The demonstration of BTV replication in endothelial cells of skin capillaries as well as in leukocytes in the skin is the most important observation of the study presented here. Previous studies have detected BTV or EHDV in the skin of infected ruminants, especially around hair follicles or in the papillae of the hair[12, 13, 17, 34]. Nevertheless, the skin has not been considered to be an important “replication” organ for these viruses, partly because virus isolation from skin samples indicated only low or inconsistent virus loads[10, 16, 20]. This may at least partly reflect the fibrous and compact nature of ovine and bovine skin, making it difficult to disrupt and effectively homogenise samples for virus isolation or RNA extraction.
Here, we report the replication of BTV in endothelial cells of the skin capillaries as well as in leukocytes present in the skin, as early as 3 dpi. Infected skin glands, which morphologically resemble sebaceous glands and on a few occasions were directly associated with a hair follicle, were also detected. Haemorrhages were observed in skin from all over the body (Figure 1b), indicating that endothelial cells of dermal capillaries in the entire skin may become infected. Since the skin represents the largest single organ in the body this suggests that collectively it may also be a major site for BTV replication at least during the acute and early stages of BTV infection investigated in this study. However no difference in the presence and/or replication of BTV in the skin was detected between skin sections taken from C. sonorensis feeding sites and sections from normal, non-feeding sites. Hence the influence, if any, of Culicoides saliva proteins on BTV replication requires further investigation.
Immuno-labelling and confocal-microscopy were used to detect non-structural viral protein NS2 showing that BTV replicates primarily in two cell types within infected sheep, endothelial cells (mainly of small capillaries) and agranular mononuclear leukocytes. Morphologically the infected leukocytes resembled monocytes,/macrophages, lymphocytes and/or dendritic cells . Although BTV infected γδ T-cells (WC-1 positive cells) were detected in the skin of infected sheep, WC-1 negative leukocyte populations were also clearly infected (data not shown), indicating that several subsets of agranular leukocytes can become infected with BTV. Unfortunately the phenotype of these BTV infected leukocytes (other than γδ T cells) could not be determined as the fixation technique employed restricted the availability of useful antibodies against ovine cell surface markers.
The detection of BTV proteins in lymphatic organs and in the endothelium of blood vessels in the skin, even at the earliest time point investigated (3 dpi), suggests that replication in mononuclear leukocytes and vascular endothelial cells was simultaneous rather than consecutive. Additionally infected endothelial cells and/or leukocytes were not evenly distributed throughout infected organs but appeared to cluster in close proximity to each other suggesting foci of infection caused by local spread of the virus within the organ itself.
Although BTV infection causes a high level of viraemia in infected ruminants the level of replication in most infected organs appeared to be relatively low (as assessed by numbers of infected cells) when compared to similar studies of other viruses such as FMDV. The high level of viraemia may reflect tropism of BTV for abundant cell types (endothelial cells) and/or widely distributed cell subsets (leukocytes). This contrasts with viruses that replicate to high levels in the cells of only a few organs or at a few specific replication sites, e.g. FMDV in the basal layer of the epithelium in the tongue, soft palate and coronary band or rotavirus in the mature epithelial cells at the apices of small intestinal villi.
The lymph nodes of the head and the tonsils contained the largest numbers of BTV infected cells. Like the skin, tonsils have not previously been identified as a major site for BTV infection and/or replication[10, 11, 16, 20]. In further contrast to other studies we detected little virus replication in the spleen of infected animals. The differences in organ tropism that were observed here compared to previous studies may reflect differences between the infecting virus strains, between individual host breeds or even between individual animals. The limited number of infected cells that we detected in the spleen compared to earlier studies could also be a reflection of the different techniques employed and/or between virus replication and virus presence. The red pulpa of the spleen is a major blood reservoir and high virus levels detected in the spleen could simply reflect a high systemic viraemia. Later in infection BTV particles are mainly embedded in the cell membrane of erythrocytes and may therefore be hidden from antibodies[37, 38] and so would remain undetected by confocal microscopy. Such particles could potentially be detected by either virus isolation or RT-PCR, techniques that do not distinguish between virus presence and replication. Further studies are needed to determine if the high level of virus commonly observed in the spleen of BTV infected ruminants is also associated with a high level of replication within this organ or due to virus particles present in the blood circulation.
Variation in endothelial cell susceptibility has been suggested as an explanation for differences in the clinical manifestation of BT in sheep and cattle[39, 40]. Endothelial cells from different tissue origins within the mammalian host also exhibit differences in BTV or AHSV susceptibility and subsequent immune responses[15, 41, 42]. This suggests that variations in the susceptibility of vascular endothelial cells within different organs of the infected host could potentially determine tissue tropism and “organ-manifestation” of BTV, particularly the localisation of haemorrhages. Endothelial cells in the skin, lymph nodes of the head, tonsils and labia may therefore be more susceptible to infection by BTV-2 and 8, and were therefore detected as positive for virus replication at an early stage of infection. However, once a high level viraemia has been established, endothelial cells of less susceptible organs (such as heart, muscle, liver and kidney) may also become infected.
In general, the observed distribution of viral replication mirrored the damage observed either in micro- (histo-) or macro- (anatomical) pathological investigations. In each organ where petechial bleeding was observed infected endothelial cells were detected by immuno-fluorescence, suggesting that infection of endothelial cells may be at least partially responsible for the haemorrhages observed. However the extent of tissue damage seen in many organs did not seem to be reflected by the number of infected cells in these organs (endothelial cells and/or leucocytes). For many haemorrhagic fevers, including BT, it has been suggested that endothelial dysfunction may, at least in part, be attributable to indirect host immune mechanisms such as the release of vasoactive and/or proinflammatory mediators and/or coagulation disorders[15, 43–45] . In this study the amount of viral antigen detected in the organs of infected sheep and the calf, at 9 or 10 dpi, was already lower than 6 and 8 dpi, suggesting a reduction of viral replication at these time points which is consistent with previous descriptions of BTV infection in ruminants[17, 46]. Nonetheless severe clinical disease in sheep is often observed in the later stages of infection (> 9 dpi) when infectious viraemia is already subsiding. Therefore it remains very likely that endothelial cell injury as a result of indirect inflammatory mechanisms rather than direct virus-mediated damage may have contributed to haemorrhagic lesions and tissue oedema observed in this study.
The only organ where histologically observed tissue damage did not reflect the level of anatomical damage and clinical dysfunction, was the lung. Despite being anatomically highly oedematous, only limited endothelial cell death and mild to moderate perivascular and peribronchal oedema were observed microscopically in the lungs. However, increased interstitial pressure can disperse oedema into nearby lower pressure areas (such as alveoli and body cavities) while the fluid is often lost during processing and thus grossly apparent pulmonary oedema may appear relatively mild or inconspicuous histologically. Alternatively, the above discussed indirect vascular leakage due to inflammatory host mechanisms could also play a role in causing the differences observed between the severe anatomical and clinical pulmonary oedema in the lungs and the limited endothelial cell damage detected histologically. In this study, high background staining, possibly caused by incomplete fixation or oedematous tissue damage, prevented the assessment of the relative level of virus replication in the lungs by confocal microscopy.
The current study was limited to only the acute and early stages of BTV infection (up to 10 dpi) in a relatively small number of animals (5 sheep in the longitudinal infection study, plus 2 animals [1 sheep, 1 calf] from a separate study). However, virus replication was clearly demonstrated in the skin of both BTV-8 infected animals, suggesting replication in the skin is a common aspect of BTV infections.
The observation of BTV replication in the skin raises questions concerning the validity of systemic viraemia determined from blood samples taken from the V. jugularis as a measure of virus levels in the periphery of the skin and therefore the amount of virus blood-feeding Culicoides spp. could potentially take up. Culicoides are “pool” feeders and do not take blood directly from blood vessels within the dermis (which are usually too deep for their proboscis to reach). Instead they lacerate the skin to feed on the effusion into this injury, which includes blood, skin cells and lymph. The replication of BTV in the skin therefore provides a local source of infectious virus for vector insects in addition to virus present in the skin capillary system as a result of systemic viraemia. Culicoides can also use the hair as a guide and route of access to the dermal layer of the skin often, feeding immediately alongside the hair shaft itself (P.S. Mellor personal communication) suggesting that infection of capillaries around perifollicular glands and of the glands themselves may also be directly important for infection of these vectors. Further research will be required to investigate if infected ruminants with a relatively low level systemic viraemia could still represent a transmission risk to individual adult Culicoides feeding directly on skin areas where the virus is replicating, especially as viral antigens have been detected in the skin as early as 2–3 dpi here and in other studies.