|Home | About | Journals | Submit | Contact Us | Français|
Superinfection exclusion is the ability of an established viral infection to interfere with a second viral infection. Using West Nile virus (WNV) as a model, we show that replicating replicons in BHK-21 cells suppress subsequent WNV infection. The WNV replicon also suppresses superinfections of other flaviviruses but not nonflaviviruses. Mode-of-action analysis indicates that the exclusion of WNV superinfection occurs at the step of RNA synthesis. The continuous culturing of WNV in the replicon-containing cells generated variants that could overcome the superinfection exclusion. The sequencing of the selected viruses revealed mutations in structural (prM S90R or envelope E138K) and nonstructural genes (NS4a K124R and peptide 2K V9M). Mutagenesis analysis showed that the mutations in structural genes nonselectively enhance viral infection in both naïve and replicon-containing BHK-21 cells; in contrast, the mutations in nonstructural genes more selectively enhance viral replication in the replicon-containing cells than in the naïve cells. Mechanistic analysis showed that the envelope mutation functions through the enhancement of virion attachment to BHK-21 cells, whereas the 2K mutation (and, to a lesser extent, the NS4a mutation) functions through the enhancement of viral RNA synthesis. Furthermore, we show that WNV superinfection exclusion is reversible by the treatment of the replicon cells with a flavivirus inhibitor. The preestablished replication of the replicon could be suppressed by infecting the cells with the 2K mutant WNV but not with the wild-type virus. These results suggest that WNV superinfection exclusion is a result of competition for intracellular host factors that are required for viral RNA synthesis.
West Nile virus (WNV) is a member of the genus Flavivirus in the family Flaviviridae. Since the outbreak of WNV in New York City in 1999, the virus has caused thousands of human infections in the United States, representing the largest meningoencephalitis outbreak in the Western Hemisphere and the biggest WNV outbreak ever reported (14). Besides WNV, many other flaviviruses are carried by arthropods and cause significant diseases in humans. These emerging and reemerging flaviviruses include the four serotypes of dengue virus (DENV), yellow fever virus (YFV), Japanese encephalitis virus (JEV), St. Louis encephalitis virus (SLEV), and tick-borne encephalitis virus. DENV alone poses a risk to 2.5 billion people worldwide and causes 50 to 100 million human infections each year (9). Human vaccines currently are available only for YFV, JEV, and tick-borne encephalitis virus. No effective antiviral therapy is available for the clinical treatment of flavivirus infections. Understanding the molecular mechanism of flavivirus replication is critical for the development of novel vaccine and antiviral therapy.
The flavivirus genome is a single-strand, plus-sense RNA of approximately 11 kb in length with a 5′ type I cap structure. Flaviviruses enter susceptible cells through a receptor-mediated endocytosis. After uncoating, the RNA genome serves as an mRNA for translation as well as a template for the transcription of the complementary minus-strand RNA, which in turn functions as a template for the synthesis of nascent genomic RNA. Flavivirus replication and assembly occur in the endoplasmic reticulum (ER). The flavivirus genome consists of a 5′ untranslated region (UTR), a long open reading frame, and a 3′ UTR (2). The single open reading frame encodes a polyprotein that is processed by viral and cellular proteases into 10 mature proteins. Three structural proteins (capsid [C], premembrane [prM] or membrane [M], and envelope [Env]) are required for viral particle formation. Seven nonstructural proteins (NS1, NS2a, NS2b, NS3, NS4a, NS4b, and NS5) are involved primarily in viral RNA replication (2); the nonstructural proteins also play roles in virion assembly (15, 18, 21, 30) and the evasion of host innate immune responses (1, 10, 20, 27, 28).
Superinfection exclusion, also known as homologous interference, is the phenomenon that a cell infected with one type of virus or transfected with a viral replicon becomes resistant to a secondary infection with the same virus, whereas infection with unrelated viruses normally is unaffected (35, 43). Superinfection exclusion has been observed among a broad range of viruses, including vaccinia virus (4), human immunodeficiency virus (25, 29), vesicular stomatitis virus (VSV) (40, 44), Borna disease virus (8), measles virus (24, 36), Sindbis virus (12), Semliki Forest virus (SFV) (41), rubella virus (5), hepatitis C virus (HCV) (35, 43), and bovine viral diarrhea virus (BVDV) (17). The exclusion could result from impairments of distinct steps of the viral life cycle during superinfection, including entry, translation, RNA replication, or virus budding (7). The mechanism of flavivirus superinfection exclusion currently is not understood.
In this study, we report that BHK-21 cells harboring the WNV replicon are 10- to 100-fold less efficient at supporting WNV infection than naïve BHK-21 cells. Such interference was also observed when the WNV replicon cells were incubated with other flaviviruses but not when the cells were incubated with nonflaviviruses. We demonstrated that WNV superinfection was blocked at the step of viral RNA synthesis, and the exclusion could be reversed by the treatment of the replicon cells with an antiviral compound. The continuous culturing of WNV in the replicon cells produced mutant viruses that can overcome the exclusion. The analysis of the mutant viruses uncovered two types of mutations: type one mutations occurred in prM (S90R) and Env (E138K) proteins that enhanced virion attachment to BHK-21 cells; type two mutations occurred in NS4a (K124R) and peptide 2K (V9M), which selectively enhanced viral RNA synthesis in the replicon-containing cells. Furthermore, mutant viruses carrying the type two mutations could suppress the preestablished replication of the replicon. Collectively, the results suggest that WNV superinfection exclusion is caused by a limitation of host factors that are critical for viral RNA replication.
Baby hamster kidney cells (BHK-21) and African green monkey kidney cells (Vero) were cultured in Dulbecco's modified Eagle medium (DMEM) with 10% fetal bovine serum (FBS) in 5% CO2 at 37°C. A BHK-21 cell line containing persistently replicating WNV replicons with a neomycin phosphotransferase gene (NeoRep; see Fig. Fig.1A)1A) was established previously (38) and was cultured in DMEM with 10% FBS plus 1 mg/ml of Geneticin (G418). Another BHK-21 cell line containing WNV replicon (RlucNeoRep) with dual reporters, a Renilla luciferase (Rluc) and a neo gene, were reported previously (22). WNV was prepared from a full-length infectious cDNA clone of epidemic New York strain 3356 (39). SLEV (strain Kern184.108.40.206), DENV-2 (New Guinea C strain), YFV (17D vaccine strain), Western equine encephalitis virus (WEEV; strain Cova 746), and VSV (New Jersey serotype) also were used in this study.
The WNV genome-length cDNA clones with specific mutations were constructed by using modified pFLWNV and two shuttle vectors (45). Shuttle vector A was constructed by engineering the BamHI-SphI fragment from pFLWNV (representing the upstream end of the T7 promoter [for the RNA transcription of genome-length RNA] to nucleotide position 3,627 of the WNV genome; GenBank no. AF404756) into the pACYC177 vector containing a modified cloning cassette. A QuikChange II XL site-directed mutagenesis kit (Stratagene) was used to engineer mutations in the prM and Env genes into shuttle vector A. The mutated DNA fragments were pasted back into the pFLWNV clone at the BamHI and StuI sites (nucleotide position 2,591). Shuttle vector B was constructed by engineering a KpnI-XbaI fragment (representing nucleotide position 5,341 through the 3′ end of the genome) into a pcDNA3.1(+) vector. The NS4a and 2K mutations were engineered into shuttle vector B using the QuikChange II XL site-directed mutagenesis kit. The mutated DNA fragments were pasted back into the pFLWNV clone and into the WNV replicon cDNA plasmid at the BsiWI and SpeI sites (nucleotide positions 5,780 and 8,022, respectively). For virus-like particle (VLP) packaging experiments, the complete C-prM-Env fragments (designated CprMEnv) containing the prM or Env mutation were PCR amplified using the above-described mutant genome-length cDNA clone as the templates; the resulting fragments then were cloned into an alphavirus SFV expression vector (33) through a unique BamHI site, resulting in the SFV-CprMEnv cDNA plasmid. All cDNA constructs were verified by DNA sequencing.
Both genome-length (39) and replicon (42) RNAs of WNV were in vitro transcribed from corresponding cDNA plasmids that were linearized with XbaI. A T7 mMESSAGE mMACHINE kit (Ambion) was used for RNA synthesis as described previously (39). The SFV-CprMEnv RNA was in vitro transcribed from a linear DNA (predigested with the SpeI) using an SP6 mMESSAGE mMACHINE kit (Ambion). The RNA transcripts were electroporated into BHK-21 cells as previously described (39). After the transfection of genome-length RNA, culture fluids were collected every 24 h until cytopathic effect was observed from days 3 to 5 posttransfection (p.t.). The culture medium containing viruses was aliquoted and stored at −80°C. The immunofluorescence assay (IFA), specific infectivity assay, and luciferase assay were performed as previously described (38, 39).
Naïve and NeoRep BHK-21 cells were seeded at 2 × 105 cells per well in 12-well plates and incubated at 37°C for 24 h. The cells were infected with WNV at a multiplicity of infection (MOI) of 10 at 4°C for 1 h, washed once with phosphate-buffered saline (PBS) and twice with DMEM, refed with DMEM, and returned to 4°C or shifted to 37°C. After incubation at 4 or 37°C for 1.5 h, one set of cells was rinsed with PBS and washed with a high-salt buffer (50 mM Na2CO3, pH 9.5, and 1 M NaCl) at room temperature for 3 min; the other set of cells was left untreated, followed by another PBS wash. Total cellular RNAs were extracted using an RNeasy kit (Qiagen) and were quantified by Nanodrop; 50 ng of extracted cellular RNA was quantified for viral RNA by real-time reverse transcription-PCR (RT-PCR), using a primer/probe set targeting viral Env region, as described below.
A transient WNV replicon assay was used to quantify viral translation and RNA replication (42). The replicon (Rluc2ARep; see Fig. Fig.3B)3B) contains a luciferase gene and a foot-and-mouth disease virus 2A protease in the position where the viral structural genes were deleted. The replicon RNA (10 μg) was electroporated into 8 × 106 BHK-21 and NeoRep BHK-21 cells (39). The transfected cells were resuspended in 25 ml of DMEM with 10% FBS, seeded into 12-well plates (1 ml/well), and assayed for luciferase activities at the indicated time points.
VLPs of WNV were prepared by the trans-supply of viral structural proteins to replicon RNAs (31). Briefly, 8 × 106 BHK-21 cells were electroporated with 10 μg of WNV Rluc2ARep RNA in a 0.4-cm cuvette with the GenePulser apparatus (Bio-Rad) using settings of 0.85 kV and 25 μF, with three pulsings at 3-s intervals. The transfected cells were resuspended in DMEM with 10% FBS and incubated at 37°C with 5% CO2 for 24 h. The cells were electroporated again with 10 μg of SFV-CprMEnv RNA at settings identical to those used for the first transfection. At 24 h after the second transfection, culture supernatants were centrifuged to remove cellular debris. The supernatants containing VLPs were aliquoted and stored at −80°C. The titers of VLPs (measured in focus-forming units [FFU]/ml) were estimated by the infection of Vero cells in a Lab-Tek Chamber Slide (Nalge Nunc International) with serial dilutions of the culture fluid, followed by the counting of IFA-positive cell foci at 18 h postinfection (p.i.). For IFA, immune mouse ascites fluid of WNV (American Type Culture Collection) and goat anti-mouse immunoglobulin G conjugated with Texas Red were used as primary and secondary antibodies, respectively. Alternatively, the VLPs were quantified by measuring replicon RNA copy numbers. The replicon RNA was extracted from 50 μl VLP and dissolved in 50 μl RNase-free water; 5 μl of resuspended RNA extract was quantified for replicon copy numbers by real-time RT-PCR using a primer/probe set targeting the viral NS5 region, as described below.
For VLP infection assays, a monolayer of naïve or NeoRep BHK-21 cells (4 × 104 cells per well in 96-well plate) was infected with approximately 1 FFU/cell of VLP. At the indicated time points, the cells were washed twice with cold PBS, lysed in 20 μl of 1× Renilla luciferase lysis buffer for 20 min, and assayed for luciferase activities using a Renilla luciferase assay kit (Promega). The luciferase signals were measured using a Veritas Microplate Luminometer (Turner BioSystems).
Naïve and NeoRep BHK-21 cells were seeded in a 12-well plate (2 × 105 cells per well) in the absence of G418. At 24 h postseeding, the cells were infected with WNV, SLEV, DENV-2, YFV, VSV, or WEEV at an MOI of 0.1. Culture medium was collected at the indicated time points and quantified for viral titers using a double-layer plaque assay on Vero cells, as described previously (33).
Because wild-type (WT) and mutant WNV did not generate plaques on NeoRep BHK-21 cells, we performed an immunostaining-based infectious center assay to visualize the infection and spread of the viruses. Briefly, the virus was serially diluted with BA-1 medium, and 100 μl of viral dilution was added to each well of 12-well plates containing confluent NeoRep BHK-21 cells (seeded 3 days earlier at 2 × 105 cells/well). After incubating the plates for 1 h at 37°C with intermittent manual shaking, an overlay of 0.5 ml MEM containing 5% FBS and 0.8% carboxymethylcellulose (Sigma) was added, and the plates were incubated further at 37°C for 2 days. After the carboxymethylcellulose layer was peeled off, the cells were washed once with PBS, fixed in acetone-methanol (1:1 volume/volume) for at least 20 min at −20°C, washed twice with PBS, and incubated with 250 μl of 1:100-diluted mouse anti-SLEV Env monoclonal antibody (cross-reactive to WNV Env protein; Mab8744; Chemicon, Millipore) for 1 h at room temperature. The cells then were washed three times with PBS, incubated with 250 μl of 1:200-diluted goat anti-mouse immunoglobulin G conjugated with horseradish peroxidase (GE Healthcare) for 1 h at room temperature, and washed three times with PBS, and color was developed by using a metal-enhanced DAB substrate kit (Thermo Scientific) according to the manufacturer's instructions.
Four independent lineages of superinfection-competent WNV were generated by the blind passaging of WNV (initially derived from an infectious cDNA clone) on NeoRep BHK-21 cells (see Fig. Fig.4A).4A). Each passage lasted for 42 to 48 h, with the first round of infection at an MOI of 0.1. All subsequent passages were performed by transferring 100 μl of culture medium from the previous passage to a T25 flask containing 90% confluent NeoRep cells. After the 10th passage, viruses from different passages were analyzed for their abilities to superinfect the NeoRep cells compared to that of the WT virus. The selections were terminated at passage 10, because viruses from the seventh passage and beyond did not further improve the superinfection. All four independent lineages from the seventh passage were subjected to genome-length sequencing. Virion RNAs were extracted from culture fluids using RNeasy kits (Qiagen). Viral RNAs were amplified by RT-PCR using SuperScript III one-step RT-PCR kits (Invitrogen). The PCR products were gel purified and sequenced. As controls, three independent lineages of WNV were passaged on naïve BHK-21 cells for seven rounds and sequenced.
The WNV RNA level was quantified by real-time RT-PCR using TaqMan One-Step RT-PCR master mixture; the assay was measured using an ABI 7500 system according to the manufacturer's protocol (Applied Biosystems). The TaqMan probes contained a 5′ 6-carboxyfluorescein (FAM) reporter and a 3′ 6-carboxy-N,N,N′,N′-tetramethylrhodamine (TAMRA) quencher. Two primer/probe sets were used: one set targeted the Env region (forward primer, 5′-TCAGCGATCTCTCCACCAAAG-3′; reverse primer, 5′-GGGTCAGCACGTTTGTCATTG-3′; and probe, 5′-FAM-TGCCCGACCATGGGAGAAGCTC-TAMRA-3′), and another set targeted the NS5 region (forward primer, 5′-GCTCCGCTGTCCCTGTGA-3′; reverse primer, 5′-CACTCTCCTCCTGCATGGATG-3′; and probe, 5′-FAM-TGGGTCCCTACCGGAAGAACCACGT-TAMRA-3′).
In addition, two sets of primers were used for standard RT-PCR amplification: one set targeted the NS3 gene (forward primer, 5′-TGAGATTGCCCTTTGCCTAC-3′; reverse primer, 5′-TTGAGCGATCAGTCCGTTTG-3′), and another set targeted the NS5 gene (forward primer, 5′-TGAAGCTAAGGTGCTTGAGC-3′; reverse primer, 5′-AGCCACATCTGGGCATAAGA-3′). The RT-PCRs were performed using SuperScript III one-step RT-PCR kits; the products were analyzed on a 1% agarose gel.
Replicon RNA was cured by the treatment of the NeoRep BHK-21 cells with a flavivirus inhibitor (unpublished data). NeoRep cells (1 × 105) in 2 ml of DMEM plus 10% FBS were seeded in a 6-well plate and treated with 50 μM of the inhibitor in the absence of G418; the medium was replenished every 3 days. At various time points after treatment, total cellular RNA was extracted using RNeasy kits, and 500 ng RNA extract was monitored for replicon levels by real-time RT-PCR using a primer/probe set targeting viral NS5 (described above). This real-time RT-PCR method has a detection sensitivity of 40 RNA molecules (37). A standard RT-PCR also was performed to confirm the cure of the replicon RNA (described above). One hundred nanograms of total cellular RNA was subjected to RT-PCR using SuperScript III one-step RT-PCR kits.
A stable BHK-21 cell line carrying a WNV replicon was used to study superinfection exclusion. The replicon contained a deletion of viral structural CprMEnv genes and an insertion of a neo gene driven by an EMCV internal ribosomal entry site at the upstream region of the 3′ UTR (NeoRep) (Fig. (Fig.1A).1A). Compared to the WT replicon, the NeoRep replicon contained a Glu→Gly change at amino acid position 249 in NS4b; this mutation improved the efficiency of NeoRep in the establishment of noncytopathic persistent replication in rodent cells (32). To examine the effect of the replicon on WNV superinfection, we infected naïve and NeoRep BHK-21 cells with WNV (MOI of 0.1) and compared the viral growth kinetics. The NeoRep cells produced 10 to 100 times less virus than the naïve cells from 16 to 48 h p.i. (Fig. (Fig.1B).1B). In addition, we transfected equal amounts of genome-length RNA (10 μg) into the naïve and NeoRep BHK-21 cells and compared the specific infectivity values, viral titer production (Fig. (Fig.1C),1C), and the expression levels of Env protein (expressed only from the genome-length RNA) (Fig. (Fig.1D)1D) between the two cells. The naïve and NeoRep cells generated 1.3 × 105 and 6.7 × 103 PFU per μg of RNA, respectively. At 24 h p.t., the cells with and without NeoRep produced 3.3 × 106 and 7.3 × 108 PFU/ml of viruses, respectively, in culture medium (Fig. (Fig.1C).1C). IFA analysis showed that >80% of the transfected cells were positive in Env protein expression at 24 h p.t., whereas <5% of the transfected NeoRep cells expressed a detectable level of Env protein (Fig. (Fig.1D).1D). These results demonstrate that the preestablished replication of replicon excludes WNV superinfection.
We examined the viral spectrum that could be excluded from superinfection on replicon cells. Three flaviviruses (SLEV, YFV, and DENV-2, representing JE, YF, and DEN serocomplexes, respectively) were analyzed for their abilities to superinfect BHK-21 cells containing WNV NeoRep. Cells with or without WNV NeoRep were infected with each virus (MOI of 0.1). As shown in Fig. Fig.2,2, the replicon cells produced about 102- and 104-fold less SLEV than the BHK-21 cells at 24 and 48 h p.i., respectively. Compared to that of SLEV, the effects of the WNV replicon on the replication of YFV and DENV-2 were less dramatic, with a viral titer reduction of 5.6- and 3.4-fold at 48 h p.i., respectively (Fig. (Fig.2).2). In contrast, similar viral titers were obtained from the two cell lines (with and without WNV NeoRep) for two nonflaviviruses, WEEV (a plus-strand RNA alphavirus) and VSV (a negative-strand RNA rhabdovirus). The results demonstrate that superinfection exclusion on WNV replicon cells was specific to flaviviruses, and among the flaviviruses, the virus closely related to WNV (i.e., SLEV) is more efficiently excluded than the more distantly related viruses (i.e., YFV and DENV).
Three sets of experiments were performed to determine the step that is blocked during WNV superinfection. The first experiment was aimed to examine viral attachment and entry. BHK-21 cells with and without WNV NeoRep were infected with WNV at an MOI of 10 at 4°C for 1 h. At 4°C, virions attach to the cells but entry is prevented. After a thorough wash to remove unattached viruses, the cells were replenished with fresh medium and maintained at 4°C or shifted to 37°C (to allow for viral internalization) for an additional 1.5 h. To distinguish between attached and internalized particles, the cells were washed with an alkaline high-salt buffer to remove virions that had attached to the cell surface but had not internalized into the cells (6). Total cellular RNAs were extracted, and viral RNA was quantified by real-time RT-PCR. As shown in Fig. Fig.3A,3A, under each treatment condition, similar amounts of viral RNA were detected between the two cell lines. At 4°C without high-salt treatment, an equal level of viral RNA was detected from the two cell lines, indicating an equal efficiency for virion attachment; at 37°C with high-salt treatment, a similar amount of intracellular viral RNA was detected, indicating a similar efficiency of viral entry. The results suggest that superinfection is not blocked at the step of virion attachment or internalization.
The second experiment was performed to examine viral translation and RNA synthesis. We previously showed that the transfection of BHK-21 cells with a WNV luciferase replicon (Rluc2Arep) (Fig. (Fig.3B)3B) generates two distinct luciferase peaks. The first peak, at 2 to 6 h p.t., represents input RNA translation; the second peak, after 12 h p.t., represents RNA synthesis (23). The replicon system allows us to quantitatively differentiate between viral translation and RNA replication. The transfection of an equal amount of Rluc2ARep RNA (10 μg) into the naïve and NeoRep BHK-21 cells yielded similar levels of luciferase signal at 1, 2, 3, 4, 5, and 6 h p.t. in both cells (Fig. (Fig.3B).3B). In contrast, the luciferase signals at 24 to 72 h p.t. from the NeoRep cells were >102- to 103-fold lower than those from the BHK-21 cells. The results indicate that superinfection is blocked at the step of viral RNA synthesis.
The third experiment used VLPs to further analyze viral attachment, entry, translation, and RNA synthesis. WNV VLPs were prepared by supplementing Rluc2ARep with structural proteins in trans (31). The infection of cells with such VLPs resulted in a single-round infection; no further VLPs would be formed due to the lack of structural proteins. Upon VLP infection, similar luciferase signals were obtained between the cells with and without NeoRep at 1, 2, and 4 h p.i. (Fig. (Fig.3C).3C). In contrast, at 24 to 72 h p.i., luciferase activities from the NeoRep cells were about 10-fold lower than those from the BHK-21 cells (P < 0.001 by Student's test), indicating a difference in VLP RNA synthesis between the two cells. The results confirmed that the exclusion of WNV superinfection occurs at the step of viral RNA synthesis.
We selected WNV variants that can overcome the exclusion imposed by the preestablished replicon replication. Such superinfecting viruses were obtained by the continuous culturing of the WT virus in the NeoRep BHK-21 cells in medium containing 1 mg/ml of G418 (Fig. (Fig.4A).4A). Four independent selections (Sel A, B, C, and D) were performed to examine the experimental reproducibility. After 10 rounds of passaging, viruses from different rounds were analyzed for their abilities to superinfect the replicon cells compared to that of the WT virus. The results showed that, after seven rounds of passaging, viruses from all four independent selections improved their ability to replicate in the NeoRep cells (Fig. (Fig.4B).4B). At 16 to 48 h p.i. (MOI of 0.1), the selected variants generated 10- to 100-fold more infectious viruses than the WT WNV in the NeoRep cells. Viruses beyond the seventh round did not further improve the superinfection (data not shown).
Both the WT and selected viruses did not form plaques on the NeoRep cells. We performed an immunostaining-based infectious center assay using an antibody against viral Env protein (Fig. (Fig.4C).4C). In agreement with the growth kinetics (Fig. (Fig.4B),4B), the selected viruses generated larger immunostaining centers than the WT virus.
The genome-length sequencing of all four selected viruses revealed four mutations (Fig. (Fig.4D):4D): (i) an E→K substitution at amino acid position 138 (E138K) of the Env protein from Sel A, B, and D variants; (ii) an S90R change in the prM protein from Sel C virus; (iii) a K124R mixed population in NS4a from Sel A and C variants; and (iv) a V9M change in peptide 2K from all four variants. The peptide 2K is a transmembrane protein located between NS4a and NS4b. The results demonstrated that WNV variants with mutations in both structural and nonstructural genes could be selected to overcome the exclusion of superinfection.
We engineered the recovered mutations (Fig. (Fig.4D)4D) into an infectious cDNA clone of WNV. Two panels of recombinant viruses were prepared. The first panel was based on Sel A and contained four mutant viruses (three variants contained a single mutation in Env, NS4a, or 2K, and one variant contained triple mutations of Env, NS4a, and 2K (designated Env+NS4a+2K). Growth kinetics of the viruses were analyzed on BHK-21 cells with and without NeoRep (Fig. (Fig.5A).5A). On naïve BHK-21 cells, WT, NS4a, and 2K mutants yielded equivalent viral titers at 24 and 48 h p.i.; however, viral titers of the Env mutant and the Env plus 2K+NS4a mutant were almost 10-fold higher than those of the WT virus (Fig. (Fig.5A,5A, left). On the NeoRep cells, all mutants replicated to levels higher than that of the WT virus in the order of Env+NS4a+2K > Env > 2K > NS4a; the triple mutant Env+NS4a+2K replicated to a level similar to that of the originally selected Sel A virus (Fig. (Fig.5A,5A, right). These results indicate that the 2K and NS4a mutations preferentially enhance viral replication in the NeoRep cells. In contrast, the Env mutation nonspecifically enhanced viral replication in both naïve and NeoRep cells; this mutation may represent a viral adaptation to BHK-21 cells rather than a specific means to overcome the exclusion of superinfection. In fact, culturing the WT WNV on BHK-21 cells for seven rounds consistently produced viruses that contained the Env E138K mutation; no other mutations were recovered from three independently passaged viruses (data not shown).
The second panel of recombinant viruses was prepared based on Sel C and contained two mutants (one variant contained a single mutation in prM, and one variant contained the double mutation prM+2K). Growth kinetic analysis showed that the prM mutant produced 5- to 10-fold more virus than the WT virus in both BHK-21 and NeoRep BHK-21 cells (Fig. (Fig.5B).5B). This result indicates that, similarly to the Env mutation, the prM mutation nonselectively enhances viral infection in BHK-21 cells with or without the replicon. The prM and prM+2K mutants replicated to equivalent levels in BHK-21 cells. In contrast, the prM+2K mutant produced a higher viral titer than the prM mutant in the NeoRep cells; furthermore, the double mutant replicated to a level close to that of the originally selected Sel C virus (Fig. (Fig.5B).5B). The results again indicate that the 2K mutation selectively enhances WNV replication in the replicon cells.
We next used VLPs to further examine the effect of mutations on viral replication. Two sets of VLPs were prepared: the first set contained WT Rluc2ARep and WT or mutant structural proteins (prM or Env), and the second set contained WT structural proteins and replicons with mutations in nonstructural genes (NS4a, 2K, and NS4a+2K). BHK-21 cells with or without NeoRep were infected with equal amounts of VLPs (normalized by replicon RNA copy numbers) and monitored for luciferase activities at 24 h postinfection. As shown in Fig. Fig.5C5C (left), the mutations in prM or Env enhanced luciferase signals in both naïve and NeoRep cells. The calculation of these luciferase signals showed that (i) the infection with Env mutant VLP generated 36- and 38-fold more luciferase activity in BHK-21 and BHK-21 NeoRep cells than the infection with the WT VLP, respectively, and (ii) the infection with prM mutant VLP generated five- and sixfold more luciferase signals in BHK-21 and BHK-21 NeoRep cells than the infection with the WT VLP, respectively (Fig. (Fig.5C,5C, right). These results indicate that, for individual Env or prM mutant VLP, similar levels of increase in luciferase signals were observed for both cell lines (Fig. (Fig.5C,5C, right). In contrast, the mutations in NS4a, 2K, and NS4a+2K resulted in a more dramatic increase in luciferase activity in the NeoRep cells than in the naïve cells (Fig. (Fig.5C,5C, right). Collectively, the results indicate that the mutations in structural genes nonselectively enhance viral infection in both cell lines, whereas the mutations in nonstructural genes more selectively enhance viral replication in the replicon cells than that in the naïve cells.
We selected the Env E138K mutation to explore the mechanism of replication enhancement. First, the effect of Env mutation on viral attachment and entry was examined using recombinant virus and VLP. (i) For recombinant virus, 31% more Env mutant viruses were attached to BHK-21 cells than WT virus (P < 0.05 by Student's t test) after incubating the cells with viruses at 4°C, suggesting that the E138K mutation increases virion attachment. In agreement with this result, after incubating the infected cells at 37°C for 1.5 h (to allow virion internalization), 21% more viral RNA was detected from the Env mutant virus-infected cells than from the WT virus-infected cells (P < 0.05). The increase in mutant virion internalization most likely is due to its increase in virion attachment. (ii) For the VLP approach, WT and mutant Env VLPs (both containing WT Rluc2ARep) were prepared. Upon the infection of BHK-21 cells with equal amounts of VLPs (normalized to viral RNA), the level of luciferase at 2 h p.i. from the Env mutant VLP was 3.6-fold higher than that from the WT VLP (Fig. (Fig.6B).6B). Since the luciferase signal at 2 h p.i. was derived from the initial translation of VLP input RNA, the difference in luciferase activity between the WT and mutant VLPs indicated the efficiency in viral entry. Collectively, the results demonstrate that the Env E138K substitution enhances virion attachment to BHK-21 cells.
Second, we examined the effect of prM and Env mutations on virion particle assembly/release. BHK-21 cells were transfected sequentially with WT Rluc2ARep RNA and SFV CprMEnv RNA (expressing WT or mutant structural proteins). At 48 h after the Rluc2ARep transfection, intracellular and extracellular levels of Rluc2ARep RNA were used to indicate replicon replication and virion assembly/release, respectively. Approximately 25% more replicon RNA was detected in the prM or Env mutant VLP-packaging cells than in the WT prM/Env VLP-packaging cells (Fig. (Fig.6C,6C, left). The yields of mutant Env VLP and prM VLP in culture medium were 51 and 87% less than that of the WT VLP, respectively (Fig. (Fig.6C,6C, right). The results suggest that the Env and prM mutations reduce virion assembly/release.
The luciferase replicon (Rluc2ARep) of WNV was used to examine the role of NS4a and 2K mutations in overcoming the exclusion of superinfection. We transected both BHK-21 and NeoRep cells with equal amounts of WT and mutant (NS4a, 2K, and NS4a+2K) Rluc2ARep and assayed for luciferase activities at various time points posttransfection (Fig. (Fig.6D).6D). In BHK-21 cells, the mutant and WT replicons produced similar luciferase signals from 2 to 72 h p.t. (Fig. (Fig.6D,6D, top). In the NeoRep cells, similar luciferase signals were detected at 2 and 4 h p.t. for all replicons; however, the mutant replicons generated much higher luciferase signals than the WT replicon at >24 h p.t.; for instance, the luciferase signals at 36 h p.t. from the NS4a, 2K, and NS4a+2K mutant replicons were 3-, 31-, and 84-fold, respectively, higher than that from the WT replicon (Fig. (Fig.6D,6D, bottom). The results indicate that the NS4a and 2K mutations more selectively enhance replicon replication in the NeoRep cells than that in the naïve BHK-21 cells.
We examined whether WNV infection could affect the preestablished replication of replicon. BHK-21 cells containing a dual-reporting replicon (RlucNeoRep) (Fig. (Fig.7A)7A) were infected with WT, mutant NS4a, and mutant 2K WNVs. The luciferase activity was used to quantify the effect of viral infection on the replicon replication (Fig. (Fig.7B).7B). In uninfected cells, the luciferase signal increased during the first 36 h because of cell proliferation; the luciferase activity peaked and decreased from 48 to 60 h, most likely due to the overgrowth of the cells. In the WNV-infected cells, the luciferase signal was dramatically reduced by the NS4a and 2K mutant viruses but not by the WT virus (Fig. (Fig.7B).7B). At 60 h p.i., the NS4a and 2K mutant viruses reduced the luciferase activity by 36 and 93%, respectively, whereas the WT virus decreased the luciferase activity by only 6% (Fig. (Fig.7C).7C). The infected cells did not exhibit apparent cytopathic effect up to 60 h p.i. (data not shown). These results indicate that the NS4a and 2K mutations enable the mutant viruses to suppress the preestablished replicon replication.
The less permissive nature of the NeoRep cells to WNV replication could result from a change(s) in host factors during the selection and maintenance of the replicon cells. To exclude this possibility, we cured NeoRep by treating the cells with a flavivirus inhibitor in the absence of G418. The residual replicon RNA was monitored by real-time RT-PCR (Fig. (Fig.8A).8A). After 17 rounds of compound treatments (3 days per round), the replicon RNA from one of the three independently treated cells (Cure 1) was undetectable by real-time RT-PCR, while the other two treated cells still had low levels of replicon (Fig. (Fig.8A).8A). These results were confirmed by standard RT-PCR (Fig. (Fig.8B).8B). The data indicate that the inhibitor had completely eliminated the replicon in the Cure 1 cells and almost eliminated the replicon in Cure 2 and 3 cells.
WT WNV was used to infect the three cured cells, naïve BHK-21 cells, and NeoRep BHK-21 cells (MOI of 0.1). Similar levels of viral growth kinetics were observed from the cured cells and the naïve BHK-21 cells (Fig. (Fig.8C).8C). Among the three cured cells, the Cure 3 cells (containing the highest level of residual replicon RNA) produced slightly less WNV than the Cure 1 and 2 cells. As expected, the NeoRep cells produced 10- to 100-fold fewer viruses than other cells. Cells that had been only partially cured of the replicon (after 10 rounds of inhibitor treatment) showed an intermediate phenotype between BHK-21 and NeoRep BHK-21 cells (data not shown). Overall, the results indicate that (i) the exclusion of superinfection requires the presence of WNV replicon RNA and viral proteins, (ii) the exclusion is not due to any changes of host factors in the replicon cells, and (iii) the permissiveness of the NeoRep cells to efficient WNV infection could be reversed by curing the replicon RNA.
Superinfection exclusion has been reported for a broad range of viruses. A number of mechanisms have been proposed to modulate the exclusion, including competition for cellular receptors or intracellular host factors, the production of interferon or interferon-like substances by the infected host cell, the production of defective interfering viral genomes from the first infecting virus, or the production of a trans-acting protease by the first infecting virus (12). In this study, we used BHK-21 cells carrying WNV NeoRep to study superinfection exclusion for flaviviruses. We show that WNV NeoRep excluded the superinfection of flaviviruses but not nonflaviviruses (Fig. (Fig.11 and and2).2). The permissiveness of NeoRep cells to nonflavivirus infections excludes the possibility that superinfection exclusion was due to the NeoRep-mediated induction of a nonspecific antiviral activity. Mode-of-action analysis showed that the superinfection exclusion of WNV occurs at viral RNA synthesis, not at virion attachment/entry or RNA translation (Fig. (Fig.3).3). Similarly to our results, the superinfection exclusion of HCV, a member of the genus Hepacivirus, was shown to be blocked at viral RNA synthesis (35, 43). Because the replicon system used in this study did not contain structural proteins, we cannot exclude the possibility that WNV superinfection also is excluded at virion assembly/release. Interestingly, we recently found that the infection of BHK-21 cells expressing viral CprMEnv proteins by DENV was much less efficient than the infection of BHK-21 cells without the expression of viral structural proteins (unpublished). The latter results suggest that the expression of structural proteins results in homologous interference in DENV.
To explore the mechanisms of superinfection exclusion, we selected superinfecting viruses that could overcome the exclusion by passaging the WT WNV on NeoRep BHK-21 cells. The sequencing of the superinfecting viruses revealed adaptive mutations in both structural (prM and Env) and nonstructural (NS4a and 2K) regions (Fig. (Fig.4D).4D). The analysis of recombinant viruses and VLPs demonstrated that both structural and nonstructural mutations contributed to overcoming the exclusion. Interestingly, the structural mutations nonselectively enhanced viral replication in both naïve and NeoRep BHK-21 cells (Fig. (Fig.5);5); the nonselective nature of enhancement also was reflected in the fact that the same Env mutation was recovered after passaging the WT WNV on BHK-21 cells (data not shown).
The Env E138K mutation previously was reported for JEV and WNV after serially passaging the viruses on human adenocarcinoma (SW13) cells (16) or after gamma-ray irradiating and recovering the JEV on BHK-21 cells (3). Similarly to our WNV results, the Env E138K JEV was shown to generate higher viral titers than the WT JEV (3). The E138 residue is located in domain I of the Env protein; this residue is conserved within the JEV serocomplex but not among other flaviviruses. Direct binding analysis showed that WNV and JEV containing the E138K mutation bound to the cell surface glycosaminoglycans (GAG) more efficiently than the corresponding WT virus. Consequently, after inoculating the viruses into mice, the E138K virus was quickly eliminated from serum (through binding to GAG found on extracellular matrices and cell surfaces), resulting in inefficient dissemination and virulence attenuation (16). These results indicate that the enhanced attachment of Env mutant WNV to cells (observed in Fig. 6A and B) was due to a higher affinity of the mutant Env for GAG. Along the same lines, since GAG is found ubiquitously on cell surfaces, the higher affinity of mutant Env/GAG binding could limit the release of progeny virions. This possibility was supported by the results that, during VLP production, the Env mutation reduced 51% of the extracellular viral RNA (in the form of VLPs in culture fluids), while it increased to 25% of the intracellular viral RNA (Fig. (Fig.6C6C).
The furin-mediated internal cleavage of prM is required for flavivirus virion infectivity. Flaviviruses share conserved basic residues at positions P1, P2, and P4 for the recognition of furin at the prM cleavage site; basic residues at positions P3, P5, and P6 of the substrate also are preferred for furin cleavage (2). The prM S90R mutation is located at the P3 position of the cleavage site that potentially could improve the prM cleavage. A previous study of DENV-2 showed that a decrease in prM cleavage was associated with higher proportions of subviral particles and prM-containing virions in culture medium (11, 13). On the other hand, enhanced prM cleavability adversely affects DENV-2 export and release (13). The latter result was in agreement with our observation that the prM S90R mutation reduced VLP yield in culture medium by 87% (Fig. (Fig.6C6C).
In contrast to the prM and Env mutations, mutations recovered in viral nonstructural genes selectively enhanced viral replication in NeoRep BHK-21 cells; the V9M mutation in 2K peptide exhibited a greater enhancement than that of the K24R mutation in NS4a (Fig. (Fig.55 and and6D).6D). Flavivirus 2K peptide consists of 23 amino acids and spans the ER membrane, with its N- and C-terminal tails on the cytoplasmic and ER lumen sides, respectively (26). The cleavage at the NS4a-2K junction by viral NS2b/NS3 protease is a prerequisite for a cleavage at the 2K-NS4b junction by a host signalase (19). The regulated cleavages at the NS4a-2K-NS4b junctions are critical for the rearrangement of cytoplasmic membrane (26, 34). In Kunjin virus, the individual expression of NS4a-2K resulted in membrane rearrangement that mostly resembled a virus-induced structure, while the removal of the 2K peptide led to a less profound membrane rearrangement (34). Although the exact function of 2K peptide remains to be determined, its role in RNA replication was indicated by the results that V9M mutation in 2K conferred on WNV to overcome the exclusion of superinfection (Fig. (Fig.55 and and6D).6D). The same 2K mutation recently was reported to confer WNV resistance to lycorine, an inhibitor of flavivirus RNA synthesis (46). These results suggest that the 2K mutation modulates the ER membrane rearrangement, leading to a better scaffold for replication complex formation. Along the same lines, the NS4a K124R mutation was mapped to the P3 position of the NS4a-2K cleavage site; although the K124R substitution is a minor side chain change, it may affect the cleavage efficiency, facilitating membrane rearrangement and replication complex formation. In addition, NS4a mutation may affect its interaction with host factors to overcome the exclusion of superinfection. Experiments are ongoing to dissect these potential mechanisms.
We found that NeoRep BHK-21 cells became permissive for WNV infection after treatment with a flavivirus inhibitor (Fig. (Fig.8).8). The results indicate that superinfection exclusion was not due to any changes of host factors during the selection of replicon cells; instead, the exclusion was due to the presence of replicon RNAs and proteins. It is possible that the preestablished replication of replicon has used up the host factors (e.g., cellular proteins or lipid membrane components) required for replication complex formation; the lack of host factors limits the replication of the superinfecting WT WNV. However, the 2K mutation and, to a less degree, the NS4a mutation could not only overcome the exclusion of superinfection but also could suppress the preestablished replicon replication. In contrast, incubating the replicon-containing cells with the WT WNV did not significantly suppress the replicon replication (Fig. (Fig.7).7). These results support the idea that superinfection exclusion is a result of competition for intracellular host factors. It remains to be determined how the mutant 2K scavenges the host factors that already have been recruited to the preestablished replication complex.
We thank the Molecular Genetics Core and the Cell Culture Facility at the Wadsworth Center for DNA sequencing and for the maintenance of BHK-21 and Vero cells, respectively.
The work was supported partially by federal funds from the National Institute of Allergy and Infectious Diseases, National Institutes of Health (NIH), under contract N01-AI-25490 and by NIH grants 1U01AI061193 and U54-AI057158 (Northeast Biodefense Center).
Published ahead of print on 2 September 2009.