Live Cell Imaging Reveals HBV Capsid Translocation from the Nucleus To the Cytoplasm Enabled by Cell Division

ABSTRACT Hepatitis B virus (HBV) capsid assembly is traditionally thought to occur predominantly in the cytoplasm, where the virus gains access to the virion egress pathway. To better define sites of HBV capsid assembly, we carried out single cell imaging of HBV Core protein (Cp) subcellular trafficking over time under conditions supporting genome packaging and reverse transcription in Huh7 hepatocellular carcinoma cells. Time-course analyses including live cell imaging of fluorescently tagged Cp derivatives showed Cp to accumulate in the nucleus at early time points (~24 h), followed by a marked re-distribution to the cytoplasm at 48 to 72 h. Nucleus-associated Cp was confirmed to be capsid and/or high-order assemblages using a novel dual label immunofluorescence strategy. Nuclear-to-cytoplasmic re-localization of Cp occurred predominantly during nuclear envelope breakdown in conjunction with cell division, followed by strong cytoplasmic retention of Cp. Blocking cell division resulted in strong nuclear entrapment of high-order assemblages. A Cp mutant, Cp-V124W, predicted to exhibit enhanced assembly kinetics, also first trafficked to the nucleus to accumulate at nucleoli, consistent with the hypothesis that Cp’s transit to the nucleus is a strong and constitutive process. Taken together, these results provide support for the nucleus as an early-stage site of HBV capsid assembly, and provide the first dynamic evidence of cytoplasmic retention after cell division as a mechanism underpinning capsid nucleus-to-cytoplasm relocalization.


RESULTS
HBV Cp accumulates preferentially in the nucleus at early time points postexpression. The productive phase of the HBV viral life cycle is initiated in the nucleus by host-mediated transcription of pgRNA that serves as both the mRNA encoding Cp and P, as well as the RNA genome substrate packaged into assembling capsids. To better define subcellular sites of HBV assembly, our first objective was to carry out single cell time course analyses of HBV Cp localization over time, comparing wild-type (WT) Cp trafficking to that of a well characterized Cp mutant (Cp-Y132A) rendered assembly-incompetent, due to removal of the phenolic side chain provided by tyrosine-132 needed by Cp to form interdimer interactions (Fig. 1A) (32).
To this end, we first transfected Huh7 liver carcinoma cells to express either the LJ144 HBV genetic clone encoding wild-type assembly-competent Cp (hereafter referred to as "WT Cp") or the J305 clone encoding Cp-Y132A co-expressed with LJ145 that encodes a Cp-minus packageable genome substrate (17,33). LJ144 expresses Cp and P from pgRNAs competent for packaging and reverse transcription, but lacks viral glycoprotein expression, validated in prior work to serve as a reliable model for studying nucleocapsid formation and viral genome replication (rcDNA synthesis) (16,34). Cells were fixed at 24, 48, or 72 h post-expression, and processed for indirect immunofluorescence (IF), detecting Cp using HBV Core Protein Trafficking mBio an anti-HBc polyclonal antibody that detects Cp's c/e1 immunodominant epitope (35) (Fig. 1B). Huh7 cells were chosen for these experiments because they have been used extensively to study HBV, and work well for both fixed and live cell imaging, being relatively flat cells that are stationary when plated on glass. Interestingly, we observed WT Cp to be predominantly localized to the nucleus at the earlier (24 and 48 h) time points, only accumulating to relatively high levels in the cytoplasm at 72 h (Fig. 1C, n = 50). In contrast, the Cp-Y132A mutant was predominantly detected in the nucleus independent of time. Consistent with capsid assembly events, WT Cp formed distinct puncta in both nuclear and cytoplasmic compartments (Fig. 1C, top panel), but with no puncta formation observed for Cp-Y132A (Fig. 1C, bottom panel). Interestingly, both nuclear WT Cp and Cp-Y132A were enriched in association with subnuclear granules (most predominantly in early periods of gene expression (Fig. 1C, 24h and 48h time points) resembling nucleoli. These results demonstrated that Cp assembly-competency is a key determinant of Cp subcellular distribution, and suggested an early bias to accumulating preferentially in the nucleus. Cp forms high-order structures consistent with capsids in the nucleus. Because assembly-competent Cp formed puncta in the nucleus at the earliest time points, we next sought to determine if nuclear Cp was forming high-order structures consistent with capsids. To this end, we developed a sequential co-labeling IF assay, wherein we first stained with a capsid-specific monoclonal antibody (mAb3120) prior to staining with the anti-HBc polyclonal antibody ( Fig. 1) that was unable to distinguish between capsids or Cp monomers/dimers (with differential targets depicted in Fig. 2A). The selectivity of mAb3120 is due to its binding to a discontinuous epitope that straddles the interface between 2 adjacent Cp dimer units (36)(37)(38). We confirmed this specificity by demonstrating that mAb3120 detected WT Cp under assembly-competent conditions but not the assembly-defective Cp-Y132A mutant protein (Fig. 2B).
Sequential co-labeling of WT Cp-expressing Huh7 cells over our 72 h time course revealed similar levels of WT Cp detected in the nucleus by both antibodies at the 24 h posttransfection time point, suggesting that the nucleus represents a site competent for formation of capsids and/or alternative high-order assemblages. Similar to our initial analysis using the polyclonal antibody in Fig. 1, we also observed a clear shift in the mAb3120 signal from the nucleus to cytoplasm over time (65% N # C by 48 h, and 92%N # C by 72 h) ( Fig. 2C and D), consistent with relocalization of capsids formed in the nucleus being relocated to the cytoplasm. Unassembled Cp, detected by the polyclonal but not mAb3210, was more likely to be detected in the nucleus at all time points, suggesting that Cp monomers/dimers are constitutively trafficked to the nucleus. Because our initial conditions lacked viral glycoprotein expression, we also evaluated Cp distribution using our dual antibody approach for cells expressing TMA153a LJ144 equivalent viral clone that also expresses HBV envelope proteins. Sequential IF detection of Cp/high-order assemblages using mAb3120 and anti-HBc polyclonal antibodies demonstrated similar nuclear-to-cytoplasm trafficking pattern of Cp/capsids (70% N # C by 48 h, and 88%N # C by 72 h) (Fig. 2E), compared to minus envelope expression conditions. These data further indicated that it is the high-order assemblage of capsid components that underpins the major transition from the nucleus to the cytoplasm over time.
To provide confidence that nucleus-associated WT Cp assemblages were indeed capsids, we compared the size and circularity of Cp nuclear signals in cells 24 h posttransfection to those of isolated, gradient-purified HBV capsids (Fig. 3). Cells expressing WT Cp and isolated capsids were subjected to dual antibody IF, and background subtracted prior to bright objects being subjected to multivariate image analysis to determine shape (circularity) and size (pixel number and maximum diameter) (Fig. 3A). As expected, isolated capsids detected using mAb3120 yielded circular puncta of uniform size that were consistent in size and shape with cell-associated WT Cp structures identified in either the nucleus or the cytoplasm (Fig. 3A, right panels, and Fig. 3B). It is important to note, however, that these assays were unable to differentiate between nuclear capsids carrying packaged and reverse transcribed pgRNA, representing replication-competent nucleocapsids, or empty capsids.
Delivery of nuclear HBV capsids to the cytoplasm occurs predominantly during nuclear membrane breakdown in conjunction with cell division. Having obtained evidence that capsids assemble in the nucleus prior to the bulk of the signal being relocalized to the cytoplasm in a time-dependent manner (over 72 h), we considered 3 competing hypotheses that might explain the nucleus-to-cytoplasm transition: (i) that nuclear capsids/high-order assemblages formed early were degraded at later time points relative to the cytoplasmic population; (ii) that capsids/high-order assemblages formed in the nucleus transitioned either gradually or en masse from the nucleus to with the indicated antibodies, to confirm binding specificity. mCherry (inset, red) was co-transfected with the Cp variant to assist in identifying transfected cells prior to fixation, staining with DAPI (blue), and staining for IF (green). mAb3120 was confirmed as unable to detect Cp-Y132A (central panels). (C) Images and analysis of dually labeled Huh7 cells expressing WT Cp and detected using mAb3120 (cyan) and polyclonal anti-HBc (green). Transect analysis illustrates differential detection of Cp with polyclonal (predominantly nuclear) and mAb3120 (predominantly cytoplasmic) at 72h in a representative cell; consistent with unassembled Cp trafficking to the nucleus, even when the assembled capsid population is predominantly in the cytoplasm. Scale bars represent 10 mm. (D) Bar graphs quantifying the subcellular localization of WT Cp for 100 cells per condition, detected using the indicated antibody and corresponding to the experiment in (C). Error bars represent the standard deviation of the mean for 3 biological replicates. (E) Bar graphs as for (D) confirming a similar subcellular localization of WT Cp distribution over time when expressed from a plasmid encoding envelope glycoproteins (Env1). the cytoplasm through nuclear pore complexes (NPCs) as recently proposed (19)(20)(21); or (iii) that capsids/high-order assemblages left the nucleus during nuclear membrane dissolution as a consequence of cell division, consistent with prior reports from Yeh, C-T et al. and Guidotti, LG et al. (19,39).
Because our understanding of Cp transport had been limited to fixed cell approaches that are unable to provide resolution of live cellular dynamics, we sought to engineer a reliable long-term (.24 h) live cell HBV imaging strategy based on fluorescently labeled Cp derivatives, again with the goal of comparing changes to the subcellular distribution of assembly-competent WT Cp to that of the assembly-incompetent Cp-Y132A control. A study from Nassal and colleagues had demonstrated previously that insertion of a 238 aa. green fluorescent protein (GFP) into the tip of the prominent Cp dimer surface spike (found between aa. 78 and 79 of Cp) can be tolerated by HBV capsids when co-expressed with untagged Cp to form Cp/Cp-GFP chimeric fluorescent particles (40). Informed by this insight, we generated analogous WT and Cp-Y132A variants carrying NeonGreen (Cp-NG and Cp-Y132A-NG) (Fig. 4A), choosing NG over GFP because it is a brighter fluorophore that also has better protein folding kinetics (41). Validation experiments confirmed that Cp-NG and unlabeled Cp, co-expressed at a 1:1 ratio, yielded Cp assemblages that co-sedimented (at fractions 7 and 8) consistent with Cp-NG being incorporated successfully into chimeric capsids (Fig. 4B). Additionally, Southern blot detection of newly synthesized HBV DNA confirmed that co-expression of Cp-NG and untagged Cp had little to no impact on HBV DNA synthesis when compared to the DNA synthesis supported by untagged WT Cp expression (positive control) (Fig. 4C). These data confirmed that coexpression of Cp-NG and untagged Cp yielded a system that both formed chimeric capsids and supported HBV genome replication, but did not allow us to determine the percentage of Cp-NG chimeric capsids competent for genome packaging and reverse transcription.
We next proceeded with live cell imaging of single Huh7 cells expressing either Cp-NG or Cp-Y132A-NG, monitored for up to 70 h post-expression. Consistent with our IF experiments ( Fig. 1 and 2), WT Cp-NG was initially observed to accumulate in the nucleus at earlier time points (up to 36 h), prior to shifting to the predominantly cytoplasmic distribution at a later time point; unlike the Cp-Y132A-NG control that did not exhibit major differences to nuclear or cytoplasmic abundances (Fig. 4D). Strikingly, our video microscopy also revealed that WT Cp-NG's re-localization to cytoplasm was not a gradual process, as would be expected if capsids were degraded in the nucleus over time or being translocated to the cytoplasm through nuclear pores (i.e., hypotheses i or ii), but instead was observed predominantly in conjunction with nuclear membrane breakdown during cell division (most consistent with hypothesis iii) (Fig. 4D, upper panel, and Movie S1). Moreover, after nuclear membrane reformation, the majority of Cp-NG remained in the cytoplasm suggestive of a capsid retention mechanism, with Cp-NG only gradually beginning to build up in the nucleus again over time (Fig. 4D, upper panel). In contrast, the Cp-Y132A-NG mutant immediately was observed to re-accumulate in the nucleus after cell division (Fig. 4D, lower panel, and Movie S2). These videos demonstrated that, under these conditions, nucleocytoplasmic re-distribution of Cp/capsids was principally facilitated by nuclear membrane breakdown during cell division, and indicated that Cp/capsids were preferentially retained in the cytoplasm after cell division in an assembly-dependent manner.
Blocking cell proliferation with aphidicolin results in nuclear entrapment of HBV capsids. We hypothesized that if nuclear envelope breakdown was capsid's predominant route of nuclear escape, then nuclear-to-cytoplasmic re-distribution would be blocked when cell proliferation was arrested. To test this hypothesis, we chose to inhibit cell division using aphidicolin (APC)an inhibitor of host DNA polymerase activity, that blocks the cell cycle in early S phase and prior to nuclear membrane breakdown (42,43). WT Cp was expressed in the absence or presence of 10 mg/mL APC by dual label IF over a 72 h time course, identical to the approach described for Fig. 2 (Fig. 5). In comparison to normal cell proliferation conditions, wherein capsids/high-order assemblages were relocalized from the nucleus to the cytoplasm, cells treated with 10 mg/mL APC exhibited a marked retention of capsids in the nucleus (.70% N . C) at all time points (Fig. 5). These results were consistent with a model wherein HBV capsids formed in the nucleus at early time points require cell division to achieve redistribution to the cytoplasm. A Cp mutant (V124W) that exhibits rapid assembly kinetics also accumulates in the nucleus at early time points. Because we observed that WT capsids/high-order assemblages were predominantly retained in the cytoplasm after cell division (Fig. 2C), we hypothesized that increasing rates of HBV capsid assembly would bias Cp to accumulate in the cytoplasm early (i.e., prior to cell division) should assembly rates outpace the rate of Cp monomer/dimer nuclear import. To test this hypothesis, we tracked a Cp mutant, Cp-V124W that carries a point-mutation at amino acid 124 within the interdimer region of Cp that kinetically favors protein-protein interactions, and has been shown to increase rates of capsid assembly in vitro. (44) Sequential IF staining of untagged Cp-V124W (p1195) showed that Cp-V124W was, similar to WT Cp, predominantly localized to the nucleus at 24 h and showed a dramatic shift to a cytoplasmic distribution (71% N , C) at 72 h post-expression (Fig. 6A, mAb3120). Unlike for WT Cp, however, the anti-HBc polyclonal antibody showed a distribution pattern similar to mAb3120 for all conditions, consistent with rapid assembly of Cp-V124W, and it being associated with capsids and/or high-order assemblages at all time points.
Live cell imaging using a fluorescently tagged Cp-V124W-NG derivative showed, similar to WT Cp-NG, rapid early accumulation in the nucleus but, unlike WT Cp-NG, an even more dramatic localization to distinct nuclear bodies that we confirmed to be nucleoli using an anti-Nucleophosmin (B23) antibody (Fig. 6C). During cell division, nucleoli-associated Cp-V124W-NG relocalized to the cytoplasm, and then the structures dissipated, followed by a gradual return of Cp-V124W-NG to the nucleus and nuclear bodies over time ( Fig. 6B and Movie S3). Taken together, our analysis of the Cp-V124W rapid assembly mutant suggested that Cp's trafficking to and accumulation within the nucleus is a fast and constitutive process and may be linked to nucleolar biology as previously proposed (24,(45)(46)(47).

DISCUSSION
In this study, we applied both single dual-antibody immunofluorescence and live cell imaging to study the intracellular distribution of HBV Cp in Huh7 hepatocellular carcinoma cells, revealing that Cp distribution changes markedly over time, dependent on Cp's ability to assemble capsids and the state of the cell cycle (see summary model in Fig. 7). Assembly-competent Cp localized to the nucleus at early time points and was shown to be able to assemble into higher-order assemblages consistent with capsids within this compartment (Fig. 2 to Fig. 4 and Fig. 7, left panel). Live cell imaging revealed that WT Cp-NG accumulated in the nucleus at early time points, and was primarily relocalized to the cytoplasm in conjunction with nuclear envelope breakdown during cell division ( Fig. 4D and Fig. 7, right panel, and Movies S1 and S2). It is important to consider that cellular levels of core protein may also be a contributing factor in influencing the intracellular distribution of Cp, as demonstrated recently by Nair and Zlotnick (24) and, in this context, it is notable that we observed Cp-NG to accumulate in the nucleus even at the lowest levels of detectable expression (Fig. 4D and Movie S1). We also showed that blocking cell division with aphidicolin (APC), a drug that causes cell cycle arrest at early S phase, resulted in the entrapment of capsids/high-order assemblages in the nucleus (Fig. 5) and a rapid-assembling Cp mutant, Cp-V124W, preferentially trafficked to nucleolar compartments at early time points (Fig. 6 and Movie S3), consistent with Cp localization to the nucleus being a strong and constitutive process occurring soon after protein synthesis, followed by nuclear assembly and relocalization of capsids/high-order assemblages to the cytoplasm where they are retained after cell division.
Our conventional understanding of the HBV life cycle describes capsid assembly taking place primarily in the cytoplasmwhere envelopment and viral egress also occur. Our work and that of others (19,22,48) now presents examples of Cp localized to the nucleus at early time points. But why? Notably, Cp harbors a nuclear localization signal (NLS) that spans the CTD (149 to 183 aa.), and is needed to direct nucleocapsids to the nucleus to initiate infection, regulated by phosphorylation (49)(50)(51)(52). This same NLS is likely responsible for Cp's rapid localization to the nucleus in monomer or dimer configurations soon after translation and prior to capsid assembly. Our findings also strongly suggest that the nucleus can represent a site of capsid assembly, regardless of the presence or absence of glycoproteins ( Fig. 2 and 3). Importantly, a prior in vivo study form Guidotti, et al. also showed an early detection of Cp in the nucleus by immunohistochemical analysis of transgenic mouse hepatocytes and, consistent with our results demonstrating mitosis-dependent relocalization to the cytoplasm (19). Moreover, a study by Rabe and Kann indicated that capsids entering the nucleus disassemble and then can reassemble into capsids in the nucleus, further demonstrating that Cp subunits in the nucleus can assemble into capsids or high-order assemblages in this compartment (48). Although our study was not designed to distinguish visually between empty capsids and genome-containing nucleocapsids, we note that Yang et al. recently also detected capsids in the nucleus of HBV-expressing cells and provided evidence for nuclear capsid association with pgRNA, implicating the nucleus as a potential site for viral genome packaging (26). Future studies of nuclear Cp transport dynamics will be needed to determine if nuclear transport is obligatory to assembly or genome packaging or, alternatively, more so a consequence of the protein encoding a strong NLS.
We also observed WT Cp and Cp-V124W to localize preferentially to nucleolar compartments with unclear implications, but consistent with a subset of prior studies (24,53). Cp-V124W's localization to the nucleolus was particularly striking (Fig. 6), and we found it remarkable that this mutant and, likely, capsids were completely dissociated from these compartments during cell division ( Fig. 6B and Movie S3). The nucleolus is the site of rRNA synthesis and processing during ribosome biogenesis (54,55), and in theory could represent an RNA-rich environment favorable for HBV Cp binding in its capacity as an RNA-binding protein. Alternatively, the nucleolus could represent an organelle where Cp associates with host factors that promote assembly or help to coordinate pgRNA packaging. For example, Nucleophosmin (NPM1), also known as B23, a core component of the nucleolus implicated in mRNA transport, chromatin remodeling, and apoptosis, has also been shown to interact with HBV Cp and to promote capsid in vitro, possibly serving as an assembly priming factor (56).
If Cp's accumulation in the nucleus (and/or in association with nucleoli) is constitutive, then how does it get out of the nucleus to reach virion egress sites in the cytoplasm? Our live cell studies reveal, for the first time, that large quantities of Cp-NG can be liberated from the nucleus during cell division and, interestingly, that these capsids/high-order assemblages end up being strongly retained in the cytoplasm, thereby preventing re-accumulation in the nucleus during anaphase (Fig. 4D). These results suggest that the cell cycle can play a more significant role in the HBV virion assembly pathway than previously appreciated, albeit with the caveat that multiple other strong prior studies have implicated the nuclear pore as a major route for Cp nuclear export (21)(22)(23)(24). The nuclear pore complex (NPC) channel spans a radius of ;2.6 nm in diameter, and small molecules up to 60 kDa can passively diffuse across the NPC (57,58). Despite these size limitations, larger molecules are actively exported by transport proteins including CRM1/exportin 1 (XPO1) (59,60), and accumulating evidence indicates that HBV Cp is regulated by a nuclear export signal (NES) that potentially accesses both CRM1-dependent and -independent (e.g., NXF1-driven) export pathways (24,49). In this context, an important observation from our study that merits follow-up is demonstration that when the cell cycle was blocked using APC, HBV capsids/high-order assemblages remained nuclear at all time points, suggesting that cell division could outweigh the NPC as the major means of Cp nuclear escape, at least under our assay conditions (Fig. 4). Consistent with these findings, a previous study indicated that cell cycle state influences whether HBV Cp is primarily found in the nucleus or cytoplasm; i.e., when the cell cycle was blocked between the G o /G 1 stages nuclear accumulation of Cp was observed (39). We propose that cell division would represent an efficient means by which HBV can leave the nucleus, but emphasize that our study does not rule out potential contributions of the NPC and CRM1 or NXF1/p15-driven nuclear export pathways. Moreover, we again emphasize the important point that our imaging techniques, to date, have been unable to differentiate between genome-loaded nucleocapsids and empty capsids. Empty capsids are likely to represent the majority of cell-associated Cp (8,9), and may traffic differently than nucleocapsids. Accordingly, visually distinguishing between these pools in both fixed and live cell imaging represents a major goal for our future research.
Overall, the relevance of the cell cycle to HBV replication and pathogenesis in vivo remains controversial. A study using liver tissue from patients with chronic HBV infection demonstrated a positive correlation between high Cp levels in the nucleus and high levels of viral DNA detected in patient serum (20). In efforts to elucidate the regulatory functions that core protein could have in the nucleus of infected cells, multi-omic studies reveal that the core protein interactome primarily consists of RNA-binding proteins involved in several aspects of mRNA metabolism, as well as alterations to the metabolic functions of hepatocellular carcinoma cells (61,62). In contrast, other studies have shown a positive correlation between disease severity and high levels of HBV Cp primarily detected in the cytoplasm of hepatocytes (21,63). HBV infection may be capable of dysregulating the cell cycle, affecting the tumorigenic environment by upregulating human oncogenes including telomerase, the Src kinase, and Neuraminidase 1 (30,(64)(65)(66). However, HBV-mediated induction of the cell cycle was, in a separate study, shown to yield reductions to viral gene expression (67) and, in vivo, rapidly proliferating liver cells in an HBV-infected humanized-mouse model created an antiviral environment, where cell proliferation was correlated with an overall decrease to HBV cccDNA levels (31,68). HBV replication was also shown to be partially affected by the cell cycle in an HBV transgenic mouse model of liver regeneration, wherein partial hepatectomy was associated with reduced HBV DNA and core protein levels, but did not affect viral RNA levels (69). However, it is important to note that cytokine levels were increased in this study, so that cytokine-mediated antiviral effects are also likely to be a confounding variable in vivo. Taken together, further studies will be necessary to fully understand differences to the HBV replication cycle in resting and dividing hepatocytes.

MATERIALS AND METHODS
Molecular clones. The HBV molecular clones used in this study (Table 1) were derived from the ayw HBV strain (GenBank accession no. V01460.1). WT Cp was expressed from plasmid pLJ144 that contains the complete HBV genome and derived from the parental clone pCH-9/3091 (17,33,70). The pLJ144 plasmid and all plasmids used in this study express the HBV pgRNA under the transcriptional control of the CMV-IE promoter. Plasmid pLJ144 does not express the envelope proteins due to two mutations: (i) the start codon of the S open reading frame (ORF) was changed from the ATG to ACG (T154C) and (ii) a premature termination codon TAA was introduced into the sixth codon of the S protein ORF (C169A) (71). Plasmid pLJ145 (Cp-minus) differs from pLJ144 due to the addition of a premature stop codon at the fourth codon of the Cp ORF, mutating GAC to TAA (16,17). The pEL43 plasmid is a derivative of parental clone pCH-3143, and is defective for pgRNA packaging due to deletion of the upstream segment of the epsilon (Ep) sequence (72). Plasmid pEL43 is also defective in P protein expression due to introduction of a premature stop codon in the P ORF's thirteenth codon, changing the CTG to TAG (16,34). The pTMA153 plasmid, equivalent to pLJ144, expresses full length HBV pgRNA, Cp, P, and all 3 envelope proteins. The pTMA153 expression profile is the equivalent to TL25, with the exception of containing the Epstein-Barr virus origin of plasmid replication and the Tet operator sequence (13). Plasmid pJ305 encoded Cp-Y132A and eliminates P protein synthesis, due to a substitution to the start codon from ATG to ACG (16). Untagged version of Cp-V124W was expressed from plasmid p1195 plasmid that encoded the Cp V124W amino acid substitution in a pLJ144 backbone (17,33,70).
NeonGreen (NG)-tagged versions of Cp were derived from pEL43 (WT, plasmid pSR007), pFU310 (Cp-Y132A, plasmid pSR009); or p1199 (Cp-V124W, plasmid pSR012), all expressed from a pEL43 backbone. All NG-tagged versions of Cp have been created on Ep-and P-pgRNA expressing plasmid backbones that make analyzing Cp function by trans-complementation assays possible with a "donor" pgRNA expressing plasmid that have been employed by the Loeb lab (70,71). The NG fluorescent protein ORF was inserted between codons encoding Cp amino acids D78 and P79, flanked by glycine-rich flexible linker sequences: GGGSGGG at the NG N-terminus and GGGG at the NG C-terminus. All pSR plasmid variants were generated by PCR. All molecular clones were confirmed by sequencing across the entire restriction fragment insert.
Cell culture and drug treatment. The human liver carcinoma cell line Huh7 (73) was cultured in Dulbecco's modified Eagle essential minimal medium nutrient mixture with Ham's F-12 medium (DMEM-F12; Gibco) supplemented to a final concentration of 10% with heat-inactivated fetal bovine serum ([FBS], Sigma) and 1% penicillin-streptomycin-L-glutamine (PSG) solution (Sigma). To arrest cells at the G1/S transition, cells were incubated in DMEM-F12 supplemented with 10 mg/mL aphidicolin (APC) (Cell Signaling Technology Cat#:32774), with this APC concentration sufficient to block Huh7 cell proliferation with low cytotoxicity confirmed using the Click-it EdU assay (ThermoFisher Cat#:C10640).
Immunofluorescence and image analysis (fixed-cell imaging). All fixed-cell imaging experiments were carried out on cells grown in 24-well glass-bottom plates (VWR). Cells were fixed using 4% paraformaldehyde (PFA, EM Sciences), washed once in phospho-buffered saline (PBS), permeabilized using 0.2% Triton X-100, and blocked with NGB (50 mM NH 4 Cl, 2% goat serum, 2% BSA) buffer. DNA was stained with 5 ng/mL 4',6-diamidino-2-phenylindole (DAPI; ThermoFisher) to visualize nuclei. For global Cp detection, a 1:1000 dilution of a polyclonal rabbit anti-core (anti-HBc) antiserum (Lot#:214-14, Austral Biologicals) was used prior to washing with PBS and staining with 1:1000 diluted secondary antibodies (goat anti-rabbit 488; Life Technologies). For differential Cp/capsid labeling, cells were first incubated with a 1:600 dilution of mouse anti-capsid antibody mAb3120 (Cat#: 2ZHC22, Cosmo Bio Co., LTD.) followed by secondary antibody (goat anti-mouse 647; Life Technologies) prior to incubation with the anti-HBc polyclonal and goat anti-rabbit 488 antisera. Images were captured using either 20Â (NA 0.75) air or 100Â (NA 1.45) oil-immersion objective lenses on a Nikon Ti-Eclipse inverted widefield epi-fluorescent deconvolution microscope (Nikon Corporation) equipped with an Orca-Flash 4.0 C11440 (Hamamatsu) camera, and with data collected using Nikon NIS Elements software (V 4.00.03). Images were processed and analyzed using FIJI/ ImageJ image analysis software (74,75), normalizing signal intensity to mock (no HBV) images to control for background. The relative subcellular distribution of Cp/capsid signal was determined for .100 cells per time point (24, 48, or 72 h), with Cp-expressing cells selected at random, and classified for nuclear (N) versus cytoplasmic (C) mean fluorescence intensity (MFI) being N . C (greater nuclear distribution), N , C (greater cytoplasmic distribution), or N = C (equal nuclear and cytoplasmic distributions).
For isolated capsid imaging, 100 mL of fraction #7 from cell lysates expressing WT Cp isolated and subjected to velocity sedimentation (see below) was plated directly onto a poly-d-lysine coated cover slip (Neuvitro Corporation; Cat#:GG-12-PDL), and left undisturbed at room temperature for 2 h. After 2 h, Puncta analysis with PyImageJ and KNIME. The puncta detection and segmentation workflow were first prototyped with the ImageJ Ops image processing framework (76) and PyImageJ (77), a Python library that wraps ImageJ's Java code base and enables access to ImageJ from the Python environment. The workflow was later ported to the Konstanz Information Miner (KNIME) (78) to take advantage of KNIME's intuitive workflow graphical user interface (GUI). Importantly the KNIME Image Processing extension (79) supports ImageJ and the ImageJ Ops framework. Briefly, dually labeled (mAb3120 and polyclonal antibody) are processed with a Gaussian convolution (sigma = 5), applied an Otsu threshold, objects are then labeled with connected component analysis (connection type = eight), label size filtered (min = 10 pixels, max = 2,000,000 pixels), and then measured for the desired statistics and geometry (e.g., circularity, diameter, intensity, etc.). The puncta analysis workflow is available on GitHub (https://github.com/shererlab/ puncta-segmentation/).
Live cell imaging and analysis. All time-lapse live cell imaging experiments were performed in 8well no.1.5H glass-bottom slides (1 cm 2 Cat#: 80826, ibidi) at 37°C, ;50% humidity, and 5% CO 2 housed in a Pathology Devices Live Cell stage top incubator (Pathology Devices, Inc). Cells were seeded and transfected as described above. After replacing transfection mix with fresh medium, imaging was carried out for 72 to 96 h starting at 4 h-posttransfection, with images acquired every 30 min. as previously described (80,81). To quantify Cp-NG signals in the nuclear versus cytoplasmic compartments for single cell analyses, MFI was measured from nuclear or cytoplasmic regions of interest (ROIs) consisting of a fixed circular area (;50 mm 2 ).
Cp and Cp-NG were detected by Western blot analysis from cytoplasmic lysates using monoclonal antibody 19C1-8 (Tokyo Future Style, Inc.) (82). For velocity sedimentation, 200 mL of the NP-40 lysate (out of 450 mL of total) was gently added to a 4-layer sucrose gradient (975 mL each: 15, 30, 45, 60%) prior to centrifugation at 52,000 rpm for 2.5 h in a Beckman SW60 rotor. Twelve fractions of equal volume (365 mL each) were then collected from top to bottom, and concentrated 10-fold (150 mL original fraction to 15 mL final volume) using methanol/chloroform-based protein precipitation. Briefly, 600 mL methanol was added per 150 mL fraction and mixed prior to adding 150 mL chloroform, vortexing, addition of 450 mL water, and centrifuging at .20,000 Â g for 5 min. Subsequently, the upper aqueous layer was discarded prior to adding 650 mL methanol, inverting tubes 3 times, and centrifuging again at .20,000 Â g for 5 min. Finally, protein pellets were dried and resuspended in 7.5 mL NP-40 lysis buffer prior to the addition of 7.5 mL 2Â Laemmli sample buffer.
Southern blot analysis. To measure rcDNA synthesis, cells were washed with 1Â PBS then lysed with 0.2% NP-40, 50 mM Tris-HCl, and 1 mM EDTA (pH 8.0) for 10 min in room temperature. The nuclei were isolated by centrifugation at 16,000 Â g for 5 min at 4°C. Supernatants were adjusted to 2 mM CaCl 2 , and then incubated with 44 units of micrococcal nuclease to digest transfected plasmid DNA and unencapsidated HBV RNA. After 2 h, these reaction mixtures were supplemented with EDTA to 10 mM, sodium dodecyl sulfate (SDS) to 0.4%, and Pronase to 400 mg/mL. These additions terminated the micrococcal nuclease digestion and digested nucleocapsids and the P. Viral DNA was then extracted using phenol-chloroform mixture and precipitated with 17 mL 5 M NaCl, 20 mg Glycogen, and 1 mL of a 100% ethanol (83,84) prior to electrophoresis using an 11-cm Â 14-cm Â 0.5-cm 1.25% agarose gel in 1Â Trisborate-EDTA buffer at 24 V for 24 h. DNA was then transferred by capillary action to a Hybond-N1 nylon membrane (Amersham Pharmacia Biotech), and cross-linked to the membrane with UV light.

ACKNOWLEDGMENTS
We are very grateful to Abena Kissi-Twum and other Loeb lab members (UW-Madison Dept. of Oncology) for protocols, reagents, and essential advice. We also thank Paul This study was supported by National Institutes of Health grants P01CA022443 to D.D.L. and N.M.S., R01AI110221 to N.M.S., P41GM135019 to K.W.E., and T32AI078985 to S.R.; and funding from a Morgridge Fellowship to E.L.E. and a SciMed/GRS Advanced Opportunity Fellowship (AOF) to S.R. AOF fellowship support was provided by the UW-Madison Graduate School, part of the Office of Vice Chancellor for Research and Graduate Education, with funding from the Wisconsin Alumni Research Foundation and the UW-Madison.