Methods

Experiments involving animals. All experiments involving animals were conducted in accordance with the guidelines and regulations put forth by the relevant institutional animal care and use committees.

Preparation and purification of horse anti-E polyclonal antibodies. Recombinant truncated WNV E protein (rWNV-Et) was expressed in Drosophila cells and purified as described elsewhere [18]. Equine anti-WNV E serum was produced by Cocalico Biologicals. A horse shown to lack antibodies against WNV E protein was immunized intradermally in multiple locations with a total of 250 µg of rWNV-Et in Titermax adjuvant (Sigma). Boosts with 125 µg of rWNV-Et in Titermax were given intradermally 56, 84, 118, and 140 days later. Serum samples were collected before each injection. All experiments described here were performed using a 2.4-L serum sample collected 70 days after the first injection.

Horse immunoglobulins were purified by affinity chromatography on HiTrap Protein G columns (GE Healthcare) as recommended by the manufacturer. Horse anti-E protein IgGs were purified by affinity column chromatography on rWNV-Et covalently bound to N-hydroxysuccinimide (NHS)-Sepharose (GE Healthcare) as recommended by the manufacturer.

Selection of horse immunoglobulins with linear peptides. Forty synthetic, overlapping, 20mer peptides corresponding to the first 410 residues of the WNV E protein were obtained from Sigma-Genosys. Selected peptides were coupled to HiTrap NHS-Sepharose (GE Healthcare) as recommended by the manufacturer. Table 1 shows the sequence and position of the 10 peptides used for affinity chromatography in this study. Horse serum samples were applied to the peptide columns after sterile filtration. Retained immunoglobulins were eluted with 100 mmol/L glycine-HCl (pH 2.7). The eluate was immediately neutralized and dialyzed against PBS.

Production and purification of recombinant DENV E proteins. We expressed truncated, soluble versions of E proteins from DENV serotypes 1, 2, and 4 (strains 98901518, New Guinea-C, and 814669, respectively). These soluble proteins are referred to as rDENV1-Et, rDENV2-Et, and rDENV4-Et. Synthetic genes encoding the first 400 aa of the DENV E proteins were obtained from GenScript. These DNA sequences were introduced into the pMTBiP/V5-HisA plasmid (Invitrogen) for expression under the control of the metallothionein promoter. The expressed recombinant proteins do not contain the V5 epitope or the polyhistidine tag encoded by the plasmid vector. The recombinant plasmids were introduced into Drosophila S2 cells (Invitrogen), and cells producing recombinant protein were selected by culture in the presence of hygromycin. Production of recombinant proteins was initiated by addition of 0.5 mmol/L copper sulphate in Drosophila serum-free culture medium (Invitrogen). Recombinant proteins were purified from the culture supernatant by ion-exchange and size-exclusion column chromatography.

ELISAs. Conditions for ELISA for antibodies against rWNV-Et have been described elsewhere [18]. ELISAs for antibodies against rDENV1-Et, rDENV2-Et, and rDENV4-Et proteins were performed as for rWNV-Et, substituting the appropriate antigen to coat the wells of assay plates. The end-point titer was the highest dilution leading to a signal significantly higher than background.

For anti-peptide ELISA, plates were coated with 1 µg of peptide per well in 0.1 mol/L carbonate buffer (pH 9.6) overnight at 4°C. After the peptide solution was removed, plates were blocked with 1% bovine serum albumin in PBS containing 0.05% Tween 20 for 30 min at room temperature. Antibodies diluted in blocking solution in dilution plates were added to the peptide-coated plates for 1 h at room temperature. After being washed with PBS-0.05% Tween 20, bound antibody was detected using a goat anti-horse IgG secondary antibody conjugated to alkaline phosphatase (Sigma) diluted 1:4000 for 45 min at room temperature. After an additional washing step, the colorimetric substrate, p-nitrophenyl phosphate (Sigma), was used to quantitate the amount of bound conjugate.

Cytopathic effect (CPE) neutralization assay. Neutralizing antibody titers were determined using a microtiter assay that was developed by the National Veterinary Services Laboratories and that has been shown to yield results identical to the more common plaque-reduction neutralization assay. All tests were performed by the Diagnostic Testing Services Laboratory of the University of Connecticut.

Serum samples were heat inactivated at 56°C for 30 min and then diluted 1:5 in Dulbecco's modified Eagle medium(DMEM) containing antibiotics and antimycotics. Two-fold serial dilutions were made in duplicate rows of a 96-well plate. A serum control well was included for each sample. WNV (200 TCID₅₀) in 25 µL of DMEM containing antibiotics, antimycotics, and 10% guinea pig complement was added to each well, except for the serum control wells; serum control wells received 25 µL of 10% guinea pig complement medium.

After the virus was added, plates were incubated for 1 h at 37°C with 5% CO₂; 100 µL of Vero cells was then added to each well. Plates were incubated for 4–5 days at 37°C in 5% CO₂, and the wells were examined for CPE by use of an inverted microscope. Back titration plates containing both positive and negative control serum were prepared with each test run. The endpoint titer was the highest serum dilution at which no CPE was seen in either duplicate sample.

Mouse passive immunization and viral challenge. Four survival experiments were performed with groups of five to ten 4–6-week-old female C3H/HeN mice, purchased from Charles River Laboratories. Mice were matched by age for each experiment. Mice were passively immunized by intraperitoneal injection of 50–200 µg of immunoglobulins. Negative control mice received an intraperitoneal injection of PBS or immunoglobulins purified from normal horse serum. Positive control mice received 50–100 µg of immunoglobulins affinity purified with rWNV-Et. Mice were challenged 24 h later with 100–1000 pfu of WNV administered intraperitoneally. Mice were monitored daily for morbidity and mortality for at least 21 days.

Indirect immunofluorescent antibody staining with anti-WNV E protein antibodies. Microscope slides containing fixed Vero cells uninfected or infected with various arboviruses and flaviviruses were purchased from Panbio. Slides were incubated with purified IgGs at 10 µg/mL in PBS containing 0.5% fetal bovine serum (FBS) for 30 min at 37°C in a moist chamber. They were rinsed with PBS for 10 min and incubated with fluorescein isothiocyanate-labeled sheep anti-horse IgG diluted as recommended by the manufacturer (Sigma) in PBS containing 0.5% FBS for 30 min at 37°C in a moist chamber. Finally, the cells were counterstained with 20 µg/mL propidium iodide for 3 min, rinsed with PBS, and mounted in Vectashield medium (Vector Laboratories) before observation with a Zeiss Axiovert 200M inverted microscope. Identical exposure times were used for all samples examined.

Results

Production of a high-titer horse antiserum against WNV E protein. The series of experiments described here required the production of large amounts of immunoglobulins against rWNV-Et. We chose to the horse as a source of immunoglobulins because this economically important animal is potentially subject to lethal WNV infections. Immunization of a horse with our rWNV-Et antigen in Titermax adjuvant led to the production of a high-titer antiserum. The end-point titer of 13 consecutive serum samples was determined by ELISA for antibody against rWNV-Et. The serum titer was 1:51,200 when tested 56 days after the first injection. Booster injections led to a progressively higher serum titer, which ultimately reached 1:409,600. All experiments described here were performed using serum collected 70 days after the first injection. The end-point titer of this production bleed was 1:204,800 against rWNV-Et. Antibodies that were cross-reactive with rDENV2-Et developed concurrently with anti-rWNV-Et antibodies. The production bleed used in these experiments had an end-point titer of 1:25,600 against rDENV2-Et.

Identification of linear determinants recognized by antiserum from an immunized horse. We tested the ability of the antiserum from the immunized horse to recognize synthetic peptides whose sequences spanned residues 1–410 of the WNV E protein. Each 20mer peptide overlapped with its neighbors by 10 aa. Yields of peptides 2, 3, 6, 7, 12, 18, 28, and 37 were too low to be included in this analysis. Reactivity of the immune serum and the preimmune serum at a 1:500 dilution in an ELISA is shown in figure 1. The immunization regimen we followed led to the production of antibodies recognizing a large number of WNV E peptides. Maximum immunoreactivity was observed with linear peptides from 5 distinct regions: residues 1–20 (peptide 1), residues 71–110 (peptides 8–10, corresponding to the region of the fusion loop), residues 141–200 (peptides 15–19), residues 231–270 (peptides 24–26), and residues 371–410 (peptides 38–40).

Purification of horse immunoglobulins that bind to linear peptides. To determine whether any of these peptides were the targets of protective antibodies, we purified and characterized subpopulations of horse IgGs capable of binding to peptides 1, 9, 10, 11, 19, 24, 26, 38, and 40. We also included peptide 29 in this analysis because it is recognized by a neutralizing recombinant monoclonal antibody against WNV [9]. Although these immunoglobulin preparations are referred to as “anti-peptide,” the immune response was elicited by immunization with the 406-aa rWNV-Et antigen.

We verified that each immunoglobulin preparation was specific for the peptide toward which it was selected. ELISA plates were coated with individual peptides, and the binding of peptide-selected immunoglobulin preparations was measured. The results of this experiment are shown in table 2. As expected, each anti-peptide immunoglobulin preparation recognized intact rWNV-Et. In addition, most immunoglobulin preparations bound only their cognate peptide. The exceptions were anti-peptide 9, 10, and 11, which cross-reacted markedly. This was expected, because peptides 9, 10, and 11 overlap with each other by 10 residues. Importantly, these results show that the peptideaffinity chromatography procedure selects subpopulations of immunoglobulins with distinct specificities.

Neutralization of WNV in vitro by immunoglobulins affinity purified with linear peptides of the WNV E protein. The 50% CPE-neutralization titer of each immunoglobulin preparation was determined. The 50% CPE-neutralization titer of the horse immune serum was 1:10,240. No neutralizing activity was observed at a 1:10 dilution of preimmune horse serum. The concentration of IgG necessary to reduce the CPE by 50% ranged from 1.9 to 7.9 µg/mL for immunoglobulin preparations selected against peptides 1, 9, 10, 11, 19, 24, and 26. The highest neutralizing activity was that for immunoglobulins binding to peptide 10, corresponding to the fusion loop of the E protein. Unlike all other immunoglobulin preparations, immunoglobulins binding to peptide 29 had no measurable neutralizing activity in vitro.

Protection of mice against a lethal challenge with WNV conferred by immunoglobulins affinity purified with linear peptides of the WNV E protein. The protective effect of horse anti-rWNV-Et immunoglobulins was tested in vivo using an established model of WNV disease. C3H/HeN mice infected withWNVstrain 2741 develop a neuroinvasive disease. Depending on the viral dose, naive mice die 6–11 days after viral challenge. We tested the protective effect of anti-peptide 1 and anti-peptide 10 immunoglobulins, because of their relatively high in vitro neutralizing activity (2.5 and 1.9 µg/mL, respectively), as well as of anti-peptide 29, because this peptide is recognized by protective recombinant antibodies [9].

Mice were injected intraperitoneally with 200 µg of peptideselected immunoglobulins 24 h before being challenged with an otherwise lethal dose of WNV. Positive control mice received 200 µg of immunoglobulins purified by affinity chromatography with rWNV-Et. Negative control mice received an injection of PBS or 200 µg of horse immunoglobulins purified from a preimmune serum sample. Survival data are summarized in table 3. As expected, mice receiving preimmune immunoglobulins uniformly died 9–10 days after challenge, whereas affinitypurified anti-rWNV-Et immunoglobulins protected 100% of the mice. Immunoglobulins affinity purified with peptides 1, 10, and 29 conferred partial protection (48%–59%) against a lethal viral challenge. This degree of protection, although incomplete, was significantly different from that for the negative control group (P < .001).

In additional experiments, we administered the anti-peptide immunoglobulins 2 and 5 days after infection with WNV, to test their therapeutic potential. None of the immunoglobulins conferred a significant protection when administered after challenge. F(ab')₂ fragments prepared from immune horse immunoglobulins provided little or no protection, whether administered prophylactically or therapeutically (data not shown).

Binding to other flaviviruses by immunoglobulins affinity purified with linear peptides of the WNV E protein. The cross-reactivity of IgGs raised against the WNV E protein were tested in an indirect immunofluorescent antibody assay. Figure 2 shows that horse immunoglobulins affinity purified against rWNV-Et stain cells infected with WNV, Japanese encephalitis virus (JEV), yellow fever virus (YFV), or DENV (serotypes 1–4). In contrast, immunoglobulins purified from preimmune serum did not produce any significant staining.

We then characterized the cross-reactivity of peptide-selected horse immunoglobulins. Uninfected Vero cells and cells infected with Venezuelan equine encephalitis virus, an alphavirus, served as negative controls. Intensity of staining was evaluated semiquantitatively (table 4). All immunoglobulin preparations bound to WNV-infected cells; however, the anti-peptide immunoglobulins differed widely in their cross-reactivity. Immunoglobulins recognizing the E protein fusion loop (anti-peptide 10 and anti-peptide 11) were broadly cross-reactive, confirming that this is a well-preserved structural element among the E proteins of flaviviruses. In contrast, anti-peptide 29 immunoglobulins were essentially specific for WNV.

We quantitatively evaluated the cross-reactivity of anti-WNV E antibodies in an ELISA format, in which antigens are expected to maintain part of their native structure. Assay-plate wells were separately coated with purified rDENV1-Et, rDENV2-Et, or rDENV4-Et. Anti-WNV horse antiserum was applied at a dilution of 1:8000, whereas each anti-peptide immunoglobulin preparation was applied at a concentration of 0.5 µg/mL. Peptide affinity-purified immunoglobulins recognized conserved epitopes in the folded E proteins (table 5). Some anti-peptide immunoglobulins were markedly more cross-reactive than others. As was the case in the immunofluorescence assay, anti-peptide 1, 9, 10, and 11 immunoglobulins were the most broadly cross-reactive.

Discussion

The goal of the present study was to identify structural determinants within the E protein of WNV that are important for protective immunity. Because immunization with short peptides often yields antibody preparation with poor titer and specificity, we adopted an indirect approach. A horse was immunized with a soluble, truncated version of the WNV E protein. Antibodies recognizing linear determinants were then selected from horse antiserum by affinity chromatography purification with linear 20mer peptides. This approach allowed us to obtain milligram quantities of polyclonal immunoglobulins with discrete specificities. Importantly, each immunoglobulin preparation was specific for the peptide that was used to select it, clearly showing that they recognized distinct epitopes. The structural elements of the E protein recognized by the immunoglobulins may be larger than the peptides used for affinity purification. As a consequence, the linear peptides we used may be only part of the complex structural determinants recognized by the immunoglobulins.

The horse immune response appeared to be very diverse, with most linear peptides being recognized by the equine antiserum. The murine response may instead be focused on a smaller number of structural determinants [20]. This could be a characteristic of the equine immune response or be linked to the immunization protocol we followed. Horse immunoglobulins bound to several peptides, such as peptide 1, which has not been predicted to be exposed to solvent on the basis of the E protein crystal structure [12, 13]. Such a phenomenon had been previously observed in immunized mice [20], suggesting that the structure of the E protein may be more flexible than expected.

The data presented here demonstrate that horse antibodies recognizing multiple epitopes in domains I and II of theWNVE protein are neutralizing in vitro, confer protection to mice in an established model of human disease, and are cross-reactive toward other flaviviruses. Our results contrast with those from a the recent study of tickborne encephalitis virus by Stiasny et al. [21]. These authors found that several monoclonal antibodies recognizing the E protein of multiple flaviviruses all bound to the fusion loop and were nonneutralizing. Our data also contrast with the results of a study by Oliphant et al. [20], who showed that all of their cross-reactive mouse neutralizing monoclonal antibodies recognizing domains I and II of the WNV E protein recognized the fusion loop. We found instead that horse immunoglobulin preparations with 7 distinct specificities (table 2) were cross-reactive and neutralizing. It seems highly unlikely that all of these immunoglobulins could bind to the fusion loop. Our results may underscore a significant difference between the equine and murine immune responses.

The neutralizing capacity of the anti-peptide immunoglobulins is relatively low compared with the entire immunoglobulin population found in serum. Efficient virus neutralization may possibly require >1 antigen-antibody interaction. Alternatively, the best neutralizing antibodies may recognize complex conformational E protein epitopes and fail to be affinity selected by linear peptides.

There is no clear correlation between the in vitro neutralization titers and the protection conferred in vivo. Immunoglobulins with a high neutralizing activity, such as anti-peptide 10, did protect mice (table 3); however, the anti-peptide 29 preparation was not neutralizing in vitro but was protective in vivo. Protection by nonneutralizing antibodies has been observed for several viruses. Passively administered antibody may protect animals through a variety of mechanisms not involving viral neutralization, such as opsonization or complement activation. However, this observation means that in vitro neutralization titers may not be an ideal surrogate marker of efficacy in studies of WNV vaccine development.

The immunoglobulin preparations we obtained cross-reacted with other flaviviruses. We found that immunoglobulins recognizing certain linear peptides of the WNV E protein reacted strongly with other flaviviruses, such as DENV and, to a lesser extent, JEV and YFV. The ability to bind multiple flaviviruses was not restricted to immunoglobulins directed against the fusion loop, as has been shown in mice [20]. In preliminary experiments, we found that horse immunoglobulins against linear WNVpeptides could inhibit the replication of DENV (serotypes 1–4) in vitro (data not shown). This result suggests that antibodies against WNV could have a broad application for the prevention and treatment of infection with other flaviviruses of worldwide significance.