Enzyme-linked immunosorbent assays using novel Japanese encephalitis virus antigen improve the accuracy of clinical diagnosis of flavivirus infections.
ABSTRACT The cross-reactive antibodies induced by flavivirus infections confound serodiagnosis and pathogenesis, especially in secondary infections caused by antigenically closely related yet distinct flaviviruses. The envelope (E) glycoprotein fusion peptide contains immunodominant cross-reactive determinants. Using a recombinant Japanese encephalitis virus (JEV) premembrane and E expression plasmid producing JEV virus-like particles (VLPs), dramatic reductions in cross-reactivity were produced by the G106K-L107D (KD) double-mutant VLP against a panel of flavivirus murine monoclonal antibodies. Human serum panels from patients with recent flavivirus infections were analyzed to compare the accuracy of JEV wild-type (WT) and KD VLPs as serodiagnostic antigens in enzyme-linked immunosorbent assays. Statistical analysis demonstrated significant differences in assay performances for accurate determination of current JEV infections between WT and KD antigens by detecting immunoglobulin M antibodies at a serum dilution of 1:4,000 (likelihood ratios = 2.74 [WT] and 22 [KD]). The application and continued development of cross-reactivity-reduced antigens should improve both flavivirus infection serodiagnosis and estimates of disease burden.
- [Show abstract] [Hide abstract]
ABSTRACT: Dengue fever and dengue hemorrhagic fever are endemic throughout tropical and subtropical countries. Four serotypes of dengue viruses (DENV-1 to DENV-4), each with several genotypes including various subclades, are co-distributed in most endemic areas. Infection-neutralizing and -enhancing antibodies are believed to play protective and pathogenic roles, respectively. Measurement of these functional antibodies against a variety of viral strains is thus important for evaluating coverage and safety of dengue vaccine candidates. Although transportation of live virus materials beyond national borders is increasingly limited, this difficulty may be overcome using biotechnology that enables generation of an antibody-assay antigen equivalent to authentic virus based on viral sequence information. A rapid system to produce flavivirus single-round infectious particles (SRIPs) was recently developed using a Japanese encephalitis virus (JEV) subgenomic replicon plasmid. This system allows production of chimeric SRIPs that have surface proteins of other flaviviruses. In the present study, SRIPs of DENV-1 (D1-SRIPs) were evaluated as an antigen for functional antibody assays. Inclusion of the whole mature capsid gene of JEV into the replicon plasmid provided higher D1-SRIP yields than did its exclusion in cases where a DENV-1 surface-protein-expressing plasmid was used for co-transfection of 293T cells with the replicon plasmid. In an assay to measure the balance between neutralizing and enhancing activities, dose (antibody dilution)-dependent activity curves in dengue-immune human sera or mouse monoclonal antibodies obtained using D1-SRIP antigen were equivalent to those obtained using DENV-1 antigen. Similar results were obtained using additional DENV-2 and DENV-3 systems. In a conventional Vero-cell neutralization test, a significant correlation was shown between antibody titers obtained using D1-SRIP and DENV-1 antigens. These results demonstrate the utility of D1-SRIPs as an alternative antigen to authentic DENV-1 in functional antibody assays. SRIP antigens may contribute to dengue vaccine candidate evaluation, understanding of dengue pathogenesis, and development of serodiagnostic systems.Vaccine 06/2014; · 3.49 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: West Nile Virus (WNV) is an emerging mosquito-transmitted flavivirus that continues to spread and cause disease throughout several parts of the world, including Europe and the Americas. Specific diagnosis of WNV infections using current serological testing is complicated by the high degree of cross-reactivity between antibodies against other clinically relevant flaviviruses, including dengue, tick-borne encephalitis (TBEV), Japanese encephalitis (JEV), and yellow fever (YFV) viruses. Cross-reactivity is particularly problematic in areas where different flaviviruses co-circulate or in populations that have been immunized with vaccines against TBEV, JEV, or YFV. The majority of cross-reactive antibodies against the immunodominant flavivirus envelope (E) protein target a conserved epitope in the fusion loop at the distal end of domain II.BMC Infectious Diseases 05/2014; 14(1):246. · 2.56 Impact Factor
- [Show abstract] [Hide abstract]
ABSTRACT: Recombinant virus-like particles (rVLPs) of flaviviruses are non-infectious particles released from cells expressing the envelope glycoproteins prM and E. Dengue virus rVLPs are recognized as a potential vaccine candidate, but large scale production of these particles is hindered by low yields and the occurrence of cytopathic effects. In an approach to improve the yield of rVLPs from transfected insect cells, several components of a dengue serotype 2 virus prM+E expression cassette were modified and the effect of these modifications was assessed during transient expression. Enhancement of extracellular rVLP levels by simultaneous substitutions of the prM signal peptide and the stem-anchor region of E with homologous cellular and viral counterparts, respectively, was further augmented by codon optimization. Extensive formation of multinucleated cells following transfection with the codon-optimized expression cassette was abrogated by introducing an E fusion loop mutation. This mutation also helped restore the extracellular E levels affected negatively by alteration of a charged residue at the pr-M junction, which was intended to promote maturation of rVLPs during export. Optimized expression cassettes generated in this multiple add-on modification approach should be useful in the generation of stably expressing clones and production of dengue virus rVLPs for immunogenicity studies.Journal of virological methods 05/2014; · 2.13 Impact Factor
CLINICAL AND VACCINE IMMUNOLOGY, May 2008, p. 825–835
Copyright © 2008, American Society for Microbiology. All Rights Reserved.
Vol. 15, No. 5
Enzyme-Linked Immunosorbent Assays Using Novel Japanese
Encephalitis Virus Antigen Improve the Accuracy of
Clinical Diagnosis of Flavivirus Infections?
Shyan-Song Chiou,1Wayne D. Crill,2Li-Kuang Chen,3and Gwong-Jen J. Chang2*
Graduate Institute of Veterinary Public Health, College of Veterinary Medicine, National Chung Hsing University, Taichung, Taiwan, Republic of
China1; Arboviral Diseases Branch, Division of Vector-Borne Infectious Diseases, National Center for Zoonotic, Vector-Borne,
and Enteric Diseases, Centers for Disease Control and Prevention, Public Health Service, U.S. Department of Health and
Human Services, Fort Collins, Colorado2; and College of Medicine, Tzu Chi University, Hualien, Taiwan, Republic of China3
Received 4 January 2008/Returned for modification 4 February 2008/Accepted 21 February 2008
The cross-reactive antibodies induced by flavivirus infections confound serodiagnosis and pathogenesis,
especially in secondary infections caused by antigenically closely related yet distinct flaviviruses. The envelope
(E) glycoprotein fusion peptide contains immunodominant cross-reactive determinants. Using a recombinant
Japanese encephalitis virus (JEV) premembrane and E expression plasmid producing JEV virus-like particles
(VLPs), dramatic reductions in cross-reactivity were produced by the G106K-L107D (KD) double-mutant VLP
against a panel of flavivirus murine monoclonal antibodies. Human serum panels from patients with recent
flavivirus infections were analyzed to compare the accuracy of JEV wild-type (WT) and KD VLPs as serodi-
agnostic antigens in enzyme-linked immunosorbent assays. Statistical analysis demonstrated significant dif-
ferences in assay performances for accurate determination of current JEV infections between WT and KD
antigens by detecting immunoglobulin M antibodies at a serum dilution of 1:4,000 (likelihood ratios ? 2.74
[WT] and 22 [KD]). The application and continued development of cross-reactivity-reduced antigens should
improve both flavivirus infection serodiagnosis and estimates of disease burden.
Japanese encephalitis virus (JEV), a member of the genus
Flavivirus in the family Flaviviridae, is the leading cause of
endemic/epidemic viral encephalitis in Asia, including India,
Thailand, Vietnam, Singapore, the Philippines, Taiwan, China,
Korea, and Japan (40). It is also one of several mosquito-borne
flaviviruses, in addition to four serotypes of dengue virus
(DENV-1 to -4), that have experienced emergence and/or re-
emergence throughout the world, especially in the tropical
regions (22, 24). Sequential infection by multiple cocirculating
flaviviruses in the affected population confounds serodiagnosis
(20), disease burden estimation (23), and the impact on patho-
Flavivirus infections elicit protective antibody responses pri-
marily against the envelope (E) glycoprotein (20). The E pro-
tein contains three structural and functional domains. E
domain I (EDI) is an eight-stranded ?-barrel; it contains two
large insertion loops forming the elongate dimerization EDII
and containing the highly conserved internal fusion peptide.
EDIII has an immunoglobulin (Ig)-like structure and contains
the primary receptor-binding motifs (16, 29). Murine mono-
clonal antibody (MAb) studies have demonstrated that EDI
contains predominately type-specific nonneutralizing (non-Nt)
epitopes, EDII contains cross-reactive epitopes eliciting both
Nt and non-Nt antibodies, and EDIII contains the majority of
the type-specific Nt epitopes (6, 31–34, 37, 38).
Diagnostic enzyme-linked immunosorbent assays (ELISAs)
are common, relatively quick, and efficient assays for clinical
diagnosis, traditionally requiring the use of suckling mouse
brain-grown (SMB) antigen and more recently utilizing non-
infectious recombinant virus-like particle (VLP) antigen. Stud-
ies have shown that VLP antigens have higher performance
accuracy than SMB antigens when used in ELISA for diagnos-
ing flaviviral infections (8, 13, 14, 28). However, both SMB and
VLP antigens contain wild-type (WT) E proteins that exhibit
the same cross-reactive epitopes as the virus responsible for
the infection. The amino acids located in the highly conserved
E glycoprotein fusion peptide, in particular Gly104, Gly106,
and Leu107, have been identified as important flavivirus cross-
reactive epitope determinants (6, 7, 37, 39). Thus, it is possible
to develop cross-reactivity-reduced antigens by introducing
substitutions for amino acids within the fusion peptide, thereby
improving virus-specific diagnostic assays (7, 39). Recently,
fusion peptide mutant VLPs for both St. Louis encephalitis
virus (SLEV) and West Nile virus (WNV) demonstrated dra-
matic reductions in the observed cross-reactivity of immuno-
globulin M capture (MAC) ELISA, producing more accurate
differentiation of both current and past WNV and SLEV in-
Here, we present results from mutagenesis in the fusion
peptide region of the JEV E protein to identify and ablate
cross-reactive E protein epitopes and utilize these mutant
VLPs as improved serodiagnostic antigens. The JEV G106K/
L107D (KD) VLP exhibited the most dramatic reductions in
cross-reactivity of the JEV fusion peptide mutants. Thus, the
JEV-KD and the previously described cross-reactivity-reduced
WNV G106R/L107H (RH) VLP (30) were used as serodiag-
nostic antigens to test a diverse group of flavivirus-infected
patients’ sera and to compare their performances for the de-
* Corresponding author. Mailing address: Division of Vector-Borne
Infectious Diseases, 3150 Rampart Road, CDC-Foothills Campus,
Fort Collins, CO 80521. Phone: (970) 221-6497. Fax: (970) 226-3599.
?Published ahead of print on 12 March 2008.
tection of virus-specific IgM and IgG in ELISA with those of
WT JEV and WNV VLP antigens.
MATERIALS AND METHODS
Cell culture, virus strain, and recombinant plasmid. COS-1 cells (ATCC CRL
1650; American Type Culture Collection, Manassas, VA) were grown at 37°C
with 5% CO2in Dulbecco’s modified Eagle’s minimal essential medium (Gibco,
Grand Island, NY) supplemented with 10% heat-inactivated fetal bovine serum
(HyClone Laboratories, Inc., Logan, UT), 110 mg of sodium pyruvate/liter, 0.1
mM nonessential amino acids, 2 mM L-glutamine, 20 ml of 7.5% NaHCO3/liter,
100 U of penicillin/ml, and 100 ?g of streptomycin/ml.
We used the recombinant expression plasmid pVJE as the template DNA both
for site-directed mutagenesis and for the transient expression of WT JEV re-
combinant antigen (see below). pVJE, derived from the pCBJE plasmid (4, 5,
14), includes the human cytomegalovirus early gene promoter, JEV signal se-
quence, JEV premembrane/membrane (prM/M) and E gene region in its en-
tirety, and bovine growth hormone poly(A) signal. JEV strain SA14 was used as
a template for amplification of JEV prM and E genes. The cloning procedures
were described in detail in a previous publication (5). The ampicillin resistance
gene in pCBJE was replaced by a kanamycin resistance gene derived from the
pVAX plasmid (Invitrogen, Carlsbad, CA) to generate pVJE. In addition, the
chimeric human ?-globin gene intron sequence derived from the pCI expression
vector (Promega, Madison, WI) was PCR amplified and inserted between nu-
cleotides 1240 and 1241 of the E gene to generate pVJEi. The intron insertion
increased the plasmid yield in the Escherichia coli host and enhanced VLP
secretion in transiently transformed COS-1 cells (G. J. Chang and J. Kim,
Site-directed mutagenesis. A homology model for the JEV E protein was
produced using the published atomic coordinates for DENV-2 and WNV and the
Swiss-model workspace (http://swissmodel.expasy.org/workspace/). We focused
on the amino acid substitutions at G106 and L107 of E protein based on a
number of criteria, as previously described (6, 7, 39). Stability calculations (??G)
were determined for all possible substitutions at residues 106 and 107 using the
PoPMuSiC server (http://babylone.ulb.ac.be/popmusic/) and Protein Data Bank
coordinates from the JEV homology model. Four and five substitutions were
selected at positions 106 and 107 to represent the diversity of biochemical and
structural properties of amino acid side chains (e.g., basic, acidic, polar, nonpo-
lar, small, and large). Individual substitutions maximizing stability (lower ??G
values) were selected from within each side chain class.
Site-specific mutations were introduced into the JEV E gene of the pVJE
plasmid using a QuikChange multisite-directed mutagenesis kit (Stratagene, La
Jolla, CA), according to the manufacturer’s recommended protocols. The se-
quences of the mutagenic primers used for all constructs are listed in Table 1.
Four or five colonies from each mutagenic PCR transformation were selected
and grown in 5-ml Luria-Bertani broth cultures, miniprepped, and sequenced
across the intended substitution to identify the correct mutant clone(s). The
transcription units, including prM/M and E gene regions and the transcriptional
and translational regulatory elements, of all purified plasmids were sequenced in
their entirety upon identification of the correct substitution(s). Automated DNA
sequencing was performed with an ABI 3130xl genetic-analysis system (Applied
Biosystems, Foster City, CA), and sequences were analyzed with Lasergene
software (DNAStar, Madison, WI).
Electroporation of tissue culture cells with plasmid DNA. For transformation,
COS-1 cells were grown to 90 to 100% confluence in 150-cm2culture flasks,
trypsinized, and resuspended in ice-cold phosphate-buffered saline (PBS) to a
final density of 1.5 ? 107cells/ml. For each reaction, 0.5 ml of this cell suspension
was electroporated with 20 ?g of plasmid DNA in a 0.4-cm-electrode-gap cuvette
with a Bio-Rad Gene Pulser II (Bio-Rad Laboratories, Hercules, CA) set at 250
V and 975 ?F. Two electroporation reaction mixtures were seeded onto a single
150-cm2culture flask containing 50 ml of growth medium and allowed to recover
at 37°C overnight. The tissue culture flasks were continuously maintained at 37°C
or at 28°C for an additional 1 to 4 days. We observed that substitutions at JEV
E Gly104, which were previously found not to secrete at 37°C in other flavivirus
systems (6, 7, 39), secreted to sufficient levels for MAb analysis when the trans-
formants were seeded into COS-1 cells and maintained at 28°C. Tissue culture
medium was harvested on day 2 (37°C) or day 5 (28°C) following electroporation,
TABLE 1. Nucleotide sequences of mutagenic PCR primers for JEV VLPs
Mutagenic primer sequence (5?–3?)a
aMutated nucleotides are shown in boldface.
bStandardized measurements of VLP secretion from transiently transformed COS-1 cells recovered at 37°C presented as percentages of the wild-type plasmid VLP
secretion (arbitrarily set at 100%). The G104H substitution prevented VLP secretion from transformed COS-1 cells at 37°C, but not at 28°C.
826CHIOU ET AL.CLIN. VACCINE IMMUNOL.
clarified by centrifugation at 10,000 rpm for 30 min at 4°C in a Sorval F-16/250
rotor (Beckman Coulter), and stored at 4°C for further analysis.
Human serum. Serum specimens were obtained from the Diagnostic and
Reference Laboratory, Arboviral Diseases Branch, Division of Vector-Borne
Infectious Diseases, Centers for Disease Control and Prevention. Panels were
assembled by selecting serum specimens collected from 1999 to 2003 having Nt
antibody titers to WNV (n ? 21), SLEV (n ? 6), or alphaviruses (n ? 12), as
determined by the 90% plaque reduction neutralization test. The serum panels
with evidence of DENV (n ? 24) or JEV (n ? 16) infection were assembled from
Taiwanese residents and provided by the Center for Disease Control—Taiwan.
The dengue serotype responsible for the most recent infection was defined by
virus isolation and/or virus-specific reverse transcriptase PCR, and the JEV
infection status was determined by IgM and IgG ELISAs (35).
MAb panel. When selecting MAbs for use in antigen characterization, we
specifically chose a variety of group-, subgroup-, complex-, and subcomplex-
cross-reactive MAbs that had been raised against a diverse assortment of flavi-
viruses (L.-K. Chen, unpublished results; 9, 17, 18, 34). MAbs 4G2, 23-1, 23-2,
and 6B6C-1 are flavivirus group cross-reactive (recognizing viruses from all
major pathogenic serocomplexes of flaviviruses) and non-Nt to moderately Nt.
MAbs 2B5B-3 and 5-2 are subgroup-reactive antibodies recognizing the JEV
complex and yellow fever virus, and only JEV and DENV-1 and -2, respectively.
MAbs 16, 6B4A-10, 1B5D-1, 109, and 203 exhibit various levels of cross-reac-
tivity with viruses within the JEV complex and are non-Nt. J3 14 H5-2, 112, and
503 are the JEV-specific MAbs used in this study (17).
Antigen characterization. Antigen capture ELISA (Ag-ELISA) was per-
formed to determine VLP secretion from plasmid-transformed cells and to
determine reductions in MAb reactivity to mutant VLP antigens as previously
described (6). Briefly, the inner 60 wells of an Immulon II HB flat-bottom 96-well
plate (Dynatech Industries, Inc., Chantilly, VA) were coated with polyclonal
rabbit-anti-JEV antibody in 50 ?l of coating buffer (0.015 M sodium carbonate,
0.035 M sodium bicarbonate, pH 9.6) and incubated overnight at 4°C. The wells
were blocked with 300 ?l of Start Block (PBS) blocking buffer (Pierce, Rockford,
IL) according to the manufacturer’s recommended procedure. Secreted WT and
mutant antigens were titered in PBS, captured with 1 h of incubation at 37°C, and
detected with anti-JEV murine hyperimmune ascitic fluid (MHIAF) at a 1:8,000
dilution in PBS with 5% milk. Anti-JEV MHIAF was detected with horseradish
peroxidase-conjugated goat anti-mouse HIAF at a 1:5,000 dilution in PBS con-
taining 5% milk. Bound conjugate was detected by adding 75 ?l of the 3,3?5,5?-
tetramethylbenzidine (Neogen Corp., Lexington, KY) substrate and incubating
the mixture at room temperature for 10 min. The substrate reaction was stopped
with 50 ?l of 2 N H2SO4, and the reactions were measured at an A450with an EL
312e Bio-Kinetics microplate reader (Bio-Tek Instruments, Inc., Winooski, VT).
Secreted antigen concentrations were standardized for MAb screening by
selecting the antigen concentration producing an optical density of ?1.0 with the
polyclonal MHIAF. The panel of MAbs was used to determine the end point
MAb reactivities of mutated and WT antigens in Ag-ELISA (Tables 2 and 3).
MAb affinity reductions were determined utilizing the same Ag-ELISA described
above, except that the standardized Ag concentrations determined above were
used to determine the end point titer of the MAb.
ELISA protocols. JEV VLPs and normal COS-1 cell culture antigen were
prepared as described above, and WNV VLPs were prepared as described in a
previous publication (30). Antigens were independently titrated against JEV or
WNV positive control serum samples with a twofold dilution series and stan-
dardized by selecting a dilution that yielded an absorbance of ?1.0 at 450 nm
(A450). The MAC-ELISA was performed as described previously (14) with some
modifications for detecting the presence of virus-specific IgM in patient’s serum
panels with the VLPs. Briefly, the inner 60 wells of Immulon II HB flat-bottom
96-well plates (Dynatech Industries, Inc., Chantilly, VA) were coated overnight
at 4°C in a humidified chamber with 75 ?l of goat anti-human IgM (Kirkegaard
& Perry Laboratories, Gaithersburg, MD) diluted 1:2,000 in coating buffer (0.015
M sodium carbonate, 0.035 M sodium bicarbonate, pH 9.6). The wells were
blocked with 300 ?l of Start Block (PBS) blocking buffer (Pierce, Rockford, IL)
according to the manufacturer’s recommended procedure. Patient sera and pos-
itive and negative antibody controls were diluted appropriately in wash buffer
(PBS with 0.05% Tween 20), added to wells (50 ?l/well), and incubated at 37°C
for 1 h in a humidified chamber. Test positive and negative human control sera
were diluted 1:400 or 1:4,000. Positive and negative control antigens were tested
with each patient serum sample in triplicate by diluting them appropriately in
wash buffer and adding 50 ?l to appropriate wells for incubation at 4°C overnight
in a humidified chamber. WT and mutant JEV or WNV antigens were added and
incubated at 37°C for 1 h in a humidified chamber. After being washed, the
captured JEV or WNV antigens were detected with anti-JEV or anti-WNV
MHIAF, respectively, at a 1:8,000 dilution. Anti-JEV and anti-WNV MHIAFs
were detected with horseradish peroxidase-conjugated goat anti-mouse HIAF
used at a 1:5,000 dilution. Bound conjugate was detected by adding 75 ?l of the
3,3?5,5?-tetramethylbenzidine (Neogen Corp., Lexington, KY) substrate and in-
cubating the mixture at room temperature for 10 min. The substrate reaction was
stopped with 50 ?l of 2 N H2SO4, and the reactions were measured at A450with
an EL 312e Bio-Kinetics microplate reader (Bio-Tek Instruments, Inc., Wi-
For detection of virus-specific IgG in patients’ serum samples using VLPs, the
IgG capture ELISA (GAC-ELISA) was performed as described above for MAC-
ELISA with the exception that the plates were coated with anti-human IgG
(Kirkegaard & Perry Laboratories, Gaithersburg, MD) diluted 1:2,000 in coating
TABLE 2. Comparison of antibody reactivitiesafor JEV WT and mutant VLPs
J3 14 H5-2
aAg-ELISA was used to determine the reciprocal end point titers (log10) for secreted VLP antigens. The numbers shown in boldface are the end point titers
decreased below that of the WT by at least two threefold dilutions; the numbers in boldface italic are the titers increased over that of the WT by at least two threefold
bGroup, recognized by all flaviviruses examined; subgroup, recognized by more than one serocomplex; complex, recognized by all members of JEV complex;
subcomplex, not recognized by all members of the serocomplex; type, recognized by JEV only.
cVirus against which antibody was raised.
dThe G104H substitution prevented VLP secretion from transformed COS-1 cells at 37°C, but not at 28°C.
eThe L107A mutation was obtained unintentionally due to misincorporation of sequence during the mutagenesis procedure.
VOL. 15, 2008 ELISAs USING A NOVEL JEV ANTIGEN 827
Test validation and calculation of P/N absorbance ratio values. Test validation
and positive/negative (P/N) ratio values were determined according to the pro-
cedure of Martin et al. (25). Briefly, internal positive and negative serum controls
were included in each 96-well plate for test validation. For a testing plate to be
considered valid, the average A450for the positive serum control reacted with
positive viral antigen had to be at least two times greater than the average A450
for the same positive serum control reacted with the negative tissue culture fluid
antigen. Each patient serum sample was validated in the same manner. This
verified that significant A450values against viral antigens were not due to non-
specific binding of serum antibodies to tissue culture fluid components.
Positive values for each specimen were determined as the average A450for the
patient serum sample reacted with positive viral antigen. Negative values were
determined for individual 96-well plates as the average A450for the normal
human serum control reacted with the positive viral antigen. A specimen was
classified as a validated positive sample if it had a P/N ratio of ?3.0.
Statistical analysis. A plot of the sensitivity versus the false-positive rate (1 ?
specificity), the receiver operator characteristic (ROC) curve analysis, was ap-
plied to determine the discriminatory accuracies of the tests using WT or cross-
reactivity-reduced JEV VLPs using GraphPad Prism version 4.00 for Windows
(GraphPad Software, San Diego, CA). The area under the ROC curve (AUC)
was used to evaluate the performance of a diagnostic test to determine the
evidence of infection. The comparative ROC was used to calculate the signifi-
cance level and to compare the paired-assay performance according to the
method described by Hanley and McNeil (11).
A P/N ratio of ?3 or ?3 for a given specimen was classified as negative or
positive, respectively. A two-by-two contingency table was prepared that catego-
rized four quadrants as true positive, true negative, false positive, and false
negative. These transformed data were applied to calculate the sensitivity, spec-
ificity, and positive likelihood ratio.
Development of cross-reactivity-reduced JEV VLP antigens.
A total of one, four, and six different amino acid substitutions
were introduced at each of the JEV E protein fusion peptide
residues Gly104, Gly106, and Leu107, respectively, into the
WT JEV expression plasmid (Tables 1 and 2). VLP secretion
levels from transiently transformed cells with G104 substitu-
tions in previous DENV-2, WNV, and SLEV studies were
below the detection level and thus excluded from MAb-map-
ping studies (6, 7, 39). Unexpectedly, we were able to detect
the secretion of JEV G104H VLPs if the plasmid-transformed
cells were initially recovered at 37°C and further incubated at
Ag-ELISA was used to determine VLP secretion levels and
to standardize VLP concentrations for MAb mapping. The
initial MAb screening identified multiple E protein residues
which, when mutated, resulted in altered recognition by MAbs
of various levels of cross-reactivity (Table 2). The G104H sub-
stitution dramatically reduced the reactivities of two of four
flavivirus group-reactive MAbs (4G2 and 6B6C-1), one sub-
group-reactive MAb (5-2), and one complex-cross-reactive
MAb (16). The reactivities of group (6B6C-1, 4G2, and 23-2)-,
subgroup (5-2 and 2B5B-3)-, and subcomplex (1B5D-1)-reac-
tive MAbs were reduced by G106Q, G106K, G106V, or G106D
mutant VLPs. All of the L107 substitutions reduced the reac-
tivities to the flavivirus group-reactive MAbs, with the excep-
tion of 6B6C-1. Only L107F dramatically reduced the reactivity
of 6B6C-1. The group-reactive MAb 4G2 exhibited dramati-
cally reduced reactivity for all substitutions introduced at
G104, G106, or L107, as was the case for the subgroup-reactive
MAb 5-2. The only L107 substitution that significantly reduced
JEV complex- and subcomplex-cross-reactive MAbs was
L107F. The reactivity of the JEV type-specific MAb (J3 14
H5-2), although not strongly reactive for the JEV-WT VLP,
was not negatively affected by either the G106 or L107 substi-
tution, but it was reduced by the G104H substitution. Since the
G104H mutation reduced VLP secretion and JEV type-specific
MAb reactivity, this mutation was eliminated from further
Based on MAb-mapping results for the individual fusion
peptide mutants (Table 2), substitutions at Gly106 (K, V, or D)
and Leu107 (D, R, or F) were combined into nine different
double-mutant constructs in an effort to maximize both cross-
reactivity reductions and antigen secretion (Table 3). The com-
binations KD, G106V/L107R (VR), and G106D/L107D (DD)
produced the most dramatic reductions in cross-reactive MAb
reactivity: these constructs either lost the ability to be detected
or were severely reduced in recognition by all of the group-,
subgroup-, and subcomplex-cross-reactive MAbs yet exhibited
only minor reductions for the complex-reactive MAbs
6B4A-10 and 16. As expected, there was no reduction in the
TABLE 3. Effect of antibody reactivityaon JEV double-amino acid mutant VLPs
GroupSubgroup Complex SubcomplexType
J3 14 H5-2
aAg-ELISA was used to determine the reciprocal end point titers (log10) for secreted VLP antigens. The numbers shown in boldface are the end point titers
decreased below that of the WT by at least two threefold dilutions.
bGroup, recognized by all flaviviruses examined; subgroup, recognized by more than one serocomplex; complex, recognized by all members of JEV complex;
subcomplex, not recognized by all members of the serocomplex; type, recognized by JEV only.
cVirus against which antibody was raised.
828CHIOU ET AL.CLIN. VACCINE IMMUNOL.
reactivity of type-specific MAb J3 14 H5-2 relative to JEV WT
(Table 3). These reactivity reductions were predictable in that
the double mutants retained any reactivity reduction observed
in either of the two corresponding single mutants. However, we
also observed nonadditive phenomena in the L107F-contain-
ing double mutants. For example, G106D-L107F restored var-
ious degrees of WT reactivities against 6B6C-1, 4G2, 6B4A-10,
and 1B5D-1 that were observed in their corresponding single-
Selection of cross-reactivity-reduced JEV VLP antigens for
serodiagnosis. Because the long-term application of this work
is to develop novel serodiagnostic antigens and VLP secretion
from plasmid-transformed cells is critical for efficient antigen
production, we focused on substitutions that did not interfere
with, or that enhanced, VLP secretion relative to that of the
WT plasmid-transformed cells. Three double-mutant JEV
VLPs, KD, VR, and DD, that showed the most dramatic re-
ductions in reactivity with cross-reactive MAbs were selected
to compare their performances as ELISA antigens against a
preliminary panel of JEV- and DENV-confirmed human sera.
Recombinant JEV VLP antigens for the WT, KD, VR, and
DD were employed in MAC-ELISA to determine their differ-
ent abilities to detect virus-specific IgM, as well as cross-reac-
tive IgM antibodies, in JEV (n ? 6)- and DENV (n ? 6)-
infected human sera. The assay results with a serum dilution at
1:400, expressed as the P/N ratio using mutant or WT VLPs,
are summarized in Fig. 1. MAC-ELISA employing the
JEV-WT antigen detected IgM antibody in all six JEV-infected
sera and also positively detected cross-reactive IgM antibody in
four of six DENV sera (Fig. 1A and B). None of the three
mutant JEV antigens detected cross-reactive antibodies in the
DENV serum panel (Fig. 1B), but all of the mutant antigens
positively detected IgM antibody in the JEV serum panel (Fig.
1A). As expected, all the P/N values employing mutant JEV
antigens were lower than those with WT antigen in either the
JEV or DENV serum panels. In the JEV serum panel, the P/N
values were very similar for all three double mutants; however,
on average, the P/N values were highest with the KD antigen.
These results suggested that the three G106 and L107 double
amino acid mutants eliminated detection of cross-reactive IgM
antibody in non-JEV DENV patient sera but maintained the
capacity to detect JEV-specific IgM antibodies in JEV patient
sera. In addition, the JEV-KD plasmid-transformed cells main-
tained the same high VLP secretion as the WT plasmid-trans-
formed cells, unlike the other two G106/L107 constructs, which
exhibited reduced VLP secretion levels (Table 1). Before pro-
ceeding with the serum screening, we decided to examine the
reactivities of the JEV-KD antigen against a very limited sup-
ply of two JEV-specific Nt MAbs (112 and 503) (17). As ex-
pected, the KD antigen maintained the same high-level reac-
tivity as the WT against MAbs 112 and 503 (data not shown).
For these reasons, the KD VLP was selected as the single
cross-reactivity-reduced JEV antigen analyzed in IgM and IgG
Detection of JEV and WNV antibodies by MAC- and GAC-
ELISA. A total of 79 arbovirus-infected human serum speci-
mens were screened with four different VLP antigens in this
study. JEV and DENV are the most medically relevant flavi-
viruses in Asia, and WNV, SLEV, and Powassan virus
(POWV) are the medically relevant flaviviruses in North
America. Additionally, there is significant geographic overlap
between WNV and JEV in Southeast Asia, the Indian subcon-
tinent, and Oceania. In order to better understand the poten-
tial complications in serodiagnosis due to antibody cross-reac-
tivity, serum specimens were randomly coded and blind tested
using JEV-WT, JEV-KD, WNV-WT, and WNV-RH antigens
in MAC- and GAC-ELISA to test for the presence of IgM and
IgG antibodies. Sera were tested concurrently for all four an-
tigens at dilutions of 1:400 and 1:4,000. We determined a priori
that a P/N ratio of ?3.0 indicated the positive presence of
serum antibody. This value has become generally accepted and
has worked well for us in the past (30).
The JEV panel consisted of 16 presumptive JEV-infected
acute patient serum specimens. The JEV-WT antigen detected
anti-JEV IgM in 16/16 of these presumptive positive samples at
either a 1:400 or 1:4,000 serum dilution (Table 4) (P/N ratios ?
3.0; range, 16.8 to 38.9, and average, 31.4 for 1:400 serum
dilution; range, 5.7 to 25.3, and average, 15.0 for 1:4,000 serum
dilution). Replacing JEV-WT with the JEV-KD antigen in the
MAC-ELISA also detected 16/16 positive samples (P/N ratios ?
3.0; range, 10.3 to 31.7, and average, 23.4 for 1:400 serum
dilution; range, 3.4 to 20.9, and average, 10.1 for 1:4,000 serum
dilution). The JEV-WT antigen detected anti-JEV IgG in
15/16 of these presumptive JEV-infected sera at a 1:400 or
1:4,000 dilution (P/N ratios ? 3.0; range, 4.1 to 26.5, and
average, 9.4 for 1:400 serum dilution; range, 3.3 to 23.6, and
average, 9.0 for 1:4,000 serum dilution). Replacing JEV-WT
FIG. 1. Determination of cross-reactivity-reduced JEV VLPs for
serodiagnosis. Six each of JEV-confirmed (A) and DENV-confirmed
(B) serum samples were tested at 1:400 dilution by ELISA using WT
and three G106 and L107 double-amino-acid-mutated JEV VLPs, DD,
KD, and VR. The bold lines represent the P/N ratio cutoff value of 3.0
used for the positive detection of serum Ig.
VOL. 15, 2008 ELISAs USING A NOVEL JEV ANTIGEN829
with the JEV-KD antigen in the GAC-ELISA detected 15/16
and 14/16 positive samples at 1:400 and 1:4,000 serum dilutions,
respectively (P/N ratios ? 3.0; range, 3.2 to 19.3, and average,
7.2 for 1:400 serum dilution; range, 3.2 to 19.8, and average, 7.0
for 1:4,000 serum dilution).
The WNV panel consisted of 21 presumptive WNV-infected
sera from North America. The WNV-WT antigen detected
anti-WNV IgM in 21/21 of these presumptive positive samples
at a 1:400 or 1:4,000 dilution (Table 4) (P/N ratios ? 3.0; range,
19.37 to 45.11, and average, 30.19 for 1:400 serum dilution;
range, 6.73 to 30.6, and average, 23.93 for 1:4,000 serum dilu-
tion). Replacing WNV-WT with the WNV-RH antigen in the
MAC-ELISA also detected 21/21 positive samples (P/N ratios ?
3.0; range, 16.88 to 36.83, and average, 24.7 for 1:400 serum
dilution; range, 4.68 to 24.52, and average, 19.27 for 1:4,000
serum dilution). The WNV-WT antigen detected anti-WNV
IgG in 19/21 or 18/21 of these presumptive WNV-infected sera
at a 1:400 or 1:4,000 dilution (P/N ratios ? 3.0; range, 3.86 to
34.27, and average, 18.82 for 1:400 serum dilution; range, 8.75
to 27.65, and average, 20.92 for 1:4,000 serum dilution). Re-
placing WNV-WT with the WNV-RH antigen in the GAC-
ELISA similarly detected 19/21 and 18/21 positive samples at
1:400 and 1:4,000 serum dilutions, respectively (P/N ratios ?
3.0; range, 3.1 to 22.5, and average, 16.02 for 1:400 serum
dilution; range, 4.78 to 26.62, and average, 17.51 for 1:4,000
Use of JEV-WT antigen in MAC-ELISA with non-JEV pa-
tient serum panels (Table 4; DENV, 24; WNV, 21; SLEV, 6;
other flavivirus, 9; nonflavivirus, 3) detected JEV-cross-reac-
tive IgM antibodies in 10/24 (DENV panel; 41.7%), 20/21
(WNV panel; 95.2%), 5/6 (SLEV panel; 83.3%), 2/9 (other
flavivirus panel; 22.2%), and 0/3 (nonflavivirus panel; 0%) sera
at a 1:400 serum dilution. For the same serum panels, when
tested at a 1:4,000 dilution using the JEV-WT antigen in MAC-
ELISA, the number of JEV-positive sera was reduced to 5/24
in the DENV panel (20.8%), 12/21 in the WNV panel (57.1%),
2/6 in the SLEV panel (33.3%), 0/9 in the other flavivirus
panel, and 0/3 in the nonflavivirus panel. As expected, these
false-positive detection rates for JEV-cross-reactive IgM anti-
bodies were reduced further when the JEV-KD antigen and a
serum dilution of 1:4,000 were used in the assay (Table 4).
However, 1 of 24 DENV-positive and 2 of 21 WNV-positive
serum specimens remained JEV-IgM positive, respectively. A
similar serodiagnostic improvement was observed when
WNV-WT antigen was replaced with WNV-RH antigen in the
MAC-ELISA and the non-WNV serum panels were tested at
a 1:4,000 dilution. Only 1/16 JEV, 1/24 DENV, 2/6 SLEV, 0/9
other flavivirus, and 0/3 nonflavivirus serum specimens, diluted
1:4,000, remained IgM positive with WNV-RH antigen (Table 4).
Similar reductions in the detection of cross-reactive antibod-
ies were less pronounced when the JEV-KD or WNV-RH
antigens were used in GAC-ELISA for non-JEV or non-
WNV panels at a 1:4,000 serum dilution (Table 4). The
JEV-KD antigen maintained positive P/N ratios for IgG in
18/24 DENV-infected and 3/21 WNV-infected patient sera.
WNV-RH antigen detected cross-reactive IgG antibodies in
4/16 and 20/24 JEV- and DENV-infected patient sera, respec-
tively. The DENV-infected patient serum specimens, provided
by the Center for Disease Control—Taiwan, were obtained
from the Taiwanese population, and the currently infecting
DENV was determined by virus isolation and/or virus-specific
nucleic acid detection and MAC-ELISA. Taiwan is in a JEV
endemic area, and mandatory nationwide JEV vaccination has
been implemented since the late 1960s (41). Thus, the pres-
ence of highly cross-reactive IgG antibodies against JEV-KD
and WNV-RH is consistent with the observation that some of
the DENV-infected sera could be from patients previously
vaccinated against or exposed to JEV. Acute DENV infection
TABLE 4. Summary of MAC- and GAC-ELISA-positive serum results grouped by infecting virus
No. of samples in which Ig was detected
JEV VLP WNV VLP
1:400 1:4,0001:4001:4,000 1:400 1:4,0001:4001:4,000
WTKDWT KDWT KDWTKD WTRHWT RHWT RH WTRH
JEVTaiwan-CDC 1616 1616 1615 1515 14161381144144
DENV Taiwan-CDC 2410551 2219221820891 242224 20
WNVU.S. CDC 21 2015122 10463 212121 21 191918 18
Total 79 53 39 351951 414738 705042 2568 496545
aWEEV, Western equine encephalitis virus; EEEV, Eastern equine encephalitis virus; LACV, La Crosse virus.
830CHIOU ET AL.CLIN. VACCINE IMMUNOL.
in JEV-immune individuals would be expected to increase the
concentration and/or the relative avidity of cross-reactive IgG
Results from the most cross-reactive sera, which were IgM
positive at 1:4,000 against WNV-WT (8 of 16 in the JEV
panel), JEV-WT (12 of 21 in the WNV panel), and WNV-WT
or JEV-WT (9 of 24 in the DENV panel), are summarized in
Table 5. Two SLEV patient specimens, although JEV-KD neg-
ative, remained positive against WNV-RH in the MAC-ELISA
and were classified as primary WNV and acute SLEV infec-
tions in a previous study (30).
Replacing the WT antigens with JEV-KD or WVN-RH an-
tigens in MAC-ELISA at a 1:4,000 serum dilution resolved the
IgM cross-reactivity and produced more accurate disease clas-
sifications in 7 of 8 specimens in the JEV panel and 10 of 12
specimens in the WNV panel. The remaining apparently false-
positive specimen, number 2 in the JEV panel, had a much
higher P/N value with JEV-KD than with WNV-RH antigen,
for which it was barely positive (Table 5) (20.92 versus 3.2,
respectively). Similarly, specimens numbers 6 and 21 in the
WNV panel had higher P/N values with WNV-RH than with
JEV-KD antigen (Table 5) (23.26 versus 3.01 and 4.68 versus
3.78, respectively). As in previously published assays, the
higher P/N ratio appears to be indicative of currently infecting
virus (30); thus, the viruses responsible for the current infec-
tion could be accurately identified for these three specimens.
Use of JEV-KD or WNV-RH in MAC-ELISA with the
DENV-infected patient sera at a 1:4,000 serum dilution also
reduced the observed cross-reactivity, producing more accu-
rate disease state classifications in eight of nine specimens,
with the exception of specimen number 12 (P/N ratios, 3.74
and 3.88 against JEV-KD and WNV-RH, respectively) (Table
5). This specimen was reverse transcriptase-PCR positive for
DENV-2. It also had the highest P/N ratios against JEV-WT
(18.52) and WNV-WT (14.07) antigens (Table 5). Thus, the
relatively low positive P/N ratios with JEV-KD and WNV-RH
antigens in this patient specimen may have resulted from a
small portion of DENV-2-induced IgM antibodies recognizing
a conserved JEV serocomplex-cross-reactive epitope(s) in both
Statistical comparison of antigen performances. The assay
results, expressed as the P/N ratio, using JEV-WT or JEV-KD
antigens were analyzed in a continuous rating scale by ROC
curves for overall assay performance and are shown in Fig. 2.
For MAC-ELISA at either a 1:400 or a 1:4,000 serum dilution,
paired-ROC-curve analysis revealed no statistical difference
TABLE 5. MAC- and GAC-ELISA P/N ratios for highly cross-reactive serum specimens diluted 1:4,000
WT KDWTKD WT RHWT RH
VOL. 15, 2008ELISAs USING A NOVEL JEV ANTIGEN 831
(P ? 0.05) in the reported AUC between WT and KD antigens
in this assay (Fig. 2A). Paired-ROC-curve analysis revealed
that the AUC for each of the GAC-ELISAs was statistically
different at a 1:400 or 1:4,000 serum dilution (P ? 0.05) (Fig.
2B). Overall, the GAC-ELISA results indicated that the assay
using the JEV-KD antigen (AUC ? 0.774 and 0.764 at 1:400
and 1:4,000 serum dilutions, respectively) more accurately dis-
criminated between true JEV IgG-positive and -negative se-
rum specimens than did the JEV-WT antigen (AUC ? 0.610
and 0.548 at 1:400 and 1:4,000 serum dilutions, respectively).
Two-by-two contingency tables were prepared to analyze the
diagnostic accuracies of WT and cross-reactivity-reduced anti-
gens for both JEV and WNV, including the sensitivity, speci-
ficity, and likelihood ratio (Table 6). In the MAC-ELISA with
the JEV panel sera, the sensitivities of JEV-KD and JEV-WT
antigens were 100%. The overall specificities at the 1:400 se-
rum dilution were 41.27% and 63.64% for WT and KD anti-
gens, respectively; at the 1:4,000 dilution, the specificities in-
creased to 63.49% and 95.45%, respectively. The sensitivities
of WNV-RH and WNV-WT antigens in the MAC-ELISA with
the WNV serum panel were 100% at both 1:400 and 1:4,000
dilutions. The overall specificities at the 1:400 dilution were
17.24% and 52.63% for WT and RH antigens; at the 1:4,000
dilution, the specificities increased to 57.78% and 91.11%,
respectively. The likelihood ratio test further indicated that the
JEV-KD and WNV-RH antigens had higher propensities to
correctly determine the disease state when the serum specimen
was tested at a 1:4,000 dilution (Table 6) (likelihood ratios, 22
and 11.25, respectively, for JEV-KD and WNV-RH versus 2.74
and 2.37 for JEV-WT and WNV-WT, respectively).
The GAC-ELISA results obtained from sera tested at the
1:4,000 dilution are summarized in Table 6. The most accurate
antigens using these testing procedures were JEV-KD and
WNV-RH. Although these antigens had sensitivities of 87.5%
for JEV-KD and 100% for WNV-RH, the assay specificities
were relatively low: 45.45% for JEV-KD and 51.72% for
WNV-RH. The likelihood ratios were 1.6 and 2.07 using
FIG. 2. Fitted ROC curves using P/N ratios for JEV WT and KD
mutated VLP antigens. A JEV-infected target serum panel and five
arbovirus-infected control serum panels were determined by MAC-
ELISA (A) and GAC-ELISA (B).
TABLE 6. Influences of WT and cross-reactivity-reduced JEV and WNV VLPs on the performances of MAC- and GAC-ELISAsa
aInfluence on abilities to distinguish JEV and WNV serum panels (disease panels), respectively, from other arbovirus-infected serum panels (control panels) using
the positive-cutoff criterion (P/N ? 3) as the evidence of infection.
832CHIOU ET AL.CLIN. VACCINE IMMUNOL.
JEV-KD and WNV-RH, respectively (Table 6), indicating that
the presence of JEV serocomplex-cross-reactive antibodies in
JEV-, WNV-, and SLEV-infected patients or JEV-immune,
DENV-infected patients complicated the disease state inter-
pretation by using GAC-ELISA results alone. Thus, further
research into the identification and ablation of a complex-
cross-reactive epitope(s) and the incorporation of this muta-
tion(s) into JEV-KD and WNV-RH antigens is critical for
improving the GAC-ELISA for accurate disease burden stud-
ies in the future.
Not only was there strong statistical support for increased
diagnostic accuracy using the JEV-KD and WNV-RH antigens
versus the WT antigen, there was also a dramatic and statisti-
cally significant improvement in assay performance when pa-
tient sera were tested at a 1:4,000 instead of a 1:400 dilution.
Although this was true regardless of the antigen used, it is best
demonstrated in the MAC-ELISA with the cross-reactivity-
reduced antigens. The overall specificity with the JEV-KD
antigen increased from 63.64% to 95.45% when sera were
diluted to 1:4,000 (Table 6) (95% confidence interval [CI] ?
47.77% to 77.59% at 1:400 and 84.53% to 99.44% at 1:4,000
dilution). This performance improvement was further sup-
ported by the increase in the likelihood ratio for JEV-KD
antigen from 2.75 to 22 when the serum dilution was increased
from 1:400 to 1:4,000 (Table 6). Similar performance improve-
ments in specificity and in likelihood ratios were observed with
the WNV-RH antigen when the serum dilutions were in-
creased from 1:400 to 1:4,000 (Table 6) (95% CI ? 38.97% to
66.02%, likelihood ratio ? 2.11 for 1:400; 95% CI ? 78.78% to
97.52%, likelihood ratio ? 11.25 for 1:4,000).
The presence of cross-reactive serum antibodies developed
from sequential heterologous flavivirus infection or previous
vaccination can dramatically complicate flavivirus serodiagno-
sis. Currently, the most accurate serologic method is the
plaque reduction neutralization test, performed by testing
paired acute- and convalescent-phase serum specimens (20).
We applied a structure-based mutagenesis algorithm for the
development of cross-reactivity-reduced WNV and SLEV mu-
tant antigens that could be used in the MAC-ELISA with
single serum specimens for the accurate identification and dif-
ferentiation of WNV and SLEV infections (6, 7, 30, 37, 39).
This strategy was applied to the development of a cross-reac-
tivity-reduced JEV VLP antigen in the present study.
Substitutions for Gly104, Gly106, or Leu107 of the JEV E
protein could significantly reduce the reactivities of flavivirus
group-, subgroup-, complex-, and subcomplex-reactive MAbs
(Tables 2 and 3). None of these substitutions significantly al-
tered the reactivities of JEV-specific MAbs. Combinations of
Gly106 and Leu107 substitutions in VLP antigens further de-
creased cross-reactive MAbs’ reactivities. In most cases, the
MAb reactivity reductions observed in the single mutants com-
bined additively in the double mutants. However, in some of
the double-mutant constructs, MAb reactivity reductions were
much greater than those observed in either of the single-sub-
stitution VLPs. Similar synergistic effects on cross-reactive
MAb reactivities were previously observed in WNV and SLEV
Gly106/Leu107 mutants (7, 39). For example, in the RH VLP,
the JEV complex-reactive MAbs 16 and 6B4A-10 were re-
duced dramatically, yet these same two substitutions alone
showed little or no reduction in the reactivities of these MAbs.
In this study, the only substitutions reducing the reactivities of
MAbs 16 and 6B4A-10 were G104H and L107F. Unfortu-
nately, these mutations also resulted in decreasing the secre-
tion of mutant VLPs by an unknown mechanism. Nevertheless,
when the L107F substitution was combined with Gly106 sub-
stitutions, the resultant double-mutant VLPs reverted to
JEV-WT levels of reactivity for both of these JEV complex-
reactive MAbs. We have not observed such negatively syner-
gistic effects on MAb reactivity when single substitutions were
combined in either WNV or SLEV studies (7, 39).
Gly104, Gly106, and Leu107, like many of the flavivirus
fusion peptide residues, are almost completely invariant across
the flaviviruses. Interestingly, L107F is known to occur in a few
different flavivirus isolates; most relevant here is its occurrence
in the JEV attenuated vaccine strain SA-14-14-2, DENV-2
strain PUO-280, POWV, and deer tick virus (3, 21, 26). It is
likely that the L107F substitution does not interfere with
flavivirus viability. However, the L107F substitution has been
shown to reduce cross-reactive antibody recognition, not only
in JEV, but also in WNV, SLEV, and tick-borne encephalitis
virus (1, 7, 39).
Among the JEV Gly106/Leu107 double mutants, the
JEV-KD combination exhibited the greatest reductions in
cross-reactivity and the highest levels of type specificity. Simi-
lar results, but with different specific substitutions, were ob-
served in WNV with RH and in SLEV with G106D/L107R
substitutions (7, 39). These findings and previous reports sug-
gest that certain fusion peptide residues can act as epitope
determinants for flavivirus subgroup- and JEV complex-reac-
tive MAbs, in addition to flavivirus group-reactive MAbs (6, 7,
The serological cross-reactivity between WNV and SLEV in
MAC-ELISA was the primary reason why the 1999 outbreak of
WNV in New York City was initially thought to be caused
by SLEV (22). Strategies to differentiate current flavivirus in-
fections have been proposed, such as by determining the IgM-
to-IgG ratio (15, 36), by using recombinant EDIII or nonstruc-
tural protein 1 (NS1) as an antigen (2, 36), or by epitope-
blocking ELISA (12, 19). All of these assays have limitations,
including requiring the simultaneous testing of serum speci-
mens for IgM and IgG, requiring paired acute- and convales-
cent-phase serum specimens, or the observation that not all
infected individuals develop antibody against EDIII or NS1
antigen. Here, we have documented that the JEV-KD antigen
proved to be superior to the JEV-WT antigen; it exhibited
greater sensitivity and specificity, demonstrated by the higher
AUC values in both the IgM and the IgG assays (Fig. 2).
Moreover, the specificity of MAC- and GAC-ELISA was im-
proved significantly using JEV-KD antigen to differentiate five
distinct non-JEV-infected serum panels (Table 4). Neverthe-
less, the results of this and previous studies (30) demonstrate
the appropriateness of using these cross-reactivity-reduced
antigens to successfully differentiate JEV-specific from WNV-
specific serum antibodies with single acute-phase serum sam-
ples and paired KD and RH antigens in the MAC-ELISA.
In the GAC-ELISA, the majority of DENV-infected pa-
tients’ sera were IgG positive whether we used the JEV and
VOL. 15, 2008ELISAs USING A NOVEL JEV ANTIGEN 833
WNV WT or the cross-reactivity-reduced antigens (Table 4).
The DENV patient sera were obtained from the Taiwanese
population, where JEV is endemic. Thus, it is expected that a
high percentage of the Taiwanese population would have IgG
antibodies against JEV. Acute dengue infection in the pres-
ence of JEV immunity would be expected to enhance the
production of flavivirus-cross-reactive antibodies derived from
shared antigenic epitopes and thus be detected by WT and
cross-reactivity-reduced JEV and WNV antigens. Neverthe-
less, even with these highly cross-reactive DENV patients’ sera,
both the JEV-KD and WNV-RH antigens exhibited improved
specificity compared to the WT antigens. In the WNV- and
SLEV-infected serum panels, the cross-reactivity of IgG anti-
body against JEV was dramatically reduced using JEV-KD
antigen and serum tested at a 1:4,000 dilution. Thus, it is
possible to estimate the disease burden more precisely using
our JEV-KD and WNV-RH antigens in the GAC-ELISA, and
assay specificity is further improved by screening all sera at a
1:4,000 dilution. The assay improvements noted at a 1:4,000
versus a 1:400 serum dilution resulted from the observations
that broadly cross-reactive antibodies make up a large propor-
tion of the flavivirus antibody response and yet the smaller
proportion of virus-specific antibodies exhibits the highest
The results presented here confirm previous reports that the
conserved fusion peptide region constitutes an immunodomi-
nant antigenic hot spot and forms a region of multiple over-
lapping cross-reactive epitopes (6, 7, 27, 37, 39). We have
demonstrated that specific substitutions in this region can be
utilized to construct novel cross-reactivity-reduced serodiag-
nostic antigens with improved assay performance for diagnos-
ing and differentiating JEV-infected from other flavivirus-in-
fected human serum specimens. This work, therefore, not only
has important implications for furthering the basic understand-
ing of immunological responses to flavivirus infections but
should also strengthen the public health response to spreading
flavivirus disease by improving serodiagnostic specificity and
estimates of the global disease burden.
We thank John T. Roehrig and his laboratory for the development
and characterization of many of the MAbs utilized in this and previous
related studies and for providing access to them, to the DVBID diag-
nostic and reference laboratory and A. J. Johnson for access to the
WNV- and SLEV-infected serum panels, and to the Taiwanese CDC
for the JEV- and DENV-infected sera.
The findings and conclusions in the study are those of the authors
and do not necessarily represent the views of the Centers for Disease
Control and Prevention.
1. Allison, S. L., J. Schalich, K. Stiasny, C. W. Mandl, and F. X. Heinz. 2001.
Mutational evidence for an internal fusion peptide in flavivirus envelope
protein E. J. Virol. 75:4268–4275.
2. Beasley, D. W., M. R. Holbrook, A. P. Travassos Da Rosa, L. Coffey, A. S.
Carrara, K. Phillippi-Falkenstein, R. P. Bohm, Jr., M. S. Ratterree, K. M.
Lillibridge, G. V. Ludwig, J. Estrada-Franco, S. C. Weaver, R. B. Tesh, R. E.
Shope, and A. D. Barrett. 2004. Use of a recombinant envelope protein
subunit antigen for specific serological diagnosis of West Nile virus infection.
J. Clin. Microbiol. 42:2759–2765.
3. Blok, J., S. Samuel, A. J. Gibbs, and U. T. Vitarana. 1989. Variation of the
nucleotide and encoded amino acid sequences of the envelope gene from
eight dengue-2 viruses. Arch. Virol. 105:39–53.
4. Chang, G. J., B. S. Davis, C. Stringfield, and C. Lutz. 2007. Prospective immu-
nization of the endangered California condors (Gymnogyps californianus) pro-
tects this species from lethal West Nile virus infection. Vaccine 25:2325–2330.
5. Chang, G. J., A. R. Hunt, and B. Davis. 2000. A single intramuscular injec-
tion of recombinant plasmid DNA induces protective immunity and prevents
Japanese encephalitis in mice. J. Virol. 74:4244–4252.
6. Crill, W. D., and G. J. Chang. 2004. Localization and characterization of
flavivirus envelope glycoprotein cross-reactive epitopes. J. Virol. 78:13975–
7. Crill, W. D., N. B. Trainor, and G. J. Chang. 2007. A detailed mutagenesis
study of flavivirus cross-reactive epitopes using West Nile virus-like particles.
J. Gen. Virol. 88:1169–1174.
8. Davis, B. S., G. J. Chang, B. Cropp, J. T. Roehrig, D. A. Martin, C. J.
Mitchell, R. Bowen, and M. L. Bunning. 2001. West Nile virus recombinant
DNA vaccine protects mouse and horse from virus challenge and expresses
in vitro a noninfectious recombinant antigen that can be used in enzyme-
linked immunosorbent assays. J. Virol. 75:4040–4047.
9. Gentry, M. K., E. A. Henchal, J. M. McCown, W. E. Brandt, and J. M.
Dalrymple. 1982. Identification of distinct antigenic determinants on den-
gue-2 virus using monoclonal antibodies. Am. J. Trop. Med. Hyg. 31:548–
10. Halstead, S. B. 1988. Pathogenesis of dengue: challenges to molecular biol-
ogy. Science 239:476–481.
11. Hanley, J. A., and B. J. McNeil. 1983. A method of comparing the areas
under receiver operating characteristic curves derived from the same cases.
12. Hawkes, R. A., J. T. Roehrig, C. R. Boughton, H. M. Naim, R. Orwell, and
P. Anderson-Stuart. 1990. Defined epitope blocking with Murray Valley
encephalitis virus and monoclonal antibodies: laboratory and field studies.
J. Med. Virol. 32:31–38.
13. Holmes, D. A., D. E. Purdy, D. Y. Chao, A. J. Noga, and G. J. Chang. 2005.
Comparative analysis of immunoglobulin M (IgM) capture enzyme-linked
immunosorbent assay using virus-like particles or virus-infected mouse brain
antigens to detect IgM antibody in sera from patients with evident flaviviral
infections. J. Clin. Microbiol. 43:3227–3236.
14. Hunt, A. R., C. B. Cropp, and G. J. Chang. 2001. A recombinant particulate
antigen of Japanese encephalitis virus produced in stably-transformed cells is
an effective noninfectious antigen and subunit immunogen. J. Virol. Meth-
15. Innis, B. L., A. Nisalak, S. Nimmannitya, S. Kusalerdchariya, V. Chongs-
wasdi, S. Suntayakorn, P. Puttisri, and C. H. Hoke. 1989. An enzyme-linked
immunosorbent assay to characterize dengue infections where dengue and
Japanese encephalitis co-circulate. Am. J. Trop. Med. Hyg. 40:418–427.
16. Kanai, R., K. Kar, K. Anthony, L. H. Gould, M. Ledizet, E. Fikrig, W. A.
Marasco, R. A. Koski, and Y. Modis. 2006. Crystal structure of West Nile
virus envelope glycoprotein reveals viral surface epitopes. J. Virol. 80:11000–
17. Kimura-Kuroda, J., and K. Yasui. 1986. Antigenic comparison of envelope
protein E between Japanese encephalitis virus and some other flaviviruses
using monoclonal antibodies. J. Gen. Virol. 67:2663–2672.
18. Kimura-Kuroda, J., and K. Yasui. 1983. Topographical analysis of antigenic
determinants on envelope glycoprotein V3 (E) of Japanese encephalitis
virus, using monoclonal antibodies. J. Virol. 45:124–132.
19. Kitai, Y., M. Shoda, T. Kondo, and E. Konishi. 2007. Epitope-blocking
enzyme-linked immunosorbent assay to differentiate West Nile virus from
Japanese encephalitis virus infections in equine sera. Clin. Vaccine Immu-
20. Kuno, G. 2003. Serodiagnosis of flaviviral infections and vaccinations in
humans. Adv. Virus Res. 61:3–65.
21. Kuno, G., H. Artsob, N. Karabatsos, K. R. Tsuchiya, and G. J. Chang. 2001.
Genomic sequencing of deer tick virus and phylogeny of powassan-related
viruses of North America. Am. J. Trop. Med. Hyg. 65:671–676.
22. Lanciotti, R. S., J. T. Roehrig, V. Deubel, J. Smith, M. Parker, K. Steele, B.
Crise, K. E. Volpe, M. B. Crabtree, J. H. Scherret, R. A. Hall, J. S. Mac-
Kenzie, C. B. Cropp, B. Panigrahy, E. Ostlund, B. Schmitt, M. Malkinson,
C. Banet, J. Weissman, N. Komar, H. M. Savage, W. Stone, T. McNamara,
and D. J. Gubler. 1999. Origin of the West Nile virus responsible for an
outbreak of encephalitis in the northeastern United States. Science 286:
23. Mackenzie, J. S. 2005. Emerging zoonotic encephalitis viruses: lessons from
Southeast Asia and Oceania. J. Neurovirol. 11:434–440.
24. Mackenzie, J. S., D. J. Gubler, and L. R. Petersen. 2004. Emerging flavivi-
ruses: the spread and resurgence of Japanese encephalitis, West Nile and
dengue viruses. Nat. Med. 10:S98–S109.
25. Martin, D. A., B. J. Biggerstaff, B. Allen, A. J. Johnson, R. S. Lanciotti, and
J. T. Roehrig. 2002. Use of immunoglobulin M cross-reactions in differential
diagnosis of human flaviviral encephalitis infections in the United States.
Clin. Diagn. Lab. Immunol. 9:544–549.
26. Nitayaphan, S., J. A. Grant, G. J. Chang, and D. W. Trent. 1990. Nucleotide
sequence of the virulent SA-14 strain of Japanese encephalitis virus and its
attenuated vaccine derivative, SA-14-14-2. Virology 177:541–552.
27. Oliphant, T., G. E. Nybakken, M. Engle, Q. Xu, C. A. Nelson, S. Sukupolvi-
Petty, A. Marri, B. E. Lachmi, U. Olshevsky, D. H. Fremont, T. C. Pierson,
and M. S. Diamond. 2006. Antibody recognition and neutralization deter-
834CHIOU ET AL.CLIN. VACCINE IMMUNOL.
minants on domains I and II of West Nile virus envelope protein. J. Virol.
28. Purdy, D. E., A. J. Noga, and G. J. Chang. 2004. Noninfectious recombinant
antigen for detection of St. Louis encephalitis virus-specific antibodies in
serum by enzyme-linked immunosorbent assay. J. Clin. Microbiol. 42:4709–
29. Rey, F. A., F. X. Heinz, C. Mandl, C. Kunz, and S. C. Harrison. 1995. The
envelope glycoprotein from tick-borne encephalitis virus at 2 Å resolution.
30. Roberson, J. A., W. D. Crill, and G. J. Chang. 2007. Differentiation of West
Nile and St. Louis encephalitis virus infections using cross-reactivity reduced
noninfectious virus-like particles. J. Clin. Microbiol. 45:3167–3174.
31. Roehrig, J. T., R. A. Bolin, and R. G. Kelly. 1998. Monoclonal antibody
mapping of the envelope glycoprotein of the dengue 2 virus, Jamaica. Vi-
32. Roehrig, J. T., A. R. Hunt, A. J. Johnson, and R. A. Hawkes. 1989. Synthetic
peptides derived from the deduced amino acid sequence of the E-glycopro-
tein of Murray Valley encephalitis virus elicit antiviral antibody. Virology
33. Roehrig, J. T., A. J. Johnson, A. R. Hunt, R. A. Bolin, and M. C. Chu. 1990.
Antibodies to dengue 2 virus E-glycoprotein synthetic peptides identify an-
tigenic conformation. Virology 177:668–675.
34. Roehrig, J. T., J. H. Mathews, and D. W. Trent. 1983. Identification of
epitopes on the E glycoprotein of Saint Louis encephalitis virus using mono-
clonal antibodies. Virology 128:118–126.
35. Shu, P. Y., S. F. Chang, Y. C. Kuo, Y. Y. Yueh, L. J. Chien, C. L. Sue, T. H.
Lin, and J. H. Huang. 2003. Development of group- and serotype-specific
one-step SYBR green I-based real-time reverse transcription-PCR assay for
dengue virus. J. Clin. Microbiol. 41:2408–2416.
36. Shu, P. Y., L. K. Chen, S. F. Chang, Y. Y. Yueh, L. Chow, L. J. Chien, C.
Chin, T. H. Lin, and J. H. Huang. 2003. Comparison of capture immuno-
globulin M (IgM) and IgG enzyme-linked immunosorbent assay (ELISA)
and nonstructural protein NS1 serotype-specific IgG ELISA for differenti-
ation of primary and secondary dengue virus infections. Clin. Diagn. Lab.
37. Stiasny, K., S. Kiermayr, H. Holzmann, and F. X. Heinz. 2006. Cryptic
properties of a cluster of dominant flavivirus cross-reactive antigenic sites.
J. Virol. 80:9557–9568.
38. Sukupolvi-Petty, S., S. K. Austin, W. E. Purtha, T. Oliphant, G. E. Ny-
bakken, J. J. Schlesinger, J. T. Roehrig, G. D. Gromowski, A. D. Barrett,
D. H. Fremont, and M. S. Diamond. 2007. Type- and subcomplex-specific
neutralizing antibodies against domain III of dengue virus type 2 envelope
protein recognize adjacent epitopes. J. Virol. 81:12816–12826.
39. Trainor, N. B., W. D. Crill, J. A. Roberson, and G. J. Chang. 2007. Mutation
analysis of the fusion domain region of St. Louis encephalitis virus envelope
protein. Virology 360:398–406.
40. Vaughn, D. W., and C. H. Hoke, Jr. 1992. The epidemiology of Japanese
encephalitis: prospects for prevention. Epidemiol. Rev. 14:197–221.
41. Wu, Y. C., Y. S. Huang, L. J. Chien, T. L. Lin, Y. Y. Yueh, W. L. Tseng, K. J.
Chang, and G. R. Wang. 1999. The epidemiology of Japanese encephalitis on
Taiwan during 1966–1997. Am. J. Trop. Med. Hyg. 61:78–84.
VOL. 15, 2008ELISAs USING A NOVEL JEV ANTIGEN 835