A low-passage insect-cell isolate of bluetongue virus uses a macropinocytosis-like entry pathway to infect natural target cells derived from the bovine host.

Bluetongue virus (BTV) causes an economically important disease in domestic and wildlife ruminants and is transmitted by Culicoides biting midges. In ruminants, BTV has a wide cell tropism that includes endothelial cells of vascular and lymphatic vessels as important cell targets for virus replication, and several cell types of the immune system including monocytes, macrophages and dendritic cells. Thus, cell-entry represents a particular challenge for BTV as it infects many different cell types in widely diverse vertebrate and invertebrate hosts. Improved understanding of BTV cell-entry could lead to novel antiviral approaches that can block virus transmission from cell to cell between its invertebrate and vertebrate hosts. Here, we have investigated BTV cell-entry using endothelial cells derived from the natural bovine host (BFA cells) and purified whole virus particles of a low-passage, insect-cell isolate of a virulent strain of BTV-1. Our results show that the main entry pathway for infection of BFA cells is dependent on actin and dynamin, and shares certain characteristics with macropinocytosis. The ability to use a macropinocytosis-like entry route could explain the diverse cell tropism of BTV and contribute to the efficiency of transmission between vertebrate and invertebrate hosts.


INTRODUCTION
Arthropod-borne viruses (Arboviruses) are a diverse group of viruses that collectively cause a wide range of important diseases in man, domesticated animals and wildlife [1]. Arboviruses are transmitted between susceptible vertebrate hosts by arthropods, such as mosquitoes, ticks, fleas, sand flies and Culicoides spp. [2]. Bluetongue virus is the type species of the arbovirus genus Orbivirus (within the family Reoviridae). Bluetongue viruses (BTV) are transmitted by their biological vector, Culicoides biting midges, and can cause a severe hemorrhagic disease [bluetongue (BT)], particularly in naïve individuals belonging to certain species of domesticated and wild ruminants. Currently, at least 27 serotypes of BTV have been recognized with further additional serotypes proposed [3][4][5][6][7]. Historically, BTV was confined to tropical and temperate areas of Africa, America, Australia and Asia. However, in the last two decades BTV has greatly expanded its geographical distribution northwards culminating in 2006 in a BTV-8 strain spreading for the first time across Northern Europe, reaching as far as Scandinavia [8,9], which has increased concerns over the potential for further significant BT outbreaks in the region.
The BTV virion comprises ten linear double-stranded RNA (dsRNA) segments encased within a three-layered capsid structure [10,11]. The outer capsid is formed by VP2 and VP5, with the surface of the inner capsid (or 'core') formed by VP7. The innermost sub-core layer is composed of VP3 and surrounds the genomic dsRNA and the three protein components (VP1, VP4 and VP6) of the transcriptase complexes [11]. The outermost protein VP2, which represents the main target for neutralizing antibodies, determines BTV serotype specificity [12] and is the major cell-attachment protein [13][14][15], while VP5 is believed to play an active role in the penetration of cellular membranes during entry [13,16,17]. BTV whole virus particles are known to exploit receptor-mediated endocytosis to enter cells [17,18]. On internalization, the low pH within acidic endosomes triggers VP2 dissociation and the fusion activity of VP5, ultimately delivering the complete core particle into the cytosol [17]. A number of endocytosis pathways have been recognized that deliver cargoes to acidic endosomes and of these, clathrin-mediated endocytosis (CME), caveolar/lipid raft-dependent endocytosis and macropinocytosis (MPC) have emerged as major cell-entry routes exploited by viruses [19]. These pathways may share some of the same molecular machinery; for example, dynamin is an important GTPase involved in endocytosis and serves as a scission factor during vesicle formation and plays a role in CME but also clathrin-independent endocytosis pathways including some types of MPC [20,21].
The ability to enter cells determines viral cell-and tissue-tropism, and ultimately both host-range and disease manifestation. Cell-entry also represents a particular challenge for BTV as it infects cells in widely diverse vertebrate and invertebrate hosts. Thus, enhancing our knowledge of cell-entry may help to develop antiviral reagents with the potential to block the initiation of BTV infections and their subsequent transmission. In its ruminant hosts, BTV infects a wide range of cell types, including several cells of the immune system such as monocytes, lymphocytes and conventional, follicular and plasmacytoid dendritic cells [22][23][24][25]. Endothelial cells of blood [25,26] and lymphatic vessels [27] are also targeted by BTV and represent major sites of virus replication. Consequently, the most severe BT disease typically results in endothelial lesions, excessive bleeding and coagulopathy [25,26,28]. Despite its known cell tropism, it is not known if in vivo BTV utilizes conserved cell surface receptors and similar cellentry mechanisms across different target cell types, or employs multiple entry strategies. In standard laboratory cell-lines, BTV has been reported to use more than one endocytosis pathway for cell-entry. Forzan et al. [17] reported that BTV-10 enters HeLa and Vero cells via CME [17], while we have shown that BTV-1 exploits a MPC-like entry pathway to infect BHK-21 cells [18]. The use of different entry pathways in these studies could be explained by the use of different BTV serotypes (which have significant variation in their cell attachment protein VP2), the use of alternative virus preparations (e.g. infectious tissue-culture supernatants or purified viral particles), or the use of different cell-lines that derive from species not normally targeted by BTV. However, although the above studies suggest BTV can use more than one entry pathway both used virus strains that had been passaged multiple times in tissue culture, thereby potentially adapting the virus to cultured cells. This could be significant as some viruses, such as foot-and-mouth disease virus and Japanese encephalitis virus, can acquire the ability to use alternative cell-attachment receptors (e.g. heparan sulphate) after cell culture adaptation, which could alter the entry route [29,30]. With this in mind, recent studies [31] have demonstrated that a highly passaged strain of BTV-8 has an increased affinity for glycosaminoglycans; hence, serial cell passage could potentially modify the cell-entry characteristics of BTV field strains. Here, to provide further insight to BTV cell-entry, we have investigated the cell-entry pathway used by BTV-1 whole virus particles to infect natural target cells (endothelial cells) derived from the bovine host. For this study we a used a low-passage, purified virus that was isolated from a clinically ill sheep using an insect cell line [32]. Our results show that this strain preferentially uses a clathrinindependent, actin and dynamin dependent MPC-like pathway as the entry route to infect bovine endothelial cells.

RESULTS
BTV-1 infection of BFA cells is partially inhibited by inhibitors of dynamin but does not require clathrinmediated endocytosis To investigate cell-entry, we used a field-isolate of BTV-1 (BTV-1GIB2007) and a bovine foetal aorta endothelial cell line (BFA cells), representing one of the major cell-types targeted by BTV in the mammalian host. BTV-1 GIB2007/ 01 was originally isolated from infected sheep blood during a BT outbreak in Gibraltar in 2007 by a single passage in KC cells. This virus (from here on known as BTV-1/KC3) was subjected to two subsequent passages and virus particle purification using KC cells. The quality of the purified virus preparation was confirmed by silver stained SDS-PAGE (Fig. S1, available in the online version of this article) and virus infectivity was determined by titrating virus on BFA cells. Sequence analysis of genome segments 2, 3, 6 and 7 of BTV-1/KC3, which encode the major structural proteins of the virus capsid VP2, VP3, VP5 and VP7 respectively, showed an identical consensus amino acid sequence to that of BTV-1GIB2007 at KC cell passage level 2 (KC2; isolate BTV-1GIB2007/06) (accession no.: KP821004, KP821126, KP821366, KP821608), suggesting that an additional passage in KC cells is unlikely to have induced cell culture adaptations that could affect the cellentry pathway. In support of this conclusion, low-passage KC cell, BTV isolates have previously been shown to retain infectivity in experimental animal infection studies, which suggest they have not acquired mutations that alter cellentry or virulence [33].
Previously, we reported that entry into BHK-21 cells by a highly passaged tissue-culture BTV-1 reference strain is most likely dynamin-dependent as entry was inhibited by dynasore, a commonly used inhibitor of dynamin [18]. However, we could not determine the effect of dynasore on infection due to a delayed cell toxicity [18]. Paradoxically, the same study showed that expression of a dominant-negative (DN) mutant of dynamin-2 (Dyn-2) had no effect on cell-entry or infection [18], suggesting that BTV entry may be dynamin-independent. Using BFA cells, our initial experiments showed that, similarly to our previous observations for BHK-21 cells [18], extended use of dynasore (at 100 µM) was toxic (data not shown). In contrast, dyngo4a (a less toxic derivative of dynasore that also inhibits dynamin) [34] was non-toxic up to 240 µM (Fig. S2). Dynamin is essential for clathrin-mediated endocytosis (CME), and consistent with this, Fig. 1(a) shows that dyngo4a (15-60µM) inhibited transferrin uptake (a commonly used ligand for CME) by BFA cells. Dyngo4a also inhibited BTV infection in a dose-dependent manner (Fig. 1b), but only when added as a pre-treatment and not when added at later times [1 and 2 h post-infection (h p.i.)] suggesting that BTV entry was dynamin-dependent. However, at the concentrations used, dyngo4a only had a partial inhibitory effect on BTV infection.
Mammalian dynamin is expressed as three isoforms (Dyn1, Dyn2 and Dyn3), and of these, Dyn2 is ubiquitously expressed [35]. Furthermore, Dyn2 exists as four splice variants ('aa', 'ab', 'ba' and 'bb') [36]. Our previous observation that BTV-1 infection of BHK-21 cells was not inhibited by a DN mutant of dynamin was made using the 'aa' splice variant [18]. To further investigate the role of dynamin in BTV infection, we transfected BFA cells to express DN mutants of three Dyn2 splice variants (DN-Dyn2-aa, DN-Dyn2-bb and DN-Dyn2-ba) and quantified the effect on transferrin uptake (Fig. 1c) and BTV infection (Fig. 1d). Cells were transfected with the above DN mutants or with the corresponding matched WT Dyn2 splice variants as green fluorescent protein (GFP) fusion proteins. At 16 h post-transfection, the cells were incubated with transferrin or infected with BTV-1/KC3. Consistent with the role of dynamin in CME, over expression of DN-Dyn2-aa or DN-Dyn2-bb inhibited transferrin uptake (Fig. 1c); however, these mutants had no effect on BTV infection (Fig. 1d). In contrast, expression of DN-Dyn2-ba appeared to increase transferrin uptake, although this was not statistically significant (Fig. 1c) and partially inhibited BTV infection (Fig. 1d). These results are consistent with our previous observations using BHK-21 cells and a DN mutant of the Dyn-2 'aa' splice variant [18], and further, show that BTV infection is selectively inhibited by expression of a Dyn2 DN mutant of the 'ba' splice variant. Taken together, the above observations show that inhibition of dynamin results in a partial inhibition of infection and suggest that BTV may be able to infect BFA cells using more than one uptake pathway that differ in their dependency of dynamin (see Discussion).
The above results suggest that BTV entry into BFA cells is partially dynamin-dependent. In addition, the observations that DN Dyn2 mutants of the 'aa' and 'bb' splice variants inhibited transferrin uptake, but not BTV infection, are consistent with virus entry not occurring by CME. To further investigate BTV entry, BFA cells were incubated with Alexa-Fluor 568-labelled transferrin (which is taken up by CME [37]) or with BTV-1/KC3 for 15 min at 37 C and the cells processed for confocal microscopy labelling for BTV-1 capsid proteins. In these experiments, the actin cortex was also labelled using Cytopainter 405-labelled phalloidin or Alexa-Fluor-647 phalloidin, respectively, to judge if transferrin or virus was at the cell surface or internalized (i.e. located outside or underneath the actin cortex). Fig. 2 shows extensive transferrin uptake by BFA cells after 15 min (Fig. 2a), while BTV particles were not present within the cytosol and remained at the cell surfaces (Fig. 2b, c). This experiment shows that BTV is internalized at a slower rate than transferrin and adds support to the conclusion that despite its partial dynamin requirement, BTV entry does not occur via CME but instead relies on a clathrin-independent pathway.
To confirm if dyngo4a blocked BTV entry, mock and dyngo4a-treated cells were infected with BTV-1/KC3 for 2 h and processed for confocal microscopy labelling for BTV-1 capsid proteins and actin. Using fields of view through the central Z-stacks, 22 BTV-positive mock-treated cells were randomly selected and all showed viral capsid labelling within the cytosol (average 13 BTV puncta/cell) in addition to virus at the cell surface (average eight BTV puncta/cell). Fig. 2(d) shows a typical mock-treated cell and labelling for BTV capsid proteins inside and outside the cytosol (note, for some images in Fig. 2 actin labeling is not shown for clarity). In contrast, central Z-stacks of dyngo4a-treated cells showed that, in addition to BTV present on the cell surfaces, only 22 of 40 BTV-positive cells had some BTV puncta in the cytosol (average 1.5 BTV puncta/cell) while BTV capsid labelling for the other 18 cells was completely confined to the cell surfaces. Fig. 2 (e, f) show dyngo4a treated cells, and for some, BTV labelling was confined to the cell surface ( Fig. 2e) while for others a small amount of virus labelling was also present within the cytosol (arrowed in Fig. 2f). Although not providing an exact quantification of BTV uptake (as we used central Z-stacks), the above observations suggest that the majority of BTV uptake is inhibited by dyngo4a and therefore dynamin-dependent. These results are consistent with the incomplete inhibitory effect of dyngo4a on BTV infection (Fig. 1b) and suggest that dynamin is most likely required for the preferred cellentry pathway used by BTV. However, the incomplete block to virus entry caused by dyngo4a suggests that virus uptake could also be occurring via an alternative, dynamin-independent endocytosis pathway (see Discussion).
To further confirm that CME was not required for BTV infection, BFA cells were treated with pitstop-2 or chlorpromazine (CPZ), drugs that inhibit CME by preventing assembly of clathrin-coated pits [38,39]. Initial experiments confirmed that pitstop-2 (up to 200 µM) and CPZ (up to 10 µg ml À1 ) were non-toxic for BFA cells (Fig. S2). Fig. 3(a) shows that pitstop-2 (25 µM), but not CPZ (10 µg ml À1 ) inhibited transferrin uptake, indicating that in BFA cells, CME was blocked only by pitstop-2. Higher concentrations of CPZ showed toxicity for BFA cells (Fig. S2), restricting its further use to investigate its effect on BTV infection. Despite inhibiting transferrin uptake, pitstop-2 did not significantly inhibit BTV infection by BTV-1/KC3 (Fig. 3b) thereby supporting the conclusion that the main entry pathway for BTV into BFA cells is independent of CME. We also evaluated the role of CME in BTV-1/KC3 infection of BFA cells using DN mutants (DIII and EH29) of eps15 (a clathrin cage recruitment factor), which are known to inhibit CME [40,41]. Both mutants inhibited transferrin uptake when compared to cells transfected to express a control eps15 construct (D3D2) that does not interfere with CME (Fig. 3c). In contrast, expression of the mutant constructs did not significantly inhibit BTV infection (Fig. 3d). Together, the above results show that although dynamin inhibitors reduced BTV-1/KC3 cell-entry and infection, infection does not require CME and suggest that a clathrin-independent endocytosis pathway serves as the major route used for BTV entry into BFA cells.
BTV-1 delivery to acidic compartments Infection by whole BTV particles is dependent on virus acidification [17,18]. To gain further insight into BTV cellentry kinetics, we determined the time taken for virus to be delivered to acidic compartments using ammonium chloride, which increases the pH within the endosomal-lysosomal system and thereby prevents BTV acidification and infection. BFA cells were mock-treated or incubated with ammonium chloride, either as a pre-treatment or at different times after infection, and the drug then remained present throughout the rest of the assay. After 16 h, infection was stopped by the addition of PFA and the level of infection quantified by flow cytometry and normalized to the mock-treated cells. As expected, a strong inhibitory effect was seen when ammonium chloride was added as a pre-treatment (Fig. 4). Infection was also significantly inhibited when the drug was added at 1 h p.i. When added at later times, infection was also inhibited, albeit to non-significant levels (Fig. 4). These results suggest that BTV-1/KC3 is delivered to acid compartments at a slower rate than expected for CME (see also Fig. 2) and add support to the conclusion that CME is not the major entry pathway [42,43] used to infect BFA cells.

BTV-1 infection of BFA cells is dependent on actin dynamics and macropinocytosis
The actin cortex reinforces the plasma membrane and is dynamically regulated during the formation of endocytic vesicles [44] and some viruses disrupt actin to facilitate uptake [45]. Therefore, we investigated the effect of BTV on actin during the early phase of infection (Fig. 5). BTV-1/ KC3 was pre-bound to BFA cells at 4 C for 1 h before warming to 37 C to initiate entry. At the indicated times post-warming, the cells were fixed and processed for confocal microscopy labelling for BTV-1 capsid proteins and actin using AlexaFluor-488-labelled phalloidin. At 0 h (i.e. before warming), virtually all virus labelling appeared to be outside of the actin cortex while some virus appeared to be in close association with actin filaments that projected away from the cells (Figs 5a, b and S3). At 0.5 h post-warming, virus labelling was still not evident within the cytosol and virtually all remained at the cell periphery, i.e. outside of the actin cortex (Figs 5c, d and S3). At these time points, all cells (BTV-positive and BTV-negative) appeared to have an intact actin cortex (Figs 5a and 5c). Similarly, the actin cortex appeared intact for virtually all (>95 %) of the BTVnegative cells at 1 and 1.5 h post-warming. By 1.5 h postwarming, virus capsid labelling was clearly evident within the cytosol (Figs 5g, h and S3). However, in contrast to the BTV negative cells, at 1 and 1.5 h post-warming the majority of BTV-positive cells showed a markedly different pattern for actin (Fig. 5e, g and S3) and the actin cortex  16 % (n=34 cells) of the BTV-positive cells had an intact cortex while for~84 % (n=184 cells) the cortex appeared disrupted. Similar results were observed at 1.5 h post-warming as~76 % (n=222 cells) of the BTV-positive cells showed a disrupted actin cortex. However, the disruption appeared to be transient as at 2 h post-warming the actin cortex appeared to be reforming for some of the BTV-positive cells (Fig. S3). These observations show that that cell-entry coincides with disruption of the actin cortex and suggest that actin rearrangements may be important for BTV internalization.
To test if actin is important for infection we used cytochalasin D (cytoD), which inhibits actin polymerization. Our initial experiments showed that cytoD was non-toxic for BFA cells up to 160 µg ml À1 (Fig. S2) and at 10 µg ml À1 , caused a major rearrangement of actin ( Fig. 6a-d). In addition, when added as a pre-treatment, or at 1 or 2 h p.i., cytoD inhibited BTV-1/KC3 infection of BFA cells. In contrast, an inhibitory effect was not seen when the drug was added at 3 h p.i. (Fig. 6e). Thus, inhibition of BTV infection by cytoD occurred with similar kinetics to virus internalization (Fig. 5) and delivery to acidic endosomes (Fig. 4), suggesting that actin dynamics plays an important role in the infection process.
Macropinocytosis (MPC) is an actin-dependent process [46] and results in increased fluid-phase uptake. In some cell types (e.g. professional phagocytes) MPC is constitutively active, while in others it requires activation [44,47]. Some viruses activate MPC for cell-entry [44]; thus, infection by such viruses can be accompanied by elevated fluidphase uptake [48,49]. Phorbol esters, such as phorbol-12myristate-13-acetate (PMA) also induce MPC and increase fluid-phase uptake. To determine if BTV induces fluidphase uptake, BFA cells were incubated with BTV-1/KC3 (m.o.i.=5) or PMA (1 µg ml À1 ) for 15 min before adding AlexaFluor-647-labelled dextran (0.5 mg ml À1 ), a marker of fluid-phase uptake [50]. After a further 15 min, the cells were fixed and dextran uptake quantified by flow cytometry. Both PMA and BTV stimulated dextran uptake compared to mock-treated cells, and to a comparable extent, suggesting that BTV may also stimulate fluid-phase uptake, possibly by MPC (Fig. 7a).
MPC is dependent on sodium/proton exchangers (NHE) [51,52] and inhibited by EIPA (5-(N-ethyl-N-isopropyl) amiloride), which targets NHE1 (55). PAK-1 is also required for MPC [50][51][52] and inhibition of PAK-1 by IPA3 (56) also inhibits MPC [44]. Thus, EIPA and IPA3 are often used as inhibitors of MPC [44,53,54] (see Discussion). In some cell types, MPC is also dependent on cellular kinases, including phosphatidylinositol-3-kinase (PI3K) [55]; thus, inhibition of PI3K by wortmannin may also inhibit MPC [56]. Our initial experiments established that EIPA (up to 200 µM), IPA3 (up to 200 µM) and wortmannin (up to 800 nM) were non-toxic for BFA cells (Fig. S2). Fig. 7(b) shows that IPA3, but not EIPA inhibited dextran uptake suggesting that, for BFA cells, IPA3 may be a more potent inhibitor of MPC/fluid-phase uptake. However, when added either as a pre-treatment or up to 3 h p.i. both EIPA (Fig. 7c) and IPA3 (Fig. 7d) had strong inhibitory effect on BTV infection. In contrast, neither drug had a significant effect on BTV infection when added at 6 h p.i. (data not shown). The same experiments were carried out using wortmannin but we did not see an inhibitory effect on either dextran uptake (Fig. 7b) or BTV infection (data not shown). These results suggest that BTV-1/KC3 may utilize MPC or a MPC-like pathway for infection of BFA cells. However, since neither dextran uptake nor BTV infection was inhibited by wortmannin, BTV entry into BFA cells may not require PI3K (see Discussion).

DISCUSSION
Here we have shown that a low-passage insect-cell isolate of a virulent BTV-1 field strain (BTV-1/KC3) preferentially utilizes an actin-and dynamin-dependent MPC-like endocytosis pathway to infect endothelial cells derived from the natural bovine host. This conclusion is based on the observations that; (i) BTV-1/KC3 triggered a major disruption of the actin cortex during entry and elevated fluid-phase uptake during the early phase of infection, (ii) infection was inhibited by actin disruption, and (iii) by known MPC inhibitors (EIPA and IPA3). Furthermore, we found no evidence for the involvement of CME in BTV infection as entry followed different kinetics to transferrin uptake and established pharmacological and DN inhibitors of CME did not appear to inhibit infection. Thus, our study shows that BTV joins a growing number of diverse viruses that can enter mammalian cells via MPC or MPClike endocytosis pathways [44,[57][58][59][60][61]. However, it should be noted that in addition to whole virus particles, BTV can exist as at least three other infectious particle types including temporarily membrane enveloped virus particles (MEVP), infectious sub-viral particles (ISVP) and core particles. These different particle types have different specific infectivity for mammalian and insect cells [62] and different outermost surface-components that are likely involved in cell attachment and could potentially influence the cell-entry route. Therefore, it is important to state that our results may only be relevant for whole virus particles and further work will be required to determine the cellentry route used by other BTV particle types.
EIPA and IPA3 are commonly used to investigate if viruses enter cells by MPC or MPC-like routes. Viruses reported to use MPC as their major cell-entry route include measles [61], influenza [58], respiratory syncytial virus (RSV) [60], and human cytomegalovirus (HCMV) [59] and in these studies treatment with EIPA or IPA3 reduced respective viral infection by about 65-70 %. Here we have shown that BTV infection is inhibited to a similar extent following treatment with either EIPA or IPA3. Our results with EIPA and IPA3 are consistent with inhibition of cell-entry, as the effects on infection were maximal when they were added during the early phase of infection (Fig. 7) and they did not significantly affect BTV infection when added at later times. Furthermore, the inhibitory effects of EIPA and IPA3 of BTV infection followed similar kinetics to inhibition by cytoD (which would inhibit actin remodelling during MPC), and with the rate of BTV delivery to acidic compartments, supporting the notion that the reduction of BTV infection by these reagents is achieved via inhibition of MPC. Nonetheless, while some studies report that EIPA selectively inhibits MPC with little or no effect on other endocytosis pathways [51], some investigations have reported that EIPA might also inhibit cellular processes such as actin remodelling, internalization of lipid rafts and CME [63][64][65][66]. EIPA has also been shown to have an indirect effect on adenovirus infection by preventing virus-induced MPC and thereby reducing virus escape from endosomes following virus uptake by CME [67]. Additionally, actin depolymerization drugs (such as CytoD) do not solely inhibit MPC, as actin is required for different types of endocytosis [68]. Thus, although we cannot completely exclude that the effects of the inhibitors used in our study on BTV infection may not result solely from effects specific to MPC, we found no evidence that BTV infects BFA cells via CME, as infection was not significantly inhibited by pitsop-2, or expression of DN inhibitors of CME that were shown to inhibit CME of transferrin (Fig. 3).
Several studies have shown that dynamin is not usually required for MPC [69]. However, dynamin has been  implicated in the regulation of actin dynamics [70] and some types of non-canonical MPC have been shown to be dynamin-dependent [58,71,72]. Thus, despite sharing some characteristics of MPC, our results show the main entry route used by BTV to infect BFA cells appears to deviate from canonical MPC, as entry was dynamin-dependent. Consistent with the existence of dynamin-independent and dynamin-dependent MPC pathways [69], infection by other viruses that use MPC for entry have been reported to have different requirements for dynamin [57-61, 71, 73]. For example, cell-entry by measles, influenza, RSV and vaccinia virus does not require dynamin, whereas entry of HCMV is dynamin-dependent [59]. Thus, similar to HMCV our results show that infection by BTV is mediated by a MPC-like pathway that is dynamin-dependent; however, for HCMV dyngo4a (25 µM) inhibited infection by a greater extent (>90 %) than we observed for BTV with BFA cells (~35 % inhibition with dyngo4a 30 µM). Thus, due to the incomplete inhibition of entry and infection caused by dyngo4a our results could be interpreted to show that BTV may be able to enter BFA cells by more than one pathway that differ in their dependency on dynamin. Further investigations will be needed to determine if this is the case and if this involves a pathway distinct to MPC or two alternative MPC-like pathways that differ in their dependency on dynamin. Nevertheless, the inability to inhibit BTV infection of BFA cells using reagents that reduce CME suggest that if a second dynamin-dependent endocytosis process mediates BTV internalization then this pathway would most likely be independent of clathrin.
Consistent with the partial inhibitory effect on infection seen for dyngo4a, infection of BFA cells was also inhibited by expression of a DN mutant of dynamin-2. Interestingly, our results also show that a DN mutant of Dyn2 'ba' splice variant (DN-Dyn2-ba) preferentially inhibited BTV infection. In support of this conclusion, expression of DN-Dyn2 mutants of the 'aa' and 'bb' splice variants inhibited transferrin uptake but did not appear to inhibit infection, whereas DN-Dyn2-ba did not inhibit transferrin uptake but did inhibit BTV infection. These results suggested that, in BFA cells, the different Dyn2 splice variants may have distinct roles in endocytosis. This conclusion is supported by studies that show that the dynamin-2 splice variants have different functions, including distinct roles in endocytosis [36,74], and are consistent with a study that showed that the 'ba' splice variant of Dyn2 is required for PDGDinduced MPC [75].
Canonical MPC is dependent on PI3K, which has been shown to be required for macropinosome formation, trafficking and fusion [54,76] and thus is sensitive to PI3K inhibition. However, it is now clear that distinct types of MPC exist, and different PI3K-dependent and PI3K-independent pathways have been described [77]. For example, different strains of vaccinia virus have been shown to use distinctive PI3K-dependent and independent MPC-like entry routes to infect HeLa cells [57]. Although further studies will be required to confirm our findings, our observation that BTV infection was not inhibited by wortmannin suggest that the entry route used by BTV to infect BFA cells does not require PI3K.
Infection by BTV whole virus particles requires virus exposure to low pH during cell-entry [17,18] and our experiments confirm that BTV infection of BFA is also dependent on low pH. While macropinosomes remain to be fully characterized, there is good evidence that they undergo a maturation process similar to early-endosomes that involves acidification [78]. In addition, macropinosomes can become acidified by fusion with acidic late-endosomes or endolysosomes [79], which could allow exposure of incoming virus to a low pH environment [80]. Thus, it is highly likely that viruses taken up by MPC or MPC-like pathways will be exposed to a low pH environment during entry.
As outlined in the Introduction, cell-entry is challenging for arboviruses as they need to infect cells from widely diverse vertebrate and invertebrate hosts. Therefore, to achieve cellentry it is possible that BTV could utilize ubiquitous receptors and entry mechanisms, or enter different cells using diverse receptors and pathways. Indeed early evidence suggested that BTV might be able to exploit more than one entry mechanism, possibly depending on virus serotype and/or the cell type [17,18]. However, these studies used high-passage tissue-culture virus isolates and cells that were not derived from the natural mammalian hosts and it is possible that cell-entry could have been influenced by virus adaptations that occur during repeated passages in cultured cells (see Introduction). Thus our studies are the first to use a low-passage virulent virus isolate and cells derived from a natural bovine host and suggest that a MPC-like pathway may be a prominent cell-entry route used by BTV to infect endothelial cells. This could have significance in vivo as MPC results in the non-specific uptake of membrane receptors and other membrane-associated particles, which could be an advantage for viruses, as they may not need to bind specific receptors for internalization [54]. In many cell types MPC requires activation [76,81] while in others, such as macrophages and dendritic cells (DC), MPC is constitutively active [82,83]. With this in mind, a number of studies have concluded that several leucocyte subsets, including macrophages and conventional dendritic cells are likely to be sites of initial virus replication and play a role in BTV dissemination in infected ruminant hosts [22][23][24][25]84]. Thus, the ability to use MPC may be an important determinant of BTV transmission between hosts and the tissue tropism within ruminants.
In conclusion, although we cannot rule out the possibility that other endocytosis pathways are used, our results show that a low-passage insect-cell BTV-1 isolate, that was derived from a virulent BTV strain, most likely utilizes a MPC-like endocytosis pathway as the preferred entry route to infect natural target endothelial cells derived from the bovine hosts. It is interesting to speculate that the ability to use MPC may be an advantage during transmission between invertebrate and vertebrate hosts and thereby facilitate the global success of BTV.

METHODS
Cell lines KC cells were originally derived from embryonic tissues of Culicoides sonorensis [32] and propagated at 28 C in Schneiders insect media (Sigma) with L-glutamine (Sigma) supplemented with 10 % FBS (Sigma), 1 % penicillin/streptomycin (Life Technologies) and 1 % amphotericin B (Sigma). Bovine Foetal Aorta Endothelial Foetal (BFA) cells were from the European Collection of Cell Cultures (ECACC 87022601) and propagated at 37 C in F12-Hams media (Sigma), supplemented as above but without amphotericin B.
Virus purification and characterization BTV-1/KC3 was derived from a virulent 'western' isolate, BTV-1 GIB2007 [The Pirbright Institute (TPI), BTV reference collection BTV-1 GIB2007/01]. This virus was isolated by inoculation of KC cells with infected sheep blood obtained during the BTV-1 outbreak in Gibralter, 2007. The recovered virus (BTV-1 GIB2007/01) was passaged twice more in KC cells before virus purification essentially as described [85], with the exception that infected cells were harvested by manual scraping from culture flasks. Viral titre was determined by fluorescent TCID50 using BFA cells in 96-well tissue culture plates. At 3 days post-infection the cells were fixed using 4 % paraformaldehyde (PFA) (Sigma) for 0.5 h, permeabilized with 0.2 % Triton X-100 (Sigma) in PBS (Gibco) and incubated with anti-BTV-1 polyclonal guinea pig (GP) sera (Orbivirus reference antibody ORAB 279 1 : 2000) in 0.5 % bovine serum albumin (BSA; from Sigma)/PBS for 1 h at room temperature (RT). Following three washes with PBS, a goat anti-GP IgG AlexaFluor-488 secondary antibody (Life Technologies) (1 : 500 in 0.5 %/ PBS) was added for 1 h at RT and the cells then washed with PBS. Plates were read to detect fluorescent cells using the ELISpot Reader (AID Elispot). Viral titre (10 6.5 TCID 50 / ml) was calculated according to previous methodology [86].
Full-length sequencing of individual BTV genome segments was carried out by Sanger sequencing using the BigDye terminator v3.1 kit (Applied Biosystems, Life Technologies, USA) on a 3730 DNA Genetic Analyzer (Applied Biosystems) as previously described [87]. Consensus sequences from each segment were analysed using DNASTAR Lasergene 11 (DNAStar).

Reagents, ligands and pharmacological inhibitors
AlexaFluor-647 labelled transferrin and dextran and Alexa-Fluor-488-labelled phalloidin were from Life Technologies and CytoPainter Phalloidin-405 from Abcam. Nuclear stains, 4¢6-Diamidino-2-Phenylindole (DAPI) and ToPro3 were from Thermo Scientific. PMA (phorbol 12-myristate 13-acetate), Cytochalasin D, Wortmannin, EIPA and cell culture grade DMSO (Dimethylsulphoxide) were supplied by Sigma. Dynasore, dyngo4a, chlorpromazine hydrochloride and pitstop-2 were from Abcam, and IPA3 and ammonium chloride were from Cabiochem and Fluka, respectively. All inhibitors were dissolved in DMSO except for ammonium chloride, which was dissolved in GMEM (Glasgow's minimum essential media) from Sigma, and chlorpromazine hydrochloride, which was dissolved in deionized water. Stock solutions were further diluted in serum-free cell culture media for final concentration given for respective experiments.
Infection conditions, inhibitor studies and cell transfections BFA cells were infected with purified virus (m.o.i.=<1) in serum-free F12-Hams media (Sigma) for 1 h at 37 C. The cells were then washed three times with PBS and infection continued in cell culture media but supplemented with 1 % FBS. Cells were pre-treated with pharmacological inhibitors for 0.5 h before infection or the inhibitors were added after infection was initiated as indicated on the figures. After addition, the inhibitor remained present throughout the assay. The control (mock-treated) cells were treated with an equivalent concentration of the appropriate drug diluent. Mammalian expression plasmids for GFP-tagged, WT and dominant-negative (DN) dynamin-2 (Dyn2) splice variant ('aa', 'bb' and 'ba') were supplied by Mark McNiven (Mayo Clinic, Rochester, USA) [53]. Mammalian expression plasmids for GFP-tagged DN eps15 (DIII and EH29) and the control plasmid (D3D2) were supplied by Alexandra Benmerah (Universte Paris Descarte, Paris, France). Cell transfections used 1 µg plasmid DNA and XtremeGene 9 (Roche) in a 6 : 1 (reagent/plasmid) ratio as per manufacturers' instructions 16 h prior to infection. Infections were stopped by incubating the cells in 4 % PFA for 1 h on ice. For entry studies infections were stopped as indicated on the figures whereas all other infections were stopped at 12 h p.i.
Flow cytometry experiments Intracellular labelling of infected cells: Infected (and mockinfected) cells in 4 % PFA were permeabilized with 0.2 % Saponin (Sigma) in 1 % BSA in PBS for 20 min. Viral capsid proteins were labelled using a primary anti-BTV-1, rabbit polyclonal sera (Orbivirus reference antibody ORAB 276) at 1 : 4000 in 1 % BSA/0.2 % Saponin/PBS (FACS Buffer) for 1 h and a goat, anti-rabbit IgG AlexaFluor-647 secondary antibody (Life Technologies) (1/500 in FACS Buffer) for 45 min. Following each incubation with antibodies cells were washed three times by 5 min agitation in FACS buffer and subsequent centrifugation at 333 g for 3 min). Following the last wash, pelleted cells were resuspended in 350 µl PBS and transferred to FACS tubes (BD Biosciences) for subsequent analysis. Control ligand uptake: BFA cells were serum starved by incubation in serum-free cell culture media for 45 min changing the media every 15 min and then allowed to take up AlexaFluor-647 transferrin (10 µg ml À1 )/dextran (0.5 mg ml À1 ) for 15 min. At the end of uptake, cells were placed on ice and fixed with 4 % paraformaldehyde for 1 h. Cells were then washed in PBS, resuspended in 350 µl PBS and transferred to FACS tubes (BD Biosciences) for subsequent analysis.
Data was collected on the LSR Fortessa (BD Biosciences) using BD FACSDivaTM software and exported as flow cytometry standard (FCS) files and analysed with DeNovo, FCS Express v5 software. Cells were stained with Near IR Live/Dead stain (Life Technologies) using manufacturers' instructions, allowing live and dead cells to be distinguishable by fluorescence staining and gating of viable cells. Cell debris was removed from the analysis by gating cell populations using forward and side scatter (SSC) and single cell populations in SSC area versus SSC height. For experiments using pharmacological inhibitors the threshold for the detection of infected cells (i.e. virus positive cells) was set using uninfected cells incubated with the primary and secondary antibodies. Infection levels of drugtreated cells were normalised to infection of mock-treated control cells. For transfection experiments using GFPtagged proteins, the level of infection of the GFP positive population for DN expressing cells was normalized to infection of the GFP positive populations expressing the matched WT or control construct. Mean fluorescent intensity (M.F.I.) was used to quantify uptake of control ligands. The background threshold was set by using cells that were exposed to ligand at 4 C.
Immunofluorescence confocal microscopy Experiments were carried out using cells on glass coverslips (VWR). For entry experiments, cells were incubated with purified virus (m.o.i.=10) in serum-free media on ice for 1 h to allow virus adsorption and entry synchronization. After adsorption, the cells were washed three times with serumfree cell-culture media and warmed to 37 C for 0.5, 1 or 1.5 or 2 h p.i. (as indicated on the figures) and then fixed and processed for microscopy labelling for BTV capsid proteins as follows. The cells were placed on ice and fixed with 4 % PFA for 1 h and then immersed in PBS. Cells were permeabilized using 0.2 % saponin in PBS for 20 min and then blocked using 10% Normal Goat Serum (Harlan Lab Sera) and 1 % Teleostean Gelatin (Sigma) in TBS (block buffer) for 0.5 h. Viral capsid proteins were detected using ORAB 276 (1 : 1000) and a goat anti-rabbit IgG AlexaFluor-568 secondary antibody (Life Technologies) (1/200) in block buffer. The cells were incubated sequentially for 1 h each with the primary antibody and secondary antibodies, with 3Â5 min PBS washes between antibody incubations. For transferrin uptake, BFA cells were incubated in serum-free cell-culture media for 45 min changing the media every 15 min and then allowed to take up AlexaFluor-568 transferrin (10 µg ml À1 ) for 15 min at 37C. At the end of uptake, cells were placed on ice and fixed with 4 % paraformaldehyde for 1 h. Cell nuclei were stained with 4¢,6-Diamidino-2-Phenylindole (DAPI) or TO-PRO-3 and coverslips mounted in Vectashield mounting medium for fluorescence (VectorLabs). The actin cortex was labelled using (1/20) AlexaFluor-488-labelled phalloidin (Life technologies) in PBS for 20 min or CytoPainter-405 Phalloidin as per manufacturers' instructions. Cells were viewed using the Leica TCS SP8 Confocal laser-scanning microscope and optical sections recorded using either the 63 or 40x oil-immersion objective with a numerical aperture of 1.4 and 1.25, respectively. The data are shown as single optical sections through the middle of the cell. All data were collected sequentially to minimize bleed-through fluorescent signals. Images were processed using Adobe Photoshop software.
Cytotoxicity assays BFA cells were seeded onto optically clear bottom, whitesided plates (Perkin Elmer) the day before the assay. When cells were 80 % confluent, cells were treated with a range of concentrations of the pharmacological inhibitors for 12.5 h. Cell viability was determined using the CellTiter-Glo luminescence assay (Promega) as per manufacturers' instructions using Hidex chameleon luminometer.

Statistical analysis
All experiments were carried out using triplicate samples for at least two independent experiments. Data from repeated experiments were combined prior to statistical analysis. The statistical significance of results of comparisons between experimental groups were determined by one-way ANOVA with Tukey's correction, which was carried out using Minitabv17.