Virus Maturation Targets the Protein Capsid to Concerted Disassembly and Unfolding*

Many animal viruses undergo post-assembly proteolytic cleavage that is required for infectivity. The role of maturation cleavage on Flock House virus was evaluated by comparing wild type (wt) and cleavage-defective mutant (D75N) Flock House virus virus-like particles. A concerted dissociation and unfolding of the mature wt particle was observed under treatment by urea, whereas the cleavage-defective mutant dissociated to folded subunits as determined by steady-state and dynamic fluorescence spectroscopy, circular dichroism, and nuclear magnetic resonance. The folded D75N α subunit could reassemble into capsids, whereas the yield of reassembly from unfolded cleaved wt subunits was very low. Overall, our results demonstrate that the maturation/cleavage process targets the particle for an “off pathway” disassembly, because dissociation is coupled to unfolding. The increased motions in the cleaved capsid, revealed by fluorescence and NMR, and the concerted nature of dissociation/unfolding may be crucial to make the mature particle infectious.

Viruses are macromolecular assemblies designed to exert their biological role in a single sequential cycle: 1) assembly inside the cells; 2) release to the environment; 3) attachment to new host cells; 4) disassembly and delivery of genome; and 5) replication of the genome and transcription of new viral proteins. Among these five stages, disassembly of the capsid and unpacking of the nucleic acid is the least understood (1)(2)(3)(4). To be effective, disassembly has to occur fast and at the correct time after endocytosis. The switch for this process is usually attributed to the acidic pH inside the endocytic vesicles, but in vitro many viruses are not uncoated by low pH, or the uncoating occurs slowly, not consistent with the requirement for rapid replication. A fundamental property of most animal viruses is the post-assembly maturation, which is required for infectivity. In picornaviruses (5)(6)(7)(8) and nodaviruses (9,10), the maturation cleavage is autocatalytic and is dependent upon appropri-ate particle assembly. A picornavirus capsid is initially assembled from 60 copies of the subunits VP0, VP3, and VP1 and an RNA molecule. The final processing of picornaviruses occurs after assembly in which the precursor protein VP0 is cleaved to release VP4 and VP2 (11). The cleavage site is not accessible to the surface, and it is proposed that the packaged RNA acts as a nucleophile (12). Like the picornaviruses, the nodaviruses are initially constructed as provirions, which mature to an infectious virion by post-assembly cleavage of ␣ protein into ␤ (43 kDa) and ␥ (4 kDa) subunits.
Flock House virus (FHV) 1 is a nonenveloped T ϭ 3 icosahedral insect virus of the family Nodaviridae (Fig. 1). The virus particle consists of 180 copies of the coat protein, which encapsidates the bipartite RNA genome (9,11,13,14). The FHV coat protein expressed in a baculovirus system spontaneously assembles into a virus-like particle (VLP) containing cellular RNA, and the ␣ capsid protein undergoes cleavage into ␤ and ␥ subunits as in the native virus (10,15). The protein capsids in VLPs and authentic particles are indistinguishable by crystallography, although the VLPs are more susceptible to proteolysis than the authentic particles (16).
In this study, we attempt to understand the changes in stability and dynamics caused by the cleavage. With this purpose, we have utilized two baculo expressed VLPs, a wild type sequence (SFHVwt) and a mutant virus (D75N) (15). This mutant was constructed by site-directed mutagenesis of Asp 75 , which is at the cleavage site; replacement of the aspartic acid residue with asparagine results in the production of noninfectious particles that do not undergo cleavage (17). Therefore, the capsid of the cleavage-defective mutant is composed of 180 copies of the intact 47-kDa precursor ␣ protein. We find that the uncleaved capsid protein can undergo dissociation not concerted with unfolding, whereas in wt VLPs, maturation cleavage favors a metastable state in which dissociation of the coat protein is coupled to unfolding. A partially folded state of the uncleaved capsid protein is characterized by fluorescence, circular dichroism, and NMR methods. In addition, cleavage may serve as part of the switch for dissociation, and the resulting "off pathway" disassembly may be important during the viral infection cycle.

EXPERIMENTAL PROCEDURES
Chemicals-All reagents were of analytical grade. Distilled water was filtered and deionized through a Millipore water purification system. The experiments were performed at 20°C in the standard buffer: 50 mM Hepes, pH 7.0. Ultra-pure urea was obtained from Sigma.
Cells and Virus-like Particles-VLPs were expressed in Drosophila cells using a baculovirus expression system and purified as described previously (15). The procapsids and capsids were stored in 10 mM Ca 2ϩ containing buffers. Fluorescence Spectroscopy and High Pressure Measurements-Fluorescence spectra were recorded on an ISSK2 spectrofluorometer (ISS Inc., Champaign, IL). The high pressure cell (18) was purchased from ISS Inc. The tryptophan residues were excited at 280 nm, and emission was observed from 300 to 420 nm. Changes in fluorescence spectra at pressure p were evaluated by the changes in spectral center of mass, where F i stands for the fluorescence emitted at wave number i . The summation is carried out over the range of appreciable values of F. The samples were allowed to equilibrate for 15 min prior to making measurements (high pressure and urea experiments). This time was chosen because the spectroscopic changes reached a plateau within the first 10 min and did not change significantly during longer times (several hours). Unless otherwise noted, experiments were performed at 20°C in 50 mM Hepes, pH 7.0.
Lifetime and Rotation Measurements-Lifetime and dynamic depolarization measurements were performed on a multifrequency crosscorrelation phase and modulation fluorometer which uses the harmonic content of a high repetition rate, mode-locked neodymium yttrium aluminum garnet laser. This laser is used to synchronously pump a dye laser whose pulse train is frequency doubled with an angle-tuned frequency doubler (19). A detailed description of phase fluorometry lifetime measurements and data analysis is found elsewhere (20). The quality of fits was assessed by 2 values and by plots of weighted residuals. Excitation wavelength was 295 nm, and the emission was observed through a long wavelength pass filter (WG 335) with a cut-off at 335 nm.
Light Scattering-Light scattering was measured in an ISS 200 spectrofluorometer (21). Scattered light (320 nm) was collected at an angle of 90°to the incident light.
Far-UV Circular Dichroism-The CD spectra were obtained in 10 mM Hepes buffer, pH 7.0, using a quartz cuvette of 0.2-cm path length. Spectra were the average of two scans from 214 to 260 nm at a 100 nm/min, and the buffer and urea base lines were subtracted. Shorter wavelengths were not analyzed because of increased noise. The spectropolarimeter used was a Jasco J-715 1505 model.
Nuclear Magnetic Resonance-NMR spectra were obtained in a Bruker 600 MHz spectrometer at 25°C. The sample was prepared in 10% D 2 O using 5 mM Hepes buffer at pH 7.0. The VLP concentration was 500 g/ml for both wt and D75N. Water suppression was achieved using the water gate sequence (22) with the composite pulse 3, 9, and 19 and a 1-ms z-pulsed field gradient at 10 G/cm. The urea peak was presaturated.
Size Exclusion Chromatography-High performance liquid chromatography was carried out in prepacked SynChropak columns (inner diameter, 250 ϫ 4.6 mm), obtained from SynChrom Inc. (Linden, IN). GPC 500 column was used for virus elution. The system was equilibrated in 50 mM Hepes, 0.2 M sodium acetate buffer, pH 7.0, at room temperature, using a flow rate of 0.3 ml/min. Sample elution was monitored by fluorescence at 330 nm (excitation at 280 nm). The equipment used was a Shimadzu model SPD 10A.

Dissociation and Unfolding of FHV Capsids-
The properties that stabilize a virus particle have to fulfill two apparently opposite requirements: 1) the stability must be great enough to keep the particle intact while it is outside the host cell and 2) the virus must be able to disassemble inside the host cell. The comparison between FHV virus-like particles of wild type and of the maturation-defective mutant can address the question of what is the role of maturation cleavage on particle stability. Direct reasoning suggests that maturation makes the particle more stable to the hostile environment. We measured both the dissociation of the capsids and the changes in tertiary structure of the capsid protein. Disassembly was monitored by light scattering, and denaturation was monitored by the shift in tryptophan fluorescence by changes in tryptophan lifetime and by circular dichroism. The urea-induced denaturation of wt mature capsids, as measured by fluorescence, overlapped the changes in light scattering ( Fig. 2A), indicating that dissociation and denaturation are concerted events. The D75N particles disassembled at comparable urea concentrations, but the tertiary structure of the subunits required much higher urea concentrations for denaturation (Fig. 2B). These data suggest that disassembled D75N coat protein can be obtained with significant tertiary structure. In fact, at 6 M urea, there were minimal changes in solvent exposure of the D75N coat protein tryptophans (Fig. 3, A and B).
To probe the tryptophan environment in the nanosecond

FIG. 1. Structural features of FHV.
A, space filling model of FHV (T ϭ 3 quasi symmetry). A, B, and C types of subunits are shown in blue, red, and green, respectively. Icosahedral symmetry elements 5-, 3-, and 2-fold are shown as pentagons, triangles, and ovals (in white) respectively, while the yellow triangle and ovals correspond to quasi-3-and 2-fold axes. The icosahedral asymmetric unit of the virus particle is outlined by a triangle, whose secondary, tertiary, and quaternary structures are shown in B (a view down the icosahedral 2-fold axis, from the outside of the virus). The ordered duplex RNA is shown in ball-and-stick representation in gold, and five calcium ions are shown as yellow spheres. C, an orthogonal view of the asymmetric unit. The ␥ peptides are colored in magenta. D, zoom in view of the ␥ peptide of the A-subunit. The only tryptophan residue (W367) on ␥ is shown in ball-and-stick representation. time frame, lifetime measurements were performed at 0, 6, and 8 M urea (Fig. 3). Fig. 3 (C and D) shows the lifetime measurements of the wt and D75N VLPs. The lifetime of the dissociated cleavage-defective coat protein was not significantly affected by 6 M urea (Fig. 3, C-E). On the other hand, 6 M urea produced a large increase in the average lifetime of the wt VLP. The changes in lifetime accompanied the increased exposure of the Trps to the solvent, shown by the decreases in the center of spectral mass (Fig. 3E).
The rotation of the tryptophan residues in the coat protein were also evaluated by phase shift and modulation (21,23,24). For both wt and mutant VLPs, the best fit to the data was obtained using a model that assigns two rotational motions ( Table I). The dynamic depolarization data showed that the mobility of the Trp residues was higher in the cleaved wt VLP than in the noncleaved mutant. This result probably reflects formation of the ␥ peptide on cleavage of the wt protein (Fig. 1,  C and D). This peptide contains one of the five Trp residues, and because of its small size (4 kDa) the Trp would acquire more freedom of motion (Fig. 1D). When 6 M urea was added, a large increase in the faster of two Trp components (f 2 , 2 ) of the wt occurred, whereas the Trps of the D75N mutant retained the motions characteristic of a globular, folded protein (Table  I). Altogether, the steady-state and dynamic fluorescence data show that the cleavage-defective mutant maintains a substantial degree of tertiary packing even in 6 M urea, in contrast to the loss of long range interactions for the wt coat protein.
Folded Conformation of the Urea-dissociated Noncleaved FHV-The fluorescence data indicate that the noncleaved mutant VLP can be readily dissociated by urea, but the coat protein retains a folded or partially folded conformation. The secondary structure of the urea-dissociated coat protein was analyzed by circular dichroism (Fig. 4, A and B). It is clear that the cleavage-defective VLP preserved a large degree of secondary structure in the presence of 6 M urea (Fig. 4B), whereas the wt VLP treated with 6 M urea lost its secondary structure (Fig.  4A).
To try to further identify the folding of the of the coat proteins dissociated by 6 M urea, high resolution proton NMR spectroscopy was utilized (Fig. 4, C-F). Because of the size of the virus, only regions with local mobility can provide sharp lines. The one-dimensional 1 H NMR spectra of wt VLP revealed a remarkable dispersion and sharp lines in the region of the amide and aromatic protons (6.8 -8.5 ppm; Fig. 4C). The appearance of these lines is consistent with the average fast rotation of Trp residues (9.3 ns; Table I) obtained by dynamic fluorescence measurements and can be explained by the presence of the small molecular weight ␥ subunit (Fig. 1). In contrast, the NMR spectrum of whole D75N mutant had a much smaller number of distinct lines and much lower resolution, which reflects the absence of the cleaved subunit and the large size of the capsid (Fig. 4D). Some of the resonance lines observed for D75N showed the same chemical shift as with wt capsid, but it has broader lines, suggestive of a similar chemical environment but lower mobility. Although the ␥ subunits are part of the capsid, they have enough motion to be sampled by NMR.
Dissociation of the wt VLPs by 6 M urea produced large changes in the NMR spectra (Fig. 4E). An increase in signal to noise ratio was observed that is consistent with VLP dissociation. The increase was more pronounced for the wt protein, probably because of the presence of the small molecular weight ␥ subunit. Even upon dissociation the NMR spectrum of the D75N mutant was typical of a large protein, as expected for the dissociated, folded ␣ subunit (47 kDa) (Fig. 4F). Because of the broad peak observed at 8.8 ppm, we did not attempt to analyze the chemical shift dispersion in either spectrum in the presence of urea.
Cleavage-defective Coat Protein Reassembles into Capsids-The overall data show that the coat protein of the noncleaved D75N capsid retained substantial secondary and tertiary structure in 6 M urea. The hydrophobic core seemed to remain intact, as revealed by the fluorescence, CD, and NMR data. The fraction of dissociation in 6 M urea was the same for wt and mutant. To determine whether reassembly could occur after dissociation, gel filtration chromatography was utilized (Fig. 5). Assembled VLPs had an elution time compatible with an icosahedral particle with 140 Å radius (25). The monomers were expected to elute close to the total volume of the column (ϳ12 min). After dissociation of D75N VLPs with 6 M urea, about 50% reassembly occurred after injection into the HPLC sizing column (Fig.  5B). On the other hand, there was very insignificant reassembly of the wt particles after dissociation and unfolding by 6 M urea (Fig. 5A). As previously found with other dissociated capsid proteins (26,27), the monomeric subunits were adsorbed to the column.
Our results agree with the general idea that a noncleaved and folded capsid protein would be ready to encapsidate the nucleic acid. In contrast, an unfolded protein would be unable to revert to a folded, associated capsid.
Dynamics Probed by Pressure Studies: Mature Versus Cleavage-defective VLPs-Intuitively, one would predict that the nonprocessed capsid would be less stable than mature particles. At very low Ca 2ϩ concentrations (i.e. when particles are purified in the presence of EDTA) the procapsids are readily degraded by 1% SDS or in a disassembly mixture (28), whereas 90% of the capsid (cleaved) particles survive. There is no sig- nificant difference in stability if the procapsids and capsids are exposed to 10 mM Ca 2ϩ (9). In the studies reported here procapsids and capsids were stored in 10 mM Ca 2ϩ containing buffers. In Fig. 2, it was shown that dissociation (measured by light scattering) was no different between cleaved and noncleaved VLPs, whereas folding (shift in fluorescence spectra) of the wt VLP was much more susceptible to denaturation than the D75N VLP.
To examine whether the concerted dissociation and unfolding of the wt would be prevented if the final products were not free capsid proteins, high pressure experiments were performed. Hydrostatic pressure is a "clean" thermodynamic method for studying the stability of macromolecular assemblies (29 -31). The main finding of the pressure studies on viruses is that the isolated capsid and the assembly intermediates assume different partially folded states in the assembly pathway. In most of the cases, pressure induces the conversion of the icosahedral particle to a ribonucleoprotein intermediate state, where the coat protein is partially unfolded but bound to RNA (26,(31)(32)(33). This ribonucleoprotein state is readily reverted to viral particles when pressure is withdrawn. Fig. 6 shows that pressure exerted reversible effects on wt and D75N VLPs. The cleavage-defective VLP was more stable against pressure than wt as measured by fluorescence (Fig. 6A) or light scattering (Fig. 6B). The reaction was almost completely reversible in both cases (Fig. 6, C and D), which agrees with previous results (27,31) showing that pressure perturbation populates partially folded conformations of the coat protein bound to RNA that return to the native state after decompression. It is noteworthy that both wt and D75N particles eluted slightly more included after pressure treatment, probably because of reassembly into a more compact particle. The fluorescence and light scattering data overlap for wt and D75N particles. These results show that nonconcerted dissociation and denaturation of the un-  cleaved coat subunits only occur when they are free in solution. The greater sensitivity to pressure of cleaved wt FHV is consistent with the higher dynamics revealed by the rotation (Ta-ble I) and NMR (Fig. 4) measurements. The direct relation between protein flexibility (because of volume fluctuations) and its isothermal compressibility (34) is well established.

DISCUSSION
The structure of several icosahedral viruses has been determined with a remarkable level of precision (1-3). The atomic structure of viruses have revolutionized the field of virology by providing insights about the interactions of viruses with cells, drugs, and the immune system (8). However, the structures of viruses provide only frozen pictures of a final product of an assembly line. To dissect the steps of this assembly line, it is necessary to add dynamic studies to the structural information. To disassemble a virus is like dismantling the pieces of an intricate puzzle, in which the parts contain an elaborate code for self-assembly. We described here that by combining previous structural information of wild type and mutant virus-like particles of FHV with spectroscopic and hydrodynamic methods in a thermodynamic approach, it is possible to identify one of the important rules of this code. An amino acid substitution (D75N) prevents maturation cleavage of the ␣ coat protein locking the capsid as a provirion. The D75N provirion dissociates to folded coat proteins, and unfolding is not coupled to the dissociation process. On the other hand, mature wild type particles dissociate and unfold in a concerted fashion.
Nodavirus processing generates an infectious particle by cleaving the full-length ␣ subunit at residue 363 to generate ␤ and ␥ subunits. The x-ray structure reveals that the first 22 residues of FHV ␥ chains are amphipathic helices that interact with RNA about the icosahedral 3-fold axes and form a helical bundle at the fivefold axes (13) (Fig. 1). Our rotation and NMR data show that it is very likely that the cleavage results in the ␥ chains gaining significant high mobility. This mobility is probably important for virus-cellular membrane interactions, perhaps in a way similar to the mechanism of fusion suggested for several enveloped viruses (35)(36)(37)(38)(39). Whether the ␥ chain can act as a conduit for RNA translocation is still to be proved; however, recent studies have demonstrated dramatic membrane activity of residues 1-22 of ␥ chain (40,41). The mechanism for the cleavage of ␣ subunit was proposed based on site-directed mutation studies (17). Assembly places Asp 75 in a buried, hydrophobic environment resulting in an abnormally high pK a of 6.8. Under physiological conditions, Asp 75 is protonated and serves as a proton donor for a hydrogen bond to the carbonyl oxygen of Asn 363 . Assembly locally destabilizes the scissile bond between residues 363 and 364, making it susceptible to attack by a water molecule, which is probably activated by the anionic environment of the directly adjacent bulk RNA. The replacement of Asp 75 with Asn prevents the peptide cleavage.
Although there was no apparent difference between wild type and cleavage-defective VLPs with regard to dissociation by urea, the concerted dissociation and unfolding of the wild type particle may render it less stable (or metastable). During infection, a trigger must exist to provoke dissociation, which would be facilitated by maturation turning the particle metastable. A coupling of unfolding to dissociation may be crucial to tunnel the activation energy and speed up the reaction.
For a given urea concentration that resulted in dissociation of wt and mutant VLPs, the yield of reassembly was much higher for the cleavage-defective particle (D75N) as measured by gel filtration chromatography. Circular dichroism confirmed the existence of a folded, dissociated D75N coat protein, whereas the CD of urea-dissociated wild type coat protein did not show any significant secondary structure. The NMR of the dissociated D75N coat protein also suggests a high molecular mass subunit, with low mobility, characteristic of a folded structure. The wt particles were less stable to pressure than the D75N mutants. The higher flexibility resulting from the cleavage of ␣ into ␤ ϩ ␥ subunits increases the compressibility and makes the particle more sensitive to pressure. Several single-chain proteins are generally less flexible and more stable than their dimeric counterparts (42)(43)(44)(45).
A free energy diagram (Fig. 7) sketches the dissociation of particles into folded subunits and RNA, followed by unfolding of the coat protein. For the cleaved, mature particle, the difference in chemical potential between folded and unfolded subunits (␤ ϩ ␥) is either nonexistent or very small. This explains the low yield of reassembly from the dissociated, cleaved proteins when urea is diluted. Overall, our results demonstrate that the maturation/cleavage process targets the particle for an off pathway disassembly, because dissociation is coupled to unfolding. Because D75N is a paradigm for a short-lived ␣ subunit, the different properties should be crucial for the viral infection cycle. On one hand, a metastable state of the cleaved coat protein would contribute to the rapid dissociation concerted to unfolding when a specific cellular switch is turned on at the early stages of viral infection. On the other hand, after RNA replication and protein synthesis, the folded noncleaved coat protein would readily assemble into a ribonucleoprotein complex. Kinetic studies should provide additional insights about the role of metastable states in virus infectivity. The pressure stability of FHV wild type (circles) and D75N (triangles) mutant virus-like particles was analyzed by the spectral center of mass (A) and light scattering (B). To better appreciate the effect of pressure, the reaction was poised by adding 1 M urea, a subdissociating concentration. Reassembly was evaluated by size exclusion high performance liquid chromatography for wt FHV (C) and mutant after decompression (D).