A Specific Splicing Variant of SVH, a Novel Human Armadillo Repeat Protein, Is Up-Regulated in Hepatocellular Carcinomas¹

HCC⁴ is the sixth most common cancer of men and eleventh most common cancer of women world-wide. However, HCC is the third cause of the cancer deaths in men and seventh in women (1). The geographic areas at highest HCC risk are eastern Asia and middle and western Africa (2). In China, HCC has ranked second in cancer mortality since the 1990s. Hepatitis B and C viruses, aflatoxin, and algal toxin in contaminated drinking water remain major etiological factors. Hepatitis G and transfusion-transmitted virus may also be involved in hepatocellular carcinogenesis (3).

Alternative splicing of gene transcripts presents in nearly all metazoan organisms as an important cellular mechanism that leads to the temporal and tissue-specific expression of unique mRNAs (4, 5). It is predicted that at least 50% of human genes are subjected to alternative splicing (6), suggesting that alternative splicing is one of the most significant components contributing to the functional complexity of the human genome (7). It has been shown that alternative splicing regulates expression and function of genes involved in almost every aspect of cell function including apoptosis (8), tumor progression (9), neuronal connectivity (10), and the tuning of cell excitation and cell contraction (11).

Many cancer-associated genes are alternatively spliced, such as Bcl-2 (12), Bcl-x (13), CD44 (14), BRCA2 (15), Wilms’ tumor suppressor gene (16), and cathepsin B (17). Alternative splicing affects not only the intracellular distribution but also the functional activity of these genes. In a number of cancer models, the ratio of the splice variants is frequently shifted so that the tumorigenesis-related splice variant predominates (18). For example, alternative splicing of CD44 results in up to 30 different spliced variants. In many human malignancies including thyroid, breast, and gastrointestinal carcinomas, certain CD44 splice variants are found to predominate (14). The specific CD44 isoforms that include variable exons 5 or 6 (CD44v5 or v6) appear to play significant roles in tumor metastasis (19). In addition, the breast cancer susceptibility gene, BRCA2, is alternatively spliced to exclude exon 12 (Δ12-BRCA2). The expression levels of Δ12-BRCA2 are elevated in steroid-negative breast tumors (15). These studies demonstrate that seeking different splicing formats is an important aspect in studying gene function and identifying molecular components involved in tumorigenesis.

In our group, genes whose expression levels were changed in paired HCC and liver tissue sample from the same patient were detected by RDA (20). Among them, a novel gene SVH, encoding an armadillo repeat protein localized in ER, was identified. Four splicing variants of SVH were observed. Interestingly, only one specific variant (SVH-B) was overexpressed in clinical HCCs, indicating its specific function related to cell growth and carcinogenesis.

Clinical HCC samples were collected from Zhongshan Hospital immediately after surgery and were diagnosed to stage II-III by pathological examination. The paired nontumorous liver samples were 3 cm away from the edge of HCC lesions from the same patients. These 46 HCC patients (male/female, 36/10) were 16–74 years old. The status of AFP, HBsAg, HBcAg, HCV, PCNA, VEGF, CD34, cytokeratin 7, cytokeratin 8, and EGFR were also examined. Access to human tissues complied with both Chinese laws and the guidelines of the Ethics Committee.

CDNA RDA was performed as described previously (20). The different products were cloned into the pGEM-5Z(+) (Promega, Madison, WI) and were confirmed by DNA sequencing (Sangon, Shanghai, China).

The cDNA sequence of SVH was obtained by aligning the ESTs from the NCBI database, and was confirmed by reverse transcription-PCR in the cDNA from the hepatoma cell line BEL-7404. Four spliced variants were found, and the coding sequence of each variant was amplified by LA Taq polymerase (TaKaRa, Dalian, China) and was cloned directly into pEGFP-N1 (Clontech, Palo Alto, CA) and pcDNA3.1A (Invitrogen, Carlsbad, CA), respectively. Primers used in this procedure were listed in Table 1. These constructs were verified by DNA sequencing.

To analyze SVH expression in various normal adult human tissues, the Multiple Tissue Northern blot (Clontech) was purchased. The blotting membrane was incubated in prehybridization solution of 5× SSC, 5× Denhardt’s solution, 0.1% SDS, and 50 μg/ml salmon sperm DNA at 68°C for 3 h, and then hybridization was performed using an 850-bp cDNA fragment of SVH-A (493–1342 bp) labeled with [α-³²P]dATP by random priming (Promega) at 68°C for 12 h. The membrane was washed in 0.1× SSC containing 0.5% SDS at 65°C for 1 h, exposed to a phosphor screen, and visualized by an FLA-3000A Plate/Fluorescent Image Analyzer (Fuji Photo Film, Tokyo, Japan). After stripping, the same membrane was similarly hybridized to the β-actin probe.

Total RNA was extracted from the paired HCC samples with TRIZOL Reagent (Invitrogen), and then was treated with amplification grade DNase I (Invitrogen) before cDNA synthesis. The mRNA for SVH and β-actin were measured by quantitative real-time PCR using the Rotor-Gene Real Time DNA Amplification System (Corbett, Sydney, Australia). Specific primers for amplification of SVH and β-actin were listed in Table 1. The primers for each splicing variant of SVH were designed to span exon junctions to prevent the amplification of other variants. The specific PCR products were confirmed by DNA sequencing and were used as templates to generate standard curves. Each assay included a no-template control, cDNA samples in duplicate, and the standard samples in duplicate. The reaction conditions were 94°C for 200 s, followed by 45 cycles at 94°C for 20 s, 58°C for 50 s, and 72°C for 20 s, and the data were acquired at 58°C.

Human hepatoma cell lines HepG2, Hep3B, SMMC-7721, BEL-7402, BEL-7404, and QGY-7703, and human liver cell lines L-02 and QSG-7701 were purchased from Type Culture Collection of Chinese Academy of Sciences, Shanghai. These cells were cultured in DMEM (Invitrogen) containing 10% fetal bovine serum, 100 μg/ml streptomycin sulfate, and 100 units/ml penicillin at 37°C in 5% CO₂.

The pEGFP-SVH constructs were transfected into QSG-7701 cells using the LipofectAMINE (Invitrogen) according to the manufacturer’s instructions. Expression of SVH spliced variants, fused with EGFP, was verified with rabbit anti-GFP polyclonal antibody (1:3000; Clontech). The transfected cells were seeded on coverslips and were cultured for 2 days. They were fixed and permeabilized in methanol for 5 min at −20°C. The rabbit polyclonal antibody against the human calnexin (1:300; Sigma, St. Louis, MO) was used as the primary antibody. The cells were stained with 1:100 dilution of Texas Red-conjugated goat antirabbit IgG (Fisher, Scientific, Cayey, PR). Specimens were visualized and photographed with a Leica DMR confocal microscope (Leica, Depew, NY).

The pcDNA3.1A-SVH constructs were transfected into QSG-7701 cells. Then, the transfected cells were switched to a selective medium containing 500 μg/ml G418 (Invitrogen). After 6 weeks of culture in the selective medium, clonal isolates were expanded, and expression of SVH fused with His₆·tag was verified with mouse anti-His₆·tag monoclonal antibody (1:2000; Santa Cruz Biotechnology, Santa Cruz, CA). To analyze the growth of QSG-7701 cells transfected with different splicing variants of SVH, the transfectants were seeded in 96-well culture plates at a density of 3 × 10³ cells per well, and the cell number were measured by a MTT (Sigma) method every day (20).

Cells (4 × 10⁶) in log growth phase were washed and suspended in 100 μl of sterile PBS on ice for s.c. injection into the right flank of 4–6-week-old BALB/c nude mice purchased from Shanghai Laboratory Animals Center of the Chinese Academy of Sciences. Mice were observed weekly for the visual appearance of tumors at injection sites, and tumors were harvested and weighed 8 weeks after injection. Each cell sample was in triplicate to avoid the individual differences in the nude mice.

Two groups of phosphorothioate ODNs were designed as shown in Table 2 and were synthesized by Sangon. The ODNs were freshly dissolved in 0.1 m PBS before use and were transfected with 20 μg/ml LipofectAMINE Reagent (Invitrogen).

Annexin V binding, PARP activity, and Bcl-2 expression were investigated to determine whether cell death was occurring by apoptosis. Annexin V-FITC binding and PI staining were performed according to the manufacturer’s protocol (Roche, Indianapolis, IN) and were analyzed by flow cytometry (Becton Dickinson, San Jose, CA; Ref. 20). To detect the PARP activity or Bcl-2 expression, 20 μg of total cellular proteins were run on 8 or 15% SDS-PAGE, electrophoretically transferred to a nitrocellulose membrane (Scheleicher and Schuell, Dassel, Germany), and blotted with mouse anti-PARP monoclonal antibody (PharMingen, San Diego, CA; 1:1000) or with rabbit anti-Bcl-2 polyclonal antibody (Santa Cruz Biotechnology; 1:2000). After binding a HRP-conjugated antimouse/rabbit antibody (1:3000; Amersham Little Chalfont, England), it was developed using the ECL luminescent system (Amersham). The membrane was reprobed by goat anti-actin polyclonal antibody (Santa Cruz Biotechnology; 1:2000) and then HRP-conjugated antigoat antibody (1:5000; Promega) as the loading control.

A modified cDNA RDA method was used to identify genes differentially expressed in HCC tissues (20). A 116-bp cDNA fragment up-regulated in the HCC sample was thus identified. The ESTs from NCBI database homology to this fragment were aligned into a 2647-bp cDNA cluster (GenBank accession no. AY150854) with an open reading frame encoding a putative protein of 343 amino acids (Fig. 1,A). The complete sequence of the putative gene, designated SVH (specific Splicing Variant involved in Hepatocarcinogenesis), was confirmed by reverse transcription-PCR and was verified by DNA sequencing (Fig. 1,B). The predicted mRNA had a typical Kozak consensus sequence preceding the ATG initiation codon, and a polyadenylation signal located 13 bp upstream of the polyadenylation site. The PROSITE analysis⁵ indicated that there were a possible armadillo repeat and a cell attachment sequence RGD from aa 33 to 35 (Fig. 1,A). Potential phosphorylation sites present in SVH included three protein kinase C sites and six casein kinase II sites. The SOSUI algorithm⁶ predicted a transmembrane domain at the NH₂ terminus of the protein (aa 7–29; Fig. 1 A).

The size of the SVH transcript and tissue specificity of its expression were estimated in a Northern blot (Fig. 1 C). The predicted length of 2647 bp was in good agreement with the major transcript size of ∼2.6 kb. It was also found that SVH was expressed in all of the tissues tested, with high levels in placenta, liver, kidney, heart, and brain.

BLASTP program⁸ revealed that the SVH protein product was closely related to the ALEX (Armadillo proteins Lost in Epithelial cancers on chromosome X) family, which contained three members: ALEX1, ALEX2, and ALEX3 (21). The homology region was confined to the COOH-terminal parts of these proteins over a length of ∼250 amino acids. In addition, the NH₂-terminal transmembrane domains of the four proteins were almost identical to each other (Fig. 2).

By comparing cDNA sequence of the SVH gene with the human genome sequences,⁹ the gene structure of SVH was assembled. It was shown that SVH gene was located on chromosome 7q11.22 and was composed of seven exons spanning a genomic sequence of ∼25 kb (Fig. 3 A).

To determine the expression status of SVH in human cell lines, we used reverse transcription-PCR with the primers P5–3 and P3–3 in Table 1. Interestingly, four bands were observed in all of the tested cell lines (Fig. 3,B). Each band was isolated, purified, and sequenced. It was indicated that the four bands resulted from alternative splicing in exon 2 and exon 5. On the basis of their molecular weight, the four splicing variants of SVH were defined as SVH-A, -B, -C, and -D (Fig. 3 C), and these four sequences were submitted to GenBank with the accession numbers of AY150854, AY150853, AY150852, and AY150851, respectively. It was also found that expression level of SVH-B was lower in the liver cell lines L-02 and QSG-7701 than in the hepatoma cell lines BEL-7402, BEL-7404, SMMC-7721, and QGY-7703 (the last was from the hepatoma tissue of the same donor as was QSG-7701; Refs. 22 and 23).

Human liver cell line QSG-7701 was used to characterize the subcellular localization of SVH proteins. Because variant-specific antibodies against SVH were not available, the four SVH variants, fused with EGFP in their COOH termini, were used in the localization assay (Fig. 4,A). Under a confocal microscopy, it was shown that the four variants shared the same localization characterization over the cytoplasm and nuclear envelope and that no apparent signals were detected in the plasma membrane (Fig. 4, B, E, H, and K). Immunofluorescence analysis with organelle-specific dyes was also performed (Fig. 4, C, F, I, and L). The majority of SVH signals codistributed with the signals of calnexin (Fig. 4, D, G, J, and M), known as an ER transmembrane protein (24).

To further analyze the effect of the transmembrane domain on the subcellular localization, we constructed SVH-T, a truncated SVH-A lacking the integrated NH₂-terminal transmembrane domain (deletion from aa 9 to 19). We observed that the SVH-T-EGFP fusion protein had the same distribution pattern as did EGFP (Fig. 4, N and O). These findings indicated that SVH was expressed predominately in the ER and that the NH₂-terminal sequence was necessary to the subcellular localization characterization of SVH.

Because the SVH gene was originally identified for its up-regulation in HCC, the expression levels of the four SVH variants were measured by quantitative real-time PCR in independent HCC samples and their paired normal liver tissues, calibrated by the expression level of β-actin. It was observed that the SVH-B was up-regulated by at least 1.5-fold in 28 of 46 HCC samples, whereas in the remaining 18 HCC samples, SVH-B was expressed at almost similar levels as in the paired liver tissues (Fig. 5,B). Moreover, the expression levels of other variants did not exhibit consistent association with HCC tissues (Fig. 5, A, C, and D), indicating that up-regulation of SVH in HCC was splicing-format-specific (P < 0.05, Pearson’s χ² test). Only SVH-B, not any of the other three formats, was up-regulated in clinical HCC.

Pathological information from the HCC samples was analyzed to detect whether they associated with an up-regulation of SVH-B. The results of using Fisher’s test suggested that the status of sex, age, and pathological stage and the expression levels of AFP, HBsAg, HBcAg, HCV, PCNA, VEGF, CD34, cytokeratin 7, cytokeratin 8, and EGFR were not significantly associated with the SVH-B mRNA level (data not shown).

To determine the effect of SVH-B overexpression on cell growth, the SVH-B expression construct was introduced into the normal liver cell line QSG-7701; SVH-A, -C, -D, and -T were used as controls. As shown in Fig. 6,A, the overexpression of SVH-A, -B, -C, -D, or -T in the transfected cells was confirmed by Western blotting using anti-His₆·tag antibody. The growth of SVH-B transfectants was significantly enhanced compared with that of control cells (Fig. 6,B), although the morphological appearance and the cell cycle distribution of the transfectants were indistinguishable from that of the control cells (data not shown). In addition, we examined whether overexpression of SVH-B would affect the tumorigenicity of QSG-7701 cells in nude mice. At 2 months after injection, tumors were visually identified at sites into which SVH-B transfectants had been injected, but no tumor was observed at sites into which control cells had been injected (Fig. 6 C). These results indicated that cells overexpressing SVH-B acquired accelerated cell growth rate and tumorigenicity in nude mice.

To examine the effect of down-regulation of SVH-B expression on hepatoma cells, we designed two groups of AS ODNs targeted to the specific exons (Table 2; Fig. 3,C). The inhibition of specific transcript expression could be achieved by a combination of these AS ODNs (Table 3). Control ODNs were also used, including SE ODNs that were complementary to AS sequences, and MI ODNs that were similar to the AS ODNs except for three or four mismatch nucleotides. For each transcript, two groups of AS ODNs were applied to inhibit the expression of each transcript to ensure the specificity of ODN treatment. A conclusion was drawn only when both AS ODNs against the same transcript had identical effects on the treated cells. The results of AS ODNs from Group 1 are shown in Fig. 7.

The AS and control ODNs were applied to hepatoma cell line BEL-7404 at the concentration of 5 μm. Measured by quantitative real-time PCR, the expression of each variant was specifically inhibited in AS ODN-treated cells but not in control ODN-treated cells (data for SVH-B are shown in Fig. 7,A, and data for SVH-A, -B, -C, and -D are represented in Table 3.). It was indicated that AS ODNs was a specific and effective approach to inhibit the expression of the spliced variants at the mRNA level.

The growth rate of BEL-7404 cells was examined at indicated times after AS ODNs treatment (Fig. 7,B). Specific inhibition of the expression of SVH-B decreased the cell numbers and then caused cell death. To investigate the mechanism of cell death induced by the inhibition of SVH-B, we measured annexin V binding (25), PARP cleavage (26), and Bcl-2 expression (27). We observed via flow cytometry that when the expression of SVH-B was inhibited, the binding percentage of FITC-labeled annexin V increased from 10% to 40% 24 h after the treatment; moreover, PI staining indicated that 20–30% of these cells became apoptotic (Fig. 7,C). We also isolated the total protein from the ODN-treated BEL-7404 cells 72 h after the ODN treatment. Western blotting showed that PARP, a M_r 116,000 nuclear chromatin-associated enzyme that was usually cleaved into M_r 85,000 and 25,000 fragments in apoptotic cells (26), was cleaved in the SVH-B-inhibited cells. Furthermore, Bcl-2 expression level was decreased in these cells (Fig. 7 D). These evidences suggested that the inhibition of SVH-B-induced apoptosis and continuous SVH-B expression was required for cell survival and growth.

Emerging data highlight the importance of alternatively spliced transcripts involved in tumorigenesis. Normal and neoplastic tissues showed differences both in the pattern of altered splicing and in the amounts of altered products. Splicing alteration may change the ligand binding of growth factor receptors (28) and cell adhesion molecules (14) and may alter the activation domains of transcriptional factors (16), the subcellular localization of the encoded protein (29), and even the phosphorylation activity of protein kinase (30).

The detail mechanism of altered splicing pattern in tumorigenesis is still unknown. Mutation at the original splice site or changes in the splicing machinery or splicing efficiency might contribute to this change in tumor development. Mutations around the intron and exon boundary would influence the possibility of splicing at the corresponding sites (31). It has also been suggested that illegitimate alternative splicing is a result of the abnormal stoichiometry of splicesome components (32). For example, down-regulation of splicing factor U2AF35 was critically involved in the misspliced form of the CCK-B receptor gene in pancreatic cancer cells (33). Furthermore, specific cellular environment plays a role in determining the nature of alternative splicing. The high temperature and low pH in the tumor tissue environment may exert an effect on the structure of the pre-mRNA and may lead to a reduced recognition by the splice machinery (34).

In our case, one of the specific splicing variant SVH-B, which lost exon 2 and exon 2 only, was up-regulated in 60% of clinical HCC samples, whereas, other splicing variants, including SVH-A, SVH-C, and SVH-D, did not exhibit the association with HCC. It was predicted that inclusion/exclusion of exon 2/5 would neither lead to a frame shift nor interfere with the integrated NH₂-terminal transmembrane domains and the COOH-terminal armadillo repeat of the four variants. It was also predicted that the exclusion of exon 2 from SVH-A might result in the loss of one protein kinase C phosphorylation site, three N-myristoylation sites, and one casein kinase II site; in addition, the inclusion of exon 5 in SVH-B generated a potential N-myristoylation site compared with SVH-D. It is speculated that these additional posttranslational modifications might determine differential activity of the corresponding proteins. Additional studies on the differences among the four spliced variants in their functional domains, protein structure, binding protein partners, and biological functions are warranted.

It was also observed that the four splicing variants of SVH were present in all of the samples examined, suggesting that multiple alternative splicing of SVH is frequent in both tumor and normal tissue. Alternative splicing may be relevant for the fine tuning of SVH actions, and balance among the different variants may modulate the overall SVH function. Molecular mechanisms on how splicing machineries are involved in carcinogenesis should be studied.

The significance of SVH-B in HCC was demonstrated by the overexpression of SVH-B gaining a fast growth rate and high tumorigenicity in nude mice, as well as by the inhibition of SVH-B causing cell apoptosis. It is suggested that the vast catalogue of cancer cell genotypes is a manifestation of six essential alterations in cell physiology that collectively dictate malignant growth: self-sufficiency in growth signals; insensitivity to growth-inhibitory signals; evasion of programmed cell death; limitless replicative potential; sustained angiogenesis; and tissue invasion and metastasis (35). Our data imply that SVH-B may be positively related to cell growth and tumorigenesis and may be involved in cell resistance toward apoptosis.

SVH encoded a novel human armadillo repeat protein, with significant similarities to the ALEX1, ALEX2, and ALEX3 genes. All of the four proteins had the NH₂-terminal transmembrane domains and the COOH-terminal armadillo repeat. Similar to the ALEX genes, SVH carried the armadillo repeat less than six times; thus, it was not a classical member of the armadillo-repeat family (36). However, the SVH gene was located on chromosome 7q11.22, and all three ALEX genes were located on the same chromosomal interval Xq21.33-q22.2. Moreover, alternative splicing was detected in the coding sequence of SVH, although the entire coding regions of ALEX genes resided in a single exon (21).

Notably, it was reported that expression of ALEX1 and ALEX2 was lost or significantly reduced in human lung, prostate, colon, pancreas, and ovarian carcinomas and also in the cell lines established from different human carcinomas. However, the involvement of the genes in HCC was not tested. Kurochkin et al. speculated that ALEX genes might play a role in suppression of tumors originating from epithelial tissues (21). It is very interesting that SVH and ALEX proteins are both related to carcinogenesis, although SVH-B is in positive correlation in HCC, whereas ALEX proteins are in negative correlation in various carcinomas. We noticed that SVH was highly expressed in human liver, whereas ALEX transcripts were barely detectable in liver (21). The armadillo repeat domain is thought to be involved in protein-protein interaction (37, 38, 39). Armadillo family members have revealed diverse cellular locations and have been implicated in a variety of processes including tumorigenesis (36). Functionally, the armadillo family members may play different roles in carcinogenesis. For example, β-catenin (40), δ-catenin (41, 42), and p120 (43) possess oncogenic activities, yet APC is a tumor suppressor gene (37). The different roles of SVH and ALEXgenes may be related to their cellular specificities and the protein partners interacting with specific armadillo family members. Seeking proteins that interact with the SVH-B variant but not with other transcription variants, thus, becomes a preferable approach in revealing the molecular mechanism related to the role of SVH-B in hepatocarcinogenesis.