**3. Discussion**

In this work, we studied differences in the pathogenesis after experimental infection of red-legged partridges with two different Mediterranean L1 WNV isolates, Morocco/2003 and Spain/2007, to elucidate the cause of the different infection course and mortality observed during the experiment and described in a previous work [17].

In both experimental groups, the WNV genome and antigen as well as macroscopic and microscopic lesions were detected as early as 3 dpi. Nevertheless, day 6 postinoculation (3 days after peak viremia, [17]) can be considered critical in the course of the infection, since at that time, the viral load in most tissues reached the highest levels, the virus antigen was more widespread and abundant, and microscopic lesions were more severe. At 14 dpi, although microscopic lesions were still observed, the low viral loads in organs and the absence of virus antigen detectable by IHC sugges<sup>t</sup> that, by this dpi, the partridges had cleared most of the infecting virus. Nevertheless, potential virus maintenance in some tissues of a low percentage of infected individuals should be considered, as persistent infection has been demonstrated in experimentally and naturally WNV-infected birds [23,24].

Microscopic lesions found in the partridges were consistent with those described in other WNV-infected gallinaceous birds [25–27]. Nevertheless, while we found endothelial cell swelling, gliosis, and neuronal necrosis in the brain, CNS lesions were not described in chukar partridges (*Alectoris chukar*) naturally infected with WNV [26,28]. In contrast, in experimentally infected ruffed grouse (*Bonasa umbellus*) who were native from the North American continent, encephalitis of differing severity was present in all, even vaccinated, individuals [29]. Differences in infection conditions, such as the specific virus strain, inoculation dose, and age or bird species, could account for the differences in pathological findings [18,30]. The marked presence of inflammatory infiltrates in feather pulps and skin from 3 dpi onwards was probably associated with the high viral load found by RRT-PCR. WNV has been detected in the skin and feather follicles in naturally and experimentally infected birds [31–33], and feather-picking has been suggested as a potential way of horizontal WNV transmission [34]. The exacerbation of coccidian oocysts infestation in the intestines, concurrent with the acute phase of WNV infection (3–6 dpi), suggests a potential WNV-mediated effect on immune function, although more studies are

necessary to support this hypothesis. Secondary pathological processes in WNV-infected birds have been documented on numerous occasions [7,13,35,36].

In general, the viral loads detected by RRT-PCR correlated well with the virus antigen detected by IHC and microscopic lesions, with some exceptions such as the feather pulp or the brain. Differences in the results between RRT-PCR and IHC can be explained by their different sensitivity (they detect different viral components), specific areas analyzed, or just by differences in sample collection. In the specific case of the brain, negative results in IHC, despite the mild to moderate neuronal necrosis, could also be explained by the induction of cell injury by the local inflammatory response rather than the direct effect of neuronal virus infection, as has been previously suggested [37–39]. However, considering our RRT-PCR results in this tissue, the lower sensitivity of IHC is a more likely explanation. In comparison free-ranging goshawks (*Accipiter gentilis*) infected with lineage 2 WNV predominantly had CNS lesions and abundant WNV antigen detection in foci associated with lesions, in addition to severe myocarditis and lesions in the liver, spleen, and kidney [40]. Additionally, free-living magpies infected with WNV L2 showed severe CNS signs while, in contrast, experimentally infected red-legged partridges, our study species, had only non-specific signs and macroscopic lesions similar as those observed in WNV L1-infected birds, although in both studies no histopathologic descriptions were included [12,41]. This suggests, on the one hand, that bird-virulent WNV L2 strains may be more neurovirulent while, on the other hand, as reported by other authors [18], WNV pathogenesis may be highly dependent on both the infecting strain and the avian host species.

Regarding the inflammatory response in the CNS, microglial cells were the most reactive population in the CNS of the WNV-infected partridges, mainly at 6 dpi (corresponding with the highest viral load), when most of these cells changed to an activated amoeboid morphology. The other resident immune effector of the CNS, astrocytes, appeared to play a limited role. Some authors have indicated that astrocytes can react later than the microglia and that their maximum activity occurs 14 days after CNS injury [42]. Therefore, it is possible that a more marked astrocytosis and/or astrogliosis would be detected later if our experimental study was extended. Microglia and astrocytes are considered essential in the anti-flavivirus response in the CNS of humans and rodent models [30,43–46]. In response to resident cell activation, among other stimuli, CD3+ T cells infiltrated the brain, especially at 6 dpi, and mainly in the cerebral pallium and molecular layer of the cerebellum. Upon antigen presentation, T cells act by directly destroying virus-infected cells and by producing cytokines that increase immune cell recruitment and stimulate other immune effectors [47]. In humans and rodents, it has been demonstrated that these cells are essential for WNV clearance and for recovery from the disease [43,48–50].

For a global interpretation of the WNV-induced encephalitis in the partridges, it is important to highlight that despite the essential defense role of all these cell components for the recovery from WNV infection; it has been indicated that a robust response can also have detrimental effects, contributing to neuron damage [43,51].

Although similar infection dynamics were found in partridges infected with Morocco/2003 and Spain/2007 strains, some differences were noted related to the viral load, virus antigen distribution, and severity of microscopic lesions. The most striking differences were found in the CNS, where encephalitis and neuronal necrosis were more acute and severe in partridges infected with Morocco/2003. The pathogenesis of WNV infection depends mainly on viral and host factors, which determine the level of viral replication and the severity of the infection [30]. As the infected partridges were homogeneous in terms of age, history of rearing, and maintenance, immunologic status should have been very similar between groups. For this reason, the observed differences were most probably due to factors related to the WNV strain. Genetic changes, particularly those leading to amino acid substitutions, can modify the virulence of WNV strains, as well as other phenotypic traits such as host and/or mosquito competence [52–55]. The Morocco/2003 and Spain/2007 WNV strains differ in 13 amino acid positions; therefore, it is possible that

one or several of these changes determined/modulated the virulence of both strains for the red-legged partridge [21].

The pathogenesis of the WNV infection in the CNS depends mainly on the capacity of the virus strain to enter the CNS (neuroinvasiveness) and to produce lesions (neurovirulence) [56]. Some authors have indicated that, once a certain viremia level is reached, the ability of a WNV strain to enter the CNS is much more important than its own neurovirulence [57,58]. Although the mean peak viremia titer was higher in the partridges inoculated with Morocco/2003, there were no statistically significant differences between groups [17], and in both, WNV was present in the brain as early as 3 dpi. At 6 dpi, despite a similar viral load in the brain, a more intense inflammatory reaction and more severe microscopic lesions were observed in Morocco/2003-infected partridges. For these reasons, our results point out that differences in intrinsic neurovirulence between strains were more important than their neuroinvasiveness. It seems plausible that a higher neurovirulence of Morocco/2003, associated with a more exacerbated inflammatory reaction, resulted in a more acute and severe pathology in the CNS. Nevertheless, the exact mechanism by which this strain results in being more neurovirulent remains unknown, and the limited number of birds analyzed in this study forces us to be cautious with the interpretation of the results obtained.

In conclusion, the higher virulence of Morocco/2003 for the red-legged partridge is probably related to a more acute and severe encephalitis in the infected birds. Further studies are needed to elucidate the specific genetic markers that determine this higher neurovirulence.

### **4. Materials and Methods**
