West Nile virus–specific immunoglobulin isotype responses in vaccinated and infected horses

Abstract

OBJECTIVE

To compare antibody responses of horses naturally infected with West Nile virus (WNV) and those vaccinated against WNV, to identify whether vaccination interferes with the ability to diagnose WNV infection, and to determine the duration of antibody responses after vaccination.

SAMPLE

Sera from horses naturally infected with WNV (n = 10) and adult WNV-naïve horses before and after vaccination with a live canarypox virus–vectored vaccine (7) or a killed virus vaccine (8).

PROCEDURES

An established WNV IgM capture ELISA was used to measure IgM responses. Newly developed capture ELISAs were used to measure responses of 8 other WNV-specific immunoglobulin isotypes. A serum neutralization assay was used to determine anti-WNV antibody titers.

RESULTS

WNV-specific IgM responses were typically detected in the sera of WNV-infected horses but not in sera of horses vaccinated against WNV. Natural infection with and vaccination against WNV induced an immunoglobulin response that was primarily composed of IgG1. West Nile virus–specific IgG1 was detected in the sera of most horses 14 days after vaccination. Serum anti-WNV IgG1 and neutralizing antibody responses induced by the killed-virus vaccines were higher and lasted longer than did those induced by the live canarypox virus–vectored vaccine.

CONCLUSIONS AND CLINICAL RELEVANCE

On the basis of these findings, we recommend that horses be vaccinated against WNV annually near the beginning of mosquito season, that both IgM and IgG1 responses against WNV be measured to distinguish between natural infection and vaccination, and that a WNV IgG1 ELISA be used to monitor anti-WNV antibodies titers in vaccinated horses.

West Nile virus is a single-stranded RNA virus of the family Flaviviridae that has been endemic in North America since 1999.¹ It is maintained in an enzootic cycle between avian and mosquito hosts, with Culex mosquitoes being the primary vector in North America.² The virus can be transmitted by infected mosquitos to other animal species, including horses and humans. Most mammals are dead-end hosts of the virus and do not further transmit the infection. Clinical signs of WNV infection are manifested after a 5- to 15-day incubation period and low, transient viremia. In horses, WNV infection is most often subclinical. Occasionally, infected horses are mildly febrile and lethargic.The neurologic form of WNV disease in horses is caused by highly virulent strains of the virus, including certain lineage 1 strains in North America and a few lineage 2 strains in South Africa.³ Affected horses develop meningoencephalitis and may have ataxia, tetraparesis, muscle fasciculations, fever, bruxism, blindness, and facial or lingual paralysis. The case fatality rate of the neurologic form of WNV disease in horses ranges from 30% to 40%.^4–6

Diagnosis of WNV infection can be made by the use of an IgM capture ELISA or Flavivirus-blocking ELISA on serum samples.^7–10 West Nile virus–specific IgM can be found in the circulation within 6 to 7 days after infection.⁶ Plaque reduction neutralization tests or SN assays have been used to confirm positive IgM ELISA results in WNV-free areas.^9–11 In horses with WNV meningoencephalitis,WNV-specific IgM can also be detected in the CSF and nonspecific mononuclear pleocytosis can sometimes be detected during cytologic evaluation of the CSF.¹² Definitive postmortem diagnosis of WNV meningoencephalitis can be made by immunohistochemical analysis of brain and spinal cord tissue specimens.

Mosquito control and vaccination against WNV remain the mainstays of WNV prevention in horses. The first commercially available WNV vaccine for horses was a killed whole-virus vaccine^a with a potent adjuvant. This vaccine was released in the United States in August 2002¹³ and has been marketed in Europe under a different name^b since 2008. Currently, several WNV vaccines are available for horses in North America, including a recombinant canarypox virus–vectored vaccine and various killed-virus vaccines. West Nile virus vaccines provide horses with adequate protection against WNV infection and rarely induce major adverse effects.^9,14

Results of a recent study¹⁵ indicate that one of the killed whole-virus vaccines induces IgM responses in WNV-naïve horses, which suggests that vaccination against WNV might interfere with the ability to reliably distinguish between naturally infected and vaccinated horses by means of WNV-specific IgM detection. Recent reviews^16,17 of WNV diagnosis and prevention techniques conclude that currently available WNV vaccines decrease clinical disease in horses and the modified-live virus vaccine induces an IgG response. However, it is unknown whether all available WNV vaccines induce an IgM response or whether vaccination and natural infection with WNV result in different immunoglobulin isotype profiles. Because it is unknown whether all the current WNV vaccines induce an IgM response in horses, it is difficult to know whether determination of anti-WNV antibody isotype responses or titers can be used to accurately distinguish horses that have been vaccinated from those that have been naturally infected or predict the onset and duration of protection against WNV induced by each vaccine.

The purpose of the study reported here was to determine whether 2 commercial WNV vaccines (a live canarypox virus–vectored vaccine^c and a killed virus vaccine^d) induce an IgM response that interferes with diagnosis of WNV infection by means of an IgM capture ELISA and to determine the predominant immunoglobulin isotypes that are induced after vaccination against or natural infection with WNV Furthermore, we aimed to identify the ideal time frame for vaccination of horses against WNV in North America on the basis of the longevity of the anti-WNV antibodies induced by those 2 commercially available vaccines.

Materials and Methods

HORSES NATURALLY INFECTED WITH WNV

Serum samples from 13 horses with neurologic signs that were submitted to the Cornell University Animal Health Diagnostic Center prior to the release of the first commercial WNV vaccine in August 2002 were evaluated to determine anti-WNV isotype responses. The neurologic signs were reported by submitting veterinarians on the submission forms and included tremors, muscle fasciculation, lethargy, forelimb or hind limb weakness, ataxia, and recumbency. Several horses were also reported to have pyrexia. All serum samples were evaluated for the presence of antibodies against WNV by the use of an IgM capture ELISA¹⁰ and SN assay.¹⁰ Ten of 13 horses were seropositive for WNV-specific IgM, had anti-WNV SN titers between 2,048 and 8,192 (median, 4,096), and were considered infected with WNV. The remaining 3 horses were seronegative for WNV-specific IgM, had anti-WNV SN titers < 4, and were considered uninfected with WNV. West Nile virus–specific immunoglobulin isotype testing was performed on serum samples from all 13 horses for comparison purposes.

WNV-NAÏVE HORSES

Sixteen Icelandic horses (15 mares and 1 stallion) with no prior exposure to WNV (ie, had not been naturally infected with or vaccinated against WNV) were imported from Iceland (where WNV has never been identified^e) into research facilities owned by the Cornell University College of Veterinary Medicine on February 4, 2012. The import quarantine was conducted at the research facility. Following the required quarantine period, the horses were housed on pasture with access to run-in sheds. The mares ranged in age from 5 to 13 years (median, 8 years), and the stallion was 16 years old. All of the mares were 6 to 7 months pregnant at the time of importation and foaled in June 2012. Serum samples were obtained from all mares immediately before and after importation and at monthly intervals prior to vaccination and evaluated by use of a capture ELISA⁹ to monitor WNV IgM status. All of the mares remained seronegative for WNV-specific IgM prior to vaccination. Additionally, review of archived results for the WNV IgM capture ELISA performed at the Animal Health Diagnostic Center of Cornell University on serum samples obtained from horses throughout the United States from 2009 to 2011 suggested that horses typically become infected with WNV between July and October. All animal procedures were approved by the Cornell University Institutional Animal Care and Use Committee.

WNV VACCINATION PROTOCOL FOR WNV-NAÏVE HORSES

The stallion was vaccinated with a killed whole WNV vaccine^a 5 days after importation on February 9, 2012, and again 25 days later on March 5, 2012. All mares were vaccinated against WNV 12 days after parturition between June 19 and July 10, 2012. Seven mares were vaccinated with a recombinant canarypox virus–vectored WNV vaccine,^c and 8 mares were vaccinated with a different killed WNV vaccine^d than that used for the stallion. All mares received a second (booster) dose of the assigned vaccine 6 weeks after administration of the first dose.The vaccines were administered IM in the right or left pectoral muscles.

Results of another study¹⁵ indicate that the vaccine^a administered to the stallion (ie, the first WNV vaccine to become commercially available in the United States and marketed under a different name^b in Europe) induces both IgM and IgG responses. On the basis of anecdotal reports, the other WNV vaccines currently available in the United States were believed to not induce WNV-specific IgM responses sufficient to yield positive results when sera are tested with a WNV IgM capture ELISA. We vaccinated the stallion with a different vaccine than those used to vaccinate the mares to confirm that the vaccine^a did induce a WNV-specific IgM response and to verify that any negative results for WNV-specific IgM for the mares vaccinated with the other 2 vaccines were not caused by differences or detection limits of the WNV IgM capture ELISA used.

SAMPLE COLLECTION

Blood samples (10 mL each) were obtained from all horses immediately before and after importation and at various times before and after vaccination against WNV. The day that the first dose of the assigned vaccine was administered was designated as day 0 for all horses. Samples were obtained from the stallion on days −10 (immediately prior to exportation from Iceland), −3 (immediately after importation), 5, 12, 25, 34, 55, 81, 145, 180, 210, 240, and 270. Samples were obtained from the mares immediately before and after importation, at monthly intervals prior to vaccination, and on days −40, −7 (5 days after parturition), 0 (immediately before vaccination), 14, 46, 74, 110, and 140. Each blood sample was obtained by jugular venipuncture with an evacuated blood sample collection system and a 20-gauge needle. Blood samples were allowed to clot and centrifuged at 700 ✕ g for 10 minutes. The serum from each sample was then harvested, frozen, and stored at −20°C until analyzed.

WNV CAPTURE ELISAS

Each serum sample was analyzed with 9 capture ELISAs, each with a different capture mAb against a specific equine immunoglobulin isotype. The serum samples obtained from the stallion prior to (day −10) and after vaccination (day 34) were used as the negative and positive controls, respectively, for the capture ELISAs. The capture ELISA for detection of WNV-specific IgM was performed as described¹⁰ with anti-equine IgM-1-22 mAb. The capture ELISAs used to detect the other immunoglobulin isotypes were performed similarly except the following purified anti-equine mAbs were used instead of anti–IgM-1-22:anti–IgG1 (CVS45),¹⁸ anti–IgG1/3-159,¹⁹ anti–IgG4/7 (CVS39),¹⁸ anti–IgG5-586 (recognizes IgG3/5),¹⁹ anti–IgG5-416,¹⁹ anti–IgG6-267,¹⁹ anti–IgE-176,²⁰ and anti–IgA-234. Briefly, 100 μL of a solution containing the coating mAb and ELISA coating buffer (concentration, 4 g/mL; 1M NaHCO₃ and 1M Na₂CO₃;pH, 9.6) was added to each well of a 96-well ELISA plate,^f and the plate was incubated overnight at 4°C. The plate was then washed 5 times with PBS solution with 0.05% Tween 20 with an automated plate washer.^g Each serum sample was diluted 1:100 in a 5% nonfat milk solution that was made with PBS solution and 0.05% Tween 20, and 50 μL of the diluted serum sample was added to each of 4 wells on the prepared culture plate. The plate was incubated for 1 hour at 37°C and washed as described. Then, 1:10 dilutions of a Vero cell supernatant containing inactivated WNV antigen and 1:10 dilutions of a supernatant from non–WNV-infected Vero cells (control) were prepared. For each set of 4 wells that contained serum from the same horse, 50 μL of the diluted Vero cell supernatant containing the WNV antigen was added to each of 2 wells and 50 μL of the diluted control supernatant was added to each of the remaining 2 wells. The plate was then incubated overnight at 4°C and washed as described. Next, 50 μL of 1:20,000 biotinylated anti–WNV-217 (an mAb directed against the envelope protein of WNV) in PBS solution with 0.05% Tween 20 was added to each well.The plate was incubated for 1 hour at 37°C and washed as described, and 50 μL of a streptavidin-peroxidase conjugate^h was added to each well. The plate was incubated for 1 hour at 37°C and washed as described, and 50 μL of a substrate buffer (33.3 mmol citric acid and 66.7 mmol NaH₂PO₄;pH,5.0) combined with 3,3′,5,5′-tetramethl-benzidineⁱ (130 μg/mL) and 0.012% hydrogen peroxide was added to each well.The plate was incubated for 15 minutes in the dark, and the reaction was stopped by the addition of an equal volume of 1N H₂SO₄ to each well. Finally, the plate was read at a wavelength of 450 nm by use of an automated microplate ELISA reader^j and data analysis software.^k The PNR for each sample was calculated by dividing the mean WNV antigen optical density by the mean control optical density. A PNR > 3.3 was considered a positive response.

The anti-WNV SN titers in serum samples obtained from the mares on days 0, 14, 46, 74, and 110 were determined at the Animal Health Diagnostic Center, Cornell University, Ithaca, NY.

STATISTICAL ANALYSIS

Descriptive statistics were generated for all study horses. For the 2 groups of mares that were vaccinated against WNV, the respective PNRs for each immunoglobulin isotype were assessed. The data distribution for each outcome variable was assessed by means of the D’Agostino-Pearson omnibus test for normality, and results indicated that all variables had a Gaussian distribution. A 2-way ANOVA for repeated measures with Bonferroni correction for multiple comparisons was used to compare the PNRs for each immunoglobulin isotype between the 2 treatment groups (ie, mares vaccinated with the recombinant canarypox virus–vectored WNV vaccine and mares vaccinated with the killed WNV vaccine). Independent variables included in the models were treatment group and day of sample collection. All analyses were performed with a commercial software program,^l and values of P < 0.05 were considered significant.

Results

HORSES NATURALLY INFECTED WITH WNV

Serum samples from 10 horses with neurologic signs and natural WNV infection indicated by positive values in a WNV IgM capture ELISA and from 3 noninfected control horses without WNV-specific IgM antibodies were tested for WNV-specific IgG isotypes (Figure 1). The noninfected horses were seronegative for all 6 IgG isotypes tested.The WNV-specific IgG response in naturally infected horses was characterized by high concentrations of IgG1 (also known as IgGa). This was confirmed by 2 isotype-specific mAbs, 1 against IgG1 only and the other against IgG1 and IgG3 (IgG1/3). Another mAb against equine IgG3 and IgG5 (IgG3/5) that was included in the isotype testing did not result in any positive anti-WNV antibody responses. It was concluded that WNV-specific IgG detected by mAb IgG1/3 was composed of IgG1 only. High anti-WNV IgG1 PNRs were found in all sera positive for anti-WNV IgM and exceeded the respective anti-WNV IgM PNRs for 8 of 10 serum samples.The PNRs for all other WNV-specific IgG isotypes in naturally infected horses were below the positive cutoff of the assays, with the exception of 1 horse that was seropositive for IgG6 (PNR, 4.6); IgG6 is also known in horses as IgGc.None of the WNV-infected, anti-WNV IgM positive horses had developed anti-WNV IgG3/5 (also known as equine IgG[T]), IgG5, or IgG4/7 (also known as equine IgGb) antibodies.

Figure 1—

Serum WNV-specific immunoglobulin isotypes for horses with neurologic signs that were (A; n = 10) and were not (B; 3) naturally infected with WNV. Serum samples were obtained from horses in the United States before WNV vaccines were commercially available. All isotype responses were measured by capture ELISAs, and results are expressed as PNRs. Horses with a WNV-specific IgM PNR > 3.3 were considered infected with WNV, whereas those with a WNV-specific IgM PNR ≤ 3.3 were considered uninfected. For all isotypes,the bars represent the mean and the brackets represent the SE. The dotted line indicates the PNR cutoff (3.3) for a positive result for all capture ELISAs.

VACCINATION OF A WNV-NAÏVE STALLION

The immunoglobulin profile for the WNV-naive stallion that was vaccinated with the first WNV vaccine^a to become commercially available in the United States was summarized (Figure 2). The stallion became seropositive for WNV-specific IgM on day 5 (5 days after administration of the first dose of vaccine) and remained seropositive until day 145. The WNV-specific IgM response peaked on day 12 (PNR, 11.1), which was 12 days after administration of the first dose of vaccine, and again on day 34 (PNR, 14.4), which was 9 days after administration of the second dose of the vaccine. The stallion was first seropositive for WNV-specific IgG1 on day 12 and remained seropositive for IgG1 throughout the remainder of the observation period (ie, 9 months).The anti-WNV IgG1 response was detected by 2 mAbs, 1 against IgG1 and the other against IgG1/3. The WNV-specific IgG1 response peaked on day 25 (PNR, 27) as determined by the IgG1/3 mAb, whereas it peaked on day 55 (PNR, 25) as determined by the IgG1 mAb. In addition, the stallion was seropositive for WNV-specific IgG4/7 between days 34 and 81; however, it was never seropositive for WNV-specific IgG3/5, IgG5, IgG6, IgE, or IgA.

Figure 2—

Serum WNV-specific IgM (white circles), IgG1 (black circles), IgG1/3 (black squares), IgG4/7 (black triangles), IgG3/5 (black diamonds), IgG5 (white squares), IgG6 (upside-down white triangles), IgE (white diamonds), and IgA (white triangles) responses over time for a 16-year-old Icelandic stallion that was imported into the United States from Iceland and subsequently vaccinated with a killed WNV vaccine^a (ie, the first WNV vaccine that was commercially available in the United States) on days 0 and 25. All isotype responses were measured by the use of separate capture ELISAs. The serum samples on days −10 and −3 were obtained immediately before exportation from Iceland and within 48 hours after importation into the United States, respectively. West Nile virus has never been identified in Iceland, and the horse was assumed to be naïve to WNV prior to vaccination. Black arrow represents the horse’s arrival in the United States. See Figure 1 for remainder of key.

VACCINATION OF WNV-NAÏVE MARES

The anti-WNV SN titers for mares that were vaccinated with the recombinant canarypox virus–vectored WNV vaccine^c (recombinant vaccine group) and the killed WNV vaccine^d (killed vaccine group) were summarized (Figure 3). Immediately before vaccination (day 0; baseline), the mean anti-WNV SN titers did not differ significantly between the recombinant (mean, 42; range, 6 to 96) and killed (mean, 43; range, 6 to 128) vaccine groups. For the horses of the recombinant vaccine group, the mean anti-WNV SN titer increased slightly from the baseline titer after administration of the first dose of the vaccine and then decreased to below the baseline titer immediately before administration of the second dose of the vaccine (mean, 20; range, 4 to 32). After the second dose of the vaccine was administered, the anti-WNV SN titer peaked on day 74 (mean, 384; range, 96 to 1,024). For horses of the killed vaccine group, the mean anti-WNV SN titer increased steadily after administration of both the first and second doses of the vaccine and peaked on day 74 (mean, 3,424; range, 768 to 8,192).The peak anti-WNV SN titer for the horses of the recombinant vaccine group was significantly (P < 0.001) lower than that for the horses of killed vaccine group.

Figure 3—

Anti-WNV SN titers over time for adult Icelandic mares that were imported into the United States from Iceland and vaccinated with a recombinant canarypox virus–vectored WNV vaccine^c (recombinant vaccine group; white circles; n = 7) or a killed WNV vaccine^d (killed vaccine group; black circles; 8) on days 0 and 46.West Nile virus has never been identified in Iceland, and all horses were tested for antibodies against WNV and found to be naïve prior to the initial vaccination. Data represent mean ± SE. *Within a day, mean values differ significantly (P < 0.05) between the 2 vaccination groups.

The mean PNRs of 9 WNV-specific immunoglobulin isotypes for the horses of the recombinant and killed vaccine groups throughout the observation period were analyzed (Figure 4). All horses in both groups were seronegative (PNR < 3.3) for all of the WNV-specific immunoglobulin isotypes prior to administration of the first dose of the assigned WNV vaccine. In addition, all horses in both groups remained seronegative for WNV-specific IgM throughout the observation period, which suggested that neither vaccine induced an IgM response. Similarly, all horses remained seronegative for WNV-specific IgG3/5, IgG5, IgE, and IgA throughout the observation period. Horses in both groups developed WNV-specific IgG1 responses following vaccination as measured by the IgG1 and IgG1/3 ELISAs. For the horses of the recombinant vaccine group, the WNV-specific IgG1 response increased from baseline levels following administration of the first dose of vaccine, peaked at day 14, and then decreased to near baseline levels before increasing again after administration of the second dose of vaccine and peaking at day 74.All horses in the recombinant group were seronegative for WNV-specific IgG1 on days 110 and 140. For the horses of the killed vaccine group, the WNV-specific IgG1 response steadily increased after administration of the first dose of vaccine and peaked at day 74, and all horses remained seropositive for IgG1 for the remainder of the observation period.Again, the IgG1 and IgG1/3 capture ELISAs resulted in a fairly similar response pattern, whereas the IgG3/5 ELISA did not detect any antibody response to vaccination. This further supports that the anti-WNV isotype response detected with the IgG1/3 capture ELISA was solely composed of IgG1. The mean PNR for WNV-specific IgG1 for the killed vaccine group as determined by the IgG1 ELISA was significantly higher than that for the recombinant vaccine group on days 46, 74, and 110. The anti-WNV IgG1 PNR for the killed vaccine group as determined by the WNV-specific IgG1/3 ELISA was significantly higher than that for the recombinant vaccine group on days 46, 110, and 140.

Figure 4—

Serum WNV-specific IgM (A), IgG1 (B), IgG1/3 (C), IgG4/7 (D), IgG3/5 (E), IgG5 (F), IgG6 (G), IgE (H), and IgA (I) responses after vaccination of WNV-naïve Icelandic mares.The horses were vaccinated with a recombinant canarypox virus–vectored WNV vaccine^c (recombinant vaccine group; white circles; n = 7) or a killed WNV vaccine^d (killed vaccine group; black circles; 8) on days 0 and 46. All isotype responses were measured by separate capture ELISAs. Data represent mean ± SE PNRs. Notice that the y-axis differs among graphs. The dotted line indicates the PNR cutoff for a positive response (3.3). *Within a day, mean values differ significantly (P < 0.05) between the 2 vaccination groups.

Four of 7 horses in the recombinant vaccine group and 5 of 8 horses in the killed vaccine group were seropositive for WNV-specific IgG4/7 on day 74 (28 days after administration of the second dose of the assigned vaccine). However, this response was transient, and all horses were seronegative for WNV-specific IgG4/7 for the remainder of the observation period.Three horses in the recombinant vaccine group and 2 horses in the killed vaccine group were seropositive for WNV-specific IgG6 on day 46 or 74 after administration of the first dose of the assigned vaccine. This response was likewise low and transient, and all horses were seronegative for WNV-specific IgG6 for the remainder of the observation period.

Discussion

Results of the present study indicated that horses naturally infected with WNV and adult WNV-naïve horses that were administered 2 doses of a recombinant canarypox virus–vectored WNV vaccine (recombinant vaccine group) or killed WNV vaccine (killed vaccine group) 6 weeks apart developed similar IgG response profiles, with IgG1 being the predominant isotype induced.The WNV-infected horses also developed a substantial IgM response, whereas the horses in the recombinant and killed vaccine groups remained seronegative for WNV-specific IgM during the approximately 5-month observation period following administration of the first dose of the assigned vaccine. This suggested that the currently available WNV IgM capture ELISA can be used to effectively distinguish WNV-infected horses from those vaccinated with the recombinant canarypox virus–vectored vaccine and the killed WNV vaccine used in the present study. However, a WNV-naïve stallion that was vaccinated with another killed WNV vaccine (ie,the first WNV vaccine to become commercially available in the United States) developed a detectable IgM response following vaccination that was similar to the IgM response induced in 20 horses and ponies that were vaccinated with the same vaccine in another study,¹⁵ which suggested that that particular vaccine may interfere with the use of the WNV IgM capture ELISA for detection of WNV-infected horses. These findings indicated that currently available WNV vaccines differ in their ability to induce an IgM response and emphasize the importance of obtaining information about which specific WNV vaccines have been administered to horses before deciding whether to use the WNV IgM capture ELISA to diagnose WNV infection or when interpreting results of WNV-specific antibody tests.

In the present study, horses in the recombinant and killed vaccine groups and the stallion that was vaccinated with the first WNV vaccine to become available in the United States developed IgG1 responses within 14 days after administration of the initial dose of vaccine. For the horses of the recombinant vaccine group, the IgG1 response peaked at day 14 (14 days after administration of the initial dose of vaccine), was undetectable by day 46, peaked again on day 74 (28 days after administration of the second dose of vaccine), and was undetectable again by day 110 in a manner that mirrored the anti-WNV SN antibody response. This suggested that the recombinant canarypox virus–vectored WNV vaccine induced only transient and short-lasting IgG1 responses in WNV-naïve horses after administration of 2 doses 6 weeks apart. Investigators of another study²¹ likewise reported that administration of the same recombinant canarypox virus–vectored WNV vaccine to WNV-naïve horses induced only a transient neutralizing antibody response. In that study,²¹ horses were administered an initial dose of the vaccine (day 0) and then received booster doses of the vaccine 28, 300, and 454 days later. Serum anti-WNV antibody titers as determined by a plaque reduction neutralization test peaked 14 days after the booster vaccination on day 28 and were undetectable by day 300 when the third dose of the vaccine was administered.²¹ Unfortunately, anti-WNV antibody titers were not determined between days 91 and 300 of that study²¹; therefore, it is unknown when or how quickly the vaccine-induced antibody response began to decline.

The killed WNV vaccine that was administered to the horses in the killed vaccine group stimulated an IgG1 response that was of significantly greater magnitude, compared with that induced by the recombinant canarypox virus–vectored WNV vaccine, and that response did not decrease between administration of the first and second doses. The IgG1 response peaked 74 days after administration of the initial dose and remained positive (PNR, > 3.3) until the end of the observation period (day 140). Results of a study²² in which serum anti-WNV antibody titers as determined by a plaque reduction neutralization test were compared between horses naturally infected with WNV and horses vaccinated against WNV with a 2-dose series of a killed virus vaccine indicate that WNV-infected horses maintained anti-WNV antibody titers > 1:100 for 5 to 7 months after the onset of disease, whereas only 28 of 84 (33%) vaccinated horses maintained anti-WNV antibody titers > 1:100 for 5 to 7 months after receiving the second dose of the vaccine. The authors of that study²² concluded that some horses may respond poorly to WNV vaccines and suggested that certain horses or horses in areas with high mosquito populations be vaccinated against WNV every 6 months.

In the present study some but not all horses in both the recombinant and killed vaccine groups developed a transient IgG4/7 response that peaked on day 74, whereas none of the WNV-infected horses were seropositive for IgG4/7, which was likely a reflection of sample collection in relation to initial WNV exposure. Horses typically develop neurologic signs 2 to 9 days after WNV infection, and 1 horse that was experimentally infected with WNV developed neurologic signs 8 days after infection.⁷ Thus, serum samples from WNV-infected horses with neurologic signs are generally obtained early during infection before a potential IgG4/7 response is induced. Immunoglobulin G3/5, IgG5, IgE, and IgA were not detected in the sera of the WNV-infected horses or the WNV-naïve horses that were subsequently vaccinated against WNV, and IgG6 was detected in low and transient levels in only 1 of the 10 horses with neurologic signs that were infected with WNV and a minority of the horses vaccinated with either vaccine. On the basis of these results, we cannot completely exclude that horses naturally infected with WNV and those that survive WNV-induced meningoencephalitis will not develop immunoglobulin isotypes other than IgM and IgG1 because all of the serum samples evaluated in the present study were likely obtained during the acute stage of the infection.

The American Association of Equine Practitioners recommends that horses be vaccinated against WNV annually in the spring prior to the peak mosquito season.²³ On the basis of a review of WNV IgM capture ELISA results for serum samples from horses throughout the United States submitted to the Cornell University Animal Health Diagnostic Center from 2009 through 2011, WNV infection is most frequently diagnosed between July and October. Results of the present study indicated that the IgG response induced by the recombinant canarypox virus–vectored WNV vaccine had a lower magnitude and lasted for a shorter duration than did that induced by the killed WNV vaccine. These results suggested that some WNV vaccines may not provide sufficient protection against infection for an entire year, and we suggest that horses be vaccinated at least 1 month before peak mosquito season so that vaccine-induced IgG1 concentrations will be maximal during the period of highest risk for WNV infection. Further investigation is necessary to determine whether annual vaccination of horses against WNV over time will have an anamnestic effect that will cause an increase in the magnitude or longevity of the WNV-specific immunoglobulin response.

Horses have 11 immunoglobulin isotypes (IgM, IgD, IgG1 to IgG7, IgE, and IgA).²⁴ The IgM and IgG isotypes likely contribute to the neutralizing effects of anti-WNV antibodies, which protect horses from infection.^7,20 The early humoral response to infection is characterized by the production of IgM, which binds to pathogens. Once IgM is bound to a pathogen, the pentameric IgM-antigen complexes activate complement and increase phagocytosis. Equine IgG1 binds to Fc receptors, induces respiratory burst from leukocytes, and activates the classical complement pathway.²⁵ Additionally, in horses, IgG1 is produced in utero,²⁶ and naïve peripheral B cells are primarily positive for IgM or IgG1.^19,27 Thus, IgG1 might contribute to the initial humoral response against pathogens in horses.²⁷ In the present study, WNV-specific IgG1 was detectable within 14 days after vaccination and long before IgG4/7 was observed. Immunoglobulin G1 and IgG4/7 provide protection against intracellular pathogens such as equine influenza virus²⁸ or EHV-1.²⁹ Therefore, a vaccine that induces a strong and long-lasting WNV-specific IgG1 response will likely provide effective protection against WNV infection, even in the absence of an IgM response. This is likely the mechanism of protection against WNV infection for many of the commercial WNV vaccines approved for use in horses.

We did not expect the immunoglobulin response to WNV infection or vaccination to be dominated by a single IgG isotype. In horses, the predominant serum IgG isotypes are IgG4 and IgG7,²⁴ and results of another study²⁵ suggest that these 2 isotypes are primarily involved in the defenses against intracellular pathogens. For example, vaccination of horses against EHV-1 induced an EHV-1–specific IgG4/7 response that was significantly increased from that prior to vaccination.²⁹ Furthermore, when vaccinated horses were experimentally exposed to a neuropathogenic strain of EHV-1, horses with low serum concentrations of EHV-1–specific IgG4/7 developed neurologic signs, whereas horses with high serum concentrations of EHV-1–specific IgG4/7 did not develop neurologic signs.²⁹ The IgG4/7 isotypes also have key roles in the defense against infection with Rhodococcus equi³⁰ and equine influenza virus.²⁸ Conversely, in the present study, a WNV-specific IgG4/7 response was not detected in any of the WNV-infected horses and was delayed and transient in the horses vaccinated against WNV Delayed class-switching from IgG1 to IgG4/7 might explain the lack of a detectable WNV-specific IgG4/7 response in the WNV-infected horses with neurologic signs. Regardless, the findings of the present study suggested that IgG4/7 does not substantially contribute to the immune response and defense against acute WNV infection in horses.

The lack of detection of WNV-specific IgG4/7 in the present study might also have been a function of the capture ELISA used. In the studies of immune responses to equine influenza virus,²⁸ EHV-1,²⁹ and R equi,³⁰ IgG4/7 was detected by the use of antigen-specific ELISAs in which the antigen was directly coated onto the culture wells and antibodies in the serum samples were allowed to bind to the antigen in a concentration-dependent manner (ie, IgG4/7, IgG3/5, and IgG1 in the serum sample are all competing for antigen binding sites on the plate). Given that IgG4 and IgG7 are the predominant IgG isotypes in equine serum, it is not surprising that IgG4/7 responses dominate assays that use antigen coating. In contrast, the WNV capture ELISA uses a different equine mAb for each specific isotype to coat the culture wells. Thus, competition among the various IgG isotypes in the serum sample does not occur in the capture ELISA, and the binding of WNV-specific IgG from serum represents the proportion of WNV-specific IgG within the total IgG isotype fraction. Consequently, binding of the WNV-specific IgG is low if an isotype fraction is composed of high concentrations of antibodies with different antigen specificities. In the absence of quantitative standards, both antigen-specific ELISAs and capture ELISAs can only provide relative estimations of the amounts of antibody isotypes within a sample.

In equine serum, the concentration of IgG3/5 is similar to that of IgG1.²⁴ Several killed virus vaccines, including those against equine influenza virus²⁸ and EHV-1,²⁹ induce predominantly IgG3/5 antibody responses. However, in the present study, vaccination of horses against WNV with either a recombinant canarypox virus–vectored vaccine or a killed virus vaccine failed to induce a detectable IgG3/5 response. It is possible that the antigenic composition of currently available WNV vaccines predominantly induces IgG1 to IgG4/7 class-switching. It is also possible that the Icelandic horses used in this study were less immunologically primed to synthesize IgG3/5 isotypes than other horses because they were raised in Iceland and were immunologically naïve to several pathogens endemic in the United States, including but not limited to equine influenza virus, EHV-1, and Streptococcus equi. Nevertheless, WNV-specific IgG3/5 responses were also absent in horses naturally infected with WNV.

Immunoglobulin G6 is generally present in low amounts in equine serum. In the present study, 3 horses in the recombinant vaccine group and 2 horses in the killed vaccine group were seropositive for IgG6 on day 46 or 74 after administration of the first dose of the assigned vaccine. Interestingly, 3 of those 5 horses were also seropositive for IgG4/7 on day 74. It is possible that the class-switching pathways from IgG1 to IgG4/7 and IgG6 share common steps, such that horses that produced IgG4/7 isotypes also produced IgG6. Regardless, IgG6 does not substantially contribute to the defense against pathogens and is not considered a good indicator of response to vaccination.²⁵

Results of the present study indicated that WNV infection induces the production of WNV-specific IgM and IgG1 in horses. Vaccination of adult WNV-naïve horses against WNV also induced a primarily IgG1 response. Some, but not all, currently available WNV vaccines also induce an IgM response, which could interfere with the ability of the WNV IgM capture ELISA to detect WNV-infected horses. In the present study, the killed virus vaccine induced WNV-specific IgG1 responses and anti-WNV SN titers of greater magnitude and that lasted for a longer duration than did the recombinant canarypox virus–vectored vaccine. Given that virus neutralization is the primary mechanism by which these vaccines are believed to protect horses against WNV infection, it is likely that the killed virus vaccine offers better overall protection against WNV infection and neurologic disease than does the recombinant canarypox virus–vectored vaccine. In the present study, the WNV-specific IgG1 response and anti-WNV SN titers peaked 74 days after vaccination and declined to nearly undetectable levels by 140 days after vaccination. Therefore, we recommend that horses be vaccinated against WNV 1 to 2 months prior to peak mosquito season to ensure that the vaccine-induced immune response will be maximal during the period of greatest risk for WNV infection (July to October in the United States) and that WNV-specific IgG1 responses be monitored annually in individual horses after vaccination and before WNV season starts.

ABBREVIATIONS

EHV

Equine herpesvirus

mAb

Monoclonal antibody

PNR

Positive-to-negative ratio

SN

Serum neutralization

WNV

West Nile virus