Expression, purification, crystallization and preliminary X-ray crystallographic studies of hepatitis B virus core fusion protein corresponding to octahedral particles

Novel hepatitis B virus-like particles of recombinant dimeric core–GFP fusion protein were expressed, purified and crystallized. The crystals diffracted to 2.15 Å resolution and belonged to space group F432, with unit-cell parameters a = b = c = 219.7 Å.

Abstract

Recombinant hepatitis B virus core proteins dimerize to form building blocks that are capable of self-assembly into a capsid. A core capsid protein dimer (CPD) linked to a green fluorescent protein variant, EGFP, at the C-terminus has been designed. The recombinant fusion CPD was expressed in Escherichia coli, assembled into virus-like particles (VLPs), purified and crystallized. The single crystal diffracted to 2.15 Å resolution and belonged to the cubic space group F432, with unit-cell parameters a = b = c = 219.7 Å. The fusion proteins assembled into icosahedral VLPs in aqueous solution, but were rearranged into octahedral symmetry through the crystal-packing process under the crystallization conditions.

1. Introduction  

Viral particles are increasingly utilized as nanoplatforms for diverse applications in materials science and biomedicine (Douglas & Young, 2006 ▶; Singh et al., 2006 ▶). One such application which has received much attention is that of virus particles as nanocarriers (Choi et al., 2011 ▶).

The hepatitis B virus (HBV) capsid is composed of dimers of a core protein (183 amino acids, ∼20 kDa; Tiollais et al., 1981 ▶). The core protein forms icosahedral capsids of two sizes, which correspond to triangulation numbers of T = 3 and T = 4 and consist of 180 and 240 monomeric subunits, respectively (Crowther et al., 1994 ▶). The recombinant core protein consisting of amino acids 1–149 (HBc149, 14 kDa) also self-assembles into icosahedral capsids and the structure of the recombinant T = 4 capsid has been solved by X-ray crystallography at 3.3 Å resolution (Wynne et al., 1999 ▶). HBV-like particles have been used as a molecular platform. An HBc149 fusion protein also assembled into virus-like particles (VLPs) that can display heterologous sequences from the surface-exposed c/e1 epitope (Skamel et al., 2006 ▶; Vogel et al., 2005 ▶) or the C-terminus (Koletzki et al., 1997 ▶; Choi et al., 2011 ▶). However, when a larger protein was fused to the monomeric HBc149 the fusion proteins could not assemble into VLPs presumably owing to steric hindrance. For example, the HBc149 fusion protein with a C-terminal Staphylococcus aureus nuclease (17 kDa) can assemble into unstable VLPs (Beterams et al., 2000 ▶), whereas a GFP (green fluorescent protein) (27 kDa) that is fused to the C-terminus of HBc149 cannot assemble into VLPs (Vogel et al., 2005 ▶).

In order to accommodate a larger protein, we designed a core protein dimer (CPD) that contained two core proteins linked to each other by a flexible peptide linker. A heterologous protein was then fused at the C-terminus of the CPD so that it could be encapsulated in VLPs. As a nanocarrier, the desired fusion proteins should be able to assemble into VLPs and the structure of the heterologous proteins should not be affected by fusion to the CPD. In order to validate these requirements, we used EGFP (enhanced green fluorescent protein) as a heterologous protein. The structure of homologous GFP (PDB entry 1gfl; Yang et al., 1996 ▶) has been solved by X-ray crystallography. Thus, based on the structural stability revealed by crystallography and a fluorescence spectrum, we can confirm whether the heterologous protein retains the function.

We constructed recombinant plasmids that encoded the CPD fused to EGFP (Fig. 1 ▶). In addition, we inserted a cysteine codon at the end of the HBc149 gene to enhance the structural stability of VLPs (Zlotnick et al., 1997 ▶), because VLPs of fusion proteins are known to show high structural plasticity caused by conformational stress (Böttcher et al., 2006 ▶).

Figure 1.

A schematic view of the construct.

Here, we report that the recombinant fusion CPD was expressed in Escherichia coli and assembled into VLPs that were indistinguishable from the wild-type T = 4 icosahedral capsids. The crystal diffracted to 2.15 Å resolution.

2. Materials and methods  

2.1. Site-directed mutagenesis and construction of plasmids  

A plasmid (PT7-SC vector) encoding HBcΔ-CW, which is truncated at residue 149, was generously donated by Dr A. G. W. Leslie (MRC Laboratory of Molecular Biology, Cambridge, England). All mutagenesis procedures were performed using a site-directed mutagenesis PCR method (Higuchi et al., 1988 ▶). HBcΔ-CW had three cysteine residues at positions 48, 61 and 107. These cysteines were replaced by alanine. A cysteine was then added to the core protein mutant at position 150 (HB150). The HB150-GGSEEEGGS gene sequence, which included the sequence of the peptide linker, was amplified by PCR and subcloned into the NcoI–BamHI sites of the pET-28a(+) plasmid (Novagen). In order to construct the CPD sequence, the PCR product of the coding sequence for (GGS)₆-HB150-GGSKL was subcloned into the BamHI–HindIII sites. In addition, the EGFP gene was amplified by PCR using the pEGFP-C1 vector (Clontech Laboratories; GenBank accession No. U55763) and the coding sequence of EGFP, which was truncated at residue 229, was subcloned into the HindIII–XhoI sites. A schematic representation of the fusion protein is shown in Fig. 1 ▶.

2.2. Expression and purification  

The recombinant plasmid was transformed into the Arctic Express (DE3) RIL E. coli strain (Agilent Technologies). Single-cell colonies were grown in 5 ml LB medium that contained kanamycin (30 µg ml⁻¹) and gentamycin (20 µg ml⁻¹) at 310 K with shaking at 250 rev min⁻¹ overnight. 5 ml of this culture was then transferred into 1 l LB medium and the cells were grown at 303 K for 3 h. Fusion protein expression was then induced by adding 0.1 mM isopropyl β-d-1-thiogalactopyranoside at 284 K for 24 h. The cells were harvested by centrifugation at 4000 rev min⁻¹ for 15 min at 277 K. The bacterial cell pellets were suspended in 60 ml lysis buffer (50 mM Tris–HCl pH 8.0, 10 mM EDTA, 200 µg ml⁻¹ phenylmethylsulfonyl fluoride, 1 mg ml⁻¹ lysozyme) and then sonicated on ice. Cellular debris and insoluble protein were separated by centrifugation at 15 000 rev min⁻¹ for 20 min. Ammonium sulfate was slowly added to the supernatant to a final concentration of 30% saturation for 30 min on ice. The solution was centrifuged at 15 000 rev min⁻¹ for 15 min. The pellet was resuspended in 20 ml 50 mM Tris–HCl buffer pH 7.5, 150 mM NaCl containing 0.1% Triton X-100 and then loaded onto a Sepharose CL-4B column (GE Healthcare). Fractions that contained the VLPs were pooled on the basis of EGFP fluorescence intensity, precipitated with 25% ammonium sulfate and centrifuged. The pellet was resuspended in 3 ml 50 mM Tris–HCl buffer pH 7.5, 300 mM NaCl. The VLP suspension was layered onto a 5–30% sucrose step gradient and sedimented by ultracentrifugation (Hitachi P28S2 rotor) to separate the VLPs and disassemble fusion proteins. 20 fractions were collected from the top (fraction 1) to the bottom (fraction 20) of the centrifuged tubes using a Gradient Master Station (Biocomp Instruments). The fluorescence intensities of the fractions were then measured. The samples from each fraction were analysed by SDS–PAGE and the molecular mass of the expressed fusion protein was determined (55 kDa). The VLP solutions were dialysed against 5 mM Tris–HCl buffer pH 7.5, 150 mM NaCl and were concentrated prior to setup in a protein crystallization experiment.

2.3. Electron microscopy  

To confirm the assembly of the VLPs, the purified CPD fusion proteins and wild-type core proteins were applied onto carbon-covered copper mesh grids and stained with 2% aqueous uranyl acetate. The grids were examined under a Hitachi H-7600 transmission electron microscope operated at 80 kV.

2.4. Crystallization and data collection  

Prior to the crystallization experiments, the protein concentration was adjusted to 10 mg ml⁻¹. The fusion protein crystallization was carried out by a hanging-drop vapour-diffusion method in which 2 µl droplets were equilibrated against 500 µl reservoir solution in a 24-well plate. Using the crystallization condition based on the reported conditions for HBcΔ-CW (Wynne et al., 1999 ▶), the VLPs were screened using 0.5–4% PEG 20000, 0.4–2 M ammonium sulfate, 0.1 M MES pH 6.0–7.5 at 292 K. The obtained crystals were observed using an IX71 fluorescence microscope (Olympus) after three rinses with the reservoir solution.

A single crystal was mounted on a cryoloop (Hampton Research) and flash-cooled in a 100 K nitrogen-gas stream after soaking in a cryoprotectant solution [22%(v/v) 2,3-butanediol in crystallization buffer]. After preliminary X-ray experiments using an R-AXIS VII system (Rigaku), a data set was collected on beamline BL17A at the Photon Factory, Tsukuba, Japan. The data were then processed using the iMOSFLM (Battye et al., 2011 ▶) and CCP4 (Winn et al., 2011 ▶) program suites. The POINTLESS package (Evans, 2006 ▶) was used for the initial space-group prediction. The average intensities were converted to structure-factor amplitudes by TRUNCATE (French & Wilson, 1978 ▶). A self-rotation function was performed using the MOLREP program (Vagin & Teplyakov, 2010 ▶).

3. Results and discussion  

SDS–PAGE analysis of the fusion protein revealed a single band that migrated at approximately 55 kDa after sucrose density-gradient ultracentrifugation (Fig. 2 ▶). Measurement of the fluorescence intensity derived from EGFP indicated that the fusion protein was present in fractions 1, 2 and 12–17. Based on previous reports of ultracentrifugation (Zlotnick et al., 1999 ▶), it is considered that the dis­assembled fusion proteins and soluble proteins from E. coli were found in fractions 1 and 2, which were at the top of the gradient, whereas fractions 12–17 contained the self-assembled VLPs. We confirmed the presence of VLPs in fractions 12–17 by TEM (transmission electron microscopy). The outer diameter (∼350 Å) of the VLPs of CPD fusion proteins was indistinguishable from that of the wild-type T = 4 icosahedral capsids. However, the TEM images showed a difference in their inner diameters (Fig. 3 ▶).

Figure 2.

SDS–PAGE analysis of aliquots from the 5–30% sucrose step gradient. 20 fractions were collected from the top (lane 1) to the bottom (lane 20). The gel was stained with Coomassie Brilliant Blue R-250. Lane M, molecular-mass marker (labelled in kDa); lanes 1 and 2 (fractions 1 and 2), disassembled core protein dimer (CPD) fusion proteins and soluble proteins from E. coli; lanes 12–17 (fractions 12–17), CPD fusion proteins that assembled into VLPs.

Figure 3.

Electron micrographs of VLPs from solution after sucrose-gradient ultracentrifugation and fractionation. The scale bar is 300 Å in length. (a) VLPs assembled by CPD fusion proteins with EGFP. (b) Empty VLPs assembled by HBc149.

The fractions that contained VLPs were pooled, dialysed and concentrated to 10 mg ml⁻¹. Crystals of diffraction quality (approximately 0.1 × 0.1 × 0.1 mm) were obtained in 3 months under the following conditions: 0.5–1% PEG 20 000, 1.0–1.4 M ammonium sulfate, 0.1 M MES pH 6.5 with a 2–4 mg ml⁻¹ protein concentration (Fig. 4 ▶ a). The crystals were rinsed with reservoir solution and the fluorescence derived from the EGFP was confirmed by fluorescence microscopy (Fig. 4 ▶ b). The crystals belonged to the cubic space group F432, with a unit-cell parameter of 219.7 Å. A summary of all X-ray crystallographic data-collection statistics is provided in Table 1 ▶.

Figure 4.

Light and fluorescence micrographs of the CPD fusion protein crystals. The scale bar is 0.1 mm in length. (a) Bright-field micrograph of the crystal. (b) Bright-field/epifluorescence micrograph of the crystal. The following fluorescence filters were used: excitation, BP460–495; emission, BA510–550; dichromatic mirror, DM505.

Table 1. X-ray diffraction data and processing statistics.

Values in parentheses are for the highest-resolution shell.

Space group	F432
Unit-cell parameters (Å, °)	a = b = c = 219.7, α = β = γ = 90
Matthews coefficient (Å3 Da−1)	2.01
Solvent content (%)	38.8
Temperature (K)	100
Wavelength (Å)	0.9800
Crystal-to-detector distance (mm)	280
Resolution range (Å)	50.4–2.15 (2.22–2.15)
No. of observed reflections	512833 (36388)
No. of unique reflections	25183 (2138)
Multiplicity	20.4 (17.0)
Completeness (%)	99.9 (99.9)
R merge † (%)	8.5 (37.0)
Mean I/σ(I)	25.9 (8.0)

^† R _merge is the quality in the scaled data set (see Diederichs & Karplus, 1997 ▶). .

In space group F432 there are 96 asymmetric units that are related by 24 different rotations and four translations in a unit cell. The 24 rotations of the asymmetric units correspond to the octahedral symmetric operation. In addition, the results from the self-rotation function map using data between 5 and 10 Å resolution and an integration radius of 100 Å showed pronounced peaks which indicated twofold (χ = 180°), threefold (χ = 120°) and fourfold (χ = 90°) axes that are consistent with octahedral symmetry expected for that space group (Figs. 5 ▶ a, 5 ▶ b and 5 ▶ c). Furthermore, there was no fivefold (χ = 72°) axis characteristic of icosahedral symmetry (data not shown). The Matthews coefficient calculation suggested the presence of one CPD fusion protein molecule per asymmetric unit (with a molecular weight of 55 kDa). The calculated Matthews coefficient and the solvent content were 2.01 ų Da⁻¹ and 38.8%, respectively (Matthews, 1968 ▶). Thus, there are four octahedra in the F432 unit cell. A scheme of the CPD subunit organization in the F432 unit cell is shown in Fig. 5 ▶(d).

Figure 5.

Self-rotation function versus (θ, ψ) for χ = 180° (a), χ = 120° (b) and χ = 90° (c) using data between 5 and 10 Å resolution and an integration radius of 100 Å. x, crystal a axis; y, crystal b axis; z (perpendicular to the plane of the page), crystal c axis. (d) Packing diagram of the organization of the core protein dimer fusion proteins in the unit cell. The 24 asymmetric units form an octahedral particle. The octahedral diagram is a schematic interpretation of the octahedral symmetry that shows the twofold axes (χ = 180°) in green, the threefold axes (χ = 120°) in blue and the fourfold (χ = 90°) axes in red.

We considered whether the VLPs formed octahedrons as a result of crystal packing, because we did not detect the octahedral particles by electron microscopy before crystallization. The unexpected octahedral virus coat protein arrangement in crystals is unknown except for the case of a bacteriophage MS2 mutant (Plevka et al., 2008 ▶). There were some similarities between the CPD fusion protein and the MS2 coat protein mutant. The MS2 mutant has been reported to be a covalent coat protein dimer (CCPD) of the MS2 phage capsomere that is assembled into icosahedral particles (Peabody & Lim, 1996 ▶). Moreover, in these studies the authors did not detect the octahedral particles by electron microscopy. However, the CCPDs were crystallized in the cubic space group F432, with unit-cell parameters a = b = c = 220.44 Å (Plevka et al., 2008 ▶). This arrangement corresponded to a particle with octahedral symmetry. Wild-type icosa­hedral MS2 particles were crystallized under these conditions, which were not phase-separation conditions (1.5% PEG 6000, 0.2 M sodium phosphate buffer pH 7.4; Valegård et al., 1986 ▶), but the icosahedral CCPD particles were disassembled and the subunits were rearranged into the octahedral structure under the phase-separation conditions (0.32 M Na₂HPO₄, 0.08 M NaH₂PO₄, 5% PEG 8000, pH 7.5; Plevka et al., 2008 ▶).

In this study, the CPD fusion proteins were also crystallized under phase-separation conditions (0.5–1% PEG 20 000, 1.0–1.4 M ammonium sulfate, 0.1 M MES pH 6.5) in which the salt concentration was higher than that under the crystallization conditions for the wild-type HBV capsids (3.5–4.0% PEG 20 000, 0.1–0.4 M ammonium sulfate, 0.1 M MES pH 6.5; Wynne et al., 1999 ▶). The phase-separation conditions may have destabilized the icosahedral VLPs that consisted of CPD fusion proteins, as in the case of CCPD. Moreover, the CPD fusion proteins formed octahedral crystals that belonged to space group F432. The resulting octahedral crystal was stable to the extent that it could tolerate soaking and diffracted to a resolution of 2.15 Å as opposed to 3.3 Å resolution for the wild-type crystal. This structural information will prove to be important in the design of optimum nanocarriers in the future. Currently, we are carrying out model building and refinement.