Reduced expression of cell cycle regulator p18INK4C in human hepatocellular carcinoma

Authors


Abstract

Cyclins, cyclin-dependent kinases (Cdks), and Cdk inhibitors (CdkIs) are frequently altered in human cancer. p18INK4C, a member of the INK4 family of CdkIs, is a potential tumor-suppressor gene product. However, the expression of p18INK4C in hepatocellular carcinoma (HCC) remains unknown. The aim of this study was to examine the expression of p18INK4C in various liver diseases including HCC and to assess its clinical significance in HCC. To that end, we examined the expression of p18INK4C by immunohistochemistry in various liver diseases, including 51 HCCs, and also studied the relationship between p18INK4C expression, the phosphorylation of retinoblastoma protein (pRb), and the activity level of Cdk4 and Cdk6. Immunohistochemical analysis revealed the frequent loss of p18INK4C expression in HCC, especially in poorly differentiated HCC. The loss of p18INK4C expression was shown to be associated with a poor prognosis compared with that associated with p18INK4C- positivity. Further, the kinase activity of Cdk4 was found to be higher in p18INK4C-negative HCCs than in p18INK4C- positive HCCs. However, the level of Cdk6 activity was similar in the 2 groups of HCCs. In p18INK4C- positive HCCs, p18INK4C dominantly interacted with Cdk4 rather than with Cdk6. pRb phosphorylated at serine(Ser) 780 was detected more frequently in p18INK4C - negative than in p18INK4C - positive HCCs. In conclusion, the loss of p18INK4C expression may play a role in the differentiation and development of HCC through the up-regulation of Cdk4 activity. (HEPATOLOGY 2004;40:677–686.)

Recently, we have revealed that the aberrant expression of cell cycle-related proteins is one of the major factors contributing to the development of hepatocellular carcinoma (HCC).1–3 The cell cycle is mainly governed by various cyclin-dependent kinases (Cdks); the activity is regulated positively by cyclins, and negatively by Cdk inhibitors (CdkIs). It is also regulated by phosphorylation and dephosphorylation events.4–6 In mammalian cells to date, at least 2 distinct families of CdkIs are known. One, the p21 family, consists of p21CIP1/WAF1, p27KIP1, and p57KIP2, which are general inhibitors of G1 to S in the cell cycle. The other known CdkI family, the inhibitor of Cdk4 (INK4) family, consists of p16INK4A, p15INK4B, p18INK4C, and p19INK4D, which specifically inhibit cyclin D-related kinase activity by binding to Cdk4 or Cdk6.4–6

The CdkI proteins, potent negative regulators for the cell cycle, are potential tumor-suppressor gene products, and their loss might play an important role in the development of human cancers.4–11 In fact, loss of INK4 family members such as p15INK4B, p16INK4A, and p18INK4C by gene mutation, deletion, and/or methylation has been observed in a variety of human cancers.6–11 Clinical findings have revealed that the loss of p16INK4A expression is associated with poor prognosis for some types of human cancers.9–11 These studies have suggested that INK4 family members might play a role in the progression and prognosis of human cancers.

Some studies on the relationship between p18INK4C and HCC, on topics such as gene methylation of p18INK4C in HCC,12 involvement of p18INK4C in troglitazone-induced cell cycle arrest of human hepatoma cell lines,13 and the rate of carcinogen-induced liver tumor in p18INK4C mutant mice,14 have been reported. However, to our knowledge, the expression of p18INK4C has not yet been thoroughly examined in various liver diseases, including HCC. In this study, therefore, we focused on the relationship between p18INK4C expression and clinical significance in various liver diseases including HCC.

We examined the expression of p18INK4C immunohistochemically by using an avidin-biotin complex plus tyramide signal amplification method for signal enhancement in normal liver (NL), chronic hepatitis (CH), cirrhosis, and HCC. We also evaluated the relationships between p18INK4C expression and the levels of Cdk4 and Cdk6 activity in HCC. In addition, we examined whether the expression of p18INK4C was an important predictor of outcome in patients with HCC. In this article, we report that loss of p18INK4C expression is involved in hepatocarcinogenesis.

Abbreviations:

HCC, hepatocellular carcinoma; Cdk, cyclin-dependent kinase; CdkI, cyclin-dependent kinase inhibitor, INK4, inhibitor of Cdk4; NL, normal liver; CH, chronic hepatitis; HBsAg, hepatitis B surface antigen; HCV, hepatitis C virus; HRP, horseradish peroxidase; IgG, immunoglobulin G; pRb; retinoblastoma protein; LI, labeling index; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis; GST, glutathione S-transferase; HBV, hepatitis B virus; TNM, tumor regional lymph nodes, distant metastasis.

Patients and Methods

Patients.

Liver biopsy specimens were obtained from 30 patients with CH (22 males and 8 females; mean age, 48.4 ± 15.5 years; range, 21-80 years). Twenty-seven patients with CH were positive for hepatititis C virus (HCV) RNA, and 3 patients with CH were positive for the hepatitis B surface antigen (HBsAg). Of these 30 patients, 8 were in F1, 7 in F2, 6 in F3, and 9 in F4 according to Desmet's classification.15 Seven NL tissues were obtained from corresponding surgical cases of liver metastasis of colon cancer (5 males and 2 females; mean age, 58.1 ± 4.9 years; range, 52-67 years). These patients were negative for HCV RNA and HBsAg. Tissue samples of HCC were obtained from 51 patients with HCC during surgery (37 males and 14 females; mean age, 62.9 ± 7.2 years; range, 44-75 years). Forty-five patients with HCC were positive for HCV RNA, and 6 patients with HCC HBsAg. The clinical–pathological data for the patients with HCC is shown in Table 1. Of these 51 patients, 8 were in stage I, 13 in stage II, 11 in stage III, and 19 in stage IV according to the criteria of the International Union against Cancer and the American Joint Committee on Cancer.16 Histological grade of HCC was determined according to the criteria of the International Working Party.17 The numbers of patients with well-, moderately, and poorly differentiated HCCs were 16, 29, and 6, respectively (Table 1). Tissues were frozen immediately at −70°C. Informed consent was obtained from each patient prior to participation, and the experimental protocol was approved beforehand by the Human Subjects Committee of Kagawa Medical University.

Table 1. Relationship Between p18INK4C Immunoreactivity and Clinicopathological Features of HCC
 Patient No. (L.I. [%])p18INK4C Staining StatusP Value
Positive (L.I. [%])Negative (L.I. [%])Negative Rate (%)
  • Abbreviations: WD, well-differentiated; MD, moderately differentiated; PD, poorly differentiated.

  • *

    Histological grading of HCC was determined using the criteria of the International Working Party.17

  • TNM stage was determined using the classification proposed by the Internal Union Against Cancer and the American Joint Committee on Cancer.16

Sex     
 Male37 (8.5 ± 7.5)19 (14.6 ± 5.3)18 (2.3 ± 2.2)48.6 
 Female14 (7.8 ± 6.7)8 (12.6 ± 4.4)6 (1.3 ± 1.2)42.9.762
Age (y)     
 <6527 (7.5 ± 6.7)15 (11.2 ± 3.7)12 (2.9 ± 1.3)44.4 
 ≥6524 (6.7 ± 5.2)12 (10.9 ± 3.7)12 (2.5 ± 1.7)50.0.577
Viral infection     
 HCV-positive45 (8.7 ± 7.2)26 (13.8 ± 5.0)19 (1.7 ± 1.6)42.2 
 HBs Ag-positive6 (6.2 ± 6.1)1 (21.0 ± 0)5 (3.2 ± 3.1)83.3.195
*Histological grade     
 WD/MD45 (9.5 ± 7.0)27 (14.0 ± 5.1)18 (2.7 ± 2.1)41.3 
 PD6 (0 ± 0)0 (0 ± 0)6 (0 ± 0)100.024
TNM stage     
 I/II21 (11.9 ± 7.4)16 (14.8 ± 5.8)5 (2.6 ± 2.1)23.8 
 III/IV30 (5.9 ± 6.1)11 (12.9 ± 3.8)19 (2.2 ± 1.8)63.3.025

Chemicals and Antibodies.

Chemicals were obtained from Sigma Chemical Co. (Tokyo, Japan) or Wako Pure Chemical Co. (Tokyo, Japan). All primary antibodies were purchased from Santa Cruz Biotechnology (Tokyo, Japan). Secondary antibodies were from Amersham Life Science (Tokyo, Japan). Optimal dilutions of antibodies used for Western blot were as follows: polyclonal antibody H303 (anti-Cdk4), 1:200; polyclonal antibody C-21 (anti-Cdk6), 1:200; polyclonal antibody serine (Ser) 780 (antiphosphoserine Rb), 1:1000; monoclonal antibody TU-02 (α-tubulin), 1:1000; horseradish peroxidase (HRP)-conjugated anti-rabbit immunoglobulin G (IgG), 1:2000; and HRP-conjugated anti-mouse IgG, 1:2000. The phosphorylated Rb polyclonal antibody (Ser 780) reacts only with phosphorylated retinoblastoma protein (pRb) at Ser 780, which is specifically phosphorylated by cyclin D1/Cdk4, and detects a 105 kd protein corresponding to human pRb that includes amino acids 774 to 786.18

Immunohistochemical Examination.

We prepared 2 μm-thick sections from formalin-fixed, paraffin- embedded tissue blocks. Sections were stained by an avidin-biotin-peroxidase complex method (Funakoshi Chemical, Tokyo, Japan). The detection of p18INK4C was performed by immunohistochemical study using polyclonal antibody M-20 (anti-p18INK4C), as described in our previous article.3 All sections were examined independently by two observers (T.M., S.W.), who were blinded to the clinical information for each case. For each sample, the percentage of nuclei-immunostained cells was estimated per 1,000 cancer cells. In each case, the nuclei staining evaluation for p18INK4C was classified as negative or positive. The decisions made by the two pathologists were fairly consistent. Furthermore, samples that resulted in disagreement on the immunostaining data between two observers (T.M., S.W.) were discussed with a third observer (A.M.), using a multiheaded microscope until agreement between at least two observers was achieved. The mean labeling index (LI) of p18INK4C was 8.4 ± 7.2 in 51 patients with HCC; therefore, for assessment of the expression of p18INK4C, we categorized HCCs into 2 groups on the basis of the percentage of HCC cells positive for p18INK4C immunoreactivity: p18INK4C expression negative (<8.4%), and p18INK4C expression positive (≥8.4%). Necrotic areas and edges of the tissue sections were not included in the counting in order to avoid possible immunohistochemical false-positive results.

Tissue Lysates.

The tissue lysate was prepared as described in our previous reports.1, 2 Protein concentration was measured by a dye-binding protein assay using the Bradford method.19

Gel Electrophoresis and Western Blot.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was performed according to the method of Laemmli,20 and Western blot was performed as described by Towbin et al.,21 using primary antibodies and HRP-conjugated secondary antibodies. Immunoreactive proteins were visualized with an enhanced chemiluminescence detection system (Amersham) on X-ray film.

Kinase Assay of Cdk4 and Cdk6.

The kinase activities of Cdk4 and Cdk6 were examined by autoradiography, as described in our previous reports.1, 2

Analysis of p18INK4C-Bound Cdk4 and Cdk6 in HCC.

Immunoprecipitation using p18INK4C was performed as described in our previous report.1, 2 Immunoprecipitates were then applied to 12.5% SDS-PAGE, and p18INK4C-bound Cdk4 and Cdk6 were detected by Western blot analysis using antibody for Cdk4 and Cdk 6, respectively.

Densitometric Analysis.

Density of the phosphorylated band of Rb fusion protein obtained by autoradiography was quantitated by densitometric scanning.

Statistical Analysis.

We performed statistical analysis of the relationship between p18INK4C expression and clinical–pathological parameters by means of the Fisher t test, as appropriate. The survival curve was plotted using the Kaplan-Meier method, and differences were analyzed statistically by log-rank test. The relationship between p18INK4C expression and the activity of Cdk4 or Cdk6 was also determined by Scheffe test. In addition, to identify independent predictors in patients with HCC, multivariate analysis using the Cox proportional hazards model was performed.

Results

p18INK4C Expression in NL, CH, and Liver Cirrhosis.

Representative immunostaining of p18INK4C in NL (Fig. 1A), CH (Figs. 1B and C), and liver cirrhosis (Fig. 1D) are shown. The stages of fibrosis in Figs. 1B, C, and D were F2, F3, and F4, respectively. The p18INK4C in NL and CH (Figs. 1B and C) was localized in hepatocellular nuclei and/or nonparenchymal cells such as infiltrating lymphocytes (Fig. 1B, arrowhead), fibroblasts (Fig. 1C, arrowhead), and endothelial cells. The expression of p18INK4C was detected in all patients (n = 30) with CH or cirrhosis, regardless of fibrosis staging. As shown in Fig. 1D, the expression of p18INK4C in liver cirrhosis was detected not only in the hepatocellular cytoplasm, but also in the hepatocellular nucleus in all 9 cases. The expression of p18INK4C was detected in the hepatocellular nucleus in all NLs examined in this study.

Figure 1.

Immunohistochemistry of p18INK4C in liver tissues of (A) normal liver, (B and C) chronic hepatitis, and (D) liver cirrhosis. Fibrosis stages in panels B, C, and D were grades F2, F3, and F4, respectively, according to Desmet's classification. Staining for p18INK4C in normal liver and chronic hepatitis was seen (A-C) in the hepatocellular nucleus (arrows)), (B) in lymphocytes (arrowhead), and (C) in fibroblasts (arrowhead). Expression of p18INK4C in liver cirrhosis was positive not only for p18INK4C staining in the hepatocellular cytoplasm but also for p18INK4C staining in (D) the hepatocellular nucleus (arrows). Scale bars, 50 μm. (Original magnification [A-D], ×200.)

p18INK4C in Malignant Liver Tissues.

As shown in Table 1, p18INK4C-negative HCCs were observed in 24 of 51 HCCs (47%; LI, 2.3 ± 2.2%) examined, and p18INK4C-positive HCCs were detected in the remaining 27 cases (53%; LI, 14.0 ± 5.1%). In well-differentiated HCCs, 10 of 15 tumor samples were positively stained in the nucleus of cancer cells for p18INK4C (67%; LI, 15.2 ± 6.7%; Figs. 2A-C), and the remaining cases were negative for staining (33%; LI, 3.8 ± 1.6%). In moderately differentiated HCCs, p18INK4C-positive HCCs were detected in 17 of 30 cases (57%; LI, 13.4 ± 3.8%) and not in the remaining cases (43%; LI, 2.7 ± 2.2%; Figs. 2D-F). In poorly differentiated HCCs, p18INK4C-positive HCCs were not detected in any of the cases examined in this study (Figs. 2G-I: LI, 0.0 ± 0.0 %).

Figure 2.

Immunohistochemistry of p18INK4C in (B and C) well-, (E and F) moderately, and (H and I) poorly differentiated HCCs. A, D and G show hematoxylin-eosin staining of the section adjacent to B, E, and H, respectively. C, F, and I represent higher magnifications of the sections of B, E, and H, respectively. T and N indicate HCC and nontumorous tissues with cirrhosis, respectively. The expression of p18INK4C in most well-differentiated HCCs was localized in the nuclei of cancer cells (arrows). Conversely, its expression in most moderately and poorly differentiated HCCs was not detected. The scar indicates the boundary between N and T tissues. Scale bars, 50 μm.(Original magnification: A, B, E, and H, ×100; D and G, ×50; and C, F, and I, ×200.)

Correlations between p18INK4C expression and clinical–pathological factors as determined by univariate analysis are summarized in Table 1. The ratio of p18INK4C-positive HCCs in poorly differentiated tissues (LI, 0.0 ± 0.0%) was significantly lower than that in well-differentiated and moderately differentiated tissues (LI, 14.0 ± 5.1%; Table 1, P = .024). In addition, the ratio of p18INK4C-negative HCCs in tumor stages III and IV (63.3%) was significantly higher than that in tumor stages I and II (23.8 %; P = .025). However, no significant relationship was seen between p18INK4C expression and gender, age, or viral markers.

Prognostic Significance of p18INK4C.

Survival analysis of HCC was performed by the Kaplan-Meier method. The tumors were divided into p18INK4C-positive and p18INK4C-negative groups. Patients with p18INK4C-negative HCC had a significantly worse prognosis than those with p18INK4C-positive HCC (P = .0007; Fig. 3). Multivariate analysis using the Cox proportional hazards model was performed. Among age, sex, histological grade, tumor stage, and hepatitis viral infection, only tumor stage and p18INK4C expression were shown to be independent prognostic factors for overall survival of patients with HCC (Table 2).

Figure 3.

Kaplan-Meier curves of overall survival according to p18INK4C expression (positive staining vs. negative staining) in 51 patients with HCC. Time to death was significantly shorter in patients with p18INK4C-negative HCCs than in those with p18INK4C-positive HCCs (P < .0007).

Table 2. Multivariate Cox Model Analysis of Overall Survival of Patients With HCC
 Hazards RatioOverall SurvivalP Value
95% CI
  1. Abbreviations: PD, poorly differentiated; WD, well-differentiated; MD, moderately differentiated.

Age (<65/≥65)0.8700.407–1.860.719
Sex (male/female)0.8210.354–1.902.645
Histology (PD/WD, MD)1.3800.443–4.306.579
Tumor stage (III–IV/I–II)2.5951.077–6.257.034
Infection (HCV/HBV)0.7720.300–1.982.590
p18INK4C (negative/positive)2.7251.155–6.427.022

Activities and Amounts of Cdk4 and Cdk6 in HCC.

p18INK4C has been shown to inhibit cyclin D1- related kinase activity specifically by binding Cdk4 or Cdk6. Therefore, to examine the role of p18INK4C in the control of cell proliferation via inhibition of Cdk4 or Cdk6 activities in the G1 to S phase, we studied the activity of Cdk4 and Cdk6 in 38 HCC cases, including 12 well-differentiated, 20 moderately differentiated, and 6 poorly differentiated tumors. Clinical–pathological data for the 38 patients with HCC are shown in Table 3. p18INK4C-negative HCCs were 21 (55 %) out of 38, with the remaining 17 (45 %) p18INK4C-positive. Glutathione S-transferase (GST)-Rb fusion protein was used as a substrate to measure the Cdk4 and Cdk6 activity. As shown in Fig. 4A, staining of the GST-Rb fusion protein after separation by SDS-PAGE showed a single band with a molecular size of 46 kd. The activities of Cdk4 and Cdk6 were measured by an in vitro kinase assay using the GST-Rb fusion protein as a substrate (Figs. 4B and C). A single band of phosphorylated GST-Rb fusion protein resulting from the level of Cdk4 and Cdk6 activity was detected in all HCCs studied. Additionally, phosphorylated GST-Rb fusion protein was not observed in the immunoprecipitate product when nonimmune rabbit IgG was used as a control (data not shown). Extremely enhanced Cdk4 activity was detected in a subset of p18INK4C-negative HCCs (Fig 4B, patients 6 and 7). In moderately differentiated HCCs, the number of p18INK4C-negative HCCs and p18INK4C-positive HCCs was 10 and 10, respectively. In moderately differentiated HCCs, Cdk4 activity of p18INK4C-negative and p18INK4C-positive HCCs were 9.8 ± 6.4 and 4.1 ± 1.4 times higher, respectively, than those of control NLs (Fig. 4D). Cdk4 activity in p18INK4C-negative moderately differentiated HCCs was significantly higher than that in p18INK4C-positive HCCs (*P < .02). Conversely, Cdk6 activity of moderately differentiated HCCs did not change regardless of expression of p18INK4C (Fig. 4 D). In well-differentiated HCCs, the number of p18INK4C-negative HCCs and positive HCCs was 7 and 5, respectively. Although Cdk 6 activity remained at a similar level regardless of p18INK4C expression in well-differentiated HCCs, Cdk4 activity in p18INK4C-negative HCCs was significantly higher than that in p18INK4C-positive HCCs (*P < .05; Fig. 4E). The kinase activity levels of Cdk4 and Cdk6 in poorly differentiated HCCs were 16.6 ± 3.8 and 2.8 ± 1.3, respectively (data not shown). We then analyzed whether one of the pRb phosphorylation sites, Ser 780, was phosphorylated. The phosphorylated pRb band was detected in p18INK4C-negative HCCs (Fig. 5A). A band corresponding to the pRb phosphorylated at Ser 780 was detected in 18 (86%) of 21 p18INK4C-negative HCCs, and in 2 (12%) of 17 p18INK4C-positive HCCs. The expression levels of Cdk4 and Cdk6 were measured by Western blot. In HCC, Cdk4 (Fig. 5B) and Cdk6 (Fig. 5C) immunoreactive bands were detected at molecular weights of 34 kd and 38 kd, respectively. As an internal control, the amount of α-tubulin was almost the same in each lane (Fig. 5D). Protein levels of Cdk4 and Cdk6 were almost the same in p18INK4C-negative and p18INK4C-positive HCCs in not only moderately differentiated HCCs (Fig. 5E) but also in well-differentiated HCCs (Fig. 5F).

Table 3. Characteristics of HCC Patients Examined for Cdk4 and Cdk6 Activity
  • Abbreviations: WD, well-differentiated; MD, moderately differentiated; PD, poorly differentiated; TNM, tumor regional lymph nodes, distant metastasis.

  • *

    The fibrosis stage of the tissue surrounding HCC was assessed according to Desmet's classification.

  • Histological grading of HCC was determined using criteria of the International Working Party.17

  • TNM stage was determined using the classification proposed by the International Union Against Cancer and the American Joint Committee on Cancer.16

Sex 
 Male25
 Female13
Age (y) 
 Mean ± SD63.8 ± 6.5
 Range44–75
Viral infection 
 HCV-positive33
 HBsAg-positive5
*Histological background 
 F36
 F432
Histological differentiation 
 WD12
 MD20
 PD6
p18INK4C expression 
 Positive17
 Negative21
TNM stage 
 I13
 II10
 III8
 IV7
Figure 4.

Activity of Cdk4 and Cdk6 in HCC. (A) SDS-PAGE profile of a glutathione-S transferase (GST)-Rb fusion protein. Staining showed a single band of 46 kd (arrow). (B) Cdk4 and (C) Cdk6 kinase assays in NL and HCC. Lysates containing 500 μg of total cellular protein were prepared as described in Materials and Methods. The protein was precipitated with excess Cdk4 or Cdk6 antibodies, incubated for 10 minutes at 30°C with [γ-32P] ATP and GST-Rb fusion protein, then resolved on 12.5% SDS-polyacrylamide gel. (B) The arrow indicates the band corresponding to the phosphorylation of GST-Rb fusion protein. (B and C) The immunostaining pattern of p18INK4C in NL and HCC is indicated as positive (+) or negative (−). W, well-differentiated HCC; M, moderately differentiated HCC. Note that extremely enhanced Cdk4 activity was seen in a subset of p18INK4C-negative HCCs, but Cdk6 activity level was similar regardless of p18INK4C expression. The relative levels of Cdk4 and Cdk6 activity in (D) moderately differentiated and (E) well-differentiated HCCs in p18INK4C-positive and -negative HCCs. Cdk4 activity of p18INK4C-negative HCCs was significantly higher than that of p18INK4C-positive HCCs when measured by the phosphorylation of GST-Rb fusion protein, whereas Cdk6 activity was not significantly different between p18INK4C-positive and p18INK4C-negative HCCs, regardless of differentiation. *P < .05 compared with HCC p18INK4C-positive HCCs.

Figure 5.

Amounts of Cdk4 and Cdk6 in HCC. (A) Representative Western blot of retinoblastoma protein (pRb) using an antibody against phosphorylated pRb (Ser 780) in normal liver (NL) and HCC. The phosphorylation of pRb was detected in HCCs negative for p18INK4C expression, but not in those positive for p18INK4C expression. Western blot of (B) Cdk4, (C) Cdk6, and (D) α-tubulin in NL and HCC. The relative amount of Cdk4 and Cdk6 in (E) moderately differentiated HCC and (F) well-differentiated HCC in p18INK4C-positive and - negative HCCs. Note that the amounts of Cdk4 and Cdk6 in p18INK4C-positive and p18INK4C-negative HCCs did not significantly differ. (G) Amounts of Cdk4-bound p18INK4C and Cdk6-bound p18INK4C in p18INK4C- positive HCCs. Note that the complex of Cdk6 and p18INK4C was not seen in the liver tissues of patients 3 and 4 with p18INK4C-positive HCC but was detected in the liver tissue of patient 2 (arrow). Cdk4-bound p18INK4C was detected in these same patients (2, 3, and 4; arrow). W, well-differentiated HCC; M, moderately differentiated HCC.

Detection of p18INK4C-Bound Cdk4 and Cdk6 in HCC.

Immunoprecipitates with an anti-p18INK4C antibody obtained from 17 HCCs were immunoblotted with an anti-Cdk4 or anti-Cdk6 antibody. Clinical–pathological data for these 17 patients with HCC are shown in Table 4. Patients 2, 3, and 4 in Fig. 5G correspond to F.O., N.S., and Y.K., respectively. As shown in Fig. 5G and Table 4, p18INK4C-bound Cdk4 was detected in 16 patients (94%), including HCCs from patients 2 (F.O.), 3 (N.S.), and 4 (Y. K.), and was not detected in one case (R.K.). p18INK4C-bound Cdk6 was detected in 2 cases (12%), including HCCs from patient 2 (F.O.), but not in any other cases (Fig. 5G, Table 4). p18INK4C-bound Cdk4 was also detected in 2 cases (F.O. and F.E.) with interaction between Cdk6 and p18INK4C (Table 4).

Table 4. Characteristics of HCC Patients Examined for Detection of p18INK4C-Bound Cdk4 and Cdk6
PatientSexAgeVirus*Histological BackgroundHistological GradeTNM Stagep18INK4C-Bound Cdk4p18INK4C-Bound Cdk6
  • Abbreviations: B, hepatitis B; C, hepatitis C; WD, well-differentiated; MD, moderately differentiated; D, detection; ND, no detection.

  • *

    The fibrosis stage of the tissue surrounding HCC was assessed according to Desmet's classification.

  • Histological grading of HCC was determined using the criteria of the International Working Party.17

  • TNM stage was determined using the classification proposed by the International Union Against Cancer and the American Joint Committee on Cancer.

F.O.M70CF4WDIDD
S.K.M54CF4WDIDND
M.T.M62CF4WDIIDND
T.K.F66CF4WDIDND
Y.S.M56BF3WDIDND
K.H.F65CF4WDIIDND
T.K.M58CF3WDIDND
N.S.M74CF4MDIIDND
Y.K.M61CF4MDIDND
R.K.M52BF3MDINDND
T.K.F71CF4MDIIDND
Y.K.M60CF4MDIIIDND
F.E.M66CF4MDIDD
S.N.F64CF4MDIIDND
T.M.M58BF3MDIIDND
Y.M.M74CF4MDIIDND
K.N.M67CF4MDIIIDND

Discussion

CdkIs regulate the progression of the cell cycle by modulating the activity of Cdks.9 Inactivation of CdkIs has been associated with neoplastic transformation in a large number of human epithelial tissues.5, 6 Moreover, recent studies have extensively demonstrated that the inactivation of p16INK4A in the members of the INK4 family leads to the development and aggression of a number of human malignancies including HCC.7–11 These previous studies suggested that the other INK4 family might also play a role in the progression and prognosis of human cancers. Recently, inactivation of p18INK4C has been reported in various human cancers.22–25 To date, very little data is available on the relationship between p18INK4C and HCC,12 though previous reports have investigated gene methylation of p18INK4C in HCC, involvement of p18INK4C in troglitazone-induced cell cycle arrest of hepatoma cell lines,13 and the rate of carcinogen-induced liver tumor in p18INK4C mutant mice.14 The expression of p18INK4C protein in HCC is not yet known. In the present study, therefore, we evaluated the expression of p18INK4C in various liver diseases including HCC. To our knowledge, this study is the first to assess the involvement of p18INK4C protein in hepatocarcinogenesis.

The major finding in this study was that the loss of p18INK4C does not occur in most NL, CH, and cirrhosis, but does occur in a subset of HCCs, especially in poorly differentiated HCCs, suggesting that loss of p18INK4C is involved in hepatocarcinogenesis. The levels of Cdk4 activity in p18INK4C-negative HCCs were significantly higher than those in p18INK4C- positive HCCs, underscoring the functional importance of p18INK4C as an inhibitor of Cdk4 in HCC. It has been shown that the loss of p18INK4C expression in HCC is a poor prognostic marker.

Evidence suggests that p18INK4C functions as a tumor suppressor. Some studies using gene knockout mice indicated that the loss of p18INK4C displays a variety of aberrant phenotypes including lymphoproliferative disorders, organomegaly, and pituitary gland hyperplasia. Double knockout mice for p18INK4C and other members of the CdkI family display more varied and pronounced phenotypes,26–28 indicating that p18INK4C is important for the proliferative control of various cell lineages. On the relationship between hepatocarcinogenesis and p 18INK4C, Bai et al.14 reported that p18INK4C-null and -heterozygous mice with chemical carcinogen results in tumor development at an accelerated rate, suggesting that loss and decrease of p18INK4C leads to malignant transformation. In addition, the p18INK4C gene has been mapped to chromosome 1q32, where chromosomal translocation or loss has been found in various human cancers including HCC.29–33 Such events might explain the loss in p18INK4C protein expression found in a subset of HCC in this study. A marked decrease in p18INK4C protein level has also been observed in testicular cancer and oligodendroglia.22, 23 These previous reports are consistent with the results of reduced p18INK4C expression in HCC found in this study. Roncalli et al.12 reported that p18INK4C was not methylated in any of the HCCs examined. Therefore, there is little possibility that the cause of loss of p18INK4C expression is related to the promoter methylation of p18INK4C. However, because there have been no other papers examining methylation of the p18INK4C gene in HCC, more studies are needed to draw any conclusions. In addition, loss of p18INK4C protein at the level of translation, or proteosome-mediated degradation, might be responsible for the down-regulation of p18INK4C in HCC.

All tissue samples used in this study were either HCV- or hepatitis B virus (HBV)-positive. In this study, the expression of p18INK4C in HCV- or HBV-induced CH was detected in all cases, whereas expression was not detected in a subset of HCV- or HBV-induced HCCs. These data suggest that the changes in p18INK4C expression in the process of HCC from CH are not affected by hepatitis viral infection but that its changes are affected by the malignant process.

Identification of the grade of tumor malignancy would facilitate treatment selection for patients and provide important information for predicting their prognosis. In HCC, clinical–pathological prognostic factors, such as tumor size, the number of tumor nodules, capsule formation, capsule invasion, and vascular invasion, have been studied.34 Cell cycle-related molecules, such as proliferating nuclear antigen, p53, p21CIP1, p27KIP2, and p73, were shown to be prognostic biomarkers in various types of human cancer including HCC.35–39 To date, the relationship between p18INK4C and prognosis in cancers has been reported in only one study on oligodendroglioma.23 However, there have been no reports on the relationship between p18INK4C expression and prognosis in human HCC. In the present study, survival analysis by the Kaplan-Meier method revealed that p18INK4C expression was associated with the overall survival of patients with HCC. Of particular importance is the finding that the loss of p18INK4C expression was significantly associated with short survival of patients with HCC. According to the multivariate analysis, p18INK4C and tumor stage were independent prognostic factors for overall survival. These data suggest that loss of p18INK4C in HCC might serve as an indicator of poor prognosis.

In the univariate analysis, p18INK4C expression correlated with the differentiation status and the tumor stage of HCCs, whereas only tumor stage and p18INK4C expression level, and not the differentiation status, were independent prognostic markers in the multivariate analysis. In recent reports, Hu et al.40, 41 showed that after multivariate analysis, the differentiation of HCC is not an independent prognostic marker for overall survival, although it was a prognostic marker for overall survival in the univariate analysis; this is consistent with the results obtained in the present study. Collectively, the differentiation of HCC is a prognostic marker dependent on other factors, such as TNM stage and p18INK4C expression.

The p18INK4C protein has been shown to interact with, and subsequently inactivate Cdk4. Half of the HCCs in this study lacked detectable levels of p18INK4C protein when assayed by immunohistochemistry. Regarding Cdk4, we found previously that increase in its activity was particularly important in the development of HCC in humans and in the Long-Evans Cinnamon rat, an animal HCC model.1, 2 In addition, in our previous report2 we found that the kinase activity of Cdk4 was markedly increased in poorly differentiated HCC compared to that in well- and moderately differentiated HCC, suggesting that Cdk4 activation may be closely related to the histopathological grade of HCC. Therefore, to determine whether reduced p18INK4C in HCC is related to increased cell proliferative activity, we studied the activities of the target proteins of p18INK4C, Cdk4, and Cdk6 in HCCs with the same degree of differentiation (i.e., in order to not reflect the differentiation in HCC) in p18INK4C-positive and p18INK4C-negative HCCs. Although Cdk6 activity in p18INK4C-positive and p18INK4C-negative HCCs was not significantly different between well- and moderately differentiated HCCs, Cdk4 activity was significantly higher in well- and moderately differentiated p18INK4C-negative HCCs than in p18INK4C-positive HCCs. On the other hand, p18INK4C-positive HCCs were not detected in any cases of poorly differentiated HCCs used in this study. Thus, we could not study the relationship between p18INK4C expression and the activities of Cdk4 and Cdk6 in poorly differentiated HCC. The kinase activity levels of Cdk4 and Cdk6 in poorly differentiated HCCs were 16.5 ± 3.8 and 2.8 ± 1.3, respectively (data not shown). Based on these data, Cdk4 activity level was also markedly increased in poorly differentiated HCCs of p18INK4C-negative HCCs. These data suggest that the up-regulation of Cdk4 activity in p18INK4C-negative HCCs might not be related to the increase of its protein but rather to the reduction of p18INK4C relative to that in p18INK4C-positive HCCs.

Why does the reduced p18INK4C lead to the up-regulation of Cdk4 activity but not of Cdk6? We hypothesized that p18INK4C might bind to Cdk4 rather than to Cdk6 in HCC tissues. To study the formation of these complexes (p18INK4C/Cdk4 and p18INK4C/Cdk6), we immunoprecipitated p18INK4C from HCC lysates with p18INK4C expression and performed Western blot analysis of Cdk4 and Cdk6 (Fig. 5G). Based on these data, p18INK4C was shown to substantially bind to Cdk4 but not to Cdk6 in HCC. Therefore, it was suggested that, in HCC, reduced p18INK4C contributed only to the up-regulation of Cdk4. The p18INK4C protein has been shown to interact with, and subsequently inactivate, Cdk4.5, 6 Conversely, there have been no reports that activation of Cdk4 does not directly decrease p18INK4C expression. Therefore, it is difficult to guess whether increased Cdk4 activity in a subset of HCC does decrease p18INK4C expression.

To date, p18INK4C has been reported to play an important role in the regulation of cell differentiation.22, 23 To investigate the possible involvement of this protein in the differentiation of HCCs, we studied the relationship between the level of p18INK4C expression and the histological grade of HCC. Expression of p18INK4C was reduced in poorly differentiated HCCs compared with levels in well- and moderately differentiated HCCs, supporting the finding in previous reports that p18INK4C accumulates at high levels in terminally differentiated cells.22, 42, 43 However, our studies have not yet clarified whether the loss of p18INK4C in HCC promotes or inhibits differentiation. These data suggest that decreased expression of p18INK4C may play a role in the regulation of tumor differentiation as well as in the malignant transformation leading to HCC. In the present study, p18INK4C was expressed in the nuclei of hepatocytes in NL and CH, and p18INK4C was also detected in the cytoplasm in liver cirrhosis. p18INK4C has been shown to localize both in the nucleus and in the cytoplasm of cells.23, 44 However, the significance of the cytoplasmic expression of p18INK4C in hepatocytes of livers with cirrhosis remains unclear.

In conclusion, loss of p18INK4C expression may be important in the process of malignant transformation and appears to be closely related to histological differentiation of HCC. In addition, loss of p18INK4C was shown to be associated with up-regulation of Cdk4 activity. Furthermore, expression of p18INK4C may be an effective predictor of clinical behavior in HCC, and, therefore, a new prognostic marker for HCC.

Ancillary