2.2. Laboratory Analysis

Antibodies against flaviviruses were detected using a commercial competitive enzyme-linked immunosorbent assay (cELISA; ID Screen^® West Nile Competition, ID Vet, Montpellier, France) based on purified whole WNV antigen for the detection of antibodies directed against the envelope E protein common to all flaviviruses. Although this ELISA test is designed for WNV detection, the high cross-reactivity observed makes this method suitable for the detection of a broad range of flavivirus antibodies [38,39]. The test was performed according to the manufacturer’s instructions and interpreted following their recommendations: Samples with a sample-to-negative control (S/N) ratio ≤40%, [40–50%] and >50% were considered positive, doubtful and negative, respectively.

To identify the virus causing the seroconversion, samples with positive or doubtful results were then tested for the presence of antibodies against WNV, USUV, TBEV, and LIV by two methods, xMAP microsphere immunoassays (MIAs) and virus micro-neutralization tests (MNTs), as detailed below. In addition, these samples were also tested for BAGV antibodies by virus micro-neutralization.

Serological screening of samples was performed using MIA technology. Purified recombinant envelope domain III (rEDIII) proteins of WNV, USUV, TBEV and LIV were used for the capture of specific IgG antibodies. A recombinant soluble ectodomain of WNV envelope (E) glycoprotein (WNV.sE), which identifies infection by all flaviviruses in the same way as the competitive ELISA test, and WNV.EDIII, USUV.EDIII, TBEV.EDIII and LIV.EDIII were coupled to MagPlex^® Microspheres using a Bio-Plex Amine Coupling Kit (Bio-Rad Laboratories, Hercules, CA, USA), according to the manufacturer’s instructions, as described in Beck et al. (2015) [39]. Five micrograms of WNV.sE, TBEV.EDIII or LIV.EDIII or 50 µg of WNV.EDIII and USUV.EDIII were coupled to the beads. The serum samples (diluted 1:100) were then tested with MIA technology using an equimolar mixture of the five beads, as previously described in Beck et al. (2015) [39]. The secondary antibody was a rabbit anti-deer IgG biotin conjugate (Interchim) for roe deer serum samples, and a goat anti-pig IgG biotin conjugate for wild boar serum samples (AB_2338429, Jackson Immunoresearch, West Grove, PA, USA), diluted 1:500. For each antigen, the diagnostic cutoff was determined from the mean of median of fluorescence (MFI) values of 25 pig (for the analysis of wild boar serum samples) or roe deer (for the analysis of roe deer serum samples) negative serum samples plus three standard deviations of the mean. The 25 pig and roe deer serum samples used to determine the cutoff were sampled in the field from several departments of France and were found to be negative by cELISA. In cases of positive reactions with several rEDIIIs for viruses belonging to the Japanese (i.e., USUV and WNV) or tick-borne encephalitis serocomplexes (i.e., TBEV and LIV), an animal was considered infected with a specific flavivirus if the corresponding bead coupled to rEDIII generated an MFI at least double that generated with the other beads.

MNTs were performed for the detection of specific neutralizing antibodies against WNV (strain Is98, Genbank ID AF481864.1, provided by Philippe Desprès, IPP, Paris), USUV (strain Italy 2012, 206795-3/2012, Genbank ID KX816653.1, provided by Davide Lelli, IZSLER, Brescia), TBEV (strain Hypr, Genbank ID U39292.1), LIV (strain Li 3/1, Genbank ID KP144331.1, provided by Nicholas Johnson, APHA, Weybridge), and BAGV (strain Spain 2010, Genbank KR108244) as described in Beck et al. (2015) [39]. A serum sample was considered positive if cells were protected at the 1:10 serum dilution for WNV, USUV and BAGV, and 1:20 for TBEV and LIV. When results indicated cross-neutralization between flaviviruses, we identified the infecting flavivirus by considering the virus with the highest neutralization capacity, and with neutralization titres that differ by at least 4-fold.

2.3. Laboratory Result Interpretation

Detection of flaviviruses. We considered a serum sample to be i) “flavivirus-negative” when the result of the pan-flavivirus cELISA was negative or was doubtful and negative in MNTs or MIAs; or ii) “flavivirus-positive” when flavivirus antibodies were detected by the pan-flavivirus cELISA (positive serum in cELISA), MNTs or MIAs (doubtful serum in cELISA).

Identification of flaviviruses. As MNTs and MIAs gave complementary and specific results and identified the same viruses (see Results 3.1), we considered that a serum sample was positive for a specific virus (i.e., USUV-, TBEV-, LIV-, WNV- or BAGV-positive) when antibodies towards this virus were detected by at least one of the methods, MNTs or MIAs, given the conditions described above. TBEV and LIV could not be differentiated by MNTs and MIAs, and hereafter results are noted as “TBEV/LIV positive.”

We considered a positive or doubtful cELISA serum sample as “confirmed” when the flavivirus responsible for the seroconversion was identified by one of the confirmatory tests, MNTs or MIAs. Positive cELISA serum samples not identified by MNTs or MIAs were considered “non-confirmed”. We calculated the proportion of confirmatory results as the proportion of confirmed serum samples among positive or doubtful cELISA serum samples. We compared the S/N ratio of the cELISA between serum samples confirmed by two or only one test to assess whether the difference in the results was correlated to the S/N ratio.

3.2. Geographical Distribution

USUV antibodies were detected in animals from six departments, located in southwestern and southeastern France by MNT and/or MIA (Figure 2). TBEV/LIV MNT- and MIA-positive antibodies were detected in three departments of eastern France (Figure 2). TBEV/LIV antibodies were detected only by MIA in three departments located in eastern (Marne, department 51), southwestern (Hautes-Pyrénées, department 65) and northern France (Yvelines, department 78) (Figure 2, Table 1). The 28 serum samples found to be positive or doubtful with the pan-flavivirus cELISA that could not be tested (low serum quantity or too hemolysed) or confirmed by MIA and MNT were from 11 departments (Figure 2). They included nine departments in which USUV or TBEV/LIV antibodies had been confirmed, and two from eastern France close to the TBEV endemic area. No flavivirus antibodies were detected in four departments with >30 serum samples tested: Three were located in western and northern France and one was located in far southeastern France.

Open in a separate window

Figure 2

Geographical distribution of USUV-positive and TBEV/LIV-positive serum samples by virus micro-neutralization test (MNT) or xMAP microsphere immunoassay (MIA), and unidentified flavivirus-positive serum samples in cELISA and negative or not tested by MNT and MIA. The number represents the number of serum samples per municipality.

3.3. Seroprevalence in USUV-Infected Departments

In USUV-infected departments, an apparent pan-flavivirus seroprevalence of 10.0% (CI_95%: 7.3–13.2%) was obtained in wild boar and 3.1% (CI_95%: 1.1–6.5%) in roe deer. The USUV seroprevalence detected by MNT and MIA was 8.0% (CI_95%: 5.6–11.0%) in wild boar and 1.0% (CI_95%: 0.1–3.6%) in roe deer. None of the juvenile roe deer (n = 38) were found to be seropositive. Five juvenile wild boar sampled in 2012 and 2013 had USUV antibodies. The USUV seroprevalence was higher in adults (12.3%, CI_95%: 8.8–16.5%) than in juveniles (4.3%, CI_95%: 1.4–9.7%). In adults, the USUV seroprevalence increased linearly with the period (p = 0.02, Wald test) from 2.6% (CI_95%: 0.3–9.1%) in 2009–2011 to 12.6% (CI_95%: 7.3–19.6%) in 2013–2015. The GLMM showed that the probability of being USUV-seropositive was higher for wild boar adults and animals sampled in 2013–2014 and 2014–2015 compared to 2009–2010 (Table 3).

4.1. Spatiotemporal Circulation of USUV in France

Our results highlighted USUV circulation in southwestern and southeastern France from s2009 to 2014. These results confirm long-term USUV circulation in southeastern France, as reported earlier in 2009–2010 by Vittecoq et al. [15], with four common magpies (Pica pica) in the Camargue area displaying USUV-neutralizing antibodies, while USUV infection could not be established firmly in two additional animals presenting USUV- and WNV-neutralizing responses. Our results support the hypothesis of continuous circulation of USUV in the study period, considering that USUV seroprevalence tended to increase over the study period and was higher in adult wild boar than in juveniles. USUV antibodies were detected every year from 2009 to 2014 in adults, and only in 2012–2013 and 2013–2014 in wild boar juveniles, likely to be older than six months. There are no data on the persistence of maternal USUV antibodies in wild boar. Considering that maternal antibodies against most viruses disappear after six months in large mammals [41], our results would suggest that USUV circulated at least in summer and autumn of 2012 and 2013. Although flavivirus antibody persistence in artiodactyl species is unknown, a longer time span of exposure and antibody persistence could be associated with the higher seroprevalence found in adults, as suggested by field observations for WNV [31,34,35,37]. Given the potential long persistence of antibodies in artiodactyls, it is impossible to know from our results whether USUV circulated every year since 2009 or was re-introduced in 2012.

Very few studies have been conducted to assess exposure of wild ungulates to USUV in infected areas in Europe. In these studies, USUV antibodies in wild ungulates were detected by MNTs to exclude cross-reactivity with WNV antibodies in areas endemic for WNV. The prevalence of neutralizing antibodies against USUV detected in wild boar (8%) and in roe deer (1%) in the infected departments studied here was slightly higher than that found in two studies conducted in endemic WNV countries: In Serbia in 2010–2011 (3.4%, n = 318 wild boar, 0%, n = 91 roe deer), and in Spain from 2003 to 2014 (0.2%, n = 4335 red deer, 0%, n = 32 roe deer) [34,42]. However, this apparent difference may only arise from the fact that the authors estimated the seroprevalence in both infected and uninfected areas, contrary to our results. Similar seroprevalence values in wild ungulates have been observed for WNV antibodies in endemic WNV areas in Serbia, Spain and the Czech Republic. This seroprevalence varied from 2% to 13% in wild boar [31,35,42,43,44], and from 4% to 15% in roe deer [42,43,44].

In our study, wild boar showed higher seroprevalence for USUV than roe deer. Among the seropositive animals by cELISA, the confirmatory rate by MNT and MIA was higher and the mean % S/N of cELISA was lower for wild boar, suggesting higher antibody titrer in wild boar. Similar differences in the seroprevalence of flavivirus infections (as regards WNV) were found between wild boar and cervids (roe deer and red deer) in Spain [28,31,35]. These results may suggest lower antibody response in roe deer to infections by flaviviruses of the Japanese encephalitis antigenic complex. Alternatively, the observed difference may arise from lower exposure of roe deer to mosquito vector bites related to a difference in terms of animal size, habitat or feeding preferences of Culex mosquitoes between wild boar and roe deer.

Surprisingly, USUV was not detected in other departments of eastern and central France in our study, whereas USUV has frequently been detected in these areas from 2015 to 2018, mainly after unusual and grouped bird fatalities [13,16,17,18]. The year 2018 was marked by a sudden increase in the reporting of USUV outbreaks, with a massive die-off of blackbirds (Turdus merula) and captive owls (Strix nebulosa and Bubo scandiacus) in most parts of France [13,18]. From 2015 to 2018, three lineages of USUV were detected, suggesting multiple introductions in France [13,16,17,19]. A Europe 3 strain, genetically close to strains circulating in Eastern Europe since 2011 (Germany, Czech Republic), was detected in dead birds from 2015 to 2017 in northeastern France (Alsace-Lorraine) [16]. Strains from the Africa 3 lineage, genetically close to the strain circulating in Germany in 2014 and in Belgium and the Netherlands in 2016 [9], were detected in central and southern France from mosquito pools in 2015 [17] and from birds in 2018 (Lecollinet, pers. comm). The Africa 2 lineage was detected on several occasions in 2015–2016 in central France (Rhône department) after bird mortalities, as well as in southern France on mosquito pools, and from a human patient suffering from a neurological disorder [16,17,19]. This lineage was previously identified on two occasions in Spain, in 2006 and 2009, from Culex mosquito samples, but was not associated with bird morbidity before its identification in European countries further east, such as France and Germany in 2015 [16,45,46,47,48]. It is impossible to infer from our results which USUV strains were circulating from 2009 to 2014 in southern France. Importantly, USUV circulation remained undetected by the French event-based surveillance network SAGIR [49] that monitors abnormal wild bird mortalities. Since their first detection in Europe, the USUV Europe 3 and Africa 3 lineages have been responsible for massive die-offs of song birds in several countries [9,16,43,50,51,52,53,54]. On the contrary, given that limited cases of bird mortality due to the Africa 2 lineage have been reported, this lineage appears to be less pathogenic for birds. Therefore, we hypothesize that USUV circulating in southern France from 2009 to 2014 may belong to the Africa 2 lineage, close to the strains found in Spain in 2006 and 2009 and in southeastern France (Bouches-du-Rhône) in 2015 [17,47,48]. The absence of detection of USUV antibodies in eastern and central France in our study suggests that both USUV lineages Europe 3 and Africa 3 were introduced more recently than 2014.

4.2. No Detection of WNV and BAGV Antibodies

No WNV antibodies were detected in wild ungulates sampled from the known endemic areas for WNV in southeastern France, the two departments Bouches-du-Rhône (department 13, Figure 1C) and Hérault (department 34, Figure 1C), whereas Vittecoq et al. [15] found WNV antibodies in a second-year common magpie in 2010 in the Bouches-du-Rhône, suggesting recent WNV infection (in 2009 or 2010) in the area. In southeastern France, WNV has been reported irregularly and is generally detected when the virus circulation is high and causes clinical outbreaks in humans and horses. Such epizootics in the Camargue area have been observed in 2000, 2004, 2015 and 2018 [11,13,55,56]. Between two epizootics, the virus could be silently maintained, as suggested by the detection of WNV antibodies in wild birds [15,57]. In our study, few animals were sampled in areas located close to the wetlands of the Hérault and Bouches-du-Rhône departments, where WNV cases were mostly detected during the outbreaks. This may have limited the detection of WNV circulation in our study. This result confirms that the virus, if present, probably circulates very locally close to wetland areas that provide sustainable habitats for the nesting of many wild resident or migrating birds, and for the multiplication of Culex vectors [58].

Bagaza virus antibodies were not detected. In Europe, this virus was only detected in southern Spain in 2010 and 2011, associated with high mortality of red-legged partridges (Alectoris rufa) and ring-necked pheasants (Phasianus colchicus) [59,60]. The main hosts of BAGV are birds and there is little evidence of BAGV antibodies in mammals. Neutralizing antibodies to BAGV have been found in serum samples from encephalitis cases in humans in India, and in experimentally infected mice [61,62].

4.3. TBEV Serocomplex Distribution

We detected TBEV/LIV antibodies in three departments of eastern France (Savoie department 73, Marne department 51, and Jura department 39 in Figure 1C), in one department in the south (Hautes-Pyrénées, department 65) and one department in northern France (Yvelines, department 78). Given the genetic and antigenic relatedness between TBEV and LIV, serological responses induced after TBEV and LIV infection could not be differentiated by MNT and MIA. Although the distribution of TBEV in France is not known precisely and clinical cases have mainly been reported in the far eastern parts of France (Alsace-Lorraine and Alpine regions), the virus is thought to be broadly distributed in the east of France. We detected TBEV/LIV antibodies in wild ungulates in one department (Savoie), where TBE clinical cases in humans have been reported [20], and in two departments (Marne and Jura), where forest workers were previously found to be seropositive to TBEV [22]. It is therefore likely that seropositive wild ungulates observed in our study in eastern France have been exposed to TBEV.

The wild boar found to be TBEV/LIV-seropositive in the Hautes-Pyrénées department by MIA was not confirmed by MNT, whereas the serum was strongly positive in cELISA (low mean % S/N). For the wild boar in the Yvelines department, the serum was too hemolysed and the result could not be confirmed by MNT. The result obtained by MIA could be interpreted, with caution, as an indicator of a tick-borne flavivirus infection. TBEV/LIV seropositivity in Hautes-Pyrénées could originate from an infection with LIV or LI-like virus, since these viruses have been reported in sheep, goats and Cantabrian chamois (Rupicapra pyrenaica parva) in close geographical areas in Spain: In the Basque and Asturias regions [23,25,63]. The presence of a virus from the TBEV-serocomplex was found at a broader scale in Spain by serology, with antibodies neutralizing TBEV reported in southern Spain and on the island of Mallorca [28,29]. The positive serum in the Yvelines might also reveal the western limit of distribution of TBEV in France. Interestingly, TBEV was recently detected in ticks in the United Kingdom, which represents the western-most distribution limit for TBEV [64]. Finally, in both cases, we cannot rule of the hypothesis that this might be due to another tick-borne flavivirus. The diversity and distribution of tick-borne flaviviruses are not well known, and new species are regularly discovered in ticks [65]. Further investigations are needed to confirm the circulation of another tick-borne flavivirus in France and to isolate this virus.

The overall TBEV seroprevalence found in wild boar and roe deer in the eastern departments (2.9% in wild boar and 0.3% in roe deer) was much lower than that found in TBEV endemic areas in Germany, Poland, Slovakia, the Czech Republic and the south of Sweden, varying from 10% to 40% [32,36,44,66,67,68,69]. However, it was similar to the seroprevalences found in areas with the same epidemiological situation as eastern France, with no or very few human clinical cases: From 2% to 6% seropositive wild boar and roe deer have been reported in Belgium, in some areas of Germany, and in the south of Denmark [66,70,71,72,73]. We did not find any significant difference in TBEV seroprevalence between wild boar and roe deer, which is consistent with previous observations [44,66,69,71,72]. These results are unexpected as roe deer are considered to be the main hosts for the nymphs and adults of Ixodes ricinus, the main tick vector for TBEV in Western Europe [74]. However, there is a lack of data on the role of wild boar as a feeding host for this tick or for other potential vector tick species of TBEV, such as Dermacentor spp. [75]

The authors are very grateful to all hunters and local hunter’s associations who helped to build the serum banking and who agreed to make samples available for our study. We are grateful to Philippe Desprès from IPP (Paris, France), Davide Lelli from IZSLER (Brescia, Italy), and Nicholas Johnson (Weybridge, United Kingdom) from APHA for providing reference flavivirus strains.