|Home | About | Journals | Submit | Contact Us | Français|
During a hepadnavirus infection, viral DNA integrates at a low rate into random sites in the host DNA, producing unique virus-cell junctions detectable by inverse nested PCR (invPCR). These junctions serve as genetic markers of individual hepatocytes, providing a means to detect their subsequent proliferation into clones of two or more hepatocytes. A previous study suggested that the livers of 2.4-year-old woodchucks (Marmota monax) chronically infected with woodchuck hepatitis virus contained at least 100,000 clones of >1,000 hepatocytes (W. S. Mason, A. R. Jilbert, and J. Summers, Proc. Natl. Acad. Sci. USA 102:1139-1144, 2005). However, possible correlations between sites of viral-DNA integration and clonal expansion could not be explored because the woodchuck genome has not yet been sequenced. In order to further investigate this issue, we looked for similar clonal expansion of hepatocytes in the livers of chimpanzees chronically infected with hepatitis B virus (HBV). Liver samples for invPCR were collected from eight chimpanzees chronically infected with HBV for at least 20 years. Fifty clones ranging in size from ~35 to 10,000 hepatocytes were detected using invPCR in 32 liver biopsy fragments (~1 mg) containing, in total, ~3 × 107 liver cells. Based on searching the analogous human genome, integration sites were found on all chromosomes except Y, ~30% in known or predicted genes. However, no obvious association between the extent of clonal expansion and the integration site was apparent. This suggests that the integration site per se is not responsible for the outgrowth of large clones of hepatocytes.
There are approximately 1012 hepatocytes in the human liver, virtually all of which can be infected during a hepatitis B virus (HBV) infection. Chronic infection leads to the development of hepatocellular carcinomas (HCC) in ~25% of patients, with tumors typically emerging after several decades. The tumors are clonal and usually contain integrated HBV DNA that was acquired at an early step in transformation, prior to tumor outgrowth. The tumors themselves typically do not support virus replication and do not contain covalently closed circular DNA (cccDNA), the virus transcriptional template. The fact that the tumors typically contain integrated HBV DNA suggests that they arise through mutation of mature hepatocytes and not from hepatic stem cells or hepatobiliary progenitor cells (37, 44, 62). Though contrary reports have appeared (3, 21), hepatocytes are currently the only liver cell type unambiguously established to be susceptible to HBV infection in most individuals.
With a few exceptions, the virus integration sites found in HCC samples from different patients have not provided insights into the mechanisms of carcinogenesis (8, 9, 53). Aside from showing a possible preference for transcriptionally active regions (36), the sites of HBV integration have appeared somewhat random in location and, by inference, not directly related to cellular transformation. Thus, it has been hypothesized that hepatocyte transformation usually results from mutations that are caused by persistent inflammation, leading to cumulative oxidative damage to the host DNA (18); from the action of a viral oncogene expressed from integrated DNA (11, 20); or even from hit-and-run mechanisms (19). In brief, the change or series of changes in host gene expression that lead to HCC in most human HBV carriers are uncertain. In contrast, inappropriate activation of N-myc2 due to nearby integration of woodchuck hepatitis virus (WHV) DNA, or to integration at the distal win and bn3 loci, is found in the majority of HCC samples from woodchucks chronically infected with WHV (4-6, 12, 14, 26, 45, 54, 56), implying that this is a critical step in the formation of woodchuck HCC. About half of the integration sites inferred to be responsible for N-myc2 activation are within a few hundred to a few thousand base pairs from N-myc2, and about half are in the win locus, which maps 150 to 180 kbp from N-myc2 (14, 45).
The earliest mutations needed to begin the transformation of hepatocytes into tumor cells, whether in chronically infected humans or woodchucks, appear to lead, first, to the formation of “premalignant” lesions, also called foci of altered hepatocytes (FAH). FAH appear prior to HCC, generally do not support virus replication, and are the morphological sites from which HCC is thought to emerge (41, 47, 49, 56). FAH may contain thousands or tens of thousands of hepatocytes, and thousands of FAH may be present in HBV- or WHV-infected livers. Interestingly, some FAH have an obviously malignant phenotype, whereas others do not, suggesting diverse causes for their emergence. In the woodchuck, a subset of FAH was shown to express N-myc2, also a characteristic of HCC in woodchucks, as noted above, but not of other liver cells (56, 57).
A possibly unifying property of HCC and most FAH may be a failure to express HBV (or WHV) (16, 41, 55, 56). Thus, it is possible that many or all FAH begin with clonal expansion of initially rare mutated hepatocytes which, for diverse reasons, cannot support virus replication and as a result are no longer targeted by the antiviral immune response. These cells would have a lower death rate than surrounding hepatocytes (34, 55) and would, as a result, undergo clonal expansion. Thus, immune evasion via loss of virus production may be an important early step in oncogenesis. In fact, even cells that have no morphological changes and are not preneoplastic should undergo clonal expansion if they are unable to support virus infection. This could, for instance, explain the decline over time in virus production and the fraction of infected hepatocytes in long-term HBV carriers (34). It may also explain the observation that woodchucks chronically infected with WHV contain similar numbers of virus-negative FAH and morphologically normal foci of hepatocytes that are evidently not FAH but, like FAH, are observed because they are virus negative (55, 56). The general progression to virus-negative hepatocytes seen in chronic HBV infections in humans may reflect a much more extreme emergence of hepatocytes that are not part of FAH (reviewed in reference 34).
Evidence has been presented that cirrhotic nodules and the more advanced lesions, dysplastic nodules (previously called adenomatous hyperplasia), contain integrated HBV DNA and are clonal in HBV patients (1, 38-40, 60, 61). It remains unclear if virus-negative foci, or even FAH, are clonal in the woodchuck. Whether hepatocyte clones are common in noncirrhotic livers of HBV patients is also unknown.
As a first step in determining possible links during chronic hepadnavirus infection between clonal expansion of hepatocytes, a decline in the fraction of infected hepatocytes, and oncogenesis, we examined the livers of woodchucks chronically infected with WHV (33). Clonal expansion was measured using inverse PCR (invPCR) to detect virus-cell junctions created by random integration of WHV DNA; thus, clonal expansion of hepatocytes leads to an increase, in the hepatocyte population, of the frequency of particular virus-cell junctions that were initially present at only one copy. We estimated that the livers of 2.4-year-old woodchucks contained greater than 100,000 clones of greater than 1,000 hepatocytes, amounting to 1 to 2% of the liver. The extent of clonal expansion may be even greater, because many clones may not contain integrated WHV DNA or may contain virus-cell junctions that were not detectable using the invPCR assay. The incidence of clones of >1,000 hepatocytes could not be explained by random death and regeneration of hepatocytes over the relatively short life span of the woodchucks (33), implying that some other explanation, such as immune evasion, is needed to rationalize their emergence (34). The morphological correlates of these clones have not yet been determined.
While studies using the woodchuck model should lead to a better understanding of liver disease in HBV carriers, analyses of disease progression in the woodchuck using well-characterized human arrays (52) only reliably detect changes in the expression of highly conserved genes. We therefore initiated studies to determine if clonal expansion of hepatocytes was also a feature of chronic HBV infections of chimpanzees and if clonal expansion was associated with particular integration sites in or near growth-promoting genes in host DNA. Like WHV infection of woodchucks, HBV infection of chimpanzees usually does not lead to cirrhosis, although recently a rare case of cirrhosis and HCC was detected in a chimpanzee (R. E. Lanford, unpublished data). Our main expectation was that much larger clones of hepatocytes should be detected in the chimpanzees because of their much longer duration of infection (2.4 years in the woodchucks versus >20 years in the chimpanzees).
Integrated HBV DNA was detected using an invPCR assay (33, 48) carried out on DNA extracted from small fragments of chimpanzee liver collected by needle biopsy. Clonal expansion of hepatocytes was found; however, the anticipated increase in hepatocyte clone size compared to the woodchuck (33) was not observed.
Next, all integration sites were mapped where possible to the human genome, for which more complete gene-mapping data are available than for the chimpanzee. Of the 209 different virus-cell junctions that were detected out of a total of 506 virus-cell junctions that were sequenced, 146 could be mapped unambiguously on human chromosomes, and of these, 58 were mapped to known or predicted host genes. Most of the rest were in repeated sequences, so their locations are ambiguous. For others, the fragment of cell DNA was too short to reliably map. No association was found between integration sites and the extent of clonal expansion of hepatocytes.
Chimpanzees were housed at the Southwest National Primate Research Center (SNPRC) at the Southwest Foundation for Biomedical Research. The animals were cared for in accordance with the Guide for the Care and Use of Laboratory Animals, and all protocols were approved by the Institutional Animal Care and Use Committee. Only a limited number of chronically HBV-infected chimpanzees were available for this study, since most animals are exposed to HBV as adults and clear the infection. All of these animals were positive for HBV surface antigen (HBsAg) and antibodies to HBV core antigen (anti-HBcAg) by commercial serological assays (Abbott Laboratories). Information on exposure to HBV was not available for some animals, since they were HBV positive at first testing. The histories of the animals were varied, since they span a period of 43 years and include housing at several different facilities. The histories relevant to this study are briefly described below. No inclusion criteria were imposed; all HBsAg-positive animals at SNPRC were included in the study.
Chimpanzee 4x0136 had no history of exposure to HBV and was HBsAg positive when first tested in 1983. Experimental treatments included exposure to hepatitis delta virus (HDV) in 1988, treatment with antibody to HBsAg in 1990, and treatment with antisense to HBV in 1994. Immunizations included pre-S peptide in 1985 and recombinant HBcAg in 1995.
Chimpanzee 4x0139 was exposed to HBV in 1979 and to HDV in 1984. Experimental treatments included exposure to recombinant plasmid encoding interleukin 12 in 1999, lamivudine therapy in 2002, and a single dose of pegylated alpha interferon in 2006. Immunizations included canary pox virus encoding all three forms of HBsAg (pre-S1, pre-S2, and S) in 1999, plasmid encoding HBcAg in 2001, and HBsAg pre-S2 protein in 2001.
Chimpanzee 4x0222 was exposed to HBV in 1984 following immunization with anti-idiotypic antibodies to HBsAg.
Chimpanzee 4x0327 had no record of HBV exposure. It was obtained from a private owner and was positive for HBsAg at first testing in 1986. Experimental therapies included antibody to HBsAg in 1990, antisense HBV oligonucleotides in 1994, antibody to HBsAg in 2001, and a small-molecule antiviral in 2005.
Chimpanzee 4x0328 was acquired from the same private owner as 4x0327 and was also positive for HBsAg at first testing in 1986. These two animals have identical sequences in the HBV core gene and may have been exposed to the same source or one may have exposed the other. Experimental therapies included antibody to HBsAg in 1990, 1994, and 2001 and treatment with the same small-molecule antiviral as 4x0327 in 2005.
Chimpanzee 4x0230 was exposed to HBV in 1980 following immunization with HBsAg in 1979. Experimental therapy included 18 doses of alpha interferon in 1982.
Chimpanzee 4x0506 had no history of HBV exposure and was HBsAg positive at first testing in 1983. Experimental treatments included an undefined exposure to HDV prior to 1983, immunization with extracts from cells infected with an HBV-carrying retroviral vector in 1995, an HBV DNA vaccine in 1997, and antibody to HBsAg in 1998.
Chimpanzee 4x0509 had no history of exposure to HBV and was positive for HBsAg upon first testing in 1984. This animal is a sibling of 4x0506 and likely had a common HBV exposure. Experimental therapies included an undefined exposure to HDV prior to 1984, antibody to HBsAg in 1991, immunization with extracts from cells infected with an HBV-carrying retroviral vector in 1995, and an HBV DNA vaccine in 1997.
Chimpanzee 4x0313 was used as an HBV-negative control animal for these studies, and samples prior to and 8 weeks after hepatitis C virus (HCV) infection were included for histology and HBcAg staining (Table (Table1).1). A liver biopsy specimen collected in a subsequent study of transient HBV infection in this chimpanzee was used as a positive control (see Fig. Fig.8A8A).
The chimpanzees live in group housing in the chimpanzee village at SNPRC with indoor/outdoor access. The housing is air conditioned and heated and has color TV. The animals have significant contact with an animal caretaker each day. A full time Ph.D. primate behaviorist and staff provide an environmental enrichment program for the animals and visit each animal daily. The animals are fed Teklad Primate Diet two times per day, as well as fresh fruit and other treats daily. Cages are cleaned two times per day. SNPRC maintains a large staff of veterinarians, and one is on call 24 h per day to provide care for the animals.
The liver biopsy tissues were collected under sedation. Similar to humans, primates recover rapidly from liver needle biopsies and do not appear to experience significant discomfort. Although recovery is uneventful, animals are monitored closely for several hours. Liver tissues were fixed with 10% formalin in phosphate-buffered saline or with ethanol-acetic acid (EAA) (3:1), dehydrated, embedded in paraffin, and sectioned at 5 μM onto glass slides. Liver tissues were also quick-frozen and stored at −80°C for subsequent nucleic acid extraction.
HBV DNA in serum was quantified by real-time PCR. The DNA was purified using the Qiagen DNA Mini Kit essentially as described by the manufacturer. Briefly, samples were incubated with the proteinase K digestion-lysis buffer, bound to the silica gel membrane of a QiaAmp Spin Column, washed with ethanol buffers, and eluted with water. HBV DNA was then quantified by a real-time, 5′ exonuclease PCR (TaqMan PCR Core Reagent Kit; Applied Biosystems) using a primer/probe combination that recognized a portion of the HBV surface gene (30). The primers and probe were selected using Primer Express software (Applied Biosystems, Foster City, CA). The fluorogenic probe was labeled with 6-carboxyfluorescein and 6-carboxytetramethylrhodamine and was obtained from Synthegen (Houston, TX). The primers and probe used were as follows: forward primer (5′-CCGTCTGTGCCTTCTCATCTG-3′; 1551 to 1571), reverse primer (5′-AGTCCAAGAGTYCTCTTATGYAAGACCTT-3′; 1674 to 1646), and Fam-Tamra probe (5′-CCGTGTGCACTTCGCTTCACCTCTGC-3; 1577 to 1602), numbered according to reference 24 (accession no. AF222323). The primers and probe were used at 10 pmol per 50-μl reaction mixture. The PCRs included a denaturation step of 10 min at 95°C and then 40 cycles of amplification using the universal TaqMan PCR standardized conditions: 15 s at 95°C for denaturation and 1 min at 60°C for annealing and extension. Standard curves for HBV DNA copy number equivalents were derived from plasmid DNA. Serum HBsAg levels were determined by end point dilution using an enzyme-linked immunosorbent assay purchased from Abbott Laboratories and purified HBsAg standards. Anti-HBsAg and anti-HBc were determined by enzyme-linked immunosorbent assay (Abbott Laboratories).
DNA was extracted from multiple ~1-mg fragments of liver cut from biopsy cores and containing ~800,000 cells. In brief, the fragments were placed in 400 μl of 0.05 M Tris-HCl, pH 7.6, 0.1 M NaCl, 0.01 M EDTA, 0.5% (wt/vol) sodium dodecyl sulfate, 1 mg proteinase K per ml at 55°C for 2 h with periodic vortexing, by which time the liver fragment had been digested. The mixture was extracted with an equal volume of phenol-chloroform (1:1), and DNA was precipitated from the aqueous phase, after the addition of 1 μl of 20 mg dextran/ml, by the addition of 2 volumes of 100% ethanol.
The substrate for integration into host DNA is predominantly double-stranded linear viral DNA (DSL DNA) (15, 59), formed as an aberrant by-product of hepadnavirus DNA synthesis (46). Integrations occur near the ends of DSL DNA, apparently via nonhomologous end joining. Some integrations also appear to occur via recombination in the cohesive-overlap region of relaxed circular (RC) DNA, possibly after conversion of RC DNA into a fully double-stranded DNA by strand displacement synthesis through the cohesive overlap (59). An even lower level of integration may occur via recombination elsewhere in the viral genome.
Detection of integrants by invPCR has been carried out by assaying for virus-cell junctions occurring near one of the ends of DSL DNA and/or in the cohesive overlap of RC DNA. In the present study, we assayed for virus-cell junctions between cell DNA and sequences near the left end of DSL DNA, i.e., the end located just upstream of the HBV core gene and corresponding to the 5′ end of viral pregenomic RNA (46). As a preliminary step necessary to define restriction endonuclease cleavage sites and to design PCR primers for invPCR, the sequence of the HBV strain present in each chimpanzee was determined by PCR amplification of the 5′ end of the pregenome at nt ~1820 and the viral core gene using the primers 5′-GGGGGAGGAGATTAGGTTAA-3′ (1744 to 1763) and5′-TTATGAGTCCAAGGAATACT-3′ (2472 to 2453) (numbered according to reference 24; accession no. AF222323), followed by direct sequencing.
Following extraction of liver DNA as described above, high-molecular-weight DNA was purified away from low-molecular-weight viral DNA by electrophoresis in 1% low-melt agarose gels in E buffer (0.04 M Tris-HCl, 0.02 M sodium acetate, 1 mM EDTA, pH 7.2, 0.5 μg/ml ethidium bromide). This step was performed to reduce cccDNA, which interferes with the invPCR assay (59). The section of the gel containing DNA with a size of 10 to 20 kbp or larger was excised and equilibrated overnight at 4°C with NEBuffer 3 (New England Biolabs; 0.05 M Tris-HCl, 0.1 M NaCl, 0.01 M MgCl2, 0.001 M dithiothreitol, 0.01% [wt/vol] Triton X-100, pH 7.9). The fluids were discarded, and the gel slice was melted for 20 min at 72°C and then cooled to 37°C. BglII was added, and the DNA was digested for 1 h at 37°C. BglII cleaves the HBV present in the chimpanzees (Table (Table1)1) at nucleotide (nt) 2425 (but not between that location and the 5′ end of DSL DNA at nt ~1820) and at an unknown location in the host DNA that depends upon the site of integration (Fig. (Fig.1).1). The DNA was then purified using a QIAquick PCR purification kit, as described by the manufacturer. Control experiments revealed ~50% recovery of DNA from the QIAquick columns using fragments of lambda DNA ranging in size from 0.5 to 2.4 kbp.
InvPCR assays for HBV DNA integrated near BgIIl in the host DNA were then carried out as previously described (33, 48). The BglII-digested DNA was incubated with T4 DNA ligase to circularize the BglII restriction fragments and then digested with BsaWI, which cleaves HBV DNA at nt 2331, to create linear DNA fragments in which chimpanzee DNA, including the virus-cell junction, was flanked by HBV DNA (Fig. (Fig.1).1). Subsequent digestion with BsaBI, which cleaves at nt 2430, was carried out to reduce the background signal from any remaining viral cccDNA. Though amplification of a full-length cccDNA did not occur under our PCR conditions, cccDNAs with internal deletions, or that religate due to rare cleavages at BglII star sites (sites with an imperfect match to the BglII recognition site), can cause significant background in the assay. The DNA was next serially diluted in 96-well trays (8 rows by 12 columns), and the diluted DNA was subjected to nested PCR (33, 48). In general, the first dilution was distributed to the first row (12 wells) in a plate, the next row contained a further threefold dilution, and so on, with the eighth row serving as a DNA-negative control. The outer primers for chimpanzees 4x0327 and 4x0328 were F1 (5′-CTCGCAGACGAAGGTCTCAAT-3′; 2390 to 2410) and R1a (5′-AGATCACGGACCGAAGGAAAGAA-3′; 1991 to 1969); the inner primers were F2 (5′-CCGCGTCGCAGAAGATCT-3′; 2413 to 2430) and R2a (5′-GTCAGCAGGCAAAAACGAGAGTAA-3′; 1968 to 1945) (Fig. (Fig.1).1). The outer primers for the remaining chimpanzees (Table (Table1)1) were F1 and R1b (5′-AGATCACGGACCGACGGAAAGAA-3′; 1991 to 1969); the inner primers were F2 and R2b (5′-TCAGAAGGCAAAAAAGAGAGTAAC-3′; 1967 to 1944).
The PCR products were subjected to agarose gel electrophoresis, and the bands were cut out of the gels for subsequent sequencing (48). Automated sequencing of virus-cell junctions was carried out using the appropriate R2 primer. Sequence alignments were carried out using the GCG program FASTA in order to locate virus-cell junctions. Cell sequences adjacent to integrated HBV DNA were then screened using Sequencher (Gene Codes Corporation) to identify virus-cell junctions present in hepatocytes that had undergone clonal expansion. The chromosome and nucleotide locations of chimpanzee DNA sequence homologues in the human genome were determined using a BLAST search of the database reference genomic sequences (refseq_genomic). The coding potential at this location in the human genome was evaluated using the Affymetrix Integrated Genome Browser (March 2006 build). The reported transcriptional activities in liver of genes in which integration occurred (see Table S1 in the supplemental material) was derived from the Genomics Institute of the Novartis Research Foundation BioGPS website and database (http://biogps.gnf.org).
Out of 826 sequences that were determined, 320 appeared to be derived from cccDNA, while 506 were from virus-cell junctions. A summary of the locations in HBV DNA of the virus-cell junctions detected in liver fragments from the eight HBV-infected chimpanzees is presented in Fig. Fig.2.2. Of the virus-cell junctions shown here, representing distinct integration events, 120 out of 176 mapped near to but downstream of the left end of DSL DNA, 40 out of 176 mapped with in the cohesive-overlap domain, and 16 out of 176 mapped upstream of the cohesive-overlap domain.
Estimates of cell numbers per milligram of liver were determined using a PCR-amplified fragment of the chimpanzee alpha interferon receptor 1 gene (accession no. NC_006488) (primers 5′-GTAACTTCAGGGTCTCCTTCTT-3′ [4909 to 4930] and 5′-CCTTGGGGCAATCATGTTAT-3′ [5438 to 5419]) as a control for quantitative PCR (qPCR). The primers used for quantifying chimpanzee DNA by real-time qPCR were 5′-TTGCTACCCTTTGGCTGCAT-3′ (5099 to 5118) and 5′-AAAGGGGAGACAGCTGAGAA-3′ (5318 to 5299).
As noted above, liver biopsy specimens were fixed in buffered formalin or EAA (28), paraffin embedded, and sectioned (5 μm). Sections of formalin- and EAA-fixed chimpanzee liver tissue were imported into Australia under a permit from the Convention on International Trade in Endangered Species of Wild Fauna and Flora. Formalin-fixed sections were used for hematoxylin and eosin staining, periodic acid Schiff after diastase digestion, Gordon and Sweets staining of reticulin, and immunostaining of HBcAg. For HBcAg detection, slides were deparaffinized in EZ-DeWax (BioGenex; HK 585) for 10 min and rinsed with water. Antigen retrieval was performed in a microwave pressure cooker for 15 min at 1,000 W and 15 min at 300 W in citrate buffer (antigen retrieval solution; BioGenex; HK 086-9K). The cooled slides were rinsed with water and phosphate-buffered saline and treated sequentially with peroxidase suppressor, universal block, and avidin (all reagents from a Pierce 36000 Immuno Histo Peroxidase Detection Kit). The primary antibody was prepared in rabbits against purified HBV core particles. HBV core protein was expressed using baculovirus, and core particles were purified from the medium of infected insect cells by pelleting them on a sucrose cushion, banding on a cesium chloride gradient, and then banding on a sucrose gradient (2, 31). The slides were incubated for 1 h at room temperature with the rabbit anti-core antibody diluted 1:200 in universal block containing biotin, for 0.5 h with biotinylated goat anti-rabbit immunoglobulin G, and for 0.5 h with avidin-biotin complex. The slides were developed with diaminobenzidine, counterstained with Mayer's hematoxylin, and mounted with Pierce mounting medium.
Sections of EAA-fixed chimpanzee liver were used for detection of HBV DNA by in situ hybridization essentially as previously described (27), incorporating a prehybridization step in 50% deionized formamide, 0.5 mg/ml tRNA, and 0.5 mg/ml carrier DNA without labeled probe for 60 min at 37°C. Full-length HBV DNA (pBluBac45HBV1.3; GenBank V01460 ) and control (pBlueBac4.5; Invitrogen) probes were labeled with digoxigenin-UTP by nick translation (Roche), denatured, and used at 2.5 ng/μl of hybridization mixture. Following in situ hybridization, the sections were counterstained with hematoxylin.
To determine the association between cumulative hepatocyte turnover and clonal expansion due solely to random death and regeneration of hepatocytes, we used the program Comp10, which simulates a liver of 1 million hepatocytes (35). Each hepatocyte at the start is uniquely identified and present only once. In each cycle of the simulation, 2.5% of hepatocytes were selected at random to die and another 2.5% were selected at random to divide to restore liver mass. It should be noted that a cycle is not a day; it is simply a computational cycle. Similar results were generated when 0.5% rather than 2.5% was used per cycle; that is, the results were not sensitive to changes in hepatocyte turnover rates. Because of the large sample population (1 million hepatocytes), differences between runs were minimal and did not alter the conclusions, which were based on the distributions of hepatocyte clone sizes. The program and Fortran code are available upon request to S. Litwin or W. Mason; the code and a more detailed explanation of the program are presented elsewhere (35).
The group of animals used in this study represented all of the chronically HBV-infected animals at SNPRC; no selection criteria were imposed. All of the chimpanzees in this study were positive for HBV DNA (i.e., virus) except for 4x0313, which served as an HBV-negative control (Table (Table1).1). Levels of HBV DNA varied from undetectable (less than 100 to 1,000 genome equivalents [GE] per ml of serum) to greater than 1 × 108 GE per ml at the time of biopsy in 2006, when virus-cell junctions were analyzed. Similarly, all animals except 4x0313 were positive for HBsAg. Serum HBsAg levels spanned a 350-fold range, from 0.2 to 70 μg per ml.
We next examined historical and more recent serum samples to determine if the serum HBV DNA levels were similar before and after the 2006 biopsy. Remarkably, we detected spontaneous clearance or multiple-log-unit reductions in the levels of HBV DNA in four of the nine animals (4x0136, 4x0327, 4x0506, and 4x0509) (Fig. (Fig.3),3), with one other animal (4x0222) having less than 1,000 GE per ml during the entire time frame examined. Levels of HBsAg in 2003 and 2006 were consistent with the findings on HBV DNA. All animals remained positive for HBsAg, and none developed anti-HBsAg antibodies.
DNA for invPCR was extracted from four ~1-mg fragments of liver cut from each needle biopsy specimen. The high-molecular-weight fraction, containing DNA of >10 to 20 kbp, was partially purified by preparative gel electrophoresis. This procedure reduces the amount of cccDNA in the preparations, which otherwise competes with detection of virus-cell junctions by invPCR, primarily through the generation of short, easily amplified viral-DNA fragments (59). The purified high-molecular-weight DNA was then cleaved with BglII, and virus-cell junction fragments were self-ligated to produce a circular DNA, as shown in Fig. Fig.1.1. Cleavage with BsaWI was then carried out to generate a linear DNA in which the virus-cell junction was flanked by viral DNA. The DNA was then serially diluted in 96-well microtiter trays, with each dilution distributed through 12 wells, and subjected to nested PCR with the HBV-specific primers (F1 and R1a or R1b, and F2 and R2a or R2b) that span virus-cell junctions. The PCR products generated in individual wells were resolved by agarose gel electrophoresis, cut out of the gel, and sequenced to identify virus-cell junctions.
Representative results of gel electrophoresis of the products of nested PCRs are shown in Fig. Fig.4.4. Figure Figure4A4A shows a highly repeated virus-cell junction fragment present in a fragment of liver from chimpanzee 4x0136. After correction for DNA losses, this fragment corresponded to a clone size of ~10,000 hepatocytes. Figure Figure4B4B shows a similar analysis of a liver fragment from chimpanzee 4x0328. No highly repeated bands were apparent; however, sequencing of the various PCR products revealed two clones of ~130 hepatocytes and one of ~400 hepatocytes. The results from analysis of four ~1-mg liver fragments from each of the eight chimpanzees are summarized in Fig. Fig.5.5. As shown, the clones ranged in size from ~30 to 50 hepatocytes, the lower-range cutoff of the assay, up to 10,000, with a median size of 200, and with 10 clones of >1,000 hepatocytes.
Since the chimpanzees had been chronically infected with HBV for at least 20 years at the time of biopsy, random death and regeneration within the hepatocyte population could explain the clonal expansion of some hepatocytes with integrated viral DNA and the loss of others, particularly if the liver is viewed as a closed system in which essentially all hepatocyte replacement occurs by division of mature hepatocytes. This was calculated, as described in Materials and Methods, for a simulated liver of 1 million hepatocytes (roughly the size of the liver fragments that were analyzed), which were considered to be initially unique with respect to viral DNA integration sites. Figure Figure66 shows the expected hepatocyte clone size distributions, assuming that all integrants were detectable. As can be seen, in this model, no clones of >1,000 hepatocytes were predicted after cumulative hepatocyte turnover equivalent to replacement of the entire liver 91 times (Fig. (Fig.6B).6B). Even after 273 turnovers (Fig. (Fig.6D),6D), only 1.44% of the clones with greater than 200 hepatocytes would be expected to be greater than 1,000 hepatocytes in size. The fact that 40% of the clones greater than 200 hepatocytes in size were also greater than 1,000 in size in our experiments suggests that the clonal expansion, particularly the appearance of clones of >1,000 hepatocytes in size, is not attributable solely to random death and regeneration in the hepatocyte population. It should be noted that random death and regeneration must still play a role in the emergence of hepatocyte clones.
An alternative, but not exclusive, explanation for clonal expansion is that integration of virus DNA near particular host genes gives the respective hepatocytes a selective growth or survival advantage. To determine if the integration sites clustered at hot spots or, for clonally expanded hepatocytes, were adjacent to genes associated with growth or survival, we attempted to map the flanking cell DNA for all 209 distinct integration sites, detected by invPCR, to the human genome. The human genome map was used because the full chimpanzee map is not yet available. As shown in Table Table2,2, out of 209 distinct integrants, 146 could be mapped to a single human chromosome. Most of the other integrations were associated with host DNA that was repeated one or more times, so that a correct integration site could not be assigned. Integrations were found on all but the Y chromosome; failure to detect Y integrants may reflect the limited number of integrations that were mapped and the fact that only two of the eight HBV carrier chimpanzees were male. Of the 146 that were mapped, 58 mapped to known or predicted genes, 4 to exons, and the remainder to introns (see Table S1 in the supplemental material). Integration in only 6 out of 10 of the clones greater than 1,000 hepatocytes in size could be mapped to a definite chromosome, and of these, only 2 mapped to a known gene.
In brief, neither selective outgrowth of clones of >1,000 hepatocytes due to integration of HBV DNA near (i.e., within a few hundred thousand nucleotides of) a known oncogene nor nonselective outgrowth due to random death and proliferation appeared to explain the emergence of these large clones of hepatocytes that takes place during chronic HBV infection of chimpanzees.
Another possibility is that clones of hepatocytes emerge that have lost the ability to express HBV and are therefore no longer targets of the host immune response. For instance, suppose that all hepatocytes had an equal probably of dividing to maintain liver size but that hepatocytes that did not support infection died at only half the rate (e.g., 0.05% per day) of hepatocytes that did support infection (e.g., 0.1% per day). In this situation, over a period of ~5 years the hepatocytes with a survival advantage would expand 1,000-fold relative to those without this advantage (34). Therefore, selective survival may explain the emergence of hepatocyte clones. To see if this possibility was at least feasible, we asked if there was a significant accumulation of uninfected hepatocytes (i.e., hepatocytes in which virus was not detected by conventional histological assays) during the course of chronic HBV infection in chimpanzees, as is also seen in human infections. We assume that most or all hepatocytes were infected following the initial exposure to HBV.
Chronically HBV- and HCV-infected chimpanzees at the Southwest Foundation for Biomedical Research have been monitored for progression of liver disease for more than a decade. In most instances, sections from formalin-fixed, paraffin-embedded liver tissue at multiple time points were available for analyses. In general, HCV-infected chimpanzees present with no fibrosis and limited histological activity (not shown), while several of the HBV-infected chimpanzees had an Ishak staging score of 2 (25) (Table (Table1),1), denoting fibrosis in most portal areas (Fig. (Fig.7).7). A single HBV-infected chimpanzee (4x0230) progressed to cirrhosis (Ishak score, 6) and HCC in 2008, both of which are rare in HBV-infected chimpanzees (R. E. Lanford, unpublished data). Histological activity in many instances included mild to moderate portal and lobular hepatitis for HBV-infected livers (Table (Table11 and Fig. Fig.7)7) and little to no inflammatory changes in the acutely HCV-infected chimpanzee (4x0313) (Table (Table1).1). To some degree, the differences in staging of liver disease may have been due to the duration of infection, since most of the HBV-infected chimpanzees had been infected for greater than 25 years and only a few HCV-infected chimpanzees were in their second decade of chronic infection.
Immunostaining for HBcAg was performed on the formalin-fixed tissues from multiple time points where available. Control tissues collected in a separate study of acute-resolving HBV infection revealed intense cytoplasmic and nuclear staining in essentially 100% of the hepatocytes at 8 weeks postinfection (Fig. (Fig.8A),8A), while the preinfection tissue from the same animal was negative (not shown). HBcAg stains from the chronically HBV-infected animals displayed several different patterns. In general, the percentage of positive hepatocytes was consistent with both the serum HBV DNA and HBsAg levels and the liver HBV DNA levels (Table (Table11 and Fig. Fig.3).3). Several of the animals, although positive for serum HBsAg and HBV DNA at low levels, did not display significant staining of hepatocytes for HBcAg: 4x0136 (Fig. (Fig.8B),8B), 4x0222 (Fig. (Fig.8D),8D), and 4x0506 (Fig. (Fig.8H).8H). Both 4x0136 and 4x0506 had rare HBcAg-positive hepatocytes, mostly staining in the nucleus, and had areas with lymphocytic infiltrates.
Of particular interest, biopsy specimens from several animals exhibited distinct foci of HBcAg-negative hepatocytes surrounded by hepatocytes that were uniformly HBcAg positive (Fig. (Fig.8C).8C). The foci were particularly well defined in 4x0327 (Fig. 8E and F) in biopsy specimens taken at multiple times during 2005, while biopsy specimens taken in 2006, the time of biopsy for viral-host junction analysis, were mostly negative for HBcAg staining, consistent with the decrease in HBV DNA at that time (not shown). A biopsy specimen taken from this animal in 2004 had only two discernible HBcAg-negative hepatocyte foci, while the biopsy specimens from 2005 had HBcAg-negative hepatocyte foci in nearly every field. A second animal that subsequently cleared HBV DNA during the study period, 4x0509, also exhibited well-defined areas of negative hepatocytes (Fig. (Fig.8I),8I), but this biopsy specimen also contained lymphocytic infiltrates and areas of lobular disarray. The animal rapidly cleared a high-level viremia the year following the biopsy for this study (Fig. (Fig.3).3). Other animals, 4x0139 (Fig. (Fig.8C)8C) and 4x0328 (Fig. (Fig.8G),8G), displayed intense nuclear and cytoplasmic HBcAg staining throughout most of the section, which tended to correlate with a stable, high-level viremia. However, even these animals exhibited areas that did not stain for HBcAg (Fig. (Fig.8C),8C), but they were not as well defined as those in other animals. Thus, a correlation existed between a progressive loss of HBV DNA in the serum during the study period and the presence of HBcAg-negative foci in the liver, consistent with the progressive emergence of mutated or epigenetically altered hepatocytes that were refractory to HBV infection.
Liver tissue from each chimpanzee was also analyzed using in situ hybridization for the lobular distribution of hepatocytes containing detectable levels of HBV DNA (Table (Table11 and Fig. Fig.9).9). Different patterns of HBV DNA were detected in individual chimpanzees. In general, the percentage of hepatocytes containing detectable levels of cytoplasmic HBV DNA was lower than the percentage containing detectable levels of HBV core antigen (compare chimpanzee 4x0328 in Fig. Fig.8G8G and and9C).9C). However, like the distribution of HBcAg-positive hepatocytes, the percentages of HBV DNA-positive hepatocytes varied widely between animals and between individual hepatocytes within each chimpanzee. For example, chimpanzee 4x0230 had high levels of HBV DNA detected in 5 to 10% of hepatocytes, while all other hepatocytes contained undetectable amounts of HBV DNA; chimpanzees 4x0136, 4x0327, 4x0328, and 4x0506 had low or undetectable HBV DNA, while chimpanzee 4x0509 had HBV DNA in >50% of hepatocytes.
In summary, as in HBV-infected humans, there is significant accumulation of apparently uninfected virus-free hepatocytes during chronic HBV infection in chimpanzees.
The invPCR assay detected proliferation of hepatocytes to produce clones of hundreds to thousands of hepatocytes in the livers of chronically HBV-infected chimpanzees. The results were not dissimilar to our findings in chronically infected woodchucks (33). This was surprising, in view of the fact that the chimpanzees had been infected ~10 times as long as the woodchucks at the time that liver biopsy specimens were analyzed. Why larger hepatocyte clones were not found in the chimpanzees than in the woodchucks is unclear. A priori, it might be concluded that this difference existed because liver disease is more severe in the woodchuck. However, we are not aware of any evidence that this is the case. The rapid occurrence of HCC in chronically infected woodchucks (29) might be an indicator of a more active hepatitis, but it more likely reflects the novel mechanism of HCC in the woodchuck, resulting from enhancer insertion with activation of N-myc2 (13, 50) or, occasionally, C-myc (22, 23). Another possibility for the similar clone sizes in the two species is that there are anatomic constraints on clonal expansion, at least in the noncirrhotic liver characteristic of infected woodchucks and chimpanzees. In contrast, more extensive clonal expansion appears to occur in cirrhotic nodules that arise in chronically infected humans (1, 42, 61), with clone sizes reaching 105 to 106 hepatocytes or more.
While various explanations can be proposed, and may exist, for clonal expansion of nontransformed hepatocytes, the actual reasons remain uncertain. At present, we believe that an immune evasion model plays a major role. Hepatocytes that did not support viral infection would have a survival advantage over hepatocytes that produce high titers of virus, as seen, for example, at the peaks of transient infections, since they would no longer be targeted by antiviral cytotoxic T lymphocytes (CTLs). A survival advantage, even without a growth advantage, would lead to clonal expansion of the affected hepatocytes (34). Considerable evidence exists that is consistent with this model.
In humans (7, 17), woodchucks (33), and, as illustrated here, chimpanzees (Fig. (Fig.8),8), a common feature of long-term infections is the emergence of large numbers of hepatocytes that no longer support hepadnavirus replication and antigen expression. In the woodchuck, these are found both in FAH, which are considered preneoplastic, and in foci of morphologically normal hepatocytes (32, 41, 55, 56, 58). This also appears to be the case for FAH in humans (16). In chimpanzees (Fig. (Fig.8),8), as in humans (34, 51), it is also clear that chronic infections can be associated with infection of only 10 to 50% or even fewer hepatocytes, despite the fact that virus is still being produced and, at least in naïve hosts, all hepatocytes would normally be assumed to be susceptible to infection (e.g., as in Fig. Fig.7A).7A). Our inference is that significant numbers of hepatocytes in long-term carriers are no longer virus susceptible, even if a clear focal organization is not evident.
However, there is an alternative or at least coexisting factor that would cause clonal expansion of normal hepatocytes and could contribute to the observation of clonal expansion via invPCR. Random death and regeneration would lead to clonal expansion (Fig. 6A to D) and genetic narrowing of the hepatocyte population (Fig. (Fig.6E)6E) as long as hepatocytes in the adult liver are an essentially closed or at least partially closed population that is maintained mostly by self-renewal. This process could fortuitously lead to the emergence of minor populations of hepatocytes that did not support virus replication, irrespective of whether this allowed them to avoid antiviral CTLs. As noted previously, the extent of clonal expansion of hepatocytes in the infected woodchucks could not be explained entirely by this model of random death and regeneration in the hepatocyte population (33). Whether this is also true in the HBV-infected chimpanzees is less clear. A simple mathematical analysis (Fig. (Fig.6)6) suggested that, as in the woodchucks, something other than random killing and division within the hepatocyte population must be responsible for the clonal expansion that we observed (Fig. (Fig.5).5). The data summarized in Table S1 in the supplemental material do not support the notion that the site of HBV DNA integration is a factor in clonal expansion. The results shown in Fig. Fig.77 to to99 are consistent with, but not proof of, an immune evasion model. Thus, immune evasion still seems a plausible option.
It is important, however, to keep in mind that there are no data that conclusively distinguish between (i) the generation of apparently virus-resistant hepatocytes and neoplastic progression via mutation and clonal expansion of mature hepatocytes and (ii) the alternative possibility that FAHs, foci of virus-negative hepatocytes, and/or HCCs all arise as a result of blocked differentiation and transformation of hepatocyte progenitor cells (43). The latter alternative rests, in part, upon the concept that mature hepatocytes become senescent during chronic HBV infection because of persistent CTL killing and compensatory regeneration; ultimately, this is assumed to mitigate their ability to grow into tumors. For instance, at a rate of turnover of 1% per day, hepatocytes equivalent to 73 livers would have died over a 20-year period of chronic infection, which amounts, in the absence of any contribution from hepatocyte progenitor cells, to all surviving hepatocytes being, on average, the result of 146 serial mitotic events. For comparison, assuming a 0.1% daily turnover in a normal liver, which is likely an overestimate, hepatocytes equivalent to only 7.3 livers would have died, with an average history among surviving hepatocytes of 14.6 rounds of mitosis. Nonetheless, the validity of the senescence argument is uncertain, in part because the amount of turnover in the normal liver, which seems to occur via self-renewal, is unknown and in part because it would be expected that progenitor cells would proliferate and differentiate to form mature hepatocytes as an ongoing process if older hepatocyte lineages have a reduced capacity to replicate. Thus, at least in theory, senescence should not occur as an observable phenomenon.
Thus, current data and concepts of liver maintenance seem most consistent with the conclusion that clonal expansion of normal-appearing hepatocytes, as well as of the altered hepatocytes that form FAH, is the explanation for the hepatocyte clones observed by invPCR. Likewise, immune evasion seems a possible contributor in both cases. By inference, chronic infection would lead indirectly to extensive evolution and repopulation of the liver with hepatocytes which, as a result of mutation or epigenetic changes, are no longer able to support hepadnavirus replication.
The present study supports but does not prove this immune evasion model. Clonal expansion of chimpanzee hepatocytes could not be solely attributed either to random death and regeneration or to integration of viral DNA at particular regions of host DNA. Identification of the morphological correlates of these clones may provide a more definitive explanation of their origin.
We are grateful to Jesse Summers (University of New Mexico), John Taylor, and Christoph Seeger (FCCC) for helpful discussions during the course of this work. We are also grateful for technical assistance from the facilities for DNA sequencing and oligonucleotide synthesis of the Fox Chase Cancer Center, to Kathleen Brasky at SNPRC for veterinary support, and to Gene Hubbard and Edward Dick in the pathology laboratory at SNPRC for preparation and evaluation of liver tissues.
W.S.M. was supported by grants from the National Institutes of Health (5R01AI018641; CA06927) and by an appropriation from the Commonwealth of Pennsylvania. A.R.J. was supported by the National Health and Medical Research Council of Australia (Project Grant 453507). This investigation was conducted in part in facilities constructed with support from Research Facilities Improvement Program grants (C06 RR016228 and C06 RR012087) from the NIH National Center for Research Resources and using resources at the Southwest National Primate Research Center (P51 RR13986). R.E.L., W.S.M., and A.R.J. were also supported in part by grant AI067455 awarded to R.E.L.
Published ahead of print on 17 June 2009.
†Supplemental material for this article may be found at http://jvi.asm.org/.