Aberrant Splicing of Hugl-1 Is Associated with Hepatocellular Carcinoma Progression

Cell polarity is a fundamental property of tissue architecture (1, 2). Many cell types in the human body, ranging from neurons to epithelial cells, must be polarized to function properly (3). Loss of cell polarity in epithelial cells is one of the hallmarks of cancer and is correlated with more aggressive and invasive cancers (4, 5). Recent evidence indicates that impairment of cell polarity can function either as an initiating event or as a cooperating event during carcinoma development (6, 7).

The hepatocyte is a polarized epithelial cell whose surface is composed of three distinct domains with different functions: the sinusoidal membrane, the lateral membrane, and the bile canalicular membrane (8, 9). Thus, the hepatocyte is polarized not only in morphology but also in function. Recently, it has been suggested that dysfunction of cell polarity proteins might be associated with liver cancer progression (10). Not surprisingly, an apolar arrangement is thought to be a key step during hepatocarcinogenesis (9). Thus, understanding the role of cell polarity proteins is important for understanding hepatocellular carcinoma (HCC) biology.

Lethal giant larvae (Lgl), an evolutionarily conserved and widely expressed cytoskeletal protein, is indispensable for the establishment and maintenance of cell polarity and is a regulator of cell proliferation (11). Lgl, dlg (discs large), and scrib (scribble) have been identified as neoplastic tumor suppressor genes in Drosophila and act in a common pathway to link cell polarity with proliferation (12–14). In lgl mutant, imaginal discs showed massive tumor-like overgrowth, and neuroblasts and ganglion mother cells in the developing nervous system exhibited excessive proliferation. The tissues of lgl mutants exhibited many similarities to human tumors, including a loss of tissue architecture and cell shape and a failure to differentiate (15). Moreover, Drosophila cells with lgl mutation exhibited invasive behavior: the tumor cells carrying the lgl mutation could propagate and migrate to remote sites in a manner similar to mammalian metastatic tumors, leading to the death of the host (16). It has also been shown that lgl inactivation combined with RasV12 expression promotes metastatic behavior, suggesting the role of loss of lgl in tumor metastasis (17).

The human homologues of lgl are called Hugl-1 (Llg11) and Llgl2. The Hugl-1 protein shares 62.5% similarity with Lgl, wherein conservative amino acid changes are considered. Like Lgl, Hugl-1 is mainly expressed in the cytoplasm, regions of cell junctions on the inner side of the cell membrane (18). A recent study has shown that Hugl-1 can substitute for the Lgl tumor suppressor function in vivo (19). The high degree of functional and sequence conservation of Lgl during evolution implies its potential role in tumorigenesis.

Accordingly, recent studies have shown that Hugl-1 transcripts are reduced or absent in a high proportion of breast cancers, lung cancers, prostate cancers, ovarian cancers, and melanomas (19–22). Moreover, reduced Hugl-1 transcription has been associated with advanced stage cancer, particularly with lymph node metastases (20, 22), and functional assays have revealed that Hugl-1 overexpression increases cell adhesion and decreases cell migration (20, 21). These findings suggest that down-regulation of Hugl-1 plays an important role in various types of human cancers. However, these data provide only indirect evidence for the tumor suppressor role of Hugl-1 in human cancer. Thus far, no mutations of Hugl-1 have been discovered in human cancers. This, to a great extent, hinders the understanding of the mechanism of Hugl-1 in tumorigenesis.

In the present study, sequence alterations of Hugl-1 were investigated in five HCC-derived cell lines and a cohort of 80 HCC specimens and matched noncancerous liver specimens. We found a high frequency of Hugl-1 transcripts with aberrant splicing and point mutations in HCC specimens and cells. In addition, the presence of abnormal Hugl-1 was clearly associated with large tumor size and poor differentiation of HCC.

Patients and tumor samples. In total, we collected 80 tumor tissue samples from patients diagnosed with HCC, who were admitted to Eastern Hepatobiliary Surgical Hospital between 1999 and 2001 or to Zhongshan Hospital, Fudan University between 2004 and 2005. The access to human tissues complied with both Chinese laws and the guidelines of the ethics committee. Paired nontumorous liver samples were obtained from a site 3 cm away from the edge of the HCC lesions.

The age of 60 HCC patients (male/female, 52:8) ranged from 25 to 72 y, with an average age of 49.2 y. α-Fetoprotein status and hepatitis B surface antigen (HbsAg) status were also examined. There are 86.7% (52 of 60) patients showed the HBsAg positive. The hepatitis C virus (HCV) status was not examined. The tumor size and the histologic grade of the 60 patients are listed in Supplementary Table S1. The baseline information of the remaining 20 patients was unavailable. Additional 11 HCC samples were used exclusively for Western blot analysis on Hugl-1 proteins expression and identification of truncated Hugl-1 proteins.

Total RNA and genomic DNA isolation. Total RNA was extracted using TRIZOL Reagent (Invitrogen Life Technologies) according to the manufacturer's instructions. Genomic DNA was obtained after total RNA was extracted according to the manufacturer's protocols.

Reverse transcription-PCR and mutation screening. cDNA was synthesized using total RNA and oligo(dT)12-18 with Moloney murine leukemia virus reverse transcriptase (Promega Corporation). Two overlapped fragments covering the reading frame of Hugl-1 were amplified by nest PCR for HCC and matched noncancerous tissues. PCR amplification was done using LA Taq polymerase (TaKaRa). The PCR conditions were 94°C for 1 min, 60°C for 1 min, and 72°C for 2.5/1.5 min for 30 cycles. After the external primers were used, a 2-μL aliquot was transferred directly to a new reaction tube for second round PCR using internal primers.

The primer sequences were used as follows (Supplementary Fig. S1):

For exons 1 to 13 of Hugl-1 (a 1982-bp PCR product),

• the first round PCR are

• Pa: 5-CGGCAAGATGATGAAGTTTC-3

• Pb: 5-AGAGACACGACTCTTGCGAATG-3

• and the second round PCR are

• P1: 5-AGTTTCGGTTCCGGCGGCAG-3

• P2: 5-AGACTGGCGCAGTGACTTCTTG-3.

• For exons 13 to 19 of Hugl-1 (a 1240-bp PCR product),

• the first round PCR are

• Pc: 5-ACTCTTCACCCCAATGACTC-3

• Pd: 5-TTCTGGCCTTCCTCATTTGATC-3

• and the second round PCR are

• P3: 5-CTCCCGGGTGAAGTCTCTCA-3

• P4: 5-GATGGCACAGGCGTGGGCCT-3.

PCR was also used to assess the Hugl-1 gene for genomic DNA deletions. The primers used for genomic DNA analysis are listed in Supplementary Table S2.

Sequencing of Hugl-1 transcripts. To verify the sequence of the normal-sized and abnormal-sized Hugl-1 PCR products, they were excised from agarose gel, purified using a DNA fragment purification kit (BioDev-Tech Co., Ltd.), and then directly sequenced (Invitrogen Biotechnology Co., Ltd.).

Cell culture. The HCC cell lines Hep G₂, Huh7, and BEL-7404 were cultured in DMEM (Invitrogen) containing 10% fetal bovine serum (Invitrogen). The SMMC-7721 cells were cultured in RPMI 1640 (Invitrogen) with 10% fetal bovine serum. The SK-HEP-1 was kept in Eagle's MEM (Invitrogen) containing 10% fetal bovine serum. All cells were maintained at 37°C in 5% CO₂.

Antibodies and Western blot. The mouse antibodies against Hugl-1 were raised in BALB/c mouse against GST-Hugl-1 (amino acids 888-1018). The purified rabbit polyclonal antibodies raised against synthetic 24 COOH terminal peptides of Hugl-1 (18) were kindly provided by Dr. Dennis Strand, Johannes Gutenberg University. The rabbit polyclonal antibodies against EGFP were purchased from SantaCruz Biotechnology. HCC tissues were lysed in radioimmunoprecipitation assay buffer (50 mmol/L Tris-HCl, 150 mmol/L NaCl, 1% Triton X-100, 0.1% SDS, 1% deoxycholic acid sodium) and sonicated. Clarified lysate supernatant was immediately shock frozen and stored at −80°C. Samples were fractionated by SDS-PAGE gels and subjected to Western blot analysis with appropriate primary and horseradish peroxidase–conjugated secondary antibodies.

Molecular cloning and stable expression of Hugl-1 and its aberrantly spliced variants. The wild-type Hugl-1 (Hugl-1/wt) gene was amplified from a cDNA of human embryonic kidney 293 cells, and Hugl-1/Δ1190-1508 was directly amplified from the SMMC-7721 cDNA. The other aberrantly spliced variants (Hugl-1/Δ1414-1567/1667-1959) were directly amplified from the patient cDNA. The coding sequence was cloned directly into pEGFP-C1 (Clontech). Constructs were transfected into SK-HEP-1 cells using Lipofectamine 2000 (Invitrogen). Subsequently, 500 μg/mL G-418 sulfate (Invitrogen) were used to select the transfected cells. After 6 wk, stable clones were obtained, and the expression of EGFP-tagged Hugl-1, Hugl-1/Δ1190-1508, and Hugl-1/Δ1414-1567/1667-1959 was verified.

Wound healing assay. The migration of cells was evaluated using a wound healing assay. Cells were grown to confluence in six-well plates. Then linear wounds were created by scraping confluent cell monolayers with a pipette tip in a definite array. The migration of cells into the wound area was documented and evaluated after 12 and 24 h. Quantitative analysis of the percentage of wound healing was calculated using distances across the wound (n = 20) at 0 and 24 h, divided by the distance measured at 0 h for each cell line.

Boyden chamber assay for migration and invasion. Quantitative cell migration and invasion assays were done using 12-well Boyden chambers containing polycarbonate filters with a pore size of 8 μm (Neuroprobe). The filters coated with Matrigel (40 μg in 20 μL; BD Biosciences) were used in the invasion assay. The lower compartment was filled with 10% fetal bovine serum–supplemented MEM. The SK-HEP-1 cells were resuspended in MEM without fetal bovine serum and seeded at a density of 2 × 10⁵/mL in the upper compartment of the chamber. After incubation at 37°C for 12 h, the filters were collected and the cells adhered to the lower surface were fixed, stained with 4% crystal violet, and counted.

Tumorigenicity assay. Male BALB/c nude mice (ages, 4-5 wk) were purchased from Shanghai Laboratory Animals Center of Chinese Academy of Sciences. The stable cell lines were suspended in PBS at a concentration of 5 × 10⁶/mL. Injections (200 μL) were given s.c. in the right lower back, with four mice per group. Growth of tumors was monitored by measuring the tumor size with calipers every 7 d. Tumor volumes were calculated using the formula: tumor size = a × b² / 2. a is the longest, and b is the shortest diameter.

Statistical analysis. Statview (SAS Institute, Inc.) software was used for data analysis. Fisher's exact test was used to analyze the relationship between clinical characteristics and aberrant Hugl-1 in 60 patients. t test was applied to compare the results obtained from functional assays. For all tests, a P value of <0.05 was considered significant.

Hugl-1 transcript profiles in 80 specimens and 5 HCC-derived cell lines (Hep G₂, Huh7, SK-HEP-1, SMMC-7721, and BEL-7404) were analyzed by reverse transcription-PCR using a panel of primer pairs encompassing the entire cDNA sequence (Supplementary Fig. S1). In all cases, a major reverse transcription-PCR product whose size was identical to that predicted from the published Hugl-1 cDNA sequence could be detected (Genbank accession no. NM 004140). However, additional smaller bands were observed in 26 of the 80 HCC specimens and in one of the five HCC cell lines (Fig. 1A). In contrast to the HCC specimens, only the normal-sized Hugl-1 product was identified in the matched noncancerous liver samples. The subsequent direct sequencing for the PCR fragments revealed that aberrant splicing mutations and point mutations happened exclusively in HCC tumor tissues. A detailed analysis of these two types of alterations is described below.

Frequent aberrant splicing of Hugl-1 mRNA in HCC. A total of 23 Hugl-1 transcript variants were detected in 32.5% (26 of 80) of HCC specimens and 20.0% (one of five) of HCC cell lines (Table 1; Supplementary Fig. S2). With the exception of 4 Hugl-1 variants, the breakpoint sites of the Hugl-1 variants were not at normal splicing sites (Supplementary Fig. S2). Eighty-seven percent (20 of 23) of the 5′ breakpoints of the Hugl-1 variants localized within a 1262-bp region of 5′Hugl-1 mRNA, which encodes the conserved WD-40 repeat motifs of Hugl-1 (Fig. 1B). The aberrant Hugl-1 transcripts included deletion and/or insertion of one or more regions of the Hugl-1 gene (Fig. 2A; Supplementary Fig. S2). Moreover, these alterations could produce 20 different truncated Hugl-1 proteins lacking one or multiple WD-40 repeat motifs, as a result of a deletion or frameshift mutation (Supplementary Fig. S3). This suggests that the observed deletions or insertions may lead to a disruption of Hugl-1 protein structure and function. Interestingly, 78.3% (18 of 23) of the aberrant Hugl-1 transcripts stand a common pattern: a short direct repeat sequence ranging from 3 to 10 bp, which existed in Hugl-1/wt cDNA, was found to flank the 5′ and 3′ breakpoints; one short repeat sequence was retained whereas the other repeat and the sequence between the two direct repeats were deleted (Supplementary Fig. S4). Such splicing happened not only within exons but also between exon and intron (Supplementary Fig. S4B). These data suggested that a common factor of the mRNA splicing machinery might be defective, leading to the formation of aberrant Hugl-1 transcripts.

Point mutations and polymorphisms in the Hugl-1 gene. Normal-sized Hugl-1 PCR products obtained from the first round of PCR was directly sequenced to identify point mutations. Six point mutations were detected in the Hugl-1 cDNA in 7.5% (6 of 80) of the patients (Supplementary Table S3; Supplementary Fig. S2D), and these mutations were confirmed in at least three independent experiments. In addition, two polymorphisms in Hugl-1 (at nucleotide 503 and at nucleotide 1921) were identified, which are currently listed in the National Center for Biotechnology Information SNP database (Supplementary Table S3).

No deletions in Hugl-1 genomic sequence. To check whether the truncation in Hugl-1 mRNAs resulted from genomic deletions of the Hugl-1 gene, PCR was used to amplify the genomic DNA containing the regions where the aberrant splicing happened from SMMC-7721 cell and 20 HCC specimens (genomic DNA was only available for 20 of 26 cases with aberrant Hugl-1 transcripts). No deletion of the Hugl-1 gene was detected at the genomic level in any of the samples tested (data not shown). It is known that accurate splicing requires recognition of cis-acting consensus sequences, including intron/exon boundaries, branch sites, and auxiliary elements, such as intronic/exonic splicing enhancers/silencers. Mutations in these elements increase the possibility of aberrant splicing at cryptic sites (23); therefore, the genomic sequences flanking the intron/exon boundary area were amplified by PCR and subsequently sequenced. We did not detect any mutations in the classic and auxiliary splicing signals in the 20 HCC samples with aberrant Hugl-1 transcripts or in SMMC-7721 cell line.

To analyze the expression of Hugl-1 protein in HCC, we used 22 pair samples for Western blot analysis. Among them, four HCC tissue samples (DF99, DF104, DF105, and DG80) were applicable for detecting the truncated proteins of Hugl-1, because the variants found in these cases contain COOH terminus of Hugl-1 (see Table 1; Supplementary Fig. S3), which could be recognized by the antibodies against synthetic 24 COOH terminal peptides of Hugl-1 in Western blot. We could detect two truncated forms, Hugl-1/Δ1414-1567/1667-1959 in DG80 and DF105 and Hugl-1/Δ496-1680 in DF99, but not in their matched noncancerous liver tissues (Fig. 2B). The presence of the truncated form Hugl-1/Δ1414-1567/1667-1959 found in DG80 and DF105 was confirmed in immunoprecipitation assay (Supplementary Fig. S5). The reason for the failure to detect the truncated protein in DF104 might be the low expression level of the protein and/or limited sensitivity of the antibody. These data showed that at least some of the aberrantly spliced Hugl-1 could be translated into their corresponding proteins.

Western blot analysis also revealed that there are patients who had reduced Hugl-1 expression level (36.4%, 8 of 22) in HCC tissues whereas others had elevated (31.8%, 7 of 22) or unaltered expressed level (Supplementary Fig. S6). With these data, we were not able to make a clear conclusion about Hugl-1/wt protein expression level in HCC, thus far. Regarding correlation between expressions of Hugl-1/wt, aberrant variants, and clinical parameters of HCC patients, more HCC samples will be needed to clarify these questions.

We used Fisher's exact test to analyze the correlation between the presence of aberrant Hugl-1 and a series of clinical parameters, such as gender, age, tumor-node-metastasis stage, and vasal invasion. As shown in Table 2, the presence of abnormal Hugl-1 was significantly correlated with large tumor size (P = 0.031 to <0.05) and poor differentiation of HCC determined according to the Edmondson-Steiner classification (ref. 24; P = 0.002 to <0.005). However, the presence of abnormal Hugl-1 was not associated with gender, age, tumor-node-metastasis stage, α-fetoprotein level, HBV infection, or vasal invasion. Nonetheless, we noted that the percentage of patients with Hugl-1 mutants who developed vasal invasion (29.2%, 7 of 24) was higher than that of patients carrying Hugl-1/wt (13.9%, 5 of 36). These results indicate that aberrant Hugl-1 splicing may be associated with HCC progression.

We next investigated whether aberrant Hugl-1 variants could accelerate hepatogenesis or promote malignancy. We established SK-HEP-1 cells stably expressing Hugl-1/wt, Hugl-1/Δ1414-1567/1667-1959 (deletion of the fifth WD-40 repeat motif), or Hugl-1/Δ1190-1508 (deletion of the fifth WD-40 repeat motif and the COOH terminus; Fig. 3A; Supplementary Fig. S3).

3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assays revealed that there was no significant difference in cell growth between control cells and cells expressing Hugl-1/wt or any of the aberrant Hugl-1 variants (Supplementary Fig. S7). However, expression of Hugl-1/wt caused reduction in the migration of SK-HEP-1 cells during wound healing (Fig. 3B), which is consistent with previous reports (20, 21). In contrast, expression of Hugl-1/Δ1414-1567/1667-1959 or Hugl-1/Δ1190-1508 increased migration of SK-HEP-1 cells (Fig. 3B).

Furthermore, we carried out migration and invasion assays using the Boyden chamber system to quantitatively evaluate these effects. Hugl-1/wt notably inhibited cell migration and invasion, whereas Hugl-1/Δ1414-1567/1673-1965 and Hugl-1/Δ1190-1508 significantly enhanced the migration and invasion of tumor cells (Fig. 3C and D).

We further studied whether expression of Hugl-1 or its aberrant variants affected growth of HCC cells in vivo. As shown in Fig. 4, tumor growth of the transfectants expressing Hugl-1/wt was significantly inhibited compared with that of control cells, whereas the Hugl-1/Δ1414-1567/1667-1959 transfectants and the Hugl-1/Δ1190-1508 transfectants grew much faster than control cells. These results suggest that Hug-1/wt could be an important tumor suppressor for tumor growth and that aberrant Hugl-1 variants in HCC might antagonize the biological effects of endogenous Hugl-1 to promote tumor progression.

Aberrant splicing of many tumor suppressors and oncogenes, such as BRCA2 (25), Wilms' tumor suppressor gene (26), and Mdm2 (27), has been associated with tumor malignancy. Seeking aberrant formats is an important aspect in studying gene function and identifying molecular components involved in tumorigenesis (23, 28). In the present study, for the first time, we provided clear evidence that aberrant splicing and point mutations of Hugl-1 occurred exclusively in HCC specimens. Moreover, clinical statistical analysis revealed a significant correlation of the aberrantly spliced Hugl-1 with HCC progressive stage. Importantly, two truncated Hugl-1 proteins were found exclusively in HCC specimens, and expression of aberrant Hugl-1 variants significantly promoted HCC cell migration, invasion, and tumorigenicity in nude mice. These data suggest the important role of aberrant splicing of Hugl-1 in the tumorigenesis and the progression of HCC.

The detailed mechanism of the aberrant splicing in the tumor is still unknown. The most common reason for splicing defects is point mutation in the genomic splice site (23). In our study, 5′ and 3′ splice sites in Hugl-1 genes obeyed the GT/AG rule, and the branch sites matched well with the consensus sequence. This indicated that the aberrant splicing of Hugl-1 in HCC was not a result of mutations in the cis-acting consensus sequence. However, most aberrantly spliced Hugl-1 variants had a common splicing pattern: a short (3-10 bp) direct repeat was detected at the 5′ and 3′ breakpoints (Supplementary Fig. S4). The position of the breakpoints in relation to the direct repeats suggests that these direct repeat sequences might mediate deletions. Deletions flanked by short direct repeats (3-10 bp) have been observed in studies of the human tumor suppressor retinoblastoma gene (29) and the oncogene Mdm2 (30). The “replication slippage” model has been proposed to explain the production of deletions in the retinoblastoma gene (31). Yet in our case, no deletion of genomic sequence between repeat sequences was detected by PCR (data not shown). Similar sequence excision between short direct repeats at the mRNA level caused by “transcription slippage” has been described previously in Mdm2 (27). Whether “transcription slippage” played a role in HCC remains to be determined. We also noticed that five of the aberrantly spliced Hugl-1 had no direct repeats at the breakpoints, suggesting that another mechanism, such as abnormal stoichiometry of splicesome components (32) or dysfunction of the nonsense-mediated decay pathway (33, 34), could be involved in the generation of aberrantly spliced Hugl-1.

Hugl-1 contains conserved functional domains, including five WD-40 repeat motifs and a cluster of serine phosphorylation sites. A common function of WD-40 repeats is to coordinate the assembly of multiprotein complexes (35). Consistent with this, the WD-40 repeat motifs of Lgl has been reported to be involved in binding to target proteins in Drosophila or yeast (36–38). Also, mammalian Lgl containing a mutation predicted to disrupt the most conserved COOH terminal WD-40 repeats was unable to rescue the salt and temperature sensitivity of SRO7/77 yeast, indicating that at least some WD-40 repeats played an important role in Lgl function (39).

Our data showed that 5 breakpoints in the most aberrant Hugl-1 mutants localized in the region of mRNA encoding the WD-40 repeat motifs (Fig. 1B), which might lead to the disruption of the protein structure of WD-40 repeats (Supplementary Fig. S3). In addition, point mutation of Asp207 in Hugl-1 to Val (which was found in cases DA83, DA104, and DA179) could also disrupt the third WD-40 repeat folding (prediction by SMAR; http://smart.embl-heidelberg.de/). Disruption of the conserved WD-40 domain might disturb the normal function of Hugl-1. Accordingly, overexpression of either of the two mutants (Hugl-1/Δ1414-1567/1667-1959 with a deletion of the fifth WD-40 repeat or Hugl-1/Δ1190-1508 with a deletion of the fifth WD-40 repeat and the COOH terminus) was able to promote cell migration and invasion and enhance the HCC-derived tumor growth in vivo. In contrast, overexpression of Hugl-1/wt resulted in the opposite effect. Moreover, these aberrant Hugl-1 variants could antagonize the biological effects of Hugl-1/wt in a dose-dependent manner, and they might disturb the activity of Hugl-1/wt through inhibiting the formation of normal Hugl-1 homo-oligomers.⁶

Hugl-1/wt could significantly inhibit the tumor growth in vivo, which gave an important evidence to support the tumor suppressor role of Hugl-1. However, the inhibitory effect of Hugl-1 on cell growth was not observed in vitro, which is consistent with previous reports (20, 21). Transformation of cells involved the regulation of both cell number and cell structure (6, 7). Hugl-1, as an important regulator of cell polarity, may not directly regulate cell proliferation but certainly affects cell architecture, a critical component of the oncogenic process (11, 18). A recent study showed that reduced expression of Scrib lead to deregulation of polarity pathways, which promoted dysplastic and neoplastic growth in mammals by disrupting morphogenesis cycle and inhibiting cell death. However, similar to Hugl-1 in our study, it had no significant effect on monolayer cell proliferation (40).

The presence of aberrant spliced Hugl-1 was significantly correlated with large tumor size and poor differentiation of HCC. Besides, aberrant Hugl-1 variants seemed to be associated with vasal invasion (Table 2). Although this correlation was not statistically significant with the current sample size, poor differentiation has been previously reported to increase the risk for HCC metastasis (41, 42). These data indicated that Hugl-1 might be useful as a biomarker for diagnosis. Nowadays, modification (mainly down-regulation) of the aberrantly spliced variants expression using antisense oligonucleotides is more and more widely accepted as potential chemotherapeutics for human cancer (43). Because the aberrant Hugl-1 variants (at least some) could be detected at the protein level and promote HCC cell migration, invasion, and tumorigenicity (Fig. 2B, 3, and 4), we have a reason to believe that these aberrant Hugl-1 variants could be potential targets for HCC therapy. A large-scale sample analysis will be needed to further verify their potential application in HCC diagnosis and therapy.

Collectively, the high frequency of aberrant splicing of Hugl-1 we observed and the abnormal functions of the Hugl-1 variants indicated that aberrant splicing might be an important mechanism through which Hugl-1 lost its tumor suppressor activity in human cancers. The common features of the aberrant Hugl-1 variants suggest that they may arise from a defect in the mRNA splicing process during the development and/or progression of HCC. Plausibly, this abnormal mRNA splicing process may apply not only to Hugl-1 but also to other oncogenes and tumor suppressor genes that could play additional roles in the development and/or progression of HCC. Nevertheless, further studies are necessary to explore the mechanism and pathologic consequences of aberrant Hugl-1 splicing in HCC. Moving forward, it will be important to understand how the dual functions of Hugl-1 (as a tumor suppressor and as a regulator of cell polarity) are coupled together and to confirm the value of using Hugl-1 mutants for diagnostic and therapeutic applications.

No potential conflicts of interest were disclosed.

We thank Dr. Dangsheng Li and Dr. Yingjie Wu for the critical reading of this manuscript.