Co-occurrence of West Nile Fever and circovirus infection in a goose flock in Hungary

Abstract

Co-apparition de la fièvre de la rive ouest du Nil et de l'infection à circovirus dans un troupeau d'oies en Hongrie

Les auteurs ont étudié un cas de fièvre de la rive ouest du Nil (WNF) caractérisé par des symptômes nerveux et de la mortalité dans un troupeau de 3600 oies âgées de six semaines. La maladie était caractérisée par de l'ataxie, des torticolis et opisthotonos intermittents, de l'incoordination, des mouvements rythmiques latéraux de la tête, des mouvements du cou, et une position anormale de la tête. La mortalité est apparue 4–5 jours après le début des symptômes. La mortalité moyenne journalière a été de 5–15, atteignant 14% (au total) sur une période de 6 semaines. Il n'y avait pas de lésions pathologiques macroscopiques évidentes, mais dans quelques cas des foyers de 3–6 mm de diamètre ont été observés en surface du cerveau ou sur sa section. L'histopathologie a révélé des infiltrations lymphohistioïdes périvasculaires et une prolifération de cellules gliales au niveau du tronc cérébral, du cervelet, du cortex, de la moelle épinière ainsi qu'une dégénérescence des fibres neurales dans la moelle épinière. En plus des lésions causées par le WNV dans le cerveau, caractéristiques de l'infection à circovirus, comme une déplétion en lymphocytes, une vacuolisation et des corps d'inclusion intra-cytoplasmiques contenant des particules ressemblant à des circovirus ont été observées en microscopie optique et électronique dans la bourse cloacale.

L'infection due au virus de la rive ouest du Nil a été confirmée par RT-PCR, amplification d'acide nucléique spécifique du virus, à partir d'échantillons de tissu du cerveau. L'analyse de la séquence nucléotidique des produits de la PCR a montré qu'il y avait, pour la région NS5 testée 99% d'identité avec la souche IS-98 ST1 isolée d'une cigogne en Israël, en 1998, et avec les souches de virus de la rive ouest du Nil apparues aux USA en 1999. Des titres élevés en anticorps fluorescents ont été détectés dans les échantillons de sérums provenant de troupeaux affectés. Ces résultats ont été confirmés par séroneutralisation à partir d'une sélection de sérums.

Gleichzeitiges Vorkommen einer West-Nil-Fieber- und einer Circovirusinfektion in einem Gänsebestand in Ungarn

Es wurde ein Ausbruch von West-Nil-Fieber (WNF) untersucht, der durch schwere neurologische Symptome und Todesfälle in einer Herde von 3600 sechswöchigen Gänsen charakterisiert war. Ataxie, intermittierender Tortikollis und Opisthotonus, Inkoordination, rhythmische Seitwärtsbewegungungen des Kopfes, Verdrehungen des Nackens und Abnormale Kopfhaltungen kennzeichneten die Erkrankung. Der Tod trat innerhalb von 4–5 Tagen nach Auftreten der ersten klinischen Symptome ein. Im Durchschnitt starben täglich 4–15 Tiere, wobei insgesamt eine Mortalitätsrate von 14% in einem Zeitraum von 6 Wochen erreicht wurde. Es gab keine einheitlichen pathologisch-anatomischen Veränderungen, aber in einigen Fällen wurden auf der Oberfläche oder auf dem Querschnitt des Hirns gelblich-graue Herde mit einem Durchmesser von 3–6 mm beobachtet. Die histopathologische Untersuchung ließ perivaskuläre lymphohistozytäre Infiltration und Gliazellproliferation in Hirnstamm, Zerebellum, Kortex und Rückenmark sowie Degeneration von Nervenfasern im Rückenmark erkennen. Zusätzlich zu den durch WNV verursachten Hirnläsionen wurden bei der licht- und elektronenmikroskopischen Untersuchung in der Bursa Fabricii für eine Circovirusinfektion charakteristische Veränderungen wie Lymphozytenverlust, Vakuolisierung und basophile intrazytoplasmatische Einschlusskörper mit Circovirus-ähnlichen Partikeln gefunden. Die West-Nil-Virusinfektion wurde durch eine RT-PCR-Amplifikation der virusspezifischen Nukleinsäure aus einer Gewebeprobe des Gehirns bestätigt. Basierend auf der Nukleotidsequenzanalyse des PCR-Produkts wurde eine 99%ige Übereinstimmung der getesteten NS5-Region mit dem IS-98 ST1-Stamm aus einem Storch in Israel im Jahr 1998 und mit West-Nil-Virusstämmen, die 1999 in den USA aufgetreten waren, festgestellt. Unter Verwendung des indirekten Fluoreszenz-Antikörpertests wurden in den aus dem betroffenen Bestand eingesandten Seren hohe Antikörpertiter gegen das Virus nachgewiesen. Dieses wurde ebenso im Neutralisationstest mit ausgewählten Seren bestätigt.

Infección simultánea de un lote de gansos con virus del Oeste del Nilo y circovirus en Hungría

Los autores investigaron un brote de infección por el virus del oeste del Nilo (WNF) caracterizado por síntomas neurológicos graves y muerte en un lote de 3600 gansos de seis semanas de edad. Se observó ataxia, tortícolis intermitente y opistótonos, incoordinación, movimiento rítmico lateral de la cabeza, torcedura del cuello y posición anormal de la cabeza. La muerte ocurría a los 4–5 días del inicio de los síntomas. La mortalidad media diaria fue de 5–15, alcanzando el 14 % (en total) durante un periodo de 6 semanas. No hubo lesiones macroscópicas consistentes, pero en algunos casos se observaron pequeños focos de un color amarillo-grisáceo de 3–6 mm de diámetro en la superficie y corte del cerebro. Histopatológicamente se observó infiltración perivascular linfohistiocítica y proliferación glial en el tronco de encéfalo, cerebelo, córtex y médula espinal así como degeneración de fibras neuronales en la médula espinal. Además de estas lesiones causadas por el WNV en el cerebro, también se observaron lesiones características de infección por circovirus, como son la depleción linfoide, la vacuolización y los cuerpos de inclusión intracitoplasmáticos basófilos en la bolsa de Fabricio mediante el estudio al microscopio óptico y electrónico.

La infección por el virus del oeste del Nilo se confirmó mediante amplificación por RT-PCR de ácido nucleico de un virus a partir de muestras de cerebro. En base al análisis de la secuencia nucleotídica de los productos de PCR, se encontró un 99% de identidad en la región NS5 con la cepa IS-98 ST1 aislada de una cigüeña en Israel en 1998, y con cepas de virus del oeste del Nilo aisladas en USA en 1999. Mediante un test de inmunofluorescencia indirecta se detectaron niveles altos de anticuerpos frente al virus en las muestras de suero provenientes del lote afectado. En algunas muestras de suero seleccionadas, estos resultados se confirmaron mediante pruebas de neutralización.

The authors investigated an outbreak of West Nile Fever characterized by severe neurological symptoms and death in a flock of 3600 6-week-old geese. Ataxia, intermittent torticollis and opisthotonus, incoordination, rhythmic side-to-side movement of the head, wriggling of the neck and abnormal head position were features of the disease. Death occurred within 4 to 5 days after the clinical signs appeared. The average daily mortality was 5 to 15, reaching 14% (in total) over a period of 6 weeks. There were no consistent gross pathological lesions, but in a few cases yellowish-grey foci of 3 to 6 mm in diameter were observed on the surface or transection of the brain. Histopathology revealed perivascular lymphohistiocytic infiltration and glia cell proliferation in the brainstem, cerebellum, cortex and spinal cord as well as degeneration of neural fibres in the spinal cord. In addition to the lesions caused by the West Nile Virus in the brain, characteristics of circovirus infection such as lymphocyte depletion, vacuolization and basophilic intra-cytoplasmic inclusion bodies containing circovirus-like particles were seen by light and electron microscopy in the cloacal bursa. West Nile Virus infection was confirmed by reverse transcriptase-polymerase chain reaction amplification of virus-specific nucleic acid from tissue samples of the brain. Based on the nucleotide sequence analysis of the polymerase chain reaction products, 99% identity was found on the tested NS5 region with the IS-98 ST1 strain isolated from a stork in Israel in 1998, and with West Nile Virus stains emerging in the USA in 1999. Using an indirect fluorescent antibody test, high antibody titres against the virus were detected in the serum samples submitted from the affected flock. In selected sera this was confirmed by neutralization antibody test as well.

Introduction

West Nile Virus (WNV) that belongs to the family of Flaviviridae in the Japanese encephalitis serocomplex group is the causative agent of West Nile Fever (WNF) (Calisher et al., 1989). It has a wide distribution in Africa, West Asia and the Middle East, but more recently outbreaks have been reported from Europe, Israel and the US as well (Hindiyeh et al., 2001; Marfin et al., 2001; Murgue et al., 2001; Zeller & Schuffnecker, 2004). WNV is transmitted through various species of adult Culex mosquitoes to a variety of mammals and birds. WNV is maintained in an enzootic cycle involving Culicine mosquitoes and birds. Viral amplification occurs in the bird–mosquito–bird cycle (Tsai, 2000; Steele et al., 2000).

Most WNV infections are mild and often clinically unapparent; however, approximately 20% of the infections result in clinical signs ranging from mild fever to fatal encephalitis in humans, as well as neurological signs with fatalities in horses (Murgue et al., 2001), free-flying and captured birds (Banet-Noach et al., 2003).

A neuroparalytic disease of young geese caused by WNV was first observed in Israel in 1997 (Guy & Malkinson, 2003). The birds exhibited acute neurological signs including paralysis, somersaulting, paddling, opisthotonus and incoordination. Morbidity was very high, ranging from 20 to 60% in large flocks of geese (Office International des Epizooties, 1999), and sick geese invariably died. Geese aged from 3 to 8 weeks were most susceptible, but older geese up to 12 weeks of age were also affected. Outbreaks of the disease occurred from mid-August until the end of November each year from 1997 to 2001 in flocks of young geese that were raised outdoors in open pens and therefore exposed to biting insects. The WNV isolate obtained from sick geese during the outbreak in 1998 showed close phylogenetic relationship with the New York isolates (Lanciotti et al., 1999).

This paper describes the first occurrence of WNV encephalitis in a commercially reared circovirus-infected goose flock in Hungary.

Materials and Methods

Case history

The outbreak of the disease occurred in a young goose flock at a farm in the flood plain of the river Danube. The geese were kept outdoors in an open pen within an area well segregated by natural boundaries from other farms, but visited frequently by different species of wild waterfowl. Other poultry species were not raised at the same farm. The distance from the nearest goose flock was about 30 km. Hatch-mates originating from the same parent flock but raised at different locations were not affected by the neurological disease; however, circovirus infection was detected in this flock as well (data not shown).

The first clinical symptoms and death occurred at 6 weeks of age and continued to occur until 12 weeks of age. The average daily mortality was 5 to 15, reaching a total of 504 (14%) over a period of 6 weeks (Figure 1). On clinical examination, geese exhibited different neurological signs including ataxia, intermittent torticollis and opisthotonus, incoordination, rhythmic side-to-side movement of the head, wriggling of the neck, abnormal head position, and paralysis (Figure 2). Death occurred within 4 to 5 days after the clinical signs appeared.

Figure 1.  Daily mortality in the goose flock affected with WNV and circovirus infection.

Figure 2.  Neurological disorders, torticollis and abnormal head position.

Laboratory investigation

Seven dead geese, 7 weeks old, were submitted for routine laboratory investigation. Based on the histological lesions observed in the brain and the cloacal bursa a tentative diagnosis of viral encephalitis and circovirus infection was made. To confirm the diagnosis and clarify the aetiology of the disease, 3 days later an additional five sick animals were received from the flock to carry out exhaustive examination (including clinical signs, serology, electron microscopy, virological and molecular biological tests). One month after the submission of the sick animals, 41 serum samples were collected from the same flock for serological testing.

Gross pathology, histopathology and electron microscopy

Postmortem examinations were performed on 12 geese (seven dead and five sick) submitted for laboratory investigation from the affected flock. For histological examinations, samples were taken from the brain, thoracic and lumbar spinal cord, liver, spleen, heart, kidney, lung, small intestine, pancreas, thymus and cloacal bursa of all animals submitted. For light microscopy, the samples were fixed in 10% buffered formalin, embedded in paraffin, sectioned at 5.0 µm, and stained with haematoxylin and eosin. For transmission electron microscopy, the samples of brain and cloacal bursa from two sick birds were post-fixed in 2.5% glutaraldehyde and osmium tetroxide, and then embedded in Durcupan (Fluka Chemie AG, Buchs, Switzerland). To select appropriate areas for electron microscopy examination, semi-thin sections (1 µm) were cut and stained with 1% toluidine blue and examined by light microscopy. From the selected areas, ultra-thin sections were prepared and placed on grids, stained with uranyl acetate and lead citrate, and examined with a Philips 208S transmission electron microscope.

Bacteriological examinations from the brain, liver and heart blood were performed by routine culturing method on blood and Drigalski agar plates under aerobic conditions and with increased carbon dioxide to exclude Riemerella anatipestifer infection.

Virus isolation

Isolation of WNV was attempted from the brain and spinal cord from five affected geese showing typical neurological symptoms of the disease. Specimens of the brain and spinal cord (one pooled sample/animal) were suspended 1:10 in phosphate-buffered saline, homogenized, and clarified by centrifugation at 5000×g for 20 min, and then the supernatants were filtered through a 0.22 µm pore size filter to exclude contamination with bacteria. The filtrates were used to inoculate monolayer of Vero cells, 6-day-old embryonated hen eggs and newborn mice.

Vero cells were grown in 25 cm² tissue culture flasks using Eagle's minimal essential medium (Sigma) with 6 to 8% foetal calf serum and an antibacterial solution. After 24 h the culture medium was removed and 0.5 ml each brain and spinal cord tissue filtrate was inoculated onto the monolayer. The inoculums were allowed to adsorb for 1 h at 37°C, then removed and the medium replaced. The monolayer was incubated for 5 to 8 days and observed daily for cytopathic effect. Five blind passages were made before the samples were considered negative for cytopathic agents.

Six-day-old specific pathogen free embryonated hen eggs (Biovo Ltd, Mohács, Hungary) were inoculated with 0.2 ml tissue filtrate via the yolk sac route and the eggs were checked daily for 10 days. Embryos dying in the first 24 h were discarded; any embryos that died later were observed for macroscopic lesions and their organs (brain and liver) were checked for the presence of WNV nucleic acid by reverse transcriptase-polymerase chain reaction (RT-PCR).

Each of the five supernatants of the brain suspensions (10% in phosphate-buffered saline, pH 7.4) were inoculated into eight newborn mice by the intracerebral route in 0.01 ml volume. Another litter of mice was inoculated the same way with phosphate-buffered saline to serve as negative controls. The inoculation site was lateral to the midline into the mid-portion of one lateral hemisphere. The mice were observed daily for symptoms of encephalitis for 3 weeks. The brains of the mice that were suspected sick or died were removed, homogenized in normal rabbit serum to give a 10% suspension (Shope & Sather, 1979), and checked for WNV by RT-PCR.

RNA extraction, RT-PCR and sequencing of WNV

Viral RNA was extracted from 140 µl brain homogenates of the geese using the QIAamp viral RNA Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The oligonucleotide primer pair designed on the NS5 and 3′-untranslated regions of WNV (forward primer, 5′-GARTGGATGACVACRGAAGACATGCT-3′ and reverse primer, 5’-GGGGTCTCCTCTAACCTCTAGTCCTT-3′; Weissenböck et al., 2002) were used on the goose brain RNA extracts in a continuous RT-PCR system, as previously described (Bakonyi et al., 2005). The brain homogenates of suckling mice infected with the reference WNV strain Eg101 were used as the positive control, and cells of goose embryo fibroblast primary cell culture served as the negative control for the RT-PCR reactions.

Where clear PCR products of the previously calculated size (725 base pairs [bp]) were observed, the fragments were excised from the gel, and DNA was extracted using the QIAquick Gel Extraction Kit (Qiagen). Fluorescence-based direct sequencing was performed in both directions on the PCR products. Sequencing of the PCR products was carried out using the ABI Prism Big Dye Terminator cycle sequencing ready reaction kit (ABI, UK) according to the manufacturer's instructions. The nucleotide sequences were identified by BLAST search against gene bank databases (http://www.ncbi.nlm.nih.gov/BLAST/). The partial nucleotide sequence of the Hungarian WNV strain WNV Hu-03 was submitted to the GenBank database. It is available under accession number AY926455.

Detection and sequencing of circovirus

To confirm the circovirus infection of the animals suspected on the basis of bursal lesions seen by histological examination, spleen samples were taken for PCR detection of goose circovirus-specific nucleic acid from the same five animals as those tested for WNV.

The viral DNA was extracted from the spleen tissue using the QIAamp DNA Mini Kit (Qiagen) according to the manufacturer's directions. The extracted DNA was eluted once with 50 µl sterile distilled water. To amplify a 565 bp fragment (nucleotides 837 to 1401) of the geese circovirus genome, the PCR primer set 1 (forward primer, 5′-TAA ATG CGA GTT TGA TGT GTC T-3′ and reverse primer, 5′-CAT TTA ACC CCT TCC AAA GAG T-3′) and the reaction condition previously described (Ball et al., 2004) were used.

The PCR products were detected in 1.5% agarose gel and their sizes were determined with reference to the 100 bp DNA Ladder (Fermentas, Lithuania).

The sequencing reaction was performed using the PCR primers in both directions with the same conditions as for WNV samples. The nucleotide sequences were aligned with the sequences of German and Taiwanese goose circovirus strains obtained from the GenBank (Chen et al., 2003; Ball et al., 2004) by BioEdit sequence alignment editor software using the Clustal W multiple alignment algorithm.

Serology

Forty-six blood samples, including the five sera collected from the sick geese submitted for the laboratory investigation, were tested for antibodies to WNV by indirect fluorescent antibody test (IFAT) on an in-house antigen. The antigen has been prepared on Vero E6 cells, infected with the first Hungarian WNV isolate (strain WN_m1) (Molnár, 1982). The identity of WN _m1 was confirmed by RT-PCR and nucleotide sequencing of the amplicon. As the second antibody, fluorescein-isothiocyanate-conjugated anti-goose immunoglobulin was used.

Five of the IFAT-positive sera were tested for the presence of neutralizing antibodies on Vero E6 cells using a microneutralization method. Briefly, each of the serial 10-fold dilutions (started from 10₋₁) of the WNV (strain WN_m1) was incubated with 1:50 dilution of individual serum samples at 37°C for 1 h. The serum–virus mixtures, as well as each of the virus dilutions without serum, were used to infect the monolayer Vero cells in two parallels. Neutralization index was calculated according to the Reed–Muench method (Reed & Muench, 1938).

Results

Gross pathology

The body condition of most geese ranged from slightly thin (5/12) to moderately thin (7/12). On postmortem examination, there were no significant gross lesions present in any of the visceral organs. In three dead animals, mild-to-moderate pulmonary congestion and oedema was seen. In a few cases, distinct parts (cerebrum, cerebellum, brainstem) of the brain contained multiple pale yellow foci measuring up to 3 to 6 mm in diameter. The wall of the cloacal bursa was thinner than in normal cases of this age.

The bacterial cultures from selected organs (including the brains) were negative.

Histopathology and electron microscopy

In the brain, all of the investigated birds (12/12) had mononuclear inflammation that formed microglial nodules, perivascular clusters and meningitis. These lesions were present in different parts (cerebrum, thalamus, optic lobe, cerebellum, medulla and brainstem) of the brain, and in some cases (7/12) in the spinal cord. Meningitis and perivascular cuffs were composed primarily of lymphocytes and plasma cells and ranged from two to five cell layers in thickness (Figure 3). Lesions in the brain were most common in the molecular layer of the cerebellum. In addition to inflammatory lesions, damage to the Purkinje cells of the cerebellum and neurons of the brainstem with focal glia cell proliferation was also present (Figure 4). In one-third of the tested animals (4/12), degeneration and necrosis of neurofibrils along with diffuse glial cell proliferation was observed in the spinal cord (Figure 5). No virus-like particles were found by electron microscopy in the affected areas of the central nervous system.

Figure 3.  Perivascular lymphocyte cuff and gliosis. Haematoxylin and eosin stain, magnification×200.

Figure 4.  Purkinje-cell necrosis and focal gliosis in the cerebellum. Haematoxylin and eosin stain, magnification×200.

Figure 5.  Degeneration of neurofibrils and restorative gliosis in the spinal cord. Haematoxylin and eosin stain, magnification×200.

In all cases various degrees of lymphocyte depletion and histiocytosis were seen in the cloacal bursa, spleen and thymus. The lymphocyte depletion of most bursa follicles was 90 to 100% in degree; however, some bursa follicles were not affected. No acute inflammatory reaction (oedema, haemorrhages, heterophil granulocyte or lymphocyte infiltration) was observed. In four cases in the affected follicles, vacuolization of the stromal reticuloepithelial cells was detectable and intracytoplasmic inclusion bodies were found in the medullar and cortical region of bursa follicles (Figure 6). These inclusions appeared as globular or coarsely granular basophilic bodies. Circovirus-like particles (about 12 to 14 nm), forming a paracrystalline array, were seen by electron microscopy in the inclusion bodies. In eight cases loss of the medullary portion and proliferation of the bursal epithelial layer was observed in the affected follicules. In six of these, numerous inclusion-like bodies were seen in the cortical remnants of the atrophic bursa follicules (Figure 7). No significant histological lesions were observed in any of the other examined organs.

Figure 6.  Severe lymphocyte depletion, vacuolar degeneration and inclusion bodies (↑) in the bursa follicles. Haematoxylin and eosin stain, magnification×400.

Figure 7.  Inclusion-like bodies in the cortical remnants of the atrophic bursa follicules (↑). Haematoxylin and eosin stain, magnification×200.

Virus isolation

WNV was not isolated from any of the five samples (pooled brain and spinal cord) tested. Neither the intracerebral inoculation of mice nor the other culture systems used yielded positive results, even after several blind passages.

RT-PCR and nucleotide sequencing of WNV

Brain homogenates of five geese were tested by RT-PCR. These animals showed clinical signs of encephalitis and the previously mentioned lesions by histopathological investigations. In the case of four animals and in the positive control, RT-PCR generated clear products corresponding to the previously estimated size (725 bp). The negative controls were negative in each reaction. The amplicons were sequenced and the 673-nucleotide sequences (without the primer sequences) were identified by a BLAST search against the GenBank database. The highest level of identity was found with the WNV strain IS-98 ST1 (accession number AF481864) isolated in 1998 in Israel from a stork (Malkinson et al., 2002); and also with several WNV isolates from the US obtained from Culex pipiens mosquitoes, humans, horse, crow, grouse or flamingo (Lanciotti et al., 2002). Only four nucleotide substitutions of the Hungarian strain were observed within the investigated genome region between nucleotide positions 10125 and 10797 (referring to AF481864). The WNV strain, which was isolated from the outbreak in Romania in 1996 (Savage et al., 1999), has shown lower identity with the Hungarian strain (26 nucleotide mismatches); but at three nucleotides where the Hungarian strain differed from the Israeli and US strains, it was identical with the Romanian isolate (Figure 8). The sequence of the positive control differed unambiguously from the clinical samples; therefore, cross-contamination was ruled out.

Figure 8.  Multiple alignment of the partial nucleotide sequences of the Hungarian strain (WNV Hu-03), the Israeli strain (IS-98 STD1), a US strain (HNY1999) and a Romanian strain (RO9750) of WNV.

PCR and nucleotide sequencing of circovirus

From each of the five tested spleen samples, a DNA product of the correct size (565 bp) was obtained with goose circovirus-specific PCR. The tested circovirus sequence showed greatest (95%) homology with the German goose circovirus reference strain (accession number AJ304456).

Serology

Forty-five of the tested sera proved to be seropositive by the IFAT for antibodies against WNV and only one serum was evaluated as negative. The serum samples taken from the five sick geese had already been found to contain a moderate to high level of antibodies to WNV (titre between 80 and ≥ 1280). Only one sample out of the five had a considerably lower titre (1:80) than those collected later during the convalescent phase. The results of IFAT are shown in Figure 9. The neutralization index was higher than 2 in four of the five sera tested by the microneutralization test.

Figure 9.  Distribution of WNV-specific antibody titres of 46 goose sera tested by IFAT.

Discussion

WNV infection causing a neuroparalytic disease of young geese was observed for the first time in Hungary in 2003. The outbreak of WNF occurred in this flock from the beginning of August until mid-September, during the months when the mosquito population was the greatest. Affected geese had weight loss and exhibited a variety of neurological signs, among which the most frequent were torticollis, opisthotonus and paralysis. The clinical signs were similar to those observed in the reported field cases in Israel (Guy & Malkinson, 2003). Morbidity and mortality was moderate, reaching about 14% over a period of 6 weeks during the course of the outbreak (Figure 1). Geese from 6 to 11 or 12 weeks of age were affected and birds showing the neurological signs of the disease invariably died. As a pre-existing condition, the concurrent infection of the flock with circovirus might be an independent risk factor through its presumed immunosuppressive effect (Soike et al., 1999) that contributed to the outcome of infection in individual birds.

Although WNV was readily detected from the brain tissues of sick geese by RT-PCR, other organs of the same animal yielded negative results and we were not able to isolate the virus either by mice inoculation or in Vero cell cultures or embryonated hen eggs. The RT-PCR and sequencing of the product, however, unambiguously identified the presence of WNV in the brain of the geese, and hence this virus was considered the causative agent of the encephalitis in the affected birds. It has been reported that the isolation of flaviviruses is mainly possible in the early viraemic phase and the early appearance of neutralizing antibodies often hinders the virus isolation, although the viruses are present in the host (Vaughn et al., 1997). Swayne et al. (2001) reported that WNV could be isolated from experimentally infected 2-week-old goslings up to day 10 post-inoculation (p.i.) (day 10 p.i. in contact animals) and anti-WNV antibodies emerged at day 5 p.i. (day 14 p.i. in contact animals). Their results correspond with previous observations; namely, that after the emergence of anti-viral antibodies, the isolation of WNV is usually not possible. They did not provide data about the last time point when WNV antigen was still detectable in the brain by immunohistochemistry and they did not perform RT-PCR on the brain samples. In our report, geese were older and have been infected naturally; therefore, the two situations are hard to compare, and the exact time of the infection in our case cannot be identified. Yet, because already the serum samples taken from the sick geese were positive, it is not surprising that the virus isolation was not successful.

Although the positive results of isolation or the nucleic acid amplification test are diagnostic, low sensitivity after the acute stage of the disease precludes their use as the only routine diagnostic tests. Therefore, histopathology can and should be used to investigate probable cases of WNV encephalitis since some lesions in the brain such as mononuclear meningitis, perivascular clusters and microglial nodules, most extensively in the brainstem and medulla, may give a strong indication of WNV encephalitis in geese. However, this should be further confirmed by the detection of specific antibodies to the virus to exclude other possible agents that may cause similar lesions in the brain.

WNV is considered to be endemic in Africa, southern Europe and the Middle East (Hayes, 1989) and there is evidence for viral dispersion between these countries and Europe by migrating birds (Malkinson & Banet, 2002). It was also reported that WNV was introduced into the Middle East by migrating white stork (Malkinson et al., 2002) for which one of the most frequent destinations are the wetlands in Hungary. The location where the reported outbreak of WNV encephalitis occurred was one of the favourite nesting places for white storks, and therefore they may be considered the most probable source of infection. This hypothesis is strongly supported by the result of the virus characterization that showed close phylogenetic relationship between the Hungarian isolate and that detected in storks in Israel.

Circulation of WNV in Hungary has been known since 1969 (Koller et al., 1969) and its presence has been proven by virus isolations. During systematic field work seeking to find natural foci of Arboviruses, two Hungarian WNV strains have been isolated from organs of trapped rodents (Molnár, 1982). In spite of these facts, no human cases caused by WNV occurred before 2003. In August 2003 WNV was proved as the causative agent of human diseases with neurological symptoms for the first time (Ferenczi et al., 2005). Serum samples of the members of the farmer's family attending the affected geese flock and living at the same area were examined retrospectively for specific WNV antibodies. The infection of one member of the family was confirmed by demonstration of a high titre of IgG antibody with vanishing specific IgM in the serum sample taken several months after his illness (data not shown). He contracted the mild form of WNV infection as a febrile illness accompanied by malaise, anorexia, headache and myalgia. During the daily care of the geese, the members of the farmer's family were obviously exposed to the mosquitoes together with the birds. Therefore, mosquitoes could be the potential source of human infections as well.

Translations of the abstract in French, Germany and Spanish are available on the Avian Pathology website.

The authors thank Dr Csaba Drén (Veterinary Medical Research Institute, Budapest) for providing fluorescein-isothiocyanate-conjugated rabbit anti-goose immunoglobulin as a generous gift, and Mrs Stoll Gyuláné, Mrs Ráczné Mészáros Ágnes, Mrs Kaposi Tamásné and Ms Edit Fodor for their technical assistance.