Figure 1. Image reconstruction of CLPs vitrified 10 months after purification.
(a) A segment of the capsid color-coded by the radial position. The length of the color key corresponds to 6 nm. For comparison, the capsid of human HBc is shown in Figure 1—figure supplement 1. A representative micrograph and class averages are shown in Figure 1—figure supplement 2. The same sample 2 weeks after purification was less well resolved (Figure 1—figure supplement 3). Conditions for imaging and image processing are summarized in Table 1. (b) An equatorial slice through the EM-density with spikes indicated by arrows (narrow sections: filled arrowheads; wide cross sections: open arrowheads). The diffuse shell of density at the capsid interior is highlighted by a white arch. The scale bar indicates 10 nm. (c) Surface representations of 3D-class averages from the classification of the asymmetric units without alignment. The two horizontal lines indicate the same radial position and are 6 nm apart. Class 1, 2 and 4 differ in their radial positions by up to 2 Å. Class five represents holes in the CLPs. Perpendicular views are shown in Figure 1—figure supplement 4. (d) Model of the asymmetric unit of DHBc fitted into the EM-map. Chain A is shown in blue, chain B in cyan, chain C in yellow and chain D in red. The model validation is summarized in Table 2. The model also fits with the map of DHBc co-expressed with FkpA two weeks after purification (Figure 1—figure supplement 5) (e,f) Perpendicular views of the map of the CD-dimer with the model of the CD-dimer fitted. Density accounted for by the extension domain (79-119) is colored white, the core of chain C in yellow and that of chain D in red. The density at the foot of the spikes (magenta) is neither accounted by the assembly domain of chain C nor of chain D and is modelled with the most C-terminal residues.
Figure 1—figure supplement 1. Organization of HBc.
(a) Surface representation of a segment of HBc CLPs (Böttcher and Nassal, 2018), with the subunits in one asymmetric unit highlighted in color (blue chain A, cyan chain B, yellow chain C, red chain D). (b) Representation of the AB-dimer of the asymmetric unit. The naming of the helices follows Wynne et al., 1999. Inter-dimer contacts are formed between the hand-regions. Inner dimer contacts in HBc can be stabilized by a disulfide-bridge between Cys61 in HBc. (c) PFAM alignment of HBc and DHBc. The alignment corresponds to Conserved Protein Domain Family pfam00906 Hepatitis_core, except that the primary sequences were adapted (lower case green letters) to match the isolate-specific DHBc and HBc proteins used here (Uniprot IDs: P0C6J7.1 and P03146.1, respectively). Identical residues are shown in red, lower case grey letters indicate non-aligning residues. Grey bars show the positions of HBc helices α1 to α5 as in pdb: 1QGT; HBc domain borders are indicated in dark green. DHBc sequence 62–134 (highlighted in yellow) is predicted to replace HBc sequence 71–89, i.e. the spike tip comprising the C terminal part of α3, the connecting loop (the c/e1 epitope) and the N terminal part of α4. DHBc sequence 78–122 (light-green box) represents the biochemically predicted and here directly confirmed extension domain at the spike tip. (d) Conservation of the extension domain sequence among hepadnaviral large type CPs: The indicated CP sequences with their Uniprot and/or NCBI accession numbers were aligned using Clustal Omega as implemented in SnapGene v. 5. Tibet Frog c refers to one of three published Tibetan Frog HBV sequences. Conservation is indicated by colored blocks, with dark blue indicating the lowest and dark red the highest conservation. Residues differing from the consensus are shown with green background. The avian HBV CP sequences are highly similar to each other, including in the extension domain (red box around DHBV16 CP aa V77-T122). The frog virus sequence only shares some interspersed key motifs; however, the sequence predicted to align with the DHBc extension domain is also exceptionally rich in proline residues (10 in 48 aa total) which also cluster mostly in the 5'- proximal part.
Figure 1—figure supplement 2. Sample preparation and 2D-image analysis of DHBc 2 weeks and 10 months after purification.
(a) Analysis of fractions obtained by step-gradient ultra-centrifugation with native agarose gel electrophoresis with 1% agarose in 2 x TAE buffer. The staining was done with SYBR Safe (Thermo Fisher Scientific) at a dilution of 1:5.000. The position of intact DHBc CLPs is indicated by an arrow. The sizes of the DNA standard are indicated in kilo base pair (kbp). Intact DHBc CLPs have an electrophoretic mobility comparable to DNA with a size of approximately 10 kbp in TAE buffer (fractions 17–18). In the following fractions (towards higher sucrose densities) all bands become gradually diffuse and shift towards lower mobility. Additionally, an increasing amount of proteins is aggregated and stuck in the wells of the agarose gel. (b) The same fractions were subjected to SDS-PAGE with 15% acrylamide and stained with Coomassie Brilliant Blue. The position of DHBc (30 kDa) is indicated by an arrow. (c, d) DHBc after 2 weeks storage. (c) A typical micrograph shows aggregated CLPs (some labelled with ‘A’) and lipid vesicles (some labelled with ‘L’). The scale bar indicates 100 nm. (d) Selected class averages of CLPs vitrified 2 weeks after purification. Most CLPs fall into one size class. Some class averages represent deformed CLPs (‘#”) and some are consistent with a smaller size class (‘*”). The scale bar in (d) indicates 10 nm. (e,f) Typical micrograph (e) and selected class averages (f) of DHBc after 10 months of storage. Class averages of particles vitrified 10 months after purification are better resolved than CLPs vitrified 2 weeks after purification and show projections of secondary structural elements in 2D-class averages. (g,h) Typical micrograph (g) and selected class averages of DHBc co-expressed with the peptidyl-prolyl cis-trans isomerase FkpA after 2 weeks of storage. The class averages are better resolved than DHBc CLPs without co-expressed FkpA after 2 weeks of storage. (i,j) Typical micrograph of DHBc-R124E-Δ i and typical class averages j. (k,l) Typical micrograph of DHBc-R124E (k) and typical class averages (l).
Figure 1—figure supplement 3. Reconstruction of CLPs vitrified 2 weeks after purification.
(a) Surface representation color-coded by the radius. The right front quarter of the CLP is not shown giving a view to the interior of the CLP at the left side. (b) Equatorial slice through the capsid. The scale bar indicates 10 nm. (c) Surface representations of 3D-classes of the asymmetric units of CLPs. The asymmetric units were classified without further alignment thus classes can also represent positional variability relative to the expected position in the capsid. The horizontal lines indicate the same radial position in the capsid. All surfaces were calculated at the same density threshold. The relative occupancies of the classes are indicated. Classes 1, 2, 6 and 8 show a weaker than expected density probably representing holes or grossly misplaced subunits. Classes 3 and 5 show one wide and one narrow spike. Classes 4 and 7 show two wide spikes.
Figure 1—figure supplement 4. Resolution of DHBc CLPs.
(a) Fourier shell correlation after gold standard refinement and masking for the asymmetric unit of DHBc prepared after 10 months (solid black line) and for the CLPs (dotted line). The FSC threshold of 0.143 for quoting the resolution is shown as thin black line. (b) Representation of selected side chain density of α4 in, at the N-terminus and in the extension domain highlighting the quality and local variation in resolution of the map. (c) Two perpendicular views of the 3D-map of the asymmetric unit after local orientational refinement of the particles of classes 1, 2 and 4 in (e). The color-coding represents the local resolution and is lowest in the extension domain. (d) Two perpendicular views of the model of the asymmetric unit of DHBc colored with the local B-factors calculated during Phenix ADP-refinement. The extension domains have the highest local B-factors, which matches the low local resolution in this region. The color key is given below. The absolute numbers for the B-factors are somewhat arbitrary and depend on the B-factor used for map-sharpening during post-processing in RELION. (e) Surface representations of 3D-class averages from the classification of the asymmetric units without alignment. Class 1, 2 and 4 show broad spikes. Class five represents holes in the CLPs. The views are perpendicular to those shown in Figure 1.
Figure 1—figure supplement 5. DHBc CLPs with co-expressed FkpA.
(a) Fourier Shell Correlation of maps of DHBc co-expressed with FkpA. The plots represent different maps of CLPs reconstructed from micrographs acquired from the same grid either in counting mode (magenta) or in linear mode (red and blue). Reconstructions in linear mode were calculated from the full data set (4039 movies, blue) or from the 1st part of the data acquisition (2038 movies, red). Reconstructions in counting mode were calculated from 2425 movies and had the same overall resolution (4.2 Å, magenta) as data acquired in linear mode from the first part of the acquisition with a similar number of micrographs (for more details on imaging see Table 1). This comparison suggests that the resolution of the final maps was not limited by the image acquisition mode, similar as reported earlier for HBc (Song et al., 2019). Doubling the data by approximately a factor of two in linear mode increased the resolution only slightly from 4.2 Å to 4 Å. The best resolved 3D-map of the CLPs of DHBC+FkpA from all movies acquired in linear mode is shown as surface representation of the B-factor sharpened map in (b) and as equatorial slice of the none-sharpened map in (c). The surface representation in (b) is color-coded by the radius as indicated by the color key. The length of the color key corresponds to a length of 6 nm. In the equatorial slice in (c) the arch indicates the approximate position of diffuse density that is mainly attributed to RNA. The similar resolution and quality of the 3D-maps from counted and linear data indicated that the limited resolution of wt DHBc prepared after two weeks is most likely a property of the sample and not related to the imaging conditions (linear, 40 e-Ų). (d-g) Analysis of the asymmetric unit of DHBc+FkpA for the counted data (classification shown in d and e) and for the linear data (best class shown in f and g). Both classification gave similar results. Surface representations calculated at the same threshold of the 3D-class averages of the counted data are shown in (d) and (e). The views in (d) and (e) are perpendicular to each other. Class1 and class 3 (38% of all asymmetric units, both 7.2 Å resolution) represented mainly holes or grossly displaced subunits. Class two was poorly resolved (15% of all asymmetric units, 6.2 Å resolution). Classes 4 and 5 (45% of all asymmetric units, 4.9 Å and 5.7 Å resolution respectively) were better resolved but had a relative radial displacement of 3 Å, which is larger than for the 10 months old DHBc CLPs (2 Å displacement) but significantly smaller than the 7 Å displacement of the 2 weeks old CLPs without co-expressed FkpA (Figure 1—figure supplement 3). Classes 4 and 5 both showed broad spikes indicating folding of the extension domain. We did not observe any classes showing consistently narrow spikes as we had observed for the 2 weeks old CLPs without co-expression of FkpA. For the best class of data acquired in linear mode, we locally refined the orientations of the asymmetric unit. The resulting 3D-map with the fitted model of the 10 months old DHBc is shown in (f and g) in two perpendicular views of the same map/model. This 3D-map has a resolution of 3.9 Å (see (a), green plot). In conclusion, data acquired in linear mode at a lower dose has a sufficient signal-to-noise-ratio to analyze the individual asymmetric units of the CLPs.