Structures of Hepatitis B Virus core- and e-antigen Immune-complexes Suggest Multi-Point Inhibition

Summary

Hepatitis B virus (HBV) is the leading cause of liver disease worldwide. While an adequate vaccine is available, current treatment options are limited, not highly effective, and associated with adverse effects, encouraging the development of alternative therapeutics. The HBV core gene encodes two different proteins: core, which forms the viral nucleocapsid, and pre-core, which serves as an immune-modulator with multiple points of action. The two proteins mostly have the same sequence, though they differ at their N- and C-termini and in their dimeric arrangements. Previously, we engineered two human-framework antibody fragments (Fab/scFv) with nano- to pico-molar affinities for both proteins. Here, by means of X-ray crystallography, analytical ultracentrifugation, and electron microscopy we demonstrate that the antibodies have non-overlapping epitopes and effectively block biologically important assemblies of both proteins. These properties, together with the anticipated high tolerability and long half-lives of the antibodies, make them promising therapeutics.

eTOC blurb

Hepatitis B virus employs two very closely related proteins; core to form its capsid and pre-core to modulate the host immune response. Eren et al. describe how two exceptionally high-affinity humanized antibody fragments bind to and block both of these proteins and propose how they may function as therapeutics.

Introduction

Hepatitis B virus (HBV) constitutes a major public health threat in many areas of the world. HBV infection can lead to chronic Hepatitis B (CHB) which can progress to end-stage liver disease, such as cirrhosis and hepatocellular carcinoma (HCC). While the widespread implementation of HBV vaccination has led to a decline in the incidence of new HBV infections, the prevalence of CHB remains high. The latest report from the World Health Organization (WHO) shows that an estimated 257 million people (3.5% of the world’s population) are living with CHB, resulting in approximately 1 million deaths annually (WHO, 2017). Coinfection with HIV is also common and the global prevalence of HBV infection in HIV-infected people is 7.4% (an estimated 2.7 million) (WHO, 2017). Liver diseases are a major cause of morbidity and mortality among people coinfected with HBV and HIV.

HBV is a small DNA virus in the hepadnaviridae family. The infectious virion, known as the Dane particle, has a spherical, double-shelled structure 42 nm in diameter (Venkatakrishnan and Zlotnick, 2016). The outer layer is a lipid envelope that contains embedded viral proteins called the surface antigen (HBsAg), which are involved in viral binding and entry into susceptible cells. The envelope surrounds an icosahedral nucleocapsid composed of the core antigen (HBcAg). The nucleocapsid encloses the viral nucleic acid and DNA polymerase. HBcAg is a 183-residue polypeptide comprising a capsid-forming region called the assembly domain (AD, residues 1149), and a basic arginine-rich domain called the nucleic acid binding domain (NABD, residues 150–183) (Fig. S1) (Venkatakrishnan and Zlotnick, 2016).

The core or C gene of HBV has two in-frame translation initiation codons (core and pre-core) and the C open reading frame encodes either the HBcAg or the e-antigen (HBeAg), depending on where the translation begins. HBeAg has an extra 29 N-terminal residues, which serve as a signal peptide directing the nascent polypeptide chain into the secretory pathway (Ou et al., 1986). HBeAg is N- and C-terminally processed in the endoplasmic reticulum (ER) prior to secretion. Mature HBeAg is C-terminally truncated variably between residues 149–154 and retains 10 N-terminal residues ([−10]-[−1]) of the signal peptide (called the pro-peptide [PP]: SKLCLGWLWG) (Fig. S1) (Messageot et al., 2003). The recombinant HBcAg AD (rHBcAg, residues 1–149) and HBeAg (rHBeAg, residues [−10]-149/150) have been used to obtain the crystal and cryo-EM structures of the core capsids (Bourne et al., 2006; Katen et al., 2013; Schlicksup et al., 2018; Tan et al., 2007; Venkatakrishnan et al., 2016; Wynne et al., 1999), and the first crystal structure of the HBeAg (DiMattia et al., 2013). These structures have shown that under oxidizing conditions the cysteine residue at position −7 forms an intramolecular disulfide bond with C61 locking rHBeAg dimers (rHBeAg_d) into an arrangement distinct from that of rHBcAg dimers (rHBcAg_d) (DiMattia et al., 2013). Analysis of HBeAg, isolated from HBV-positive patient plasma samples, by SDS-PAGE under reducing and non-reducing conditions, has confirmed the presence of the disulfide bond between C(−7) and C61 in HBeAg (Zhuang et al., 2017).

HBeAg is a non-particulate protein, and is not required for viral replication, infection or capsid assembly. However, HBeAg plays an important role in the establishment and progression of CHB by modulating the host immune response through multiple pathways. Secreted HBeAg preferentially elicits non-inflammatory Th₂ cells (Milich et al., 1997) and deletes inflammatory Th₁ cells by Fas-mediated apoptosis (Milich et al., 1998). In addition, studies with mouse models have shown that HBeAg plays a role in the impairment of CD8⁺ cytotoxic T lymphocytes (CTL) following vertical transmission of HBV (Tian et al., 2016). Although the majority of HBeAg is secreted, ~15–30% of the mature protein is recycled to the cytoplasm (Duriez et al., 2008; Garcia et al., 1988). Cytoplasmic HBeAg suppresses the Toll-like receptor (TLR) signaling pathway by interacting with the Toll/IL-1 receptor (TIR)-containing proteins Mal and TRAM and disrupting homotypic TIR:TIR interactions critical for TLR signaling (Lang et al., 2011). Under the reducing conditions of the cytoplasm, HBeAg can undergo a conformational shift to form empty “decoy” capsids that can be enveloped and released, presumably to exhaust the host immune system (Duriez et al., 2014; Kimura et al., 2005). More recently, HBeAg was also shown to suppress the IFN/JAK/STAT signaling pathway by stimulating the expression of suppressor of cytokine signaling 2 (SOCS2) (Yu et al., 2017).

An absolute cure (total elimination of HBV DNA) for CHB is currently not achievable but antiviral treatments involving administration of pegylated-interferon or nucleo(s)tide analogues have demonstrated clinical benefit to CHB patients (Chang et al., 2014; Loggi et al., 2015). Sustained suppression of HBV improves CHB by slowing or reversing the degree of fibrosis and by decreasing the incidence of HCC. Interferon-α, which has mainly immunomodulatory effects and limited direct antiviral effects, can be administered for short periods of time, whereas nucleos(t)ide analogues, which function by inhibiting the reverse transcriptase activity of HBV polymerase, require lifelong use in most cases. However, these therapies are often associated with contraindications or adverse effects and are not always effective in achieving a sustained viral response (Chang et al., 2014; Loggi et al., 2015). In addition, resistant mutants emerge after prolonged use of some of these drugs including the low genetic barrier drugs Lamivudine and Adefovir (Lim, 2017; Zoulim, 2007). Therefore, alternative therapeutic strategies are required that would ideally eradicate intra-hepatic HBV or at least increase the efficacy of the therapy. Complete eradication of HBV from the host system requires overcoming two major challenges: one is the elimination or silencing of the particularly stable covalently closed circular DNA (cccDNA) from infected hepatocytes to prevent transcription and future replication cycles, the other is the restoration of an effective immune response to destroy already infected cells and to ensure protection from reactivation of the virus (Chang et al., 2014). Current strategies include development of capsid assembly inhibitors and new classes of immune modulators as HBV drugs (Sharpel, 2017). Capsid assembly inhibitors target the HBcAg and function by destabilizing the nucleocapsid and blocking RNA packaging, thereby preventing the transport of viral genetic material between cells. Immune modulators aim to restore host innate and adaptive immunity to induce HBV clearance. One of the primary objectives of such drugs is the prevention of host immune modulation by the viral HBeAg. For instance, a TLR agonist has been shown to induce long term suppression of HBV DNA and a reduction in serum HBsAg and HBeAg (Lanford et al., 2013).

Antibodies and their fragments have been useful as tools in a number of therapeutic applications to treat cancer, autoimmune disorders, and viral infections (Wang et al., 2008). Chimeric, humanized, and fully human therapeutic antibodies are associated with a much lower frequency of immunogenicity (anti-antibody responses (AAR) against variably engineered Mab have been observed in the range of <1% to ~10% of patients), which increases both their tolerability and half-life (Hwang and Foote, 2005; Wang et al., 2008). In addition to their direct use as therapeutics, antibodies can also be used to design therapeutics such as cyclic peptides or even smaller molecules. However, such strategies require the knowledge of epitopes on the antigen and paratopes on the antibody, which currently can be best obtained by structural methods, such as X-ray crystallography. In this study, we describe the structures of two high-affinity chimeric rabbit/human antibody fragments that recognize both the HBcAg and HBeAg and discuss their potential use as viral inhibitors.

Results and Discussion

In this study, we have used rHBcAg and rHBeAg. Supplementary Figure S1 shows a schematic representation of these constructs in comparison to the wild-type (WT) proteins. rHBcAg comprises the AD (residues 1–149) and lacks the C-terminal NABD (residues 150–183). rHBeAg comprises the PP and AD (residues [−10]-149) and lacks the short arginine-rich C-terminal domain (residues 150–152/154) of the WT HBeAg. These C-terminally truncated recombinant proteins provide some advantages over the WT proteins in terms of protein solubility, purification and crystallization and therefore similar constructs have been used by other groups for biochemical, structural and drug design studies (Bourne et al., 2006; Ceres et al., 2004; Cho et al., 2014; Jeong et al., 2014; Venkatakrishnan et al., 2016).

Previously, we identified several diverse chimeric rabbit/human Fabs against rHBeAg_d using a phage display library (Zhuang et al., 2017). Sequence alignment of the complementarity determining regions (CDRs) of the 24 Fabs revealed two clusters (Fig. S2). The first cluster is comprised of 16 Fabs which have high sequence conservation among their CDRs (97–66% sequence identity), suggesting these Fabs might recognize similar epitopes. The second cluster includes the remaining 8 Fabs which have less sequence conservation among their CDRs (50-37% sequence identity) and where the least conservation is observed amongst the heavy chain CDR3 (HCDR3). Fab e13 and Fab e21, from the first and second cluster, respectively (41% CDR sequence identity), were among the highest-affinity partners for HBeAg (K_d = 3.1 × 10⁻¹¹ M and K_d = 1.1 × 10⁻¹⁰ M, respectively; Table 1). We tested both Fabs for their potential use as HBV diagnostic tools due to their high affinity and designed a test kit that incorporates a murine monoclonal antibody, Mab me6, for initial capture of HBeAg_d from patient sera and Fab e13, which did not compete with Mab me6, for quantification (Zhuang et al., 2017). In contrast, Fab e21 showed reduced binding to HBeAg_d in the presence of Mab me6, suggesting an overlap of the epitopes (Zhuang et al., 2017). To define the epitopes on both antigens and the paratopes on the antibody fragments we characterized the immune-complexes by surface plasmon resonance, analytical ultracentrifugation, and X-ray crystallography.

Table 1.

rHBcAg_d and rHBeAg_d immune complexes: molecular weights and binding affinities. See also Figure S2, Figure S3 and Figure S4.

Immune Complex a	Mw (AUC) b kDa	Mw (Sequence) c kDa	Binding affinity d Kd (M)
rHBcAgd-Fab e13	117.1 (±1.5)	128.66	3.4 ×10−12
rHBcAgd-scFv e13	80.0 (±1.3)	86.92	nd
rHBeAgd-Fab e13	127.00 (±1.5)	130.87	3.1 ×10−11
rHBeAgd-scFv e13	84.51 (±1.4)	89.20	nd
rHBeAgd-Fab e13-scFv me6	182.5 (±3.5)	187.50	nd
rHBcAgd-Fab e21	126.6 (±1.1)	128.44	2.0 × 10−12
rHBeAgd-Fab e21 e	136.41 (±2.5)	130.73	1.1 × 10−10
rHBcAgd-Fab me6	138.52 (±2.8)	133.11	2.7 [2.0] d × 10−9
rHBcAgd-scFv me6	93.7 (±1.4)	90.11	[1.1] × 10−9
rHBeAgd-Fab me6	138.86 (±3.0)	135.2	1.1 × 10−8
rHBeAgd-scFv me6	99.4 (±1.3)	92.21	1.6 × 10−9

^aChimeric antibody fragments: e13 and e21 (rabbit/human); me6 (murine/human).

^bMolecular weights were determined by sedimentation equilibrium. The equilibrium profiles for scFv e13 and Fab e21 bound to rHBcAg_d and rHBeAg_d are shown in Figure S3.

^cMolecular weights were calculated from the amino acid sequences for a 1:2 complex (i.e. 1 rHBc/eAg_d plus 2 antibody fragments). In all cases, there is a close match between the experimental and expected molecular weights, indicating that the binding stoichiometry is 1:2. The mass of the rHBeAg_d-Fab e13-scFv me6 complex corresponds to a stoichiometry of 1:2:2 (i.e. 1 rHBeAg_d plus 2 Fab plus 2 scFv), indicating that Fab e13 and scFv me6 have non-overlapping epitopes.

^dBinding affinities were determined by surface plasmon resonance (Biacore) with the antibody fragments immobilized (ligands) and the HBV proteins (analytes) titrated. The two values in brackets were determined with the HBV proteins immobilized and the antibody fragments titrated.

^eBinding stoichiometry for the rHBeAg_d-Fab e21 complex was previously reported to be 1:05 [12], compared to 1:2 in the current study. This discrepancy is due to incomplete saturation of the rHBeAg_d with Fab e21 in the earlier study. The binding affinity remains the same.

Characterization of Immune Complexes

Single chain variable fragments (scFvs) consist of the smallest antigen binding domain of an antibody where the variable heavy (V_H) and variable light (V_L) chain domains are joined by a flexible peptide linker. Like Fabs, they retain the binding specificity of the parent antibody and their smaller size gives them several advantages such as better tissue penetration, lower immunogenicity and cost-efficient production in high quantities (Monnier et al., 2013). In addition, they have proven to be effective crystallization chaperones, sometimes improving the resolution (Griffin and Lawson, 2011). Therefore, we prepared scFv versions of some of the Fabs to aid in our crystallization studies (Zhuang et al., 2017).

For crystallization, we prepared immune complexes of rHBcAg_d and rHBeAg_d with either chimeric rabbit/human Fab or scFv antibody fragments (Table 1). For biochemical and structural comparisons, we also included chimeric murine/human Fab me6 and scFv me6 which were derived from Mab e6 and retain the original CDRs (Ferns and Tedder, 1984). A rHBeAg_d-Fab me6 crystal structure has been solved previously and, therefore, the epitope-paratope regions are well defined (DiMattia et al., 2013). The chimeric antibody fragments bind to either rHBcAg_d or rHBeAg_d with (in most cases) similar affinities (Table 1). The fact that a given antibody binds to both rHBcAg_d and rHBeAg_d with similar affinities indicates that the protein epitopes are freely (and equally) accessible despite their rather different dimeric arrangements (DiMattia et al., 2013). Analytical ultracentrifugation (AUC) analysis of antibody fragments in complex with rHBcAg_d or rHBeAg_d shows that both proteins are bivalent with respect to antibody binding, whereby each subunit of the dimers quantitatively binds an antibody molecule without steric hindrance (Table 1 and Fig. S3). This, again, suggests similar structural epitopes on both antigens despite distinct overall conformational differences. The AUC results also show that rHBeAg_d + Fab e13 + scFv me6 forms a ternary complex with a ratio of 1:2:2 (1 rHBeAg_d + 2 Fab + 2 scFv), indicating distinct epitopes for these two antibody fragments, which allows simultaneous binding (Table 1 and Fig. S4).

Crystallization of rHBcAg and rHBeAg in Complex with scFv e13

The rHBcAg_d-scFv e13 crystals (space group: C₂) diffracted to 3.33 Å (PDB ID: 6CWD) and the asymmetric unit contained two copies of a 1:2 rHBcAg_d-scFv e13 complex (Fig. 1A and Table 2). The rHBcAg forms a dimer similar to those previously crystallized from the soluble form (PDB: 3KXS) (Packianathan et al., 2010) and when assembled as capsids (PDB: 1QGT) (Wynne et al., 1999). The dimer has the characteristic central four-helix bundle, but with the α3 - α4a loop on each monomer, the two together corresponding to the spike apices in the context of the capsid, less well ordered than in the previous structures. The intra-dimer helical bundle is also slightly distorted relative to these structures as it has a greater crossing angle between the two α3 helices.

Figure 1. Crystal Structures of Immune Complexes.

(A) rHBcAg_d-scFv e13 at 3.33 Å, (B) rHBeAg_d-scFv e13 at 2.38 Å, and (C) rHBcAg_d-Fab e21 at 3.15 Å resolution. The rHBcAg and rHBeAg monomers are shown as gold and blue ribbons. Surface representations of the variable heavy chain fragments (V_H) and variable light chain fragments (V_L) are shown in pink and gray, respectively. The N-terminal 10 amino acid pro-peptide (PP) in rHBeAg is shown in magenta. (D) Overlaid structures of scFv e13, Fab e21 and Fab me6 bound to rHBeAg (for clarity only a single chain is shown for the rHBeAg although the protein is dimeric). rHBeAg is represented as a ribbon where the alpha-helical regions are color-coded (DiMattia et al., 2013). The panel below shows the corresponding amino acid ranges for each region. α4 is divided into two helices, α4a and α4b, separated by a kink at residues V99-G104. Only the variable chains of the antibody fragments are shown as surfaces (scFv e13 in pink, Fab e21 in yellow, and Mab me6 in blue). See also Figure S1, Figure S3, Figure S5 and Figure S6.

Table 2.

X-ray Diffraction Data Collection and Refinement Statistics

	rHBeAgd-scFv e13 PDB 6CVK	rHBcAgd-scFv e13 PDB 6CWD	rHBcAgd-Fab e21 PDB 6CWT
Data Collection
Beamline	APS 22-ID-D	APS 22-ID-D	APS 22-ID-D
Wavelength	1.0	1.0	1.0
Space Group	P21	C2	C2
Cell dimensions
a, b, c (Å)	46.5, 192.1, 65.0	219.4, 65.1, 148.2	238.4, 66.5, 104.2
α, β, γ (°)	90.0, 99.0, 90.0	90.0, 101. 7, 90.0	90.0, 110.3, 90.0
Resolution (Å)	20.00-2.38 (2.48-2.38) *	19.92-3.33 (3.45-3.33)	39.98-3.15 (3.21-3.15)
Multiplicity	5.1 (5.2)	2.1 (2.1)	6.4 (5.1)
Completeness (%)	100.0 (100.0)	98.6 (99.5)	97.2 (82.7)
I/σ	10.3 (2.4)	5.0 (1.5)	12.9 (1.9)
R-meas	0.16 (1.20)	0.25 (0.82)	0.18 (1.45)
CC1/2	0.99 (0.59)	0.96 (0.67)	0.94 (0.67)
No. of unique reflections	44849 (4459)	30123 (3029)	26112 (2196)
Refinement
Resolution (Å)	20.00-2.38	19.92-3.33	39.98-3.15
No. reflections used in refinement	44851 (4460)	30118 (3029)	26052 (2190)
Rwork/Rfree	0.1745/0.2196	0.2424/0.2821	0.2305/0.2831
No. atoms
Protein/Water	5896/266	10853/0	8078/0
B-factors
Protein/Water	48.2/51.6	57.5	120.4
RMS deviations Bond lengths (Å) and angles (°)	0.015/1.33	0.005/0.96	0.003/0.82
Ramachandran (%) favored/allowed/outliers	96.6/3.0/0.4	91.3/8.1/0.6	90.2/9.1/0.7

^*Statistics for the highest-resolution shell are shown in parenthesis.

rHBeAg_d-scFv e13 crystals (space group: P₂₁) diffracted to 2.38 Å, the highest resolution obtained for HBeAg so far (PDB ID: 6CVK, Table 2). In agreement with the AUC results, the structure shows a 1:2 rHBeAg_d-scFv e13 complex with one copy in the asymmetric unit (Fig. 1B). The structure shows that rHBeAg forms a dimer similar to the rHBeAg_d observed in the previously obtained rHBeAg-Fab me6 structure (PDB: 3V6Z) (DiMattia et al., 2013) with some conformational shifts (the Cα overall root mean square deviation between the 3V6Z monomers and 6CVK monomers is 0.98 Å, while that between the dimers is ~5.0 Å [Fig. S5]) (DiMattia et al., 2013). In 6CVK, the distal ends of the C-terminal helices of the two monomers are rotated by 23° in opposite directions compared to 3V6Z (Fig. S5). This conformational difference between the two dimers is due in part to both a slight rigid-body rotation between the monomers and to conformational shifts within the monomers, and can probably be attributed to the high physical flexibility of rHBeAg_d, as previously shown by hydrogen-deuterium exchange-mass spectrometry (Bereszczak, 2014). The rHBeAg_d-scFv e13 complex shows the full-length PP with a more ordered helical conformation (3₁₀-helix) in contrast to the extended loop structure observed in the 3V6Z structure while still retaining the C(−7)-C61 disulfide bond (Fig. 1B and Fig. S5). The presence of the C(−7)-C61 disulfide bond was verified by generating a composite omit map to remove model bias (data not shown).

Our structures confirm bivalent scFv e13 binding to both rHBcAg_d and rHBeAg_d where the antibody fragment recognizes epitopes on each monomer that are not located on the dimerization interfaces and, therefore, the binding is not affected by the structural differences between the two molecules.

Cystallization of HBcAg in Complex with Fab e21

Although the CDR sequences of Mab e6/Fab me6 and Fab e21 are quite different (24.8 % sequence identity), previous antibody competition assays indicated that Mab e6/Fab me6 competes with Fab e21 for the rHBeAg_d binding site (Zhuang et al., 2017). Interestingly, Fab e21 has significantly higher affinity for both rHBcAg_d and rHBeAg_d (Table 1). We obtained rHBcAg_d-Fab e21 crystals that diffracted to 3.15 Å (space group: C₂, PDB ID: 6CWT, Table 2). The structure shows a 1:2 rHBcAg_d-Fab e21 complex consistent with that determined by AUC analysis (Fig. 1C and Fig. S3).

Previous structural comparisons of HBcAg_d in the capsid (PDB: 1QGT) and free soluble HBcAg_d (PDB: 3KXS) have shown substantial differences between these dimers, where the intradimer helical bundle in the free form was distorted and α5 adopted a conformation unfavorable for capsid formation (Zlotnick et al., 2013). This supports the view that an “allosteric change” of the HBcAg dimer interface is required for capsid assembly (Bourne et al., 2009). We compared our two rHBcAg_d structures with the capsid HBcAg_d (PDB: 1QGT) (Fig. S6). As noted above, in rHBcAg_d-scFv e13, the intradimer helical bundle was distorted (Fig. S6, left). In contrast, in the rHBcAg_d-Fab e21 structure the α3-α4 bundle had a very similar conformation to that of capsid HBcAg_d (Fig. S6, right). The difference between our two structures may be attributed to the high flexibility of the protein and to the binding of the different antibodies. In both structures, however, there was a substantial movement of α5 (~ 15 Å in HBcAg_d-scFv e13 and ~10-11 Å in HBcAg_d-Fab e21) which would likely prevent capsid assembly without invoking other conformational shifts.

Epitope-paratope Interfaces

The crystal structures of rHBcAg_d-scFv e13 and rHBcAg_d-Fab e21 show that scFv e13 and Fab e21 recognize different epitopes on rHBcAg_d (Fig. 1D and Fig. 2). On the other hand, the Fab e21 epitope overlaps with the Fab me6 epitope (21 of the 26 Fab me6 epitope residues, and of the 28 Fab e21 epitope residues, are identical (Table S1), which is in agreement with our previous biochemical data (DiMattia et al., 2013). Fab e21 binds to αl and α5 with its V_L CDRs (LCDR) and α5 with its V_H CDRs (HCDR) 2 and 3 (Fig. 2 and Table Sl). The interaction is primarily CDR mediated and only two non-CDR residues are found at the interface. Hot-spot residues (those residues identified by computational alanine scanning mutations that theoretically change the binding energy (ΔΔG) by more than 2 kcal/mol, as calculated with KFC2) (Darnell et al., 2007) are distributed mainly between LCDRs 1–3 and HCDR 3 (Table Sl), suggesting that these CDRs are required to maintain high-affinity binding between Fab e21 and rHBcAg_d. The scFv e13 interacts with α2, α4b and α5 primarily with its HCDRs (Fig. 2 and Table S2). There is also some weaker interaction between α2 and LCDRs 1 and 2 (Table S2). A non-CDR scFv loop region (I197-S201) also interacts with the HBcAg C-terminus but this interaction does not contribute significantly to the overall binding affinity (data not shown). In the case of scFv e13, hot-spot residues are concentrated in HCDR2 and mediate mainly hydrophobic interactions with α2 and α4b (Table S2). Analysis of the binding interfaces of rHBcAg_d with PISA (Krissinel and Henrick, 2007) shows a binding energy of −6.5 kcal/mol for the rHBeAg-e13 HCDR2 interface alone, which is half of the binding energy of the full rHBeAg-scFv e13 complex (−13.1 kcal/mol).

Figure 2. rHBcAg_d-scFv e13 and rHBcAg_d-Fab e21 Epitope-paratope Regions.

Overlaid structures of rHBcAg in complex with scFv e13 and Fab e21. rHBcAg is depicted as in Fig. 1D. The variable chains of scFv e13 and Fab e21 are represented as gray ribbons where the V_L CDRs are shown in yellow and labeled from L1-L3, and the V_H CDRs are shown in cyan and labeled from H1-H3. Individual residues of the epitopes and paratopes are given in Table S1 and S2. See also Figure S2.

Both biochemical and structural studies show that scFv e13 and Fab e21 epitopes are accessible on rHBcAg_d and rHBeAg_d. We checked whether these epitopes still remain accessible on capsid assemblies. In HBcAg_c, dimers are arranged in a pattern of fivefold vertices and quasi-sixfold vertices. Icosahedral HBcAg_c can have 90-dimer T = 3 or 120-dimer T = 4 symmetries (Schlicksup et al., 2018). The in vitro assembly of HBV capsids appears to proceed via the formation of a triangular trimer-of-dimers nucleus, followed by the further addition of dimers (Zlotnick et al., 1999). The HBcAg_c dimer-dimer interfaces are formed by weak interactions between α1 and α2 of HBcAg from one dimer and α2, α4b and α5 of HBcAg from the other dimer. These weak but multiple interactions eventually lead to a globally stable capsid (Ceres and Zlotnick, 2002). To compare the scFv e13 and Fab e21 epitopes with the capsid dimer-dimer interface, we overlaid rHBcAg-antibody fragment structures with the HBcAg_c structure (PDB ID: 1QGT) (Fig. 3). This structural comparison showed that both scFv e13 and Fab e21 epitopes are partially buried at the capsid dimer-dimer interface (Fig. 3A and B). Therefore, both antibody fragments can potentially compete with rHBcAg_d for binding to the next dimer and prevent capsid assembly. We calculated the capsid dimer-dimer interface surface using the Fast Atomic Density Evaluation (FADE) algorithm (Mitchell et al., 2001) on the KFC2 server to obtain a dimer-dimer interface surface map. The resulting map enabled us to determine which scFv e13 and Fab e21 CDR binding sites predominantly overlap with the dimer-dimer interface. With scFv e13, HCDR1 and HCDR2 project directly towards the dimer-dimer interface (Fig. 3C). With Fab e21, LCDR1 and LCDR3 predominantly project towards the interface while only some residues from HCDR2 and HCDR3 are engaged (Fig. 3D).

Figure 3. The e13 and e21 Epitopes Overlap with the HBV Capsid Dimer-dimer Interface.

(A) Overlaid structure of the rHBcAg-scFv e13 complex with a rHBcAg_c dimer-dimer interface (PDB ID = 1QGT). (B) Overlaid structure of the rHBcAg-Fab e21 complex with a rHBcAg_c dimer-dimer interface. The two interacting rHBcAg monomers from individual core dimers are shown as blue and gold ribbons. For clarity, the remaining rHBcAg monomers of each dimer are not shown. The variable chains of scFv e13 and Fab e21 are shown as pink semi-transparent surfaces. The rHBcAg_c dimer-dimer interfaces are indicated with arrows. (C) Locations of scFv e13 CDRs with respect to the rHBcAg_c dimer-dimer interface. (D) Locations of Fab e21 CDRs with respect to the rHBcAg_c dimer-dimer interface. The variable chains of scFv e13 and Fab e21 are shown as pink and yellow ribbons, respectively. The heavy chain CDRs are shown in orange and numbered from H1-H3, and the light chain CDRs are shown in green and numbered from L1-L3. The rHBcAg_c dimer-dimer interfaces, as calculated with FADE, are shown as blue semi-transparent surfaces.

Capsid Formation Inhibition by Antibody Fragments

HBcAg_c dimer-dimer interfaces have been targeted for HBV drug design by other groups. Thus, capsid assembly modulators (CAMs) have been identified (Venkatakrishnan et al., 2016; Wang et al., 2011). The heteroaryl-dihydropyrimidines (HAPs) induce formation of aggregated and aberrant capsid structures (Venkatakrishnan et al., 2016; Wang et al., 2011). They have also been shown to inhibit viral replication in vitro and in cell culture by interfering with pre-genomic RNA (pgRNA) encapsidation (Tan et al., 2015). Other drugs, such as phenyl-propenamide and sulfamoyl-benzamide derivatives, accelerate formation of capsid-like particles (Campagna et al., 2013; Katen et al., 2013). These two different classes of CAMs target the same hydrophobic pocket located at the capsid HBcAg_c dimer-dimer interface (Katen et al., 2010; Klumpp et al., 2015; Qiu et al., 2016; Zhou et al., 2017).

As both scFv e13 and Fab e21 bind to rHBcAg_d with nano- to pico-molar affinity, and their epitopes lie on the HBcAg_c dimer-dimer interface, we determined their ability to prevent capsid assembly. For this, we used an in vitro system where the capsid assembly from soluble HBcAg_d or HBeAg_d could be controlled by simply exchanging buffer conditions. rHBcAg is soluble and dimeric at pH 9.5 and low ionic strength. The soluble rHBcAg_d represent HBcAg_d which transiently exist in the cytoplasm during the viral replication cycle. Fig. 4A shows soluble rHBcAg_d as visualized by negative stain electron microscopy. Upon shifting to quasi-physiological conditions (pH 7.5 and 250 mM NaCl), rHBcAg_d assemble as rHBcAg_c (Fig. 4B). In the presence of scFv e13 or Fab e21, rHBcAg_c assembly is totally inhibited under the same conditions (Fig. 4C and D, respectively). The corresponding experiments with rHBeAg_d, which rearranges to the assembly competent rHBcAg_d-like form upon reduction and subsequently forms core-like capsids, are shown in Fig. 5. In the case of rHBeAg, the switch between the soluble dimers and the core-like capsids is controlled by the redox state and therefore, the experiments were performed under reducing/oxidizing physiological conditions. Addition of either antibody fragment to pre-assembled rHBcAg_c or rHBeAg_c did not cause the capsids to fall apart since the epitopes are partially buried, preventing access of the antibody fragments to their respective binding sites (Fig. S7). Next, we confirmed the inhibition of HBcAg capsid assembly in the presence of antibody fragments by determining the sedimentation velocity of immune complexes under assembly and non-assembly conditions. At pH 9.5 and low ionic strength, rHBcAg is dimeric with a sedimentation coefficient (s) of ~ 2.4 (Fig. 6A). At pH 7.5 and 250 mM NaCl, assembly occurs, and the protein has s values of 42–45, corresponding to rHBcAg_c with ~ 10% unassembled dimer remaining (Fig. 6B) (Zlotnick et al., 1996). The immune complex rHBcAg_d-scFv e13 had the same s value (4.3) under both assembly and non-assembly conditions, indicating no change in mass or shape, consistent with the prevention of capsid formation (Fig. 6C and 6D). Similarly, we measured the sedimentation coefficients of rHBcAg and rHBeAg complexed with Fab e21 under assembly conditions (Fig. 7A and 7B). Both complexes were homogenous (single boundaries) with s values of 5.7 and 5.6, respectively. The absence of high molecular weight material indicated that the Fab effectively blocked assembly. In the absence of Fab e21 and under reducing conditions we observed a 40% conversion of HBeAg_d to core-like capsids (data not shown). The very similar s values of the soluble HBcAg_d and HBeAg_d immune complexes indicate that they have similar molecular compositions.

Figure 4. Inhibition of rHBcAg_c Formation by scFv e13 and Fab e21.

Negative-stain EM of (A) rHBcAg_d, (B) rHBcAg_c, (C) rHBcAg_d-scFv e13, and (D) rHBcAg_d-Fab e21. (A) Nonassembly conditions (pH 9.5), (B-D) capsid assembly conditions (pH 7.5 and 250 mM NaCl). Antibody fragments alone, not shown. All images are at the same magnification. Scale bar, 50 nm. See also Figure S7 and Figure S8.

Figure 5. Inhibition of rHBeAg_c Formation by scFv e13 and Fab e21.

Negative-stain EM of (A-B) rHBeAg_d, (C-D) rHBeAg_d-scFv e13, and (E-F) rHBeAg_d-Fab e21, where (A, C, and E) were oxidized, and (B, D, and F) were reduced. All were under capsid assembly conditions. All images are at the same magnification. Scale bar, 50 nm. See also Figure S7 and Figure S8.

Figure 6. Sedimentation Velocity Analysis of rHBcAg Alone and in Complex with scFv e13.

Raw absorbance scans of (A) rHBcAg at pH 9.5, (B) rHBcAg at pH 7.5 and 250 mM NaCl, (C) rHBcAg-scFv e13 at pH 9.5, and (D) rHBcAg-scFv e13 at pH 7.5 and 250 mM NaCl. All analyses were performed at 40,000 rpm except for rHBcAg at pH 7.5 and 250 mM NaCl, which was performed at 25,000 rpm. The red arrow indicates the direction of sedimentation from top (meniscus) to bottom of the analytical cells. At pH 9.5 the rHBcAg is dimeric with a sedimentation coefficient (s) of ~ 2.4 whereas at pH 7.5 and 250 mM NaCl (assembly conditions) the protein has s values of 42 – 45, corresponding to capsids [45]. Approximately 1015% unassembled protein is also present. The rHBcAg-scFv e13 complex, previously resolved by gel filtration at pH 9.5, has an s value of 4.3. The same complex under assembly conditions also has an s value of 4.3. See also Figure S7 and Figure S8.

Figure 7. Sedimentation Velocity Analysis of rHBcAg and rHBeAg in Complex with Fab e21.

Raw absorbance scans of the (A) rHBcAg-Fab e21 and (B) rHBeAg-Fab e21 complexes, both at pH 7.5 and 250 mM NaCl. All analyses were performed at 40,000 rpm. The red arrow indicates the direction of sedimentation from top (meniscus) to bottom of the analytical cells. Despite being under assembly conditions the rHBcAg-Fab e21 and rHBeAg-Fab e21 complexes remain soluble, with sedimentation coefficient (s) of ~ 5.7 and 5.6, respectively. See also Figure S7 and Figure S8.

In the preceding experiments, we mixed the rHBcAg_d and rHBeAg_d with a saturating amount of scFv/Fab to demonstrate that the fully formed complexes were assembly incompetent (Figs. 4 - 7). When antibody, specifically Fab e21, is mixed with excess rHBcAg_d we see the formation of a mixture of partially and fully saturated rHBcAg_d, where partial saturation refers to only one subunit of a rHBcAg_d binding a Fab (Fig. S8, peak 3). These unsaturated rHBcAg_d are also assembly incompetent. At a high rHBcAg_d:Fab e21 ratio (5:1), the rHBcAg_d not bound to Fab form rHBcAg_c as normal (Fig. S8, peak 1). In these experiments, we only see assembled rHBcAg_c, immune complexes and small amounts of free rHBcAg_d (Fig. S8). We do not detect any Fab e21 associated with the rHBcAg_c by SDS-PAGE (Fig. S8), as also observed previously by electron microscopy (Fig. S7). These results are consistent with Fab e21 binding with high affinity to the rHBcAg_d - rHBcAg_d assembly interface and completely abrogating rHBcAg_c assembly. The Fab also inhibits assembly of excess rHBcAg_d, unless the rHBcAg_d is present it great excess (Fig. S8).

Conclusions

The severe clinical consequences of CHB infections and the problems associated with current treatment regimens emphasize the need for better therapeutics. The HBcAg and HBeAg amino acid sequences 1–149 are identical but, due to the additional N-terminal PP of the HBeAg and the C-terminal NABD of HBcAg, these proteins play very different roles during the viral replication cycle. HBcAg_c assembly is required for virion formation, completion of genome duplication, and transfer of the viral genome between cells. HBeAg, on the other hand, is an immune-modulator that acts on multiple host immune pathways and plays an important role in progression of CHB. Therefore, it is another important target for anti-HBV drug design. Although there are promising drug candidates that can act on either HBcAg or HBeAg (Campagna et al., 2013; Katen et al., 2013; Lanford et al., 2013; Venkatakrishnan et al., 2016; Wang et al., 2011), currently there are no known candidates that might act on both proteins simultaneously. Antibodies directed against the same epitopes on the distinct dimeric surfaces of HBcAg and HBeAg can potentially have dual action against the virus. Chimeric rabbit/human Fabs consist of rabbit variable domains and human constant domains. Such antibodies often have both high affinity and specificity and, as they can be fully humanized, which increases their tolerability and half-life, they can have therapeutic potential. In this study, we identified two chimeric antibody fragments, scFv e13 and Fab e21, that can bind to distinct epitopes, on both HBcAg and HBeAg. Our in vitro studies have also shown that these antibody fragments can effectively block both HBcAg_d and HBeAg_d association. Hence, these antibody fragments have the potential to interfere with both the production of virions and with the immune-modulation by HBeAg. Figure 8 summarizes some of the viral replication pathways that can potentially be inhibited by using Mab, Fab, or scFv versions of e13 and e21. When targeted to the cytoplasm, as either Fab or scFv, these antibodies could bind to transiently existing HBcAg_d and thereby prevent capsid assembly. This can block the downstream formation of Dane particles, which is essential for the transport of viral genomes and infection of new hepatocytes (Fig. 8A). The same antibody fragments can also bind to cytoplasmic HBeAg_d and prevent the assembly of core-like “decoy” particles, which are thought to exhaust the immune system, in two ways. First, by locking reduced HBeAg in transition state intermediates preventing formation of core-like HBeAg_d (Figure 8B, left); supporting this, we obtained rHBeAg-scFv e13 crystals under reducing conditions where HBeAg was extensively unfolded (~ 50%) suggesting that any further conformational shift was effectively blocked by the high-affinity scFv (Eren, unpublished). Second, by binding to any core-like HBeAg_d that do form, thereby preventing assembly of the decoy particles (Figure 8B, right). In the case of both HBcAg_c and HBeAg_c, pairs of high-affinity Fab/scFv capable of binding to the two assembly faces of the dimer simultaneously, such as e13 and e21, should prove to be particularly effective antivirals as the need for both escape mutations effectively preventing binding and compensatory mutations still allowing correct assembly presents a four-fold barrier that must be overcome for replication to take place.

Figure 8. A Schematic of Viral Replication Pathways that can Potentially be Inhibited by Fab/scFv or Mab Forms of e13 and e21.

(A) Inhibition of HBcAg_c assembly in the cytoplasm. The scheme shows that antibody fragments e13 and e21 can bind to transiently existing HBcAg_d before they can assemble into HBcAg_c in the cytoplasm, which in turn can block the downstream formation of infectious Dane particles, preventing transport of the viral genome and infection of healthy cells. (B) Inhibition of HBeAg_c assembly in the cytoplasm. Cell-penetrating antibody fragments of e13 and e21 can bind to reduced HBeAg_d, preventing its conversion into the HBcAg_d-like arrangement (left), or binding to any HBeAg_d that has already converted to the assembly competent arrangement, blocking the assembly of decoy particles (right). (C) Inhibition of cytoplasmic HBeAg_d-mediated immune-modulation by binding to HBeAg and blocking its interaction with host proteins such as Mal and TRAM. (D) Inhibition of circulating HBeAg_d-mediated immune-modulation by binding of Mab versions of e13 and e21, which could trigger clearance of HBeAg through antibody-dependent mechanisms. In this scheme, HBcAg_d and HBeAg_d are represented as gray cartoons and antibody fragments as magenta cartoons. Disulfide bonds are represented by yellow lines while the non-bonded Cys residues are shown as yellow dots. The PP is represented by a red undulating line. Red crosses on blue arrows indicate pathways that can potentially be inhibited by antibody fragments.

In addition to capsid assembly prevention (as demonstrated here) these antibody fragments may also block HBeAg mediated immune-modulation through additional pathways as summarized in Figures 8C and D. For instance, antibody fragments may block the interaction of cytoplasmic HBeAg with host proteins such as Mal and TRAM, preventing suppression of TIR-mediated induction of inflammatory transcription factors such as NF-κB (Figure 8C) (Lang et al., 2011). In addition to the intracellularly targeted Fab and scFv, fully humanized Mab forms of e13 and e21 could facilitate the direct clearance of the circulating HBeAg via antibody-dependent mechanisms (Figure 8D). Finally, antibody-HBeAg complexes could bind to Fc receptors that are expressed by immune effector cells that can trigger a multitude of innate and adaptive responses against HBV, neutralizing the immune-modulatory effect of HBeAg.

STAR Methods

Reagent and resource sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Paul T Wingfield (wingfiep@mailnih.gov).

Experimental Model and Subject Details

All proteins were expressed in E. coli BL21-CodonPlusRIL cells (Agilent) in Luria Broth in a 2 L fermenter (Sartorius Stedim) operated at 37°C.

Method Details

Purification of rHBcAg_d and rHBeAg_d

The expression in E. coli and purification of HBV proteins were performed as previously described (Zhuang et al., 2017). The construct Cp1–149.C48A.C107A, was used to produce the capsid assembly domain rHBcAg (Figure S1) and the construct Cp(−10) 149 C48A.C107A to prepare rHBeAg (Figure S1). Briefly, both proteins were expressed in E.coli and cell pastes (~50 g) were resuspended in 50 mM Tris-HCl (pH 7.5), 2 mM EDTA, 2 mM DTT (Buffer A). Cell suspensions were lysed with a French Press followed by 2 minutes of sonication. Suspensions were centrifuged at 25,000 × g for one hour. At this stage, HBcAg was in the supernatant and HBeAg in the pellet. HBcAg was concentrated by precipitation with 40% (NH4)₂SO₄. The pellet was dissolved in 100 mM NaHCO₃ (pH 9.5) (Buffer B) containing 3 M urea, then centrifuged at 25,000 × g for 2 hours at 4 °C. The supernatant was applied to a Superdex 200 column (GE Healthcare) equilibrated with Buffer B. The pooled fractions from the Superdex 200 column were loaded onto a Q-Sepharose column (GE Healthcare) equilibrated with 50 mM Tris-HCl, pH 8.0 (Buffer C). After a two-column volume wash with Buffer C, a linear gradient from 0 to 1 M NaCl in Buffer C was applied to elute the protein. Pooled fractions were titrated to 9.5 with NaOH and urea was added to a final concentration of 3 M. The rHBcAg was then re-chromatographed on a Superdex 200 column equilibrated in Buffer B as described above. Purified rHBcAg_d was assembled into capsids by dialysis against 50mM Tris-HCl, pH 7.5, containing 250 mm NaCl (Zhuang et al., 2017).

rHBeAg pellets from the initial cell fractionation (see above), were homogenized in Buffer B containing 3 M urea and processed by gel filtration and ion-exchange as described for rHBcAg. To mediate complete oxidation, CuCl₂ was added (1 μM), followed by incubation for 30 minutes at 4 °C and then 5 mM EDTA was added. The protein solution was applied to a Superdex S200 column equilibrated in Buffer B.

All protein samples were checked by SDS-PAGE under reducing (DTT in sample buffer) and oxidizing conditions (no DTT). Purified and reduced proteins ran with expected monomeric molecular weights of 16,706 (rHBcAg) and 17,194 (rHBeAg). Under oxidizing conditions, rHBcAg runs as a dimer (due to intermolecular disulfide C61-C61) and HBeAg runs as monomer with a slightly faster mobility (lower apparent mass) than reduced protein due to intramolecular disulfide C(−7)-C61 (Zhuang et al., 2017).

Purification of Antibody Fragments

The scFv e13 and Fab e21 antibodies were expressed in E. coli and purified as previously described (Zhuang et al., 2017). Briefly, the expression plasmids for Fab or scFv production were transfected into E. coli and the resulting transfectants were grown in a 1-L fermenter. Bacterial cultures were clarified by centrifugation at 14,000 × g for 1 h. The secreted antibody in the supernatant had a Carboxyl-terminal His-tag and was captured using Ni-Sepharose resin (GE Healthcare) at 4 °C for 1 – 2 h, washed with PBS (pH7.4) plus 20 mM imidazole. Antibodies were eluted with PBS plus 0.5 M imidazole and then dialyzed against 25 mM HEPES buffer (pH7.4), 0.15 M NaCl, 0.2 mM TCEP, 10% glycerol. To maintain solubility during purification, ~ 1 M urea was often included in buffers. In the studies described, the Carboxyl-terminal His-tag was not removed.

Kinetics of Antibody Binding Using Surface Plasmon Resonance

All experiments were performed on a Biacore ×100 (GE Healthcare) instrument at 25 °C. HBS-EP (10 mM HEPES, pH 7.4, 150 mM sodium chloride, 3 mM EDTA, 0.05% Polysorbate 20) was used as the running buffer and data were analyzed using Biacore ×100 evaluation software (GE Healthcare). Cell 1 was left untreated to serve as a reference surface and cell 2 was used as the experimental surface. Fabs and scFvs were diluted (10 – 20 μg/ml) in 10 mM sodium acetate buffer (pH 4.5 – 5.0) and immobilized on CM5 sensor chips by the standard amine coupling method (Amine Coupling kit, GE Healthcare) at a flow rate of 5 μl/min. The immobilization levels of the proteins on the sensor chip surfaces were approximately 1500 RU. For kinetic analysis, analytes were prepared by serial dilution with HBS-EP buffer over a range of typically 10 nM – 1 μM and injected over both the reference and experimental surfaces at a flow rate of 30 μl/min. Sensor chips were regenerated by a 60-s injection of 50 mM sodium hydroxide.

Preparation of Immune Complexes for Structural Studies

A two-fold molar excess of Fab or scFv was mixed with rHBcAg_d or rHBeAg_d and the mixture was then applied to a Superdex S200 column equilibrated in 100 mM sodium bicarbonate, pH 9.5. The column fractions were monitored by SDS-PAGE and the immune complex identified (usually the main peak). This was used for characterization and for crystallization screening.

Analytical Ultracentrifugation

A Beckman Optima XL-I analytical ultracentrifuge, absorption optics, an An-60 Ti rotor and standard double-sector centerpiece cells were used. Equilibrium measurements were made at 20 °C at 11,500 rpm for Fab complexes and 14,500 rpm for scFv complexes. The ternary complex of rHBeAg + scFv me6 + Fab e13 was measured at 10,000 rpm. Concentration profiles were recorded every 4 h for 16 h and then baselines were established by over-speeding at 45,000 rpm for 3 h. Data (the average of 8 – 10 scans collected using a radial step size of 0.001 cm) were analyzed using the standard Optima XL-I data analysis software v6.03. Sedimentation velocity measurements at 20 °C were made at 40,000 rpm with data collection every 5 min to 3 h. Measurements on capsids were made at 25,000 rpm. As indicated in text and Figure legends, all analytical centrifugation determinations were either performed at pH 9.5 (non-assembly conditions) or at pH 7.5 (assembly conditions) and the buffers used were: 100mM sodium bicarbonate, pH 9.5; 50mM Tris-HCl, pH 7.5, 250mM NaCl (plus or minus 5mM DTT).

Crystallization of rHBcAg-scFv e13, rHBeAg-scFv e13 and rHBcAg-Fab e21 Immune Complexes

Immune complex buffer was exchanged to 20 mM Tris, 50 mM NaCl, pH 8.0 by dialyzing overnight at 4°C. All crystals were obtained using the hanging drop vapor diffusion technique by mixing 1 μl of immune complex with 1 μl of mother liquor. rHBcAg-scFv e13 thin plate crystals appeared in 3 weeks in 12% (w/v) PEG 20000, 0.1M Imidazole, pH 7.0 (PEG RX [D11], Hampton Research). rHBeAg-scFv e13 thin plate crystals appeared within 1 week in 15% (w/v) PEG 550 MME, 0.1 M MES, pH 6.5 (ProPlex [A5], Molecular Dimensions). rHBcAg-Fab e21 crystals appeared in 4 weeks as square thick plates in 20% (w/v) PEG 3350, 0.2 M Potassium thiocyanate, 0.1 M Bis-tris propane, pH 6.5 (PACT Premier [F4], Molecular Dimensions). The crystals were protected with 30% glycerol before flash-freezing in liquid nitrogen. All data collections were carried out at the Advanced Photon Source (APS), Lemont, IL beamline 22-ID-D using the high-speed Mar300HS detector. Data were collected from multiple crystals in each case and the final structures were obtained using data from either best diffracting individual crystals or merged data from multiple crystals using BLEND (Foadi et al., 2013).

Data processing and integration was done with XDS (Kabsch, 2010) and scaling was done with XSCALE (Kabsch, 2010) for rHBcAg-scFv e13 and rHBeAg-scFv e13 crystals. Data processing and integration was done with HKL2000 (Otwinowski, 1997) and scaling was done with SCALEPACK (Otwinowski, 2006) for rHBcAg-Fab e21 crystals. Molecular replacement was carried out with Phaser (McCoy et al., 2007) using PDB 2KH2 (Wilkinson et al., 2009) as an scFv e13 search model, PDB 4D3C (Tolbert et al., unpublished) as an Fab e21 search model, and PDB 3V6Z (DiMattia et al., 2013) as a rHBcAg and rHBeAg search model. Final structures were obtained after multiple rounds of model building with Coot (Emsley et al., 2010) and refinement with Phenix Refine (Afonine et al., 2012). For refinement, XYZ coordinates, real-space, individual B-factors, occupancies and TLS strategies were applied. For refinement of rHBcAg-scFv e13 and rHBeAg-scFv e13 structures one TLS group was assigned per macromolecule. For refinement of rHBcAg-Fab e21 the TLS groups were automatically assigned using Phenix. The crystallographic data and refinement statistics are listed in Table 2. The statistics were validated using MolProbity (Chen et al., 2010). Figures involving structures were prepared using Chimera (Pettersen et al., 2004). RMSD values were calculated using PyMol (DeLano, 2002).

Negative Stain Electron Microscopy

Samples (0.05 mg/ml with respect to rHBcAg or rHBeAg) were applied to carbon-coated grids glow-discharged (Fischione) immediately prior to use in a plasma of 25% oxygen and 75% argon. Grids were rinsed with distilled water and negatively stained with 1% uranyl acetate. Micrographs were recorded on a CCD camera at 35,000× nominal magnification with a FEI CM120 electron microscope. For rHBcAg_c and rHBeAg_c samples the grids were coated with 0.1 % poly-lysine to prevent capsid aggregation.

Assembly of rHBcAg_c in the Presence of Varying Amounts of Fab e21

Fab e21 in 25 mM HEPES, pH 7.5, 0.15 M NaCl and rHBcAg_d in 50 mM sodium bicarbonate, pH 9.5 were mixed in various molar ratios Fab:rHBcAg_d (1:1, 1:2 and 1: 5), then added to an equal volume of 25 mM Tris-HCl, pH 7.5, 250 mM NaCl (assembly buffer). A control of rHBcAg_d was processed in same manner. Samples were concentrated to 1 ml and loaded on a Superdex 200 column (1.0 cm × 30 cm) equilibrated in assembly buffer and 1 ml fractions were collected.

Quantification and Statistical Analysis

Surface Plasmon Resonance

Signals from the reference surface and an ensemble of buffer blank injections were subtracted to correct for nonspecific binding and injection artifacts. The corrected results were globally fitted to a 1:1 binding model and the association rate constant (k_a), and dissociation rate constant (k_d), were used to determine the equilibrium dissociation constant (K_d) in units of M.

Analytical Ultracentrifugation

Protein partial specific volumes (ν-bar), calculated from the amino acid compositions, and solvent densities were estimated using the program SEDNTERP (http://www.rasmb.bbri.org/). Data analysis was done using DCDT+ 2.4.3 (Philo, 2006).

Data and Software Availability

The final structures were deposited in the RCSB PDB with the accession codes: 6CWD (rHBcAg-scFv e13), 6CVK (rHBeAg-scFv e13), and 6CWT (rHBcAg-Fab e21).

Highlights.

• HBcAg and HBeAg are 77% colinear but differ in quaternary structure and function

• Humanized antibody fragments e13 and e21 have separate epitopes, on both proteins

• e13/e21 block HBcAg and HBeAg self-assembly, and likely host protein interactions

• Mab/Fab/scFv forms of e13 and e21 have high potential as anti-HBV therapeutics