|Home | About | Journals | Submit | Contact Us | Français|
A critical feature of a viral life cycle is the ability to selectively package the viral genome. In vivo, phosphorylated hepatitis B virus (HBV) core protein specifically encapsidates a complex of pregenomic RNA (pgRNA) and viral polymerase; it has been suggested that packaging is specific for the complex. Here, we test the hypothesis that core protein has intrinsic specificity for pgRNA, independent of the polymerase. For these studies, we also evaluated the effect of core protein phosphorylation on assembly and RNA binding, using phosphorylated core protein and a phosphorylation mimic in which S155, S162, and S170 were mutated to glutamic acid. We have developed an in vitro system where capsids are disassembled and assembly-active core protein dimer is purified. With this protein, we have reassembled empty capsids and RNA-filled capsids. We found that core protein dimer bound and encapsidated both the HBV pregenomic RNA and heterologous RNA with high levels of cooperativity, irrespective of phosphorylation. In direct competition assays, no specificity for pregenomic RNA was observed. This suggests that another factor, such as the viral polymerase, is required for specific packaging. These results also beg the question of what prevents HBV core protein from assembling on nonviral RNA, preserving the protein for virus production.
Hepatitis B virus (HBV) is the leading cause of liver cancer and has infected approximately 2 billion people worldwide, of which approximately 350 million people have developed chronic infections (19, 65). HBV has a partially double-stranded DNA (dsDNA) genome that replicates through a single-stranded RNA intermediate, called pregenomic RNA (pgRNA). Infectious virions have a lipid envelope studded with surface protein, which surrounds the virus core (16, 19). The core is composed of an icosahedral capsid that contains the DNA genome, a DNA-associated viral polymerase (Pol), and several host proteins. Upon infection, the partially double-stranded DNA genome is converted to a covalently closed circular DNA (cccDNA) and takes up residence in the nucleus (8, 40, 52). Viral mRNAs are transcribed from cccDNA. One of these is pgRNA, which also codes for core protein and Pol. Full-length core protein (Cp183) is 183 amino acids in length and consists of an assembly domain (amino acids 1 to 149) and a nucleic acid-binding domain (amino acids 150 to 183) (11, 41). The 34-residue nucleic acid-binding domain is extremely basic, with 17 arginines, consistent with its function. Core protein is dimeric in solution; these dimers self-assemble into ~95% T=4 (120 copies of core protein dimer) and ~5% T=3 (90 dimer) icosahedral capsids (14, 56) and specifically encapsidate both Pol and pgRNA (5, 23). Once encapsidated, pgRNA serves as the template for reverse transcription, which regenerates the partially dsDNA genome (6, 8, 60). Reverse transcription requires an interaction between Pol and a highly conserved stem-loop in the 5′ sequence of pgRNA, called (9, 34, 42, 61, 62).
Specific encapsidation of pgRNA is a critical part of the viral life cycle. Pol, the nucleic acid-binding domain of core protein, and the stem-loop of pgRNA are all involved in specific packaging of pgRNA (23, 26, 30, 41). The sequence is present at both the 5′ and 3′ ends of pgRNA, but only the 5′ copy is required for specific packaging of pgRNA (26, 30). Elements of pgRNA downstream from the 5′ stem-loop do not seem to contribute to specificity (30, 48). When preceding a nonviral RNA, can direct encapsidation of the nongenomic RNA in the presence of Pol and core protein (5, 26). Pol is known to bind tightly to ; however, it is unknown whether Pol plays a direct or indirect role in determining specificity for RNA encapsidation (49, 60). One attractive hypothesis suggests that Pol first binds pgRNA, and then the complex interacts specifically with Cp183 (23). Alternatively, Cp183 may have a preference for pgRNA, independent of Pol. Specificity of a capsid protein for its genome has been observed for many viruses. Retrovirus genomes have a packaging signal that is specifically bound by Gag (1, 18). Similar strategies are employed by bromoviruses and certain phages (17, 37, 51, 64, 69).
The phosphorylation state of HBV core protein plays an important role in specificity for RNA binding. Phosphorylation is required for specific packaging of pgRNA in transfected cells (20, 27, 33). HBV capsid protein expressed in Escherichia coli is unphosphorylated and packages heterologous RNA (11). There are thought to be three major phosphorylation sites and several minor ones (33, 36). Mimicking phosphorylation of the nucleic acid-binding domain of Cp183 by replacing serines S155, S162, and S170 with aspartate or glutamate has been shown to direct specific packaging of pgRNA (20, 33). When these residues are mutated to alanine, mimicking the unphosphorylated state, pgRNA is not specifically packaged. It is not known what nucleic acid, if any, is packaged (20, 33). The core protein phosphorylation state also correlates with subsequent activities, including reverse transcription of pgRNA and intracellular trafficking of the HBV cores (20, 27, 31, 33, 35, 36, 47, 67, 68, 71). Both protein kinase C and serine-arginine protein kinase 1 (SRPK1) have been implicated as the encapsidated kinase (15, 29). Recombinant Cp183 has been phosphorylated in vitro using both of these kinases (15, 27).
To study core protein assembly, RNA binding, and specificity for RNA, an in vitro disassembly/reassembly system is required. Though expression of recombinant, unphosphorylated capsids in E. coli is well established (63), the development of an in vitro assembly system has been hampered by difficulties in purifying assembly-competent Cp183 dimer. Disassembly of HBV capsids in which the core protein is truncated to the first 149 amino acids (Cp149), i.e., the assembly domain only, is well characterized. As the assembly domain lacks the nucleic acid-binding moiety, it does not package nucleic acid (11). Cp149 capsids disassemble to dimers of core protein in mild urea. Reassembly of empty Cp149 capsids is triggered by increased ionic strength (13, 63). Full-length core protein (Cp183) capsids are more stable to dissociation than Cp149 capsids (63), presumably because they are stabilized by Cp-Cp and Cp-RNA interactions. At high ionic strength, the Cp-Cp interactions are strengthened (13). At low ionic strength, the Cp-RNA interactions are strengthened. For efficient disassembly, both of these interactions must be weakened. Additionally, Cp183 can form covalent interdimer bonds between C-terminal Cys183, which may stabilize the capsids further (43). The effect of these disulfides can be ameliorated by the presence of a reducing agent, like dithiothreitol (DTT).
Here, we tested the hypothesis that HBV Cp183 binds pgRNA specifically as a function of only core protein phosphorylation and pgRNA sequence. We provide a system for disassembly of Cp183 capsids using mild treatment with guanidine HCl (GuHCl) and simultaneous precipitation of the prepackaged E. coli RNA. From this disassembled material, we purified assembly-competent Cp183 dimer and reassembled it into empty and RNA-filled capsids (Fig. (Fig.1).1). Both Cp183, phosphorylated by coexpression in E. coli with SRPK1 (P-Cp183), and a mutant Cp183, with three serine-to-glutamate mutations at positions 155, 162, and 170 (Cp183-EEE), were similarly disassembled and reassembled into empty and RNA-filled capsids. We find that core protein, independent of phosphorylation state, binds RNA nonspecifically but with high cooperativity.
Four RNAs were produced by in vitro transcription from purified plasmid DNA templates: HBV pgRNA (3.2 kb), Cowpea chlorotic mottle virus (CCMV) RNA1 (3.2 kb), E. coli LacZ RNA (3.2 kb), and Xenopus elongation factor RNA (1.9 kb). The HBV pgRNA (GenBank accession number V01460) was transcribed from the plasmid 1135. This plasmid has the SP6 promoter juxtaposed to the HBV sequence such that the 5′ end of the transcript has the sequence GACUUUUUCAC instead of the canonical AACUUUUUCAC. An SalI site is 3,296 nucleotides (nt) downstream of the transcription start site. Linearization of the plasmid with SalI prior to transcription results in a pgRNA with the following sequence at its 3′ terminus: 5′-GAAUUUGGAGCGUCGA-3′. The italicized letters represent HBV sequence, and the underlined residues represent the remnants of the SalI site. Unlike natural pgRNA, this RNA does not have a 5′ cap or a 3′ poly(A) tail. The pCC1TP1 plasmid encodes CCMV RNA1 (2). RNA1 transcription starts at a T7 promoter and terminates at an XbaI restriction site. The LacZ RNA was transcribed from an ApaI-linearized pcDNA3.1/V5-His-TOPO/LacZ vector (Invitrogen) starting at a T7 promoter. The linearized template for the Xenopus elongation factor (Xef) RNA was provided with the T7 MEGAscript kit. In vitro transcription from each linearized DNA was performed using the appropriate MEGAscript kit (Ambion) according to the manufacturer's protocol. RNA was precipitated with LiCl per the MEGAscript kit protocol, ethanol precipitated, and then resuspended in nuclease-free water and stored at −20°C until used. 32P-radiolabeled RNA was produced using MEGAscript kits by supplementing the in vitro transcription reaction with [α-32P]UTP (Perkin-Elmer), per protocol.
The plasmid coding for Cp183 was made by modifying a pET11c plasmid containing the sequence for Cp149 (74). The sequence coding for the C-terminal 34 amino acids of Cp183 was incorporated by inserting a 145-bp sequence into existing SalI and BamHI restriction sites. This insert was created using a single round of PCR with GoTaq polymerase (Promega) with the following overlapping oligomers: 5′-ACGCGTCGACGCTTCCGGAGACTACTGTTGTTCGTCGCCGTGGCCGTTCCCCGCGTCGCCGTACTCCGTCGCCGCGTCGC-3′ and 5′-CGGGATCCTAACATTGAGATTCACGAGATTGAGAACGACGGCGACGCGGCGATTGAGAGCGACGGCGACGCGGCG-3′. The inserted sequence matches the subtype adyw amino acid sequence, codon optimized for expression in E. coli. Amplification of this insert was accomplished by multiple rounds of PCR using the primers 5′-ACGCGTCGACGCTTCCGGAGACTAC-3′ and 5′-CGGGATCCTAACATTGAGATTCACG-3′.
The insert and Cp149-pET11c vector were double digested using SalI and BamHI (Roche), purified by native agarose gel electrophoresis, and extracted from the gel using a QiaQuick Gel Extraction kit (Qiagen). Ligation of vector and insert was accomplished with T4 DNA ligase (Roche), and the newly created vector was transformed into BL21 Gold(DE3) cells (Stratagene). This vector, coding for adyw Cp183, is referred to as pET11cCp183.
The vector coding for Cp183-EEE, pET11cCp183EEE, was generated from pET11cCp183 by exchanging the inserted region created for pET11cCp183 with an insert containing mutations of serine to glutamic acid at amino acid residues 155, 162, and 170. Rather than produce this insert by a single round of PCR (as above), the 143-bp dsDNA insert was obtained from IDT. The sequence of this insert is as follows: 5′-CCTGTCGACGCTTCCGGAGACTACTGTTGTTCGTCGCCGTGGCCGTGAACCGCGTCGCCGTACTCCGGAACCGCGTCGCCGTCGCTCTCAAGAACCGCGTCGCCGTCGTTCTCAATCTCGTGAATCTCAATGTTAGGATCCGG-3′. This insert was amplified by PCR using the primers 5′-CCTGTCGACGCTTCCGGAGACTACTGTTGTTCG-3′ and 5′-CCGGATCCTAACATTGAGATTCACGAGATTGAGAACG-3′. Incorporation of this insert into pET11cCp183 was performed as described above.
Phosphorylation of Cp183 in E. coli was accomplished by cotransfection of BL21 Gold(DE3) cells with both pET11cCp183 and pROS-SRPK1a5, a derivative of pLysRARE, which codes for expression of SRPK1, chloramphenicol resistance, and several rare tRNAs. The SRPK1 clone was provided by Peter Kreivi (72). Successful cotransformation was selected for with 100 μg/ml ampicillin and 34 μg/ml chloramphenicol LB agar plates.
Large-scale expression and purification were based on the protocol by Wingfield et al. (63) and was the same for all three Cp183 variants, except that for phosphorylated Cp183, 34 μg/ml chloramphenicol was added to maintain the SRPK1 plasmid. In all cases, we took advantage of uninduced expression. Freezer stocks of transformed BL21 Gold(DE3) cells were used to inoculate 2 liters of Terrific Broth (Amresco) with 100 μg/ml carbenicillin. Following overnight growth at 37°C, uninduced cells were pelleted and resuspended in 0.25 mg/ml Pefabloc AEBSF [4-(2-aminoethyl)-benzenesulfonyl fluoride; Roche], 50 mM Tris (pH 7.5 at 4°C), and 5 mM DTT. Typical growth yielded 15 g of cell paste. The cells were lysed by sonication and spun at 8,000 × g for 1 h to pellet cell debris. A spin of the supernatant at 120,000 × g for 14 h was used to pellet RNA-filled Cp183 capsids from this supernatant. The pellet was resuspended in purification buffer (5% sucrose, 5 mM EDTA, 50 mM Tris, pH 7.5 at 4°C, 2 mM DTT) and loaded onto a 300-ml CL4B Sepharose column. Fractions containing capsid protein were then passed over a 60-ml Blue Sepharose HP (GE Healthcare) column equilibrated in purification buffer. The flowthrough contained the majority of the protein and was further purified using step gradients of 40-50-60% sucrose buffered with 50 mM Tris (pH 7.5)-2 mM DTT. The gradients were centrifuged in an SW-28 Ti Beckman Coulter rotor at 120,000 × g for 4 h at 4°C. The gradients were fractionated, and fractions containing capsid were dialyzed against purification buffer. Protein purity was assessed at each step using SDS-PAGE. Recovery of protein varied between 30 and 60 mg, which was pressure concentrated using YM100 ultrafiltration disks (Amicon) to working concentrations of 0.5 to 2 mg/ml. These RNA-filled capsids were stored at −80°C.
Mass spectra were collected using reversed-phase high-performance liquid chromatography electrospray ionization time of flight mass spectrometry (HPLC-ESI-TOF-MS). The capsid samples were dialyzed into 50 mM HEPES, pH 7.5, and then separated on a BioBasic 4, C4 reverse phase column (ThermoFisher), using a linear gradient from 5% acetonitrile-0.1% formic acid to 95% acetonitrile-0.1% formic acid. Mass spectra were collected using a Waters LCT Classic mass spectrometer in line with the HPLC system.
Capsids expressed and purified from E. coli were disassembled by dialysis at 4°C into disassembly buffer (1.5 M guanidine HCl, 0.5 M LiCl, 50 mM HEPES, pH 7.5, 2 mM DTT). Following dialysis, encapsidated RNA from the expression system that was precipitated by the LiCl was pelleted with a spin of 20,000 × g for 15 min at 4°C. Cp183 dimer from the supernatant of this spin was purified by size exclusion chromatography (SEC) using an analytical grade Superose 6 column (GE Lifesciences) equilibrated in disassembly buffer. Fractions containing core protein were identified by SDS-PAGE. Cp183 dimer was either used immediately or stored for short periods at 4°C.
Empty capsids were reassembled by dialysis of purified Cp183 dimer into empty capsid assembly buffer (0.25 M NaCl, 50 mM HEPES, pH 7.5, 2 mM DTT). RNA-filled capsids for electrophoretic mobility shift assays (EMSAs) were produced by adding purified Cp183 dimer to 150 ng of various in vitro transcribed RNAs over a range of molar ratios of protein dimer to RNA polymer. In each reaction, the final buffer conditions were 0.5 M guanidine HCl, 167 mM LiCl, and 16.7 mM HEPES pH 7.5, which allowed reassembly to occur. E. coli capsid and RNA standards were also adjusted to these buffer conditions. After 15 min at room temperature, samples were loaded onto 1% agarose gels in Tris-acetate-EDTA (TAE) with 0.0001% ethidium bromide in both the gel and the running buffer. Gels were run at 80 V for 1.5 to 2 h. We observed that ethidium bromide improved separation of free RNA from RNA packaged into capsids. Reassembly of RNA-filled capsids by dialysis into 150 mM NaCl, 50 mM HEPES, pH 7.5, and 2 mM DTT yielded nearly identical results to reassembly by dilution (above) but required significantly more material for each reaction.
Competition assays were performed exactly as for EMSAs, except that core protein dimer was held at a constant 120:1 molar ratio with 50 ng of α-32P-labeled pgRNA for each reaction, while unlabeled competitor RNA was added in increasing amounts. Following electrophoresis, gels were dried using a vacuum gel dryer at 62°C for 50 min. Dried gels were exposed to a storage phosphor screen overnight and were imaged using a Typhoon 9210 Variable Mode Imager (Amersham Biosciences).
UV absorption spectra were obtained using an Agilent 8453 UV-Visible Spectrophotometer. Signal due to light scattering was estimated using absorbance at 320 nm and 340 nm, assuming a 1/λ4 relationship between light scattering and wavelength. The calculated scattering was subtracted from all spectra. The concentration of in vitro transcribed RNA was determined using 260-nm absorbance (1 A260 is equivalent to 40 μg/ml). Protein and RNA concentrations from mixed protein and RNA samples were calculated using extinction coefficients of 38,000 M−1 cm−1 (260 nm) and 60,900 M−1 cm−1 (280 nm) per Cp183 dimer and 8,000 M−1 cm−1 (260 nm) and 4,000 M−1 cm−1 (280 nm) per nucleotide for RNA. The extinction coefficients for protein were calculated using sequence according to the method of Pace et al. (45).
Samples for electron microscopy (EM) were taken from EMSAs at a dimer-to-RNA ratio of 120:1. Capsid samples were applied to glow-discharged, carbon-coated copper grids (EMS) and immediately blotted dry. The grids were negatively stained with 2% uranyl acetate for 1 min and then blotted dry. Images were obtained at ×50,000 magnification using a JEOL JEM-1010 transmission electron microscope and recorded using a charge-coupled-device (CCD) camera.
To study capsid assembly, nucleic acid binding, and specificity of RNA encapsidation in vivo, it is necessary to purify assembly-competent Cp183 dimer and then reassemble capsids in the presence or absence of RNA (Fig. (Fig.1).1). We began by expressing and purifying Cp183 capsids from E. coli according to established methods (63). Capsids purified from E. coli are filled with RNA from the expression system (11). The E. coli purified capsids had characteristics typical for HBV capsids as determined by negative-stain electron microscopy (Fig. (Fig.2A)2A) and size exclusion chromatography (Fig. (Fig.2B)2B) (63).
We used guanidine HCl (GuHCl) to induce disassembly of capsids and LiCl to precipitate the encapsidated E. coli nucleic acid. Cp183 dimer was then purified from the disassembly reaction mixture by size exclusion chromatography (Fig. (Fig.2B).2B). This approach was inspired by protocols for the disassembly of plant viruses that take advantage of unstable capsid protein-protein interactions at pH 7 in combination with divalent cations for RNA precipitation (4). For the disassembly reaction, we found that 1.5 M guanidine HCl disrupted capsid quaternary protein structure. At <1.25 M guanidine HCl, disassembly was inefficient. Denaturation of the core protein assembly domain does not occur at <2.5 M guanidine HCl (55). The high ionic strength of the GuHCl also weakens Cp183-RNA electrostatic interactions allowing the E. coli RNA, packaged during expression, to be precipitated by the LiCl. Samples of the disassembly reaction mixture were visibly cloudy prior to sedimentation of RNA. Following sedimentation, capsid dissociation and dimer purification were confirmed and completed by size exclusion chromatography. Three peaks were observed in the size exclusion chromatograph, referred to as capsid peak, RNA peak, and dimer peak (Fig. (Fig.2B).2B). The elution volume of these peaks and their UV absorption spectra were used to identify them. RNA contamination was judged using UV absorption. RNA has an absorption maximum at 260 nm, whereas tyrosine and tryptophan contribute the majority of absorption for protein and maximally absorb near 280 nm, with a characteristic shoulder at 290 nm.
The dimer peak from the dissociation procedure had a UV absorbance spectrum with 260-nm/280-nm ratios of 0.6, consistent with nucleic acid-free protein (Fig. (Fig.2C)2C) (21). This peak eluted from a Superose 6 column slightly earlier than the control peak for 34-kDa Cp149 dimers (data not shown), consistent with a 42-kDa Cp183 dimer. The capsid peak coeluted with RNA-filled, E. coli-expressed capsids and likely represents RNA-filled capsids that have remained intact despite the disassembly protocol. The UV spectrum for the capsid peak is typical for a mixture of protein and RNA (Fig. (Fig.2C).2C). It had a significant absorbance at 260 nm but a clear 290-nm shoulder. The RNA peak eluted at an intermediate volume and appeared to be largely nucleic acid by UV absorption (Fig. 2B and C). This peak varied in height from quite prominent to almost undetectable between core protein preparations. Consistent with the critical concentration-like nature of capsid assembly (28), the fraction of capsids that were disassembled to dimer was concentration dependent. At higher capsid concentrations, the capsid peak was increased while the dimer peak remained about the same. At low concentrations of capsid, the ratio of dimer to capsid peaks was increased, but the absolute concentration of protein was less. Typical disassembly yielded 5 to 15 μM Cp183 dimer. Critically, we also observed that purified Cp183 dimer was not soluble in the low-ionic-strength buffers typically used for Cp149 dimer. Also, the protein precipitated if the concentration of guanidine HCl was decreased below 0.5 M.
We assembled empty Cp183 capsids to demonstrate the activity of the assembly domain of the core protein and to determine if the polycationic C terminus inhibited assembly. We tested for salt-induced reassembly by dialysis of protein from the dimer peak into buffers lacking GuHCl but with various ionic strengths. Below 0.25 M ionic strength, Cp183 precipitated. At higher ionic strength, however, solutions remained clear, and the protein was observed by size exclusion chromatography and electron microscopy (Fig. 3A and C) to reassemble into empty capsids. Particles consistent in size with both T=3 (32-nm diameter) and T=4 (36-nm diameter) capsids were observed. Reassembled capsids eluted at the same positions as the capsid peak (Fig. (Fig.3A)3A) by size exclusion chromatography. The UV absorption spectrum for these particles suggests that they were pure protein compared to the nucleoprotein spectrum of the E. coli capsids or the capsid peak (Fig. (Fig.2B).2B). By negative-stain EM, these reassembled particles had the correct size and morphology for HBV capsids but had slightly thinner walls than the E. coli purified capsids (Fig. (Fig.2A2A and and3C).3C). Thus, ≥0.25 M NaCl is sufficient to attenuate electrostatic repulsion from the basic C termini while supporting the protein-protein interactions of the assembly domain and allowing reassembly of empty capsids with substantially greater solubility than free Cp183 dimer.
To evaluate the contribution of phosphorylation to specificity of RNA packaging, we produced a phosphorylated version of Cp183 (P-Cp183) and a Cp183 mutant where the critical phosphorylated serines (S155, S162, and S170) (31, 33) were replaced by glutamic acid (Cp183-EEE). P-Cp183 was produced by cotransformation of the E. coli expression system with SRPK1, which is capable of phosphorylating HBV in vitro (15). The phosphorylation state of our core protein was demonstrated by LC-ESI mass spectrometry. The unphosphorylated monomeric Cp183 had a mass of 21,041.9 Da (Fig. (Fig.4A),4A), consistent with the mass calculated from the amino acid sequence. The mass spectrum for P-Cp183 contained peaks consistent with the addition of zero to seven phosphates (Fig. (Fig.4B).4B). Comparison of ion intensities is not necessarily quantitative due to possible differences in ionization efficiency. However, given that all species are of similar height and that the phosphate itself makes only a small contribution to the net charge of the protein, it seems likely that phosphorylation of core protein does not have a dramatic effect on the ionization efficiency. If we make the assumption that all forms of P-Cp183 were detected with equal sensitivity, the relative abundance of each species is listed in Fig. Fig.4C.4C. Cp183 with seven phosphates is the most intense peak, consistent with the highly processive nature of SRPK1 and the sequence of the nucleic acid-binding domain, which contains seven serines and one threonine (Fig. (Fig.4D).4D). The identities of the phosphorylated residues cannot be deduced from these data; however, we anticipate that the nucleic acid-binding domain is preferentially phosphorylated, based on the sequence specificity of SRPK1 (22). Both P-Cp183 and Cp183-EEE were amenable to the disassembly protocol used for Cp183. Size exclusion chromatograms and the UV absorption spectra of the resulting peaks for the disassembly reaction for these two capsid types were almost indistinguishable from those shown in Fig. Fig.22 (data not shown). Purified P-Cp183 and Cp183-EEE dimer could be reassembled into empty capsids using the protocol established for Cp183. These reassembled particles were characterized by SEC and negative-stain electron microscopy and were indistinguishable from Cp183 (data not shown).
After demonstrating assembly competence of Cp183 without nucleic acid, we evaluated the ability of purified Cp183, P-Cp183, and Cp183-EEE dimer to bind and package RNA. Several single-stranded RNAs (ssRNAs) were evaluated: HBV pgRNA, CCMV RNA1, E. coli LacZ RNA, and Xenopus elongation factor RNA. It has been suggested that viral RNA may be more compact, in general, and more easily encapsidated (70). We chose CCMV RNA1, which has approximately the same length as pgRNA, as a non-HBV viral RNA (25). We chose LacZ RNA, which codes for the expression of β-galactosidase and is similar in length to pgRNA, as a nonviral RNA. To examine the role of RNA size in HBV assembly, we used Xef RNA (coding for Xenopus elongation factor 1α gene), which is approximately 1,900 nt in length.
We used electrophoretic mobility shift assays (EMSAs) to probe RNA binding. These assays can differentiate low cooperativity binding (Fig. (Fig.5A)5A) from high cooperativity binding (Fig. (Fig.5B).5B). Under the conditions we used for electrophoresis, unbound pgRNA, RNA1, and LacZ RNA migrate slightly more slowly than a 1-kb dsDNA molecular weight (MW) marker. Xef RNA migrates significantly faster than the 1-kb dsDNA MW marker (Fig. (Fig.5C).5C). Capsids containing RNA migrate at approximately the position of a 1.5-kb dsDNA MW marker, regardless of which RNA is encapsidated (Fig. (Fig.5C).5C). This is consistent with the theory that capsid migration is largely determined by external charge and size, not contents (25, 53, 54). These relationships between the migration of ssRNA, dsDNA, and capsids were consistent under the conditions we used and were used to compare results between experiments.
RNA packaging was evaluated at multiple protein/RNA ratios, using a constant RNA concentration. In each gel, an MW standard, an RNA standard, and an E. coli-expressed capsid standard were used to identify bands (Fig. (Fig.5C).5C). Cp183, P-Cp183, and Cp183-EEE appeared to bind all RNAs with high levels of cooperativity. At low concentrations of dimer, RNA appeared unbound. In every experiment, as the concentration of dimer protein was increased, a bimodal distribution of RNA was observed. The slower band comigrated with RNA-filled capsids; the faster band comigrated with free RNA. At protein/RNA ratios of >90:1, a slight amount of fluorescence (<5% of the total fluorescence) was detected in the loading wells, consistent with a small amount of protein-RNA aggregation, likely due to the solubility issues inherent to core protein. At lower protein/RNA ratios, no fluorescence was detected in the wells. Solution experiments demonstrated that protein and RNA aggregation was minor (data not shown). In all EMSAs, a vast majority of the total fluorescence was partitioned between the free-RNA band and the RNA-filled capsid band. The approximate point at which the ethidium bromide fluorescence associated with RNA (and presumably the amount of RNA) in these two pools was equal was at a ratio between 45 and 60 dimers per RNA molecule. This ratio varied slightly between experiments, but the range was the same for pgRNA, RNA1 (Fig. (Fig.5C),5C), and LacZ RNA (data not shown). For Xef, the crossover occurred between 15 and 30 dimers per RNA molecule (Fig. (Fig.5C).5C). In each case, the RNA was completely packaged at protein concentrations below the stoichiometric ratio of 120 dimers to each RNA for a T=4 capsids. This suggests that multiple RNAs may be packed into each capsid.
The EMSAs indicate that RNA is bound and migrates at a position consistent with RNA-filled capsids. Electron microscopy demonstrates that capsids are formed in the reassembly reaction (Fig. (Fig.6).6). No obvious assembly intermediates were observed, consistent with the high positive cooperativity seen in the experiments shown in Fig. Fig.5.5. Capsids reassembled with RNA have an appropriate size and morphology and appear very similar to capsids purified from E. coli. Both T=3 and T=4 capsids are observed in approximately the same ratio seen for E. coli purified capsids (Fig. (Fig.6).6). However, the walls of RNA-filled capsids appear to be thicker than those seen in empty capsids (Fig. (Fig.3)3) or Cp149 capsids (59).
To further test for specificity of RNA packaging, purified Cp183, P-Cp183, and Cp183-EEE dimer were presented with a mixture of 32P-labeled pgRNA and unlabeled competitor RNA. If P-Cp183 has no specificity for pgRNA, we anticipate that the fraction of bound radiolabeled RNA will be the same as its fraction overall. The amount of competitor RNA it takes to prevent radiolabeled pgRNA from being encapsidated is a measure of the specificity P-Cp183 has for pgRNA. We chose LacZ RNA as a nonviral competitor RNA of similar length to pgRNA. By EMSA, all three core protein variants package LacZ RNA with the high levels of cooperativity binding observed for all tested RNAs (an example of assembly is shown by EMSA in Fig. Fig.7A;7A; other data not shown). In the design of this competition assay, the concentrations of P-Cp183 dimer and radiolabeled pgRNA were held constant, and the concentration of the competitor RNA was varied. We used unlabeled pgRNA as the competitor for control reactions (Fig. (Fig.7B7B).
Qualitatively, we see no evidence for specific encapsidation of pgRNA by P-Cp183 (Fig. 7B and C) or other core protein variants (data not shown). Here, in all lanes except that of the RNA standard, there are 120 protein dimers present for each 32P-labeled pgRNA, more than enough protein to encapsidate the radiolabeled pgRNA, as seen in Fig. Fig.55 and in Fig. 7B and C (lanes 0). As competitor RNA is added and the concentration increased, the amount of free 32P-labeled pgRNA increases, indicating that progressively more of the P-Cp183 has been appropriated to encapsidate unlabeled, competitor RNA. The gels in which unlabeled pgRNA is the competitor RNA (Fig. (Fig.7B)7B) are essentially identical to those using unlabeled LacZ RNA (Fig. (Fig.7C).7C). This argues against specificity of P-Cp183 for viral RNA under these conditions. Similar experiments with Cp183 and Cp183-EEE also showed no preference for pgRNA over LacZ RNA (data not shown). There are some caveats to this experiment. We observe that free pgRNA migrates through the native gel as a smear, suggestive of multiple conformations (Fig. 7B and C, RNA Std). Also, in the titration, some of the 32P-labeled RNA is at the top of the gel, suggestive of a small amount of nucleoprotein aggregate. While these caveats confound quantitation, they do not effect qualitative interpretation of this experiment.
We have demonstrated a successful strategy for the disassembly and reassembly of Cp183 HBV capsids. Dimer purified from this process is free of nucleic acid (Fig. (Fig.2C)2C) and is assembly competent. Both empty particles and RNA-filled particles were reassembled. We find that Cp183, phospho-Cp183 (P-Cp183), and the phospho-mimic Cp183-EEE all bind HBV pgRNA and nonviral RNA with similar affinities and levels of cooperativity. This suggests that specificity for viral RNA is not inherent in the core protein and is consistent with the hypothesis that specificity of RNA packaging is derived from interaction with the pgRNA-Pol complex (23, 49). The ability of Cp183 to spontaneously assemble on nonviral RNA (Fig. (Fig.5C5C and and6)6) suggests that some additional factor, for example, an as yet unidentified chaperone, may have a role in preventing assembly on the wrong RNA.
In transfected cells, specific packaging of viral RNA requires correctly phosphorylated Cp183 though a mimic of P-Cp183, such as Cp183-EEE, will also work (32, 33). However, in our in vitro experiments, we did not observe evidence of specificity regardless of phosphorylation state (Fig. (Fig.5).5). Competition EMSAs (Fig. (Fig.7B7B and Fig. C), which were designed to be sensitive to even weak specificity, demonstrate the same general trend. Although we did not observe measurable specificity for pgRNA, the phase space of reaction conditions and reactants remains suitably large for additional evaluation. In these experiments, we may have distorted the necessary structure of either pgRNA or core protein. pgRNA may require a cofactor, such as Mg2+, or a chaperone for proper folding (46). Our nonphysiological conditions may distort the normal interaction of HBV with its genome. We expect that capsids are very stable once formed (55); thus, bound RNA may be kinetically trapped rather than in its most stable state. However, we do not anticipate rearrangement of Cp to replace packaged heterologous RNA with pgRNA in our in vitro system.
In assembly EMSAs, we observe a bimodal distribution of free and packaged RNA, indicating highly cooperative assembly (Fig. (Fig.5).5). Both high-cooperativity and low-cooperativity nucleic acid binding by capsid proteins have been observed for other viruses (25, 64). Cooperativity can be qualitatively differentiated by EMSAs (Fig. 5A and B). If the binding of core protein to RNA occurs with low cooperativity, protein is bound randomly, irrespective of other binding events, which results in a Gaussian distribution of protein among available RNA. As the protein concentration is increased, a progressively more complete ensemble of RNA-bound assembly intermediates will be formed until all RNA strands are encapsidated (Fig. (Fig.5A).5A). In the case of high cooperativity (Fig. (Fig.5B),5B), the binding of a single protein to an RNA creates adjacent high-affinity sites on the same RNA. Thus, RNA strands that randomly bind a single coat protein tend to bind additional subunits and become encapsidated while other RNA remains unbound. The result is a bimodal distribution of free RNA and completely encapsidated RNA, with low concentrations of intermediates. McGhee-von Hippel theory breaks nonspecific nucleic acid binding into an association constant (KA) for protein binding to nucleic acid and a cooperativity parameter, ω, for protein-protein interaction, with high cooperativity associated with high values of ω (39). CCMV, which has very weak protein-protein interactions, is an example of low-cooperativity binding (25). Cp183, P-Cp183, and Cp183-EEE all demonstrate high-cooperativity binding to both pgRNA and heterologous RNAs (Fig. (Fig.5C),5C), consistent with equating the ω term with the association constant for the assembly of empty capsids (13).
The EMSAs also indicate that Cp183 and variants are able to package multiple RNAs per capsid, as all RNA is encapsidated at less than the expected 120:1 protein dimer-to-RNA ratio. We observe complete packaging of genome-sized RNA molecules at an ~60:1 dimer-to-RNA polymer ratio, suggesting, on average, two genome-sized RNAs per capsid. This judgment is based on the disappearance of the free-RNA band. This result would seem to contradict the arguments that genome packaging is electrostatically regulated in HBV (44). However, at two genomes per capsid, corresponding to 3,840 C-terminal arginines to ~6,400 RNA phosphates (Fig. (Fig.5C),5C), the ratio of arginine to RNA phosphate is 0.6, similar to the ratio of ~0.7 arginines per encapsidated RNA phosphate typical in other viruses and consistent with the counterion condensation model of ionic interactions with DNA and with Poisson-Boltzmann screening (10, 38, 57, 58). We should note that our assembly conditions are nonphysiological and at very high ionic strength, which may mask the differences in the positive charge of the RNA-binding domain to some degree. Both small ions and electrostatic screening by proximal charges must be taken into account in evaluating electrostatic interactions (24). The excess guanidine HCl and LiCl remaining in the buffers of our assembled capsids may have unintended effects on Cp183-RNA interactions. Nonetheless, the high ionic strength would be expected to weaken interactions and decrease the amount of bound RNA. In vivo, other factors are likely to play critical roles in RNA packaging, resulting in the higher than typical 1:1 ratio of protein-associated positive charge to nucleic acid-associated negative charge in HBV.
We have developed an in vitro system for isolating relatively large amounts of highly purified, full-length HBV core protein (Cp183) in an assembly-competent form suitable for investigating its interaction with nucleic acid and other proteins. We found this protein to have poor solubility, contributing to our difficulties and those of others trying to purify Cp183 dimer. Other approaches have been taken for the development of such a disassembly/reassembly system for HBV capsids (44, 50). Rabe et al. (50) reported that small amounts of dimer could be isolated from E. coli-derived capsids by column chromatography at low ionic strength. Newman et al. (44) observed disassembly of Cp183 capsids by digesting the E. coli RNA from the expression system with micrococcal nuclease, with reassembly triggered by mixing the disassembled capsids with polyanions; unfortunately, this system did not allow evaluation of contaminants or characterization of the dimer itself. By obtaining purified dimer, we have been able to investigate the mechanism of RNA binding and observe the solubility issues of Cp183.
All three versions of core protein dimer (Cp183, P-Cp183, and Cp183-EEE) precipitated if the GuHCl concentration was less than 0.5 M. This is remarkable as the RNA binding domain found in Cp183, but absent in Cp149, is extremely ionic in nature and would be expected to enhance solubility. By comparison, Cp149 dimer is stable at very low ionic strength. It is possible that the positively charged nucleic acid-binding domain interacts with negative charge on the core protein itself, such as the patch at the spike tip (12, 66), leading to aggregation. The solubility problems encountered with Cp183 dimer can be circumvented by providing appropriate reassembly conditions; in vitro, either sufficient ionic strength to yield empty capsids (Fig. (Fig.3)3) or nucleic acid for RNA-filled capsids would be required (Fig. (Fig.55 and and6).6). In the case of empty capsids, we found that the ionic strength of the buffer must be maintained above 0.25 M although GuHCl could be completely removed. This is likely due to the need to shield the electrostatic repulsion of the very positively charged nucleic acid-binding domains to allow the capsids to form. In contrast, the RNA-filled capsids are very soluble following reassembly, even in the absence of salt. Although these assays are performed under nonphysiological conditions, the solubility issues encountered for HBV dimer suggest the need for a mechanism to ensure dimer solubility in vivo, either through a chaperone or assembly into capsids shortly after translation.
Phosphorylation is critical for packaging the correct RNA in transfected cells (32, 33). We produced a phosphorylated capsid protein in which up to seven phosphates have been added by coexpression of Cp183 with SRPK1 (Fig. (Fig.4B).4B). SRPK1 is highly processive and has a high affinity for serine- and arginine-rich sequences (15, 22, 73). There are seven serines in the arginine-rich nucleic acid-binding domain of Cp183, including three SPRRR repeats which fit the putative consensus sequence for SRPK1 (Fig. (Fig.4D)4D) (22). The high frequency of proteins with seven phosphorylations (Fig. (Fig.4B4B and Fig. C) is consistent with the high processivity of SRPK (3) though phosphorylation of serines on noncanonical sites in the assembly domain cannot be ruled out. As previously noted, P-Cp183 shows the same characteristics in terms of solubility and RNA specificity as do the phospho-mimic Cp183-EEE and unphosphorylated Cp183.
The results presented in this paper suggest the need for two sets of chaperones in HBV assembly. One set of chaperones maintains Cp183 solubility and prevents inappropriate assembly. The lack of specificity observed here supports the idea that specific packaging of pgRNA in vivo relies on an additional factor/chaperone. The front-runner for such a factor is Pol, which specifically binds the stem-loop of pgRNA (7, 9, 49). Tests of specificity in transfected cell lines have confirmed that the presence of Pol correlates with specific packaging (23). However, Pol has yet to be successfully purified, preventing a more complete examination of its role in specificity. The stability of the HBV nucleoprotein complex raises another quandary. In vivo, assembly must be reversible for the genome to be released upon infection; this suggests that another set of chaperones, perhaps associated with the nuclear pore complex (50), is involved in disassembly. The reagents developed in this paper will support efforts to address these questions.
We thank Gillian Air for temporary laboratory accommodation, advice, and co-mentorship for J.Z.P. We acknowledge the contributions of Maren Vogel to these studies. We also thank Jonathan Karty of the Indiana University Chemistry Department Mass Spectrometry Facility and Barry Stein of the Indiana Molecular Biology Institute.
This work was supported by National Institutes of Health grants R01-AI077688 to A.Z. and R56-AI068883 to A.Z. and Gillian Air.
Published ahead of print on 28 April 2010.