Imaging core-LHBs interaction by FLIM.
Recently, we, among others, documented the intracellular co-localization of LHBs and core proteins in HBV-permissive cells and concluded that LHBs would serve as a platform for HBV virion assembly by recruiting core and SHBs proteins [35, 41, 42]. Here we used a FRET assay to visualize in cell the core-LHBs interaction when these two partners are co-expressed using a set of tagged proteins: core-eGFP, eGFP-LHBs or mCherry-LHBs as a tracer (Fig. 1A). Insertion of eGFP between residues 78 and 80 of the core protein was shown permissive for capsid assembly [36, 43]. As shown in Fig. 1B, the results of a western blot analysis revealed that each chimeric protein was expressed at the expected molecular weight for full length chimeric protein. Note that both eGFP-LHBs and mCherry-LHBs lack the N-terminal myristate, a post-translational modification essential for viral entry but dispensable for HBV morphogenesis [44, 45].
The eGFP-mCherry fusion protein was used here as a FRET-positive control (not shown) [46, 47]. FRET between eGFP and mCherry occurs only when the two proteins are less than 8 nm apart, a distance assumed to indicate a direct protein-protein interaction [48, 49]. FRET is measured by calculating the lifetime (τ) of eGFP (FLIM), an intrinsic parameter of the chromophore, which is independent of its concentration. However, the eGFP lifetime value is highly sensitive to its environment and decreases during FRET. It is calculated pixel-by-pixel on a cell expressing the eGFP partner only, and it is compared to the lifetime calculated on a cell expressing both eGFP and mCherry partners. An image is then reconstructed using an arbitrary color scale so that the color differences provide a spatial representation of the interaction between the two partners.
Figure 2 shows representative images of cells expressing core/core-eGFP + mCherry (upper row) or core/core-eGFP + mCherry-LHBs (lower row). Images were obtained by confocal microscopy (panel A), and by lifetime measurements (panel B). In cell co-expressing core and core-eGFP, together with mCherry, core proteins accumulate in the nucleus (Fig. 2a) whereas mCherry diffuses into the entire cell (Fig. 2b). This led to a yellow staining in the nucleus when the channels were superimposed (Fig. 2c). By FLIM, the nuclear yellow staining corresponds to a τ value egal to 1.90 ± 0.1ns (orange arrow on the arbitrary color scale), equivalent to that of free core-eGFP, since eGFP does not interact with mCherry. It should be noted that the eGFP lifetime is usually at 2.3–2.4 ns [50], however we previously observed that core-eGFP had a shorter-than-expected lifetime, probably as a result to its involvement into capsid assembly [36].
Huh7 cells were transfected with a mix containing 0.35 µg of core-expressing plasmid and 0.15 µg of core-eGFP-expressing plasmid (70/30) and 0.5 µg of plasmid expressing either mCherry (row 1, images a-d) or mCherry-LHBs (row 2, images e-h). 72h. post transfection, Williams medium was substituted for Leibovitz’s L-15 medium supplemented with 5% fetal bovine serum and placed at 37°C in thermostatic chamber. Cells were observed simultaneously by confocal microscopy (panel A) and FLIM (panel B). The fluorescence lifetime τ was measured pixel per pixel and converted into color using a color scale ranging from orange (0.7 ns = short fluorescence lifetime) to red (2.5 ns = long fluorescence lifetime). Each image is representative of 10 cells/experiment repeated three times. A scale bar in white is reported image a and represents 10 µm. The white dots delineate the cell nucleus.
In cells co-expressing core + core-eGFP + mCherry-LHBs (Fig. 2, second row), core proteins accumulated at the nuclear periphery (Fig. 2e), and co-localized with large clusters of LHBs (Fig. 2f and 2g). An LHBs-mediated delocalization of core at the rim of the nucleus was previously reported [35, 41, 51]. The average core-eGFP lifetime for the whole cell was 1.65 ± 0.07ns (Fig. 2h) for a FRET efficiency of 14.69 ± 3.4%. Interestingly, the spatial distribution of this lifetime was heterogeneous throughout the cell. A short lifetime (τ = 1.3–1.4 ns, turquoise arrow) localized in the cytoplasm at the periphery of the nucleus, whereas a longer lifetime (τ = 1.8–1.9 ns, yellow arrow) localized within the nucleus (yellowish color). This observation is in support of a direct interaction between the core and LHBs, in the cytoplasm, at the periphery of the nucleus, while the free core protein remained localized within the nucleus. Other examples that highlight this heterogeneous distribution are shown in Fig S1.
To further investigate an intracellular interaction between LHBs and core by FRET, we substituted i) envelopment deficient mCherry-SHBs or mCherry-L-ΔMD-HBs [28] to mCherry-LHBs, and ii) a capsid assembly deficient Y132A-core to wt-core [52, 53]. When transfected cells were observed by confocal microscopy (not shown) and FLIM, nuclei appeared as yellow corresponding to long lifetime (Fig. 3A, images b, c and d) and, consequently, to a low FRET efficiency at around 5% (Fig. 3B). These results are in support of i) SHBs not interacting with core in absence of LHBs [35] and ii) the MD domain of LHBs being required for interacting with core [28]. The lack of FLIM observed when Y132A-core was substituted to wt-core strongly indicate that LHBs interact with core in a preassembled capsid. Note that the deletion of MD has no impact on LHBs oligomerization [54] (Fig S2) and that capsid-assembly deficient Y132A-core poorly oligomerizes (Fig S3) [36, 52, 53]. Our results also show that co expression of the eGFP- or mCherry-fusion proteins with their wt counterpart does not prevent interaction between capsid and LHBs.
HBV capsids and LHBs co-localize in large clusters cells.
An immuno-transmission electron microscopy (TEM) approach was used for a better resolution of the core-LHBs interaction. TEM can reach nanometric resolution that allows for the observation of cell ultrastructure and viral particles. In this approach cells were transfected with DNA vectors coding for i) core/core-GFP (90/10) (Fig. 4a), ii) LHBs/eGFP-LHBs (95/5) (Fig. 4b), and iii) core/LHBs/eGFP-LHBs (50/45/5) (Fig. 4c). Post-transfection, cells were sorted by FACS for eGFP expression, and selected cells were then treated for immuno-TEM using anti-HBc or anti-preS1 antibodies.
In cells expressing core only, we observed that core-specific 6 nm-gold beads surrounded spherical and electron dense material corresponding to native HBV capsid (Fig. 4a, yellow arrows). The distance observed between capsid and the gold beads was compatible to the size of the primary-secondary antibody pair. Note that capsids (identified by their apparent icosahedral shape) were not always decorated with gold beads. We interpret this as a lack of antibody accessibility, as is often the case with this type of technique.
In agreement with the fluorescent microscopy images (Figs. 2 and 3), capsids were mainly observed in cell nucleus (Fig. 4a), but occasionally in the cytoplasm (Fig S4). But whatever their intracellular localization, capsids were always evenly dispersed and distanced from each other. Interestingly, we could not document any proximity between capsids and cell membranes. This implies that, unlike assembly of HIV Gag protein [55], HBV core assembly into capsid would not depend on membrane association.
Huh7 were transiently transfected by a mixture of DNA plasmids expressing core and core-eGFP (90/10) (a), LHBs and eGFP-LHBs (95/5) (b) and core, LHBs and eGFP-LHBs (50/45/5) (c) for 72h. Cells were sorted by monitoring eGFP and fixed prior to gelatin inclusion. Ultrathin sections were cut (80 nm) and incubated with human anti-HBc (images a and c), rabbit anti-PreS1 (images b and c) and then with a 6-nm gold-conjugated anti-human antibody (image a) or a 10-nm gold-conjugated anti-human antibody (image c) or a 6-nm gold conjugated anti-rabbit antibody (images b and c). Scale is reported at the bottom left. Some capsids are presented with yellow arrows, and membrane rich domains with white arrows. Nucleus are delimited by a blue line.
Using anti-preS1 primary antibody and secondary antibody bound to 6 nm bead for immuno-TEM of cells expressing LHBs and eGFP-LHBs, the latter were always detected in large clusters, exclusively in the cytoplasm at the periphery of the nucleus (Fig. 4b and enlargement). Furthermore, clusters of LHBs proteins co-localized with convoluted membranes (designed by white arrows in Fig. 4b).
When immuno-TEM was performed on cells expressing core + LHBs + eGFP-LHBs, capsids were identified by their apparent icosahedral structure and by anti-core specific followed by anti-human bound to 10 nm beads, while envelope proteins were detected with anti-PreS1 and a 6 nm beads-labeled secondary antibodies (Fig. 4c). Numerous capsids (yellow arrows) appeared as clusters, whereas LHBs remained associated to membrane rich region (white arrows). Our interpretation of the data is that the encounter between envelope proteins and capsid is mediated by membrane associated LHBs. As shown in Fig. 4a, capsids remain individually dispersed in the absence of LHBs, whereas in its presence, they cluster with LHBs (Fig. 4c).
To corroborate that the observed capsids were equally associated with LHBs and eGFP-LHBs, we performed immuno-TEM only with an anti-GFP antibody and a secondary antibody bound to 6 nm beads, capsid being clearly identified by their icosahedral structure. As shown in Fig. 5, capsid clusters (yellow arrows) associate with LHBs (gold beads) in a convoluted membrane environment (white arrows) close to the nucleus. All together, these data are in support of LHBs condensing capsids at intracellular membranes.
Huh7 cells transfected with the mix of plasmids expressing core, L-HBs and eGFP-L-HBs (50/45/5) were sorted by FACS, centrifuged and the pellet treated for observation by immuno-TEM with an anti-eGFP antibody and secondary anti-rabbit antibody labelled with 6 nm gold. This electron micrograph shows that the LHBs protein detected by an anti-eGFP antibody condenses numerous HBV capsids identified by their shape (yellow arrows) in a membrane-rich region (white arrows) at the periphery of the nucleus delimited by a blue line.
In support of this interpretation, we showed that L-ΔMD-HBs protein could also cluster (Fig. 6a) in a membrane-rich region (white arrows) close to the nucleus but, as expected, co-expressed capsids remained dispersed within the nucleus (yellow arrow). This agrees with the absence of FRET efficiency reported above, and it confirms the essential role of LHBs MD in LHBs-induced capsid recruitment and virion assembly [32, 56, 57].
Huh7 were transiently transfected by a mixture of DNA plasmid expressing core, L-ΔMD-HBs and eGFP-L-ΔMD-HBs (50/45/5) (a) or Y132A-core, LHBs and eGFP-LHBs (50/45/5) (b) for 72h. Cells were sorted by monitoring eGFP and fixed prior to gelatin inclusion. Ultrathin sections were cut (80 nm) and incubated first with human anti-HBc and rabbit anti-PreS1 and then with a 6-nm gold-conjugated anti-human antibody and with a 10-nm gold-conjugated anti-rabbit antibody. Scale is reported at the bottom left. Capsid found in nucleus is indicated with yellow arrow, core proteins in the nucleus are indicated in dashed yellow arrows while L-ΔMD-HBs proteins (a) or LHBs proteins (b), observed in membrane rich domains are indicated with white arrows. Nucleus and cytoplasm are delimited by a blue line.
When cells co-expressing wt-LHBs and Y132A-core proteins were subjected to immuno TEM, the LHBs-specific 10 nm beads clustered in the cytoplasm (Fig. 6b, white arrows), but core-specific 6 nm beads remained dispersed in the nucleus in the absence of assembled capsids (yellow dotted arrow) with no evidence of LHBs-associated clusters. This emphasizes the requirement of core proteins being assembled into capsids prior to interaction with LHBs.
HBV capsids-LHBs complexes accumulate in large intracytoplasmic compartments
Imunogold labeling above described requires specific cell preparation, in particular lighter chemical fixation to preserve the antigenic properties of the proteins. As a result, this leads to lower contrast of acquired images and it is associated with less precise definition of cell structures. To precisely visualize the intracellular architecture of capsid-LHBs clusters, regular TEM was chosen because of its property to preserve the overall intra-cellular architecture. LHBs + eGFP-LHBs (95/5) + core proteins were co-expressed in Huh-7 cells and transfected cells were sorted for eGFP expression; selected cells were then prepared for TEM. Large electron-dense areas containing numerous intact capsid shells (yellow arrows), often close to the nucleus, were observed (Fig. 7a, 7b and Fig. 5S). All capsids appeared in dense clusters as seen by immuno-TEM. Under these conditions, there was evidence of membrane-surrounded capsid clusters (white arrows in Fig. 7a, 7b and Fig. 5Sb). Taken together, TEM and immuno-TEM observations highly favor the notion that HBV capsid-LHBs interactions are formed in intracellular compartments containing internal membranes.
Huh7 cells were transfected with a mixture of DNA plasmids expressing core, LHBs and eGFP-LHBs (50/45/5) for 72h. The cells were sorted, fixed and ultrathin sections were cut and contrast-stained with uranyl acetate/citrate for regular transmission electron microscopy. Images a and b are two examples of large clusters of HBV capsids (yellow arrows) observed in intracellular compartments. White arrows indicate that HBV capsids accumulate within an intracellular compartment surrounded by a membrane. Some other cellular compartments or organelles are named in turquoise.
HBV core-LHBs complexes localize in late endosomes/MVB
Given that our TEM images clearly show that the capsid-LHBs complex is associated with membranes, we next carried out experiments to further identify the intracellular compartment in which capsid-LHBs complexes are formed. We used markers of the endoplasmic reticulum (ER) (calreticulin, calnexin and PDI), early endosomes (Rab5), late endosomes/MVB (Rab7, Rab9, CD81, HA-TSG101) and late endosomes/lysosomes (Lamp 1 and Lamp 2) in cells co-expressing core and LHBs. Transfected cells were then imaged by confocal microscopy and the co-localization between the cellular markers (green) and core (yellow) and LHBs (magenta) was quantified by the Pearson correlation coefficient (Fig. 8B for core and Fig. 8C for LHBs). As shown in Fig. 8, there is a strong co-localization between calreticulin (or calnexin, not shown) and core and LHBs, with a R value of 0.75 ± 0.09 (first row). These two proteins are ubiquitous chaperone proteins that promote assembly and folding of proteins at the ER membrane [58]. ER is also the site of the HBV envelope proteins synthesis and oligomerization via multiple disulfide bounds [59–61], and PDI has been proposed to be involved in the assembly process [62]. To test this hypothesis, we performed a PDI-targeted IF and noted, in agreement with previous studies [27], an efficient co-localization of core-LHBs with PDI with a R value of 0.65 ± 0.2 (second row).
The knowledge of an intimate relationship between ER and endosomal pathway [63] prompted us to test other membrane markers. We confirmed endosomal localization of core-LHBs with Rab 5 and Rab 9 (R value ~ 0.60) (Fig. 8, third and fourth rows) [19, 64, 65]. The Rab family consists of small GTPases that serve as master regulators of membrane trafficking in eukaryotes and localizes from early to late endosomes/MVB [66]. In fact, we identified an efficient co-localization of core-LHBs complex and CD81 in cells co-expressing core and LHBs with a R value of 0.61 ± 0.1 (fifth row). CD81 belongs to the tetraspanin family and is a marker of endosome and MVB. Interestingly, another marker of MBV, TSG101, was tested on cell co-tranfected with core + LHBs + HA-TSG101 and no co-localization between core-LHBs and TSG101 was detected (sixth row). TSG101 also referred to as Vps23 protein is a subunit of ESCRT-I (endosomal sorting complexes required for transport) [67] belongs to membrane of exosomes [68] and is involved in membrane protein sorting that facilitates the final stages of virus release [69]. The absence of co-localization of core-LHBs complex with TSG101 could results from the lack of SHBs that is dramatic for virion egress [70].
As sorting occurs, endosomal subdomains mature and, ultimately, fuse with a degradative compartment named lysosome [71]. Nevertheless, the co-localization between core-LHBs complex and Rab 7 was less effective with R = 0.47 ± 0.14 (seventh row). This protein is a marker of late endosome maturation toward the degradation pathway of endocytosed cargos. In fact, lysosomes, and often autophagosomes, are associated with Rab7 effectors [72]. In this context, we observed a strong co-localization between core-L-HBs and endogenous Lamp1 with R = 0.76 ± 0.06 (eighth row), but a lesser co-localization of membrane-cloaked capsids with Lamp 2 (ninth row, R = 0.50 ± 0.14). In contrast, there was no co-localization of core or LHBs and the lysotracker, with R values of 0.14 ± 0.09 and 0.27 ± 0.17 respectively (Fig. 8D and 8E). We could not show in addition evidence of core-LHBs being able to recruit the autophagy machinery [16] since core-LHBs and LC3 did not co-localize in cell expressing the viral proteins together with LC3-auquamarine (R = 0.30 ± 0.2, tenth row in Fig. 8A).
Huh7 Cells were co-transfected with plasmids expressing core and LHBs. After 72h, cells were fixed, permeabilized and immunostained. (A) In the green channel, images were obtained by immunofluorescence using anti-endosomal markers targeting endogenous proteins or using anti-HA for TSG101 or using aquamarine signal of LC3-Aquamarine protein in cyan. In the yellow, the human anti-HBc polyclonal antibody targeting core, and in the magenta, the goat anti-HBs monoclonal antibody (Fitzgerald 70HG15) or rabbit anti-pre S1 R271 targeting LHBs protein. DAPI nuclear staining is shown in blue in the merge panel. (B and C) The bar graphs show the mean values of the Pearson’s correlation coefficient (PCC) from 15–20 cells. The PCC was calculated independently for the green (or cyan) and yellow channels (B), or the green (or cyan) and magenta channels (C) for each cell. (D) After 72h., cells transfected with Core/Core-mCherry (90/10) plasmids, or LHBs/ mCherry-LHBs (95/5) plasmids were washed and incubated with DAPI and green Lysotracker in Leibovitz’s L-15 medium at 37°C for 30 minutes. Images show the acidified vesicles staining in green, with core (upper panels) or LHBs (lower panels) in red. (E) The mean values of the PCC of 20–25 cells are shown. In all cases scatter dots represent each cell analyzed from two independent experiments. The bars represent the mean values and the error bars the ± SD of the PCC. Representative images are shown.
Finally, the strong co-localization between core, LHBs and Lamp1 observed by confocal microscopy (eight row) was confirmed by IF on cell expressing Lamp1-mTurquoise and LHBs (Fig. 6S) and by immuno-TEM on cell expressing core + LHBs (Fig. 9). In this latter figure are two electron micrographs where capsids are identified by their shape (yellow arrows), the smaller gold beads are targeting LHBs (black arrow) while the larger Lamp1 (green arrows). As aforementioned, capsids are clustered together with LHBs associated with membranes (white arrows) expressing Lamp1 proteins. Thus, core-LHBs complexes are associated with membranes gold-labelled with anti-Lamp1.
All the experiments were also carried out on cells expressing LHBs alone (Fig S7) and images, and Pearson coefficients, were very similar to those shown in Fig. 8. All together, these results show that LHBs drive core-LHBs clusters in the ER/endosomal pathway and their accumulation in neutral vesicles.