APOBEC3-Independent Interferon-Induced Viral Clearance in Hepatitis B Virus Transgenic Mice

Interferon (IFN) has been part of the standard treatment ofchronic hepatitis B infection for more than 2 decades, yet themechanism of action of this antiviral remains poorly understood.It was recently observed that members of the human APOBEC familyof cytidine deaminases endowed with anti-hepatitis B virus (HBV)activity are upregulated by type I and II IFNs. However, wedemonstrated that, in tissue culture, these cellular enzymesare not essential effectors of the anti-HBV action of thesecytokines. Here, we show that murine APOBEC3 (muA3) can alsoblock HBV replication. While expressed at low levels in themouse liver at baseline, muA3 is upregulated upon IFN induction.However, in HBV-transgenic muA3 knockout mice, IFN inductionblocked HBV DNA production as efficiently as in control HBV-transgenicmuA3-competent animals. We conclude that APOBEC3 is not an essentialmediator of the IFN-mediated inhibition of HBV in vivo.

INTRODUCTION

Top

INTRODUCTION

Acute hepatitis B virus (HBV) infection is rapidly cleared inmost immunocompetent adults through a combination of cytotoxicand cytokine-mediated noncytolytic responses (6, 38). In contrast,perinatal and early-childhood contaminations usually lead tochronic infections, which constitute the ground for the developmentof cirrhosis, liver insufficiency, and hepatocellular carcinoma.In acutely infected chimpanzees, HBV-specific CD8⁺ T cells producinghigh levels of gamma interferon (IFN- ${gamma}$ ) are essential for viralclearance (38, 43), and IFN- ${alpha}$ has been the treatment of choicefor chronic HBV infection, in spite of its limited efficacyin many patients (15). While the exact mechanisms by which IFNblocks HBV remain unclear, HBV-infected hepatocytes exposedto this cytokine exhibit reduced levels of pregenomic RNA-containingviral cores (27, 44, 45). A remarkably similar effect has beennoted upon overexpression of several members of the human APOBEC3(hA3) family of cytidine deaminases (3, 21, 29, 39, 40), a groupof antiviral proteins also active against exogenous retrovirusesas well as endogenous retroelements (2, 8, 13, 14, 17, 18, 21,22, 24, 32). Because several of these enzymes are induced byIFN in various cell types including hepatocytes, it has beensuggested that they might be responsible for the antiviral effectof the cytokine (3, 21, 30, 37). However, we recently demonstratedthat, at least in tissue culture, IFN efficiently blocks HBVreplication even when human APOBEC3F (hA3F), hA3G, and hA3Bare inhibited (21). The present study was undertaken to addressthis point in vivo.

Animal models of HBV infection are limited to chimpanzee andtupaia, but transgenic mice bearing an integrated terminallyredundant HBV genomic construct produce high levels of HBV antigensand DNA in the liver, kidneys, and serum (17). The high levelsof HBV production, albeit comparable to those measured in chronicallyinfected humans, are not hepatotoxic in these animals. The HBVtransgenic (HBV-Tg) mouse thus represents a suitable systemto assess the effect of antiviral agents, at least those actingon viral production. IFN efficiently blocks HBV replicationin HBV-Tg mice (5, 12, 16, 27, 35, 45). In this species, thereis only one A3 orthologue (9, 19) that can inhibit HIV (1, 11,22), human T-cell leukemia virus (25, 31) and mouse mammarytumor virus (26). We thus asked whether it acts as the downstreameffector of IFN against HBV.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Cells and DNA. Ear primary fibroblasts were isolated and maintained as previouslydescribed (36), while Mus dunni tail fibroblasts (MDTF) andhuman embryonic kidney 293T cells were purchased from the AmericanType Culture Collection. To induce the IFN pathway, cells induplicate were treated with 1 µg/ml murine recombinantIFN- ${gamma}$ (Peprotech) or with 50 µg/ml double-stranded RNApoly(I·C) for 20 h before RNA was extracted. The pCMVaywHBV plasmid, kindly provided by S. Wieland and F. V. Chisari(Scripps Research Institute), expresses the pregenomic RNA undercontrol of a cytomegalovirus (CMV) promoter. The Vif-defectivehuman immunodeficiency virus type 1 (HIV-1 ${Delta}$ Vif) proviral clonewas previously described (42). The plasmid pCMV4-hA3G-HA expressinga hemagglutinin (HA)-tagged form of hA3G was a kind gift fromM. Malim (Imperial College, London, United Kingdom) (33). ThepCMV4-muA3-HA was constructed in the same backbone by swappingthe hA3G cassette with the murine APOBEC3 (muA3) gene excisedfrom the pCDNA3-muA3-HA plasmid (provided by N. Landau, SalkInstitute) (22). The muA3 cDNA contained in this plasmid wasoriginally derived from C57BL/6 mice and corresponds to themost commonly found exon V-deleted form of the transcript.

Virus production, infection, and titration. HBV and HIV-1 ${Delta}$ Vif viruses were produced by transient transfectionin 293T cells using Fugene 6 (Roche). Inhibition of HIV-1 ${Delta}$ Vifinfectivity and of HBV replication by APOBEC proteins was determinedby a single-round infectivity assay on HeLa-CD4-LTRLacZ indicatorcells and quantitative analysis of cytoplasmic core-associatedHBV DNA, respectively, as described previously (7, 21).

Generation and genotyping of HBV-Tg A3^–/– double transgenic mice. HBV-Tg mice (kindly provided by U. Protzer) replicate HBV froman integrated terminally redundant HBV genome (HBV 1.3) (17).A3^–/– mice were the generous gift of C. Rada (23).To generate double-transgenic animals, HBV-Tg mice were crossedwith muA3-deficient mice, and offspring were further crossedto obtain HBV-Tg-muA3^–/– and HBV-Tg-muA3^+/–animals. Genomic DNA was extracted from tail samples using aChargeSwitch genomic DNA mini tissue kit (Invitrogen), adaptedfor the automated Tecan Evoware station. Aliquots of DNA (50to 100 ng) were tested by PCR. The wild-type muA3 allele wasdetected with primers O.mAPO3.5 (5'-GACAACATCCACGCTGAAATCTGC-3')and O.mAPO3.6 (5'-GCGGGAGCTGAAGATGTCCAGGCTC-3'), the knockoutmuA3 allele was detected with primers O.mAPO3.5 and O.neo.1(5'-GCGTTGGCTACCCGTGATATTG-3'), and the HBV transgene was detectedwith primers O.HBV.9 (5'-ACCTGTCTTTAATCCTCATTG-3') and O.HBV.2(5'-AGGCGGATTTGCTGGCAAAG-3').

Protein analyses. Cells were lysed with radioimmunoprecipitation assay buffer(1% NP-40, 0.5% sodium deoxycholate, 0.1% sodium dodecyl sulfatein phosphate-buffered saline) supplemented with anti-proteasecocktail (Calbiochem). Lysates were centrifuged for 30 min atmaximum speed in a refrigerated tabletop microcentrifuge; proteinswere quantified by bicinchoninic acid assay (Pierce) beforeseparation by standard sodium dodecyl sulfate-polyacrylamidegel electrophoresis and Western blotting with mouse monoclonalantibodies against HA (anti-HA antibody 3F10 conjugated to peroxidase;Roche Applied Science) and PCNA (Oncogene Research Products).

In vivo analyses. Blood (100 µl) was harvested by retro-orbital phlebotomy,mixed with 25 µl of heparin, and centrifuged twice for10 min at 2,300 rpm in a refrigerated tabletop centrifuge. Aliquotsof the upper phase, corresponding to serum, were treated withDNase I for 15 min at room temperature to get rid of contaminatingtransgene DNA. After DNase inactivation, viral capsid-associatedDNA was extracted using a DNeasy kit (Qiagen) and subjectedto standard PCR with the O.HBV.9/O.HBV.2 primers. For quantification,2 µl of DNA was used in a real-time quantitative PCR usingprimers and a probe specific for HBV (O.HBV1, 5'-AAGGTAGGAGCTGGAGCATTCG-3';O.HBV2; and O.FAM.HBV.1, 5'-FAM-AGCCCTCAGGCTCAGGGCATAC-3'-TAMRA,where FAM is 6-carboxyfluorescein and TAMRA is 6-carboxytetramethylrhodamine).Each sample was tested in triplicate, and a standard curve withserial dilutions of the HBV-expressing plasmid pCMVayw was includedin each TaqMan run. Serum levels of HBV surface antigen (HBsAg)were measured in duplicate using a Monolisa HBsAg Plus enzyme-linkedimmunosorbent assay kit (Bio-Rad). Poly(I·C) (PO93; Sigma)was injected intraperitoneally once a day for 3 days (5 µg/gof body weight). For cellular RNA analyses, heparinized bloodsamples (100 µl) were resuspended in 500 µl of erythrocyteslysis buffer (Qiagen) and kept on ice for 15 min. Cells wereharvested by two consecutive centrifugations of 10 min eachat 2,300 rpm in a refrigerated tabletop centrifuge. A totalof 30 mg of liver harvested from sacrificed mice was cut intosmall pieces, subjected to mechanical homogenization through18-gauge (0.5-in.) and then 22-gauge (.25-in.) needles, andmicrocentrifuged for 3 min at maximum speed at 4°C. RNAwas then purified from the clarified supernatant. For both lymphocytesand liver cells, cytoplasmic RNA was extracted using an RNeasyMini Kit (Qiagen), including an on-column DNase step, and convertedinto double-stranded cDNA by SuperscriptII reverse transcriptase(Invitrogen) using random hexamers (Roche) as primers. Quantificationwas performed by real-time PCR using primers and probes specificfor muA3 and amplifying an exon junction (O.mA3.1, 5'-CAGAATCTTTGCAGGCTGGTT-3';O.mA3.2, 5'-CTTCCAACACTTTTTAAATTCGTATAGG-3'; O.FAM.mA3.1, 5'-FAM-AGGAAGGAGCCCAGGTGGCTGC-3'-TAMRA).The following housekeeping genes, with the primers and probes,were used for normalization: murine connexin 6a1 (mCox6a1),O.mCox6a1.1 (5'-CTCTTCCACAACCCTCATGTGA-3'), O.mCox6a1.2 (5'-GAGGCCAGGTTCTCTTTACTCATC-3'),and O.FAM.mCox6a1 (5'-FAM-CCCACTTCCGACCGGCTATGA-3'-BHQ); murineperipheral myelin protein 22 (mPmp22), O.mPmp22.1 (5'-TTCGTCAGTCCCACAGTTTTCTC-3'),O.mPmp22.2 (5'-ACTCGCTAGTCCCAAGGGTCTA-3'), and O.FAM.mPmp22.1(5'-FAM-CGGTCGGAGCATCAGGACGAGC-3'-BHQ, where BHQ is Black Holequencher); or the 18S normalization gene (using the 20x 18SrRNA mix from Applied Biosystems). Each sample was tested intriplicate, establishing a standard curve with serial dilutionsof muA3 expressing plasmid pCDNA3-muA3-HA. For each sample,a normalization factor was obtained by calculating the geometricmean of the values obtained for the housekeeping genes, andthis value was subsequently used to normalize the relative amountsof the RNAs of interest. Hepatic levels of metallothionein II(MTII) RNA were measured by reverse transcription-PCR (RT-PCR)using the primers O.mMTII.1 (5'-TTGCGCTCGACCCAATACTC-3') andO.mMTII.2 (5'-CGGAAGCCTCTTTGCAGATG-3').

RESULTS AND DISCUSSION

Top

RESULTS AND DISCUSSION

We first tested whether muA3 is active against HBV. For this,we cotransfected 293T cells with either an HIV-1 ${Delta}$ Vif- or anHBV-expressing plasmid together with a control vector or vectorsencoding HA-tagged versions of muA3 and hA3G. As previouslyobserved (1, 4, 11, 22), at comparable expression levels, muA3was as efficient as hA3G at blocking HIV-1 ${Delta}$ Vif (Fig. 1, graybars). Both proteins also had similar antiviral activities againstHBV (Fig. 1, white bars).

View larger version (15K):
[in this window]
[in a new window]

FIG. 1. muA3 is a potent inhibitor of HBV. 293T cells were transfected with HBV or HIV-1 ${Delta}$ Vif plasmid together with empty vector or with vectors encoding hA3G-HA or muA3-HA. Antiviral activity against HIV-1 ${Delta}$ Vif was determined by measuring the infectivity of virions produced in the supernatant of transfected cells in a single-round assay by titration on HeLa-CD4-LTRLacZ indicator cells. Values were normalized for input virus (as determined by RT activity). The effect on HBV was scored by quantitative real-time PCR of core-associated HBV DNA produced from transfected cells. Results obtained in the presence of the empty vector were given the arbitrary value of 100% in both cases. Data are representative of duplicate experiments. Below the graph is a Western blot analysis of extracts from virus-producing 293T cells using antibodies against the HA tag of A3 protein derivatives or PCNA.

There is a high degree of variability in serum levels of HBVantigen and DNA among HBV-Tg mice, possibly due to modulationof transgene expression by hormonal and metabolic factors (20,34). In order to explore the possibility that muA3 might participatein this regulation, we crossed HBV-Tg mice with muA3 knockout(A3^–/–) mice (17, 23). We then selected HBsAg- andage-matched male mice differing in their muA3 genotype and quantifiedcapsid-associated HBV-DNA in the serum by real-time PCR. Becausethe production of HBV DNA-containing capsids initially appearedmore heterogeneous among A3^–/– mice, we examined15 of these versus 7 mice each of the A3^+/– heterozygousand A3^+/+ homozygous controls. This analysis revealed no dramaticdifferences between the three groups, with A3^–/–,A3^+/–, and A3^+/+ mice producing on average 32,318, 19,360,and 17,080 HBV DNA molecules per µl of serum, respectively(Fig. 2).

View larger version (8K):
[in this window]
[in a new window]

FIG. 2. Equal levels of HBV production in HBV-Tg muA3-defective and muA3-competent mice. HBV particles were purified from the serum, and core-associated HBV DNA was quantified by quantitative real-time PCR. A total of 15 A3^–/–, 7 A3^+/–, and 7 A3^+/+ mice were analyzed. Bars represent the mean values for each group.

As observed for human APOBEC proteins, muA3 expression is tissuespecific (11, 22, 23), but its presence in murine hepatocyteshas not been formally documented. We thus first probed bothperipheral blood lymphocytes (PBLs) and liver from four C57BL/6mice by real-time quantitative RT-PCR. In all four animals,baseline hepatic levels of muA3 RNA were at least 15 times lowerthan those measured in PBLs (Fig. 3). We then checked whetherIFN can stimulate muA3 expression. We treated duplicates ofprimary fibroblasts freshly isolated from two mice as well asMDTF with murine IFN- ${gamma}$ at 1 µg/ml and measured levels ofmuA3 mRNA 20 h later by quantitative real-time RT-PCR. We observeda 2.5- to 5-fold induction of muA3 mRNA in IFN-treated cellscompare to untreated control cells (Fig. 4) in a dose-dependentfashion (data not shown).

View larger version (9K):
[in this window]
[in a new window]

FIG. 3. muA3 is expressed at low levels in the mouse liver. Cytoplasmic RNA was extracted from both liver and PBLs isolated from four different mice. Levels of muA3 gene expression were determined by quantitative real-time RT-PCR and expressed relative to the levels of the 18S normalization gene. To ensure that only RNA was detected by amplification, control reactions were performed with omission of the reverse transcriptase at the RT step (wo RT). Amounts represent means and standard deviations of values obtained from the four animals.

View larger version (14K):
[in this window]
[in a new window]

FIG. 4. muA3 expression is induced by IFN in vitro. Primary fibroblasts (fibro) isolated from two different mice and MDTF were treated in duplicate with 1 µg/ml murine recombinant IFN- ${gamma}$ for 20 h before RNA was extracted. muA3 gene expression was recorded by quantitative real-time RT-PCR and expressed relative to the housekeeping Cox6a1 and Pmp22 genes. Normalized muA3 expression levels obtained from untreated cells were set as 1.

Finally, we examined the IFN sensitivity of viral replicationin HBV-Tg mice defective for A3. For this, we first measuredserum HBV changes in response to the IFN inducer poly(I·C)in 20 age-matched mice, half A3^–/– and half A3^+/+,selecting five females and five males in each group. Baselineserum levels of HBsAg were heterogeneous among mice but withouta significant difference between A3^–/– and A3^+/+animals (Table 1). Mice were injected intraperitoneally withthe double-stranded RNA poly(I·C) (5 µg/g of bodyweight) on three consecutive days, a procedure previously demonstratedas optimal for inducing the IFN- ${alpha}$ /β pathway (10, 28). Theday after the last injection (day 1), two mice in each group(mice E and F and mice M and P) as well as four HBV-Tg untreatedcontrol mice were sacrificed. Induction of the IFN pathway andof muA3 mRNA levels was checked by RT-PCR. As previously described(41), the IFN-responsive metallothionein mRNA was upregulatedin mice treated with poly(I·C) (Fig. 5a). The muA3 mRNAwas also induced 2- to 2.5-fold in the two HBV-Tg A3^+/+ poly(I·C)-treatedmice as measured by quantitative real-time RT-PCR (Fig. 5b).The efficiency of HBV replication in the 10 A3^+/+ and 10 A3^–/–poly(I·C)-treated mice was monitored by measuring serumlevels of HBV DNA-containing capsids either before treatmentor 1 and 4 days after the last injection of poly(I·C),first by standard PCR (Fig. 6a) and then by quantitative real-timePCR (Fig. 6b). In all mice for which HBV DNA was detected atbaseline, the level decreased after poly(I·C) treatment,irrespective of the A3 genotype. Noteworthy, this response wasalso seen in A3^–/– mice for which HBV DNA quantificationyielded an estimate of more than 10⁷ genome equivalents/ml,comparable to levels found in the serum of chronically infectedpatients.

View this table:
[in this window]
[in a new window]

TABLE 1. Characterization of the 20 double-transgenic mice^a

View larger version (23K):
[in this window]
[in a new window]

FIG. 5. Induction of cellular IFN-induced mRNAs in the liver of poly(I·C)-injected mice. At day 1 after the last poly(I·C) injection, two A3^–/– (E and F) and two A3^+/+ (M and P) mice were sacrificed, and RNA was extracted from the livers. (a) The IFN-induced MTII mRNA and the Pmp22 control mRNA were detected by RT-PCR. To check for the specificity of the RNA detection, the F and M RNAs were also included in a control reaction where the RT enzyme was omitted (C– wo RT). RT-PCR was also performed without RNA (C–). (b) The induction of muA3 was quantified by real-time RT-PCR in the A3^+/+ poly(I·C)-treated mice (M and P), in four HBV-Tg untreated mice, and in one C57BL/6 mouse. The mean obtained for the HBV-Tg control mice was set as 1.

View larger version (35K):
[in this window]
[in a new window]

FIG. 6. Poly(I·C)-mediated IFN induction inhibits HBV replication as efficiently in A3^+/+ as in A3^–/– mice. Serum levels of HBV-DNA associated with secreted capsids were monitored the day before the first poly(I·C) injection (0) and on days 1 and 4 after the last injection by RT-PCR (a) and further quantified by quantitative real-time PCR (b). Mice that were sacrificed at day 1 after the last poly(I·C) injection are indicated with an asterisk. neg, DNA extracted from serum of mice not transgenic for HBV; pos, DNA extracted from serum of parental HBV^+/+ mice; C–, PCR performed without DNA; 10 ng, PCR performed with 10 ng of HBV-expressing plasmid; p.i., postinjection.

These experiments unequivocally demonstrate that, in HBV-Tgmice, IFN-mediated inhibition of HBV production occurs independentlyof A3. This strongly suggests that in humans as well, in spiteof similarities between the molecular characteristics of APOBEC-and IFN-induced blockade of HBV replication, the cytokine doesnot exert its antiviral effect through induction of the cytidinedeaminases. The downstream effectors of the anti-HBV effectof IFN thus remain to be identified.

ACKNOWLEDGMENTS

We thank Patrick Descombes from the Genomics Platform of theNCCR Frontiers in Genetics at the University of Geneva for thedevelopment of the TaqMan analysis macro and the design of primersfor cellular controls, Johan Jakobsson for providing murineear fibroblasts, and Severine Reynard and Sujana Nylakonda fortechnical help.

This work was supported by the Strauss Foundation and the SwissNational Science Foundation.

FOOTNOTES

* Corresponding author. Mailing address: Global Health Institute, School of Life Sciences and Frontiers in Genetics National Center for Competence in Research, Ecole Polytechnique Fédérale de Lausanne, CH-1015 Lausanne, Switzerland. Phone: 41 21 693 17 51. Fax: 41 21 693 16 35. E-mail: didier.trono{at}epfl.ch

Published ahead of print on 23 April 2008.

Present address: Departments of Dermatology and Venereology,Microbiology and Molecular Medicine, University Hospital andMedical School of Geneva, Geneva, Switzerland.

REFERENCES

Top