Enhancement of protein expression by alphavirus replicons ... · Enhancement of protein expression by alphavirus replicons by designing self-replicating subgenomic RNAs Dal Young

Enhancement of protein expression by alphavirusreplicons by designing self-replicatingsubgenomic RNAsDal Young Kima,1, Svetlana Atashevaa,1, Alexander J. McAuleyb, Jessica A. Plantec,2, Elena I. Frolovaa,David W. C. Beasleyb,d, and Ilya Frolova,3

aDepartment of Microbiology, University of Alabama at Birmingham, Birmingham, AL 35294; bDepartment of Microbiology and Immunology, cDepartment ofPathology, and dSealy Center for Vaccine Development and Institute for Human Infections and Immunity, University of Texas Medical Branch, Galveston,TX 77555

Edited by Diane E. Griffin, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, and approved June 13, 2014 (received for review May 10, 2014)

Since the development of infectious cDNA clones of viral RNAgenomes and the means of delivery of the in vitro-synthesizedRNA into cells, alphaviruses have become an attractive systemfor expression of heterologous genetic information. Alphavirusesreplicate exclusively in the cytoplasm, and their genetic materialcannot recombine with cellular DNA. Alphavirus genome-based,self-replicating RNAs (replicons) are widely used vectors for ex-pression of heterologous proteins. Their current design relies onreplacement of structural genes, encoded by subgenomic RNAs (SGRNA), with heterologous sequences of interest. The SG RNA istranscribed from a promoter located in the alphavirus-specific RNAreplication intermediate and is not further amplified. In this study,we have applied the accumulated knowledge of the mechanism ofalphavirus replication and promoter structures, in particular, toincrease the expression level of heterologous proteins from Ven-ezuelan equine encephalitis virus (VEEV)-based replicons. DuringVEEV infection, replication enzymes are produced in excess to RNAreplication intermediates, and a large fraction of them are notinvolved in RNA synthesis. The newly designed constructs encodeSG RNAs, which are not only transcribed from the SG promoter,but are additionally amplified by the previously underused VEEVreplication enzymes. These replicons produce SG RNAs andencoded proteins of interest 10- to 50-fold more efficiently thanthose using a traditional design. Amodified replicon encodingWestNile virus (WNV) premembrane and envelope proteins efficientlyproducedsubviral particles and, aftera single immunization, elicitedhigh titers of neutralizing antibodies, which protected mice fromlethal challenge with WNV.

expression vectors | vaccines

Alphaviruses are a group of enveloped viruses with a positive-strand RNA genome that replicate in most commonly used

cell lines to titers exceeding 1010 infectious units (inf.u)/mL (1,2). Upon infection, the genomic RNA serves as a template fortranslation of viral nonstructural proteins that form replicationcomplexes (3). Within a few hours postinfection, these com-plexes synthesize large amounts of viral genomic and sub-genomic (SG) RNA (3). The SG RNA is transcribed from theSG promoter and serves as a template for translation of viralstructural proteins: capsid, E2 and E1, which ultimately assemblewith genomic RNA into infectious viral particles. This highlyefficient virus-specific RNA and protein synthesis, coupled withthe availability of infectious cDNA clones, have made alphavi-ruses an attractive system for designing self-replicating vectorsfor delivery and expression of heterologous genetic information.The most widely used alphavirus-based expression systems arebased on replacement of viral structural genes by a gene(s) ofinterest (4). These modified viral genomes, termed replicons, canbe synthesized in vitro and delivered into cells either by trans-fection or in infectious viral particles, which deliver essentially

every packaged RNA molecule into the cells both in vivo andin vitro.In recent years, significant progress has been made in our

understanding of the mechanism of alphavirus replication. De-tailed studies have elucidated the structure and function of theRNA promoters, critical aspects of virus–host cell interactions,and the composition of the replication complexes (5–12). Thesemechanistic studies of alphavirus replication raised the questionof whether we are using their entire expression potential, andwhether the traditional replicon design can be further improvedto achieve higher levels of heterologous protein production.In this project, we sought to apply the latest advances inunderstanding of alphavirus RNA replication to design a newgeneration of Venezuelan equine encephalitis virus (VEEV)genome-based expression systems. The distinguishing feature ofthese constructs is the modification of the SG RNAs. These SGRNAs have been engineered to contain the cis-acting promoterelements, which are normally present at the 5′ end of the viralgenome and mediate genomic RNA replication (8, 13, 14). Thus,in these newly designed VEEV replicons (VEErep), the SGRNAs were not only transcribed from the SG promoter, but werecapable of replication/amplification by the VEEV replicationcomplexes. As a result, the heterologous gene expression wasmore efficient than that of the existing constructs, which use

Significance

One of the goals of modern molecular medicine is delivery andexpression of heterologous genes in living organisms. RNA-based delivery vectors are a safer choice than DNA vectors, butthey are prone to degradation and are highly dependent onefficient delivery methods. One of the ways to improve RNAvector performance is to increase the level of expression of theencoded proteins. We followed this approach and modifiedstandard alphavirus replicon-based expression systems tomake the transcribed subgenomic RNA additionally amplifiableby viral replication enzymes. Higher levels of subgenomic RNAsynthesis increased the replicons’ expression efficiency at least10-fold. Such replicons can be widely applied for developmentof efficient DNA and RNA vaccines and protein productionin vitro.

Author contributions: E.I.F., D.W.C.B., and I.F. designed research; D.Y.K., S.A., A.J.M., J.A.P.,and I.F. performed research; E.I.F., D.W.C.B., and I.F. analyzed data; and D.Y.K., S.A., E.I.F.,D.W.C.B., and I.F. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1D.Y.K. and S.A. contributed equally to this work.2Present address: Department of Epidemiology, University of North Carolina at ChapelHill, Chapel Hill, NC 27599.

3To whom correspondence should be addressed. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1408677111/-/DCSupplemental.

10708–10713 | PNAS | July 22, 2014 | vol. 111 | no. 29 www.pnas.org/cgi/doi/10.1073/pnas.1408677111

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

18, 2

020

replicons with the standard SG RNAs. The expression level ofheterologous protein encoded by the improved replicons wasalso found to be dependent on coexpression of VEEV capsidprotein. The VEEV replicons, which use both amplification ofthe SG RNA and express capsid protein, provide a platform fordevelopment of a variety of more efficient expression systemsand have numerous applications. To illustrate this, we have gen-erated a VEEV replicon encoding the premembrane and envelope(prM/E) proteins of West Nile virus (WNV). Particles containingthe newly designed replicons induced high levels of WNV E proteinexpression in vitro and elicited robust protective immunity in mice.

ResultsDesign of More Efficient VEEV-Based Expression System. Our pre-vious study demonstrated that during VEEV and Sindbis virusinfections only a small portion of viral nonstructural proteins(nsPs) is colocalized with dsRNA replication intermediates.Thus, it appears that a large fraction of nsPs are not involved inRNA replication (15–17). This provides an opportunity to ex-ploit the underused ns proteins for amplification of the SGRNAs encoding proteins of interest, which is normally tran-scribed from the SG promoter and is not further amplified (18).The accumulated experimental evidence has demonstrated

that replication/amplification of VEEV and other alphavirusgenomes and their defective interfering (DI) RNAs is de-termined by three promoter elements: (i) the conserved 3′-terminal sequence element (3′ CSE) and the following poly(A)tail; (ii) the 5′ UTR, which functions as a key promoter elementfor both negative- and positive-strand RNA synthesis; and (iii)the 51-nt conserved sequence element (51-nt CSE), which islocated in the nsP1-coding sequence and functions as an en-hancer of alphavirus genome replication (5–9). The SG RNAencodes the same 3′ CSE and poly(A) as the viral genome, butcontains a different 5′ UTR, and lacks the 51-nt CSE. To test thepotentially beneficial effects of the genomic RNA-specific pro-moter elements on SG RNA synthesis, the 5′ UTR of the VEEVgenome and the following fragment, containing the 51-nt CSE,were cloned under control of the SG promoter in the VEEVstrain TC-83–based replicon (Fig. 1 and Fig. S1). In this design,we preserved the natural sequence of the SG promoter and thefirst 2 nt of the subgenomic RNA, but entirely replaced theendogenous SG RNA 5′ UTR with that of the viral genome.Following the new 5′ UTR, we inserted the fragment of thensP1-coding sequence containing the VEEV-specific 51-nt CSE.The designed nucleotide sequence also had an nsP1-specific

initiating AUG (19), and the ORF was fused with the heterol-ogous genes, which we intended to express.To achieve the expression of the protein of interest without

additional amino terminal peptides, the remaining nsP1 fragmentand heterologous sequences of interest were separated by genes,whose products are capable of processing the polyprotein in cis.Such genes included ubiquitin (Ubi) (in VEErep/DI-Ubi-GFP),2A protease of foot-and-mouth-disease virus, FMDV 2A (inVEErep/DI-2A-GFP), and VEEV capsid protein (VEErep/DI-Cm-GFP). The encoded capsid protein sequence used for de-signing of the latter replicon contained no nuclear localizationsignal (NLS), which is responsible for its transcription inhibitoryfunctions, and thus, the expressed capsid protein was non-cytopathic (10, 19, 20). We also tested whether the encepha-lomyocarditis virus (EMCV) derived internal ribosome entrysite (IRES) could be used for heterologous gene expression. Itwas used in the VEErep/DI-IRES-GFP construct. To measurechanges in heterologous protein expression, in the initial experi-ments, we used GFP as a protein of interest in our various vectordesigns. The design of all of the constructs is presented in Fig. 2A.It should be noted that these SG RNAs lacked the packagingsignal required for RNA packaging into viral particles. Becausethe SG RNA strategy resembled that of the previously describedDI RNAs (21–23), in this and the following sections, such con-structs are referred to as DI replicons. The VEEV TC-83–basedVEErep/GFP replicon, having a traditional design (24) and en-coding GFP and unmodified SG 5′ UTR under control of theSG promoter, was used as a control to demonstrate a standardlevel of RNA synthesis and heterologous protein expression.

DI Replicons Demonstrate Higher Levels of Protein Expression.VEEVreplicons used in the study were packaged into viral particles usingthe previously designed packaging systems (25) and applied forinfecting cells at the same multiplicity of infection (MOI). In babyhamster kidney (BHK-21) cells, all of the constructs (Fig. 2A),except that using the EMCV IRES, were more efficient in GFPexpression than the standard VEErep/GFP, suggesting amplifica-tion of the originally transcribed SG RNA. Unlike VEErep/GFP,they expressed GFP protein to a level readily detectable on Coo-massie blue-stained gels as a major protein band, and FACS anal-ysis also demonstrated a strong increase in GFP fluorescence (Fig. 2B and C). The performance of FMDV-derived 2A protease was notas efficient as we expected, and only ∼50% of GFP was present ina cleaved and free form on the gels (Fig. 2C, see also Fig. 4). ThensP1–capsid and nsP1–GFP fusions were readily detectable on thegels. However, the expressed VEEV capsid protein mediated 100%

Fig. 1. The schematic representation of alphavirusvectors and protein expression strategies. (A) Thealphavirus genome. (B) Standard alphavirus repli-cons. Viral structural genes are replaced by heter-ologous gene of interest. (C) The newly designedVEEV DI replicons. The DI SG RNAs contain all of thepromoter elements, which are normally present onlyin the 5′ end of VEEV genomic RNA (Fig. S1 fordetails). Presence of these regulatory sequencesallows nsPs not only to transcribe the DI SG RNAs butalso to amplify these RNA as viral genomes.

Kim et al. PNAS | July 22, 2014 | vol. 111 | no. 29 | 10709

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

18, 2

020

processing, which did not depend on its presence either in nsP1fusion or in completely processed form. The lower GFP expressionlevel from the VEErep/DI-IRES-GFP was unexpected, particularlyconsidering the high level of SG RNA synthesis (Fig. 3 for details).

There were noticeable differences in the cytopathogenicityof the designed replicons in all of the cell types that were used.Replicons expressing mutated capsid protein were poorly cyto-pathic and demonstrated only a cell growth arrest, whereas theVEErep/DI-Ubi-GFP and VEErep/DI-IRES-GFP replicons in-duced very strong morphological changes of the cells within 12–16 h postinfection. For the VEErep/DI-Ubi-GFP replicon, theabnormally high intracellular concentration of Ubi most likelyhad a deleterious effect on cell biology.We also noticed that the level of heterologous protein ex-

pression mediated by the designed constructs was specific to thecell type. Most of the constructs produced less GFP in HEK293,Vero, and National Institutes of Health (NIH) 3T3 cells thanthey did in BHK-21 cells (Fig. S2). However, the VEErep/DI-Cm-GFP was always more productive than a standard VEErep/GFP, suggesting that in the previous studies, the entire potentialof VEEV replicons in terms of expression of heterologous pro-teins was not entirely exploited.

Higher Expression Level of Heterologous Genes Is Not Specific to GFP.To rule out the possibility that the increased level of proteinexpression is specific to GFP, we also designed DI repliconsexpressing a firefly luciferase (Fig. 2A and Fig. S3). The designedreplicons demonstrated an even more profound increase in het-erologous protein (Luc) expression. At 20 h postinfection, theDI replicons, but not the control VEErep/Luc/GFP, producedquantities of luciferase that were readily detectable on Coomassieblue-stained gels (Fig. S3). Accordingly, at 24 h postinfection, theactivity of luciferase in VEErep/GFP/DI-Cm-Luc–infected cellswas 50-fold higher than in cells infected with VEErep/GFP/Luc(Fig. S3).

DI Replicons Are Capable of Replication of Subgenomic RNA.Next weattempted to directly demonstrate that presence of RNA repli-cation promoter elements had a positive impact on subgenomicRNA synthesis. In all cases, the DI RNA-encoding repliconsproduced their SG RNAs at levels higher than those found in thecells infected with replicons with a standard SG RNA design(Figs. 2A and 3), (see control VEErep/GFP and VEErep/Luc/GFP). The stimulatory effect of the new 5′ promoter sequenceson DI SG RNA synthesis was especially evident in the case ofLuc-encoding constructs; the DI SG RNAs were synthesized 14-to 24-fold more efficiently than those transcribed from anotherSG promoter and having natural 5′UTR. Because both SG RNA

Fig. 2. VEEV DI replicons express heterologous proteins more efficientlythan those having standard design. (A) The schematic representation ofVEEV replicons encoding GFP or Luc genes under control of the subgenomicpromoters. (B) BHK-21 cells were infected with the indicated packagedreplicons at an MOI of 20 inf.u/cell, and GFP fluorescence was evaluated byFACS analysis at 20 h postinfection. (C) Cells were infected with indicatedreplicons at the same MOI, harvested at 20 h postinfection, lysed in loadingbuffer, and samples were analyzed by SDS/PAGE, followed by Coomassieblue staining. Cm, noncytopathic VEEV capsid protein; nsP1Δ, a peptide,whose coding sequence contains a 51-nt CSE.

Fig. 3. DI replicons demonstrate higher levels of SG RNA synthesis in theinfected cells. BHK-21 cells were infected with the indicated packagedreplicons at an MOI of 20 inf.u/cell. Replicon-specific RNAs were metaboli-cally labeled with [3H]uridine in the presence of ActD and analyzed byagarose gel electrophoresis (Materials and Methods for details). Panelsrepresent fragments of the same gel and the same X-ray film.

10710 | www.pnas.org/cgi/doi/10.1073/pnas.1408677111 Kim et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

18, 2

020

promoters were identical, the detected differences in SG RNAsynthesis were most likely the result of additional replication ofthe SG DI RNAs. Moreover, in the cells infected with capsid-expressing replicons VEErep/DI-Cm-GFP and VEErep/DI-Cm-Luc/GFP (Fig. 3), the [3H]-labeled replicon genomic and SGRNAs were found at higher concentrations.

Intracellular Accumulation of Heterologous Proteins Is Not ExclusivelyDetermined by Translation Efficiency. The final levels of heterolo-gous protein accumulation could be dependent not only on theefficiency of their synthesis, but also on the cytopathic effect(CPE)-associated changes in the intracellular environment, thedegradation rates of replicon RNAs, and individual proteins andother parameters. To directly compare the rates of GFP andLuc translation, we metabolically labeled cells infected withdifferent DI replicons with [35S]methionine at 7 h postinfection.The results presented in Fig. 4, demonstrate that GFP and Lucsyntheses were particularly efficient in case of Ubi- andFMDV 2A-dependent constructs. At 7 h postinfection, thesereplicons produced heterologous proteins at rates 30-fold higherthan standard VEErep/GFP and VEErep/GFP/Luc (Fig. 4). TheCm-dependent cassettes also synthesized 10- to-20-fold higherlevels of the protein of interest. As shown in Fig. 2 and Figs. S2and S3, the latter constructs were very productive in long-termexpression, suggesting that high levels of intracellular proteinaccumulation were determined both by more efficient synthesisand lower cytopathogenicity of the replicons. Thus, the rate ofCPE development appears to be an important parameter, whichneeds to be considered during optimization of alphavirus repli-cons for expression of particular proteins.At 7 h postinfection, all of the used constructs demonstrated

different abilities to interfere with cellular translation (Fig. 4).Moreover, expression of capsid protein made replicons incapableof overexpressing VEEV nonstructural proteins nsP1, nsP2, andnsP3. It is likely that in the presence of capsid protein, replicongenomes are efficiently packaged into nucleocapsids, makingRNA both resistant to degradation and no longer involved intranslation of nsPs. This correlates with the RNA replication

data (Fig. 3) and indirectly indicates that the increase in accu-mulation of VEErep/DI-Cm-GFP and VEErep/DI-Cm-Luc/GFPgenomes in the RNA labeling experiments was likely a result oftheir efficient encapsidation.

DI VEEV Replicons Are More Potent IFN-β Inducers than VEErep/GFP.Importantly, the designed DI replicons remained potent type IIFN inducers. It has been previously demonstrated that both lackof capsid protein expression in VEEV replicons (26) and in-activation of the capsid-specific NLS in capsid-coding sequenceof VEEV (19) made these self-replicating RNAs efficient inducersof antiviral responses. This proinflammatory characteristic allowsVEEV replicons to function as potent self-adjuvants. Replicationof the subgenomic RNA did not affect replicons’ ability to induceIFN-β. To the contrary, VEErep/DI-Cm-GFP reproducibly in-duced levels of IFN-β higher than those induced by VEErep/GFP(Fig. S4).

VEEV DI Replicons Can Be Used in Vivo. It was important to furthertest whether the newly developed replicons can be applied for invivo studies. We designed a replicon (Fig. 5A) that encodedWNV prM/E proteins in the context of self-replicating SG DIRNA (VEErep/DI-Cm-WNV). This replicon was packaged intoVEEV TC-83–derived structural proteins, and cells infected withthis replicon demonstrated accumulation of WNV E protein,some of which was released to the media in form of subviralparticles (SVPs) (Fig. 5B), which were readily detected by EM.Intracellular and extracellular E protein was also clearly visibleon Coomassie blue-stained gels (Fig. 5C). Efficient accumulationof E protein in the replicon-containing cells was visibly damagingfor the endoplasmic reticulum and Golgi networks and wascausing profound CPE within 12–16 h postinfection.An initial assessment of the safety and immunogenicity of the

VEErep/DI-Cm-WNV replicon was undertaken in 8- to 10-wk-old Swiss Webster mice. Groups of four female mice were in-oculated via the i.p. route with a single dose of 108 or 106 inf.u ofthe packaged VEErep/DI-Cm-WNV replicon, or 104 pfu of anattenuated WNV NY99ic L107F mutant (27). There were nosigns of illness in any animals infected at any dose with packagedreplicon. Serum samples collected ∼6 wk later showed high titersof neutralizing anti-WNV antibodies in all animals (Table S1).Fifty percent neutralization antibody titers in mice receiving 108

inf.u of packaged VEErep/DI-Cm-WNV ranged between 1,280and 5,120 and were similar to those in animals receiving the

Fig. 4. DI replicons demonstrate higher levels of heterologous proteinsynthesis, which correlates with the increase of SG RNA production. Infectedcells were metabolically labeled with [35S]methionine at 7 h postinfection(Materials and Methods for details). Fragments of the same gel are pre-sented. Radioactivity in the GFP- and Luc-containing bands was measured onStorm phosphorimager. The data were normalized to the radioactivity in theGFP and Luc bands detected in the VEErep/GFP and VEErep/Luc/GFP samples,respectively.

Fig. 5. VEErep/DI-Cm-WNV expresses high level of WNV-specific E protein,which is released from the cells in the SVP form. (A) The schematic repre-sentation of VEErep/DI-Cm-WNV replicon, encoding prM and E proteins ofWNV. (B) Fractions of the concentrated samples of the media, harvested at12 h postinfection, were used for negative staining and analyzed by EM. Barcorresponds to 50 nm. (C) SDS/PAGE analysis of WNV E protein present inthe cells or released in SVPs at 12 h postinfection. Gel was stained withCoomassie blue.

Kim et al. PNAS | July 22, 2014 | vol. 111 | no. 29 | 10711

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

18, 2

020

attenuated live virus, whereas titers for the 106 group weredetectably lower.Subsequently, groups of five or eight 3- to 4-wk-old mice were

inoculated i.p. with single doses of VEErep/DI-Cm-WNV rangingbetween 108 and 102 inf.u, or with PBS only. All mice were bled at19 d postimmunization and challenged at 22 d with a lethal doseof WNV strain NY99 (100 pfu, which is equivalent to ∼100 LD50).Dose-dependent differences in neutralizing antibody titers andsurvival rates were observed, with 100% seroconversion and sur-vival in the 108 and 106 groups (Fig. 6). Neutralizing antibody titerswere lower in those groups than had been observed in the initialexperiment, which could likely be explained by the shorter intervalpostimmunization for sample collection.Progressively fewer seroconvertors/survivors were seen at the

two lower doses of VEErep/DI-Cm-WNV replicon. However,despite only three of eight mice in the 104-dose group having de-tectable prechallenge neutralizing antibodies, an 85% survival ratewas observed. All but one of the animals that survived challengeshowed a more than twofold increase in antibody titer in thepostchallenge sample, although increases in titer were slightly lowerin the 108-dose group compared with the lower dose groups.

DiscussionIn recent years, the accumulated knowledge of the molecularmechanisms of alphavirus replication and virus–host interactionshas reached a level, which facilitates targeted manipulations oftheir genomes to achieve a programmed phenotype (28). Besidescausing infections of significant public health threat (29, 30),alphaviruses have been previously modified into RNA vectorsystems (4), which can deliver and express additional geneticinformation both in vivo and in vitro. To date, strategies used foralphavirus vector development have been very straightforward.Their structural genes in the SG RNA were either replaced bygenes of interest or the subgenomic promoters were duplicatedto control the expression of additional genes (4). Similarly toalphavirus genomes, replicon RNAs, which encode no structuralgenes, replicate exclusively in the cytoplasm and cannot intro-duce their genetic material into the cellular genome. They canbe used either for transient expression of proteins of interest orfor producing stable cell lines expressing heterologous proteinsfrom persistently replicating RNAs (24, 31). However, the mostimportant use of replicons appears to be their application fordevelopment of recombinant vaccines (32–36). Among thealphaviruses, VEEV replicons became more widely used forthese applications because of their higher levels of protein ex-pression, their abilities to serve as self-adjuvants (37, 38), andbecause they efficiently replicate in human cells.

Our data suggest that replacement of the structural genes inthe subgenomic RNA is not the end point of VEEV genome-based vector development. They can be additionally modified toachieve higher levels of protein expression. The coding strategyof VEEV replicon genomes can be changed in such a way thatthe transcribed SG RNAs function as templates for replicon-encoded viral replication complexes. Thus, the number of in-tracellular SG RNAs can be significantly increased and, conse-quently, the encoded proteins are synthesized more efficiently.This approach was found to be successful and the newly designedDI replicons were capable of synthesizing proteins of interestto 10- to 20-fold higher levels than standard VEEV replicons.Importantly, they remain efficient inducers of type I IFN and,probably, the proinflammatory cytokines. Single immunizationof mice with replicons expressing prM/E proteins of WNV wassufficient to induce readily detectable levels of neutralizingantibodies and protect mice against subsequent infection withhighly pathogenic WNV. VEErep/DI-Cm-WNV was able notonly to express high levels of intracellular WNV-specific pro-teins, but also to produce SVPs, which presumably contributedto the robust immunogenicity of this replicon. Taken together,the data demonstrated that these replicons retained all of theadvantages of previously designed VEEV-based expression sys-tems, but also became more productive in terms of heterologousprotein synthesis.Despite these significant improvements, there appears to be

possibilities for further increase of the efficacy of these newVEEV-based replicons. Unexpectedly, the VEEV capsid proteinwas found to be applicable not only as a self-protease, but also asan enhancer of replicon genome and SG RNA synthesis. First, itis possible that similarly to the capsid protein of rubella virus(39), it may function as a component of the replication complex.Second, it may package viral genomes and subgenomic RNAsinto nucleocapsids and thus increase their stability. The SGRNAs contain no packaging signal (40). Nevertheless, the pos-sibility of their partial encapsidation in the presence of capsidprotein cannot be ruled out. If this is the case, higher levels ofprotein expression from the designed replicon also raises thequestion as to whether RNA packaging into nucleocapsids isa reversible process and whether there is a dynamic balancebetween RNA packaging and release.In summary, this study demonstrated that VEEV-based vec-

tors are flexible in terms of introducing additional modificationsthat increase the expression levels of the encoded proteins.Depending on the cell line used for expression, it is possible touse different genes in the constructs, whose products mediateprocessing of the initially translated fusion protein and to modifythe replicon’s ability to induce CPE. We have tested only a lim-ited number of such genes. However, the modified, noncytopathicVEEV capsid protein used in this study appears to be reasonablyuniversal in promoting efficient cleavage in multiple cell linesand additional up-regulation of RNA replication. The enhancedsynthesis of SG RNAs and the encoded protein can be used toincrease the efficiency of VEEV replicon-based DNA vaccines,which are currently in development, and RNA vaccines in par-ticular. The low stability of RNA vaccines during their delivery invivo can be at least partially compensated by achieving higherlevels of heterologous protein expression.

Materials and MethodsPlasmid Constructs. The DI SG RNAs encoded 228 nt of the 5′ terminus of VEEVTC-83 genome followed by sequences of different proteases and heterolo-gous genes of interest. All of the constructs were designed using standardPCR-mediated techniques. To increase translation, all of the AUG codons inthe nsP1 sequence cassette were placed in-frame with the capsid gene bydeleting nucleotides 129 and 191 (19). WNV cassette encoding signal pep-tide, prM and E sequences of WNV NY99, was derived from the plasmiddescribed elsewhere (41). Standard recombinant DNA techniques were ap-plied for the construction of the plasmids. pVEErep/GFP and pVEErep/Luc/GFP plasmids were designed based on VEEV TC-83 replicons and have beendescribed elsewhere (13, 24). Helper genome constructs encoding VEEV TC-83capsid protein and glycoproteins have been described elsewhere (25).

Fig. 6. Neutralizing antibody titers (50% focus reduction) and protectionamong 3- to 4-wk-old Swiss Webster mice inoculated with a single dose ofVEErep/DI-Cm-WNV and challenged with 100 LD50 of WNV NY99. Animalswere bled at 19 d postimmunization and challenged at 22 d postimmunization.n/a, not applicable.

10712 | www.pnas.org/cgi/doi/10.1073/pnas.1408677111 Kim et al.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

18, 2

020

Analysis of Protein and RNA Synthesis. The 5 × 105 (BHK-21, NIH 3T3, Vero, orHEK293) cells in six-well Costar plates were infected with packaged replicongenomes at an MOI of 20 inf.u/mL. For analysis of long-term protein ex-pression, cells were harvested at 20 h postinfection and lysed in proteinloading buffer. Equal amounts of samples were analyzed by SDS/PAGE andthen stained with Coomassie blue. For FACS analysis, cells were trypsinizedat 20 h postinfection, fixed with 4% paraformaldehyde, and fluorescence wasmeasured by flow cytometry using LSR II (BD Biosciences). For analysis of therates of protein synthesis, BHK-21 cells were metabolically labeled with [35S]methionine at 7 h postinfection for 30 min in 0.8 mL of DMEM lackingmethionine, supplemented with 0.1% FBS and 20 μCi/mL of [35S]methionine.Equal amounts of proteins were loaded onto SDS/PAGE. After electropho-resis, gels were dried and autoradiographed. Quantitative analysis of ra-dioactivity in the specific bands was performed using a Storm phosphorimager.To analyze synthesis of replicon-specific RNAs, at 3 h postinfection, media inthe wells were replaced by 0.8 mL of αMEM supplemented with 10% FBS,actinomycin D (1 μg/mL), and [3H]uridine (20 μCi/mL). After 4 h of incubation at37 °C, total cellular RNAs were isolated with TRizol (Invitrogen) according tothe manufacturer’s protocol, then denatured with glyoxal in dimethyl sulfox-ide, and analyzed by agarose gel electrophoresis using the previously describedconditions (17). Gels were impregnated with 2,5-diphenyloxazole (PPO), dried,and autoradiographed.

WNV SVP Analysis. BHK-21 cells were infected with packaged VEErep/DI-Cm-WNV replicon at an MOI of 20 inf.u/cell. At 6 h postinfection cells werewashed and overlaid with serum-free medium (VP-SFM; Gibco). Media washarvested at 12 h postinfection, andparticleswere concentrated using centrifugalUltracel-100K filters (Millipore). Then they were either pelleted by ultracentri-fugation at 50,000 rpm for 1 h at 4 °C in a TLA-55 rotor using a TL-100 tabletopultracentrifuge (Beckman) for further analysis by 10% SDS/PAGE or used for EManalysis (42). Droplets of virus suspension were placed onto copper electronmicroscope grids and stained with 1% uranyl acetate. Images were acquired atmagnification of 50,000× using a FEI Tecnai F20 electron microscope in theUniversity of Alabama at Birmingham’s cryo-EM core facility.

ACKNOWLEDGMENTS. The authors thank Maricela Ramos for technicalassistance and Dr. Niall Foy for critical reading of the manuscript. Stud-ies at the University of Alabama at Birmingham were supported byNational Institutes of Health (NIH) Grants R01AI070207, R01AI095449, andR01AI073301. Studies at the University of Texas Medical Branch (UTMB)were supported by funding from the Institute for Human Infections andImmunity (to D.W.C.B.). A.J.M. was supported by a predoctoral fellowshipfrom the Jeane B. Kempner scholar program at UTMB. J.A.P. was supportedby a predoctoral fellowship from the Biodefense Training Program, NIHGrant T32-AI060549.

1. Griffin DE (2001) Alphaviruses. Fields’ Virology, eds Knipe DM, Howley PM (Lippincott,Williams and Wilkins, New York), 4th Ed, pp 917–962.

2. Weaver SC, Frolov I (2005) Togaviruses. Virology, eds Mahy BWJ, Meulen VT (ASMPress, Salisbury, UK), Vol 2, pp 1010–1024.

3. Strauss EG, Strauss JH (1986) Structure and replication of the alphavirus genome. TheTogaviridae and Flaviviridae, The Viruses, eds Schlesinger S, Schlesinger MJ (Plenum,New York), pp 35–90.

4. Frolov I, et al. (1996) Alphavirus-based expression vectors: Strategies and applications.Proc Natl Acad Sci USA 93(21):11371–11377.

5. Frolov I, Hardy R, Rice CM (2001) Cis-acting RNA elements at the 5′ end of Sindbis virusgenome RNA regulate minus- and plus-strand RNA synthesis. RNA 7(11):1638–1651.

6. Fayzulin R, Frolov I (2004) Changes of the secondary structure of the 5′ end of theSindbis virus genome inhibit virus growth in mosquito cells and lead to accumulationof adaptive mutations. J Virol 78(10):4953–4964.

7. Gorchakov R, Hardy R, Rice CM, Frolov I (2004) Selection of functional 5′ cis-actingelements promoting efficient sindbis virus genome replication. J Virol 78(1):61–75.

8. Michel G, Petrakova O, Atasheva S, Frolov I (2007) Adaptation of Venezuelan equineencephalitis virus lacking 51-nt conserved sequence element to replication in mam-malian and mosquito cells. Virology 362(2):475–487.

9. Strauss JH, Kuhn RJ, Niesters HGM, Strauss EG (1990) Functions of the 5′-terminal and3′-terminal sequences of the Sindbis virus genome in replication. New Aspects ofPositive-Strand RNA Viruses, eds Brinton MA, Heinz FX (American Society of Micro-biology, Washington, DC), pp 61–66.

10. Atasheva S, Fish A, Fornerod M, Frolova EI (2010) Venezuelan equine encephalitisvirus capsid protein forms a tetrameric complex with CRM1 and importin alpha/betathat obstructs nuclear pore complex function. J Virol 84(9):4158–4171.

11. Garmashova N, et al. (2007) Analysis of Venezuelan equine encephalitis virus capsidprotein function in the inhibition of cellular transcription. J Virol 81(24):13552–13565.

12. Garmashova N, et al. (2007) The Old World and New World alphaviruses use differentvirus-specific proteins for induction of transcriptional shutoff. J Virol 81(5):2472–2484.

13. Kulasegaran-Shylini R, Atasheva S, Gorenstein DG, Frolov I (2009) Structural andfunctional elements of the promoter encoded by the 5′ untranslated region of theVenezuelan equine encephalitis virus genome. J Virol 83(17):8327–8339.

14. Kulasegaran-Shylini R, Thiviyanathan V, Gorenstein DG, Frolov I (2009) The 5’UTR-specific mutation in VEEV TC-83 genome has a strong effect on RNA replication andsubgenomic RNA synthesis, but not on translation of the encoded proteins. Virology387(1):211–221.

15. Gorchakov R, Garmashova N, Frolova E, Frolov I (2008) Different types of nsP3-con-taining protein complexes in Sindbis virus-infected cells. J Virol 82(20):10088–10101.

16. Frolova EI, Gorchakov R, Pereboeva L, Atasheva S, Frolov I (2010) Functional Sindbisvirus replicative complexes are formed at the plasmamembrane. J Virol 84(22):11679–11695.

17. Gorchakov R, et al. (2008) A new role for ns polyprotein cleavage in Sindbis virusreplication. J Virol 82(13):6218–6231.

18. Strauss JH, Strauss EG (1994) The alphaviruses: Gene expression, replication, andevolution. Microbiol Rev 58(3):491–562.

19. Atasheva S, et al. (2013) Pseudoinfectious Venezuelan equine encephalitis virus: Anew means of alphavirus attenuation. J Virol 87(4):2023–2035.

20. Atasheva S, Garmashova N, Frolov I, Frolova E (2008) Venezuelan equine encephalitisvirus capsid protein inhibits nuclear import in mammalian but not in mosquito cells.J Virol 82(8):4028–4041.

21. Levis R, Weiss BG, Tsiang M, Huang H, Schlesinger S (1986) Deletion mapping ofSindbis virus DI RNAs derived from cDNAs defines the sequences essential for repli-cation and packaging. Cell 44(1):137–145.

22. Lehtovaara P, Söderlund H, Keränen S, Pettersson RF, Kääriäinen L (1981) 18S de-fective interfering RNA of Semliki Forest virus contains a triplicated linear repeat. ProcNatl Acad Sci USA 78(9):5353–5357.

23. Lehtovaara P, Söderlund H, Keränen S, Pettersson RF, Kääriäinen L (1982) Extremeends of the genome are conserved and rearranged in the defective interfering RNAsof Semliki Forest virus. J Mol Biol 156(4):731–748.

24. Petrakova O, et al. (2005) Noncytopathic replication of Venezuelan equine enceph-alitis virus and eastern equine encephalitis virus replicons in Mammalian cells. J Virol79(12):7597–7608.

25. Volkova E, Gorchakov R, Frolov I (2006) The efficient packaging of Venezuelan equineencephalitis virus-specific RNAs into viral particles is determined by nsP1-3 synthesis.Virology 344(2):315–327.

26. Konopka JL, Thompson JM, Whitmore AC, Webb DL, Johnston RE (2009) Acute in-fection with venezuelan equine encephalitis virus replicon particles catalyzes a sys-temic antiviral state and protects from lethal virus challenge. J Virol 83(23):12432–12442.

27. Zhang S, et al. (2006) A mutation in the envelope protein fusion loop attenuatesmouse neuroinvasiveness of the NY99 strain of West Nile virus. Virology 353(1):35–40.

28. Kim DY, et al. (2011) Design of chimeric alphaviruses with a programmed, attenuated,cell type-restricted phenotype. J Virol 85(9):4363–4376.

29. Griffin DE (1986) Alphavirus pathogenesis and immunity. The Togaviridae and Flaviviridae,The Viruses, eds Schlesinger S, Schlesinger MJ (Plenum, New York), pp 209–250.

30. Weaver SC, Tesh RB, Shope RE (1998) Alphaviruses (VEE). Tropical Infectious Diseases:Principles, Pathogens and Practice, eds Guerrant RL, Krogstad DJ, Maguire JH,Walker DH, Weller PF (Churchill Livingstone, New York), Vol 2, pp 1281–1287.

31. Frolov I, et al. (1999) Selection of RNA replicons capable of persistent noncytopathicreplication in mammalian cells. J Virol 73(5):3854–3865.

32. Baric RS, et al. (2002) Expression and self-assembly of Norwalk virus capsid proteinfrom Venezuelan equine encephalitis virus replicons. J Virol 76(6):3023–3030.

33. Johnston RE, et al. (2005) Vaccination of macaques with SIV immunogens delivered byVenezuelan equine encephalitis virus replicon particle vectors followed by a mucosalchallenge with SIVsmE660. Vaccine 23(42):4969–4979.

34. Mok H, et al. (2007) Venezuelan equine encephalitis virus replicon particles encodingrespiratory syncytial virus surface glycoproteins induce protective mucosal responsesin mice and cotton rats. J Virol 81(24):13710–13722.

35. Thomas CE, et al. (2006) Vaccination of mice with gonococcal TbpB expressed in vivo fromVenezuelan equine encephalitis viral replicon particles. Infect Immun 74(3):1612–1620.

36. ZhuW, et al. (2005) Comparison of immune responses to gonococcal PorB delivered asouter membrane vesicles, recombinant protein, or Venezuelan equine encephalitisvirus replicon particles. Infect Immun 73(11):7558–7568.

37. Thompson JM, et al. (2006) Mucosal and systemic adjuvant activity of alphavirus re-plicon particles. Proc Natl Acad Sci USA 103(10):3722–3727.

38. Thompson JM, Whitmore AC, Staats HF, Johnston RE (2008) Alphavirus repliconparticles acting as adjuvants promote CD8+ T cell responses to co-delivered antigen.Vaccine 26(33):4267–4275.

39. Tzeng WP, Matthews JD, Frey TK (2006) Analysis of rubella virus capsid protein-mediated enhancement of replicon replication and mutant rescue. J Virol 80(8):3966–3974.

40. Kim DY, Firth AE, Atasheva S, Frolova EI, Frolov I (2011) Conservation of a packagingsignal and the viral genome RNA packaging mechanism in alphavirus evolution.J Virol 85(16):8022–8036.

41. Mason PW, Shustov AV, Frolov I (2006) Production and characterization of vaccinesbased on flaviviruses defective in replication. Virology 351(2):432–443.

42. Lulla V, Kim DY, Frolova EI, Frolov I (2013) The amino-terminal domain of alphaviruscapsid protein is dispensable for viral particle assembly but regulates RNA encapsi-dation through cooperative functions of its subdomains. J Virol 87(22):12003–12019.

Kim et al. PNAS | July 22, 2014 | vol. 111 | no. 29 | 10713

MICRO

BIOLO

GY

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

18, 2

020