|Home | About | Journals | Submit | Contact Us | Français|
Chronic Hepatitis B virus (HBV) infection is linked to development of hepatocellular carcinoma (HCC). The HBV X protein (pX) is implicated in HCC pathogenesis acting as a weak oncogene or a co-factor in hepatocarcinogenesis. pX induces DNA re-replication, DNA damage and partial polyploidy in a poorly differentiated, immortalized hepatocyte cell line. In this study, we employed sorted, pX-induced polyploid cells to investigate their growth and oncogenic transformation potential, over the course of 70 cell doublings. Immediately after live cell-sorting, nearly 40% of pX-induced polyploid cells undergo apoptosis, while the surviving cells exhibit proliferation sensitive to p53. After 40 cell generations, the pX-expressing polyploid cultures exhibit loss of p53 function and become growth factor- and anchorage-independent, indicative of oncogenic transformation. The pX-induced polyploid cultures in the course of 70 cell generations undergo progressively increasing DNA damage, propagate damaged DNA to daughter cells, and display increased expression of a cluster of proliferation genes shown to be elevated in human HCC, including HBV-HCC. One of these genes is the mitotic kinase Polo-like kinase 1 (Plk1). Oncogenic transformation is suppressed in the absence of pX expression, and significantly, by inhibition of Plk1. These results identify Plk1 as crucial in pX-mediated oncogenic transformation.
Partial polyploidy induced by pX is not immediately associated with oncogenic transformation. Continued DNA damage for 40 cell generations is reproducibly associated with loss of p53 function, enhanced expression of Plk1 and oncogenic transformation. Since Plk1 expression is also elevated in HBV-HCC tumors, this in vitro cellular model simulates liver cancer progression and pathogenesis in chronic HBV patients. Inhibition of Plk1 activity suppresses pX-mediated oncogenic transformation, identifying Plk1 as a promising therapeutic target for HBV-mediated HCC.
Chronic hepatitis B virus (HBV) infection is associated with increased risk of hepatocellular carcinoma (HCC) development (1). HBV-HCC exhibit frequent and specific chromosomal aberrations (2), by mechanism(s) not yet understood. Studies have shown that pre-neoplastic human specimens exhibit markers of DNA damage (3–5), that defective DNA repair causes chromosomal instability, accelerating liver carcinogenesis (6), and genomic instability is likely the primary cause of carcinogenesis, based on the slow kinetics from carcinogen exposure to cancer development (7).
The mechanism of HCC pathogenesis by chronic HBV infection involves effects of chronic inflammation and regeneration of the liver (8), and effects of two HBV proteins, the X (9) and S (10) proteins. HBV DNA integrates into the host genome during early steps of tumor expansion (11), and most tumors continue to express the X protein (pX). pX is multifunctional, essential for the viral life cycle (12), and implicated in HCC pathogenesis (9, 11), acting as a weak oncogene (13) or a co-factor in hepatocarcinogenesis (14). pX increases the activity of the cellular mitogenic pathways (15), and enhances transcription of select viral and cellular genes (9). Activation of mitogenic pathways by pX deregulates hepatocyte gene expression. Depending on growth conditions, this deregulation results in either accelerated cell cycle entry (16) or apoptosis (17). Specifically, pX induces p53 apoptosis, only when pX-expressing cells are challenged with additional sub-apoptotic signals such as growth factor deprivation (17, 18). By contrast, in optimal growth factor conditions, pX-expressing cells do not undergo apoptosis, but rather exhibit accelerated and unscheduled S phase entry, transient S phase pause, activation of the G2/M checkpoint, and eventual progression through mitosis (16). Moreover, in optimal growth conditions, pX induces DNA re-replication and DNA damage by deregulating expression of replication initiation factors Cdt1 and Cdc6, and geminin, the negative regulator of re-replication (19). Despite pX-induced DNA re-replication and the ensuing DNA damage, these hepatocytes proceed through mitosis, propagating DNA damage to daughter cells and generating pX-expressing cells with aberrant DNA content (> 4N DNA or partial polyploidy). Significantly, pX-induced partially polyploid cells have been isolated by live cell sorting (19).
Herein, we investigate the growth properties and oncogenic transformation potential of pX-induced partially polyploid cells. Immediately after cell-sorting, 40% of pX-induced polyploid cells undergo apoptosis. The surviving cells, in the course of 70 cell generations, exhibit increasing DNA damage propagated to daughter cells, and progressively increasing expression of a cluster of proliferation genes that are elevated in human HCC tumors, including HBV-HCC (20). This cluster includes replication factors MCM4-6, PCNA, Thymidilate Synthase (TYMS) and Polo-like kinase 1 (Plk1). Importantly, Plk1 is overexpressed in many human tumors (21, 22). Constitutive Plk1 expression causes focus formation in NIH3T3 cells and tumors in nude mice (23), whereas Plk1 silencing inhibits xenograft tumor growth (24). Herein, we demonstrate that after 40 cell generations the polyploid cultures have characteristics of oncogenic transformation, dependent on pX expression. Significantly, inhibiting Plk1 suppresses pX-induced oncogenic transformation, suggesting Plk1 could provide a therapeutic target for HBV-HCC. These observations, together with accumulating evidence that genomic instability is the primary cause of carcinogenesis (3–7), are congruent with the weak oncogenic potential of pX (13, 14) and the slow progression of HBV-mediated HCC, following chronic HBV infection (1).
4pX-1 (25) and 4pX-1-p53kd (19) cell lines were grown in the presence of 5µg/ml tetracycline. pX expression was initiated by removal of tetracycline. Live cell-sorting of polyploid cells was performed as described (19). Immediately after sorting, 30,000 cells containing >4N DNA, referred to as Passage0 (P0), were plated in media containing 10% fetal calf serum (25) without tetracycline addition. On day6, P0 cultures were re-plated (30,000 cells/dish), generating Passage 1 (P1). Similarly, Passage2 through Passage5 (P2-P5) cultures were generated consecutively by plating 30,000 cells from the previous day6 culture.
Dual parameter flow cytometry for caspase8 or 9 was performed using single cell suspensions of P0 cells, employing Cytomics FC-500 (Beckman-Coulter). Fluorogenic substrates for caspase8 (FAM-LETD-FMK) or caspase9 (FAM-LEHD-FMK) were added to trypsinized cells (4,000 cells), per manufacturer’s instructions (Biocarta). Incubation was for 60 minutes at 4oC, followed by staining with propidium iodide (PI).
Real-time PCR quantifications were performed as described (16). PCR primers for Plk1: Fwd 5' AACCAAAGTGGAATATGAAGGG 3'; Rev 5' GAGGTCTCAAAGGGAGGCT 3'. PCR primers for pX have been described in (18), MCM4-6, PCNA and TYMS (19), and the 16-gene signature set (26). For unsupervised clustering analysis using the 16-gene expression profile, mouse fetal liver microarray data (27) and real-time PCR data from 4pX-1, BMEL cells and post-natal mouse livers were normalized using dChip v1.6 software (26).
p53 (CM5, 1:1000) from Vector Laboratories; H2AX and γH2AX (1:1000) from Cell Signaling; Plk1 (1:1000) Abcam; BubR1 (1:500) Affinity Bioreagents; and MRE11 (1:500) Abcam.
105 cells of P1-P5 cultures were plated per 6cm dish on a layer of 1% agarose in 2ml growth media (25). Cultures were fed with fresh media every 4 days for 14 days, with or without addition of 5µg/ml tetracycline, or 20µM BTO-1 (Sigma).
To study the role of pX in hepatocyte transformation, we employ the tetracycline-regulated pX-expressing 4pX-1 cell line (25). The 4pX-1 cell line is immortalized and resembles less-differentiated hepatocytes (19, 25). Employing the expression profile of 16 marker genes that recognize liver developmental stage (26), we determined that the 4pX-1 cell line co-clusters with fetal hepatocytes between E12.5 and E13.5, and bipotential mouse embryonic liver (BMEL) cells (28) (Fig.1A), thus modeling poorly differentiated hepatocytes. Since hepatic progenitors participate in liver regeneration after chronic liver injury (29), and are likely precursors of HBV-HCC (30), the 4pX-1 cell line is physiologically relevant to study pX-mediated oncogenic transformation. Moreover, the 4pX-1 cell line expresses low level of pX (25), resembling pX levels in natural infection (12). 4pX-1 cells grown in optimal conditions with pX expression generate a partially polyploid cell population (6–8% of total cells) containing >4N DNA (19). pX-induced polyploid cells have been isolated by live cell-sorting (ref. 19 and Supplementary fig. 1), re-plated and propagated generating the consecutive P1-P5 cultures (Experimental Procedures and Fig. 1B). P1-P5 cultures continue to express pX (Fig. 1C), thus enabling study of the role of pX in oncogenic transformation.
Since p53 maintains genomic stability (31), and pX sensitizes cells to p53 apoptosis (18), we quantified the apoptotic cells in the P0 population immediately after sorting, until 24h after re-plating. Dual parameter FACS analysis, employing fluorogenic substrates for activated caspase8 or 9 and propidium iodide (PI) staining (Supplementary fig. 2), demonstrates 45% of P0 cells undergo apoptosis immediately after cell-sorting (Fig. 2A). Apoptosis was decreased to less than 10%, 24h after re-plating (Fig. 2A). Fluorescence microscopy using fluorigenic caspase9 substrate, confirmed these results (Fig. 2B). Interestingly, apoptosis increased in the absence of pX, suggesting pX rescues a small, but significant number (p<0.005 for indicated pairs) of P0 cells from apoptosis.
Next, we investigated whether p53 modulates the proliferation of pX-expressing polyploid cells. P0-P5 cultures were treated with or without the p53-specific inhibitor PFT-α (32) and counted on day6. Cell doublings on day6 were estimated based on the consideration that 30,000 cells were seeded for each consecutive culture (Fig. 3 and Supplementary fig. 3). We observed a small but progressive increase in cell number of P0-P2 cultures (Fig. 3A). Inhibition of p53 by PFT-α further increased the cell number of P0-P2 cultures (Fig 3A). Interestingly, the cell number of P3-P5 cultures increased by 50%, even without PFT-α addition, indicating loss of p53 function (Fig. 3A and Supplementary fig. 4).
Since growth of P3-P5 cultures was insensitive to PFT-α (Fig. 3A), we quantified by real-time PCR the expression of p53 mRNA in P0-P5 cultures (Fig. 3B). P0 cells immediately after cell-sorting have the highest level of p53 mRNA, in agreement with the increased apoptosis for that interval (Fig. 2), and the enhanced p53 protein level in P0 lysates (Fig. 3C). The mRNA level of p53 in P1-P3 cultures is similar to the p53 mRNA level in 4pX-1 cells, 10h following pX expression. Interestingly, the mRNA of p53 decreased in P4 and P5 cultures (p<0.05), suggesting either defects in p53 transcription after 40 cell doublings, or post-transcriptional mechanisms such as microRNA expression targeting p53 mRNA.
To confirm the growth profiles of P1-P5 cultures (Fig. 3A), we quantified on day6 the percent of P1-P5 cells that synthesize DNA by monitoring incorporation of BrdU (Fig. 4A), and the mitotic index by immunostaining with α-tubulin to detect mitotic spindles (Fig. 4B). In P1-P3 cultures 25–35% of cells incorporate BrdU. Interestingly, in P4 and P5 cultures as well as in P1 cultures from the 4pX-1- p53kd cell line nearly 60% of the cells incorporate BrdU. Likewise, the percent of cells undergoing mitosis (Fig. 4B) increased by 2.5-fold from P2 to P3 cultures (p<0.05) and 5-fold from P3 to P5 cultures (P<0.005).
Next, we investigated whether BrdU incorporation in P1-P5 cultures is associated with DNA damage (Fig. 5A). Phosphorylated H2AX (γH2AX) is a marker of DNA damage induced by DNA re-replication (19, 33). The number of γH2AX-positive cells progressively increased (p<0.05) from P1 to P5 cultures. Immunoblots of γH2AX confirmed this increase (Fig. 5C). To investigate whether DNA damage occurs in cells that synthesize DNA, P1-P5 cultures were co-stained for BrdU and γH2AX. Nearly 50% of γH2AX-positive cells were also BrdU-positive. Compared to P1 cultures, the number of P5 cells positive for both markers displayed a statistically significant (p<0.05) increase (Fig. 5A). Based on our earlier studies (19), the BrdU-positive and γH2AX-positive cells are undergoing DNA re-replication.
To determine whether P1-P5 cells propagate re-replication-induced DNA damage, day6 cultures were co-immunostained for γH2AX and α-tubulin to detect mitotic cells. The number of P4 and P5 cells immunostained for γH2AX while undergoing mitosis was significantly (p<0.05) increased (Fig. 5B), demonstrating enhanced propagation of DNA damage. Since DNA damage induces a cellular response that initiates DNA repair (34), we investigated the potential of pX-expressing polyploid cells to repair DNA damage. Immunoblots of the DNA repair protein MRE 11 (35) demonstrate a progressive decrease in MRE11 level from P2 to P5 cultures (Fig. 5D), indicating defective DNA repair.
Since propagation of DNA damage is linked to genomic instability, we determined the DNA content of P0-P5 cultures by FACS. Nearly 16% of P5 cells contained >4N DNA, representing a 2-fold increase in polyploidy from P0 to P5 cultures (Fig. 6A and Supplementary fig. 5). Although pX induces DNA re-replication (Fig. 5A), a distinct peak of cells with 8N DNA was not observed in the polyploid cultures, suggesting the DNA is partially re-replicated and/or mitotic defects contribute to aberrant DNA content. Employing confocal microscopy, we examined the mitotic spindles of P3-P5 cultures. The most frequent mitotic aberrations are improperly formed mitotic spindles, associated with misaligned chromosomes and unequal division of the DNA (Fig. 6B).
To assess whether DNA re-replication and DNA damage induced by pX in P0-P5 cultures are relevant to HBV-HCC pathogenesis, we made use of microarray data (20) derived from clinical HCC samples, including HBV-HCC. Specifically, the proliferation cluster of genes (MCM4-6, PCNA, TYMS and Plk1) was shown to be elevated in human HCC (20). Accordingly, we quantified by real-time PCR expression of these genes, demonstrating a progressive increase in their mRNA level in P1-P5 cultures (Fig. 7A). The elevated expression of Plk1 is quite intriguing because Plk1 mediates recovery from the G2/M checkpoint (36), is required for mitosis (37), and exhibits highest expression in G2 and M phases (37). Immunoblots of lysates from unsynchronized and G2/M sorted polyploid cells demonstrate a progressive increase in Plk1 from P2 to P5 cultures (Fig. 7B), confirming the real-time PCR data. Intriguingly, BubR1, required for stable attachment of chromosomes to spindle microtubules (38) and activation of the mitotic checkpoint (39), is decreased from P3 to P5 cultures (Fig. 7B). In comparison to other kinetochore proteins, BubR1 depletion has the most severe phenotype of chromosome misalignment (38), as observed in P3 to P5 mitotic spindles (see Fig. 6B).
Genomic instability and loss of p53 are hallmarks of cancer. Moreover, elevated expression of Plk1 (21, 22) and loss of BubR1 (40) are associated with various human cancers. Accordingly, we examined whether P0-P5 cultures display criteria of oncogenic transformation, including growth factor independence, loss of contact inhibition and anchorage independence.
Phase contrast images of day6 P1-P3 cultures grown in 2% serum showed reduced proliferation compared to those grown in 10% serum (Fig. 8A). By contrast, P4 and P5 cultures retained the ability to proliferate in low serum, and in 10% serum, they overgrew the monolayer and formed foci, as it occurs with 4pX-1-p53kd cells (Fig. 8A), indicating loss of contact inhibition. To investigate anchorage independent growth, P1-P5 cells were grown on soft agar as a function of pX expression (Fig. 8A). P1 and P2 cells did not form colonies on soft agar. P3 cells formed small colonies, whereas P4 and P5 cells displayed robust anchorage-independent growth, forming large colonies on soft agar, similar to P1 cells from the 4pX-1-p53kd cell line (Fig. 8B and C). Inhibition of pX expression by tetracycline addition, suppressed colony formation on soft agar by P3-P5 cells, demonstrating that oncogenic transformation is pX-dependent (Fig. 8A). Since Plk1 is elevated in P3-P5 cultures (see Fig. 7), we examined whether inhibition of Plk1 had an effect on pX-mediated oncogenic transformation. Addition of the Plk1 inhibitor BTO-1 (41) suppressed formation of colonies on soft agar (Fig. 8A and Supplementary fig. 6), indicating that Plk1 mediates effects of pX in oncogenic transformation.
Genomic instability is the most consistent marker of malignancy. However, it remains unresolved whether genomic instability is cause or consequence of cancer development (42). Herein, we employed partially polyploid pX-expressing immortalized hepatocytes isolated by live cell-sorting and demonstrate that propagation of DNA damage for approximately 40 cell doublings precedes pX-induced oncogenic transformation. As in human HCC tumors (20), our in vitro cellular model exhibits enhanced expression of a cluster of genes involved in proliferation, including Plk1. Significantly, Plk1 inhibition suppresses pX-mediated oncogenic transformation. Since Plk1 inhibitors are currently in clinical trials for various human cancers (43), Plk1 could be a target for therapy of HBV-HCC. Thus, the significance of this study is establishment of an in vitro liver cancer progression model that simulates liver cancer pathogenesis in chronic HBV patients.
The pathogenesis of HBV-HCC in chronic HBV patients proceeds slowly, involving effects of chronic inflammation of the liver (8) as well as effects of the weakly oncogenic pX (11, 13, 14). Moreover, genomic aberrations occur frequently in HBV-HCC (2), but mechanisms that initiate and propagate these genomic changes remain to be understood. We have shown that pX induces DNA re-replication and DNA damage, resulting in partial polyploidy (19). Herein, we demonstrate that pX continues to propagate re-replication-induced DNA damage to successive generations of polyploid cells (Fig. 5), resulting in oncogenic transformation after 40 cell generations (Fig. 8). We conclude that pX-induced DNA re-replication initiates genomic aberrations leading to oncogenic transformation. These results agree with observations that pX potentiates c-myc-induced liver carcinogenesis in transgenic mice (13), because c-myc overexpression has been linked to DNA damage (44), and pX likely amplifies c-myc-induced DNA damage, thereby accelerating liver carcinogenesis. Whether continuing liver inflammation in chronic HBV patients synergizes with pX to promote re-replication-induced DNA damage, polyploidy, and oncogenic transformation remains to be determined. This mechanism of re-replication-induced DNA damage is relevant to HBV-HCC pathogenesis. We base this conclusion on microarray analyses (http://genome-www.stanford.edu/hcc) which revealed genes involved in DNA synthesis (MCM4-6, PCNA and TMYS) and G2/M phase progression (Plk1) to be elevated in human HCC tumors including HBV-HCC (20). These genes are also up-regulated by pX in 4pX-1 cells progressing through S and G2 phases (19) and in pX-expressing P1-P5 cultures (Fig. 7). Importantly, genes up-regulated in conditions of DNA damage are the same as those induced in normal S phase progression (45).
Thorgeirsson and Grisham (46), based on clinical data of preneoplastic lesions and various types of HCC, concluded that loss of mechanism(s) that maintain genomic integrity propagates chromosomal aberrations randomly, while also selecting for genetic changes that support high proliferation rates. Our results support these conclusions. Specifically, more than 50% of P0 cells undergo apoptosis within 24h after re-plating (Fig. 2). Interestingly, pX rescued a statistically significant number of cells from apoptosis (Fig. 2), suggesting pX down-regulates mechanisms that maintain genomic integrity. Furthermore, loss of p53 function coincides with a statistically significant increase in the mitotic index (Fig. 4B), suggesting that abrogation of p53 function and deregulation of mitosis occur by the same mechanism. A candidate molecule for this mechanism is Plk1, known to mediate recovery from the G2/M checkpoint (36). Plk1 is clearly increased in G2 phase of P3 cultures, and intriguingly, the reverse occurs for BubR1 (Fig. 7). Since BubR1 is necessary for p53 activation in mitosis (47), we propose that loss of BubR1 results in misalignment of chromosomes as well as absence of p53 activation, resulting in cell survival (despite DNA damage) and oncogenic transformation. The mechanisms mediating the decrease of BubR1 and MRE11 in the polyploid cultures are currently under investigation. Since Plk1 inhibition suppresses oncogenic transformation in our cellular model (Fig. 8) and Plk1 is also up-regulated in HBV-HCC (20), inhibition of Plk1 could provide a target for prevention and/or therapy of HBV-HCC in chronic HBV patients.
Live cell-sorting of cells containing >4N DNA from A. 4pX-1 and B. 4pX-1-p53kd cell lines. Circled areas indicate cells with >4N DNA. Cells were grown ± tetracycline as described (25), harvested by trypsin treatment, resuspended in DMEM/F12 medium, and stained with the Hoechst 33422 (500ng/ml) 1h prior to cell sorting. Cells were sorted for DNA content employing the Epics Altra cell sorter (Beckman-Coulter). Calibration (not shown) was performed using chick erythrocytes for each assay.
The DNA content of X-expressing cells without cell fixation was determined by dual parameter analysis employing the viable DNA dye Hoechst. We monitored total fluorescence intensity (X-axis) and peak integral fluorescence intensity (y-axis), which distinguishes polyploid cells from cells in the G2 phase (Muehlbauer PA, Schuler MJ, 2005). Parallel assays using the non-viable DNA dyes SYBRGREEN and propidium iodide (PI), ensured Hoechst staining accurately reflected the estimated DNA content (data not shown).
Dual parameter flow cytometry employing fluorigenic substrates for caspase8 (FAM-LETD-FMK) or caspase9 (FAM-LEHD-FMK) and propidium iodide (PI) for fluorescence detection using: A. (left panel) non-sorted 4pX-1 cells; (right panel) polyploid cells immediately after cell sorting. The lower left quadrant identifies cells that are negative for both markers. The right lower quadrant identifies cells negative for PI and positive for activated caspase. The upper right quadrant identifies double-positive cells. B. Polyploid cells grown for the indicated time course after cell-sorting. A representative experiment is shown.
Cell numbers of day6 P0–P5 polyploid cultures grown as shown in diagram of Fig. 1B. 30,000 cells were re-plated in the same culture dish or a new culture dish, to exclude the possibility of selecting cell variants displaying enhanced growth during cell passage. Cells were counted in triplicates using a Beckman Coulter counter. Histograms are from at least three independent culture preparations.
(Left panel) Growth curves of day6 P0-P5 cultures from 4pX-1 and 4pX-1-p53kd cell lines, grown as in Fig. 1B. Cells were counted in triplicates. (Right panel) Cell doublings of P0–P5 cultures, estimated based on 30,000 cells seeded per dish. n=3
FACS analyses of control chick erythrocytes, 4pX-1 cells grown with pX for 10h, and day6 P0-P5 cultures. Arrows indicate the end of the symmetrical G2 peak. Cells to the right of arrow are quantified as partially polyploid (>4N in DNA). The percent of cells containing >4N DNA is shown in each panel. A representative experiment is shown from at least two independent cell-sorting preparations.
In vitro Plk1 immunokinase assays using lysates from 4pX-1 cells in the G2 phase, synchronized by the double thymidine block ±pX expression and BTO-1 addition. BTO-1 was added one hr prior to cell harvesting. WCE (100µg) were immunoprecipitated with Plk1 antibody. Plk1 immunokinase assays used recombinant Topoisomerase II (100ng) as substrate and γ-32P-ATP. Plk1 kinase reactions were performed as described in Li et al, 2008.
This work was supported by NIH grant NIDDK 044533 to OMA. LR was supported by the Purdue University Bilsland graduate student fellowship. LS is supported by a National Science Foundation graduate student fellowship. Acknowledgments: The authors thank Dr. N. Prasad for critical review of the manuscript, and Drs Liu X. and Tang J. for the Plk1 kinase assays.