INTRODUCTION

Top Abstract Introduction Materials and methods Results and discussion References

The presence of extrahepatic replication of hepatitis B virus (HBV) remains controversial. While several earlier studies reported the presence of viral DNA, RNA, and replicative intermediaries in a wide range of extrahepatic sites including peripheral blood mononuclear cells (PBMCs) (5, 13, 17), in a recently published work Köck and colleagues failed to detect the presence of covalently closed circular DNA (cccDNA), an early replicative form of HBV, in PBMCs and concluded that all of the previous findings could be explained by passive viral adsorption on these cells (11). However, the presence of cccDNA and viral mRNAs was later detected in PBMCs from highly viremic patients by another group of researchers (22).

In contrast to extensive studies on liver integration, the possibility of HBV DNA integration in extrahepatic sites, although suggested by some early studies (5, 15), has received little attention. Recently, we reported that PBMCs from chronic hepatitis B patients often harbor HBV DNA sequences with rearrangements in the core promoter region compatible with the intracellular action of topoisomerase I (Topo I) (14). Since this enzyme has been implicated in the liver integration of HBV and of a closely related woodchuck hepatitis virus (25), we speculated that a similar process might take place in PBMCs. However, since HBV replication and subsequent integration in PBMCs is expected to occur at much lower rates than in the liver, detection of viral integration might require exceptionally sensitive techniques. Accordingly, we optimized a two-step PCR-based assay originally described by Sørensen et al. (20). This technique uses biotinylated virus-specific and partly degenerate arbitrary primers with fixed 3' ends which are designed to hybridize to the cellular DNA flanking the integration. Using this technique, we have demonstrated the presence of integrated HBV DNA in PBMCs and analyzed for the first time the sequences of virus-cell junctions.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results and discussion References

Ten patients with chronic hepatitis B were studied. All had been hepatitis B surface antigen positive for at least 1 year and had had recent liver biopsies compatible with chronic active hepatitis. None of the patients had serological evidence of hepatitis D virus, hepatitis C virus, or human immunodeficiency virus type 1 infection, and none had a history of drug or alcohol abuse. The control group consisted of 12 chronic hepatitis C patients who did not have any serological markers of past or current HBV infection.

PBMCs were isolated by Ficoll-Hypaque (Pharmacia, Uppsala, Sweden) density gradient centrifugation, washed three times with phosphate-buffered saline (pH 7.4), and stored frozen until use. DNA was extracted from 5 × 10⁶ to 10⁷ cells or 50 µl of serum with the Isoquick nucleic extraction kit (Microprobe Corporation, Garden Grove, Calif.) and finally dissolved in 20 µl of water. Ten microliters of this DNA solution, or serial dilutions of a plasmid containing the full-length HBV sequence (see Results and Discussion), was added to a 50-µl PCR mixture containing 10 mM Tris (pH 8.3), 10% glycerol, 2.5 mM MgCl₂, 0.2 mM (each) deoxynucleoside triphospates, 1.25 U of Taq DNA polymerase Gold (Perkin-Elmer), 100 pmol of random primer P1 (for plasmid amplification) or P2 (for PBMC amplification), and 50 pmol of the HBV-specific primer HBV1 (Table 1). Amplification was performed in a model 480 Perkin-Elmer thermal cycler with a three-step cycling profile, as follows: 45 s of denaturing at 94°C (after initial denaturing and Taq polymerase activation for 7 min), 45 s of annealing at 55°C, and 1 min of extension at 72°C for a total of 35 cycles, followed by a final extension of 7 min at 72°C.

The amplified biotinylated fragments were isolated with streptavidin-coated magnetic beads (Dynabeads M-280 Streptavidin; Dynal, Oslo, Norway), after which the nonbiotinylated strand was removed by alkali denaturation, neutralized with HCl as suggested by the manufacturer, and subsequently purified on spin columns (Microcon 30; Amicon, Beverly, Mass.). The final volume was adjusted to 30 µl, 2 µl of which was loaded into the second PCR mixture, containing 50 pmol (each) of primers HBV2 and P3 (Table 1). The conditions for second-round PCR were identical to those for the initial amplification except that the number of cycles was limited to 25. A 15-µl volume of the final PCR mixture was fractionated on a 2.5% agarose gel, transferred to a nylon membrane by Southern blotting, and hybridized to virus-specific ³²P-labelled oligonucleotide probe HBV3 (Table 1). This probe was predicted to hybridize to a region immediately downstream of the virus-specific primer HBV2.

PCR products which were positive upon hybridization were cloned with a TA cloning kit (Invitrogen), after which bacterial colonies were further screened for recombinant plasmids containing HBV sequences by the Grunstein-Hogness method (7). The probe used for colony screening was identical to that used for screening of PCR products. Positive clones were subsequently sequenced with a modified T7 DNA polymerase (Sequenase 2.0; United States Biochemical Corporation). DNA sequences were analyzed with the DNASIS program (Hitachi), and the EMBL/GenBank databases were searched with the Blast 2.0 program.

	RESULTS AND DISCUSSION

Top Abstract Introduction Materials and methods Results and discussion References

In designing the assay for detection of HBV integration, we followed the overall strategy outlined by Sørensen et al. (20), which used a similar system to amplify unknown DNA sequences flanking integrated proviruses. The first round of amplification was done with a biotinylated primer specific for the virus (primer HBV1) and a partly degenerate arbitrary primer, P2, which was designed to anneal to the adjacent cellular DNA (Table 1). The latter primer contained 5 fixed nucleotides on its 3' end, followed by 5 or 7 random nucleotides and 24 nucleotides needed for the second-round amplification. In theory, the partly degenerate primer should anneal every 1,000 bp in the genome, keeping the amplification product within the range of standard PCR. After purification of the specific products with streptavidin-coated beads, the second PCR ensued with a nested HBV-specific primer (primer HBV2) and a primer corresponding to the fixed 5' end of the degenerate primer (primer P3). HBV-specific primers were designed to flank the vicinity of direct repeat 1 (DR1) and the distal part of the core promoter region, which are the common viral integration sites in the liver (9, 16).

To test and optimize the assay, we used serial dilutions of a plasmid containing a full-length copy of the HBV genome as a template. The degenerate primer P1 contained a GGCCT sequence at its 3' end, and the predicted amplified fragment was 500 bp. We found that the optimized technique eventually enabled reliable detection of around 10 target copies of the plasmid (Fig. 1).

View larger version (72K): [in this window] [in a new window]   FIG. 1.   Sensitivity of the optimized two-step PCR assay for detection of HBV integrations. This technique uses biotinylated virus-specific and partly degenerate arbitrary primers with fixed 3' ends (see text). Tenfold serial dilutions of the full-length HBV genome cloned into a plasmid were used as a template; the number of template copies was calculated from absorbance readings. A 15-µl volume of the final PCR mixture was fractionated on a 2.5% agarose gel, transferred to a nylon membrane by Southern blotting, and hybridized to a virus-specific 32P-labelled oligonucleotide probe. The approximate limit of detection was about 10 target template copies.

However, as human PBMCs are likely to contain adsorbed and/or replicating HBV sequences, to prevent detection of nonintegrated HBV DNA, the degenerate primer used for the actual experiments had a different 3' sequence (CGCGT), as this motif is rare in wild-type HBV strains. When the outlined technique was used on PBMC samples from 10 HBV-infected patients, a positive hybridization signal with the HBV-specific probe was repeatedly detected in two patients (Fig. 2). Serum samples from all 10 patients analyzed and PBMC samples from all 12 control patients were consistently negative (not shown).

View larger version (37K): [in this window] [in a new window]   FIG. 2.   Detection of virus-human integrations in PBMCs from 10 patients with chronic hepatitis B. The sequences spanning the junctions were amplified with biotinylated virus-specific and partly degenerate arbitrary primers with fixed 3' ends, as described in the text. A 15-µl volume of the final PCR mixture was fractionated on a 2.5% agarose gel, transferred to a nylon membrane by Southern blotting, and hybridized to a virus-specific 32P-labelled oligonucleotide probe. The two positive reaction products (patients 2 and 5) were subsequently cloned and sequenced; these sequences are shown in Fig. 3.

The PCR products were directly cloned, and clones hybridizing to the HBV-specific probe were sequenced. The putative virus-cell junctions of these sequences are presented in Fig. 3. Among the eight clones from patient 2, six different sequences were found representing likely viral integrations upstream of the DR1 sequence (Fig. 3). Similarly, in patient 5, sequencing of five positive clones revealed two different sequences consisting of the HBV fragment fused to putative human DNA. However, since both parent sequences were not known, a region of homology at the crossover site could effectively preclude precise identification of the crossover points. The length of presumably human DNA in the sequenced clones varied from 56 to 314 bp. The relatively small size of these sequences was most probably directly related to the PCR technique used, which preferentially amplifies short sequences and discriminates against longer products. Thus, comparison of these sequences with those deposited in EMBL/GenBank, while supporting their eukaryotic origin, did not allow for specific allocation to any chromosome.

View larger version (15K): [in this window] [in a new window]   FIG. 3.   Nucleotide sequences of the HBV-human junctions found in PBMCs from patient 2 (sequences 1 through 6) and patient 5 (sequences 7 and 8). The sequence of a closely related wild-type strain (GenBank accession no. V00866) is shown at the top. The viral sequence is numbered from a unique EcoRI site; in this system, the 11-bp direct repeats DR1 and DR2 start at positions 1826 and 1592, respectively. Likely Topo I breakage sites (2, 18) are shown in underlined lowercase letters.

To determine the approximate sizes and structures of integrated viral fragments, a series of primers were designed to match the putative human DNA on one side of the junction and HBV DNA on the other side of the junction. Viral primers, which were deduced from the closely related wild-type HBV sequence (GenBank accession no. V00866), were designated to anneal every 400 to 500 bp from the site of integration. Using hemi-nested PCR, we were able to amplify a 0.8-kb virus-cell hybrid DNA fragment in integration 6 from patient 2 and a 1.2-kb fragment from integration 7 in patient 5. Thus, the fragments contained at least a near-full-length X open reading frame and its transcriptional regulating elements (27). Direct sequencing of the amplified products excluded any rearrangements of the integrated viral DNA within about 600 bp upstream of the junction.

In an extensive compilation of somatic cell illegitimate crossover regions, Konopka (12) pointed to the common presence of Topo I preferential sites and the overrepresentation of purine (pyrimidine) runs or sequences rich in thymine and adenine in the near vicinity of recombination points. Thus, identified putative virus-chromosome integrations share many characteristics deemed typical for illegitimate recombinations, and the sites of integration on the side of the virus are similar to those described for the liver (9, 16). However, it should be emphasized that the HBV-specific primers were designed to detect integrations only in the vicinity of DR1. Obviously, integrations at different sites would remain undetectable by the technique used.

The mechanisms by which HBV integrates into human chromosomes are still not fully understood. Wang and Rogler (25), working with the woodchuck hepatitis virus model, provided evidence that integration near DR1 is directly related to the nearby presence of Topo I cleavage motifs where cleavage of the plus strand generates linearized viral molecules which can subsequently be joined to nonhomologous DNA. There is indeed strong evidence that Topo I is directly involved in the process of illegitimate recombination and that it can generate rearrangements at the ligation site (3, 8). In addition, it has been reported to reach particularly high activity in human lymphocytes (10). However, it is unclear why Topo I would effect HBV DNA sequences in the first place, one of the possibilities being that it is recruited by a reverse transcriptase complex, as suggested for some retroviruses (24, 26). Another possibility is that HBV DNA integration is the result of aberrant linear replication (6, 21).

Although our results are compatible with the presence of active viral replication in PBMCs at some time point, caution should be exerted in interpreting the findings of viral mRNAs as evidence of active replication since, hypothetically, these could be transcribed from integrated sequences. Interestingly, Stoll-Becker and colleagues detected unexpectedly large amounts of X mRNA in PBMCs (22), and these transcripts were reported to reach high levels in liver cells with integrated HBV sequences (23). Integrated HBV sequences could also account for situations in which viral DNA is detected in PBMCs but not in serum and, when transcribed, could be responsible for the described persistence of the cytotoxic T-lymphocyte response after clearance of HBV infection (19). Although it remains unclear whether any of the PBMC-integrated HBV DNA sequences are actively transcribed, the presence of a near-full-length X open reading frame and its promoter suggests such a possibility. X protein is likely to play a key role in liver carcinogenesis (4, 23); however, as the eventually translated X protein would be truncated, its transactivating capabilities are likely to be diminished (1).

In summary, our study provides evidence for at least the occasional presence of HBV integration in PBMCs. However, the consequences of this process are currently unclear.