Venezuelan Equine Encephalitis Virus nsP2 Protein Regulates Packaging of the Viral Genome into Infectious Virions

Abstract

Alphaviruses are one of the most geographically widespread and yet often neglected group of human and animal pathogens. They are capable of replicating in a wide variety of cells of both vertebrate and insect origin and are widely used for the expression of heterologous genetic information both in vivo and in vitro. In spite of their use in a range of research applications and their recognition as a public health threat, the biology of alphaviruses is insufficiently understood. In this study, we examined the evolution process of one of the alphaviruses, Venezuelan equine encephalitis virus (VEEV), to understand its adaptation mechanism to the inefficient packaging of the viral genome in response to serial mutations introduced into the capsid protein. The new data derived from this study suggest that strong alterations in the ability of capsid protein to package the viral genome leads to accumulation of adaptive mutations, not only in the capsid-specific helix I but also in the nonstructural protein nsP2. The nsP2-specific mutations were detected in the protease domain and in the amino terminus of the protein, which was previously proposed to function as a protease cofactor. These mutations increased infectious virus titers, demonstrated a strong positive impact on viral RNA replication, mediated the development of a more cytopathic phenotype, and made viruses capable of developing a spreading infection. The results suggest not only that packaging of the alphavirus genome is determined by the presence of packaging signals in the RNA and positively charged amino acids in the capsid protein but also that nsP2 is either directly or indirectly involved in the RNA encapsidation process.

INTRODUCTION

Members of the Alphavirus genus in the Togaviridae family are a group of widely distributed human and animal pathogens (1, 2). They continuously circulate on all continents between mosquito vectors and vertebrate hosts (1). The New World alphaviruses, which include the Venezuelan (VEEV), eastern (EEEV), and western (WEEV) equine encephalitis viruses, represent a serious public health threat and are continuously isolated from mosquitoes in the North, South, and Central Americas and the Caribbean islands (3). These viruses induce severe, highly debilitating diseases in humans and equids with frequent lethal outcome (2). Nevertheless, in spite of a continuous public health threat, no approved vaccines or therapeutic means have been developed for any encephalitogenic alphaviruses. The experimental, live VEEV vaccine TC-83 was developed more than 4 decades ago, and the attenuated phenotype of this strain is a result of two point mutations in the viral genome (4, 5). One of these is located in the 5′ untranslated region, and the second was identified in the E2 glycoprotein. Thus, a possibility of reversion to the wt phenotype during virus replication in vivo remains a strong concern. Development of new, more attenuated alphaviruses with a stable, irreversible phenotype is complicated by our limited knowledge about the mechanism of their genome replication, formation of infectious virions, and virus-host interactions. Thus, further mechanistic studies of alphavirus replication and pathogenesis are in strong demand. Moreover, the reasonably simple strategy of alphavirus genome replication and virus assembly, the exceptionally high mutation rates, and the availability of reverse genetics systems (6, 7) make these viruses a very important tool for studying development of new characteristics and understanding the underlying molecular mechanisms of viral evolution.

VEEV contains a single-stranded, ∼11.5-kb RNA genome of positive polarity (8). It mimics the structure of cellular messenger RNAs, in that it has a cap structure at the 5′ terminus and a poly(A) tail at the 3′ terminus (1). Genomic RNA encodes only a few proteins. Four nonstructural proteins are translated directly from the genomic RNA and assemble the replication complex (RC), which mediates viral genome replication and transcription of the subgenomic RNA. The subgenomic RNA serves as a template for translation of virus-specific structural proteins, which after multiple steps of processing, form the infectious, genomic RNA-containing virions. One of the structural proteins, capsid protein (CP), has numerous functions in VEEV replication. Its primary function, as with CPs of other alphaviruses, is to bind to viral genomic RNA (1, 9). This RNA binding is mediated by the amino-terminal CP domain, which contains a high concentration of positively charged amino acids. Its other function, also related to virion formation, is to assemble icosahedral nucleocapsid, which in the process of virus budding, attains a lipid envelope with imbedded glycoprotein spikes (10). The third role of VEEV CP is in virus pathogenesis and development of the cytopathic effect (11). This function is mediated by a short peptide located in the CP amino terminus (12, 13). This peptide contains both a nuclear localization signal (NLS), which interacts with nuclear import receptors, importin-α and -β, and a supraphysiological nuclear export signal (supraNES), also termed helix I, which interacts with nuclear export receptor CRM1. The tetrameric complex of CP, importin-α/β, and CRM1 inhibits the nuclear pore function and blocks active transport of NLS-containing proteins through the nuclear pore complex. This inhibition of nucleocytoplasmic trafficking profoundly affects transcription of cellular ribosomal and messenger RNAs (12, 13).

In our recent studies, we have designed a VEEV variant, in which all of the positively charged amino acids in the CP-specific RNA-binding domain were replaced by glycines, alanines, serines, and asparagines. These mutations did not abrogate virion assembly; however, the released particles were predominantly genome-free and are referred to as subviral particles (SVPs). The virus stocks generated by these subviral particle-producing mutants had infectious titers a few orders of magnitude below those normally measured in the stocks of the wild-type (wt) virus encoding natural CP. Moreover, the introduced mutations also destroyed the capsid-specific NLS and made virus replication incapable of interfering with the development of the antiviral response, resulting in release of high levels of cytokines by the infected cells. Importantly, further studies of this virus demonstrated that it represents a very interesting system for investigating virus evolution and the mechanisms of virus-specific protein function in the formation of infectious virions.

We present here the results of a detailed investigation of the evolution of the SVP-producing, noncytopathic VEEV mutant toward a more efficient packaging of viral RNA and a more cytopathic phenotype. VEEV variants encoding mutated CP accumulated adaptive point mutations, which were identified in the CP- and nsP2-coding sequences. Our data demonstrate that mutations in the previously described, CP-specific helix I can increase virus genome packaging almost 100-fold, and adaptive mutations in nsP2 have a positive effect not only on infectious titers but also on the development of cytopathic effect. These results suggest new roles for VEEV nsP2 and CP-specific helix I in virus replication.

MATERIALS AND METHODS

Cell cultures.

The BHK-21 cells were kindly provided by Paul Olivo (Washington University, St. Louis, MO). Vero cells were kindly provided by Scott Weaver (University of Texas Medical Branch at Galveston, TX). The NIH 3T3 cells were obtained from the American Type Culture Collection (Manassas, VA). These cell lines were maintained at 37°C in alpha minimum essential medium (αMEM) supplemented with 10% fetal bovine serum (FBS) and vitamins.

Plasmid constructs.

Standard recombinant DNA techniques were applied for the construction of the plasmids used in the present study. The modified CP gene was assembled from oligonucleotides using PCR. The resulting CP-coding sequence contained 26 codons of basic amino acids replaced by those coding glycines, alanines, serines, and asparagines. The resulting plasmid was named pVEEV/Cm/GFP. The pH/2A-C1 plasmid contained cDNA of a helper genome encoding noncytopathic capsid (14) and lacked the subgenomic promoter. pVEErep/GFP/Pac replicon-containing plasmid and pVEEV/GFP have been described elsewhere (14, 15). Adaptive mutations were cloned into the originally designed mutant virus genome- and replicon genome-containing plasmids using PCR and other standard cloning techniques. In all of the plasmids, cDNAs of viral and replicon genomes were placed under the control of the SP6 promoter. Sequences were always verified by sequencing of the cloned fragments in the final constructs. The sequences of the recombinant genomes and details of cloning procedures can be provided upon request.

In vitro RNA transcription and electroporation.

Plasmids were purified by centrifugation in CsCl gradients. Before the transcription reaction, the viral and replicon genome-coding plasmids were linearized using a MluI restriction site located immediately downstream of the poly(A) sequence. RNAs were synthesized by SP6 RNA polymerase in the presence of a cap analog according to the manufacturer's recommendations (Invitrogen). The yield and integrity of the transcripts were analyzed by agarose gel electrophoresis. Aliquots of transcription reactions were used for electroporation without additional purification. Electroporation of BHK-21 cells was performed under previously described conditions (16). Media were harvested at ∼24 h postelectroporation. Packaging of the defective viral genomes was performed by coelectroporating their in vitro-synthesized RNA and helper RNA genomes. Stocks of infectious viral particles, containing defective viral genomes, were also harvested within 24 h postelectroporation.

Titers of plaque-forming viruses were determined by a standard plaque assay on Vero cells (17). The infectious titers of noncytopathic viruses and packaged replicons were determined by infecting BHK-21 cells (5 × 10⁵ cells/well) in six-well Costar plates with 10-fold dilutions of the samples and counting the number of green fluorescent protein (GFP)-positive cells after a 6 h of incubation at 37°C in the CO₂ incubator. In some experiments with VEEV mutants lacking GFP gene, Vero cells in six-well Costar plates were infected with 10-fold dilutions of the samples and after an overnight incubation at 37°C under agarose cover, fixed with paraformaldehyde, and immunostained with antibodies specific to either VEEV nsP2 or VEEV structural proteins and corresponding fluorescent secondary antibodies.

The colony-forming activity of replicons was determined by electroporating BHK-21 cells with 1 μg of the in vitro-synthesized RNA. Different aliquots of the electroporated cells were seeded into 100-mm dishes and incubated in the presence of puromycin at a concentration of 5 μg/ml. Colonies were stained by crystal violet after 5 to 6 days of incubation at 37°C.

Analysis of virion release.

BHK-21 cells were infected at an MOI of 20 infectious units (inf.u)/cell for 1 h and then overlaid with a complete medium. At 6 h postinfection, cells were washed three times with phosphate-buffered saline (PBS) to remove the FBS and overlaid with serum-free medium (VP-SFM; Gibco). Media were harvested at 20 h postinfection. Viral particles and SVPs were pelleted from 1 ml of media by ultracentrifugation at 50,000 rpm for 1 h at 4°C in a TLA-55 rotor using a TL-100 tabletop ultracentrifuge (Beckman). Pelleted samples were analyzed by SDS-PAGE, followed by Western blotting. Western blots were performed using anti-VEEV TC-83 antibodies (a generous gift from R. Tesh, University of Texas Medical Branch), followed by treatment of the membranes with infrared dye-labeled secondary antibodies. For quantitative analysis, the membranes were scanned on a LI-COR imager.

Analysis of viral RNA synthesis.

To analyze the synthesis of virus-specific RNAs, 5 × 10⁵ Vero cells in six-well Costar plates were infected at the same multiplicity of infection (MOI) of 20 inf.u/ml, with samples of mutant viral genomes packaged into infectious virions. At 3 h posttransfection, media in the wells were replaced by 1 ml of αMEM supplemented with 10% FBS, ActD (1 μg/ml), and [³H]uridine (20 μCi/ml). After 4 h of incubation at 37°C, total cellular RNAs were isolated by TRIzol (Invitrogen) according to the manufacturer's protocol and then denatured with glyoxal in dimethyl sulfoxide and analyzed by agarose gel electrophoresis using the previously described conditions (18). Gels were impregnated with 2,5-diphenyloxazole (PPO), dried, and autoradiographed. For quantitative analysis, bands were excised from the dried gels and radioactivity was measured by liquid scintillation counting.

Analysis of protein synthesis.

A total of 5 × 10⁵ Vero cells in six-well Costar plates were infected at the same MOI of 20 inf.u/ml with samples of viral genomes packaged into infectious virions. At 6 h postinfection, proteins were metabolically labeled by incubating for 30 min in 0.8 ml of Dulbecco modified Eagle medium lacking methionine, supplemented with 0.1% FBS, and 20 μCi of [³⁵S]methionine/ml. After this incubation, the cells were harvested and dissolved in standard protein loading buffer. Equal amounts of proteins were subjected to SDS–10% PAGE. After electrophoresis, the gels were dried and autoradiographed. Quantitative analysis of the radioactivity in the specific bands was performed using a Storm phosphorimager.

IFN-β measurement.

NIH 3T3 cells were infected with different viruses at an MOI of 20 inf.u/cell. Media were harvested at 20 h postinfection, and the pH in the media was stabilized by adding HEPES buffer (pH 7.5) to 0.01 M. Concentrations of beta interferon (IFN-β) in the samples were measured with the VeriKine Mouse Interferon Beta ELISA kit according to manufacturer's recommendations (PBL InterferonSource).

RESULTS

Mutations in CP-specific helix I can strongly increase infectious titers of VEEV mutants.

In a previous study, we designed a recombinant, TC-83-based VEEV variant, which encoded a strongly mutated CP. The CP gene contained 26 mutations, which led to replacement of most of the positively charged amino acids (aa) in the amino-terminal domain by glycines, alanines, serines, and asparagines (Fig. 1A). These mutations were expected to have a deleterious effect on the CP's ability to interact with viral genomic RNA and package it into infectious virions. The introduced mutations also inactivated the previously identified CP-specific NLS and abrogated CP's ability to function in the inhibition of nucleocytoplasmic trafficking. These mutations made the VEEV mutant incapable of inducing the characteristic cytopathic effect (CPE) in vertebrate-derived cell lines. After the mutations were introduced, two positively charged aa were left unaltered in helix I/supraNES, positioned between aa 37 and 52 of CP Fig. 1A), as it was surmised that they may be important in dimerization of capsid protein. Helix I is known to be responsible for dimerization of capsid protein during Sindbis virus nucleocapsid assembly (19, 20). Four more positively charged residues remained in the conserved peptide, located between aa 110 and 127 (Fig. 1A). This conserved, VEEV CP-specific peptide (helix 2) was previously proposed to fold into a short α-helix (21) and to be involved in binding to the ribosomes, and recognition of viral genomic RNA (9, 22). Multiple functions proposed for helix 2 need further, more detailed investigation, which is beyond the scope of the present study. The constructed recombinant genome, termed VEEV/Cm/GFP, also contained a GFP gene under the control of a subgenomic promoter (Fig. 1A). GFP expression was used for evaluation of virus titers, because the designed mutant virus was incapable of inducing plaque formation, even after prolonged incubation. Under agarose cover, VEEV/Cm/GFP developed very small, pinpoint foci, which were formed by dividing cells, containing replicating viral genomes (data not shown).

Fig 1.

Mutations in CP-specific helix I have a positive impact on infectious titers of VEEV encoding CP with mutated RNA-binding domain. (A) Schematic representation of the VEEV genome encoding wt and mutated CP. The amino acid sequences of helix I and putative helix 2 are presented. The residual positively charged amino acids, which were not mutated in VEEV/Cm/GFP are indicated with blue letters. The adaptive N47Y mutation, identified in helix I of the variant capable of developing spreading infection, is indicated. (B) BHK-21 cells were electroporated with 4 μg of in vitro-synthesized RNA of VEEV/GFP and VEEV/Cm/GFP variants. Media were replaced at the indicated time points, and titers of infectious viral particles were determined as described in Materials and Methods. (C) BHK-21 cells were electroporated with 4 μg of in vitro-synthesized RNA of VEEV/Cm/GFP and VEEV/Cm1/GFP variants. Media were replaced at the indicated time points, and titers of infectious viral particles were determined as described in Materials and Methods. The dashed line indicates the limit of detection.

The abundant mutations introduced into the CP-coding sequence had a deleterious effect on the ability of the virus to produce infectious, genome-containing virions. At 24 h after electroporation, BHK-21 cells transfected with the in vitro-synthesized VEEV/Cm/GFP RNA, produced infectious virus of almost 6 orders of magnitude lower titer than cells transfected with the control VEEV/GFP RNA (Fig. 1B). This difference in titers did not result from less efficient transfection, because at 6 h after electroporation, in both samples, essentially all of the cells demonstrated GFP expression, indicative of virus replication. At the next passage, samples harvested after VEEV/Cm/GFP RNA electroporation demonstrated no detectable presence of virus capable of developing a spreading infection, even in highly permissive BHK-21 cells.

We have seen in our previous studies that alphaviruses consistently exhibit outstanding abilities for evolution and adaptation to the selection conditions (23–26). Introduction of 26 mutations, which eliminated the CP's positive charge, were expected to make reversion to a more efficient genome packaging phenotype an impossible event. Nevertheless, we tested whether the newly designed mutant could evolve to a phenotype characterized by higher infectious titers. Electroporation-derived, low-titer samples of VEEV/Cm/GFP were passaged five times in BHK-21 cells. At each passage, cells were infected with 1 ml of a 10-ml portion of viral particle-containing media harvested from the previous passage at 48 h postinfection. After this procedure, we detected the presence of variants capable of developing a spreading infection. These variants also formed larger foci of GFP-positive cells under agarose cover (see Materials and Methods for details). Virus from one of the largest GFP-positive foci was isolated, and its entire genome was sequenced. The only putative adaptive mutation, N47Y, was identified in the CP-specific helix I. It was introduced into the originally designed construct, and the VEEV/Cm1/GFP variant (Fig. 1C) demonstrated a strong increase in infectious titers, likely resulting from the more efficient RNA packaging (Fig. 1C). However, the titers remained 4 orders of magnitude below those achieved by VEEV/GFP, encoding the wt CP. VEEV/Cm1/GFP variant remained noncytopathic for BHK-21 cells and thus was incapable of forming plaques (see Fig. 4 for details).

Fig 4.

Adaptive mutations in nsP2 can increase infectious titers of VEEV CP mutant independently of the CP-specific adaptive mutations. (A) Schematic representation of the genomes of VEEV CP mutants used in these experiments. The introduced adaptive mutations are indicated. (B) Viral genomes containing indicated mutations were packaged to high titers into infectious viral particles, and 5 × 10⁵ Vero cells in six-well Costar plates were infected at an MOI of 20 inf.u./cell as described in Materials and Methods. At the indicated times, media were replaced, and infectious titers were determined by plaque assay on Vero cells.

Mutations in VEEV nsP2 cause an increase in infectious titers.

Next, we modified the selection system and continued passaging of the newly designed variant VEEV/Cm1/GFP in Vero cells, which are less productive than BHK-21 cells in terms of expression of virus-specific proteins (IF unpublished data). To select the most efficiently replicating variants, passaging of VEEV/Cm1/GFP was performed by infecting Vero cells every 24 to 48 h, using decreasing volumes of samples harvested at the previous passages. We used 10-fold-lower volume of samples at every next passage. Interestingly, virus evolution continued and, after five more passages, we detected the appearance of new variants demonstrating replication levels higher than those of the originally applied VEEV/Cm1/GFP. Moreover, the selected variants became cytopathic for both BHK-21 and Vero cells. Sequencing of the genomes of a few plaque-purified mutants revealed the presence of additional mutations summarized in Fig. 2A. Surprisingly, the new adaptive mutations were detected in the nsP2-coding sequence, but not in the CP, where it was more logical to expect them. Most of the mutations were found in the amino terminus of VEEV nsP2, at positions 3, 5, and 17 (Fig. 2A and B). The T5A, T5I, and G17V mutations were located downstream of the amino-terminal peptide required for the nsP2 protease-mediated cleavage of the P123 polyprotein precursor (1). Other selected variants contained mutations Q471L and N545D in the protease domain of VEEV nsP2 (Fig. 2A and B). Notably, the nsP2 amino terminus has been previously proposed as a protease cofactor (27, 28).

Fig 2.

Adaptive mutations in VEEV nsP2 protein increase the release of infectious virions from cells containing replicating viral genomes encoding mutated CP. (A) Mutations identified in cytopathic, plaque-purified variants of passaged VEEV/Cm1/GFP. (B) Schematic representation of the VEEV nsP2 domain structure and positions of the identified mutations. (C and D) The schematic representation of recombinant VEEV/Cm1/GFP variant genomes with the indicated adaptive mutations in nsP2 and CP and their replication rates. BHK-21 cells were electroporated with 4 μg of in vitro-synthesized RNAs of the indicated variants. Media were replaced at the indicated time points, and titers of infectious viral particles were determined as described in Materials and Methods.

Only one of the plaque-purified viruses contained in conjunction with the nsP2-specific mutation, an additional mutation in the CP gene, L41S (Fig. 1A and 2A). Like the above-described N47Y, the L41S substitution was also located in helix I, and it was reasonable to expect that it would produce an additional positive impact on the infectious virus titers. However, this was not the case. Cloning of this CP-specific mutation into VEEV/Cm1/GFP did not cause a significant increase in infectious virus titers (Fig. 2C and D) or CPE development. Replication rates of the VEEV/Cm1(L41S)/GFP variant were essentially the same as those of the original VEEV/Cm1/GFP. In contrast, introduction of the T5A mutation into nsP2 of VEEV/Cm1/GFP [VEEV(nsP2/T5A)/Cm1/GFP variant] or both nsP2- and CP-specific mutations [VEEV(nsP2/T5A)/Cm1(L41S)/GFP variant] caused a 10-fold increase in virus titers (Fig. 2D), suggesting that the amino terminus of nsP2 has an additional, previously undetermined function in virus replication and/or virion formation.

Adaptive, nsP2-specific mutations increase virus cytopathogenicity.

The selected mutants were characterized not only by their replication to higher infectious titers but also by their ability to induce CPE, at least in Vero and BHK-21 cells. In contrast to the originally designed VEEV/Cm/GFP or the previously described VEEV/GFP/C1 virus (14) (the latter virus had mutations only in the CP-specific NLS and the peptide connecting helix I and the NLS), the selected variants were able to form plaques in Vero cells. Cloning of the identified mutations into VEEV/Cm1/GFP demonstrated conclusively that the mutations found in nsP2 were responsible for the increase in cytopathogenicity. The T5A, G17V, V3A, T5I, Q471L, and N545D mutation-containing viruses readily formed distinct plaques (Fig. 3) and developed a spreading infection in Vero cells, which resulted in complete destruction of cell monolayers. However, for all of the designed mutants, despite complete CPE, the infectious titers remained at the level of 2 × 10⁷ PFU/ml.

Fig 3.

Adaptive mutations in nsP2 make VEEV variants encoding mutated CP more cytopathic. The schematic representation of VEEV genomes encoding wt and mutated nsP2 and CP. Mutant genomes were packaged into infectious virions as described in Materials and Methods, and these stocks were titrated on Vero cells. Plaques were stained with crystal violet after 3 days of incubation at 37°C.

In the next experiments, we tested whether CPE development and higher titers of the released viruses required the originally identified CP mutation N47Y, which was used for construction of VEEV/Cm1/GFP. The nsP2-specific T5A and Q471L mutations were cloned into genome of the originally designed VEEV/Cm/GFP, whose CP gene contained no adaptive mutations. To achieve higher titers required for cell infection in the single step growth curve analysis, all of the mutant genomes were packaged into infectious virions using wt CP-encoding helper RNAs. The in vitro-synthesized genomic RNAs of the mutants and helper construct encoding RNA packaging-competent capsid protein were coelectroporated into the cells (see Materials and Methods for details), and stocks were harvested at 24 h posttransfection. The applied helper genome contained no packaging signal and, therefore, was packaged with at least 4 orders of magnitude lower efficiency and thus did not have a detectable effect on the results of the experiments described below. Vero cells were infected at the same MOI, and the released viruses were harvested at the indicated time points (Fig. 4). VEEV(nsP2/T5A)/Cm/GFP and VEEV(nsP2/Q471L)/Cm/GFP mutants were viable and replicated to higher titers than the parental, originally designed VEEV/Cm/GFP (Fig. 4). They also were cytopathic and induced plaque formation in Vero cells (data not shown). However, their replication rates and final titers remained lower than those of the variant containing the adaptive mutations in both capsid-specific helix I and nsP2 (VEEV(nsP2/T5A)/Cm1/GFP). Thus, the nsP2-specific mutations functioned synergistically or additively with the CP-specific adaptive mutations and increased both the infectious titers of the released viruses and made them capable of causing CPE.

In contrast to infectious wt VEEV TC-83, replication of VEEV replicons lacking viral structural genes is dramatically less cytopathic, and upon delivery into the cells defective in type I IFN synthesis or signaling, they readily establish persistent replication (15). Therefore, to additionally evaluate the effect of accumulating nsP2-specific mutations on cytopathogenicity of self-replicating VEEV-specific RNAs, we introduced one of the above-described mutations into the nsP2 gene of VEEV replicon (Fig. 5). The wt VEEV replicon, VEErep/GFP/Pac, was based on the VEEV TC-83 genome and encoded a Pac gene under the control of a second subgenomic promoter. The second subgenomic promoter drove GFP expression, which was used to measure the efficiency of electroporation. The VEErep(nsP2/T5A)/GFP/Pac had exactly the same design but contained a T5A mutation in the nsP2 gene. Equal amounts of the in vitro-synthesized RNAs were electroporated into BHK-21 cells, and puromycin selection was applied as described in Materials and Methods. At 7 days posttransfection, numbers of colonies of Pur^r cells were evaluated: VEErep(nsP2/T5A)/GFP/Pac caused a profound CPE and was almost 2 orders of magnitude less efficient in establishing persistent replication and colony formation (Fig. 5). The data presented in the following section demonstrate that T5A mutation increases the level of replication of virus-specific RNAs and expression of the proteins encoded by the subgenomic RNA. Therefore, the reduced efficiency of colony formation cannot be explained by the decrease in Pac expression. Based on our extensive previous experience, the incomplete death of all of the replicon-containing cells and formation of the low numbers of colonies of Pur^r cells was always a result of accumulation of adaptive mutations in the replicons' nsPs (29–31). This was most likely the case in the present study. Sequencing of the replicons' genomes was not performed, because identification of the mutations was unlikely to be useful to the progress of this particular study.

Fig 5.

The nsP2-specific adaptive mutations identified make VEEV TC-83-based replicons more cytopathic. (A) Schematic representation of VEEV replicons encoding either wt or mutated nsP2 and containing GFP and Pac genes under the control of the subgenomic promoters. Equal amounts of the in vitro-synthesized RNAs were electroporated into BHK-21 cells, and the indicated colony-forming efficiencies were measured as described in Materials and Methods. (B) Representative dishes, in which equal numbers of electroporated cells were seeded. Colonies of replicon-containing cells were stained with crystal violet after 6 days of incubation at 37°C in the presence of puromycin at a concentration of 5 μg/ml.

nsP2-specific mutations increase synthesis of virus-specific RNAs.

The observed higher cytopathogenicity of the developed mutants needed an explanation. Therefore, we characterized the various mutant viruses in terms of virus-specific RNA and protein synthesis. Such experiments require infection of all of the cells in the monolayers at the same MOI. Therefore, to achieve higher titers of stocks for infection, particularly of the poorly replicating VEEV/Cm/GFP and VEEV/Cm1/GFP, all of the in vitro-synthesized mutant genomes were packaged into infectious viral particles (see Materials and Methods for details). For all of the constructs, titers of packaged viral genomes were always >10⁹ inf.u/ml.

Vero cells were infected at an identical MOI of 20 inf.u/cell with the particles containing either the originally designed genomes with mutated CP, or those having additional adaptive mutations in nsP2 and/or CP (Fig. 6A). Viral structural proteins and virus-specific RNAs were metabolically labeled with [³⁵S]methionine and [³H]uridine, respectively (Fig. 6B and C), and analyzed as described in Materials and Methods. As expected, the mutations found in CP, N47Y and L41S, had no noticeable effect either on RNA or protein synthesis. However, the nsP2-specific mutations had a readily detectable, 3- to 4-fold, stimulatory effect on both genome replication and transcription of the subgenomic RNAs (Fig. 6B). This higher level of replication appeared to result in an increased level of synthesis of viral structural proteins and GFP, detected in protein synthesis analysis (Fig. 6C). Thus, the adaptive mutations demonstrated distinct roles in virus evolution to higher titers and cytopathic phenotype: the CP-specific mutations likely increased the efficiency of genome packaging, and the nsP2-specific mutations led to more efficient synthesis of viral genomes and structural proteins. The mutations in nsP2 correlated with the appearance of the cytopathic phenotype of these constructs, thus, indirectly indicating that the mutations caused major changes in virus-host cell interactions, including the cells' ability to respond to viral replication. They noticeably decreased the ability of VEEV, encoding the defective CP, to induce type I IFN production (Fig. 7).

Fig 6.

Adaptive mutations in nsP2 increase synthesis of virus-specific RNAs and structural proteins. (A) The schematic representation of VEEV genomes containing adaptive mutations in CP and nsP2. (B) The indicated genomes were packaged to high titers into infectious viral particles as described in Materials and Methods, and Vero cells were infected at an MOI of 20 inf.u/cell. Viral RNAs were metabolically labeled with [³H]uridine between 3 and 7 h postinfection and analyzed as described in Materials and Methods. (C) Vero cells were infected as described above, and proteins were metabolically labeled with [³⁵S]methionine at 6 h postinfection and analyzed as described in Materials and Methods. This experiment was repeated twice with very similar results.

Fig 7.

Adaptive mutations in nsP2, but not in CP, have negative effects on IFN-β induction. NIH 3T3 cells were infected with the indicated viruses as described in the Fig. 6 legend (see also in Fig. 6A the schematic representation of their genomes). Media were harvested at 24 h postinfection, and concentrations of IFN-β were measured as described in Materials and Methods. The experiment was repeated twice with very similar results.

Accumulation of the adaptive mutations in the main protease domain of nsP2 and in its amino terminus, which was previously proposed as a protease cofactor, suggested that they may interfere with the nsP2 polyprotein cleavage. However, in both an in vitro translation system and in the infected cells, we were able to detect only a very small decrease in the rates of polyprotein processing (data not shown), which we considered inconclusive. Analysis of nsP2 distribution in the infected cells also failed to detect any changes. In contrast to the Old World alphaviruses, VEEV-specific nsP2 is present in the cytoplasm of the infected cells and demonstrates no nuclear distribution (13). We observed that after attaining adaptive mutations, VEEV nsP2 remained in the cytoplasm and did not relocalize to the nucleus.

nsP2 in VEEV capsid mutants continues to evolve.

The above-described data demonstrated that the increase in infectious titers of VEEV variants with mutated CP strongly correlated with development of the ability of viruses to induce CPE. However, there was an additional detectable change in virus phenotype. The distinguishing feature of the originally designed VEEV/Cm/GFP and then selected VEEV/Cm1/GFP was their ability to release SVPs, the genome-free subviral particles, as efficiently as wt VEEV TC-83 releases infectious virions. It was noted that a number of the originally selected nsP2 mutants demonstrated a 2- to 3-fold decrease in SVP release compared to VEEV/Cm/GFP (data not shown). Therefore, the next round of the experiments was aimed at further testing (i) whether downregulation of the SVP release was a real, nsP2-dependent effect, and (ii) whether the mutant phenotype described above was an endpoint to the in vitro evolution and adaptation of VEEV mutants to replication in cell culture. Given that after the first round of selection, VEEV CP mutants became cytopathic, further selection was performed using viruses no longer expressing GFP [VEEV(nsP2/T5A)/Cm1(L41S)].

Based on the data presented above, the most anticipated result was either further increase in RNA replication, or CP evolution to more efficient RNA packaging. After five passages of the designed VEEV mutant VEEV(T5A)/Cm1(L41S) in Vero cells, we indeed detected appearance of more cytopathic variants and their phenotype correlated with new mutations in the nsP2 protease domain. In three randomly selected plaques, it was found that in addition to the originally introduced nsP2-specific T5A and CP-specific L41S and N47Y mutations, new amino acid substitutions were present. One of the plaque-purified variants contained K480T, the second A474V, and the third T13A, T589K, and L789H mutations in nsP2. Cloning of these mutations back into the viral genome (Fig. 8A) did not have a strong positive impact on virus replication (Fig. 8C) but caused an increase in plaque size Fig. 8B) and an additional boost in RNA replication (Fig. 8D). However, at the same time, to our surprise, these additional mutations in nsP2 reduced synthesis of viral structural proteins 3- to 5-fold (Fig. 8E). They also caused a profound decrease in the SVP release (Fig. 8F) without affecting the infectious titers. Thus, the selected nsP2 mutants became less efficient in production of genome-free virions, SVPs, which provide no benefit for infection spread.

Fig 8.

Further virus evolution leads to a decrease in SVP production. (A) Schematic representation of VEEV genomes containing adaptive mutations in CP and nsP2. (B) Mutant genomes were packaged into infectious virions as described in Materials and Methods, and these stocks were titrated on Vero cells. Plaques were stained with crystal violet after 3 days of incubation at 37°C. (C) Vero cells in six-well Costar plates were infected with the indicated viruses at an MOI of 20 inf.u/ml. At the indicated times, media were replaced, and infectious titers were determined by plaque assay on Vero cells. Titers of VEEV/Cm were determined by immunostaining, using mouse antibodies, specific to VEEV structural proteins. (D) Vero cells were infected with the indicated mutants at an MOI of 20 inf.u/cell. Viral RNAs were metabolically labeled with [³H]uridine between 3 and 7 h postinfection and analyzed as described in Materials and Methods. (E) Vero cells were infected as described above, and proteins were metabolically labeled with [³⁵S]methionine at 6 h postinfection and analyzed as described in Materials and Methods. (F) Vero cells were infected with the indicated viruses as described above. At 6 h postinfection, media were replaced by serum-free VP-SFM media, and the released particles were harvested at 20 h postinfection. They were pelleted by ultracentrifugation and samples corresponding to 1 ml of media were analyzed by SDS–10%PAGE, followed by Western blotting with VEEV-specific antibodies. Quantitative analysis was performed on a LI-COR imager. The experiments were repeated three times with reproducible results. This figure represents one of the experiments.

An additional test for assessment of the nsP2-associated increase of cytopathogenicity was made by cloning of the identified mutations into VEEV replicon, which encodes GFP and Pac under the control of the subgenomic promoters. BHK-21 cells were transfected with equal amounts of the in vitro-synthesized RNAs, and we evaluated the efficiencies of Pur^r cell colony formation of the replicons, containing the newly found mutations, with that of the replicon having a T5A mutation in nsP2 (see Materials and Methods for details). Despite equal transfection efficiency and comparable levels of GFP expression, the newly designed constructs produced 4- to 5-fold-fewer Pur^r cell colonies than the above-described VEErep(nsP2/T5A)/GFP/Pac (data not shown), which was already more cytopathic than the construct encoding wt nsP2 (Fig. 5). This was an additional indication that nsP2-specific mutations can regulate VEEV cytopathogenicity.

DISCUSSION

Similarly to most RNA viruses, the alphavirus genome encodes only a few proteins. It was generally believed that their functions are reasonably well separated: nonstructural proteins mediate viral RNA synthesis, and structural proteins package viral genome and form infectious viral particles. However, the most recent studies started to demonstrate that current hypotheses about alphavirus proteins functions appear to be oversimplified, and at least some of the viral nonstructural and structural proteins have additional activities in virus replication and virus-host interactions (11, 13, 24, 32–34). The best examples are the CPs of the New World alphaviruses and the nsP2 proteins of the Old World alphaviruses. The Old World alphavirus nsP2 proteins accumulate not only in the replication enzyme complexes, where they exhibit NTPase, helicase, and protease functions, but also in the cell nuclei, where they induce rapid degradation of the Rbp1 subunit of cellular DNA-dependent RNA polymerase II (32). The New World alphavirus CPs interact in the infected cells with importin-α/β and CRM1, and this tetrameric complex interferes with the nuclear pore function, blocks the nuclear cytoplasmic traffic, and thus induces transcriptional shutoff. A distinguishing feature of VEEV, the representative member of the New World alphaviruses, is the noncytopathic phenotype of its replicon, which encodes no structural proteins (15). Thus, in contrast to the Old World alphaviruses (33, 34), neither VEEV RNA replication nor its nsP2 expression in the context of the ns polyprotein has a deleterious effect on cell viability.

In a previous study, we developed a VEEV TC-83 mutant, in which almost all of the positively charged amino acids in the RNA-binding CP domain were replaced with neutral amino acids. These 26 mutations had a deleterious effect on viral genomic RNA packaging and infectious titers. The high number and strategic placement of the introduced mutations in CP were also expected to make reversion to more efficient RNA packaging an impossible event. However, this was not the case. VEEV demonstrated an incredible ability for evolution and developed ways to increase infectious titers and produce a spreading infection. Within a few passages, these defective in RNA packaging mutants developed adaptive mutations and surprisingly, none of these found mutations caused an increase in the CP's positive charge. One of the mutations modified CP-specific helix I, which was previously proposed to be important for capsid protein dimerization during nucleocapsid assembly, but not for RNA packaging (19, 20). Further evolution was associated with accumulation of point mutations in the nonstructural protein nsP2, which has to date, not been proposed as a protein that might regulate viral genome packaging. Both CP- and nsP2-specific mutations independently increased the infectious titers, but utilized different mechanisms for achieving this goal. The helix I-specific mutation most likely directly enhanced RNA genome packaging into nucleocapsids. The nsP2-specific mutations, at least those initially identified, demonstrated a strong stimulatory effect on both viral RNA synthesis and production of VEEV-specific structural proteins. Initially, the accumulation to higher concentrations of both components of the infectious virions (RNA genome and structural proteins) looked as a plausible explanation for the infectious titer increase.

However, the next experiments with the mutants suggested other possibilities. During further passaging, VEEV nsP2 continued to accumulate adaptive mutations, but further evolution of nsP2 was no longer associated with increase in structural protein production. To the contrary, VEEV structural proteins were translated less efficiently, and these variants demonstrated a profound decrease in the release of genome-free SVPs (Fig. 8). Thus, the evolution of the mutant virus was likely driven toward accuracy and/or efficiency of viral genome packaging. In previous studies, numerous mutations in the nsP2 protein always produced a strong negative effect on alphavirus RNA replication (15, 30, 31). The present study presents the first example of mutations having a positive impact on the efficiency of viral RNA synthesis. The obvious question is, if the newly developed mutations increase virus replication, then why they were not selected during the course of natural VEEV evolution. Their effect on wt virus replication cannot be experimentally tested because of the biosafety concerns. Nevertheless, we can speculate that such mutations are not beneficial in the wt virus context. In combination with the naturally cytopathic VEEV capsid protein, they might make virus too cytopathic and induce CPE before optimal virus release. Alternatively, the identified mutations might decrease formation of viral particles by the wt capsid. The direct or indirect interactions of nsP2 and CP have been experimentally confirmed, at least for VEEV and SINV. CPs of both viruses are very efficiently coprecipitated with nsP2 from virus-infected cells (27), and both nsP2 and capsid proteins are associated with ribosomes in coimmunoprecipitation and polysome fractionation experiments (35–37).

To date, the efficiency of RNA packaging into viral particles is thought to be dictated by RNA packaging signals, which specifically interact with CP (38–41). The results displayed here raise the question: is this interaction the only regulatory mechanism in genome packaging? It seems likely that the answer is no. Alteration of packaging by mutations in either CP or packaging signal does not lead to encapsidation of cellular mRNA, many of which are present in the cells at high concentration. Thus, we cannot rule out the possibility that, similar to what has already been proposed for some other viruses (42, 43), only replicating, VEEV-specific RNA is competent for packaging. This proposition also implies that during the replication process, viral RNA must be properly presented for encapsidation, and nsP2 is one of the candidates, which might mediate this function. Other viruses with RNA-positive genomes have also demonstrated accumulation of mutations in the helicase/protease-encoding nonstructural proteins in response to profound CP modifications. One of the best examples is CSFV (44), in which compensatory mutations in the NS3-specific helicase domain adapted the virus to deletion of the entire capsid-coding gene and made it capable of RNA packaging. In the present study, the adaptive mutations in VEEV genome were found either in the main protease domain or in the amino-terminal fragment of nsP2, which is an essential cofactor of the nsP2 protease. However, at this point we cannot rule out the possibility that the identified adaptive mutations affect helicase rather than protease function. This uncertainty is due to a number of factors, not least of which is the lack of a crystal structure of the entire alphavirus nsP2. In addition, our data from another line of research demonstrate that without the protease domain, nsP2-associated helicase is inactive (data not shown). We were also unable to detect significant changes in nsP2-mediated polyprotein processing, which would more strongly suggest an affect on the protease domain. Moreover, in our previous studies, in response to modifications in the genome, which caused a profound decrease in expression of structural proteins, VEEV demonstrated accumulation of mutations in both nsP2 helicase and protease domains (26, 45). These adaptive mutations also increased the release of infectious virions without a significant change in CP and glycoprotein synthesis.

As we described above, another consequence of accumulation of mutations in nsP2 of VEEV variants encoding strongly mutated CP was an increase in virus cytopathogenicity. In agreement with the previously published data, the originally designed constructs, which had a wt sequence of nsPs, were noncytopathic and did not form plaques even after acquiring CP-specific mutations, which made virus capable of developing a spreading infection. Only the defined mutations in nsP2 had a strong positive impact on both virus and replicon's cytopathogenicity. None of the tested mutations noticeably affected polyprotein processing or nsP2 compartmentalization, but they obviously changed the protein's function in virus-host cell interactions. These data strongly suggest that our knowledge about this protein's roles in replication is incomplete. The results also imply that VEEV replication is a process characterized by the balance between replication and cytopathogenicity, and the development of a more cytopathic phenotype is not necessarily beneficial for virus replication.

Taken together, the results presented here demonstrate that the RNA-binding domain of VEEV CP and the RNA packaging signal are not the only determinants of RNA packaging into virions. Mutations in both the CP-specific helix I, which has not been previously proposed to have a direct role in the RNA binding, and in nsP2 protein strongly increase the infectious titers of the VEEV CP mutants. The process of RNA packaging into nucleocapsid is therefore as complex and multifaceted as all other aspects of alphavirus biology.