Enhanced hepatocarcinogenesis in mouse models and human hepatocellular carcinoma by coordinate KLF6 depletion and increased messenger RNA splicing


  • Diana Vetter,

    1. Department of Medicine/Division of Liver Diseases, Mount Sinai School of Medicine, New York, NY
    2. Department of Abdominal Surgery, University Hospital of Zurich, Zurich, Switzerland
    Search for more papers by this author
  • Michal Cohen-Naftaly,

    1. Department of Medicine/Division of Liver Diseases, Mount Sinai School of Medicine, New York, NY
    Search for more papers by this author
  • Augusto Villanueva,

    1. HCC Translational Research Laboratory, Barcelona-Clinic Liver Cancer Group, Liver Unit. Institut d'Investigacions Biomediques August Pi i Sunyer (IDIBAPS), Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBEREHD), Hospital Clinic, Barcelona, Spain
    Search for more papers by this author
  • Youngmin A. Lee,

    1. Department of Medicine/Division of Liver Diseases, Mount Sinai School of Medicine, New York, NY
    Search for more papers by this author
  • Peri Kocabayoglu,

    1. Department of Medicine/Division of Liver Diseases, Mount Sinai School of Medicine, New York, NY
    Search for more papers by this author
  • Rebekka Hannivoort,

    1. Department of Medicine/Division of Liver Diseases, Mount Sinai School of Medicine, New York, NY
    2. Department of Gastroenterology and Hepatology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands
    Search for more papers by this author
  • Goutham Narla,

    1. Department of Medicine/Division of Liver Diseases, Mount Sinai School of Medicine, New York, NY
    2. Departments of Genetics & Genomic Sciences, Mount Sinai School of Medicine, New York, NY
    Search for more papers by this author
  • Josep M. Llovet,

    1. Department of Medicine/Division of Liver Diseases, Mount Sinai School of Medicine, New York, NY
    2. HCC Translational Research Laboratory, Barcelona-Clinic Liver Cancer Group, Liver Unit. Institut d'Investigacions Biomediques August Pi i Sunyer (IDIBAPS), Centro de Investigación Biomédica en Red de Enfermedades Hepáticas y Digestivas (CIBEREHD), Hospital Clinic, Barcelona, Spain
    3. Institució Catalana de Recerca i Estudis Avançats, Hospital Clinic, Barcelona, Spain
    Search for more papers by this author
  • Swan N. Thung,

    1. Department of Pathology, Mount Sinai School of Medicine, New York, NY
    Search for more papers by this author
  • Scott L. Friedman

    Corresponding author
    1. Department of Medicine/Division of Liver Diseases, Mount Sinai School of Medicine, New York, NY
    • Division of Liver Diseases, Box 1123, Mount Sinai School of Medicine, 1425 Madison Ave., Room 11-70C, New York, NY 10029-6574
    Search for more papers by this author
    • fax: 212-849-2574

  • Potential conflict of interest: Nothing to report.


KLF6-SV1 (SV1), the major splice variant of KLF6, antagonizes the KLF6 tumor suppressor by an unknown mechanism. Decreased KLF6 expression in human hepatocellular carcinoma (HCC) correlates with increased mortality, but the contribution of increased SV1 is unknown. We sought to define the impact of SV1 on human outcomes and experimental murine hepatocarcinogenesis and to elucidate its mechanism of action. In hepatitis C virus (HCV)-related HCC, an increased ratio of SV1/KLF6 within the tumor was associated with features of more advanced disease. Six months after a single injection of diethylnitrosamine (DEN), SV1 hepatocyte transgenic mice developed more histologically advanced tumors, whereas Klf6-depleted mice developed bigger tumors compared to the Klf6fl(+/+) control mice. Nine months after DEN, SV1 transgenic mice with Klf6 depletion had the greatest tumor burden. Primary mouse hepatocytes from both the SV1 transgenic animals and those with hepatocyte-specific Klf6 depletion displayed increased DNA synthesis, with an additive effect in hepatocytes harboring both SV1 overexpression and Klf6 depletion. Parallel results were obtained by viral SV1 transduction and depletion of Klf6 through adenovirus-Cre infection of primary Klf6fl(+/+) hepatocytes. Increased DNA synthesis was due to both enhanced cell proliferation and increased ploidy. Coimmunoprecipitation studies in 293T cells uncovered a direct interaction of transfected SV1 with KLF6. Accelerated KLF6 degradation in the presence of SV1 was abrogated by the proteasome inhibitor MG132. Conclusion: An increased SV1/KLF6 ratio correlates with more aggressive HCC. In mice, an increased SV1/KLF6 ratio, generated either by increasing SV1, decreasing KLF6, or both, accelerates hepatic carcinogenesis. Moreover, SV1 binds directly to KLF6 and accelerates its degradation. These findings represent a novel mechanism underlying the antagonism of tumor suppressor gene function by a splice variant of the same gene. (HEPATOLOGY 2012)

Hepatocellular carcinoma (HCC) is comprised of several molecular subclasses.1 We previously reported that allelic loss of the KLF6 tumor suppressor enhances chemical carcinogenesis in mice, and the molecular signatures of the resulting tumors closely mimics aggressive human HCC.2 KLF6 belongs to the family of Krüppel-like factors, which are characterized by the presence of Cys2/His2 zinc finger motifs in their carboxy-terminal domains, which confer binding to GC/GT-rich sequences in gene promoter and enhancer regions.3 KLF6 regulates cellular pathways that inhibit tumor cell proliferation, migration, angiogenesis, and invasion,2, 4-7 while enhancing apoptosis8 and differentiation.9 Reduction of KLF6 messenger RNA (mRNA) expression in HCCs due to chronic hepatitis B virus (HBV)10 and hepatitis C virus (HCV)2, 10 is frequent, and correlates with advancing stage; moreover, extremely low KLF6 mRNA levels are linked to reduced survival.2

KLF6 activity in human cancer can be attenuated by loss of heterozygosity,5, 11-14 somatic mutation,11, 12 and promoter methylation.15 Additionally, alternative splicing of KLF6 into an antagonistic splice form, SV1, is increased in HCC10, 16 and other cancers.9, 17-19 Specifically, ratios of SV1/KLF6 in tumors from HBV10 and HCV2, 10, 16 -related HCCs are increased compared to surrounding tissues.

SV1, the major KLF6 splice variant, lacks the DNA binding domain, is pro-proliferative, and facilitates tumor invasion by antagonizing the transactivation of p21 and E-cadherin by KLF6.5, 6 SV1 also displays proapoptotic caspase activity and accelerates degradation of the antiapoptotic protein NOXA.20, 21 Moreover, silencing of SV1 in ovarian cancer models decreases invasiveness and angiogenesis, with reduced VEGF protein.9

Mechanisms driving splicing of KLF6 and accounting for its antagonism of full-length KLF6 are largely unknown. Activation of the Ras oncogene stimulates KLF6 splicing, which promotes proliferation.15, 22 The specific ratio of SV1/KLF6 appears to regulate proliferative and tumorigenic activity, but it is unclear whether the effect is due solely to increased SV1, decreased KLF6, or both. Accordingly, in this study we have first established the clinical relevance of an increasing ratio of SV1/KLF6 as a predictor for HCV-associated HCC behavior, and then modeled the key features of KLF6 dysregulation in human HCC using mouse models, including loss of KLF6 expression through hepatocyte-specific deletion, increased SV1 through hepatocyte-specific transgene expression, and a combination of the two defects. These findings confirm KLF6 dysregulation in human HCC and provide novel insights into this tumor suppressor gene's regulation and impact on hepatocarcinogenesis.


ALT, alanine transaminase; AST, aspartate transaminase; DEN, diethylnitrosamine; HCC, hepatocellular carcinoma; HCV, hepatitis C virus; LOH, loss of heterozygosity.

Materials and Methods

Human Data.

We analyzed SV1- and KLF6 mRNA levels in 149 HCV-infected human liver samples covering the entire hepatocarcinogenic spectrum: normal liver (n = 9), cirrhosis (n = 9), dysplastic nodules (n = 27), very early HCC (n = 16), early (n = 17), advanced HCC (n = 51), and very advanced HCC (n = 20) as described.2 The SV1-/KLF6 mRNA expression ratio was further correlated with clinical and pathological variables in a subset of 55 HCC patients. Real-time polymerase chain reaction (PCR) was performed as described.2


Klf6fl(+/+) mice provided by Genentech were bred with Albumin-Cre mice.23 The TTR-flag-humanSV1-PolyA construct was cloned with a three-fragment recombination into the pcDNA6.2/V5-pL destination vector using the MultiSite Gateway Pro system from Invitrogen. The construct was injected into Klf6fl(+/+) fertilized eggs. The resulting SV1 Klf6fl(+/+) mice were bred with AlbCre Klf6fl(+/+) mice.

Male Klf6fl(+/+)-, AlbCre Klf6fl(+/+)-, SV1 Klf6fl(+/+)-, and SV1 AlbCre Klf6fl(+/+) mice were injected with 5 mg/kg body weight diethylnitrosamine (Sigma, #N0258) intraperitoneally at 2 weeks of age. Tumors were measured macroscopically and analyzed microscopically as described.2

Primary Hepatocytes.

Primary hepatocytes were isolated by in situ perfusion with Liberase (Roche 05-401-119-001).2 Twelve hours later, either AdenoCre- or LacZ-expressing control virus was added at a concentration of 10 multiplicity of infection. Twenty-four hours later, media was replaced with a lentivirus expressing pBabe- or pBabe-KLF6. After 12 hours, fresh virus-containing media was added and the cells were collected 24 hours later.

Incorporation of 3H-thymidine was used to measure DNA synthesis.5 Hepatocytes were trypsinized and counted 5, 24, and 48 hours after isolation, and the number of nuclei per hepatocyte were counted in triplicate by ImageJ64 in ten 10× fields of isolated primary hepatocytes from all four mice lines.

For cell cycle analysis, ≈106 hepatocytes were suspended in 0.5 mL phosphate-buffered saline (PBS), fixed with 4.5 mL of ice-cold 70% ethanol, stained with PI solution (propidium iodide, RNAseA, PBS), strained through polystyrene cell strain tubes, incubated in the dark for 20 minutes at room temperature, and fluorescence-activated cell sorting (FACS) analysis performed with Calibur cell sorter.


Proliferating cell nuclear antigen (PCNA) immunostaining was performed using sodium citrate 10 mM, pH 6.0, Dako Kit Envision System HRP labeled Polymer, antimouse (Dako K4000) and the sc-56 α-PCNA antibody.

Cell Culture.

293T and HUH7 cells were cultured in DMEM+GlutaMAX GIBCO 31985 with 10% fetal bovine serum (FBS) and transfected with Lipofectamine 2000 (Invitrogen 11668-019) according to the manufacturer's instructions. Cells were transfected with pCI-neo-GFP, pCI-neo-FLAG-KLF6, pCIneo-FLAG-SV1, a p21 luciferase promoter,5 and Renilla luciferase vector (Promega, Madison, WI) as internal control.

Protein was collected in RIPA Buffer with added protease (Roche Complete Mini 04693124001) and phosphatase inhibitors (Thermo Scientific #78428), and the following antibodies were used: α-KLF6 (sc7158), α-FLAG (Sigma, F7425), α-calnexin (ab75801), α-p21 (sc397), and α-Cyclin B1 (sc752). For coimmunoprecipitation studies, protein was collected in CoIP Buffer (50 mM Tris, 150 mM NaCl with PI 1:10, PPI 1:100, PMSF 100 mM 1:100) 24 hours after transfection. Then 500 mg of protein was incubated with 40 mL of washed FLAG agarose beads (Sigma Red ANTI-FLAG M2 Affinity Gel F246) overnight. Beads were washed with PBS, 20 mL of 2× sample buffer was added per sample, and an immunoblot with α-KLF6 (sc7158) was performed.

RNA was collected, reverse transcribed, and amplified as described.23 The following primers were used: mCyclophillin F/R (CAGACGCCACTGTCGCTTT/TGTCTTTGGAACTTTGTCTGCAA), hKLF6 F/R (CGGACGCACACAGGAGAAAA/CGGTGTGCTTT CGGAAGTG), SV1 F/R (CCTCGCCAGGGAAGG AGAA/CGGTGTGCTTTCGGAAGTG).

Cycloheximide (10 mg/mL) was added 24 hours after cotransfection and protein collected at 0, 15, 30, and 60 minutes. MG132 (Sigma C2211) was added (5 mM) 4 hours prior to adding cycloheximide.

Cotransfection of p21 luciferase and Renilla plasmids was performed as described.5 Cells were washed with PBS twice and 30 mL of Lysis Buffer (Promega passive Lysis Buffer 5x, #E194A) was added per well.

Statistical Analysis.

Data are expressed as mean ± standard error of the mean (SEM) or standard deviation (SD). Student's t test, analysis of variance (ANOVA), Mann-Whitney, and chi-square (3-way contingency table) were calculated to compare experimental groups. Differences were considered statistically significant if P < 0.05. Analyses were conducted using R statistical package and SPSS software (v. 14) for the human data,2 and GraphPad prism for the remaining data.


Correlation of SV1/KLF6 Ratio in HCV-Associated HCC with Aggressive Clinical Features.

The SV1/KLF6 mRNA splicing ratio is increased in 18% of HBV-associated10 and 0%-76% of HCV-positive10, 16 HCCs. Here we analyzed the SV1/KLF6 mRNA splicing ratio in liver tissues from 149 HCV-positive patients with progressive stages of HCV-associated liver disease.2 The splicing ratio was significantly increased in HCC samples compared to nontumoral tissues, including normal liver (P = 0.03), cirrhotic liver (P = 0.01), or dysplastic nodules (P < 0.001). In addition, the ratio linearly increased with progressive stages of HCC (P < 0.001) (Fig. 1A). This finding raised the possibility that increased KLF6 splicing might contribute to tumor behavior or clinical outcomes.

Figure 1.

SV1/KLF6 ratio is increased in HCV-associated HCC and is associated with aggressive clinical behavior. The SV1/KLF6 mRNA ratio in human HCV-associated liver disease samples is significantly increased in HCC compared to nontumoral tissues (P < 0.001; N: normal liver (n = 9), Ci: cirrhosis (n = 9), D: dysplastic nodules (n = 27), VE: very early HCC (n = 16), E: early HCC (n = 17), A: advanced HCC (51), AA: very advanced HCC (n = 20) (Fig. 1A). The SV1/KLF6 ratio was significantly higher in larger tumors (P = 0.04) and in HCC with vascular invasion (P = 0.01, Fig. 1B).

We examined whether the SV1/KLF6 mRNA ratio was correlated with features of more advanced disease. Accordingly, we correlated SV1/KLF6 mRNA ratio with clinical and pathological variables in a subset of 55 HCCs from whom these data were available. Increased SV1/KLF6 ratio was significantly associated with larger tumors (P = 0.04) and vascular invasion (P = 0.01) (Fig. 1B). The KLF6 splicing ratio was not correlated with survival, however (data not shown). These findings are consistent with earlier reports in prostate, lung,21 ovarian,20 and pancreatic19 cancers, where SV1 has been correlated with more aggressive disease.

Hepatocyte-Specific SV1 Transgenic Mice Have More Advanced Tumors 6 Months After Diethylnitrosamine (DEN) Treatment.

To clarify the role of an increased KLF6 splicing ratio in hepatocarcinogenesis, we generated mice with different SV1/KLF6 ratios by first crossing double floxed Klf6 mice (Klf6fl(+/+)) with albumin-Cre transgenic animals (AlbCre).23, 24 Then we generated hepatocyte-specific human SV1-expressing mice in a Klf6fl(+/+) background (SV1 Klf6fl(+/+)). Hepatocyte-specificity was conferred by using the transthyretin (TTR)25 promoter. These mice were then bred with the AlbCre Klf6fl(+/+) mice. Altogether, this breeding strategy yielded four lines of mice: (1) Klf6fl(+/+) mice (used as controls) with endogenous Klf6; (2) AlbCre Klf6fl(+/+) mice with hepatocyte-specific Klf6 depletion and no SV1; (3) SV1 Klf6fl(+/+) mice with hepatocyte-specific SV1 overexpression and endogenous Klf6, and; (4) SV1 AlbCre Klf6fl(+/+) mice with hepatocyte-specific SV1 overexpression on a background of Klf6 depletion. All mice appeared phenotypically normal (Supporting Fig. 1A,B), with normal liver architecture and no spontaneous tumorigenesis. Hepatocyte-specific overexpression of SV1 and endogenous KLF6 levels were validated in the SV1 Klf6fl(+/+) transgenics by immunoblot and immunostaining (Supporting Fig. 1C,D). Of note, due to an inverted distal LoxP site AlbCre Klf6fl(+/+) mice have a partial Klf6 depletion and are effectively hypomorphic rather than complete knockouts (F. DeSauvage, pers. commun.; see Supporting Fig. 1C,D).

To assess the propensity of each of these four mouse lines toward hepatocarcinogenesis, animals were injected with a single intraperitoneal dose (5 mg/kg) of DEN at 2 weeks of age,2 and tumor development was assessed at 3, 6, and 9 months and compared to nontransgenic littermates.

At 3 months there were no macro- or microscopically visible tumors. At 6 months 57% (4/7) of the Klf6fl(+/+) controls had microscopic tumors compared to 83% (5/6) in the SV1 Klf6fl(+/+) transgenics and 100% of both the AlbCre Klf6fl(+/+)) (13/13) and the SV1 AlbCre Klf6fl(+/+) (7/7) animals (not significant) (Supplemental Table 1). Tumors were significantly larger in the AlbCre Klf6fl(+/+) mice (P < 0.05) (Fig. 2A,B) and contained a significantly higher tumor grade in the SV1 Klf6fl(+/+) transgenics (P < 0.05) and even more so in SV1 AlbCre Klf6fl(+/+) mice (P < 0.005) (Fig. 2C,D) at 6 months after DEN injection. These findings indicate that both Klf6 depletion and SV1 overexpression independently promote tumorigenesis after DEN treatment by increasing the size and advancing the histologic grade of the tumors.

Figure 2.

Hepatocyte-specific Klf6-depleted mice have larger tumors and SV1 transgenic mice have more progressed tumors than controls after 6 months of DEN treatment. AlbCre Klf6fl(+/+) mice have significantly larger tumors than the Klf6fl(+/+) controls 6 months after DEN treatment (P < 0.05) (A,B). SV1 Klf6fl(+/+) mice have tumors with higher grading. This is even more pronounced in SV1 AlbCre Klf6fl(+/+) animals (P < 0.0001) (C). (D) Representative hematoxylin and eosin images of Klf6fl(+/+) livers 6 months after DEN treatment with vaguely nodular parenchyma with large cell changes (grade 0-1) (i) and tumor cells with hyperchromatic nuclei and increased nuclear/cytoplasmic ratio with microvascular invasion in SV1 AlbCre Klf6fl(+/+) livers (grade 2) (ii).

Hepatocyte-Specific SV1 Transgenic mice, in Combination with Klf6 Depletion, Have a Greater Tumor Burden and Heavier Livers 9 Months After DEN.

The tumorigenic activities of Klf6 depletion and/or SV1 overexpression at 6 months were more clearly evident at 9 months, with both more (Fig. 3A, P < 0.001; 3B, P < 0.05) and larger tumors (Fig. 3C, P < 0.05) in SV1 AlbCre Klf6fl(+/+) mice, resulting in a significantly greater tumor burden (Fig. 3D) and heavier livers (P < 0.01) (Supporting Fig. 2) than Klf6fl(+/+) controls. Interestingly, this trend was also observed in both Klf6-depleted mice (AlbCre Klf6fl(+/+)) and in SV1-transgenic mice (SV1 Klf6fl(+/+)), but only the combination of both defects led to an additive and significant effect, reinforcing the contribution of the SV1/Klf6 ratio in hepatocarcinogenesis.

Figure 3.

Hepatocyte-specific SV1-transgenic mice with Klf6 depletion have a greater tumor burden than Klf6 wildtype control mice 9 months after DEN treatment. SV1 AlbCre Klf6fl(+/+) mice have significantly more (A) (P < 0.001), (B) (P < 0.05), and larger tumors (C) (P < 0.05), resulting in a significantly higher tumor load (D) than Klf6fl(+/+) controls 9 months after DEN treatment.

Both SV1 Overexpression and Klf6 Depletion Increase Proliferation in Primary Hepatocytes.

To determine whether increased tumorigenesis arose from altered cellular growth we analyzed primary hepatocytes from the four transgenic mouse models. We also explored the impact of acute, rather than chronic Klf6 depletion, by treating primary hepatocytes from Klf6fl(+/+) mice with either LacZ- or AdenoCre virus to deplete Klf6 in culture, and additionally treating cells with either pBabe- or pBabeSV1 lentivirus to overexpress SV1 in order to recapitulate the phenotypes of the four different mouse lines. This approach yielded a consistent Klf6 mRNA depletion of ≈50% and ≈5-fold overexpression of SV1, as assessed by quantitative PCR (Supporting Fig. 3A-C).

Primary hepatocytes from both chronic (AlbCre Klf6fl(+/+)) and acute (Klf6fl(+/+) and adenoCre virus) Klf6-depleted hepatocytes displayed a significant increase in DNA synthesis, based on 3H-thymidine incorporation assay (Fig. 4A, P < 0.05; 4B, P < 0.05). Although SV1 overexpression in hepatocytes both from SV1 transgenic animals and pBabeSV1-infected Klf6fl(+/+) hepatocytes only displayed a trend toward increased 3H-thymidine incorporation, the combination of Klf6 depletion and SV1 overexpression led to an additional and significant increase in DNA synthesis in both models (Fig. 4A, P < 0.05; 4B, P < 0.0005). These findings correlated with a significantly increased hepatocyte count from either SV1 Klf6fl(+/+) transgenics 24 hours after isolation (P < 0.05), and of AlbCre Klf6fl(+/+)- (P < 0.05) or SV1 AlbCre Klf6fl(+/+) hepatocytes (P < 0.02) 48 hours after isolation (Fig. 4C). Thus, Klf6 depletion increases proliferation, which is further augmented when SV1 is simultaneously overexpressed.

Figure 4.

SV1 overexpression and Klf6 depletion increase proliferation in primary hepatocytes. In primary hepatocytes isolated from Klf6fl(+/+) mice with induced acute Klf6 depletion through adenoCre virus infection ex vivo, 3H-thymidine incorporation is increased significantly (P < 0.05), with an additional effect when cells are treated with both adenoCre- and pBabeSV1-lentivirus (P < 0.05) (A). Correspondingly, hepatocytes from AlbCre Klf6fl(+/+) (P < 0.05), SV1 Klf6fl(+/+) (P = 0.09), and SV1 AlbCre Klf6fl(+/+) (P = 0.0005) have more 3H-thymidine incorporation than Klf6fl(+/+) control cells (B). The cell count of SV1 Klf6fl(+/+) hepatocytes 24 hours after isolation (P < 0.05)- and of AlbCre Klf6fl(+/+)- (P < 0.05) and SV1 AlbCre Klf6fl(+/+) (P < 0.02) mice 48 hours after isolation was significantly higher than that of Klf6fl(+/+) control hepatocytes (C).

SV1 Overexpression and Klf6 Depletion Increase Ploidy in Primary Hepatocytes.

To examine if SV1 overexpression or Klf6 depletion had an impact on cell cycle distribution, we performed FACS analysis of primary hepatocytes. There was a significant increase of cells with 4N DNA, but without any measurable differences in S-phase distribution (Fig. 5A) in SV1 overexpressing and/or Klf6 depleted cells compared to Klf6 wildtype hepatocytes. This corresponded with very few PCNA-positive hepatocytes in untreated hepatocytes of all four murine models (Supporting Fig. 4A,B). The increased number of 4N cells was consistent with an increase in cell ploidy, which was confirmed by nuclear quantification per cell (Fig. 5B,C). The control mice had significantly more cells with a single nucleus at baseline compared with the SV1 transgenic or Klf6-depleted hepatocytes. Forty-eight hours after isolation there was a shift toward higher ploidy of the hepatocytes in all groups. This shift was significant for hepatocytes from SV1 AlbCre Klf6fl(+/+) mice (P < 0.05) (Fig. 4B). The increased ploidy is DEN-independent, as it was observed in hepatocytes isolated from untreated mice and in age-matched animals between 10-12 weeks of age.

Figure 5.

SV1 overexpression and Klf6 depletion increase ploidy in primary hepatocytes. FACS analysis of freshly isolated hepatocytes from AlbCre Klf6fl(+/+), SV1 Klf6fl(+/+) and SV1 AlbCre Klf6fl(+/+) mice shows a significant increase of hepatocytes with 4N as opposed to 2N when compared to Klf6fl(+/+) hepatocytes (A) (P < 0.05) without a corresponding increase in S-phase (A). Ploidy in hepatocytes from AlbCre Klf6fl(+/+), SV1 Klf6fl(+/+), and SV1 AlbCre Klf6fl(+/+) animals compared with Klf6fl(+/+) controls is significantly increased 5 hours, and even more pronounced 48 hours, after hepatocyte isolation. At 48 hours after isolation there were significantly more hepatocytes with 3 or 4 nuclei as compared with 5 hours after isolation in SV1 AlbCre Klf6fl(+/+) hepatocytes (P < 0.05). (C) Representative phase contrast microscopic images of hepatocytes from Klf6fl(+/+) hepatocytes 5 hours (ai) and 48 hours (aii) and from SV1 Klf6fl(+/+) again at 5 (bi) and 48 hours (bii) after hepatocyte isolation. The arrows point out hepatocytes with one nucleus. Asterisk indicates a hepatocyte with three nuclei.

Cyclin B1, which is a key regulator of G2/M transition, was consistently down-regulated in SV1-transgenic and Klf6-depleted hepatocytes, indicating reduced G2/M transition in the SV1-transgenic and Klf6-depleted hepatocytes (Supporting Fig. 5).

SV1 Binds to KLF6 and Inhibits Its Transcriptional Activity (p21).

We next examined how SV1 antagonizes KLF6 tumor suppressor function. To do so, we tested the hypothesis that SV1 might physically sequester KLF6 to prevent it from transactivating growth-inhibitory genes. 293T cells were transfected with either FLAG-SV1 or FLAG-KLF6. Coimmunoprecipitation using α-FLAG beads, followed by immunoblotting with a KLF6 polyclonal antibody that detects both KLF6 and SV1 protein, was performed. In the cells transfected with FLAG-SV1, endogenous KLF6 was recovered and, conversely, in the cells transfected with FLAG-KLF6 SV1 was recovered, documenting physical interaction between SV1 and KLF6 (Fig. 6). To date, Ras-induced SV1 has been shown to decrease KLF6 promoter occupancy.16, 22 We confirmed the antagonistic function of SV1 to KLF6 by assessing the impact of SV1 on the transcriptional activity of KLF6 in cell culture using the p21 promoter, a well-validated direct transcriptional target of KLF6.5 In both 293T (Fig. 7A), and HUH7 cells (Fig. 7B), transfection of FLAG-SV1 alone did not affect p21 activity, whereas cotransfection of FLAG-SV1 with FLAG-KLF6 significantly reduced the increase in p21-luciferase activity induced by FLAG-KLF6 alone (P < 0.05).

Figure 6.

SV1 physically binds to KLF6. In 293T cells transfected with either empty pCI-neo, FLAG-KLF6, or FLAG-SV1 cDNA. Immunoprecipitation was performed with α-FLAG, followed by an immunoblot with α-KLF6. In cells transfected with FLAG-KLF6, SV1 was pulled down and in cells transfected with FLAG-SV1, KLF6 was pulled down.

Figure 7.

SV1 inhibits KLF6 transcriptional activity. Transfection of FLAG-KLF6 significantly increases p21-luciferase activity in 293T- (A) (P < 0.05) and HUH7 cells (B) (P < 0.01). This increase is inhibited when FLAG-SV1 is cotransfected (A) (P < 0.05) (B) (P < 0.05). FLAG-SV1 alone does not affect p21 luciferase activity (A,B). Similarly, in isolated primary hepatocytes from Klf6-depleted mice (AlbCre Klf6fl(+/+), SV1 AlbCre Klf6fl(+/+)) versus Klf6fl(+/+) controls (C) as well as hepatocytes from Klf6fl(+/+) mice treated with adenoCre- versus lacZ control virus (D) p21 protein levels were decreased.

Similar to transfected cells, p21 protein levels were decreased in primary hepatocytes from our AlbCre Klf6fl(+/+)- and SV1 AlbCre Klf6fl(+/+) mice in comparison with the Klf6fl(+/+) hepatocytes with endogenous Klf6 (Fig. 7C). Moreover, in primary hepatocytes from Klf6fl(+/+) mice transduced with either AdenoCre virus or pBabe- or pBabeSV1 lentivirus, respectively, Klf6 depletion was also associated with a decrease in p21 protein levels (Fig. 7D). These findings confirm p21 as a transcriptional target of KLF6 and establish a bona fide antagonistic function of SV1.

SV1 Accelerates the Degradation of KLF6.

Because SV1 is localized primarily in the cytoplasm26 and can interact with KLF6, we hypothesized that SV1 antagonizes KLF6 by accelerating its degradation in the cytoplasm. To test this hypothesis, increasing amounts of FLAG-SV1 cDNA were cotransfected with a constant amount of FLAG-KLF6. This led to an SV1-dependent decrease in KLF6 protein in both 293T (Fig. 8A), and HUH7 cells (Fig. 8C), without affecting KLF6 RNA levels (not shown), which was quantified in four 293T immunoblots (Fig. 8B, P < 0.05).

Figure 8.

SV1 accelerates KLF6 degradation. Cotransfecting increasing amounts of FLAG-SV1 to a constant amount of FLAG-KLF6 leads to a decrease in KLF6 protein levels in 293T- (A) and HUH7 cells (C). (B) The immunoblot quantification in 293T cells (P < 0.05) (n = 4). 293T- (D) and HUH7 cells (F) were cotransfected with either FLAG-KLF6 + pCI-neo-GFP or FLAG-KLF6 + FLAG-SV1 and protein was collected 0, 15, 30, or 60 minutes after adding cycloheximide. Cotransfection of FLAG-SV1 accelerated the degradation of KLF6 in both 293T cells (D). This degradation can be inhibited by adding the proteasome inhibitor MG132 (D). The quantification of six immunoblots in 293T cells (E) shows lower KLF6 protein levels at timepoint 0 (P < 0.05) and accelerated KLF6 degradation (P < 0.005) in the presence of SV1 as seen at timepoint 60 minutes after cycloheximide (E). Corresponding can be observed in protein from primary hepatocytes of Klf6fl(+/+) versus SV1 Klf6fl(+/+) mice (F).

To assess if decreased KLF6 protein was due to accelerated degradation of KLF6, 293T cells were cotransfected with either pCI-neo-GFP+FLAG-KLF6 or FLAG-SV1+FLAG-KLF6, and protein was collected 24 hours after transfection and at 0, 15, 30, and 60 minutes after adding cycloheximide (Fig. 8D,E). In the presence of SV1, KLF6 protein levels at baseline were significantly lower (Fig. 8E, P < 0.05) and KLF6 degradation was accelerated 60 minutes after adding cycloheximide (Fig. 8E, P < 0.005). The accelerated degradation was inhibited by adding the proteasomal inhibitor MG132 (Fig. 8D,F), indicating that SV1 accelerates the proteasomal degradation of KLF6. These findings were reproducible in primary hepatocytes isolated from Klf6fl(+/+) control mice compared to SV1 Klf6fl(+/+) transgenics (Fig. 8F). The findings establish that SV1 antagonizes KLF6 at least in part by accelerating its degradation.


Our findings demonstrate that the SV1/KLF6 mRNA ratio is significantly increased in patients with HCV-associated HCC relative to surrounding tissue, and correlates with features of more aggressive disease. This observation is consistent with findings in gastric,18 lung,21 prostate,27 and pancreatic19 cancers, where increased splicing has been associated with worse patient outcomes. Consistent with these findings, KLF6 depletion in mouse models resulted in significantly increased tumor size, whereas hepatocyte-specific overexpression of SV1 led to more advanced tumor grade, consistent with findings of higher-grade tumors associated with an increased SV1/KLF6 ratio in both ovarian9 and pancreatic19 cancers. The increased tumorigenesis after DEN resulted in a higher tumor burden, with both larger and more numerous tumors in mice with hepatocyte-specific Klf6 depletion and SV1 overexpression 9 months after DEN treatment, supporting a role of increased KLF6 splicing in hepatocarcinogenesis.

The proliferative effect of an increased SV1/KLF6 ratio was directly validated in freshly isolated primary hepatocytes from four different mouse lines and in Klf6fl(+/+) hepatocytes in which Klf6 was depleted or SV1 was overexpressed in culture. Here, Klf6 depletion was associated with significantly increased DNA synthesis and cell number, and was further enhanced by additional SV1 overexpression. However, hepatocyte-specific SV1 overexpression alone was not sufficient to significantly affect cell count. Together, these findings suggest that the loss of KLF6 and increased SV1 confer separate, complementary effects on tumor propensity, with hepatocyte-specific Klf6 depletion driving enhanced proliferation, whereas hepatocyte-specific SV1 overexpression largely provokes an increased tumor grade. Of equal importance, the ratio of SV1/KLF6, and not just Klf6 depletion or SV1 overexpression alone, is additive in promoting tumorigenesis.

Additionally, whereas KLF6 mRNA levels in HCV-associated liver disease decreased progressively and significantly from noncirrhotic to cirrhotic liver tissue, with a further significant decrease in dysplastic liver tissue,2 the ratio of SV1/KLF6 mRNA did not change between normal, cirrhotic, and dysplastic liver tissue. It increased significantly from dysplastic to very early HCC, with a further significant increase in advanced and very advanced HCC. This finding suggests that a decrease of KLF6 precedes onset of splicing, and that splicing coincides with malignant transformation. This finding is consistent with evidence of increased KLF6 splicing only in malignant tissues from gastric18 and pancreatic19 cancers.

In aggregate, the findings indicate that reduction in KLF6 mRNA and increased SV1 expression occur independently and sequentially, ultimately accelerating the development of HCC when both are present. The association of the onset of KLF6 splicing with malignant transformation, as well as the inhibition of tumor progression by siSV1 in ovarian,9 gastric,18 and lung28 cancer models underscore the importance of KLF6 splicing in hepatocarcinogenesis, and reinforce its antagonism as a potential therapeutic target, possibly by antagonizing Ras16 or other drivers of KLF6 splicing.

The distinct phenotypes between SV1-overexpressing mice after DEN treatment and those with Klf6 depletion indicate that in addition to KLF6-dependent functions, SV1 likely has KLF6-independent functions, although little is known about these activities. Because SV1 has been localized to the nucleus as well as cytoplasm,22 one possibility is that SV1 functions as a transcriptional cofactor. Alternatively, SV1 might bind to other proteins and influence their degradation and/or cytoplasmic-nuclear partitioning.

There is a growing recognition that functional antagonists of tumor suppressors may contribute to cancer progression, including those of p53,29 or cell cycle checkpoint kinases like Chk2.30 Interestingly, for both p53 and Chk2, heterodimerization with their splice variants is essential for their antagonistic function29, 30 and can have an impact on cellular localization.29 Whereas SV1 binds to KLF6, and can increase the degradation of KLF6 by the proteasome, it is uncertain whether this interaction is required for SV1′s tumor promoting activity, or if KLF6-SV1 heterodimerization affects cellular localization. Finally, both SV1 overexpression as well as Klf6 depletion in hepatocytes each increases cell ploidy, implying a role of SV1- and KLF6 in G2/M cell cycle checkpoint regulation.

Our findings in HCC confirm KLF6 splicing as a mechanism to inactivate KLF6 full length and further reinforce findings in a growing list of tumors in which splicing is enhanced in cancer, and in which an increased SV1/KLF6 ratio has been associated with poorer outcome. In vivo cancer models employing small interfering RNA (siRNA) to knock down SV1, for example in ovarian,9 lung,28 and gastric18 cancers, emphasize the therapeutic potential of blocking SV1 and justify efforts to elucidate mechanisms of KLF6 splicing regulation in hopes of developing splicing antagonists.


We thank Sigal Tal-Kremer for technical assistance.