During a 9 day period, lethargy, low activity, anorexia and ruffled feathers were observed among some of the jackdaws inoculated with NY99, Greece-10, Ita09 and 578/10. Birds died within 24–48 h after onset of clinical symptoms. Among the five jackdaws that were inoculated with NY99 and Ita09 and followed for survival, three (60 %) died; among the five jackdaws inoculated with 578/10, two (40 %) died; and among the six jackdaws inoculated with Greece-10, three died (50 %). In contrast, all five birds inoculated with FIN survived the infection (Table 1). Comparison of the survival curves of the birds inoculated with the different virus strains revealed a significant difference between the survival curves of jackdaws infected with FIN compared with Ita09 (Fig. 1; P=0.05). Median day of death was 7 days post-infection (p.i.) for birds that succumbed due to infection with NY99, Ita09 and 578/10, and 8 days p.i. for birds that died due to infection with Greece-10 (Table 1).

We determined WNV viraemia profiles for the jackdaws inoculated with the five different virus strains and found that all birds developed viraemia within 96 h p.i (Table 1; Fig. 2). For NY99-infected birds, peak serum viraemia titres ranged from 4.1 to 9.6 log₁₀RNA copies ml⁻¹ (mean 6.2 log₁₀RNA copies ml⁻¹). Birds infected with the Greece-10 strain had peak serum viraemia titres ranging from 4.2 to 8.4 log₁₀RNA copies ml⁻¹(mean 5.9 log₁₀RNA copies ml⁻¹). For FIN-infected birds, peak serum viraemia titres ranged from 3.3 to 5.1 log₁₀RNA copies ml⁻¹(mean 4.0 log₁₀RNA copies ml⁻¹). Ita09-infected birds displayed peak serum viraemia titres ranging from 3.4 to 7.5 log₁₀RNA copies ml⁻¹ (mean 4.7 log₁₀RNA copies ml⁻¹). Among jackdaws infected with the 578/10 strain, peak serum viraemia titres ranged from 3.1 to 6.9 log₁₀RNA copies ml⁻¹ (mean 5.0 log₁₀RNA copies ml⁻¹).

These data demonstrate that birds inoculated with NY99 and Greece-10 had the highest mean peak viraemia titres, followed by 578/10 and Ita09, with the lowest viraemia peaks for FIN-infected birds. Mean peak viraemia titres were significantly different between Greece-10 and FIN (P=0.02) and NY99 and FIN (P=0.03). Onset of viraemia occurred significantly later for FIN-infected birds compared with Greece-10 (P=0.02), NY99 and 578/10 (P=0.03, for both) (Table 1). The peak of viraemia was reached the earliest by birds inoculated with Greece-10, followed by the other viral strains (Table 1), but differences were not statistically significant (P=0.8). In terms of duration, viraemia lasted the longest for FIN and 578/10, followed by Greece-10, NY99 and Ita09 (Table 1), but these differences were also not statistically significant (P=0.4).

Determination of the viraemia profiles in terms of infectious virus (TCID₅₀) titres was only successful for a limited set of birds. The infectious titres recorded successfully were for the birds with the highest peak viraemia titre (in terms of viral RNA) in each group. These titres were specifically 10^6.3 TCID₅₀ ml⁻¹ (10^9.6 RNA copies) for the NY99-infected bird with the highest peak viraemia, 10^5.3 TCID₅₀ ml⁻¹ for the Greece-10- and Ita09-infected birds (10^8.4 and 10^7.5 RNA copies, respectively), and 10^2.3 TCID₅₀ ml⁻¹ for the 578/10-infected bird (10^6.9 RNA copies).

Viral loads were determined in the heart, liver, spleen, kidney, bone marrow and brain for all birds. In order to assess the spread of virus to the different organs at the approximate peak of viraemia, two birds per group were euthanized on day 4 p.i. Virus was detected in all organs of these birds (Fig. 3), with the highest mean tissue viral titres found in the spleen and bone marrow (7.1 and 6.9 log₁₀RNA copies g⁻¹, respectively), followed by the liver, kidney and heart (5.8, 5.7 and 5.5 log₁₀RNA copies g⁻¹, respectively). The lowest tissue viral titres were found in the brain (3.8 log₁₀RNA copies g⁻¹). Between the different virus strains, higher tissue viral RNA titres were found in the organs infected by Greece-10 (7.0 log₁₀RNA copies g⁻¹), followed by 578/10 (6.4 log₁₀RNA copies g⁻¹), NY99 and Ita09 (5.4 and 5.3 log₁₀RNA copies g⁻¹, respectively), whilst FIN-infected birds had the lowest tissue viral RNA burden (4.8 log₁₀RNA copies g⁻¹). Viral titres in the organs were significantly different for Greece-10 compared with NY99 (P=0.05), FIN (P=0.004) and Ita09 (P=0.02), and in the organs infected with 578/10 compared with FIN (P=0.01).

For birds euthanized due to morbidity, virus was also found in all the organs (Fig. 4), with the tendency for the spleen, kidney and heart to contain the highest mean tissue viral RNA load (6.7, 6.6 and 6.2 log₁₀RNA copies g⁻¹, respectively), followed by the liver, brain and bone marrow (5.5, 5.4 and 5.2 log₁₀RNA copies g⁻¹, respectively). Viral RNA titres were highest in organs of birds infected by Greece-10 (6.2 log₁₀RNA copies g⁻¹), followed by 578/10 and NY99 (6.1 log₁₀RNA copies g⁻¹, for both), with the lowest titres exhibited by Ita09-infected birds (5.4 log₁₀RNA copies g⁻¹). However, these titres were not found to be significantly different (P=0.4).

All birds that had survived infection by day 14 were euthanized and necropsied to determine whether virus could still be found in any of the organs (Fig. 5). Viral RNA was detected in all organs of the three Greece-10-infected survivor birds except in the bone marrow of one bird, whilst NY99-infected birds had a minimum of three positive organs (out of six), with the liver being completely negative for both birds. FIN-infected survivor birds had a minimum of two organs (out of six) positive for viral RNA, whilst the bone marrow was consistently negative for all five birds. The two birds that survived infection with Ita09 were both positive for viral RNA in five out of six organs, with the liver negative in one bird and the bone marrow negative in the other bird. Viral RNA was detected in all organs of the two 578/10-survivor birds except for in the liver of one bird. Between the different virus strains, mean viral RNA titres in the organs were significantly higher in Greece-10-infected birds compared with birds infected with FIN (P=0.03).

Overall, mean tissue viral titres in the organs of the survivor birds were lower (2.7 log₁₀RNA copies g⁻¹) compared with the birds necropsied on day 4 (5.8 log₁₀RNA copies g⁻¹) or after illness/death (5.9 log₁₀RNA copies g⁻¹), with most virus persisting in the kidney and spleen (4.1 and 3.7 log₁₀RNA copies g⁻¹, respectively), followed by the brain and heart (3.0 and 2.3 log₁₀RNA copies g⁻¹, respectively), and the liver and bone marrow (1.7 and 1.6 log₁₀RNA copies g⁻¹, respectively).

Sections of organs of birds euthanized on day 4 p.i. were stained with polyclonal anti-WNV NS3 to determine replication of the virus in these tissues and rule out any positive quantitative real-time (qRT)-PCR detection as a result of spillover from blood (Table 2). Tissues most consistently positive for WNV antigen were the kidney (100 %), followed by the heart, liver, spleen (90 %) and bone marrow (80 %), with the brain only positive in 50 % of the cases. Antigen was most abundant in the spleen, followed by the kidney, bone marrow, heart and liver (Fig. S1, available in the online Supplementary Material). The abundance of viral antigen found in the positive brains was minimal. Overall, organs of birds most often positive, as well as most abundant for viral antigen, were those infected with Greece-10 and 578/10, followed by FIN and Ita09, with the lowest amount of antigen found in the organs of the two NY99-infected birds.

Viral antigen appeared to be slightly more abundant in birds euthanized due to morbidity (Table 3) as compared with the birds euthanized on day 4, with the spleen and kidney consistently positive for viral antigen (100 %), followed by the heart and brain (91 %), whilst the bone marrow (64 %) and liver (45 %) were the least often positive. In terms of antigen abundance, staining was most prominent in the heart, kidney and spleen, whilst lower amounts of antigen were present in the brain, bone marrow and liver (Fig. S1). Between the different virus strains, the abundance of viral antigen in the organs was very similar in Greece-10, 578/10 and Ita09-infected birds, and was also higher than the amount of antigen found in the organs of NY99-infected birds.

Five different isolates of WNV were utilized (Table 4). The NY99-4132 strain (NY99; lineage 1a) (Brault et al., 2004) was isolated originally from the brain of an American crow and was passed two times in Vero cells before being used in this study. The lineage 2 Greek strain Nea Santa-Greece-2010 (Greece-10; GenBank accession number HQ537483.1) was isolated from a pool of C. pipiens mosquitoes (Papa et al., 2011) and passaged once in Vero cells. Two lineage 1a Italian strains, FIN (provided by Dr Vittorio Sambri, University of Bologna, Italy; two Vero E6 passages; GenBank accession number KF234080) and Ita09 (provided by Dr Luisa Barzon, University of Padova, Italy, one Vero E6 passage; GenBank accession number GU011992.2) were both isolated from a patient with neuroinvasive disease (Barzon et al., 2009). The Hungarian lineage 2 strain 578/10 (provided by Dr Tamás Bakonyi, Szent István University, Hungary, two passages on Vero E6 cells; GenBank accession number KC496015) was isolated from the brain of a horse that died of WNV neuroinvasive disease. Virus stocks used in this study (not including NY99 and Greece-10) were prepared by growing the viruses once in C6/36 insect cells (Table 1).

Jackdaws were captured using walk-in traps in the municipality of Rotterdam, The Netherlands. They were transported to indoor housing where they were kept in groups of seven or eight in isolators under negative pressure. Only seronegative birds were used in this study. All birds were cared for in animal holding facilities at the National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands.

To confirm that jackdaws had not been exposed previously to WNV, the birds were bled before experimental infection and serum was tested for neutralizing antibodies using TCID₅₀ neutralization assays. Serum was heat-inactivated at 56 °C for 30 min, serially diluted twofold and incubated with an equal volume of virus (strain NY99, originally isolated from a dead Chilean flamingo at the Bronx Zoo in New York, obtained from the Health Protection Agency, Porton Down, UK; P5 on Vero E6 cells; GenBank accession number AF196835.2) to a final concentration of 100 TCID₅₀ per 0.1 ml. Samples were incubated at 37 °C for 1 h and subsequently added to an 80 % confluent monolayer of Vero E6 cells in CELLSTAR 96-well plates (Greiner Bio-One). Plates were incubated at 37 °C for 5 days. Samples were read and a 100 % reduction in cytopathic effect (CPE) as compared with the serum-negative control was used for the determination of neutralization. Detection of any neutralizing activity to WNV in the serum of any bird precluded its use for experimental inoculation.

Jackdaws were subcutaneously inoculated in the thigh region with 2000 TCID₅₀ virus in 0.1 ml Dulbecco's modified Eagle's medium (DMEM) containing no FBS. Jackdaws were injected with NY99 (n=7), Greece-10 (n=8), FIN (n=7), Ita09 (n=7) and 578/10 (n=7). Blood was collected from all birds (~0.1 ml) at 2 day intervals for a period of 8 days p.i. Coagulated blood was centrifuged at 1300 g for 5 min in MiniCollect vials (Greiner Bio-One) in order to separate serum, which was stored subsequently at −80 °C. All jackdaws were examined for signs of disease twice daily for 14 days following inoculation and euthanized under isoflurane anaesthesia upon display of clinical symptoms. Additionally, two birds per group were euthanized at day 4 p.i. for monitoring the dissemination of virus to the organs at the approximate time of peak viraemia, as well as all surviving birds at day 14 p.i.

Necropsies were performed on all euthanized birds and the following tissues were collected: heart, liver, spleen, kidney, bone marrow and brain. A small section of each tissue was collected, and subsequently weighed and homogenized using a metal bead in 1 ml DMEM containing antibiotics (100 U penicillin ml⁻¹, 100 μg streptomycin ml⁻¹). The remaining portion of the tissues was collected in formalin for use in immunohistochemical staining.

To determine viral loads in the serum samples and tissue homogenates, we used qRT-PCR to measure viral RNA titres (serum and tissue) and TCID₅₀ titration for the calculation of infectious virus titres (serum only). Briefly, RNA was isolated from 50 μl serum or 100 μl homogenized tissue using the MagNA Pure LC Total Nucleic Acid Isolation kit (Roche) and an automated nucleic acid robotic workstation (Roche) according to the manufacturer's instructions. RNA was eluted in 100 μl elution buffer (Roche) and stored at −80 °C until assayed. RNA copy numbers were quantified using unmodified primers as described previously (Lim et al., 2013a). The limit of detection of the assay was 0.95 log₁₀RNA copies.

Infectious titres in the serum were determined by log₁₀ titration of the serum samples on Vero E6 cells and calculating the TCID₅₀ using the Spearman–Kärber method (Kärber, 1931; Spearman, 1908) after the determination of CPE 5 days p.i. Initial 1 : 10 dilution of serum resulted in a limit of detection of 10^1.75 TCID₅₀ ml⁻¹.

Sagittal organ paraffin sections (4 μm thick) were processed for peroxidase immunohistochemistry of virus non-structural protein markers. Sections were deparaffinized in xylene, rehydrated in descending concentrations of ethanol and incubated for 10 min in 3 % H₂O₂ diluted in PBS in order to block endogenous peroxidase activity. Antigen exposure was performed by 15 min incubation at 121 °C in citrate buffer (0.01 M, pH 6.0). Sections were incubated overnight at 4 °C with primary goat anti-WNV NS3 (1 : 100; R&D Systems) or goat serum (1 : 100; Dako) for isotype controls, and detected with secondary rabbit anti-goat IgGPO (Dako) antibody. Sections were counterstained with Mayer's haematoxylin, mounted with Kaiser's glycerol gelatin and analysed using a light microscope.

Survival curves were analysed using the log-rank (Mantel–Cox) test. All other statistical analyses were performed by Kruskal–Wallis ANOVA and any significant difference found was more closely analysed between the groups using the Mann–Whitney U test.