Role of HSP90, CDC37, and CRM1 as modulators of P16INK4A activity in rat liver carcinogenesis and human liver cancer

Authors

  • Rosa M. Pascale,

    1. Department of Biomedical Sciences, Division of Experimental Pathology and Oncology, University of Sassari, Sassari, Italy
    Search for more papers by this author
  • Maria M. Simile,

    1. Department of Biomedical Sciences, Division of Experimental Pathology and Oncology, University of Sassari, Sassari, Italy
    Search for more papers by this author
  • Diego F. Calvisi,

    1. Department of Biomedical Sciences, Division of Experimental Pathology and Oncology, University of Sassari, Sassari, Italy
    Search for more papers by this author
  • Maddalena Frau,

    1. Department of Biomedical Sciences, Division of Experimental Pathology and Oncology, University of Sassari, Sassari, Italy
    Search for more papers by this author
  • Maria R. Muroni,

    1. Department of Biomedical Sciences, Division of Experimental Pathology and Oncology, University of Sassari, Sassari, Italy
    Search for more papers by this author
  • Maria A. Seddaiu,

    1. Department of Biomedical Sciences, Division of Experimental Pathology and Oncology, University of Sassari, Sassari, Italy
    Search for more papers by this author
  • Lucia Daino,

    1. Department of Biomedical Sciences, Division of Experimental Pathology and Oncology, University of Sassari, Sassari, Italy
    Search for more papers by this author
  • Maria D. Muntoni,

    1. Department of Biomedical Sciences, Division of Experimental Pathology and Oncology, University of Sassari, Sassari, Italy
    Search for more papers by this author
  • Maria R. De Miglio,

    1. Department of Biomedical Sciences, Division of Experimental Pathology and Oncology, University of Sassari, Sassari, Italy
    Search for more papers by this author
  • Snorri S. Thorgeirsson,

    1. Laboratory of Experimental Carcinogenesis, Center for Cancer Research, National Cancer Institute, National Institutes of Health, Bethesda, MD
    Search for more papers by this author
  • Francesco Feo

    Corresponding author
    1. Department of Biomedical Sciences, Division of Experimental Pathology and Oncology, University of Sassari, Sassari, Italy
    • Dipartimento di Scienze Biomediche, Sezione di Patologia Sperimentale e Oncologia, Università di Sassari, Via P, Manzella 4, 07100 Sassari, Italy
    Search for more papers by this author
    • fax: (39) 079-228485


  • Potential conflict of interest: Nothing to report.

Abstract

Current evidence indicates that neoplastic nodules induced in liver of Brown Norway (BN) rats genetically resistant to hepatocarcinogenesis are not prone to evolve into hepatocellular carcinoma. We show that BN rats subjected to diethylnitrosamine/2-acetylaminofluorene/partial hepatectomy treatment with a “resistant hepatocyte” protocol displayed higher number of glutathione-S-transferase 7-7(+) hepatocytes when compared with susceptible Fisher 344 (F344) rats, both during and at the end of 2-acetylaminofluorene treatment. However, DNA synthesis declined in BN but not F344 rats after completion of reparative growth. Upregulation of p16INK4A, Hsp90, and Cdc37 genes; an increase in Cdc37-Cdk4 complexes; and a decrease in p16INK4A-Cdk4 complexes occurred in preneoplastic liver, nodules, and hepatocellular carcinoma of F344 rats. These parameters did not change significantly in BN rats. E2f4 was equally expressed in the lesions of both strains, but Crm1 expression and levels of E2f4-Crm1 complex were higher in F344 rats. Marked upregulation of P16INK4A was associated with moderate overexpression of HSP90, CDC37, E2F4, and CRM1 in human hepatocellular carcinomas with a better prognosis. In contrast, strong induction of HSP90, CDC37, and E2F4 was paralleled by P16INK4A downregulation and high levels of HSP90-CDK4 and CDC37-CDK4 complexes in hepatocellular carcinomas with poorer prognosis. CDC37 downregulation by small interfering RNA inhibited in vitro growth of HepG2 cells. In conclusion, our findings underline the role of Hsp90/Cdc37 and E2f4/Crm1 systems in the acquisition of a susceptible or resistant carcinogenic phenotype. The results also suggest that protection by CDC37 and CRM1 against growth restraint by P16INK4A influences the prognosis of human hepatocellular carcinoma.(HEPATOLOGY 2005;42:1310–1319.)

Human and rodent hepatocarcinogenesis is characterized by the progressive development of foci of altered hepatocytes (FAH), neoplastic nodules, and hepatocellular carcinoma (HCC).1, 2 Epidemiological evidence suggests a polygenic predisposition for human HCC.2 Studies on rodents allowed mapping the loci responsible for the progression of preneoplastic hepatocytes to malignancy.3–7 However, the nature and temporal occurrence of the events underlying resistance to HCC remain uncertain. The evaluation of these mechanisms could help in understanding the molecular pathways affected by susceptibility genes to evaluate cancer risk and identify potentially reversible phases of carcinogenesis.

Previous research has shown cell cycle deregulation in neoplastic liver lesions induced by the “resistant hepatocyte” protocol in susceptible Fisher 344 (F344) rats.8–10 Lower or no changes were found in resistant Brown Norway (BN) rats in which overexpression of p16INK4A, a well-known inhibitor of the cyclin-dependent kinase (Cdk) 4/6, occurs. Formation of p16INK4A-Cdk4/6 complexes decreases Cdk4/6 availability for the formation of CDK4–cyclin D1 complexes that inactivate pRb suppressor activity.11 Recent studies focused on the role of the cell division cycle 37 (Cdc37) gene in maintaining Cdk4/6 in an active state. Cdc37 forms a chaperone complex with heat shock protein (Hsp) 90, which contributes to the stability and activity of numerous protein kinases.12–14Cdc37 may behave as an oncogene: MMTV-Cdc37 transgenic mice develop breast tumors and collaborate with MMTV-c-myc or cyclin D1 in the transformation of multiple tissues.15 Another molecule that interferes with p16INK4Aactivity is “required for chromosomal-region-maintenance 1” (Crm1), a receptor for various proteins that promotes the cytoplasmic redistribution of E2f transcription factor 4 (E2f4).16 E2f4 is a transcription repressor acting as a downstream mediator of p16INK4A.16–19

The present study evaluated whether differences in the regulation of p16INK4A, Cdc37, Hsp90, E2f4, and Crm1 genes influence the development of preneoplastic and neoplastic liver lesions in rat strains with different genetic predisposition to hepatocarcinogenesis, as well as the progression of human HCCs. The results revealed that resistance of BN rats to hepatocarcinogenesis depends on the incapacity of initiated cells to grow autonomously upon exhaustion of the promoting stimulus. p16INK4A is overexpressed in preneoplastic and neoplastic lesions of both BN and F344 rats as well as in human HCC with a better prognosis (HCCB), while being downregulated in human HCC with a poorer prognosis (HCCP). Cdc37/Hsp90 and Crm1 protect more efficiently the lesions of susceptible rats and human HCCB against cell cycle inhibition by p16INK4A.

Abbreviations

BN, Brown Norway; F344, Fisher 344; FAH, foci of altered hepatocytes; HCC, hepatocellular carcinoma; Cdk, cyclin-dependent kinase; Hsp, heat shock protein; HCCB, better prognosis of HCC; HCCP, poorer prognosis of HCC; Crm1, required for chromosomal-region-maintenance 1; PH, partial hepatectomy; BrdU, 2-bromo-3-deoxyuridine; siRNA, small interfering RNA.

Materials and Methods

Animals and Treatments.

F344 and BN rats (140–160 g; Charles-River-Italia, Calco, Italy) were fed and housed as previously reported.3 Seventy F344 and BN rats were treated according to a “resistant hepatocyte” protocol10 including a 150-mg/kg intraperitoneal dose of diethylnitrosamine followed by a 15-day feeding of a hyperprotein diet containing 0.02% 2-acetylaminofluorene, with partial hepatectomy (PH) at the midpoint of this feeding. For adducts analysis, 10 F344 and BN rats received a diethylnitrosamine dose (150 mg/kg intraperitoneally) or three 2-acetylaminofluorene daily doses (30 mg/kg each by gavage) 24 hours before being killed. Rats were killed by way of bleeding through the thoracic aorta under ether anesthesia. Preneoplastic liver (4–15 weeks after initiation), nodules (32 weeks), and HCCs (57 weeks) were collected and used for experiments. Animals received humane care, and study protocols were in compliance with our institution's guidelines for the use of laboratory animals.

Human Tissue Samples.

Six normal livers, 42 HCCs, and corresponding surrounding tissues were used. Clinicopathological features of the patients are shown in Table 1.20 Liver samples were kindly provided by Dr. Z. Sung (National Laboratory of Molecular Oncology, Cancer Institute, Beijing, China) and Liver Tissue Procurement and Distribution System (Minneapolis, MN; Pittsburgh, PA; Richmond, VA), funded by National Institutes of Health Contract N01-DK-9-2310. Institutional review board approval was obtained at participating hospitals and the National Institutes of Health.

Table 1. Clinicopathological Features of HCC Patients
VariableFeatures
HCCBHCCP
No. of patients2121
 Male2019
 Female12
Age, yrs (mean ± SD)54.2 ± 8.059.5 ± 12.4
Etiology  
 Hepatitis B virus1412
 Hepatitis C virus36
 Ethanol01
 Hemochromatosis01
 Not available41
Cirrhosis  
 +1417
 −74
Tumor size  
 <5 cm106
 >5 cm1115
Survival after partial liver resection>3 yrs<3 yrs
Average (mo)67.519.19

Histology and Immunohistochemistry.

Preneoplastic liver and isolated nodules were processed for glutathione-S-transferase 7-7 (GST7-7), p16INK4A, Cdc37, E2f4, and Crm1 immunohistochemistry as previously described3 using specific antibodies (Table 2), followed by incubation with biotinylated secondary antibodies and a standard avidin–biotin peroxidase complex method (Vector Laboratories, Burlingame, CA) and counterstaining with hematoxylin.

Table 2. Primary Antibodies Used for Western Blot and Immunoprecipitation Analysis
ProteinAntibodyEpitope Mapping
  • *

    Provided by Santa Cruz Biotechnology (Santa Cruz, CA).

  • Provided by Stressgen Biotechnologies (Victoria, Canada).

  • Provided by Chemicon International (Temecula, CA).

Rat  
 p16INK4ARabbit polyclonalFull length*
 p50CDC37Mouse monoclonalAmino acids 108–378*
 E2f4Rabbit polyclonalCOOH terminus*
 Crm1Goat polyclonalCOOH terminus*
Human  
 P16INK4ARabbit polyclonalFull length*
 P50CDC37Mouse monoclonalAmino acids 108–378*
 E2F4Mouse monoclonalAmino acids 108–300*
 CRM1Rabbit polyclonalCOOH terminus*
 CDK4Rabbit polyclonalCOOH terminus*
 HSP90Rat monoclonalFull length
 ACTINMouse monoclonalAmino acids 50–70

The number of single cells, doublets, triplets, microfoci (4–10 cells), and minifoci (11–100 cells) per cm2 in the liver, and the number and mean volume of FAH and nodules in the entire liver were calculated via computer-assisted morphometric analysis.21 The percentage of cells incorporating 2-bromo-3-deoxyuridine (BrdU) (labeling index) was evaluated 2 hours after intraperitoneal injection of BrdU (5 mg/100 g body weight)22 by determining BrdU nuclear incorporation using a cell proliferation kit (Amersham Biosciences, Cologno Monzese, Italy). Remodeling lesions were identified and quantified as previously described.22

Small Interfering RNA Treatment.

Human HepG2 liver carcinoma cells were cultured as reported.23 Cultures were incubated for 12 hours at 37°C before transfection with the following double-strand RNA oligos specific for human CDC37 coding region: 5′-CATGAGTGCACATTTCTCCtc-3′, 5′-TCCAGCTCCTTCGTTTCCtc-3′, 5′-GTCCAGTTCCTCCTTCTCCtt-3′ (Ambion, Austin, TX). The small interfering RNA (siRNA) duplexes and scramble oligo sequence (final concentration 100 nmol/L) were transfected using the siPORT NeoFX system (Ambion). [3H]thymidine DNA incorporation was used to measure cell proliferation.23

DNA Adduct Determination.

32P-postlabeling assay of DNA adducts was performed according to Gupta24 by digesting DNA from F344 and BN rat liver with micrococcal nuclease (Roche Diagnostics, Monza, Italy) and spleen exonuclease (Calbiochem, Inalco, Milan, Italy). Digested DNA was converted to 32P-labeled 2′-deoxynucleotide-3′,5′-diphosphates by incubation with [γ-32P]-adenosine triphosphate (Amersham) and T4 polynucleotide kinase (Roche Diagnostics). Adducts and unmodified nucleotides were separated on polyethyleneimide, and radioactivity was counted with an InstantImager (Packard Instrument Co., Meriden, CT). Standardized deoxyguanosine and deoxythymidine adducts were prepared as previously described.25

Real-Time Reverse-Transcriptase Polymerase Chain Reaction.

Primers for p16INK4A, Cdc37, Hsp90, E2f4, Crm1, and RNA ribosomal 18S (RNR-18) were chosen using Assays-on-Demand products or Custom Assay-on-Demand (Applied Biosystems, Foster City, CA). Polymerase chain reaction was performed with 75–300 ng of complementary DNA using an ABI Prism 7000 and Taq Man Universal PCR Master Mix (Applied Biosystems). Cycling conditions were: 10 minutes of denaturation at 95°C and 40 cycles at 95°C for 15 seconds and at 60°C for 1 minute. Quantitative values were calculated using PE Biosystems Analysis software (PE Biosystems, San Jose, CA).

Western Blotting.

Human liver samples were homogenized in lysis buffer and sonicated.26 Protein aliquots of 40 μg were denatured, separated via SDS-PAGE, and transferred onto nitrocellulose membranes via electroblotting. The membranes were blocked in 5% nonfat dry milk, probed with specific antibodies (Table 2), incubated with secondary antibodies, and revealed with the Super Signal West Pico (Pierce Chemical Co., New York, NY). Densities of the protein bands were normalized to actin levels and calculated by ImageQuant Software.

Immunoprecipitation Analysis.

Five hundred milligrams of normal rat liver, preneoplastic liver (7 weeks after initiation), and nodules; and 300 mg human normal liver, HCCs, and surrounding liver were homogenized and processed.9 Immunoprecipitations were performed using 4 μg agarose-conjugated antibodies against CDK4, p16INK4A, p50CDC37, E2f4, and Crm1. Immunoprecipitated samples were separated via SDS-PAGE and treated with biotinylated secondary antibody.9 Immunocomplexes were revealed and quantified via enhanced chemiluminescence (Amersham). In rat samples, Cdk4-p16INK4A, Cdk4-Cdc37, and E2f4-Crm1 complexes were assessed by immunoprecipitating p16INK4A, Cdc37, and E2f4 with specific antibodies. Immunoprecipitated proteins were subjected to Western blotting, and the membranes were probed with anti-Cdk4 or anti-Crm1 antibodies. In human liver samples, immunocomplexes of CDK4 with P16INK4A, CDC37, and HSP90 were determined by immunoprecipitation of CDK4 followed by Western blotting of immunoprecipitated proteins, and the membranes were probed with anti-P16INK4A, anti-CDC37, and anti-HSP90 antibodies.

Statistical Analysis.

Results are expressed as the mean ± SD. Comparisons between means were made using Student t and Tuckey-Kramer tests. A P value of less than .05 was considered significant.

Results

Analysis of Initiation and Promotion Steps.

The number/cm2 of liver of single GST7-7(+) hepatocytes, doublets, triplets, microfoci, and minifoci showed insignificant interstrain differences before 2-acetylaminofluorene administration. However, a 1.5- to 2.6-fold and 3.9- to 10-fold increase in the number of all liver lesions occurred 12 hours after PH and at the end of 2-acetylaminofluorene treatment, respectively. These lesions, except for minifoci 24 hours after PH, were more numerous in BN rats than in F344 rats (Supplementary Fig. 1).

Liver reparative growth in untreated rats revealed a peak 48 hours after PH followed by a decrease to resting liver values at 168 hours, without interstrain differences (Fig. 1A). However, the labeling index was significantly higher in carcinogen-treated BN rats than in F344 rats 48 hours after PH (Fig. 1A, inset). The number/cm2 of GST7-7(+) hepatocytes, roughly corresponded to that of BrdU(+) hepatocytes and was significantly higher in BN rats compared with F344 rats (Fig. 1B). The coincidence of GST7-7 positivity and BrdU incorporation in liver lesions was suggested by the double positivity for BrdU and GST7-7 in minifoci 24 hours after PH (Fig. 1C). This indicates that a higher number of hepatocytes was resistant to 2-acetylaminofluorene mitoinhibition10 in BN rats.

Figure 1.

DNA synthesis in post-PH liver of F344 and BN rats. (A) Percentage of BrdU(+) cells in rats not treated with carcinogens and in rats subjected to the “resistant hepatocyte” model 48 hours after PH (inset); 1,000 to 4,000 GST7-7(+) hepatocytes were counted per liver. Data are expressed as the mean ± SD of 4 rats. (B) Number/cm2 of GST7-7(+) and BrdU(+) cells. Data are expressed as the mean ± SD of 6 rats. Student t test (A, inset) and (B): *BN versus F344 (P < .001). (C) BrdU incorporation into nuclei of GST7-7(+) minifoci 48 hours after PH. Original magnification ×140. F344, Fisher 344; BN, Brown Norway; LI, labeling index; PH, partial hepatectomy; BrdU, 2-bromo-3-deoxyuridine.

The analysis of DNA adduct formation in carcinogen-treated rats showed the absence of interstrain differences in alkyldeoxyguanosine and alkyldeoxythymidine adducts. Twenty-four hours after 2-acetylaminofluorene treatment, liver dG-AF adduct levels were 6.7 times higher in BN rats than in F344 rats (Supplementary Fig. 2).

Figure 2.

Labeling index of GST 7-7(+) foci (black diamonds and triangles) and surrounding liver (white diamonds and triangles) at different times from initiation and in nodules (32 weeks) in F344 and BN rats treated according to the “resistant hepatocyte” model. Data are expressed as the mean ± SD of 4–8 rats. Tuckey-Kramer test: ●GST7-7(+) versus surrounding liver (at least P < .05); *BN versus F344 (P < 0.001). LI, labeling index; F344, Fisher 344; BN, Brown Norway; N, nodules.

We timed the switch from an apparently susceptible to resistant phenotype in preneoplastic liver of BN rats by evaluating the labeling index during 2-acetylaminofluorene treatment (4th week), at the end of treatment (5th week), and subsequently up to 32 weeks. At the 4th week, the labeling index of surrounding liver was low in both strains (Fig. 2). It increased in surrounding liver of F344 and BN rats at 5 and 6 weeks, respectively, then decreased to normal liver values (0.26 ± 0.08) at 7 to 32 weeks. A relatively high labeling index occurred at 4 weeks in the FAH of both strains. It followed a decrease between 5 and 32 weeks in BN rats, up to values slightly higher than those of resting liver. In contrast, the labeling index of F344 rat lesions remained high and further increased at 15 and 32 weeks. Consequently, 7 weeks after initiation the number of FAH was still 1.8 times higher, but their volume was 8.3 times lower in BN rats than in F344 rats (Table 3). Twenty-four percent of these lesions remodeled, with progressive loss of GST7-7 positivity in resistant rats, against 13.2% in F344 rats. At 32 weeks, most lesions were atypical nodules/well-differentiated HCCs in F344 rats and clear/eosinophilic cell nodules with rare atypical nodules in BN rats. These lesions were slightly more numerous and bigger in F344 rats than in BN rats. In BN rats, 40.2% of nodules remodeled compared with 9.7% in F344 rats (Table 3). At 57 weeks, both incidence and multiplicity of HCCs were significantly lower in BN rats than in F344 rats.

Table 3. Development of Foci of Altered Hepatocytes and Preneoplastic Nodules in F344 and BN Rat Liver
LesionStrainIncidence (%)Number/LiverMean Volume (cm3 × 104)Remodeling
  • NOTE. Values are expressed as the mean ± SD of 5–10 rats.

  • *

    BN versus F344 (P < .0001).

FAH (7 wk)     
 F344 10066,300 ± 6,5407.60 ± 1.8113.2 ± 2.4
 BN 100136,200 ± 11,200*0.91 ± 0.44*24.0 ± 2.3*
Nodules (32 wk)     
 F344 100128,600 ± 8,27818.78 ±3.249.7 ± 2.3
 BN 100100,800 ± 7,460*3.96 ± 2.40*40.2 ± 3.7*
HCC (57 wk)     
 F344 1003.7 ± 0.82  
 BN 330.7 ± 0.20*  

Expression of Cell Cycle Regulators.

We next determined the expression of p16INK4A and some genes modulating p16INK4A activity, such as Hsp90, Cdc37, E2f4, and Crm1 in preneoplastic liver, nodules, and HCCs. No interstrain differences in the expression of these genes occurred in normal liver (Fig. 3). p16INK4A mRNA increased 19- to 27-fold with respect to normal liver in preneoplastic liver of both strains 5 to 15 weeks after initiation. At 32 and 57 weeks, nodules and HCCs exhibited relatively high p16INK4A mRNA contents, with a slight prevalence in BN rats compared with F344 rats (Fig. 3).

Figure 3.

Expression of p16INK4A, Cdc37, Hsp90, E2f4, and Crm1 genes in preneoplastic liver (5–15 weeks after initiation), nodules (32 weeks), and HCC (57 weeks) of F344 and BN rats treated according to the “resistant hepatocyte” model. Data are expressed as the mean ± SD of N target of 4–6 rats. N target = 2−ΔCt, where the ΔCt value of the sample was calculated by subtracting the average Ct value of the target gene from the average Ct value of the RNR-18 gene. Tuckey-Kramer test: ●carcinogen-treated versus control (P < .01); *BN versus F344 (at least P < .05). NT, N target; LI, labeling index; F344, Fisher 344; BN, Brown Norway; C, control; N, nodules; H, hepatocellular carcinoma.

Cdc37 and Hsp90 mRNAs increased 1.8- to 2.7-fold and 1.9- to 4.3-fold, respectively, in preneoplastic liver, nodules, and HCCs of F344 rats, compared with normal liver (Fig. 3). Cdc37 overexpression was detected in preneoplastic liver (all timepoints), nodules, and HCCs of BN rats. However, Cdc37 mRNA levels were 27% to 38% lower in the nodules and HCCs of BN rats compared with F344 rats. A relatively low increase or no change of Hsp90 mRNA occurred in preneoplastic liver and nodules of BN rats, except at 7 weeks and in HCC, where it increased to levels equivalent to those of F344 rats. A rise in E2f4 expression in preneoplastic liver of F344 rats 5 weeks after initiation was followed by a decrease to normal liver values at 7 and 15 weeks and in nodules. In preneoplastic liver of BN rats, a significant increase in E2f4 expression occurred only at 5 and 7 weeks, with respect to normal liver. E2f4 mRNA was relatively high in HCCs of both strains, without interstrain difference. Crm1 expression increased 2- to 2.9-fold in preneoplastic liver, nodules, and HCCs of F344 rats compared with normal liver. A significant Crm1 increase occurred in preneoplastic liver of BN rats only at 5 weeks. Crm1 mRNA was higher in HCCs than in normal liver of BN rats but was still significantly lower than in HCCs of F344 rats.

These results were roughly confirmed at the protein level. p16INK4A was approximately 1.5 to 3 times higher in preneoplastic liver 7 weeks after initiation and nodules of both strains compared with control liver (Fig. 4). A consistent rise in p50CDC37 occurred in preneoplastic liver and nodules of F344 rats, whereas no change or relatively low increase occurred in corresponding lesions of BN rats. Low changes of E2f4 level occurred in the lesions of both F344 and BN rats, whereas Crm1 increased 1.6- to 2-fold exclusively in the lesions of F344 rats. The inactive complex p16INK4A-Cdk4 did not appreciably change in preneoplastic liver and nodules of F344 rats compared with normal liver, whereas it increased in the lesions of BN rats (Fig. 5). Accordingly, p50Cdc37-Cdk4 and E2f4-Crm1 complexes underwent marked increase in preneoplastic liver and nodules of F344 rats, with respect to normal liver.

Figure 4.

Detection via immunoprecipitation of p16INK4A, p50CDC37, E2f4, and Crm1 in normal liver and preneoplastic liver and nodules 7 and 32 weeks after initiation, respectively, in rats treated according to the “resistant hepatocyte” model. Sample proteins and blocking peptides were coimmunoprecipitated with antibodies against p16INK4A, p50CDC37, E2f4, and Crm1 and separated via SDS-PAGE. Enhanced chemiluminescence was used to visualize the immune complexes. Upper panel: representative immunoprecipitation analysis, with control proteins (blocking peptides) in the last lane. Lower panel: chemiluminescence analysis showing mean ± SD of 3 and 6 rats for preneoplastic liver and nodules normalized to control values. Tuckey-Kramer test: ●carcinogen-treated versus control (P < .001); *BN versus F344 (at least P < .05). F344, Fisher 344; BN, Brown Norway; C, control; PL, preneoplastic liver; N, nodules.

Figure 5.

Detection of p16INK4A-Cdk4, p50CDC37-Cdk4, and E2f4-Crm1 complexes in preneoplastic liver and nodules 7 and 32 weeks after initiation, respectively, in rats treated according to the “resistant hepatocyte” model. Proteins were immunoprecipitated using antibody against p16, p50, and E2f4; separated via SDS-PAGE; and probed with antibodies against Cdk4 or Crm1. Enhanced chemiluminescence was used to visualize the immune complexes. F344, Fisher 344; C, control; PL, preneoplastic liver; N, nodules; BN, Brown Norway.

p16INK4A(+) nuclei were present at 7 weeks in FAH (Fig. 6C -D) and in surrounding parenchyma (not shown) of F344 and BN rats. Cdc37 immunohistochemistry revealed cytoplasmic and nuclear immunoreactivity in FAH of F344 rats, with relatively few Cdc37(+) nuclei and absence of cytoplasmic positivity in BN rats (Fig. 6E-F). E2f4 was both nuclear and/or cytoplasmic in the lesions of F344 rats, but prevalently nuclear in BN rats (Fig. 6G-H). An analogous situation of Cdc37 and E2f4 expression occurred in surrounding liver of both strains (not shown). Finally, Crm1 was almost exclusively present in the GST7-7(+) lesions (Fig. 6I-J), and prevalently localized in the cytoplasm of FAH of F344 (not shown) and BN (Fig. 6J, inset) rats.

Figure 6.

Representative immunohistochemical analysis of (A-B) a GST7-7(+) lesion for (C-D) p16INK4A, (E-F) Cdc37, (G-H) E2f4, and (I-J) Crm1 expression in F344 (left column) and BN (right column) rats 7 weeks after initiation according to the “resistant hepatocyte” model. Brown nuclei and yellow-brown cytoplasms indicate positivity for the proteins tested. (A,B,I,J) Original magnification ×25. (C-H) Original magnification ×120.

Gene Expression in Human HCC.

Western blot analysis results are shown in Fig. 7, with statistical analysis in Table 4. All genes were 1.4 to 11 times more expressed in surrounding liver than in normal liver, without significant differences between cases with better and poorer prognosis (Table 1)—except for E2F4, which was slightly more expressed in poorer compared with better prognosis. Gene expression further increased in HCCB. HCCP showed a great decrease in P16INK4A expression compared with normal liver and poorer prognosis of surrounding liver. The expression of all other genes, except for CRM1, was much higher in HCCP than in poorer prognosis of surrounding liver and HCCB. CRM1 expression increased only with respect to poorer prognosis of surrounding liver.

Figure 7.

Detection via Western blot analysis of p16INK4A, CDK4, p50CDC37, HSP90, E2F4, and CRM1 in normal liver (NL), nonneoplastic liver surrounding HCC with better and poorer prognosis, and HCC with longer (HCCB) and shorter (HCCP) survival rate. Upper panel: representative Western blot analysis. Lower panel: enhanced chemiluminescence analysis showing mean values ± SD of 6 normal livers, 21 HCCs, and corresponding surrounding livers with better or poorer prognosis. Each HCC was compared with surrounding liver from the same patient. Optical densities of the peaks were normalized to actin levels and expressed in arbitrary units. C, control; S, surrounding liver; HCCB, better prognosis of HCC; HCCP; poorer prognosis of HCC; SLB, better prognosis of surrounding liver; SLP; poorer prognosis of surrounding liver.

Table 4. Statistical Analysis of the Data in Figure 7
ProteinNL vs. SLBNL vs. SLPSLB vs. SLPSLB vs. HCCBSLP vs. HCCPHCCB vs. HCCP
  1. Abbreviations: NL, normal liver; SLB, better prognosis of surrounding liver; SLP, poorer prognosis of surrounding liver; NS, not significant.

P160.0120.026NS2.78 × 10−94.31 × 10−81.54 × 10−12
CDK41.75 × 10−52.88 × 10−6NS0.0318.9 × 10−192.83 × 10−17
CDC378.17 × 10−56.73 × 10−4NS1.23 × 10−62.04 × 10−164.05 × 10−12
HSP902.13 × 10−34.81 × 10−10NS4.88 × 10−82.87 × 10−201.65 × 10−16
E2F40.0190.0290.0482.50 × 10−129.87 × 10−133.04 × 10−13
CRM11.36 × 10−82.85 × 10−7NS2.63 × 10−33.56 × 10−3NS

P16-CDK4 complex increased in HCCB and decreased in HCCP compared with normal and surrounding liver (Fig. 8). HSP90-CDK4 and CDC37-CDK4 complexes greatly increased in HCCP, and at a lower extent in HCCB, with respect to surrounding liver.

Figure 8.

Detection of complexes of CDK4 with P16INK4A, HSP90, and p50CDC37 in normal liver, nonneoplastic surrounding liver, and HCC with longer (HCCB) and shorter (HCCP) survival rate. No differences between complexes of surrounding livers of HCCs with better and poorer prognosis were found, and only a representative experiment with surrounding liver of the HCC with better prognosis was included. Proteins were immunoprecipitated using antibody against CDK4, separated via SDS-PAGE, and probed with antibodies against P16INK4A, HSP90, and p50CDC37. Enhanced chemiluminescence was used to visualize the immune complexes. C, control; S, surrounding liver; HCCB, better prognosis of HCC; HCCP; poorer prognosis of HCC.

Anti-CDC37 siRNA Treatment.

The culture of HepG2 cells for 72 hours with siRNA(A) and (B) against CDC37 resulted in a 31% to 33% decrease in cell proliferation. After 48 and 75 hours of incubation with siRNA(C), 50% and 75% decreases in cell growth occurred, respectively. Scramble mRNA was ineffective. A significant decrease in CDC37 expression at 48 hours only occurred with siRNA(C), whereas at 72 hours a 25% decrease in CDC37 expression occurred with siRNA(A) and (B), and a 50% decrease occurred with siRNA(C). CDC37 siRNAs did not significantly affect P16INK4A expression (Supplementary Fig. 3).

Discussion

It has been hypothesized that the genetically transmitted resistance of rats to hepatocarcinogenesis depends on the poor capacity of preneoplastic liver lesions to progress to HCC. However, a higher number of GST7-7(+) lesions, suggestive of enhanced hepatocarcinogenesis initiation, occurs in the early stages of hepatocarcinogenesis in BN rats.3, 4 Indeed, the present results show that the number of GST7-7(+) lesions—including single cells scattered in liver parenchyma—strongly increases in both strains during 2-acetylaminofluorene treatment, and more markedly in BN rats. No interstrain difference in the growth of initiated hepatocytes suggestive of differences in 2-acetylaminofluorene selective pressure was detected under the promoting stimulus represented by PH. Furthermore, previous observations showed marked 2-acetylaminofluorene genotoxicity in rats subjected to resistant hepatocyte protocol.27 These findings, together with our observation of a high capacity of hepatocytes of BN rats to form dG-AF-DNA adducts, suggest an interference of 2-acetylaminofluorene treatment with hepatocarcinogenesis initiation in liver of diethylnitrosamine-treated rats. Seven weeks after initiation, in coincidence with the exhaustion of the reparative growth, a sharp decline in growth capacity and pronounced phenotypic reversion occurred in GST7-7(+) lesions of BN rats. Therefore, the acquisition of a resistant phenotype by these rats is a relatively late event, linked to the incapacity of initiated cells to grow autonomously and progress to HCC.

Previous work showed p16INK4A overexpression and low formation of cyclin D1–Cdk4 and E2f1-Dp1 complexes in lesions of resistant rat strains.9 Higher amounts of cyclin D1–Cdk4 and E2f1-Dp1 complexes occurred in HCCs of F344 rats, which also expressed p16INK4A, although at a lower level than BN rats.9 pRb was equally expressed in the two strains but was largely hypophosphorylated in HCCs of resistant rats.9 Promoter methylation, mutations, and loss of heterozygosity inactivate P16INK4A in human HCC.28–33 However, P16INK4A expression in early preneoplastic liver, nodules, and HCCs of resistant and susceptible rats, as well as HCCB, excludes its inactivation in these lesions. P16INK4A downregulation in HCCP is probably linked to the high degree of malignancy reached by these tumors. What causes a rise in P16INK4A expression in rat liver lesions and HCCB remains a matter of conjecture: P16INK4A is a stress-responsive gene whose expression is activated by signals from bmi134 and 14-3-3σ,35 and inhibited by junB, Erk signaling pathway,36 and hSNF5.37 Many properties of precancerous and cancerous cells can be regarded as adaptations to stress conditions38 that induce bmi1-dependent derepression of the Ink4A/Arf locus.34, 39 This stress situation may be initially triggered by DNA damage induced by diethylnitrosamine/2-acetylaminofluorene27 and further supported by the accumulation of DNA damage, hypoxia, oxidative damage, improper intercellular contacts, etc.39, 40 during the evolution of the process. The behavior of the signaling pathways involved in regulation of p16INK4A in various stages of hepatocarcinogenesis is not known and should be the focus of future studies.

No interstrain difference in p16INK4A expression in early preneoplastic liver could differentiate the susceptible from the resistant phenotype. p16INK4A overactivity could brake the cells at G1 to allow DNA repair or, if sufficient damage occurs, to favor apoptosis. Nonetheless, fast cell proliferation in the presence of P16INK4A overactivity occurs in recurrent prostate cancers,40 cervical tumors, and preneoplastic liver lesions during HPV infection,41 and HPV-driven immortalization of ectocervical cells.42 The question arises of how preneoplastic and neoplastic liver cells overexpressing P16INK4A undergo autonomous growth. It is well established that CDK4 forms inactive complexes with P16INK4A and active complexes with HSP90 and CDC37 chaperons.14, 15 Upregulation of Hsp90 and Cdc37 genes in preneoplastic and neoplastic lesions of F344 rats results in protection against Cdk4 inhibition by p16INK4A, as shown by the increase in active Cdc37-Cdk4 complex and maintenance of the p16INK4A-Cdk4 complex at normal liver levels. This is in agreement with the observation of high levels of pRb phosphorylation, E2f1-Dp1 complex, and DNA synthesis in these lesions.9 Cdc37 overexpression and eventual protection of Cdk4 from inhibition by p16INK4A occur in fast-growing GST7-7(+) lesions and surrounding liver of BN rats at 5 weeks. In later stages, Cdc37 expression is lower in GST7-7(+) lesions and HCCs of resistant rats compared with susceptible rats. This behavior, which is associated with high p16INK4A expression, could make Cdc37 protective action insufficient with consequent decrease in cell cycle activity, as suggested by the increase in p16INK4A-Cdk4 complex, and decrease in DNA synthesis of the lesions. Consistent with a protective role of CDC37 against growth inhibition is the association between a decrease in its expression and inhibition of DNA synthesis in HepG2 cells treated with anti-CDC37 siRNAs. This suggests that CDC37 could be a target for HCC chemoprevention and therapy. In the late stages of hepatocarcinogenesis, absence of interstrain differences in Hsp90 expression and low decrease in Cdc37 expression in HCCs of BN rats indicate an attenuation of interstrain differences in the protection by Hsp90/Cdc37 against p16INK4A.

Deregulation of E2f4 and Crm1 expression represents another mechanism contributing to the phenotypic behavior of GST7-7(+) lesions. E2f4 forms repressor complexes with pocket proteins p107 and p130 at E2f-responsive promoters in G0 and early G1.19, 43, 44 In late G1, pocket proteins dissociate from E2f4 that relocates to the cytoplasm. E2f4 is an essential downstream mediator of p16INK4A-induced growth arrest.18, 19 Because Crm1 performs the nuclear export of E2f4,18 its overproduction could override p16INK4A-induced G1 block. No marked changes in E2f4 expression occurred in preneoplastic liver and nodules, with respect to control liver in both strains, except at 5 weeks when a significant increase in E2f4 mRNA occurred. In contrast, Crm1 expression increased in preneoplastic liver and nodules of F344 rats, compared with normal liver, while exerting low protection in resistant rats. Indeed, a significant increase in Crm1 expression occurred only at 5 weeks in BN rats, but not at 7 to 15 weeks and in nodules, which are characterized by a relatively low growth rate. At 7 weeks, E2f4 was localized in the nuclei and cytoplasms of preneoplastic lesions of F344 rats, and only in nuclei of BN rat lesions. The levels of Crm1-E2f4 complexes in preneoplastic liver and nodules were lower in BN than F344 rats.

The protective mechanisms based on CDC37 and CRM1 expression may also play a role in human hepatocarcinogenesis. P16INK4A overexpression in HCCB was paralleled by a relatively modest rise in CDK4, HSP90, CDC37, and CRM1 levels, which was probably insufficient to exert a high protection from inhibition by P16INK4A. In these tumors, a high level of CDK4-P16 complex was associated with relatively low increases in CDK4-HSP90 and CDK4-CDC37 complexes. Under this aspect, HCCBs resemble HCCs of resistant rats characterized by slow progression and high Cdk4-p16 complex.9 This could reflect the behavior of tumors with a slow progression rate depending on both genetic and/or environmental factors. A different behavior characterized HCCP, which exhibited low P16INK4A expression and very high CDK4, HSP90, and CDC37 expression. The protective action of the HSP90/CDC37 system was evident, taking into account the presence of low levels of CDK4-P16 complexes and high levels of CDK4 complexes with HSP90 and CDC37. It cannot be excluded that the marked upregulation of HSP90 and CDC37 also protects other signal transduction pathways in which P16INK4A is not involved.45 Less clear is the role of E2F4 overexpression in HCCP. E2F transcription factors govern the expression of several genes controlling S-phase entry and progression, response to DNA damage, apoptosis, and differentiation.44 Thus, E2F4 deregulation may account for apoptosis, altered DNA damage response, and resistance to differentiating agents of HCCs. A better knowledge of multiple E2F4 activities could help in understanding its role in HCC. The behavior of P16INK4A and other cell cycle regulatory genes in HCCP is different, in various aspects, from that in rat liver cancer. This could at least partially depend on differences in the developmental stages of HCCP and HCC induced by the “resistant hepatocyte” protocol in rats.

In conclusion, the regulation of p16INK4A activity by Hsp90, Cdc37, E2f4, and Crm1 gene products is involved in the acquisition of a phenotype susceptible or resistant to hepatocarcinogenesis. The interaction of Hsp90/Cdc37 with p16INK4A and E2f4 with Crm1 prevents Cdk4 inhibition by p16INK4A during hepatocarcinogenesis promotion in both F344 and BN rats, thus allowing the growth of preneoplastic lesions. The failure of these protective mechanisms in the resistant rats, after cessation of the promoting stimulus, is associated with the poor capacity of preneoplastic lesions to undergo autonomous growth and progression to HCC. The mechanisms underlying the loss of Cdk4 protective systems in preneoplastic lesions of resistant rats are unknown and should be the object of future studies. At the very least, the protective mechanism based on the Hsp90/Cdc37 system operates in human hepatocarcinogenesis and could contribute to determine the prognosis. The inhibition of HepG2 cell growth by CDC37 silencing suggests that this gene may play a central role in HCC progression and prognosis, although the involvement of other mechanisms cannot be excluded.

Ancillary