|Home | About | Journals | Submit | Contact Us | Français|
To understand subcellular sites of hepatitis B virus (HBV) replication, we visualized core (Cp), polymerase (Pol), and pregenomic RNA (pgRNA) in infected cells. Interestingly, we found that the majority of Pol localized to the mitochondria in cells undergoing viral replication. The mitochondrial localization of Pol was independent of both the cell type and other viral components, indicating that Pol contains an intrinsic mitochondrial targeting signal (MTS). Neither Cp nor pgRNA localized to the mitochondria during active replication, suggesting a role other than DNA synthesis for Pol at the mitochondria. The Pol of duck hepatitis B virus (DHBV) also localized to the mitochondria. This result indicates that localization of Pol to mitochondria is likely a feature of all hepadnaviruses. To map the MTS within HBV Pol, we generated a series of Pol-green fluorescent protein (Pol-GFP) fusions and found that a stretch spanning amino acids (aa) 141 to 160 of Pol was sufficient to target GFP to the mitochondria. Surprisingly, deleting aa 141 to 160 in full-length Pol did not fully ablate Pol's mitochondrial localization, suggesting that additional sequences are involved in mitochondrial targeting. Only by deleting the N-terminal 160 amino acids in full-length Pol was mitochondrial localization ablated. Crucial residues for pgRNA packaging are contained within aa 141 to 160, indicating a multifunctional role of this region of Pol in the viral life cycle. Our studies show an unexpected Pol trafficking behavior that is uncoupled from its role in viral DNA synthesis.
IMPORTANCE Chronic infection by HBV is a serious health concern. Existing therapies for chronically infected individuals are not curative, underscoring the need for a better understanding of the viral life cycle to develop better antiviral therapies. To date, the most thoroughly studied function of Pol is to package the pgRNA and reverse transcribe it to double-stranded DNA within capsids. This study provides evidence for mitochondrial localization of Pol and defines the MTS. Recent findings have implicated a non-reverse transcription role for Pol in evading host innate immune responses. Mitochondria play an important role in controlling cellular metabolism, apoptosis, and innate immunity. Pol may alter one or more of these host mitochondrial functions to gain a replicative advantage and persist in chronically infected individuals.
Infection with hepatitis B virus (HBV) can lead to a chronic infection that persists for life and significantly increases the risk for HBV-associated liver disease, including hepatocellular carcinoma (1). HBV is a global health problem, with 350 million people chronically infected worldwide, among which about 700,000 die every year from HBV-related complications (2, 3). Current drug regimens include interferon (IFN) therapy and treatment with nucleoside analogs, neither of which is curative (4). A better understanding of the viral life cycle is needed to guide new drug development for complete viral eradication.
HBV is an enveloped double-stranded DNA (dsDNA) virus that preferentially replicates in hepatocytes (5). At only 3,200 bp, it is the smallest known virus with a dsDNA genome, and it maximizes its coding capacity by employing overlapping open reading frames (ORFs) to encode the core (Cp), polymerase (Pol), envelope (env; S, M, and L), and X proteins. During an infection, HBV enters the host cell by interacting with sodium taurocholate cotransporting polypeptide (NTCP), a cell surface receptor, depositing the viral capsid into the host cell cytoplasm (6). Next, the capsid traffics to the nuclear pore and disassembles, releasing the dsDNA genome into the nucleus, where it is converted by host enzymes to a covalently closed circular DNA (cccDNA) (7,–9). cccDNA serves as a transcriptional template for the synthesis of pregenomic RNA (pgRNA) and subgenomic RNAs (sgRNAs). The pgRNA is translated to make Cp and Pol. Pol interacts with a stem-loop structure (epsilon [Ep]) at the 5′ end of pgRNA to initiate packaging of the pgRNA-Pol complex within a capsid formed by repeating Cp subunits, in a process called encapsidation (10, 11). Reverse transcription (RT) occurs within capsids. Cp is an active participant in genome replication, synthesizing first the minus-strand DNA and subsequently the dsDNA from the pgRNA template (12,–15). Three distinct template switches occur during genome replication, among which the second template switch determines if the major form of relaxed circular (RC) or the minor form of duplex linear (DL) dsDNA is synthesized (16). Envelope proteins translated from the subgenomic RNA are cotranslationally inserted into the post-endoplasmic reticulum (post-ER), pre-Golgi compartments, where they preferentially interact with dsDNA-containing capsids for envelopment and virion secretion in a noncytopathic manner (17).
HBV Pol is an 832-amino-acid (aa) multifunctional protein with four domains or regions (see Fig. 3A). Starting at the N terminus, the terminal protein (TP) domain—unique to hepadnaviruses—contains residues that form a binding surface for pgRNA during encapsidation (18, 19). In addition, the TP domain contains the primer for the initiation of synthesis of minus-strand DNA by covalently linking the first nucleotide to tyrosine at position 63 (20, 21). The TP domain is followed by the spacer domain, a poorly conserved region within Pol that can withstand various deletions and insertions without significantly altering the genome replication functions of Pol (22, 23). Reverse transcriptase and RNase H domains in the C-terminal portion are responsible for reverse transcribing and degrading the pgRNA, respectively, during genome replication within capsids. Several host factors have been shown to interact with Pol during viral replication. Heat shock protein 90 (Hsp90) is required to maintain Pol in an active conformation to interact with Ep. Apobec3G, a host restriction factor, has been shown to interact with Pol in an RNA-independent manner and is packaged into capsids (24). Multiple studies have also linked Pol to host immune modulation by prevention of IRF3 phosphorylation and blocking of translocation of NF-κB and STAT1 to the nucleus, which is thought to dampen host innate immune sensing (25,–27).
Although many aspects of the HBV life cycle have been elucidated, the subcellular locations of HBV during replication remain largely unknown. Only Cp has been explored significantly in subcellular localization studies. Histopathological studies of infected livers identified two distinct subsets of cellular staining, where the Cp distribution was either predominantly nuclear or cytoplasmic (28,–31). A previous study showed that arginine-rich domains (ARD) within Cp play a role in Cp's nuclear export and import (32). Also, Cp has been shown to localize predominantly to the nucleus when only Cp is expressed but to switch to a predominantly cytoplasmic distribution upon expression of pgRNA and Pol, suggesting that active intracellular replication can influence Cp distribution (33). Previously, immunofluorescence identified Pol as a predominantly cytoplasmic protein with granular staining (34). In addition, studies with duck HBV (DHBV) showed that the majority of Pol is not located within capsids but associates with an unidentified cytoplasmic structure (35).
In this study, we demonstrate that HBV Pol localizes to mitochondria. Mitochondrial localization is also found with DHBV Pol, occurs independently of other viral components, and, remarkably, is regulated by multiple mitochondrial targeting signals (MTSs) mapped to Pol's N-terminal TP domain. We define a discrete, transferable MTS in Pol (aa 141 to 160) that overlaps residues involved in packaging of pgRNA.
The HBV molecular clones were derived from the ayw strain (GenBank accession no. V01460.1). The green fluorescent protein (GFP) ORF flanked by a GGGGSGGGG linker was inserted into the unique EcoRI site of the Pol gene in the previously described HBV expression plasmid LJ96 (HBVCp+Pol+X+) to generate plasmid HBVCp+PolGFP+X+ (36). The plasmid HBVCp+PolGFP+X+ bears a deletion in the Ep sequence that prevents pgRNA from being packaged. In addition, the preS1 and preS2 proteins are prematurely truncated by the GFP insertion. Also, the start codon of the S gene was changed to ACG. To study intracellular replication, the previously described pgRNA donor plasmid NL84 (HBVEp+), which does not express Cp, Pol, X, and env, was used (37). All variants generated from plasmid HBVCp+PolGFP+X+ are listed in Table 1. For a subset of studies, we used the plasmid HBVEp+Cp+PolY63F+X+env+, which is similar to LJ96 except that it does not have an Ep deletion and instead introduces a Y63F change which ablates the DNA priming capacity of Pol but does not affect its pgRNA packaging function (21). In addition, HBVEp+Cp+PolY63F+X+env+ has the nucleotides at positions 484 to 487 changed from CAGG to GTCC to disrupt the splice acceptor site (spl−), preventing formation of spliced RNAs (38). Plasmids HBVEp+Pol+X+env+ and endo-HBVEp+Pol+X+env+ were used to compare Pol expression levels from the cytomegalovirus (CMV) immediate early promoter and the endogenous promoter, respectively. Both plasmids express pgRNA and all viral proteins except Cp, but they differ in the promoter driving pgRNA expression.
DHBV plasmids were derived from DHBV3 (GenBank accession no. DQ195079.1). Wild-type (WT) DHBV3 expression plasmid D1.5G (DHBVEp+Cp+Pol+env+), from which all other clones were derived, has been described previously (39). Plasmid DHBVPol+, derived from D1.5G, has a deletion from nucleotides (nt) 7059 to 7090 (N-terminal Ep deletion to prevent RNA packaging), a 4-nt deletion within the NsiI site, and a T1327A substitution to prematurely truncate the Cp and S genes. Plasmid DHBVPolGFP+ was generated from DHBVPol+ by introducing the GFP ORF flanked by a GGGGSGGGG linker into the Pol gene, between the codons encoding amino acids 292 and 293 of Pol. DHBVPolGFP+ expresses only the Pol gene product. To engender intracellular replication, DHBVPolGFP+ was cotransfected with plasmid DHBVEp+Cp+env+, derived from D1.5G by deleting T424 to prematurely truncate the Pol gene product at aa 88. DHBVEp+Cp+env+ expresses pgRNA, Cp, and the env proteins.
The HBV Pol-only expression plasmid Pol-GFPonly was generated by placing the Pol gene sequence from Pol-GFP downstream of the CMV immediate early promoter by using the HindIII site in the multiple-cloning site (MCS) of the pCR3.1 plasmid (Invitrogen). A stop codon was introduced after the GFP gene in the Pol-GFP plasmid, and the resulting Pol 1-292-GFP-stop sequence was PCR amplified and cloned into the pCR3.1 background at the HindIII site to generate plasmid Pol 1-292-GFP. All subsequent clones for the truncation analysis were derived from the Pol 1-292-GFP plasmid, whose HindIII and KpnI sites were used to clone various PCR-generated fragments. In addition, the deletion clones in the full-length Pol context were derived from Pol-GFPonly. The clones generated are highlighted in Table 1.
To label the mitochondria with a fluorescent protein, the Mito-GFP plasmid was generated from a CMV-GFP expression plasmid by appending the N-terminal 33-aa mitochondrial targeting signal of Tom20 (Gene ID 9804) followed by a 10-residue glycine linker to the N terminus of GFP.
The human hepatoma-derived cell lines Huh7 and HepG2 have been described previously (40, 41). LMH is a chicken liver hepatocellular carcinoma cell line (42). HEK 293 cells and cos7 cells are human embryonic kidney-derived and African green monkey kidney-derived cells, respectively (43, 44). Cells were cultured in Dulbecco's modified Eagle essential minimal medium nutrient mixture with Ham's F-12 medium (DMEM-F12; Gibco) and were supplemented to final concentrations of 5% (Huh7) or 10% (HepG2, 293, and cos7) fetal bovine serum (SAFC Biosciences) and 10 U/ml penicillin-streptomycin (Gibco). HepG2-NTCP12 cells stably expressing the sodium taurocholate cotransporting polypeptide (NTCP) were a gift from Haitao Guo and have been characterized previously (45). HepG2-NTCP12 cells were cultured in the same medium as that for HepG2 cells, but under selection with 8 μg/ml blasticidin (MP Biomedicals). HepAD38 cells also used the same medium as that for HepG2 cells but were maintained with 0.4 mg/ml G418 (Gibco) and 3 ng/ml doxycycline-HCl (Dox-HCl) (Fisher) to repress HBV expression (46). All HepG2 cells and derivatives were maintained on collagen-coated dishes for optimal growth and morphological characteristics. Briefly, tissue culture dishes were overlaid with 50 μg/ml rat tail collagen (BD Biosciences) in 0.02 N acetic acid solution and left at room temperature for at least 1 h for coating. The plates were then washed with 1× phosphate-buffered saline (PBS) and used immediately or stored at 4°C for up to a week, until use. In addition, for all imaging studies on cells plated on glass coverslips, collagen coating was used to aid in cell adherence.
All transfections were performed with linear polyethyleneimine (PEI) with a molecular weight (MW) of 25,000 (Polysciences). Briefly, cells were seeded the day before transfection in culture dishes with or without glass coverslips (18 mm, no. 1.5; neuVitro). On the day of transfection, the cells were 60 to 70% confluent. The total mass of DNA transfected per condition was 1 μg/3.8 cm2 cell culture growth area and was adjusted accordingly for different cell culture growth areas. The final ratio of DNA to PEI was 1:3 (μg:μg) in the final DNA–PEI–Opti-MEM mix. Medium on cells was replaced with fresh prewarmed medium between 6 and 18 h posttransfection.
For intracellular viral DNA (icDNA) isolation, cell cultures were washed with 1× PBS and then lysed with 0.2% NP-40, 50 mM Tris-HCl, and 1 mM EDTA, pH 8.0, for 20 min at 37°C. Nuclei were removed by centrifugation at 4,000 rpm for 5 min at 4°C and discarded. Supernatants were adjusted to 2 mM CaCl2 and 44 U of micrococcal nuclease (New England BioLabs) to digest transfected plasmid DNA for 2 h at 37°C. The samples were adjusted to 10 mM EDTA and 0.4% SDS, and pronase (Roche) was added to a final concentration of 400 μg/ml for 2 h at 37°C to release the encapsidated icDNA and to deproteinize the icDNA. The samples were then extracted with phenol-chloroform, ethanol precipitated, and resuspended in TE solution (10 mM Tris, 0.1 mM EDTA, pH 8.0) supplemented with 1 μg RNase A (total) per sample.
Southern blotting was performed as previously described (47). Briefly, icDNA underwent gel electrophoresis in a 1.25% TBE gel in 1× TBE buffer (90 mM Tris-borate, 2.5 mM EDTA, pH 8.0). DNA was transferred to a Hybond-N membrane (GE Life Sciences) by a passive upward capillary transfer method using 10× SSC buffer (1.5 M NaCl, 0.15 M sodium citrate) overnight. The membranes were air dried, and the nucleic acid was cross-linked to the membranes with UV light. The membranes were then hybridized with 10 pM 32P-end-labeled oligonucleotide probes at 46°C in Church's hybridization buffer (5 mM EDTA, 1% bovine serum albumin [BSA], 0.25 M Na2HPO4, 7% SDS, pH 7.2). The blots were washed with Church's wash buffer (1% SDS, 20 mM Na2HPO4, 1 mM EDTA), and then the membranes were air dried and exposed to a phosphor screen (Molecular Dynamics). Scans were carried out on a Typhoon 8600 phosphorimager (Molecular Dynamics) and subsequently analyzed with ImageQuant 5.2 software (GE Life Sciences).
Denaturing gel protein analysis and Western blotting were performed as previously described (48). Briefly, cytoplasmic lysates were loaded onto 12% SDS-PAGE gels for protein separation and then transferred to modified polyvinylidene fluoride membranes (Immobilon-FL; Millipore) by a wet transfer system using Towbin transfer buffer (25 mM Tris, 192 mM glycine, 20% [vol/vol] methanol, pH 8.3). Posttransfer, the membranes were blocked with a 1:2 dilution of Li-Cor blocking buffer in 1× PBS. Blocked membranes were probed with a rabbit anti-core antibody (1:1,000; Austral Biologicals) followed by an IR 800-conjugated goat anti-rabbit secondary antibody (1:20,000; Li-Cor). The membranes were scanned on a Li-Cor Odyssey infrared imager and analyzed with Li-Cor imaging software.
For native gel analysis of capsids, cytoplasmic lysates were electrophoresed through a 1.0% TBE gel and then transferred by a passive upward capillary transfer method using 1× TNE buffer (10 mM Tris, 150 mM NaCl, 1 mM EDTA, pH 8.0), with two membranes in the sandwich. The front membrane was an Immobilon-FL membrane, and the back membrane was a Hybond-N membrane. This method allowed the detection of both proteins and nucleic acids on the two membranes individually in a single transfer. The Immobilon-FL membrane was treated in the same way as the denaturing gel transfer membrane for immune detection. The Hybond-N membrane was air dried, and the capsids were denatured in 0.2 N NaOH, 1.5 M NaCl for 1 min and then neutralized in 1 M Tris-HCl, 1.5 M NaCl for 5 min. The nylon membranes were then air dried and treated with UV light. These membranes were treated like Southern blot membranes for hybridization and exposure. The only difference was that the end-labeled probes for native gel capsid analysis were designed to detect plus-strand nucleic acids.
HepAD38 cells were seeded at 70% confluence in a 10-cm dish, derepressed by removal of Dox-HCl, and also maintained without G418. The medium was changed every 2 days, and the spent medium was spun at 1,000 × g to remove cell debris and stored at −70°C until use. Cultures were maintained until day 20, after which the plates containing HepAD38 cells were washed and frozen at −70°C until further analysis. To measure the amount of virus, polyethylene glycol 8000 (PEG 8000) was first added to a fraction of medium supernatant, to a final concentration of 10%, and the sample was left on a shaker overnight at 4°C for virion precipitation. The next day, tubes were spun at 1,000 × g for 30 min to pellet virions. Each pellet was resuspended in a 1/100 volume of serum-free DMEM-F12 and used for native gel analyses or treated with digestion buffer (1.2 μg/μl proteinase K, 0.5% SDS) for at least 1 h and analyzed by Southern blotting.
For infection, HepG2-NTCP12 cells were seeded onto collagen-coated glass coverslips in 12-well plates (Costar). Fully confluent cells were infected by overlaying spent HepAD38 medium adjusted to 4% PEG 8000 and 2% dimethyl sulfoxide (DMSO), which represented 500 viral genomic equivalents (VGE)/cell. The plates were then spun at 1,000 × g for 1 h at room temperature for spinoculation (45). For control infections, either fresh medium or spent medium supplemented with Myrcludex B (MyrB) at a final concentration of 500 nM was used. MyrB was previously described as a potent HBV entry inhibitor and was a gift from Stephan Urban (49). For the MyrB conditions, cells were exposed to the drug for at least 1 h prior to infection and maintained until 18 h postinfection, at which point the inoculum was removed and fresh prewarmed medium was added to each culture. Cells were maintained until 7 days postinfection and then processed for immunofluorescence or fluorescence in situ hybridization (FISH) analysis.
All fixed-cell imaging studies were carried out on cells grown in a 12-well plate and seeded onto collagen-coated 18-mm, no. 1.5 glass coverslips. For staining of mitochondria, 200 nM Mitotracker Red CMXRos (Life Technologies) diluted in culture medium was added to the cells and incubated for 25 min. Cells were then fixed with 4% paraformaldehyde (PFA; EM Sciences), and nuclei were stained with 5 ng/ml 4′,6-diamidino-2-phenylindole (DAPI; Thermo Fisher). The coverslips were then mounted on glass slides by use of Mowiol mounting medium (120 mg/ml Mowiol 4-88 [MW, 31,000]; Sigma) and were left in the dark overnight at room temperature to harden before imaging.
A subset of fixed-cell imaging samples was analyzed by immunofluorescence assay. Samples were treated similarly to the fixed-cell samples mentioned above until PFA fixation. Cells were then washed once with 1× PBS, permeabilized with 0.2% Triton X-100, blocked with a 2% BSA solution, and incubated with a primary antibody overnight at 4°C. For Cp detection, a 1:500 dilution of a rabbit anti-core antibody (Dako) was used, whereas for Pol detection 1:100 dilutions of the mouse anti-Pol 2C8 and 8D5 antibodies (Santa Cruz) were used in combination to obtain the best sensitivity of detection. The next day, cells were stained with 1:500-diluted secondary antibodies (goat anti-rabbit 594, goat anti-mouse 488, and goat anti-mouse 594; Life Technologies). Cells were then stained for nuclei with 5 ng/ml DAPI, mounted with Mowiol mounting medium, and left to harden overnight in the dark before imaging.
For fluorescence in situ hybridization (FISH), 26 oligonucleotide FISH probes labeled with Quasar-570 dye were generated. The FISH probes were designed based on strain ayw coordinates between nt 1987 and nt 2805, rendering the probes specific to pgRNA and spliced RNAs but not subgenomic RNAs. FISH was performed as recommended by the Stellaris FISH protocol (Biosearch Technologies), using Stellaris hybridization and wash buffers. All reagents were prepared with RNase-free water. Briefly, cells were fixed with 3.7% formaldehyde in 1× PBS. Cells were then permeabilized with 70% (vol/vol) ethanol. Coverslips were inverted on 100 μl of 125 nM FISH probe mix for at least 4 h in a humid chamber at 37°C, followed by nuclear staining with 5 ng/ml DAPI. The coverslips were mounted on glass coverslips with Mowiol mounting medium and imaged.
A Zeiss Axio Observer.Z1 wide-field epifluorescence microscope was used for all imaging studies. A 10× (numerical aperture [NA] 0.3) objective and a 63× (NA 1.4) oil-immersion objective in conjunction with an X-cite 120Q mercury-arc lamp light source (Excelitas Technologies) and filter set combinations (Zeiss) were used for fluorescence imaging. Images were captured by use of a Zeiss AxioVision 506 monocamera, and postacquisition image analysis and adjustments were performed with Zen Pro 2012 and Fiji (ImageJ2) image analysis software.
For particle analysis by FISH, raw images were converted to binary images by applying a threshold in the Fiji image analysis software. Binary images were then processed to define the particle boundaries. The particle boundaries were then applied to the original raw images, and fluorescence intensity measurements were obtained for each particle. The sum of all the intensities within each particle was plotted as the integrated fluorescence intensity.
For successful intracellular replication, Cp, Pol, and pgRNA must converge to package the Pol-pgRNA complex into capsids. It is not understood where these components are localized during formation of replication-competent capsids. Previous studies have not simultaneously visualized Cp, Pol, and pgRNA during an infection. To this end, we imaged Cp, Pol, and pgRNA in HBV-infected HepG2-NTCP12 cells by indirect immunofluorescence and RNA fluorescence in situ hybridization analyses (Fig. 1). By performing immunofluorescence analysis of Cp, we showed that HepG2-NTCP12 cells were successfully infected with medium acquired at day 18 from HepAD38 cells and that the infection was almost completely blocked by 500 nM Myrcludex B, an inhibitor of HBV entry (Fig. 1A). We found that the Cp distribution varied from cell to cell (Fig. 1B). We scored cells on their subcellular distribution of Cp staining and found that 49% of cells showed a predominantly cytoplasmic staining (C > N), 39% stained equally well in the cytoplasm and the nucleus (C = N), and 12% had a strong nuclear staining (N > C). Previous studies of liver biopsy specimens from patients infected with HBV demonstrated both nuclear and cytoplasmic Cp and showed that cytoplasmic Cp correlated with active virus replication (80, 81). Other studies suggested that core is nuclear in resting cells and that nuclear breakdown during cell division releases the Cp into the cytoplasm, making it predominantly cytoplasmic (29, 30). We also performed FISH to identify viral pgRNA in infected cells (Fig. 1C). The FISH probes were 26 singly labeled oligonucleotides which should permit detection of RNA at a single-molecule sensitivity (52, 53). Using a particle-counting algorithm (54), we found that cells had 50 to 500 discrete RNA signals per cell. When we measured the intensities of a large number of particles from the same cell, we found that the intensity distribution was unimodal (Fig. 1C, histogram). Such a distribution likely indicates that each particle signal corresponds to a single RNA molecule. Infected cells, which can contain spliced derivatives of pgRNAs, would be detected in our FISH analysis (50).
Unexpectedly, in the cells costained for Cp and Pol, cells positive for Cp expression did not show detectable Pol staining (Fig. 1C, bottom row). This may have been due to a previously reported masking effect of Cp on Pol (51) and/or due to the level of Pol being below the limit of detection. To determine if low levels of expression could explain the inability to detect Pol in infected cells, we transfected HepG2-NTCP12 cells with HBVEp+Pol+X+env+ or endo-HBVEp+Pol+X+env+ and stained the cells for Pol. In the transfections, Pol was detectable with the plasmid expressing HBV from its endogenous promoter (Fig. 1D, top row) and, as expected, was present at a level much lower than that of Pol expressed from the CMV promoter construct (Fig. 1D, bottom row). With the transfection protocol we used, we estimated that 500 to 700 copies of plasmid DNA were present per nucleus, whereas for an infection the estimates for cccDNA copy number range from 1 to 20 copies per cell (55,–57). This difference in copy numbers would be expected to lead to lower levels of Pol in the infected cells and may explain our inability to detect Pol in an infection. Interestingly, in the transfected cells, Pol was exclusively cytoplasmic and its distribution asymmetric, suggesting that Pol was associated with subcellular structures (Fig. 1D, bottom row, inset). This finding is consistent with a previous observation with DHBV that showed that a vast majority of Pol was associated with an unidentified cytoplasmic structure (35).
To determine the subcellular localization of Pol, we cotransfected Huh7 cells with HBVEp+Cp+PolY63F+X+env+ and expression constructs for fluorescently labeled proteins to visualize different subcellular organelles, such as endosomes, lysosomes, mitochondria, ER, peroxisomes, and stress granules. Of all the organelles visualized, Pol colocalized only with the mitochondria, as seen by cotransfecting the Mito-GFP expression plasmid (Fig. 2A). For successful intracellular replication, the Pol-pgRNA complex is encapsidated. Also, it has been shown that RNAs themselves can localize to specific subcellular locations for local protein production (58). If the majority of Pol is at the mitochondria, we wanted to determine whether Cp and pgRNA also localize to the mitochondria. To this end, we exploited two important features of the HBVEp+Cp+PolY63F+X+env+ expression plasmid. First, PolY63F does not synthesize DNA, halting viral replication after pgRNA packaging. Second, splice acceptor site mutations inhibit the generation of spliced RNAs, which can be detected by FISH probes, allowing detection of only pgRNAs in cells (38). To label mitochondria, the Mito-GFP expression plasmid was cotransfected into Huh7 cells. Upon immunofluorescence analysis of Cp, the cytoplasmic core was punctate in appearance, and more importantly, cytoplasmic Cp puncta did not colocalize with Mito-GFP in these cells (Fig. 2B). We also performed FISH analysis of pgRNA in a parallel transfection. Discrete pgRNA signals were observed in cells, with an even distribution across the cytoplasm. On visualization of pgRNA with Mito-GFP, we did not notice any significant colocalization with the mitochondria (Fig. 2C). In summary, these studies showed that Pol was localized to the mitochondria and that Cp and pgRNA did not colocalize with Pol.
To corroborate the immunofluorescence studies and to further study Pol at the mitochondria, we modified HBV Pol by inserting the GFP ORF into the spacer domain (Fig. 3A). The spacer domain is the least-conserved region of HBV Pol, as shown by the conservation score depicted in Fig. 3A, which is an alignment of 7 hepadnaviruses (4 mammalian and 3 avian viruses) where the grayscale range depicts residue conservation from white (least conserved) to black (highly conserved). Previous studies have shown that insertions and deletions within the spacer can have little to no impact on the RT function of Pol (22, 23). Hence, we inserted the GFP ORF flanked by a 4-glycine–serine–4-glycine (G4SG4) flexible linker into the spacer region of Pol (59, 60). First, we analyzed the capacity of HBVCp+PolGFP+X+ to support intracellular replication by measuring viral icDNA levels compared to those for HBVCp+Pol+X+. To prevent the influence of a longer-than-WT-length pgRNA on icDNA synthesis, we expressed the viral proteins and pgRNA from two separate plasmids in Huh7 cells. By Southern blotting, we observed that GFP-labeled Pol supported the synthesis of 73% of the total level of icDNA seen with wild-type Pol, with the synthesis of all three major forms of viral replicative intermediates (RC, DL, and SS [single stranded] DNAs) in their normal proportions (Fig. 3B). This result shows that Pol-GFP is functionally active to generate viral icDNA and, more importantly, that the insertion of GFP did not have a deleterious effect on the proper folding of Pol.
We cotransfected Huh7 and HepG2 cells with HBVCp+PolGFP+X+ and HBVEp+ and saw that the majority of the Pol-GFP localized to the mitochondria as visualized by staining of the cells with 200 nM Mitotracker Red CMXRos (Fig. 3C, left panels). HBV is a preferentially hepatotropic virus. To determine whether mitochondrial localization of Pol was specific to cells of hepatic origin, we transfected two nonhepatic cell lines: 293 and cos7. As shown in Fig. 3C, right panels, Pol-GFP localized to the mitochondria in both 293 and cos7 cells.
To determine whether localization of Pol to the mitochondria was dependent on viral DNA replication, we used the expression plasmid Pol-GFPonly. Pol-GFPonly expressed only GFP-labeled Pol and no other viral protein or pgRNA. We then compared cotransfection of HBVCp+PolGFP+X+ and HBVEp+ (replication positive) to transfection with Pol-GFPonly (replication negative) alone and found that neither active intracellular replication nor any viral protein or pgRNA was required for mitochondrial targeting of Pol in both Huh7 and HepG2 cells (Fig. 3D). This result indicated that Pol contains an intrinsic MTS.
Hepadnaviruses are highly species specific and infect a variety of mammals and birds. If the Pol of a distantly related hepadnavirus localized to the mitochondria, this would indicate evolutionary conservation of this property and be consistent with an important role for Pol at mitochondria. To examine this possibility, we chose a distantly related but well-characterized hepadnavirus: DHBV. DHBV Pol shares only 30% amino acid sequence identity with HBV Pol (Fig. 4A). We generated a DHBVPolGFP+ expression plasmid that has the GFP ORF inserted between the codons for amino acids 292 and 293 in the DHBVPol+ expression plasmid. A strategy similar to the one used with HBV Pol was used to evaluate the intracellular replication capacity of DHBVPolGFP+. Pol expression plasmids (DHBV−, DHBVPol+, and DHBVPolGFP+) were cotransfected with a plasmid expressing pgRNA, Cp, and env (DHBVEp+Cp+env+), and icDNA synthesis was measured. We observed by Southern blotting that DHBVPolGFP+ synthesized 36% of the level of total icDNA seen with DHBVPol+ (Fig. 4B). In addition, DHBV Pol-GFP was competent for synthesis of RC DNA, as shown in Fig. 2B. Next, we determined the subcellular localization of DHBVPolGFP+ in LMH cells, a cell line that is typically used for the study of DHBV. DHBV Pol localized to the mitochondria in LMH cells, as seen by colocalization with Mitotracker Red staining (Fig. 4C, top row). Interestingly, DHBV Pol also localized to the mitochondria in the human Huh7 cell line (Fig. 4C, bottom row). This shared property of HBV and DHBV suggests that the Pol enzymes of all hepadnaviruses localize to the mitochondria to carry out an important function for these viruses.
We demonstrated that Pol contains an intrinsic MTS. Defining the amino acid sequence of the MTS could allow the disruption of Pol's mitochondrial targeting by introduction of mutations within the MTS and observations of their effects on the viral life cycle. To locate the MTS, we first generated a Pol 1-292-GFP construct to determine whether the MTS is within the first 292 amino acids of Pol. Pol 1-292-GFP localized to mitochondria in Huh7 and HepG2 cells (Fig. 5A). To further define the MTS, we generated a series of deletions, as shown in Fig. 5B. From this analysis, we identified the region spanning aa 141 to 160 as necessary for the localization of Pol to the mitochondria. To determine if this region was sufficient for mitochondrial targeting, we generated the Pol 141-160-GFP construct and expressed it in Huh7 cells. We found that the region containing aa 141 to 160 was sufficient to target GFP to the mitochondria (Fig. 5C). Next, we determined whether deletion of aa 141 to 160 within the context of full-length Pol ablated mitochondrial targeting. We analyzed the Δ141–160 construct and, surprisingly, found that its mitochondrial targeting, although weaker than that of WT Pol, was not completely ablated (Fig. 5D, top row). We found that only by deleting the first 160 amino acids (Δ1–160 construct) was the mitochondrial targeting of Pol completely ablated (Fig. 5D, bottom row, and Table 2). The results of our efforts to define the MTS indicate that the mitochondrial targeting signal within Pol is both evolutionarily conserved and multipartite.
Pol binding at the Ep stem-loop on the pgRNA initiates encapsidation. Previous studies identified residues within Pol that are important for pgRNA binding. One of these regions is called T-3 and spans aa 153 to 160 (61, 62). T-3 is within the minimal MTS (aa 141 to 160) that we identified. Because the boundary of T-3 was not precisely mapped, it was possible that other residues within the minimal MTS would play a role in the pgRNA-binding and/or RT function of Pol. To determine if this possibility is the case, we generated a series of di-alanine substitutions within aa 141 to 160 of Pol-GFP (Fig. 6A). These alanine mutants, generated in HBVCp+PolGFP+X+, were cotransfected with HBVEp+ to study intracellular viral replication in Huh7 cells. By Southern blotting of these mutants, we found that the RH-AA, HT-AA, and KA-AG mutants were moderately affected in icDNA synthesis, whereas the YL-AA and LW-AA mutants were severely affected (Fig. 6B). The QT-AA mutation had no impact on icDNA levels. Consistent with our results, Y147 was previously implicated to play a role in icDNA synthesis (63). The icDNA synthesis defect may have occurred during DNA synthesis or at the earlier step of pgRNA packaging. To differentiate between these possibilities, we performed native gel electrophoresis on cytoplasmic lysates to measure the relative levels of packaged plus-strand nucleic acids (pgRNA and plus-strand DNA). Consistent with our Southern blot analysis, the RH-AA and HT-AA mutations had moderate effects and the YL-AA mutant had a severe defect, whereas the LW-AA mutant had plus-strand nucleic acid levels within capsids that were below the limit of detection (Fig. 6C). The KA-AG mutant was similar to WT Pol in its capsid plus-strand nucleic acid content, suggesting that the KA-AG mutation-induced moderate defect in icDNA synthesis occurs during DNA synthesis. These results identify new residues within Pol that contribute to pgRNA packaging. It is possible, if not likely, that these results indicate that the T-3 region extends upstream of aa 153.
To better understand how the N-terminal portion of Pol might act as a mitochondrial targeting signal, we performed algorithm-guided secondary structure analyses of the sequence of Pol. The secondary structure prediction algorithm PSIPRED predicts an alpha-helix from aa 134 to 153 (Fig. 6D) (64, 65). This region overlaps the minimal MTS that we identified. To determine whether this region forms an alpha-helix, we mutated Q143 and T144 to either alanines or prolines. Alanines are predicted to preserve alpha-helices, whereas prolines introduce kinks and break alpha-helices (66). We already showed that the QT-AA change did not affect icDNA synthesis (Fig. 6B and andC),C), indicating that glutamine and threonine at those positions are not required for icDNA synthesis. Imaging revealed that mitochondrial targeting of the QT-AA and QT-PP mutants was unaffected. Southern blot analysis showed a severe decrease in icDNA synthesis for the QT-PP construct. To determine if the defect of the QT-PP mutant occurred during packaging of pgRNA, we compared it to the Y63F-Pol-GFP plasmid, which does not synthesize icDNA, to measure the level of encapsidated pgRNA. Under both conditions, the HBVEp+ plasmid was cotransfected to express the pgRNA. On analysis by native gel electrophoresis and probing for pgRNA, a severe defect in pgRNA packaging was seen for the QT-PP mutant, which explains its subsequent icDNA synthesis defect (Fig. 6F). Western blot analysis did not show a difference in the accumulation of Cp, indicating that low Cp levels are not the reason for the pgRNA packaging defect of the QT-PP mutant (Fig. 6F, top panel).
Our studies show that the previously unidentified subcellular site of HBV Pol accumulation is the mitochondrion. We showed that mitochondrial localization is also a feature of DHBV Pol, indicating an evolutionary conservation of this property. This conservation is consistent with an important role for Pol at mitochondria. Also, we showed that the accumulation of Pol at mitochondria is independent of viral replication and that Pol contains a mitochondrial targeting signal.
Pol interacts with the Ep stem-loop on the pgRNA, and Cp is thought to assemble into a capsid around this complex. Interestingly, neither Cp nor pgRNA colocalized with the mitochondria. It is likely that mitochondria do not represent sites of pgRNA encapsidation for the virus. Although Cp did not colocalize with Pol at the mitochondria, we identified that in an infection of cell cultures, the distribution of Cp was either predominantly cytoplasmic (C > N) or equal in both the nucleus and the cytoplasm (C = N). A previous study of Cp localization in transfected cells showed that almost all cells had predominantly cytoplasmic staining (33). It is possible that Cp subcellular localization is differentially regulated in an infection compared to a plasmid transfection-based assay, in which the viral entry pathways are not used to initiate intracellular replication.
We were not able to detect Pol in HepG2-NTCP12 cells infected with HBV by using the anti-Pol monoclonal antibodies 2C8, 8D5, and 9F9, though it was readily detected when these cells were transfected with HBV expression plasmids (34). Hence, the level of Pol was below our limit of detection during infection. It is possible that the pattern of localization of Pol observed in our transfection experiments is not the same as that during an infection. The localization of Pol during an infection will be elucidated once more sensitive methods for detection become available.
We found that HBV Pol has a minimal MTS spanning aa 141 to 160, but we also found other dependencies within the N-terminal region (aa 1 to 140). Deleting only the sequence from aa 141 to 160 within Pol does not completely ablate mitochondrial targeting. The other well-studied mitochondrial targeting signals are either a 10- to 80-amino-acid presequence at the N terminus or an internal sequence that forms a transmembrane domain for anchoring the protein to the outer or inner membrane of the mitochondria (67). Algorithm-guided protein sequence analysis predicts neither a presequence nor a transmembrane domain in Pol. As such, the MTS of Pol represents a previously unidentified signal sequence for mitochondrial targeting.
The mitochondrial targeting signal is within the TP domain of Pol. The TP domain is unique to hepadnaviruses, whereas the RT and RNase H domains are homologous to their retroviral counterparts. The TP domain has other previously described functions in Ep binding and pgRNA packaging. Specifically, T-3, a sequence first identified in DHBV Pol, forms an important contact point for pgRNA binding, along with the RT domain of Pol (61, 62). T-3 corresponds to residues 153 to 160 (KAGILYKR) in the HBV Pol sequence and is conserved across hepadnaviruses (19). T-3 is within the minimal MTS we identified in this study. We hypothesize that pgRNA binding to Pol occludes the MTS and subsequently blocks Pol's trafficking to mitochondria. As such, Pol would exist in two separate pools within cells: pgRNA-bound Pol that is destined for encapsidation and pgRNA-free Pol that traffics to the mitochondria. This hypothesis is consistent with our finding that core and pgRNA are not enriched at the mitochondria.
We identified residues in Pol, distinct from the T-3 sequence but within aa 141 to 160, that are important for pgRNA packaging. Specifically, the Y147A+L148A and L151A+W152A mutations had significant impacts on pgRNA packaging. In addition, a protein secondary structure algorithm predicts an alpha-helix spanning aa 134 to 153. We showed that perturbing this predicted alpha-helix by changing residues to prolines had a severe impact on pgRNA packaging. In all, our findings show that the MTS is embedded within other functional regions of Pol, making it unlikely to ablate mitochondrial targeting without altering other important functions of Pol.
Why might Pol localize to the mitochondria? Mitochondria play important roles in metabolism, apoptosis, and innate antiviral cell immune defenses (68, 69). Viruses frequently perturb those functions to gain a replicative advantage (70, 71). There are multiple reports implicating HBV Pol in modulating the host innate immune response, but none of these reports indicate the involvement of the mitochondria in manifesting these effects of Pol. Specifically, HBV Pol has been reported to reduce the response to type I IFN signaling by blocking NF-κB, IRF3, and Stat1 nuclear translocation (25, 26, 72). Further, RIG-I has been identified as a direct antiviral sensor for the Ep stem-loop on the pgRNA to repress HBV replication. Also, HBV Pol has been reported to interact directly with the stimulator of interferon genes (STING) protein to block IFN-β responses, and the RT and RNase H domains of Pol are sufficient for this effect (73). STING has been shown to be associated with mitochondrion-associated membranes (MAM) (74). Since our findings show that the TP domain contains the MTS, it still remains to be seen if localization of Pol to the mitochondria plays a role in the reported interaction between Pol and STING (73). HBx, a regulatory protein specific to mammalian hepadnaviruses, has also been detected at mitochondria in some studies (75,–78). HBx has been linked to both pro- and antiapoptotic roles, mitochondrial membrane potential modulation, generation of oxidative stress, and downregulation of innate immune responses (79). One possibility is that Pol and HBx work in concert at the mitochondria to modulate host physiology. However, the fact that Pol's mitochondrial localization is conserved in nonmammalian HBV species that lack HBx suggests a broader role.
We thank Yongna Xing at the McArdle Laboratory for Cancer Research for the use of her Zeiss Axio Observer Z1 wide-field epifluorescence microscope.
This work was supported primarily by NIH grant P01 CA022443. In addition, we thank the University of Wisconsin Carbone Cancer Center (UWCCC) for NIH/NCI P30 CA014520-UW Comprehensive Cancer Center Support funds to complete this project.