Direct-acting HBV antivirals

HBc binding pockets and mode of action of dimeric binders.

A) Left: Close-up view of the three addressable effector sites within HBc-dimers (shown as cartoon model with transparent surface in grey) together with representative ligands shown as stick models: SLLGRM peptide (marine blue, PDB: 7PZN); Geraniol, resolved here (Cyan); heteroaryldihydropyrimidine (HAP ((2S)-1-[[(4R)-4-(2-chloranyl-4-fluoranyl-phenyl)-5-methoxycarbonyl-2-(1,3-thiazol-2-yl)-1,4-dihydropyrimidin-6-yl]methyl]-4,4-bis(fluoranyl)-pyrrolidine-2-carboxylic acid), green, PDB: 5WRE). HAP is a representative example of a canonical CAM, that targets a hydrophobic pocket mediating HBc-dimer multimerization, an essential step in capsid assembly. A blue arrow indicates how dimeric peptide-based ligands may induce aggregation. Right: The general architecture of an HBc-dimer is depicted as a cartoon with transparent surface model in grey and the three ligands that target distinct binding pockets are in color. The binding sites of two HBc dimers can be linked by dimeric ligands, here exemplified with the peptide ligand. B) Hypothetical mode of action of HBc aggregation triggered by cross linking the spikes of individual HBc dimers, HBc multimers or the whole capsid.

The central hydrophobic pocket of HBc-dimer is targeted by hydrophobic molecules containing isoprene units.

A) Structures of different substances used for the ITC and cryo-EM experiments. N-Decyl-beta-D-maltopyranoside (DM) (1) and geraniol (2). Using geranic acid we synthesized geranyl dimer (3), a dimeric binder forked by a lysine and having a linker of six dioxaoctanoic units. B) Representative ITC heat signatures of DM (1), geraniol (2) and the geranyl dimer (3) with HBc capsids. Heat release is detected upon titration of the ligands to the HBc solution, indicating stoichiometric binding interaction. 4 mM geraniol (2) was titrated into a solution of 210 µM HBc. A solution of 2 mM geranyl dimer (3) was titrated into a solution 200 µM HBc. 1.6-2 mM solutions of DM (1) were titrated into solutions with 90, 100 and 150 µM HBc, respectively. The control experiments where geraniol, geranyl dimer and DM were titrated into buffer are shown in Supplementary Figure 7. Integrated heat signatures in kcal⋅mol-1 plotted against the molar ratio of titrants to HBc. Binding isotherms (solid lines) were determined using a curve fitting procedure based on a one-site model. Among the ligands, the geranyl dimer has the strongest affinity to HBc, expectedly surpassing the monovalent geraniol by 2-fold. C) Structure of the geraniol (magenta) within the HBc binding site (yellow and red) together with close-up view of the binding site with the EM-densities. Geraniol and residues (P5, L60, K96, E64 and V13) involved in HBV’s envelopment with natural phenotypes are depicted in stick representation. The EM density of geraniol is shown in the zoom-out in blue. D) Side-by-side comparison of the overlapping HBc geraniol and TX100[31] binding sites suggests conformational flexibility and the ability of the hydrophobic pocket to accommodate larger hydrophobic molecules.

Dimeric peptide spike binders display strong low micromolar and sub-micromolar affinity.

A) Chemical structures of the dimeric peptides, all contain the core binding sequence -SLLGRM and share the same PEG linker and a lysine as the branching element of the dimer. B) Exemplary ITC thermograms showing the titration heat signature of HBc with dimers. A solution of 1500 µM (4) was titrated into a solution 150 µM HBc. A solution of 125 µM (5) was titrated into a solution 25 µM HBc. A solution of 200 µM (6) was titrated into a solution 25 µM HBc. A solution of 100 µM (7) was titrated into a solution 25 µM HBc. C) The peptide dimers display low micromolar to sub-micromolar affinity to HBc, the affinity increases with the elongation of the binding sequence. D) Sequence requirements of the HBc Spike binding site. Full positional scan of the P1 peptide sequence in microarray format, in which each residue was varied to each other proteogenic amino acid. Note that a drop in binding intensity upon variation of the core motif SLLGRM (highlighted in bold) substantiates its critical involvement in HBc binding. Refer to supplementary table 3 for the corresponding absolute greyscale values. Affinity gains observed for exchanging positively charged for negatively charged amino acids may be assay-specific false-positives as highlighted previously.[33]

P1dC aggregates HBc in living HEK293 cells.

A) A polyarginine cell-penetrating peptide containing a cysteine with a TNB-activated thiol (gray highlight, (8)). B) The live cell experiment flow. First, mammalian cells are transfected with HBc coding plasmid. Then, after the cells express the protein, a mix of (8) and (7) is applied. The excess CPP facilitates membrane permeation, allowing (7) to enter the cell after a brief incubation. Once inside, (7) is separated from the CPP and can interact with the capsids. C) After 1 hour incubation with (7) or scr(7) the cells were immediately washed, fixed and labelled with anti HBc mAb16988 and a secondary DyLight650 conjugated antibody. The cells were visualized on wide-field fluorescent microscope with identical conditions and are presented with the same grayscale range. Transfected and untreated cells display diffuse HBc distribution, with clear fluorescence at the nucleus. Transfected cells treated with (7) display bright aggregates, whereas transfected cells treated with scr(7) have similar diffuse labelling as the untreated cells. Non-transfected cells are non-fluorescent. Scale bar 20 µm.

Peptide dimers binding the spike tips.

Close up of the surface representation of the EM-maps of capsid-like particle (CLP) incubated with geraniol (2), with SLLGRM-dimer (4) and P1dC (7). (A) The surface of the map is colored according to the local resolution. The map of (7) has a lower overall resolution, which is consistent with the lower number of particles in the reconstruction (Supplementary Table 4). In all three maps the tips of the spikes are less well resolved than the capsid shell regardless of whether peptides are bound or not. This is in line with the general flexibility of the protruding spikes in HBc-CLPs.[37,48] (B) The surface of the maps is coloured according to the relative occupancy based on the grey value distribution as determined with OccuPy.[49] Low relative occupancy cannot be distinguished from local flexibility. As the tips of the spike are flexible, they show generally lower occupancy than the protein shell. Comparing the relative occupancies in samples incubated with (4) and (7) suggests a lower occupancy with (4) than with (7). (C) Fit between the model and the map (grey, translucent) at the tips of spikes. Binding of an (4) or of (7) splays the helices at the tips apart similar as previously reported for binding of a P2-monomer.[33] (4) binds to both quasi equivalent sites in contrast to SLLGRM-monomers, which binds only to the CD-dimer and does not show such a prominent splaying.[33] Geraniol binds at the center of the spikes and does not change the conformation at the tips of the spikes.

Rational design of multivalent binders.

This figure illustrates the rationale behind the design of the PEG linker length in the peptide- and geraniol-dimers. The binding site A, where the peptide or geraniol dimer binds, is located at the tip of a capsid spike or in a hydrophobic pocket (depicted as a yellow circle). The possible additional interactions of the dimers are shown as yellow circles labeled B1, B2, B3, and B4. The PEG linker, depicted as a dotted blue line, has been carefully chosen to allow simultaneous binding between two opposing sites. (I) The distance between the tip of a “central” capsid spike (A) and the surrounding four spikes (B1, B2, B3, and B4) is approximately 6 nm (60 Å). The PEG linker, with a length of about 8 nm (80 Å), was selected to provide flexibility and ensure the dimer can potentially bridge two adjacent spike tips, optimizing binding avidity by enabling interaction with any combination of the adjacent spikes. (II) For geraniol dimers, the hydrophobic pockets are separated by a distance of approximately 4 nm (40 Å), and the designed PEG linker (∼3.8 nm or 38 Å) was chosen to match this distance, allowing for optimal interaction with two pockets in close proximity.

High turbidity is observed upon addition of HBc binding dimers to HBc solution.

Assessment of turbidity of capsid solutions induced by peptide dimers. The optical densities (OD) of HBc solutions (10 or 50 µM) with and increasing peptide-dimer concentrations were measured at a wavelength of 350 nm and plotted against peptide-dimer concentrations. All peptides showed increased turbidity with increasing concentrations, with P1 (7) and P2 (5) dimers inducing turbidity at low micromolar concentrations, while the SLLGRM dimer (4) induced turbidity at high micromolar concentrations.

scrP1dC does not interact with HBc.

A) 0.1 mM solution of the scrP1dC-dimer (scrambled version of the P1dC-dimer) was titrated into a solution of 0.025 mM HBc. B and C) 0.1 mM solutions of scrP1dC-dimer and P1dC-dimer were titrated into buffer A as additional controls. In all cases no heat change was examined, validating the lack of HBc – scrambled peptide interaction and excluding residual binding interactions from the handle or linker. D) Heat fluctuations from Fig3B on identical y-axis scaling.

HBc aggregates appear after treatment with P1dC.

Live HEK293 cells expressing HBc were treated with either P1dC or with the scrambled (scr) P1dC analogue together with the cell penetrating peptide. After treatment, the live cells were fixed and labelled with HBc Antibody and a secondary DyLight650 conjugated antibody (1:500). The imaging showed HBc aggregates in P1dC treated cells (A), while scrP1dC treated cells showed little to none aggregates (B). All images shown with identical grayscale range. Scale bar 10 μm.

Overview of equilibria between the asymmetric unit of HBc and peptide-binders.

The asymmetric unit of HBc capsids (T =4) consists of a tetramer which is composed of the A/B (closest to the 5-fold symmetry axes in the icosahedron) (in blue) and the C/D-dimer (closest to the 3-fold symmetry axes) (in red). The peptide moiety of the dimers is depicted as yellow filled circles connected by a flexible PEG linker symbolized as a dotted line. Peptide dimers interact with the asymmetric unit in 4 different states (S1, S2, S3 and S4). The concentration of every state is dictated by the energetics of the respective state and the concentration of the peptide-dimer. At low concentrations, S1 and the possible degenerative permutations could be expected to be favored and at high concentrations S4. For the sake of simplicity, only a single (abstract) asymmetric unit is depicted here which represents capsid with 60 asymmetric units.

Electron microscopy and image processing of HBc-CLPs with binders: A) CLPs with bound Geraniol, B) HBc-CLPs with bound SLLGRM-dimer and C) HBC-CLPs with bound P1dC

A) shows a representative micrograph. All micrographs are shown at the same scale as indicated in Aa; B) shows the 2D-Class averages of the 5 most populated classes after automated template picking. All class averages are shown at the same scale as indicated in Ab; C) shows a close-up of the surface representation of the final map after post-processing with relion (filtered by Fourier Shell Correlation (FSC), B-factor sharpened). One unit cell is colored according to the density covered by the model (HBc chains A, B, C, D in blue, cyan, yellow and red respectively, and binders (geraniol, SLLGM and P1 in green); D) Fourier Shell Correlation plot of the final map. FSC=0.143 is marked by a thin, solid line. Green curve: FSC between unmasked half-maps, blue curve: FSC between masked half-maps; red-curve FSC between phase-randomized masked half-maps, black curve: FSC corrected for the contribution of the mask.

Control titrations of substances used for ITC experiments.

All substances were dissolved in buffer A and titrated into buffer A as a control for experiments where equal concentrations of these substances were titrated into solutions of HBc (see Figures 2 and 3 in the main text). A) 4 mM geraniol, B) 2 mM geraniol dimer, C) 1.7 mM DM, D) 0.3 mM P1d, E) 0.5 mM P2d and F) 1.5 mM SLLGRM dimer. X and Y axes are scaled differently in the panels.

Geraniol binding mode to the hydrophobic pocket at the base of the capsid spike.

(A) Model of the CD-dimer (yellow, red) with bound geraniol (purple) inside the EM-map (transparent). The density accounted by geraniol is shown in blue. Right: close-up of the hydrophobic pocket with residues labelled in the vicinity of geraniol and at the entrance of the pocket. Residues P5, L60 and K96 are implicated with naturally occurring envelopment phenotypes. Close-up of the quasi-equivalent hydrophobic pockets with bound geraniol. (B) Slice of the EM-map with fitted model: The slice shows the center of the two quasi-equivalent spikes with the fitted models of Geraniol in brown. One Geraniol molecule is bound to each of the quasi-equivalent sites. The surface of the EM-map is transparent and coloured according to the chains. For clarity the density attributed to Geraniol is highlighted in green (colour blob option of Chimera[68]). Geraniol binds to the same site as Triton X100[31] but does not change the rotamer conformation of F97. Binding of geraniol is not linked to conformational changes in the HBc-dimers. (C) Slices of the EM-map at the centre of the spikes shown above. The surface of the map is coloured according to the relative occupancy estimated with “OccuPy”[49] based on the grey value distribution. The relative occupancies at the geraniol moiety are somewhat lower than the surrounding protein. Considering that flexibility and occupancy have a similar effect on the grey value distribution, the geraniol has an increasing flexibility towards the outside of the pocket and has at least 80-90% occupancy at the interior of the pockets. The colour key for the relative occupancy is shown below.

Cryo-EM confirms strong capsid aggregation with peptide dimers.

Low magnification cryo-EM images of CLPs + P1dC (7) (A), (B) CLPs + SLLGRM-dimer (4) and (C) CLPs + geraniol (2). The micrographs are part of the grid-atlas of the respective data acquisition. Each image shows 4 meshes of the respective grid atlas at a similar ice thickness. For representation, the images were aligned to show a similar orientation of the meshes. CLP aggregates are seen as dark speckles (yellow arrow). The size of the aggregates is largest in P1dC-treated samples, while aggregates are frequent and smaller in samples treated with SLLGRM-dimer. Geraniol treated samples have very few aggregates which are generally smaller than 1 µm.

Nanoscale resolution of the dimer binding sites by Cryo-EM.

HBc capsids with bound SLLGRM-dimer (4) or P1dC-dimer (7) imaged by electron cryo-microscopy. (A) and (B) show selected areas of micrographs of CLPs treated with (4) or with (7). One exemplary aggregate of multiple HBc capsids is indicated by an arrow in each micrograph. (C) and (D) show close-ups of the asymmetric unit of HBc capsids with bound SLLGRM dimers or with bound P1dC. Models of a single asymmetric unit consisting of two HBc dimers is fitted into the asymmetric unit. Both maps show a density at the tips of the spikes (arrow) that accounts for approximately six amino acids of the peptide-dimer. The flexible linker between the peptides was not resolved. The position of the symmetry axes of the icosahedral capsid is labelled with numbers in (C).

ITC200 specifications

ITC200 instrument’s specifications used for the interaction analysis between peptides and capsids.

Thermodynamic parameters of HBc capsids interactions.

Summary of thermodynamic parameters obtained by ITC experiments using the peptide dimers and fl wt HBc capsids. In case of the P2 dimer the deviations represent deviations of fit since only one ITC experiment was performed. N represents stoichiometry.

Microarray positional scan data.

Obtained raw greyscale values from P1 full positional scan in µSPOT format. Each box corresponds to a single point variation of the P1 peptide sequence (horizontal) as indicated in the first column. The raw intensity values presented here were used for calculating the fold intensity change of each point variation against the wildtype sequence. Data are presented as mean of n=3 microarray slides with standard deviation.

Cryo-EM data.

Summary of Cryo-EM data acquisition and image processing of the HBc CLPs with bound geraniol, P1dC and SLLGRM-dimers.