The degradation of cell cycle regulators by SKP2/CKS1 ubiquitin ligase is genetically controlled in rodent liver cancer and contributes to determine the susceptibility to the disease

Previous work showed a genetic control of cell cycle deregulation during hepatocarcinogenesis. We now evaluated in preneoplastic lesions, dysplastic nodules and hepatocellular carcinoma (HCC), chemically induced in genetically susceptible F344 and resistant Brown Norway (BN) rats, the role of cell cycle regulating proteins in the determination of a phenotype susceptible to HCC development. p21WAF1, p27KIP1, p57KIP2 and p130 mRNA levels increased in fast growing lesions of F344 rats. Lower/no increases occurred in slowly growing lesions of BN rats. A similar behavior of RassF1A mRNA was previously found in the 2 rat strains. However, p21WAF1, p27KIP1, p57KIP, p130 and RassF1A proteins exhibited no change/low increase in the lesions of F344 rats and consistent rise in dysplastic nodules and HCC of BN rats. Increase in Cks1‐Skp2 ligase and ubiquitination of cell cycle regulators occurred in F344 but not in BN rat lesions, indicating that posttranslational modifications of cell cycle regulators are under genetic control and contribute to determine a phenotype susceptible to HCC. Moreover, proliferation index of 60 human HCCs was inversely correlated with protein levels but not with mRNA levels of P21WAF1, P27KIP1, P57KIP2 and P130, indicating a control of human HCC proliferation by posttranslational modifications of cell cycle regulators.

Hepatocellular carcinoma (HCC) is the fifth commonest malignant tumor worldwide and the third cause of cancercaused death. 1,2 The study of genetic and molecular events during hepatocarcinogenesis could lead to the identification of prognostic markers of the disease, which could eventually be used as therapeutic targets.
Recently, different hepatocarcinogenesis susceptibility and resistance genes controlling the growth, progression and redifferentiation of preneoplastic and neoplastic lesions, have been mapped in mice and rats, indicating an inherited predisposition to HCC, 3 as also suggested by human HCC epidemiology. 4 Analysis of the effector mechanisms of susceptibility genes indicates that only few preneoplastic lesions of resistant rats acquire the capacity to grow and progress autonomously to high-grade dysplastic nodules and HCC. 5 Overexpression of c-myc, Cyclin D1, E and A and E2f1 genes associated with pRb hyperphosphorylation occurs in neoplastic nodules and HCCs of genetically susceptible F344 rats but not in corresponding lesions from resistant Brown Norway (BN) rats. 6 Moreover, Cdc37, Hsp90 and Crm1, which protect cell cycle kinases from inhibition by p16 INK4A , are upregulated in the lesions of F344 rats but not in BN rats. 5 Interestingly, analogous changes were found in human HCCs, with a HCC subtype with better prognosis (based on survival length after surgical resection) resembling HCCs of resistant rats and a subtype with poorer prognosis similar to the lesions of susceptible rats. 5,6 Furthermore, higher expression of extracellular signal-regulated kinase (ERK) occurs in nodules and HCC of F344 rats and human HCC with poorer prognosis, whereas ERK is slightly induced in corresponding lesions from BN strain and human HCC with better prognosis. 7,8 Emerging evidence indicates a posttranscriptional regulation of oncosuppressors by the S-phase kinase-associated protein 1 (SKP1)/CUL1/F-box protein (SCF) complex, an ubiquitin ligase implicated in the regulation of G1-S transition. [9][10][11] The SKP1/CUL1/F-box protein SKP2 (SCF SKP2 ) complex is involved in ubiquitination and proteasomal degradation of P21 WAF1 , P27 KIP1 , P57 KIP2 , PRb-related P130, free Cyclin E, E2F1, RASSF1A and FOXO1 in various tumors. [9][10][11] Furthermore, recent results have shown that the control of G1/S phases of cell cycle largely depends on the abundance of p21 WAF1 , p27 Kip1 , p57 KIP2 , p130 and RASSF1A proteins, as determined by their ubiquitination dependent on the CDC28 protein kinase b1 (CKS1) and SKP2 ligase component of SCF SKP2 complex (Calvisi et al., submitted).
Decrease in growth ability and/or rise in redifferentiation of preneoplastic lesions characterize rodent strains resistant to hepatocarcinogenesis. 5,6 Consequently, studies on the mechanisms underlying the acquisition of a phenotype susceptible/resistant to hepatocarcinogenesis in rodent strains, carrying preneoplastic lesions differently prone to progress to HCC, may lead to discovery of prognostic markers and therapeutic targets for the human disease. In this article, we comparatively investigated, in preneoplastic and neoplastic liver lesions of genetically susceptible F344 rats and resistant BN rats, the genetic control of CKS1-SKP2 ubiquitin ligase and the role of the ubiquitination of p21 WAF1 , p27 Kip1 , p57 KIP2 , pRb/p130 and RASSF1A cell cycle regulators in the determination of a phenotype susceptible to HCC development. We also evaluated the correlation of cell cycle regulator expression with the proliferation rate of human HCCs.

Animals and treatments
F344 and BN rats (Charles-River-Italia, Calco, Italy) were fed, housed and treated according to the ''resistant hepatocyte'' protocol, 12 consisting of a 150 mg/kg i.p. dose of diethylnitrosamine followed by a 15-day feeding a 0.02% 2-acetylaminofluorene containing hyperprotein diet, with a partial hepatectomy at the midpoint of this feeding. Preneoplastic liver, including foci of altered hepatocytes 4-6 weeks after initiation, early nodules (12 weeks), dysplastic nodules (32 weeks) and HCCs (50-60 weeks) were collected. Total preneoplastic liver, at 4/6 weeks, and pools of isolated early or dysplastic nodules and single HCCs from each animal were used for mRNA and protein analyses. Animal received human care, and study protocols were in compliance with the National Institutes of Health guidelines for use of laboratory animals. Six normal livers and 60 HCC and corresponding surrounding nontumorous livers were used. Supporting Information Table I shows patients' clinicopathological features. Liver tissues were kindly provided by Dr. S. S. Thorgeirsson (NCI, Bethesda, MD). Institutional Review Board approval was obtained at participating hospitals and the National Institutes of Health.

Histology and immunohistochemistry
Liver tissues were fixed in neutral buffered paraformaldehyde and processed for hematoxylin/eosin staining or glutathione-S-transferase, 7-7 isoform (GST 7-7) or KI67 immumnohistochemistry as published. 6 Number and volume of GST 7-7 positive rat preneoplastic and neoplastic lesions were calculated by morphometric analysis. 6

Labeling and proliferation indices
The labeling index (LI) of rat lesions was evaluated 2 h after intraperitoneal injection to rats of 2-bromo-3-deoxyuridine (BrdU) (5 mg/100 g body weight) by determining nuclear incorporation of BrdU, using the ''cell proliferation kit'' (Amersham Biosciences, Cologno Monzese, Italy), in 3,000 hepatocytes. The proliferation index was determined in human HCC by counting Ki-67-positive cells on 3,000 hepatocytes. 5 Quantitative real-time RT-PCR Primers for p21 WAF1 , p27 Kip1 , p57 KIP2 and pRb/p130 genes and Ribonucleic acid ribosomal 18S (RNR-18) were chosen by the ''Assays-on-DemandTM Products'' (Applied Biosystems, Foster City, CA). PCR reactions were performed with 75-300 ng of cDNA, using an ABI Prism 7000 and Taq Man Universal PCR Master Mix (Applied Biosystems) as reported. 5 Immunoprecipitation Proteins were extracted from tissues as reported. 5 For immunoprecipitation analysis, 500 lg of hepatic tissue lysates were used. Samples were immunoprecipitated with specific antibodies (Supporting Information Table II). The immunocomplexes were separated by SDS-PAGE and treated with biotinylated secondary antibody. 5 The complexes of Skp2 or ubiquitin with Cks1, p21 WAF1 , p27 KIP1 , p57 KIP2 , P130 and RassF1A were determined by immunoprecipitating SKP2 or ubiquitin with anti-Skp2 and anti-ubiquitin antibody, respectively, and probing the membranes with antibodies against Cks1, p21 WAF1 , p27 KIP1 , p57 KIP2 , P130 and RassF1A. Phosphorylation of RassF1A was determined with a rabbit antiphosphorylated (pSer203) RassF1A antibody. 10 Bands were quantified in arbitrary units by Molecular Imager ChemiDoc XRS, using the Quantity One 1D Analysis Software, and normalized to b-actin levels. To test the specificity of the primary antibodies, negative controls were made, in which the antibodies used to immunoprecipitate were incubated 2 h at room temperature with the respective immunogen peptide (in 200-fold molar excess to antibodies), before immunoprecipitation. In coimmunoprecipitation studies, the specificity of coimmunoprecipitation was evaluated by preincubating the second primary antibody, before immunoblotting, with the respective immunogen peptide.

Statistical analysis
Student's t and Tuckey-Kramer tests were used to evaluate statistical significance. Values of p < 0.05 were considered to be significant. Linear regression analyses were done by GraphPad Instat 3 software. Data are expressed as means 6 SD.

Development of liver lesions
Four weeks after initiation, GST 7-7-positive foci of altered hepatocytes occupied $12% of the liver in both rat strains. Further evolution of foci to early nodules, larger than a liver lobule and compressing surrounding parenchyma, was more pronounced in F344 than BN rats, and at 12 weeks, the lesions occupied 54% and 16% of liver in F344 and BN rats, respectively (p < 0.0001, n ¼ 5). At 32 weeks, dysplastic nodules occupied only 15% of liver in BN rats against 74% in F344 rats, and at 50-60 weeks, HCCs were present in all F344 rats but only in 35% of BN rats. Foci of altered hepatocytes and early nodules prevalently consisted of clear/eosinophilic cells in both rat strains, but at 32 weeks low-and high-grade dysplastic nodules occurred in liver of BN and F344 rats, respectively (Supporting Information Fig. S1). At 50-60 weeks, HCC were well-differentiated/moderately differentiated in BN rats and moderately/poorly differentiated in F344 rats (Fig. S1). At 4 and 6 weeks, the LI of liver foci was 14.08 6 2.6 and 11.44 6 1.9 in preneoplastic liver of F344 rats and 12.64 6 2.2 and 11.06 6 2.0 in BN rats. The LI of early nodules, dysplastic nodules and HCC was 9.07 6 1.2, 11.24 6 1.09 and 9.62 6 1.3, respectively, in F344 rats, and 3.2 6 0.91, 5.01 6 0.78 and 6.78 6 1.06 in BN rats (means 6 SD, n ¼ 5; Tuckey-Kramer test: F344 vs. BN at least p < 0.005 for early and dysplastic nodules and HCC). Figure 1a shows progressive increase in p21 WAF1 , p27 KIP1 , p57 KIP2 and p130 mRNA level in preneoplastic lesions, nodules and HCCs of F344 rats. Lower or no increases occurred in BN rats. A similar behavior of RassF1A mRNA was found previously in the 2 rat strains. 7 A different situation occurred at the protein level (Figs. 1b and 1c). Indeed, p21 WAF1 , p27 KIP1 , p57 KIP , p130 and RassF1A exhibited no change or relatively low increase in preneoplastic liver, compared with normal liver, without interstrain differences. However, in

Short Report
Calvisi et al.
early and/or dysplastic nodules and HCC of F344 rats, the absence of significant further increases in protein levels contrasted with the consistent rise in corresponding BN lesions, suggesting interstrain differences in posttranscriptional regulation of protein levels resulting in higher levels of cell cycle inhibitors in slowly growing lesions of these rats.

Ubiquitination of cell cycle regulators
To assess the posttranslational mechanism determining the protein levels of cell cycle regulators in liver lesions of F344 and BN rats, we evaluated the expression of 2 key component of the SCF SKP2 complex, Skp2 and Cks1 and Skp2 interaction with cell cycle regulators. The results in Figure 2 show a progressive increase in Cks1, Skp2, Skp2 complexes with Cks1, p21 WAF1 , p27 KIP1 , p57 KIP and p130 in preneoplastic liver, nodules and HCC of both F344 and BN rats. Consistent increase in Cdk4-RassF1A complex, phosphorylated RassF1A and Skp2-RassF1A complex only occurred in early nodules and/or dysplastic nodules and HCC of both strains. Highest levels of all protein and protein complexes tested occurred in early nodules and/or dysplastic nodules and HCC, but significantly lower levels were found in nodules and HCCs of BN than F344 rats. Consistent with these observations was the presence of lower amounts of ubiquitinated p21 WAF1 , p27 KIP1 , p57 KIP2 , p130 and RassF1A proteins in 12 weeks nodules and/or dysplastic nodules and HCC of BN than F344 rats (Fig. 3).

Correlation studies
The aforementioned results clearly indicate that posttranslational regulation of cell cycle inhibitors is associated with rapid growth and progression of rat HCC. To further substantiate the role of this phenomenon in cancer progression, we comparatively evaluated the correlation of regulators expression, at protein and mRNA levels, with growth rate of a population of human HCCs with highly heterogeneous proliferation index (Ki67 protein, 21.12 6 12.03; mRNA, 158.62 6 131.65; mean 6 SD, n ¼ 60). Linear regression analysis (Table 1) indicated the existence of a strong and highly significant negative correlation of Ki67 protein levels with P21 WAF1 , P27 KIP1 , P57 KIP2 and P130 protein levels. In contrast, very weak (P21 WAF1 ) correlation or no correlation was found for mRNA levels.

Discussion
G1 phase progression of mammalian cells is mainly controlled by the cyclin-cyclin-dependent kinase (CDK)-CDK inhibitor-retinoblastoma protein (pRb) regulatory pathway. Cell cycle regulators controlling G1 phase progression are involved in the development of many cancer types, including HCC. 13 In HCC, the CDK inhibitors encoded by p16INK4, WAF/KIP family genes, p130 and RASSF1A may be downregulated by methylation of gene promoters. 14,15 However, in numerous human and rodents, HCCs cell cycle regulators are overexpressed and predominantly inactivated by posttranscriptional regulation. 5,16,17 Mounting evidence underlines the role of CKS1-SKP2 ligase in ubiquitination and proteasomal degradation of P21 WAF1 , P27 KIP1 , P57 KIP2 and RASSF1A in different tumor types, 10,11,18,19 including human HCC (Calvisi et al., submitted). The present results strongly suggest that the expression of CKS1-SKP2 ligase and the ubiquitination and proteasomal degradation of WAF/KIP family regulators of cell cycle as well as p130 and RassF1A are under control of cancer susceptibility genes. The lack of knowledge of the nature of genes predisposing to hepatocarcinogenesis precludes the possibility to identify the epistatic interactions  Calvisi et al. regulating the activity of Ubiquitin ligase and proteasome. Furthermore, it must be considered the possibility that the influence of susceptibility genes could be indirect. Indeed, the FoxM1 gene, whose expression is under the control of HCC susceptibility genes, 20 may activate the Cks1-Skp2 ligase. 9 Moreover, Skp2 is targeted by E2F 21 and, consequently, its upregulation could simply reflect the deregulation of Rb pathway in HCC. 6 Whatever the polygenic mechanism controlling Cks1-Skp2 ligase is, our data indicate that Cks1 and Skp2 upregulation is associated with the acquisition of a susceptible phenotype in rats. Indeed, the activity of Cks1-Skp2 ligase sharply increases 6/12 weeks after initiation, in F344 rats, namely in the period of time when preneoplastic lesions acquire the capacity to grow autonomously. 5 Conversely, the activity of Cks1-Skp2 ligase is relatively low and the expression of cell cycle inhibitors relatively high in BN rat lesions after the 6th week, namely in coincidence with the acquisition of the resistant phenotype by these rats. 5 Notably, our data demonstrate the existence of an inverse correlation of HCC cell proliferation with protein levels, but not with mRNA levels, of cell cycle regulators. This indicates a role of posttranslational modifications of cell cycle regulators in the control of the proliferation of human HCC and suggests that posttranslational modifications of cell cycle inhibitors may have a prognostic value. This is consistent with recent observations (Calvisi et al., submitted) indicating that similar to HCC of F344 rats, human HCC with poorer prognosis exhibit active ubiquitination and degradation of cell cycle regulatory proteins, analogous to those found in HCCs of F344, whereas cell cycle regulators undergo lower changes in human HCC with better prognosis, as in the lesions of BN rats. Furthermore, a correlation of SKP2 protein levels with human HCC prolifera-tion was found (Calvisi et al., submitted). The existence of an association of growth rate and prognosis with CKS1-SKP2 ligase activity in human HCC does not necessarily imply a genetic regulation of CKS1-SKP2 ligase in humans. Nevertheless, epidemiologic studies on families at risk and segregation studies on human population 4,22-24 clearly indicate that genetic susceptibility is one of the factors involved in familial aggregations of HCCs, even in HBV endemic area. 2 These observations support a polygenic model of autosomal recessive inheritance, involved in the genetic predisposition of HCC, with a major locus and various low-penetrance genes, at play in different subsets of population. An analogous genetic model controls hepatocarcinogenesis in rodents. 3 This suggests a possible link between susceptibility genes and the mechanisms determining HCC onset at earlier age, growth rate and progression in humans, analogous to that discovered in rodents HCC. However, although numerous susceptibility/resistance loci governing rodents hepatocarcinogenesis are syntenic to human genome sites where frequent aberrations (i.e., loss of heterozygosis, translocations, deletions, amplifications) occur, 3 there is no proof that the same genes predispose to liver cancer in rodents and humans and further studies are necessary to address this pathogenetic aspect of HCC.

Short Report
Collectively, previous 5 and present results indicate that the predominance of susceptibility or resistance genes in individuals largely influences the molecular control of HCC cell proliferation and progression. The susceptible individuals develop various adaptive mechanisms for protection against tumor suppressor inhibitors of cell cycle, such as p16 INK4A,5 and WAF1/KIP family genes, p130 and RassF1A (this work). This confers to HCC cells the ability to proliferate and progress even under stressful conditions. The expression of cell cycle regulators at protein and mRNA levels was determined by immunoprecipitation and quantitative RT-PCR, respectively. Data are means 6 SD of 60 human HCC. Linear regression analysis of the correlation of the proliferation index (percentage of Ki67-positive cells) with protein and mRNA levels of cell cycle regulators gave negative R values indicating a negative correlation.