Potential conflict of interest: Nothing to report.
Hypermutations in hepatitis B virus (HBV) DNA by APOBEC3 cytidine deaminases have been detected in vitro and in vivo, and APOBEC3G (A3G) and APOBEC3F (A3F) have been shown to inhibit the replication of HBV in vitro, but the presumably low or even absent hepatic expression of these enzymes has raised the question as to their physiological impact on HBV replication. We show that normal human liver expresses the mRNAs of APOBEC3B (A3B), APOBEC3C (A3C), A3F, and A3G. In primary human hepatocytes, interferon alpha (IFN-α) stimulated the expression of these cytidine deaminases up to 14-fold, and the mRNAs of A3G, A3F, and A3B reached expression levels of 10%, 3%, and 3%, respectively, relative to GAPDH mRNA abundance. On transfection, the full-length protein A3BL inhibited HBV replication in vitro as efficiently as A3G or A3F, whereas the truncated splice variant A3BS and A3C had no effect. A3BL and A3BS were detected predominantly in the nucleus of uninfected cells; however, in HBV-expressing cells both proteins were found also in the cytoplasm and were associated with HBV viral particles, similarly to A3G and A3F. Moreover, A3G, A3F, and A3BL, but not A3BS, induced extensive G-to-A hypermutations in a fraction of the replicated HBV genomes. In conclusion, the editing enzymes A3BL, A3F, and most markedly A3G, which are expressed in liver and up-regulated by IFN-α in hepatocytes, are candidates to contribute to the noncytolytic clearance of HBV. (HEPATOLOGY 2006;43:1364–1374.)
The hepatitis B virus (HBV) infects more than 350 million people worldwide and is a leading cause of end-stage liver disease and of hepatocellular carcinoma.1 HBV is non-cytopathic for hepatocytes; however, most newly HBV infected adult patients develop acute hepatitis because of a strong immune response that clears HBV from the liver, whereas approximately 5% of newly HBV-infected adult patients generate insufficient immunity and become chronically infected.1, 2 Administration of interferon alpha (IFN-α) is a mainstay of therapy for chronically HBV-infected patients.2 Interferons restrict the replication of HBV by inducing the expression of antiviral proteins that inhibit the formation of replication-competent HBV nucleocapsids, and ultimately can result in the resolution of the chronic HBV infection.2–7 HBV and other hepadnaviruses replicate their partially double-stranded DNA genome within cytoplasmic core particles by reverse transcription of encapsidated pregenomic RNA and thus are related to retroviruses.8, 9
The cytidine deaminase APOBEC3G (A3G), which is encoded within a cluster of seven related editing enzymes (APOBEC3A-G) on chromosome 22, provides broad innate immunity against exogenous and endogenous retroelements.10–16 Encapsidated into the retroviral particle A3G deaminates dCs of the retroviral minus strand cDNA immediately on reverse transcription within the infected cell, and the formation of dUs leads to premature degradation of the retroviral cDNA or to non-functional hypermutated proviruses.10–16 HIV overcomes this innate defense barrier in T-cells with the help of the HIV-encoded Vif (virion infectivity factor) protein that specifically targets A3G to proteosomal degradation.17–19 In addition, the editing enzymes APOBEC3F (A3F) and APOBEC3B (A3B) inhibit HIV-1, and A3B and APOBEC3C (A3C) restrict simian immunodeficiency virus replication by cytidine deamination of the viral minus strand cDNA, and A3B and to a lesser extent also A3F are resistant to HIV Vif–induced proteosomal degradation.20–23 Recent evidence suggests that some of the HIV restriction exerted by A3G may be independent of its cytidine deaminase activity.24–26
G-to-A hypermutated HBV genomes have been detected in the plasma of HBV-infected patients, suggesting that APOBEC3 editing enzymes have edited the minus strand cDNA during HBV replication.27–29 In hepatoma cells, transfection of A3B, A3C, A3F, or A3G induced extensive hypermutations in a minor fraction (approximately 10−3) of HBV genomes that was detected by a novel polymerase chain reaction technique (3D-PCR) designed to selectively amplify AT-hypermutated sequences.29 In addition, two groups showed that A3G and also A3F, but not rat APOBEC-1, can inhibit the replication of HBV and of duck hepatitis B virus in transfected hepatoma cells in vitro.30–33 Whether this inhibition of HBV replication in vitro is linked to catalytic cytidine deamination is unknown, and a non-editing mechanism has been proposed as the mode of inhibition.30–33 The low or even absent expression of APOBEC3 genes in human liver, however, has raised the important question as to their impact on HBV replication in vivo. Because IFN-α induces cellular pathways in hepatocytes that lead to a noncytolytic clearance of HBV, we investigated whether any of the APOBEC3 genes might be up-regulated in human primary hepatocytes in response to IFN-α and studied their effect on HBV replication, core-particle association, and HBV DNA editing in vitro.
A3, APOBEC3; HBV, hepatitis B virus; IFN-α, interferon alpha; Vif, virion infectivity factor; 3D-PCR, differential DNA denaturation polymerase chain reaction; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; HA, hemagglutinin; HBcAg, hepatitis B core antigen; HBsAg, hepatitis B surface antigen.
Materials and Methods
Liver Specimens and Cultivation of Primary Hepatocytes and Hepatoma Cells.
Normal human liver tissue was taken from the periphery of liver specimens from patients undergoing surgical resection for colorectal metastases that were HBV and HCV negative and had no apparent liver disease. Informed consent of the patients was obtained in accordance with institutional guidelines and the local ethics committee. Primary human hepatocytes were isolated from the liver specimens following a two-step enzymatic perfusion protocol.34 The viability of the isolated hepatocytes was determined by trypan blue exclusion, and only preparations of over 90% viability were used. The hepatocytes were seeded onto tissue culture plastic coated with rat tail collagen in Dulbecco's modified Eagle medium containing 10% fetal bovine serum, left to attach for 1 to 2 hours, and then washed twice with phosphate-buffered saline (PBS) to remove any remaining non-viable cells from the culture. The hepatocytes were cultured in arginine-free Williams E medium supplemented with insulin (0.015 IU/mL), hydrocortisone (5 μmol/L), penicillin (100 IU/mL), streptomycin (100 μg/mL), glutamine (2 mmol/L), and ornithine (0.4 mmol/L) for 24 hours before use. Recombinant IFN-α (Sigma-Aldrich, cat. No. I4401, St. Louis, MO) was then added to the cultures and subsequently replaced every 24 hours. Human hepatoma HuH-7 and HepG2 cells were cultivated in Dulbecco‘s modified Eagle medium containing 10% fetal calf serum.
RNA was prepared from normal human liver and cultivated human liver cells using TrizolR as described.35 Five micrograms total RNA from liver, primary hepatocytes or hepatoma cells, was reverse transcribed using a commercially available cDNA synthesis kit and oligo-dT primer (StrataScriptR, Stratagene, La Jolla, CA). Conventional polymerase chain reaction (PCR) amplification of target cDNAs was performed as described.35 Real-time quantification of cDNA targets was done using Qiagen QuantiTect SYBR Green Kit (QIAGEN, Hilden, Germany) and the ABI Prism 7700 Sequence Detection System. The primer efficiency E was determined for each target sequence by calculation of the slope of the respective amplification curves as described.36, 37 ΔCt values were standardized against an internal GAPDH control, and fold increase was calculated as (1+E)−Δ ΔCt, relative to an unstimulated control at each time point.33, 34 The following primers were used: A3B: 3B4 (GGTCAGCAATTCATGCCTTGGTAC, nt 574-597), 3B5 (CCCTG TAGATCTGGGCCGGGTCC, as, nt 750-728). A3C: 3C4 (CCCCTCCACCCTGGACCC, nt 738-61); 3C5 (CGCAGGCTGGAGGAACGGGGTCTGT, as, nt 925-901). A3F: 3F10s (CCAGAGTGTGCAGGGGAGGT, nt 919-938), 3F11as (TCCTGCAGCTTGCTGTCCAG, nt 1187-1168). A3G: 3G10s (GCTGTGCTTCCTGGACGTGA, nt 819-838), 3G11as (GGTGGTCCACAAAGGTGTCCC, nt 1102-1182). GAPDH: GAPDH 356s (TTCACC ACCATGGAGAAGGC, nt 356-376) and GAPDH 572as (GGCATGGACTGTGGTCATGAG, nt 593-572). The PCR with A3B4/A3B5 discriminates A3BL mRNA (330 bp product) from alternatively spliced A3BS mRNA with skipped exon 5 (184 bp product). All primers are gene specific and do not cross-amplify any other APOBEC3 cDNA apart from their specific target sequence.
APOBEC3 Expression Plasmids.
Full-length cDNAs for A3BS (IRAKp961N03115Q2), A3C (IRALp962P2327Q2), A3F (IMAGp998L1211571Q3), and A3G (IRALp962I0642Q2) were obtained from the German Resource Center for Genome Research (RZPD). The full-length cDNA of A3BL was provided by Michael M. Malim, London U.K.. A3BL, A3BS, A3C, A3F and A3G were amplified by PCR and inserted in frame into a modified pcDNA6 expression plasmid (Invitrogen, Carlsbad, CA) (provided by Robert Friis, University of Berne, Switzerland) in which the V5 and 6XHis epitopes have been replaced by a carboxyterminal hemagglutinin- (HA-) epitope. The resulting plasmids were entirely sequenced. pCMV-HBV, constructed by Heinz Schaller, Heidelberg, Germany, and provided by Hans Will, Hamburg, Germany, contains a 1.3-fold overlength HBV genome from which pregenomic RNA is synthesized by the CMV promoter.38
HBV Replication in Hepatoma Cells In Vitro.
Human hepatoma HuH-7 cells (2 × 105 cells per 25-mm dish) were transfected with 3 μL FuGENE6 (Roche, Switzerland) and 0.2 μg pCMV-HBV, 0.7 μg pcDNA6, and 0.1 μg pCMV-LacZ per milliliter culture medium. Forty-eight hours after transfection, the cells were lysed with 1 mL lysis buffer (50 mmol/L Tris-HCl pH 7.4, 1 mmol/L EDTA, 1% NP-40), and the lysates were digested with DNaseI (100 μg/mL) at 10 mmol/L MgCl2, for 30 minutes at 37°C. Intracellular HBV replicative intermediates were immunoprecipitated using 3 μL rabbit polyclonal anti-hepatitis B core antigen (HBcAg) antibodies (Dako, Denmark, or Abcam, UK) as described.39 Ninety percent of the isolated DNA were separated by electrophoresis on an 0.8% agarose gel, transferred to HybondN+ nylon membranes, and hybridized with radiolabeled full-length HBV DNA using ExpressHyb (BD Biosciences, Franklin Lakes, NJ). The HBV DNA was quantified by radiophosphorimaging of the southern blots. Beta-galactosidase was determined in the cell lysates by luminometry using Galacto-Light Plus TM (Tropix, Bedford, MA). Secreted hepatitis B surface antigen (HBsAg) was determined in the cell supernatant by the commercially available enzyme-linked fluorescence assay HBsAg Ultra VIDASR (BioMerieux, Lyon, France).
Immunoprecipitation was performed exactly as described, except that the ProteinA-Sepharose was blocked overnight with cell extract of untransfected HuH-7 cells containing 1% bovine serum albumin before use. The immunoprecipitated proteins were separated on a 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis and electroblotted onto Hybond-P (Amersham, Buckinghamshire, UK). For immunodetection of APOBEC3 proteins, an anti-HA antibody [6E2 mouse monoclonal antibody (mAb), Cell Signaling Technology, Danvers, MA], diluted 1:3,000 in blocking solution for 12 hours at 4°C, a polyclonal horseradish peroxidase–conjugated anti-mouse IgG antibody (Bio-Rad, Hercules, CA) diluted 1:2,000 in blocking buffer for 1 hour at 4°C, and the ECL Western Blotting Analysis System (Amersham Biosciences) were used.
HuH-7 cells or HepG2 cells were grown on 60 × 15-mm tissue culture dishes at a density of 5 × 105 cells per dish and transfected with pcDNA6-APOBEC3 without and with pCMV-HBV. After 48 hours, the cells were washed twice with PBS, fixed for 3 minutes in ice-cold acetone-methanol (1:1), and air-dried for at least 30 minutes. The monoclonal HA-specific antibody (6E2 mouse mAb, Cell Signaling Technology), diluted 1:50 in cell culture medium, was applied for 1 hour at 37°C. Cells were washed three times in PBS, and fluorescein isothiocyanate–conjugated anti-mouse antibody (Sigma-Aldrich), diluted 1:200 in culture medium, was applied for 30 minutes at 37°C. Cells were then washed three times in PBS, air-dried in the dark, and mounted with VectashieldR containing 4′,6′-diamidino-2-phenylindole to counterstain the nuclei.
Detection of Hypermutated HBV Genomes.
The HBV DNA from immunoprecipitated replicative intermediates was amplified by PCR with oligonucleotides seqHBVs and rtHBVas and with oligonucleotides XNs (AAGCGGCCGCTATGGCTGCTAGGCTGTGCTGCCAA) and XHas (GCAAGCTTGGCAGAGGTGAAAAAGTTGCATGG). The respective PCR products for the HBsAg/HBV DNA polymerase and the X-protein were cloned and sequenced.
For the selective amplification of hypermutated genomes, differential DNA denaturation (3D-) PCR with hot start and degenerate primers in a two-round procedure was performed exactly as described.29, 40 Briefly, for the first PCR the primers 3D1 (CGCAAATATACATCGTATCCAT) and 3D2 (AAGAGTYYTYTTATGTAAGACYTT), and for the second PCR primers 3D3 (ATGGCTGCTARGCTGTGCTGCCAA) and 3D4 (AAGTGCACACGGTYYGGCAGAT) were used (Y = T/C and R = A/G). The first PCR used standard conditions. The second PCR with 0.5 μL of the first reaction as template was performed with initial denaturation at 87°C for 5 minutes, followed by 35 cycles (87°C for 60 seconds, 45°C for 30 seconds, and 72°C for 30 seconds) and finally 10 minutes’ extension at 72°C under standard buffer conditions. The 3D-PCR products were cloned, and up to 20 recombinant clones were sequenced.
Expression of APOBEC3 Genes in Human Liver.
Oligo-dT primed cDNA pools were synthesized from total RNA of normal human liver specimens and were analyzed for the expression of APOBEC3 genes by PCR. A full-length A3B mRNA form, A3BL, that encodes two cytidine deaminase domains, and an alternatively spliced variant with skipped exon 5 (A3BS), that lacks the carboxyterminal deaminase domain, have been deposited in the NCBI data base. In cDNA pools from three human liver specimens, the A3B PCR amplified two products with sizes of 330 bp and 184 bp, whereas in one liver only faint products of 184 bp and in another no products could be generated (Fig. 1). Both the 330-bp and the 184-bp product were cloned and sequenced, and encoded, as expected, the mRNA forms of A3BL and A3BS, respectively. For A3C and A3F, only faint products were generated, whereas strong products for A3G cDNAs were generated in all five liver tissue specimens (Fig. 1). The identities of the PCR products for A3C, A3F, and A3G cDNAs were confirmed by DNA sequencing.
Interferon alpha–Induced Expression of APOBEC3 Genes in Human Hepatocytes.
Primary human hepatocytes and human hepatoma HepG-2 and HuH-7 cells were cultivated without and with increasing concentrations of IFN-α. In oligo-dT primed cDNAs from these cells, the A3B PCR amplified exclusively the 330-bp product encoding A3BL without evidence of the A3BS product of 184 bp. IFN-α stimulation (1,000 units/ml for 24 hours) appeared to increase the intensity of this product, most markedly in HuH-7 cells (Fig. 2). In these cDNA pools, PCR generated strong products for A3F and A3G, respectively, and their abundance appeared to be further increased by IFN-α stimulation (Fig. 2). For A3C, strong products were generated in HepG2 cells, whereas in primary hepatocytes or in HuH-7 cells very weak products were visible only after stimulation with IFN-α (Fig. 2).
Real-time PCR was applied to quantify the expression of these four APOBEC3 genes in primary hepatocytes or in human hepatoma HuH-7 and HepG2 cells cultivated for 12, 24, and 48 hours without or with increasing concentrations of IFN-α (10, 100, 250, 1,000 units/mL). In all three liver cell types, the expression of these four APOBEC3 genes was stimulated by IFN-α treatment in a dose-dependent manner (Fig. 3A). In primary hepatocytes, the strongest induction was observed for A3G mRNA, with a 14-fold increase compared with unstimulated cells, whereas the respective increases for A3B, A3C, and A3F were between fourfold and sixfold (Fig. 3A). In primary hepatocytes, the response of all four APOBEC3 genes to IFN-α treatment appeared to cease at the 48-hour time point. In HuH-7 cells, a similar dose-dependent increase of the mRNA expression of A3B, A3C, A3F, and A3G, respectively, was observed, whereas in HepG2 cells the overall IFN-α response of all four APOBEC3 genes was diminished in comparison with the primary hepatocytes and the HuH-7 cells (Fig. 3A). The strongest increase (60-fold) was observed for A3G mRNA expression in HuH-7 cells stimulated for 12 hours with 1,000 U IFN-α/mL (Fig. 3A). The real-time PCRs were normalized for the amplification efficiencies of each transcript and were normalized to that of GAPDH cDNA in each sample.36, 37 These calculations allowed us to directly compare the expression of the four APOBEC3 transcripts.36, 37 In unstimulated primary hepatocytes, A3B and A3G mRNAs were expressed to a level of approximately 0.8% relative to GAPDH mRNA, while A3F and A3C transcripts were present only at 0.4% and 0.05%, respectively, relative to GAPDH mRNA (Fig. 3B). After IFN-α stimulation for 12 hours, the abundance of A3G mRNA reached 10% of the GAPDH mRNA level, and that of A3B and A3F approximately 3%, whereas A3C mRNA was induced to only 0.3%, relative to GAPDH mRNA (Fig. 3B).
Inhibition of HBV Replication In Vitro by APOBEC3 Proteins.
A3BL, A3BS, A3C, A3F and A3G were co-transfected into HuH-7 cells together with pCMV-HBV to induce a strong HBV replication in vitro,38 and with pCMV-LacZ to normalize for the transfection efficiency. A3BL inhibited the accumulation of intracellular HBV replicative intermediates as efficiently as A3G and A3F, whereas A3C and A3BS had no inhibitory effect (Fig. 4A). In a series of six independent experiments, the expression of A3BL, but not of A3BS, A3C, A3F, or A3G resulted in a consistent decrease of secreted HBsAg even after correction for transfection efficiency by determination of the co-transfected β-galactosidase (Fig. 4B). Compared with the empty vector control and normalized to the amount of secreted HBsAg, A3BL decreased the formation of intracellular HBV replicative intermediates in this series to approximately 40% and thus achieved a similar level of inhibition as A3G and A3F (Fig. 4C). In all of these experiments, A3BS and A3C showed no apparent inhibition of HBV replication (Fig. 4C).
Association of APOBEC3 Proteins With HBV Replicative Intermediates.
The next question addressed was whether the APOBEC3 proteins associate with HBV replicative intermediates. The rabbit polyclonal anti-HBcAg antibody co-precipitated all five APOBEC3 proteins in strict dependence of the presence of replicating HBV (Fig. 5, lower panel). This experiment further demonstrated that HBV expression did not affect the synthesis of the transfected APOBEC3 proteins (Fig. 5, upper panel). Thus, there is no evidence that any of the HBV-encoded proteins inactivates the APOBEC3 proteins in a similar fashion as Vif affects A3G or A3F.17, 18
A3BL and A3BS Redistribute From the Nucleus Into the Cytoplasm on HBV Replication.
Immunofluorescence microscopy was used to localize the HA-tagged APOBEC3 proteins in transfected hepatoma cells. In contrast to A3F and A3G that were detected predominantly in the cytoplasm as described,12, 18 A3BL, A3BS, and A3C were found nearly exclusively in the nuclei of HuH-7 cells (Fig. 6, upper panel, and data not shown). However, after co-transfection with pCMV-HBV, both A3BL and A3BS were also detectable in the cytoplasm of HuH-7 and HepG2 cells (Fig. 6, middle and lower panel, and data not shown). Approximately 85% of the co-transfected HuH-7 or HepG2 cells contained A3BL predominantly within the cytoplasm, and only 10% of the co-transfected HuH-7 or HepG2 cells contained A3BL predominantly in the nucleus, in contrast to HuH-7 or HepG2 cells transfected only with A3BL where 95% of the cells contained A3BL exclusively in the nucleus. For this assessment, more than 300 transfected cells were analyzed and counted in each cell line. Moreover, after transfection of A3BL into HepG2.2.15 cells, which continuously replicate the HBV genome, A3BL was found predominantly in the cytoplasm of most transfected cells, and most of the remaining transfected cells contained these two proteins both in the cytoplasm and in the nucleus (data not shown).
A3BL, A3F and A3G, But Not A3BS, Induce G-to-A Hypermutations in Replicating HBV Genomes.
Finally, to what extent did APOBEC3 proteins induce G-to-A hypermutations in replicating HBV genomes? The S-region and the X-region of immunoprecipitated HBV replicative intermediates were amplified by PCR, cloned, and sequenced. Three of 52 recombinant bacterial clones from A3G expressing HuH-7 cells contained strongly G-to-A hypermutated HBV sequences with a clear preference for CpC dinucleotides (data not shown), the preferred target for A3G mediated deamination in retroviral minus strand cDNAs.41 In contrast, 50 clones recovered from A3BL expressing HuH-7 cells, and 51 clones from A3BS, A3C, or A3F expressing HuH-7 cells showed only infrequently single base changes without evidence for hypermutations.
Therefore, 3D-PCR was employed to selectively amplify AT-rich hypermutants in this background of wild-type sequences.29, 40 The lowest temperature that still allows PCR amplification of HBV genomes from HuH-7 cells transfected with pCMV-HBV and the empty vector was 88°C. By performing the PCR at a denaturation temperature of 87°C, we were able to amplify HBV genomes from HuH-7 cells cotransfected with A3BL, A3C, A3F, or A3G, but not with A3BS or empty vector control (Fig. 7A). These PCR products were cloned and sequenced. As expected, almost all sequences were extensively hypermutated (Fig. 7B). Most of the sequences were unique, indicating that they represented independent hypermutation events, with the exception of A3C, where only a few independent sequences were recovered (Fig. 7B). In A3F-expressing HuH-7 cells, we also detected edited HBV genomes with extensive C-to-T hypermutations in the plus strand sequence, indicating that the HBV plus strand or the HBV RNA must have been edited by A3F.29 Examination of the dinucleotide context preferences showed typical traits for A3G, A3F, and A3BL, whereas no significant preferences could be noted for A3C (Fig. 7C).
This study demonstrates the mRNA expression of the four APOBEC3 editing enzymes A3B, A3C, A3F, and A3G in normal human liver and their concerted up-regulation in primary hepatocytes in response to IFN-α stimulation. In addition, we demonstrate inhibition of HBV replication in vitro not only by A3G and A3F, but also by A3BL, whereas the alternative splice form A3BS and A3C are inactive in this respect. All five APOBEC3 cytidine deaminases are associated with HBV replicative intermediates, and A3BL, A3F, and A3G, all of which contain two cytidine deaminase domains, can edit replicating HBV DNA to generate extensively G-to-A hypermutated HBV genomes. These results indicate that the editing enzymes A3G, A3F, and A3BL have the capacity to contribute to noncytolytic clearance of HBV and suggest a concerted action of these enzymes in the human liver.
Previous investigations have shown that A3G and A3F inhibit HBV replication in vitro.30–33 Inhibition of pregenomic HBV RNA packaging by A3G was originally proposed by Turelli et al.30 as the mode of action. This is in line with a recent investigation by Rösler et al.,33 who reported an increased nuclease sensitivity of core-protein associated full-length pregenomic HBV RNA on expression of A3G. The extent of hypermutations in replicating HBV DNA induced by APOBEC3 editing enzymes is another matter of debate.29, 30, 32 The most important issue in this research field—the expression of APOBEC3 editing enzymes in the human liver and their potential up-regulation in response to IFN-α—had not been addressed in detail so far. The initial report of the APOBEC3 gene cluster failed to detect expression of A3G or A3B in human liver by Northern blotting; however, two subsequent studies demonstrated co-expression of A3G and A3F in many organs, including the human liver, by Northern blotting or RT-PCR.20, 42, 43 Our study confirms the expression of A3G and A3F in normal human liver and proves the hepatic expression of A3B mRNA. Thus, the concluding statement of Rösler et al.33 that a role of APOBEC3 cytidine deaminases in the HBV life cycle is speculative because expression of cytidine deaminases in untransformed liver has not been shown has to be amended. All four cytidine deaminases expressed in normal human liver are up-regulated in primary hepatocytes on IFN-α stimulation, and A3B, A3F and A3G achieve at least moderate mRNA expression levels. Most notably, the mRNA of A3G reached a level of 10% of the GAPDH mRNA abundance and was the predominantly expressed editing enzyme in primary hepatocytes stimulated with IFN-α. A previous study in the human leukemic T cell line H9 found unaltered A3G mRNA expression on stimulation with IFN-α, but an up-regulation by phorbol esters.44 The mitigated response of the APOBEC3 mRNA expression to IFN-α in HepG2 cells suggests that tumor cell lines may not be the best model to study the IFN-α responsiveness of APOBEC3 editing enzymes.
Here, we demonstrate that A3BL can inhibit HBV replication in vitro and in this respect is at least as efficient as A3G or A3F. The gene regulation of A3B in human liver appears to involve transcriptional and also posttranscriptional mechanisms, because alternative mRNA splicing leads to the synthesis of the inactive A3BS. Moreover, we show that A3B and A3C are localized in the nucleus, in contrast to the cytoplasmic enzymes A3F and A3G.12, 18 On HBV expression, the A3B proteins redistribute from the nucleus into the cytoplasm, where HBV DNA replication takes place.2 The co-immunoprecipitation experiments further demonstrate a physical association of A3BL, A3BS, and A3C with HBV replicative intermediates that are present predominantly within the cytoplasm. The observed redistribution of A3B has two possible explanations: The A3B proteins may be co-exported from the nucleus together with the HBV pregenomic RNA, or, alternatively, newly synthesized A3B proteins are sequestered in the cytoplasm by binding to viral proteins. The reduction of HBsAg secretion by A3BL suggests that A3BL does not only affect HBV DNA replication but also may interfere with HBV protein synthesis or processing. Certainly, the role of A3BL in the HBV life cycle warrants further studies.
The dual deaminase domain enzymes A3BL, A3G, and A3F can edit replicating HBV DNA to generate extensively G-to-A hypermutated HBV genomes, whereas the single-domain enzymes A3BS and A3C were less efficient in this activity. Similarly, editing of retroviral minus strand cDNAs by A3G and A3F has been shown to require their respective carboxyterminal deaminase domain.20, 24 The observed link of inhibitory and editing activity of A3BL and the absent editing and inhibitory activity of A3BS suggest that the carboxyterminal cytidine deaminase domain might be necessary not only for editing of HBV but also for inhibition of HBV replication. The finding that A3C had no inhibitory effect on HBV replication supports this assumption. C-to-T hypermutated plus-strand sequences were observed only for A3F, suggesting differences in the HBV editing activities amongst these enzymes. Our results strongly argue against the notion that A3G requires cellular cofactors for HBV DNA editing that might be missing in HuH-7 cells as previously proposed.32 Deducing the frequency of hypermutated sequences within a background of wild-type sequences from PCR analyses with non-degenerate primers is difficult.30–33 G-to-A hypermutations are extensively present in edited HBV genomes and certainly interfere with primer annealing. PCR amplification with non-degenerate primers therefore is likely to select for wild-type sequences.
In conclusion, the current investigation demonstrates the concerted and interferon-inducible expression of A3BL, A3F, and A3G in human hepatocytes and shows that these three cytidine deaminase editing enzymes possess anti-HBV activities. The induction of A3G mRNA expression in human primary hepatocytes by IFN-α is confirmed by a recent study that also identified IFN regulatory factor elements in the A3G promoter.45 Another recent investigation demonstrates IFN-α as a potent inducer of A3G in human primary monocyte-derived macrophages to override HIV Vif neutralization, suggesting that IFN-α–induced inhibition of HIV may be mediated by APOBEC3 editing enzymes.46 The different cellular localization of A3B as opposed to A3G and A3F, and the editing of plus-strand sequences of HBV only by A3F, suggest that besides their common properties these enzymes may have unique features in their interference with the HBV life cycle. Therefore, a concerted action of these three cytidine deaminases may be required for an efficient contribution to IFN-α–induced non-cytopathic clearance of HBV. This hypothesis should now be tested in appropriate experimental settings, including human hepatocyte chimeric mice and in HBV-infected chimpanzees, where gene-expression analyses so far have failed to detect an up-regulation of APOBEC3 mRNA expression during viral clearance.5
The authors thank Markus Reichmuth and Franziska Suter-Riniker, Institute for Infectious Diseases, University of Berne, for determination of HBsAg concentrations, and Anne-Catherine Andres, Department for Clinical Research, University of Berne, for providing FITC-conjugated antibodies and for help in the immunofluorescence experiments.