**4. Discussion**

This study characterized and quantified the distribution of immune cells in the CNS of lambs experimentally infected with SGEV. The inflammatory cells were predominantly microglia, with a moderate number of T lymphocytes and a smaller number of B lymphocytes. These findings are similar to those of previous studies on SGEV in goats [6] and on LIV and other flaviviruses in horses, humans, and non-human primates [7–12,14,15]. Microglia, which are of mesodermal origin, play a key role in the innate and adaptive immune responses in the CNS [10]; in fact, they are the first cells that respond to CNS infection [12,16]. Infection itself and the resulting neuronal damage activate the recovery in microglia of their amoeboid properties such as motility and of their macrophagic function [10,12,16]. This activation increases neuronal cell death, astrogliosis, and production of pro-inflammatory cytokines [10,12,16].

The proportion of microglia was higher in lambs in the present study than in goats in previous work [6], which may reflect the greater susceptibility of lambs to tissue damage [17]. In this regard, 5 out of 9 and 2 out of 9 lambs showed type III and type IV microscopic lesions, respectively, whilst only 4 out of 9 goats showed grade III lesions, and none grade IV lesions with an identical challenge [3,4]. Moreover, in the present study the proportion of microglia was higher in cluster B and was related to more extensive tissue damage. In studies of human and animal infection with Japanese encephalitis virus, microglia predominated in the thalamus and hippocampus, the primary a ffected brain regions [18]. We also observed the greatest relative abundance of microglia in thalamus and hippocampus, although these regions were not the most seriously a ffected by SGEV [4].

T lymphocytes were the second most abundant cell population, as reported in flavivirus infections in goats, lambs, humans, non-human primates, horses, and mice [6,7,9,11,15,17,19]. T lymphocytes control viral infections in the CNS by destroying infected cells, producing cytokines, stimulating phagocytic activity of microglia, and stimulating local antibody production by B lymphocytes [12,20]. High numbers of T lymphocytes have been linked to fatal outcome in human patients infected with tick-borne encephalitis virus [21]. T lymphocytes were more abundant than B lymphocytes, but the di fference was not significant, similar to the case in goats infected with SGEV [6] and in lambs infected with LIV [22]. A higher abundance of T lymphocytes than of B lymphocytes has been associated with acute disease in horses infected with West Nile virus [9].

B lymphocytes were the less numerous cellular type and were more abundant in perivascular spaces (6.3% of all immunostained cells) than in parenchyma (2.5%), similar to what has been reported in goats, humans, and non-human primates infected with flavivirus [6,11,12,17,23]. This might reflect antibody-mediated tissue damage in lambs [17]. The proportion of B lymphocytes was lower in lambs in the present study than in goats in previous work [6], which may reflect a lower specific humoral cell-mediated response in lambs.

Reactive astrogliosis was observed in the CNS of mice, lambs, goats, humans, and mice infected with flaviruses [6,23–25], especially in persistent infections [25]. In our study of lambs, most CNS sections received an astrogliosis score of 2, characteristic of di ffuse innate immune activation in response to viral infection [13]. An astrogliosis score of 3 was observed mainly in the spinal cord near the pia mater, which is the most likely entry route for the virus into the CNS, where astrocytes are in the lining of brain capillaries and pia mater and are the first cells to come in contact with the virus [25]. GFAP expression has been associated with the greatest damage in humans infected with West Nile virus [23]; this is similar to the present study's finding of a particularly strong GFAP immunostaining in spinal cord and medulla oblongata, where histological lesions following SGEV infection are more frequent and severe [4].

In our SGEV-infected lambs, glial foci were significantly more abundant in thalamus, hypothalamus, corpus callosum, and medulla oblongata; SGEV-infected goats, in contrast, showed more foci in medulla oblongata and spinal cord [6]. Therefore, immunohistochemistry can help identify tissues preferentially a ffected by di fferent flaviviruses, otherwise di fficult to detect using conventional hematoxylin-and-eosin staining, as already demonstrated for tick-borne encephalitis and Japanese

encephalitis [14,16]. We observed lesion predominance on the same side of the medulla oblongata as that of the subcutaneous virus injection, similar to results obtained in goats [6]. This preference supports a neurotropic route of virus movement [4].
