Capsid Assembly in a Family of Animal Viruses Primes an Autoproteolytic Maturation-That Depends on a Single Aspartic Acid Residue*

Maturation of noninfectious nodavirus provirions oc- curs by autoproteolytic cleavage of most of the 180 copies of the a-protein that make up the icosahedral capsid. This maturation, which is much slower than viral assembly, produces an infectious particle that is more stable than the provirion and makes viral uncoating thermodynamically distinct from assembly, allowing assembly and (a time-delayed) uncoating to occur under similar conditions. The results of structural, computational, and molecular genetic studies suggest that maturation depends both on intrasubunit strain, produced during assembly, and on a critical aspartic acid residue. This residue lies in a hydrophobic pocket that is stabilized by intersubunit contacts. It is close to the scissile bond and exhibits an environmentally elevated pK,. The apparent involvement of a single acidic residue in the hydrolytic cleavage of a peptide bond contrasts with the involvement of 2 such residues in acid proteases. The nodaviruses are a family of small icosahedral viruses infecting insects, mammals, and fish (1-3). They are among the simplest of all animal viruses. The virus particle consists of 180 copies of the coat protein, which encapsidates the bipartite RNA genome. Only three proteins are Conditions

Maturation of noninfectious nodavirus provirions occurs by autoproteolytic cleavage of most of the 180 copies of the a-protein that make up the icosahedral capsid. This maturation, which is much slower than viral assembly, produces an infectious particle that is more stable than the provirion and makes viral uncoating thermodynamically distinct from assembly, allowing assembly and (a time-delayed) uncoating to occur under similar conditions. The results of structural, computational, and molecular genetic studies suggest that maturation depends both on intrasubunit strain, produced during assembly, and on a critical aspartic acid residue. This residue lies in a hydrophobic pocket that is stabilized by intersubunit contacts. It is close to the scissile bond and exhibits an environmentally elevated pK,. The apparent involvement of a single acidic residue in the hydrolytic cleavage of a peptide bond contrasts with the involvement of 2 such residues in acid proteases.
The nodaviruses are a family of small icosahedral viruses infecting insects, mammals, and fish (1)(2)(3). They are among the simplest of all animal viruses. The virus particle consists of 180 copies of the coat protein, which encapsidates the bipartite RNA genome. Only three proteins are encoded in the viral genome. The a-protein (the coat protein precursor) is encoded on RNA2; RNA1 encodes a replicase and a small protein of unknown function (1, 2). The simple particle and genome have made this system an attractive subject for the study of viral capsid structure and assembly as well as viral gene expression. Like the more common and more complex picornaviruses (which include important pathogens such as poliovirus, rhinovirus, and foot-and-mouth disease virus) (4), nodaviruses are initially constructed as provirions, which mature to an infectious virion by post-assembly cleavage of the subunits (5). In both virus families, the proteolysis reaction occurs near the interior surface of the protein shell a n d is apparently autocatalytic (5).
Nodavirus capsids display T = 3 icosahedral symmetry: the 60 icosahedral asymmetric units contain three copies of the a-protein, each in a slightly different, quasi-equivalent envi-* This work was supported in part by United States Public Health Service Grants AI22813 (to R. R. R.) and GM34220 (to J. E. J.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "uduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ** To whom correspondence should be addressed. ronment (see Fig. IA) (6). The full-length coat protein of Flock House virus (407 amino acids) assembles rapidly in uivo (within 5 min of synthesis), followed by a slow autoproteolytic cleavage of most subunits (5). The 363-residue NH,-terminal fragment @-protein) forms the virus capsid; the 44-residue COOH-terminal fragment (?-peptide) remains associated with the interior of the capsid. Maturation proteolysis is required for infectivity (7) and results in a marked increase in virion stability (5). At present, structural information is available only for nodaviruses in >he cleaved mature state (8)(9)(10). The recently refined 2.8-A structure of black beetle virus (10) has provided us the opportunity to examine the cleavage site in greater detail.
MATERIALS AND METHODS Structural Observations-Black beetle virus was crystallized in 50 II~M sodium phosphate, 0.55 M ammonium sulfate, 1% polyethylene glycol 8000 (Sigma) at pH 7 (11). The structure was solved by multiple isomorphous replacement and molecular averaging (8). The 2.8-A structure of the virus, including 10 ribonucleotides of the RNNicosahedral asymmetric unit, was refined to a final R-factor of 22.1% as described by Wery et ul. (10).
Development of Model a-Protein and Electrostatic Calculations-The procedure of Yang et al. (12) was followed to calculate the pK, ofAsp-75 using the atomic coordinates from the refined structure of black beetle virus (10) with hydrogen atoms generated by C H A R " (13). To model the uncleaved a-protein, the COOH terminus of the P-protein and the NH, terminus of the y-peptide were explicitly connected and then subjected to energy minimization by C H A R " (13). Amino acids within a 20-A radius of the cleavage site, including all hydrogens, were used in this calculation 1100 steps of Powell minimization; harmonic constraints applied to the positions of non-hydrogen atoms). Final coordinates of the non-hydrogen atoms, including the NH, terminus of the y-peptide, were shifted by 0.3-A r.m.s' from the refined crystallographic coordinates (10).
The resultant structure, with C H A R " partial charges (all atoms included) on the respective atoms, was used to calculate the pK, shifts by numerically solving a linearized Poisson-Boltzmann equation by the finite difference method with the program DELPHI (DELPHI Version 3.0, courtesy of Barry Honig, Columbia University) (14). Calculations were done with a grid spacing of -0.4 A. All bound waters were considered explicitly as part of the protein. The dielectric constants of the solvent and protein were set to 78.5 and 4.0, respectively, and the ionic strength was 0.145 M.
p H Profile of Autoproteolysis-Drosophila cells (Schneider's line 1) were suspended to 4 x lo' celldml in a complete growth medium containing Schneider's insect medium with 15% fetal bovine serum (CGM). Flock House virus was added at a multiplicity of 120 plaque-forming unithell and was allowed to attach for 1 h at 26 "C. Cells were then sedimented and resuspended to 5 x lo6 celldml in CGM. Aliquots were distributed onto 100-mm tissue culture plates and incubated at 26 "C. At 15 h postinfection, the medium was removed, and monolayers were rinsed with 10 ml of 25 n w PIPES (pH 6.8), 100 n m NaCl, 0.1% BSA.
Cells were then covered with 5 ml of methionine-deficient Grace's insect medium (Life Technologies, Inc.) with 50 pCi of [36Slmethionine (Amersham Corp.).
Freshly isolated [35Slmethionine-labeled provirions, resuspended in buffers at different pH values, were incubated at room temperature. Samples (30 pl) were withdrawn immediately a h r resuspension and 3, 6,5, and 24 h later. The samples were mixed with an equal volume of 2 x electrophoresis buffer and stored frozen until analysis on 12% SDSpolyacrylamide gel (15). Gels were fixed and exposed to Kodak X-ARE, film. The extent of a-protein cleavage was quantified by densitometry using a Zenith densitometer.
Mutagenesis and Analysis of Mutants and pH Profile of Cleavage-Conditions for site-directed mutagenesis, transfection of Drosophila cells, [35S]methionine labeling, purification of virus particles, and gel electrophoresis were as described (7). Briefly, 200 ng of purified virion RNA1 plus 200 ng of RNA2 transcripts were used to transfect lo7 Drosophila cells. After 15 h, 50 pCi of [%]methionine was added to the medium, and incubation was continued for another hour at 26 "C. Virus particles were purified by pelleting the cell lysate through a 30% (w/w) sucrose cushion, followed by a 5-20% (w/w) sucrose gradient. Peak fractions were pooled and incubated for 24 h at room temperature to allow cleavage (5). Viral proteins were separated on 12% SDS-polyacrylamide gel (15). The gel was fluorographed after treatment with Amplify (Amersham Corp.).

RESULTS AND DISCUSSION
X-ray Structure of Mature Particle CP + ?)-The sites of maturation cleavage in the three identical gene products that form the icosahedral asymmetric unit of black beetle virus lie close to subunit interfaces and are near the internal surface of the shell (Fig. 1, A and B ) . The environments of each of the three unique clecvage positions are very similar, with a r.m.s. deviation of 0.5 A for a-carbons of residues shown in Fig. 2 (A and B ) and for a-carbons within 15 a of the cleavage site. The COOH terminus of the P-protein (Asn-363) lies within a hydrophobic pocket, and the NH,-terminal amine of the y-peptide (Ala-364) is exposedJo the aqueous environment in the interior of the virus, 8.9 A from the COOH-terminal carbon of the P-protein for the C subunit ( Fig. 2, A and B ). Neither residue is accessible to exogenous proteases. In the mature virion, the carboxyl group of the cleavage-produced COOH-terminal Asn-363 and the side chain of Asp-75 share a proton (0-H-0; 2.5 A).
Model of a-Protein-The mechanism for the cleavage-induced maturation of nodaviruses proposed here is based on observations of the mature subunit and a model of the intact a-protein in the procapsid. The latter was reconstructed from the refined x-ray model of the cleaved Pand y-proteins in the mature crystalline virus. In the reconstructed a-protein, the portion corresponding to P is essentially unchanged from it! position in the mature protein (overall r.m.s. movement of 0.4 A for all residues within 15 A of the cleavage sites in the three subunits). By contrast, the r.m.s. shifts in the positions of the amino-terminal residues of the y-protein are 2.7,0.9, and 0.3 A for residues 364-366, respectively, averaged over the three subunits. This model of a-protein suggests that the separation between Asn-363 and Ala-364 caused by the autoproteolytic maturation can largely be accounted for by local movement of the amino terminus of the y-protein and a minimal change in Asn-363. This conclusion seems reasonable, partly because the observed interactions of the carboxyl-terminal residue of the The cleavage sites between A and B subunits are near quasi-2fold axes. The quasi-2-fold contacts form an angle of 144" between the planes of the subunits. The C-C contact at the true 2-fold axis forms a flat contact, a distinctly different geometry that utilizes the otherwise disordered residues 20-30 of the capsid protein and structural RNA to fill the groove between subunits (9). B , a stereo ribbon drawing of the C subunit showing the tertiary and secondary structures of the cleavage site. The top of the ribbon drawing is at the outer surface of the virus. The subunit is caged in a 2O-A-thick trapezoid corresponding to the C subunit trapezoid of the T = 3 icosahedron in A . The y-peptide on the bottom is exposed to the aqueous interior of the virion. The bottom of the P-barrel forms the roof of the cleavage site. Residues 20-31 of the C subunit form a complex web of interactions involving the adjacent asymmetric unit (8, 10). These residues (which are disordered in the A and B subunits) help occlude the cleavage site. The disordered amino acids 32-54 must cross this face of the subunit near Am-363 and may help to limit access to the cleavage site. Except residues 20-30, the tertiary structure of the A and B subunits is nearly identical to that of the C subunit (6). The role of the C subunit NH, terminus in occluding the cleavage site is filled by residues in the @-barrel of a neighboring subunit in the A and B subunits. The ribbon diagram was produced with MOLSCRIPT (20). P-protein (Am-363) with Asp-75, Tyr-176, and Ala-360 ( Fig.  2 B ) , each of which is conserved in the four characterized nodaviruses (16), can be maintained in the proposed model of the a-protein. The amino-terminal residue of the y-peptide in the mature virus interacts less extensively with adjacent residues. In addition, major structural changes would be required to form the a-protein by maintaining the NH, terminus of the y-peptide and moving the COOH-terminal segment of the P-protein into position where a peptide bond could be formed between Asn-363 and Ala-364. Hence, there is good reason to believe that the conformation of the COOH-terminal portion of the cleavage site in the mature P-protein is very similar to that in the a-protein of the procapsid.
Cleavage Mechanism-The mechanism that we have proposed for the cleavage reaction depends on four features observed in the x-ray structure and/or in the model of the a-protein. (a) Hydrogen bonds formed by the amide side chain of Asn-363 direct the carbonyl oxygen of the scissile peptide bond toward Asp-75 (Fig. 2 B ) . The functional groups that interact with Asn-363 and Asp-75 (Fig. 2 B ) are conserved within the nodavirus family (16). ( b ) Asp-75 is at least partially protonated in the a-protein. The environmental effects of the hydro-A / A364 C subunit of black beetle virus crystallized at pH 7.0 (11). The relationship between the COOH-terminal carboxylic acid of the &protein and the Fro. 2. S t r u h of solventoccluded cleavage site. A, the electron denaity map and the refined molecular model of the cleavage site of the side chain of Asp75 (red) is accentuated by the close proximity of the corresponding electron density. The cleavage products, the COOH-terminal carboxylic acid of the &protein, and the NH&erminal amine of the ypeptide are in green. Although density from the C subunit is shown, the structures of the cleavage sites of all three subunits of black beetle virus are very similar (10). B, the molecular model highlighting the network the cleavage site that are shown are blue. F'unctional groups that participate in hydrogen bonding interactions are conserved within the nodavirus of hydrogen bonds that direct interactions betweenhn-363 (the COOH terminus of the &protein) andhp-75. The hydrophobic amino acids amund family (12). Asn-363 is the last residue in a type 1 &turn, where both the side chain and peptide amides donate hydrogens to the carbonyl of Ala-360. The side chain carbonyl ofhn-363 accepts a hydrogen bond from the phenolic hydroxyl of Tyr-176. Tyr-176 is on the D strand of the coat protein subunit's &barrel. These interactions direct the COOH terminus ofAen-363, created by proteolysis, toward the side chain carboxylic acid ofhp-75, forcing protonation of one or both of the two acids at physiological pH and formation of a hydrogen bond. The COOH-terminal carboxyl a h accepts a hydrogen bond from the peptide amide of Ser-365 of the ypeptide. Asp75 accepts a hydrogen bond from the peptide amide of Met-366. The black beetle virus structure has been refined to an R-factor of 21% the extraordinary similarity between quasi-symmetrically related subunits is investigated in greater detail elsewhere (10). This figure was generated using MACINPLOT (21) and FRODO (22). phobic cleavage site, including the lack of an identifiable countenon and the peptide carbonyl acting as a hydrogen bond acceptor, are expected to raise the free energy of ionization (and the PK,) of Asp-75 (17, 18). (c) Protonated Asp-75 acts as a general acid, polarizing the main chain carbonyl ofAsn-363 and making it susceptible to nucleophilic attack by a water molecule (Fig. 3A). The active water molecule is incorporated into the COOH-terminal carboxylic acid of the p-protein and cannot be identified in the structure of the mature virus. In a hydro-vate proteolysis by destabilizing the geometry of the scissile peptide bond. ( d ) "he increased stability of the mature capsid is thermodynamically linked to cleavage and helps drive proteolysis. "he absence of cleavage prior to assembly suggests that quaternary interactions initiate autoproteolytic maturation. Provirion assembly probably has a role in stabilizing the hydrophobic cleavage site (altering the p K n of Asp-75) and adding strain to the scissile bond of Asn-363-Ala-364.
To test the above hypothesis, the protonation state of Asp-75 phobic environment, like the cleavage site, where a proton can-was analyzed using a linearized Poisson-Boltzmanu equation not be readily transferred, general acid catalysis may be an and estimated solvation energy as implemented in the program important hydrolytic mechanism. It is also possible that inter-DELPHI (14). The calculations were based either on the model action of Asp75 with the Asn-363 peptide carbonyl may acti-of the uncleaved a-protein or on the 2.8-A refined coordinates of

. Proposed mechanism for autoproteolytic maturation of nodavirus capsid protein results from chemical (A) and thermodynamic ( B ) factors affected by assembly and maturation. A,
Asp-75 is highly protonated even a t neutral pH by virtue of its burial in a hydrophobic environment stabilized by the association of subunits during assembly (1). It forms a hydrogen bond with the carbonyl of the Asn-363-Ala-364 peptide bond, and this is sufficient to make it susceptible to nucleophilic attack by water (2) to form a tetrahedral intermediate (3). The water molecule is not trapped within the cleavage site, but probably originates from within the hydrophilic interior of the virus capsid. The intermediate can relax by loss of the amine from the nascent peptide (4,5), yielding the hydrolyzed peptide bond at the cost of one water molecule (6). The hydrophobic protonationlactivation of a catalytic acid residue resembles the activation of the catalytic Asp in lysozyme (23). Conversely, in the typical acid protease, it is the close interaction between 2 aspartates that is responsible for the elevated pK, of one of the catalytic acids. B , shown is a scheme describing the thermodynamics of virus assembly and maturation. The cartoon relates the activation and relaxation of the coat protein subunits in the course of capsid assembly and maturation. The free energy of activated a* (aprotein in the provirion, but not including the stabilizing free energy from quaternary interactions) is greater than that of the free a-protein by AGl, and this energy may stabilize an intermediate in the cleavage. Quaternary interactions stabilize formation of the relatively unstable provirion favoring spontaneous assembly. Autoproteolytic maturation is energetically driven by AG;, the combination of energy gained by relaxation of the strained provirion conformation of the coat protein and the change in the free energy of intersubunit interactions in the mature capsid. Although the provirion and mature particle are drawn as substantially different structures, the two particles are nearly identical in their physical parameters (sedimentation coefficient and diameter). The gain in particle stability is accompanied by a very subtle change in quaternary structure. black beetle virus modified to eliminate the charge on the COOH terminus of the p-protein. In both model systems, the pK, of Asp-75 is elevated to pH -6.0 in all three quasi-equivalent subunits. Thus, a significant fraction of Asp-75 is protonated near pH 6, the optimal pH of the maturation cleavage reaction in Flock House virus (data not shown).
The plausibility of this mechanism was tested by site-directed mutagenesis of Asp-75 in Flock House virus. The coat proteins of black beetle virus and Flock House virus share 87% identity. Residues near the cleavage site are more strictly conserved (16). Replacement of Asp-75 with glutamic acid (D75E),  (D75(wt FHV)). Only the a-protein is observed when Flock House virus Asp-75 is mutated to valine (D75V), glutamate (D75E), asparagine (D75N), or threonine (D75T). The y-peptide (5 kDa) is not visible on this gel. The identity of bands migrating near the 14-kDa marker is not known. asparagine (D75N), threonine (D75T), or valine (D75V) resulted in the production of noninfectious particles that did not undergo maturation cleavage (Fig. 4). The D75N mutant is an isosteric mutation, while the D75E mutant retains the carboxylic acid of the wild-type protein, but alters the relative position of the catalytic group. Drosophila cells were transfected with purified Flock House virus RNA1 plus mutant transcript RNA2 to initiate one round of infection. Plaque assays showed that progeny were not infectious within the detectable limit (<0.1% of the wild-type transfection). The loss of infectivity correlated with the absence of maturation cleavage. After 24 h at room temperature, which is sufficient for 90% cleavage of the wildtype a-protein, only uncleaved protein was observed in purified mutant capsids (Fig. 4).
When Asn-363, which positions the scissile bond with respect to Asp-75 (Fig. 2B), was replaced by alanine or aspartic acid, the synthesis of noninfectious, cleavage-defective particles (7) was again observed. The mutant N363T does undergo cleavage imder some conditions, but the cleavage is drastically slowed down under all conditions. Cleavage Kinetics: Subunit Communication or Quasi-eguivalent Environments?-Maturation kinetics of Flock House virus do not follow a simple first-order rate law ( 3 , as expected for an intramolecular reaction, since the cleavage rate slows more rapidly than would be expected for such a reaction. If the cleavage reaction at a given site depends on a local strain, communicated to that site by the structure of the procapsid (Fig. 3B), the relief of such strain in the cleavage process could cause the excess slowing of the overall cleavage reaction. Such a rationale requires considerable intersubunit communication: on the average, each cleavage event must allow the capsid to relax to a slightly more stable form, thus decreasing the cleavage rate of the remaining a-proteins. Although a unique interpretation of cleavage kinetics cannot be formulated because the three quasi-equivalent cleavage sites could give rise to three distinct first-order rate constants (see below), the above possibility is attractive, especially since cleavage sites are located near subunit interfaces. In such a case, the provirion could be considered as a reactive intermediate, the instabilities of which help to overcome the energy barrier for proteolysis. Equation 1 is a rate expression for one process of this type. In the reaction described by Equation 1, successive cleavage events progressively increase the stability of the procapsid, thus progressively increasing the energy barrier for proteolysis (AGS), so that the rate of cleavage will decrease more rapidly than for a first-order process.

d[a]ldt = -k([a]/[a"])[a]
(Eq. 1) In a reaction whose progress is described by this rate law, the effective rate constant for the cleavage of the a-protein (k([a]/ [a"])) will decrease from a n initial value of k at [a]/[ao] = 1 to essentially zero when the fraction of uncleaved a-protein approaches zero. The cleavage kinetics observed by Gallagher and Rueckert (5) can be described by Equation 1, where k = 0.315 h-', which would correspond to a half-time of 2.2 h if the effective rate constant for the reaction remained unchanged throughout. In the simplest case, shown here, the same rate constant (k) would apply to all three classes of subunits in the provirion. In addition, the progressive gain in stability would have to be distributed uniformly over the entire capsid so that, in effect, the cleavage rate on one side of the virus would be affected by the cleavage of a subunit on the other side.
Alternatively, the multiphasic first-order kinetics can be interpreted as a n effect of the T = 3 geometry (Fig. lA), where 120 subunits cleave with a fast rate (tin = 2.2 h) and 60 subunits cleave with a slow rate ( t , = 13.1 h) (5). This interpretation suggests that the scissile Asn-Ala bond in two classes of subunit is cleaved at nearly the same rate and that one of the three classes (A, B, or C) cleaves more slowly. In fact, the A and C subunits do possess features that distinguish them from the B subunit. The C subunit is located near an icosahedral 2-fold axis ( Fig. 1 B ) ; the protein loop that includes Asp-75 interacts with the peptide in the groove coming from the same C subunit.
On the other hand, the y-peptide of the A subunit is in a distinctly different environment than those of the B and C subunits. The five A subunit y-peptides around each icosahedral &fold axis form a five-helix bundle (10). Unfortunately, these two extreme cleavage models cannot be distinguished easily by kinetics since both fit the data well at t < 24 h. Although these models predict different behavior a t much longer time intervals, a n accurate assessment of reaction kinetics after several half-times is very difficult due to contributions that could arise from an even slight heterogeneity of the sample. And it goes without saying that more sophisticated models (involving a more localized intersubunit communication or a combination of intersubunit communication and different initial rates for quasi-equivalent subunits) require too many variables to be tested critically. In spite of this ambiguity, we strongly favor the model based on subunit communications since at very long time intervals, i.e. weeks to months, a few percent of a-chains remain. Although one could argue that the uncleaved a-chains arise from a minor noninfectious population of defective provirions that do not undergo cleavage or from a population of defective provirions that undergo only a limited cleavage, we discount this possibility on the basis that intersubunit communication has also been implicated in the autoproteolytic maturation of hepatitis A (191, a picornavirus. Bishop and Anderson (19) point out that cleavage of the 60 copies of the precursor protein V P O to the mature proteins VP4 and VP2 is linear with time, i.e. it does not follow simple firstorder kinetics. In fact, the observed kinetics suggest that early cleavage events accelerate subsequent cleavage. This rate-enhancing effect contrasts with the rate-attenuating effect pro-posed for the nodaviruses. It is likely that the chemical details of picornavirus cleavage differ from those of nodavirus cleavage. It is important that, in both virus families, there appears to be a cooperative effect influencing cleavage as reflected by the failure to follow first-order kinetics for what should be first-order reactions.
In summary, we believe that assembly-based instability of the virus capsid and the alteration of the pK, ofAsp-75 provide an attractive thermodynamic and chemical basis for the observed cleavage of a normally inert peptide during viral maturation. We hypothesize that maturation-induced cleavage will occur efficiently only within a provirion complex in which there are localized "high energy" regions and the protonated aspartic acid. If such regions do exist and if they contribute to postassembly modification of the coat protein, viral uncoating would become thermodynamically distinct from viral coating, rather than simply the reverse of the coating process. Thus, in the life cycle of the virus, it seems reasonable that both assembly and disassembly should be thermodynamically favorable under the conditions where each occurs and that both may occur under similar conditions. In addition, it may be advantageous for the disassembly process to involve a pair of pand y-proteins, where y may be able to dissociate from the particle during the uncoating process (24).