HCC and Tip30 mutations in those HCC. One mutation results in significantly decreased expression of TIP30, and another mutation changes cellular localization of TIP30. Therefore, it is possible that Tip30 deficiency as a result of mutations in the Tip30 gene may contribute to the pathogenesis of human HCC. Together, these results strengthen a conclusion that TIP30 plays an important role in tumorigenesis.

We thank C. Yang and the Transgenic Facility of the Rockefeller University for help with ES cell manipulation and blastocyst injection, K. Ge for SV40 T-antigen expression vector, P-C. Cui, and K. Wagner for technical help and useful discussions, and C. Eischen and A. Diehl for critical reading of the manuscript.