INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Within the genus Flavivirus of the family Flaviviridae, yellow fever (YF) and Japanese encephalitis (JE) viruses are distinguishable by a number of properties. The viruses are antigenically distinct (2), lack common mosquito vectors and vertebrate reservoirs (22, 24), and cause dissimilar disease syndromes in humans (15). Such differences presumably reflect evolutionary divergence of these viruses from common ancestors (16). To begin investigation of the molecular basis for the differences in biological properties of these two flaviviruses, we have engineered chimeric YF/JE viruses in which the structural proteins prM and E of JE virus were exchanged for the homologous proteins of YF virus within a molecular clone of the YF17D strain (21). This strategy was based on the previous observation that chimeric viruses between distantly related members of this family, such as TBE and DEN, can be recovered from engineered cDNA templates (19). Because of the conserved features of flavivirus genome organization and replication (4), it may be possible to genetically engineer a range of such chimeric viruses. This approach is relevant for the potential use of recombinant flaviviruses as live-attenuated vaccine candidates (1, 14) and also for studying protein-protein and RNA-protein interactions important for flavivirus replication and pathogenesis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and viruses. SW-13 (derived from human adenocarcinoma originating from adrenal cortex), Vero, LLC-MK₂, C6/36, and NB41A3 (mouse neuroblastoma) cells were originally obtained from the American Type Culture Collection (ATCC). YF5.2iv (a molecular clone of the YF17D strain), has been previously described (21). The JE Nakayama (JE-N) strain was obtained from ATCC and was amplified in LLC-MK₂ cells. JE SA₁₄-14-2 was obtained at passage level PHK-5 (courtesy of Kenneth H. Eckels) and amplified in LLC-MK₂ cells.

Plasmid constructions. JE SA₁₄-14-2 cDNA was derived by reverse transcription (RT)-PCR from infected LLC-MK₂ cells based on the published nucleotide sequence of this strain (18). The structural region was cloned into pBluescript-KS(+) (pBS) (Stratagene) by using two primer sets. The 5' terminus through nucleotide 1132 was cloned into pBS with primers containing EcoRI sites to yield pBS/JE(1-1131). The 3' primer contained a nested NheI restriction site for subsequent insertion of an engineered chimeric YF/JE fragment containing the YF nucleotide sequence from 1 to 481 joined to the JE nucleotide sequence from 477 to 1131 (see below). The region from nucleotide 1108 to nucleotide 2472 was cloned by using primers containing XbaI restriction sites to produce pBS/JE(1108-2472). The 5' primer contained a nested NsiI restriction site, and the 3' primer contained a nested KasI site to facilitate insertion of an NsiI-KasI digestion fragment of pBS/JE(1108-2472) into YFM5.2[KasI] (see below). The cDNA for the JE-N has been previously described (13).

Recovery of infectious virus. Full-length cDNA templates were assembled by in vitro ligation of restriction fragments after digestion of the YF/JE plasmids with NheI and BspEI and gel purification of the appropriate DNA products. The in vitro-ligated templates encoding chimeric YF/JE viruses were used for synthesis of SP6 RNA transcripts based on methods described previously (21). RNA transfection of Vero cells was done in the presence of Lipofectin (Gibco/BRL), using between 100 and 250 ng of RNA transcript. Virus was harvested after onset of cytopathic effect and titrated by plaque assay on LLC-MK₂ cells.

Protein labelling. Vero cells or LLC-MK₂ cells were infected with virus at a multiplicity of 1 PFU/cell and labelled at approximately 48 h postinfection with [³⁵S]methionine for 10 to 12 h. The media from the culture or lysates made from the cell monolayers were used for immunoprecipitation of the viral proteins under either denaturing conditions for YF polyclonal anti-E antisera (5) or nondenaturing conditions for JE monoclonal antibody (12) and with hyperimmune antisera against YF and JE. Immunoprecipitates were processed and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and fluorography as previously described (5).

Neutralization assays. Mouse hyperimmune antisera to YF, JE, and nonimmune ascitic fluid were obtained from the ATCC. Anti-YF immunoglobulin G (IgG) to the YF E protein and a nonimmune isotype-matched IgG were purified from ascites fluid containing monoclonal antibody 2E10 or 2C5 (23 [courtesy of Jack Schlesinger]) by using a monotrap protein A column (Pharmacia). IgG concentrations were estimated by using stained protein markers (Sigma) as standards for quantitation by SDS-PAGE. For neutralization assays, antisera or IgGs were diluted in minimal essential medium plus 3% fetal bovine serum. Plaque reduction titers were calculated as the highest dilution of serum or IgG which neutralized 50% of the input virus (100 PFU). Plaque assays were performed on LLC-MK₂ cells.

Growth curves. Growth curves were done by infecting confluent LLC-MK₂, C6/36, SW-13, or NB41A3 cells at a multiplicity of 0.5 PFU/cell and harvesting the media at successive 12 or 24-h intervals postinfection. Yields of virus in each sample were then quantitated by plaque titration on LLC-MK₂ cells.

Mouse experiments. Three- to four-week-old outbred male and female mice (ICR and C57BL/6 strains; Harlan-Sprague Dawley) were used. Neurovirulence was assessed by intracerebral inoculation of virus diluted in sterile phosphate-buffered saline plus 5% fetal bovine serum into the left cerebral hemisphere of anesthetized mice. Virus doses were confirmed by back-titration of the inocula on LLC-MK₂ cells. Neuroinvasiveness was assessed by inoculation of virus by the intraperitoneal route. Endpoints were scored as either the day of onset of a moribund condition or, alternatively, survival to 3 weeks after inoculation. Mice found in a moribund state were euthanized.

Nucleotide sequence analyses. DNA sequences of the plasmids encoding the YF/JE chimeras were determined by dideoxy sequencing by the Sequenase (U.S. Biochemical Corp.) protocol. Duplicate clones were sequenced for each chimeric virus clone.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Recovery of chimeric YF/JE viruses. We chose to engineer chimeric viruses encoding the prM and E proteins of the JE SA₁₄-14-2 human vaccine strain and the virulent JE-N strain. The SA₁₄-14-2 strain is an attenuated derivative of the JE SA-14 strain, obtained after serial passage in cell culture (6, 17). JE SA-14 is a virulent strain originally isolated from mosquitoes in China (17). The JE-N strain was selected rather than JE SA-14 because of availability of a well-characterized cDNA clone encoding the structural proteins of this virus (12, 13). Figure 1 shows the structural region of the plasmids encoding the chimeric viruses and the recovery of virus from cells transfected with synthetic RNA derived from the plasmid templates. High titers of infectious virus were recovered from transcripts derived from templates encoding the YF capsid protein together with the JE prM and E proteins of the SA₁₄-14-2 strain (YF/JE-S) or the Nakayama strain (YF/JE-N): 6.8 and 7.3 log₁₀ PFU/ml, respectively. The yield of YF5.2iv virus after transfection of Vero cells under similar conditions was 6.1 log₁₀ PFU/ml (data not shown). Templates encoding the entire JE structural region (C-prM-E) did not yield detectable virus by this assay. Virus recovered from the transfections was passaged onto new cell monolayers, and total intracellular RNA was used for RT-PCR to verify the chimeric structure. Primer pairs which amplify the YF genome from nucleotide 1 to nucleotide 2980 were used to generate PCR products which were analyzed by restriction enzyme digestion to verify the presence of JE sequences in the recovered virus (data not shown). Moreover, nucleotide sequence analysis of the PCR products across the chimeric C/prM junction revealed the predicted structure of the chimeric virus (data not shown).

View larger version (24K): [in this window] [in a new window]   FIG. 1.   Structure of cDNA templates for chimeric YF/JE viruses (truncated within the NS1 protein for clarity). YF/JE-S and YF/JE-N refer to the two chimeric viruses as described in the text. The 5' nontranslated region is derived from YF5.2iv (6). Hatched regions are from JE SA14-14-2, and solid regions are from JE-N. The remainder of the nonstructural region is derived from YF5.2iv. The amount of virus recovered is indicated as the titer of virus (log10 PFU per milliliter) in the media at the time of harvest at 96 h after RNA transfection of Vero cells.

Properties of viral proteins. Figure 2A illustrates the immunoprecipitation of viral E proteins present in the media of cells infected with either the parental or chimeric viruses. The YF E protein migrates with a molecular mass of approximately 50 kDa, consistent with previous observations (5). The E proteins of the chimeric viruses and parental JE viruses migrated with molecular masses of approximately 52 kDa, consistent with the presence of an N-linked glycan on the E protein (12). To determine if this difference in the apparent molecular masses of the JE E proteins relative to the YF E protein resulted from N-linked glycosylation, the profiles of intracellular viral glycoproteins produced in the presence and absence of tunicamycin were analyzed. Figure 2B illustrates that the apparent molecular mass of the YF E protein is not affected by tunicamycin, whereas the molecular masses of the JE E proteins are reduced from 52 kDa to 50 kDa when labelled in the presence of this drug. This is consistent with utilization of a single N-linked glycosylation site on these proteins. The YF prM protein migrated with an apparent molecular mass of 24 kDa, consistent with addition of two N-linked glycans (5), whereas the JE prM protein migrated with a molecular mass of 19 kDa, consistent with a single N-linked glycan (12). In the presence of tunicamycin, the molecular mass of the JE prM protein was reduced by 2 kDa.

View larger version (63K): [in this window] [in a new window]   FIG. 2.   Immunoprecipitation of viral proteins from chimeric YF/JE viruses. (A) Parental or chimeric viruses were grown in LLC-MK2 cells, labelled with [35S]methionine, and harvested from the media at 48 to 60 h postinfection. Viral E proteins were immunoprecipitated from the media as described in Materials and Methods, and proteins were analyzed on an SDS-9% polyacrylamide gel. The E protein of YF5.2iv virus was immunoprecipitated with rabbit polyclonal antiserum against YF. The E proteins of the JE-S (JE SA14-14-2), YF/JE-S, YF/JE-N, and JE-N viruses were immunoprecipitated with mouse hyperimmune ascitic fluid against JE virus. (B) Viral proteins produced in Vero cells. Proteins were labelled with [35S]methionine for 6 h, and lysates were prepared at 48 h postinfection as described in Materials and Methods. Virus in the media was harvested and immunoprecipitated with hyperimmune ascitic fluid to either YF or JE, as described for panel A. Mock-infected lysates were immunoprecipitated with a mixture of YF and JE hyperimmune ascitic fluids. Proteins were analyzed on 13% SDS-polyacrylamide gels. Tunicamycin was added (+) or not added () at the time of labelling at a concentration of 7.5 µg/µl. Molecular mass markers, indicated by small lines in the right margin, represent 220, 97.4, 66, 46, 30, and 14.3 kDa, respectively.

View larger version (16K): [in this window] [in a new window]   FIG. 3.   Immunoprecipitation of YF and JE proteins with monoclonal and polyclonal antisera. LLC-MK2 cells were infected and labelled as described for Fig. 2A. Viral proteins were immunoprecipitated from the media with monoclonal antibody to JE (A) or polyclonal antisera to the YF and JE viruses (B and C), as described in Materials and Methods. In panel A, YF, YF/JE-N, and YF/JE-S refer to media from cells infected with these respective viruses. In panel B, YF-infected cells were used. In panel C, YF/JE-N-infected cells were used. NI refers to nonimmune ascites fluid. YF-HI and JE-HI refer to hyperimmune ascites fluid to the YF and JE viruses, respectively. Proteins were analyzed on 10% SDS gels.

Growth efficiency in cell culture. Growth kinetics of the chimeras were initially examined on LLC-MK₂ and C6/36 cells. As shown in Fig. 4A and B, differences were observed between the chimeras and the parental viruses in terms of the maximal yield of virus production on LLC-MK₂ cells, based on single-step growth analysis. All viruses reached a peak of production at approximately 48 h postinfection. The maximal virus production from the chimeric viruses was higher than that of the parental YF5.2iv virus. YF/JE-S replicated more efficiently than JE SA₁₄-14-2, whereas YF/JE-N replicated less efficiently than JE-N. Both chimeric viruses replicated better than YF5.2iv. On mosquito cells, the YF/JE chimeric viruses also exhibited efficient replication compared to the parental YF5.2iv and JE SA₁₄-14-2 viruses (Fig. 4C). The growth properties of the chimeras were examined on two other cell lines. Comparison of the growth rates of the chimeras on SW-13 cells (Fig. 4D) revealed a difference in peak titer, with the YF/JE-N virus exhibiting a higher replication efficiency than YF/JE-S in this cell line. This virus also exhibited higher plaque efficiency than YF/JE-S on SW-13 cells (data not shown). On NB41A3 cells (Fig. 4E), the YF/JE-S and parental YF5.2iv viruses exhibited roughly similar growth kinetics, whereas YF/JE-N appeared to replicate slightly less efficiently.

View larger version (39K): [in this window] [in a new window]   FIG. 4.   Growth curves of chimeric viruses in cell culture. Cells were infected at a multiplicity of 0.5 PFU/cell, and media were collected at 12- or 24-h intervals, followed by plaque titration on LLC-MK2 cells. (A) YF/JE-S compared to its parental viruses on LLC-MK2 cells. (B) YF/JE-N compared to its parental viruses on LLC-MK2 cells. The titers represent averages of triplicate samples for both experiments. (C) Chimeric virus growth on C6/36 cells, with values representing an average of two samples for each virus. (D) Growth of YF/JE-S and YF/JE-N viruses on SW-13 cells. (E) Growth of YF5.2iv, YF/JE-S, and YF/JE-N on NB41A3 cells. The titers represent averages of triplicate samples, except for the last time point, which was determined in duplicate.

Mouse virulence testing. Mice 3 to 4 weeks old were used to assess virulence phenotypes. Figure 5A compares the results obtained with the YF/JE-S chimera with those obtained with the YF5.2iv and JE SA₁₄-14-2 viruses. At a dose of 10⁴ PFU delivered intracerebrally, parental YF caused 100% mortality within 7 to 12 days, whereas neither the JE SA₁₄-14-2 strain nor the YF/JE-S chimera caused any mortality. No signs of illness were observed over the period of 3 weeks after inoculation for the two latter viruses. The neurovirulence of the YF/JE-N virus in this assay is compared to the YF5.2iv and JE-N virus in Fig. 5B. The YF/JE-N chimera exhibited a level of neurovirulence similar to the YF5.2iv virus in this model (average time of survival, 7 days), whereas the JE-N virus required less time to cause fatal disease (average time of survival, 4 days). The attenuation phenotype of the YF/JE-S virus was also demonstrated at a higher-dose intracerebral challenge (Table 2). Compared to the 100% mortality of mice receiving 10⁴ PFU of YF5.2iv virus, no mortality or signs of illness were observed in mice receiving between 10⁴ and 10⁶ PFU of the YF/JE-S chimera. In contrast, the YF/JE-N virus was lethal for 60% of mice at a dose of only 10 PFU (Table 2).

View larger version (20K): [in this window] [in a new window]   FIG. 5.   Mouse neurovirulence assay. A fixed-dose intracerebral challenge with 104 PFU was carried out in 4-week-old ICR mice. (A) YF/JE-S compared with its parental viruses. (B) YF/JE-N compared with its parental viruses. Differences in mortality between YF/JE-S and YF5.2iv were significant (P < .005), based on 2 analysis of the proportion of survivors.

Nucleotide sequence analysis of chimeric viruses. Plasmids encoding the chimeric YF/JE-N and YF/JE-S viruses were sequenced through the JE prM-E portion of the respective clones (Table 4). The E protein sequence of YF/JE-N agreed with that of the JE-N strain (13). The sequence of YF/JE-S agreed with that of JE SA₁₄-14-2 (18), except at two positions (residues 177 and 264, as discussed below). Comparison of the predicted amino acid sequences reveals a total of 10 differences between the E proteins of the chimeric viruses. Four of these residues (227, 244, 315, and 439) are common to YF/JE-S and its virulent JE SA-14 parent. The remaining six residues (107, 138, 176, 177, 264, and 279) are common between the sequences of the YF/JE-N virus and the virulent JE-SA14 virus. In the prM region, three residues of the YF/JE-S chimera (I₁₄, I₁₂₉, and V₁₄₀) differed from those in YF/JE-N (V₁₄, V₁₂₉, and I₁₄₀) (data not shown).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

This study demonstrates that a molecular clone of YF17D virus can be engineered to express the structural proteins prM and E of the heterologous flavivirus, JE virus. These findings are similar to those originally reported for a chimeric DEN/TBE virus showing that the prM-E region of TBE could be expressed in the context of the dengue-4 virus genome (19). Although chimeric YF/JE(prM-E) viruses could be readily recovered, failure to generate YF/JE(C-prM-E) chimeric virus could be due to several possible causes. These include incompatibility of predicted cis-acting RNA structures, such as conserved sequences found within the capsid region (7), inefficient processing of the JE C/prM junction by the YF protease, or other steps, such as defective packaging or deleterious effects of the JE capsid on RNA synthesis driven by YF nonstructural proteins. It has been shown, for instance, that Kunjin virus RNA replication requires between 2 and 20 amino acid residues of the homologous capsid protein, although it is not known whether nonhomologous sequences can substitute in the capsid region (11). Similar requirements may exist for other flaviviruses.

Immunoprecipitation and plaque-reduction neutralization assays were used to determine whether the JE E protein is expressed in an antigenically intact form in the context of chimeric YF/JE viruses. The JE prM and E proteins appeared to be properly glycosylated, as indicated by detection of radiolabelled proteins in infected cells, and the E protein exhibited both JE-specific as well as cross-reactive flavivirus epitopes. The neutralization data suggest that protective antibody epitopes of the JE E protein are likely to be preserved in the chimeric viruses, although we cannot rule out subtle structural differences between their E proteins relative to those of the parental JE viruses. Thus, it may be possible to elicit JE-specific neutralizing antibodies by using these chimeras as experimental vaccines.

Differences in growth efficiency of the chimeric viruses were observed in cell culture in LLC-MK₂ cells. Both chimeric viruses exhibited higher growth efficiency than the YF5.2iv parent. This may reflect the higher efficiency of the JE prM-E proteins for steps involved in replication and spread of virus in these cells, particularly since infection was done at relatively low multiplicity. Efficient replication of the chimeric viruses was also observed in mosquito cells (C6/36). Growth of the chimeric viruses in NB41A3 cells was examined to determine if the difference in neurovirulence of the viruses would be reflected by different replication efficiencies in this cell line, but no correlation could be established. However, a difference was observed in SW-13 cells, suggesting that these cells may serve as a surrogate for investigation of molecular mechanisms governing the difference in mouse neurovirulence which exists between the two chimeric viruses. Taken together, these preliminary analyses have not revealed any unexpected changes in host range due to the combination of YF nonstructural proteins and JE structural glycoproteins.

A striking difference between the neurovirulence levels of the YF/JE-S and YF/JE-N viruses was observed in young adult mice. Previous studies have demonstrated that the JE SA₁₄-14-2 virus is noneurovirulent compared to its JE SA-14 parent (6, 8), but sequence differences have been identified throughout their genomes (18), making it unclear which protein (or proteins) is a principal virulence factor. The data reported here with the chimeric viruses suggest that the prM-E proteins may be major determinants of neurovirulence, since YF/JE-N was more virulent than YF/JE-S. The YF/JE-N virus exhibited partial neuroinvasiveness in these experiments, which is consistent with other studies indicating that the JE E protein contains determinants of mouse neuroinvasiveness (3, 9). Failure to observe neuroinvasiveness in ICR mice presumably indicates that there are strain-specific differences in the susceptibility of young mice to the chimeric YF/JE-N virus. Further studies are needed to fully characterize the neuropathogenic properties of this chimeric virus.

Nucleotide sequence analysis of the clones encoding the YF/JE chimeric viruses suggests that 10 amino acid residues within the E protein and possibly 3 residues within the prM protein account for the difference in neurovirulence. We cannot exclude the potential contribution of mutations outside of the structural region as determinants of this difference; however, some of the predicted differences in the E proteins map to positions which have been implicated as virulence determinants among neuropathogenic flaviviruses (20). In particular, these include position 138, which has been reported to be a critical determinant of JE virulence in the context of the JaOArS982 strain (25). Since the 10 residues which differ in the YF/JE chimeras are distributed through all three structural domains of the E protein as predicted from the TBE model (20), it is possible that the difference in neurovirulence is dependent on more than a single functional property of the E protein. Such steps may include those involved in virus attachment, the acid pH-catalyzed conformational change, and/or membrane fusion events associated with virus entry. In this regard, attenuation of the JE SA-14 strain is believed to involve sequential accumulation of several mutations in the E protein (17). Substitutions at positions 107 (LF), 138 (EK), 176 (IV), and 279 (KM) occurred early and remained stable during subsequent passage of attenuated derivatives of the JE SA-14 strain. Substitutions at positions 177 (TA) and 264 (QH) occurred during passage in PHK cells and were unstable upon subsequent passage in PDK cells, reverting to the original residues (18). Since the YF/JE-S virus was derived from JE virus at the PHK passage level, its E protein contains alanine at position 177 and histidine at position 264, rather than those of the JE SA₁₄-14-2 virus which was produced from the PDK passage (6). This suggests that the four positions 107, 138, 176, and 279 may harbor residues which are critical for determining the neurovirulence phenotype. As a first step toward investigating these possibilities, the genetic basis for the attenuation can now be defined by constructing a series of intertypic YF/JE chimeric viruses containing one or more reversions of the YF/JE-S virus to the sequence of the JE-N strain and testing these engineered viruses for their mouse neurovirulence phenotypes.

The high degree of attenuation of the YF/JE-S virus, demonstrated by lack of virulence even at very high doses given by either the intraperitoneal or intracerebral route, is of interest because it mimics the properties of the attenuated JE SA₁₄-14-2 virus (6). This virus has been used extensively for vaccine production in the People's Republic of China; however, there is evidence that complete protection against JE may require multiple-dose immunization (10, 26). Incorporation of the protective antigens against JE into the YF virus may circumvent problems with replication efficiency that could explain the less than optimal vaccine properties of JE SA₁₄-14-2. This raises the possibility that the YF/JE-S virus could be used as a second-generation live, attenuated vaccine for JE. Because of incomplete knowledge of the mechanisms of protection associated with live, attenuated JE vaccine, this question warrants further investigation by careful assessment of the properties of YF/JE-S as an experimental vaccine in murine and primate models. If this system proves successful, use of engineered YF17D chimeric viruses may have promise in vaccine development for other important flavivirus diseases, including dengue fever and tick-borne encephalitis.