These authors contributed equally to this work.
Telomere shortening and inactivation of cell cycle checkpoints characterize human hepatocarcinogenesis†
Article first published online: 28 MAR 2007
Copyright © 2007 American Association for the Study of Liver Diseases
Volume 45, Issue 4, pages 968–976, April 2007
How to Cite
Plentz, R. R., Park, Y. N., Lechel, A., Kim, H., Nellessen, F., Langkopf, B. H. E., Wilkens, L., Destro, A., Fiamengo, B., Manns, M. P., Roncalli, M. and Rudolph, K. L. (2007), Telomere shortening and inactivation of cell cycle checkpoints characterize human hepatocarcinogenesis. Hepatology, 45: 968–976. doi: 10.1002/hep.21552
Potential conflict of interest: Nothing to report.
- Issue published online: 28 MAR 2007
- Article first published online: 28 MAR 2007
- Manuscript Accepted: 2 NOV 2006
- Manuscript Received: 27 AUG 2006
- Deutsche Forschungsgemeinschaft (Heisenberg-professorship). Grant Numbers: Ru 745/8-1, Ru 745/4-1, KFO119
- Deutsche Krebshilfe e.V.. Grant Numbers: 10-2236-Ru 2, R13-2002-054-02003-0
- Basic Research Program of the Korea Science & Engineering Foundation
Telomere shortening and inactivation of cell cycle checkpoints characterize carcinogenesis. Whether these molecular features coincide at specific stages of human hepatocarcinogenesis is unknown. The preneoplasia–carcinoma sequence of human HCC is not well defined. Small cell changes (SCC) and large cell changes (LCC) are potential precursor lesions. We analyzed hepatocellular telomere length, the prevalence of DNA damage, and the expression of p21 and p16 in biopsy specimens of patients with chronic liver disease (n = 27) that showed different precursor lesions and/or HCC: liver cirrhosis (n = 25), LCC (n = 26), SCC (n = 13), and HCC (n = 13). The study shows a decrease in telomere length in nondysplastic cirrhotic liver compared with normal liver and a further significant shortening of telomeres in LCC, SCC, and HCC. HCC had the shortest telomeres, followed by SCC and LCC. Hepatocytes showed an increased p21 labeling index (p21-LI) at the cirrhosis stage, which remained elevated in most LCC. In contrast, most SCC and HCC showed a strongly reduced p21-LI. Similarly, p16 was strongly expressed in LCC but reduced in SCC and not detectable in HCC. γH2AX-DNA-damage-foci were not detected in LCC but were present in SCC and more frequently in HCC. These data indicate that LCC and SCC represent clonal expansions of hepatocytes with shortened telomeres. Conclusion: The inactivation of cell cycle checkpoints coincides with further telomere shortening and an accumulation of DNA damage in SCC and HCC, suggesting that SCC represent more advanced precursor lesions compared with LCC. (HEPATOLOGY 2007;45:968–976.)
HCC represents the fifth most common neoplasm in humans.1 Treatment options for HCC patients are limited, and the disease is often diagnosed at an advanced stage.2 The molecular pathogenesis of hepatocarcinogenesis is poorly understood,3 thus impairing the development of molecular markers, efficient screening procedures, and molecular therapies.
HCC rarely evolve in noncirrhotic liver but the risk sharply increases at the cirrhosis stage of chronic liver diseases, with the yearly incidence of HCC ranging from 3% to 6%.2 The preneoplasia–carcinoma sequence during hepatocarcinogenesis remains under debate. Dysplastic nodules (DNs) and dysplastic foci are regarded as early preneoplastic lesions.4 Because only a fraction of these lesions progress into HCC,5, 6 characterizing the DN–HCC transition on a molecular level is virtually impossible. Hepatocyte dysplasia is an important feature of DNs and dysplastic foci, associated with a high risk of HCC development.6–12 These lesions can be divided into 2 groups: (1) large cell changes/dysplasia (LCC) and (2) small cell changes/dysplasia (SCC).13, 14 Whether both of these lesions represent true precursor lesions and which of these precursors is more closely related to HCC are intensely debated.7–12
On the molecular level, HCC are characterized by massive chromosomal instability (CIN) in more than 95% of cases,15–17 and telomere shortening18, 19 and loss of p53-checkpoint function in more than 70% of the cases.3, 20 Experimental data from telomerase-deficient mice show that telomere shortening represents one mechanism that leads to induction of CIN and increasing initiation of various cancer types21, 22 including HCC.23 These studies indicated that telomere shortening acts as a co-factor, which increases CIN and cancer risk. Telomere shortening occurs during human aging and is accelerated during chronic diseases.24 Telomere shortening and an accumulation of senescent hepatocytes also characterize liver cirrhosis25 and could give a plausible explanation for the decline in regenerative reserve26 and the sharply increased cancer risk at this stage.2
Research work in the past has improved our understanding of the molecular consequences induced by telomere shortening. When telomeres reach a critically short length, they lose capping function, and a DNA-damage response is induced, involving the p53–p21 pathway, that leads to a permanent cell cycle arrest at the senescence stage.27–29 Senescence is regarded as a tumor suppressor mechanism that prevents proliferation of genetically unstable precancerous cells. In line with this hypothesis, p53 deletion accelerated CIN in primary human cells, bypassing the senescence checkpoint in cell culture.28In vivo, p53 deletion led to an increase in CIN and cancer formation in telomerase-deficient mice.30
The activation of oncogenes also can induce a senescence checkpoint, which shares many molecular pathways with the senescence checkpoint induced by telomere dysfunction.31, 32 Oncogene-induced senescence in early neoplastic lesions and inactivation of this senescence response in subsequent cancer lesions have been demonstrated in a variety of cancer mouse models.33–35 Furthermore, an activation of senescence-associated DNA damage checkpoints has been observed in preneoplastic lesions during carcinogenesis in humans, and these checkpoints were inactivated during progression into cancer.36, 37 In liver, previous studies have demonstrated that p21 checkpoint activation was associated with an increased HCC risk at the cirrhosis stage.38 Whether the inactivation of DNA damage checkpoints coincides with telomere shortening during carcinogenesis in humans has not been analyzed, however. A combined analysis of these parameters in defined pathological lesions associated with hepatocarcinogenesis could help to delineate the preneoplasia–carcinoma sequence of human HCC.
We analyzed telomere length and expression of p21 and p16 during human hepatocarcinogenesis at preneoplastic stages (cirrhosis, LCC, and SCC) and in HCC. Our data provide direct evidence that telomere shortening coincides with loss of cell cycle checkpoints in SCC and HCC.
Materials and Methods
We collected paraffin blocks from 27 specimens from the Departments of Pathology at the University of Milan and Humanitas Clinical Institute of Rozzano, Milan, Italy, and from the Liver Cancer Specimen Bank supported by the National Research Bank Program of the Korean Science & Engineering Foundation in the Ministry of Science & Technology.
Quantitative Fluorescence In Situ Hybridization.
Paraffin sections (4 μm) were deparaffinised with xylol and rehydrated. Samples were boiled for 30 minutes in sodium citrate buffer (pH 6.0). Fixation and washing steps were repeated as described previously.24 In brief, samples were washed with phosphate-buffered saline (PBS) and fixed in 4% formaldehyde followed by enzymatic unmasking for 10 minutes at 37°C (enzyme mix: 125 mg pepsin/105 μl concentrated HCl/125 ml distilled water). The samples were denatured for 3 minutes at 80°C followed by hybridization for 2 hours at room temperature in the dark [hybridization mix: 1.5 μl 1 mμ Tris-Cl, pH 7.2; 10.7 μl MgCl2 (25 mM MgCl2/9 mM citric acid/8.2 mM NaH2PO4/pH 7.4); 87.5 μl deionized formamide; 6.2 μl 10% (wt/wt) blocking reagents; 2.5 μl 25 μg/ml PNA Cy3-telomere probe; and 16.6 μl distilled water]. The slides were washed twice in washing solution I (70 ml formamide/1 ml 1 M Tris-Cl, pH 7.2/1 ml 10% bovine serum albumin stock solution/28 ml distilled water), followed by 3 washes in washing solution II [15 ml 1 M Tris-Cl, pH 7.2/15 ml 1.5 M NaCl/120 μl Tween 20 (0.08% final)/120 ml distilled water]. After dehydration in 70%, 90%, and 100% ethanol, samples were air-dried for approximately 20 minutes in the dark and then mounted with DAPI mounting medium. Pictures were taken at 4000 ms for the Cy3 images and at 100 ms for the DAPI images. Quantification of the telomere fluorescence intensity (TFI) by quantitative fluorescence in situ hybridization (Q-FISH) analysis was performed using TFL-TELO V1.0.39 To facilitate day-to-day comparison, a centromere probe was photographed and analyzed for each individual session as an internal standard. Finally, the mean ratio of telomere and centromere probes was calculated.
Immunohistochemical Analysis of p21.
Paraffin sections (4 μm) were deparaffinized with xylene and rehydrated with graded alcohol. After washing in distilled water, the sections were immersed in 3% hydrogen peroxide to block endogenous peroxidases. Antigen retrieval was performed by boiling the sections in 100 mM sodium citrate (pH 6.0) for 15 minutes in a microwave oven. Monoclonal primary antibody (p21WAF1/Cip1, Clone SX118, DAKO, Glostrup, Denmark) at a 1:50 dilution was applied for 30 minutes at room temperature followed by washing in PBS. Incubation with the secondary antibody was carried out using the DAKO EnVision Rabbit/Mouse kit for 30 minutes at room temperature, developed with diaminobenzidine (DAKO). Sections were counterstained with hematoxylin.
For assessment of the immunohistochemical stain results, positive and negative hepatocytes were counted in at least five random high-power fields (400× magnification) so that at least 500 hepatocytes were counted for each lesion (HCC, SCC, LCC, and non-dysplastic cirrhosis) in every case. Dark brown staining of hepatocytes' nuclei was considered positive. The number of positive hepatocytes was expressed as a percentage of the total to give a labeling index (p21-LI). The analysis was conducted by two blinded, independent pathologists.
Immunohistochemical Analysis of p16 and γH2AX.
Paraffin sections (4 μm) were deparaffinized and rehydrated with graded alcohol. Antigen retrieval was performed by boiling sections in 100 mM sodium citrate (pH 6.0) for 15 minutes in a microwave oven. Primary antibody (p16, Clone M-156, Santa Cruz, CA; γH2AX, Clone JBW301, Upstate) was applied at a 1:200 dilution overnight at 4°C followed by washing in PBS. Incubation with the horseradish peroxidase–conjugated secondary antibody was performed at room temperature (p16: 60 minutes; γH2AX: 10 minutes), and the cells were developed with diaminobenzidine (DAKO) or 3-amino-9- ethyl-carbazole (Sigma) and counterstained with hematoxylin. The analysis was conducted by 2 blinded, independent pathologists.
Patient Characteristics and Histology.
In total, 27 samples with defined clinicopathological features were analyzed (Table 1). In most cases, HCC and/or the preneoplastic changes were associated with liver cirrhosis (25 of 27 cases). LCC and SCC were morphologically identified on the basis of previously described standards (Fig. 1): (1) LCC showed nuclear and cellular enlargement, normal or slightly increased nuclear and cytoplasmic ratios, nuclear pleomorphism with hyperchromasia, and multinucleation (Fig. 1A)12, 13; (2) SCC were characterized by crowded small hepatocytes with high nuclear and cytoplasmic ratios, and slight nuclear pleomorphism (Fig. 1B)14; (3) HCC were characterized by an irregular hyperchromatic nucleus, thick trabeculae of more than three cell plates, increased cellularity of more than 2 times that of surrounding liver, and stromal invasion (Fig. 1C).4 The HCC grading was determined according to the Edmonson and Steiner criteria.40 The status of chronic hepatitis in the background liver was evaluated semiquantitatively by grading the severity of necroinflammation: 0 (none), 1 (minimal), 2 (mild), 3 (moderate), and 4 (severe); and fibrosis: 0 (no fibrosis), 1 (portal fibrosis), 2 (periportal fibrosis), 3 (septal fibrosis), 4 (cirrhosis), according to previously described standards.41 The average age of all patients with chronic liver disease, cirrhosis, SCC, LCC, or HCC was 55 years (range, 36–74). The cases included 16 men and 11 women (Table 1).
|Case Number||Age (years)||Sex||Etiology||Pathological Diagnosis*||Background Liver Pathology†||p21-IHC||qFISH|
|1||68||M||HBV||LCC1, HCC(2)||Cirrhosis (3,4)||Done||Done|
|2||47||M||HBV||LCC1, HCC(3)||Cirrhosis (2,4)||Done||Done|
|4||52||M||HBV||LCC, SCC, HCC(2)||Cirrhosis (3,4)||Done||–|
|5||54||M||HBV||LCC, SCC||Cirrhosis (2,4)||Done||–|
|6||52||F||HBV||LCC, SCC||Cirrhosis (2,4)||Done||Done|
|7||52||F||HBV||LCC, SCC, HCC(1)||Cirrhosis (2,4)||Done||Done|
|8||60||F||HBV||LCC, SCC||Cirrhosis (2,4)||Done||Done|
|9||60||F||HBV||LCC, SCC, HCC(1)||Cirrhosis (3,4)||–||Done|
|10||53||M||HCV||LCC, HCC(3)||Cirrhosis (3,4)||Done||Done|
|11||56||F||HCV||LCC, SCC, HCC(1)||Cirrhosis (3,4)||Done||Done|
|12||61||M||HCV||LCC, SCC, HCC(3)||Cirrhosis (3,4)||Done||Done|
|14||58||M||HCV||LCC, SCC||Cirrhosis (3,4)||Done||Done|
|15||43||F||HCV||LCC||Chronic hepatitis (3,3)||Done||Done|
|16||59||M||HBV||LCC, SCC||Cirrhosis (3,4)||Done||–|
|17||59||F||HBV||LCC, HCC(2)||Cirrhosis (2,4)||Done||Done|
|18||61||F||HBV||LCC, HCC(2)||Cirrhosis (3,4)||Done||Done|
|20||48||F||HBV||LCC, SCC||Cirrhosis (3,4)||Done||Done|
|21||51||M||HBV/HDV||LCC, SCC||Cirrhosis (3,4)||Done||Done|
|24||74||F||Cryptogenic||LCC, HCC(2)||Cirrhosis (3,4)||Done||Done|
|25||64||M||Alcoholic||LCC, HCC(1)||Cirrhosis (3,4)||Done||Done|
|26||49||M||Cong. hepatic fibrosis||LCC||Cong. hepatic fibrosis (3,3)||Done||Done|
|27||72||M||HBV||LCC, HCC(2)||Cirrhosis (3,4)||–||Done|
Telomere Shortening During Multistep Hepatocarcinogenesis.
Telomere length was analyzed by quantitative fluorescence in situ hybridization (qFISH), allowing a cell type–specific analysis of the hepatocellular telomere length in different lesions: non-dysplastic hepatocytes of cirrhotic liver (n = 22) and chronic hepatitis (n = 2), LCC (n = 24), SCC (n = 10), and HCC (n = 12) (Fig. 2A, Table 1). In addition, we analyzed a control set of 10 liver biopsy specimens from patients without chronic liver disease or cirrhosis. The average age in this cohort was 56.4 years (range, 37–71). The cases included five men and five women. Using a telomere- and a centromere-specific probe for qFISH allowed optimal comparison between different samples because the centromeric probe was used as an internal standard, and telomere length was calculated as a ratio of TFI to centromere fluorescence intensity (CFI). In line with previous studies, this method showed significantly shorter telomeres in liver cirrhosis compared with noncirrhotic liver (Fig. 2B),25 as well as significantly shorter telomeres in HCC compared with surrounding cirrhosis (Fig. 2B).17, 18 Comparing telomere length in different types of lesions among all patients revealed shorter telomeres in preneoplastic lesions, LCC, and SCC, compared with nondysplastic hepatocytes located in surrounding chronic liver disease tissue (Fig. 2B). However, both lesions (SCC and LCC) had longer telomeres compared with neoplastic hepatocytes located in HCC (Fig. 2B). These data were in agreement with the hypothesis that LCC and SCC represent premalignant lesions that characterize the transition of cirrhosis toward HCC. Hepatocytes in SCC had significantly shorter telomeres compared with hepatocytes of LCC (Fig. 2B).
The mean age was similar in groups of patients with liver cirrhosis: 54.3 years; LCC: 55.3 years; SCC: 55.4 years; HCC: 60.5 years. Within each group we found no significant correlation between age and telomere length (cirrhosis: P = 0.497, LCC: P = 0.903, SCC: P = 0.103, HCC: P = 0.485). In most patients, liver disease was attributable to HBV or HCV infection (Table 1). There was no significant difference in telomere length between HBV and HCV patients (cirrhosis: P = 0.462, LCC: P = 0.381, SCC: P = 0.095, HCC: P = 0.629).
p21-Checkpoint Inactivation During Multistep Hepatocarcinogenesis.
p21 is a marker of p53-checkpoint function, which is necessary to establish a cell cycle arrest at the senescence stage.29 The labeling index (LI = percentage of labeled nuclei per lesion) of p21 was analyzed by immunohistochemistry on the same set of samples that was used for telomere length measurement (Table 2). In agreement with previous studies, the p21-LI was low in normal liver samples (mean p21-LI, 0.65%; data not shown), which were derived from living related donors of liver transplantation patients. An increased p21-LI was present in most of the samples in nondysplastic cirrhosis (areas not showing SCC or LCC) (Table 2, Fig. 3D, Fig. 4; mean p21-LI, 20.6%; P = 0.001 compared with p21-LI in normal liver).38 The intracellular localization of p21 expression was mainly nuclear (Fig. 3D,F,J). The expression of p21 at the cirrhosis stage stands in agreement with the hypothesis that hepatocyte telomere shortening and senescence impair liver regeneration during cirrhosis development.25 The p21-LI remained elevated in LCC (Fig. 3F, Fig. 4, mean value: 21.9%; P = 0.6 compared with p21-LI in non-dysplastic cirrhosis). In contrast to LCC, a significant decrease of the p21-LI was seen in SCC (Table 2, Fig. 3H, Fig. 4; mean value: 4.97%, P < 0.001 compared with p21-LI in nondysplastic cirrhosis) and HCC (Table 2, Fig. 3J, Fig. 4; mean value: 5%, P < 0.001 compared with p21-LI in non-dysplastic cirrhosis). Similar results were obtained using a different primary antibody against p21 (monoclonal primary antibody, Santa Cruz Biotechnology).
p16 is a marker of Rb-checkpoint function. Similar to the p21-staining pattern, cirrhosis and LCC showed strong p16 expression, which was gradually lost in SCC and absent in HCC (Fig. 5A, B). Together, these data indicated that telomere shortening in SCC and HCC correlated with a loss of Rb-checkpoint and p53-checkpoint function.
Telomere dysfunction leads to induction of DNA damage responses, including the formation of γH2AX-foci.27 The prevalence of γH2AX-foci was low in LCC, gradually increased in SCC, and highest in HCC (Fig. 5C, D). These data indicated that loss of cell cycle checkpoints in SCC and HCC allowed proliferation of cells with short telomeres, thereby leading to DNA damage accumulation.
This study provides experimental evidence that loss of p21 and p16 expression coincides with critical telomere shortening at a specific stage during multistep hepatocarcinogenesis in humans. An inactivation of cell cycle checkpoints has recently been identified as a hallmark feature characterizing the transition of premalignant to malignant lesions in mouse models of cancer.33–35 In addition, inactivation of DNA damage checkpoints has been observed during early stages of human carcinogenesis.36, 37 A current hypothesis indicates that telomere shortening and loss of DNA damage checkpoints cooperate during cancer initiation in humans. A combined analysis of telomere length and DNA damage checkpoint integrity on defined lesion during multistep hepatocarcinogenesis in humans has not been reported to support this hypothesis. Our data on human hepatocarcinogenesis indicate that both LCC and SCC represent clonal expansions of hepatocytes with shortened telomeres at the cirrhosis stage. Telomere shortening is more pronounced in SCC and coincides with loss of p21 and p16 expression in SCC and HCC. p21 and p16 are markers of p53- and Rb-checkpoint function. Our data indicate that these checkpoints are active in LCC but inactive in SCC and HCC. The data represent a first example connecting inactivation of cell cycle checkpoints and telomere shortening at a defined stage of hepatocarcinogenesis in humans. These results represent a significant extension of previous studies showing that telomere shortening characterizes the cirrhosis stage25 and that further telomere shortening coincides with the evolution of CIN in HCC.18, 19 High levels of p21 expression in cirrhotic liver have previously been correlated with an increased cancer risk.38 That telomere shortening and consequent activation of cell cycle checkpoints repress liver regeneration at the cirrhosis stage and that in this molecular context loss of p16/p21-expression represents a selective advantage allowing clonal expansion of hepatocytes with dysfunctional telomeres seem possible.
Studies in telomerase-deficient mice have shown that telomere shortening cooperates with loss of p53-checkpoint function to induce CIN and cancer initiation.30 In human hepatocarcinogenesis, an increase in CIN is seen in SCC but not in LCC.42 Moreover, CIN in SCC is similar to the type of CIN observed in HCC.42 Our study shows that LCC contain short telomeres; however, in contrast to SCC and HCC, they show intact cell cycle checkpoints. Intact checkpoint responses may prevent proliferation of LCC with shortened telomeres, thereby preventing the evolution of DNA damage and CIN. In contrast, loss of cell cycle checkpoints in SCC may allow proliferation of hepatocytes with short telomeres and thus may be associated with the accumulation of telomere dysfunction and DNA damage. Because abrogation of cell cycle checkpoints and telomere shortening induce CIN in cell culture,28 the current data provide a plausible explanation for the sharp increase of CIN in SCC and HCC.42
The presented results indicate that SCC represent a more advanced precancerous lesion compared with LCC during human hepatocarcinogenesis. Both lesions contain shorter telomeres compared with surrounding non-dysplastic hepatocytes of cirrhosis, but only SCC show a high prevalence of checkpoint abrogation. Previous studies in mouse models and human cancer have shown that inactivation of DNA-damage checkpoints characterize the transition of precancerous lesions toward cancer.33–35 In addition to an inactivation of DNA damage checkpoints, most human HCC show an upregulation of telomerase,43 which is necessary for tumor progression in mouse models.22, 23 In human hepatocarcinogenesis, telomerase activation occurs in high-grade dysplastic nodules—a lesion characterized by the evolution of SCC.44, 45 Together, these data suggest that telomere dysfunction, inactivation of DNA damage checkpoints, and telomerase reactivation coincide at the SCC/HCC transition.
Whether premalignant lesions with intact DNA-damage checkpoints can progress into more advanced lesions that have lost a DNA-damage checkpoint function, or whether both types of lesion develop in parallel but only one type can progress toward cancer, remain to be determined. In terms of hepatocarcinogenesis, whether LCC can progress into SCC or HCC remains to be analyzed. The observed heterogeneity in p21-LI of LCC could indicate that this might occur. Alternatively, LCC could represent a dead end of cancer development during human hepatocarcinogenesis. These data suggest that SCC and LCC carry intrinsic differences that are associated with a higher risk of checkpoint abrogation in SCC compared with LCC.
This study provides evidence that the “crisis” model of cancer initiation characterized by telomere shortening and loss of cell cycle checkpoints applies to human hepatocarcinogenesis. The study shows that telomere shortening and checkpoint abrogation coincide at a specific stage of hepatocarcinogenesis—the SCC. These data improve our understanding and will help to molecularly characterize multistep hepatocarcinogenesis in humans.
- 2Hepatobiliary neoplasm. In: DeVitaVT, HellmanS, RosenbergSA, eds. Cancer: Principles & Practice of Oncology. 4th ed., Philadelphia, PA: Lippincott Williams & Wilkins; 1993: 883–887., , .
- 4International Working Party. Terminology of nodular hepatocellular lesions. HEPATOLOGY 1995; 22: 983–993
- 24Telomeres and telomerase in aging, regeneration and cancer. Mol Cells 2003; 30; 15: 164–175., , , .
- 44Quantitative assessment of hTERT mRNA expression in dysplastic nodules of HBV-related hepatocarcinogenesis. Am J Gastroenterol 2006; 101: 831–838., , , , , .Direct Link: