Aedes albopictus C6/36, C7-10 and U4.4 and Aedes aegypti Ae cells were grown at 28°C in Leibovitz 15 medium (Gibco) supplemented with 10% foetal calf serum (FBS) and 10% tryptose phosphate broth. Vero E6 cells were maintained at 37°C in Dulbecco's modified Eagle's medium (Gibco) supplemented with 10% FBS. Working stocks of wild-type Bunyamwera virus (wtBUNV), the recombinant NSs deletion virus (rBUNdelNSs2; [17]), rBUNGcGFP [41] and rBUNdelNSs-GcGFP [42] were prepared as described previously [18], [43].

Mosquito cells were infected at an MOI of 1 PFU/cell. After 1 h incubation at 28°C, the inoculum was removed and the cells were washed once to remove unattached virus. Supernatants were harvested at various times post-infection and assayed for virus by plaque assay on Vero E6 cells as previously described [13], [44]. In our laboratory, we routinely titrate BUNV on Vero E6 cells as these give the most easily discernible plaques. We have also performed immunostaining assays of viral foci produced in C6/36 cells, and observed that the efficiency of plating compared to Vero E6 cells is in the range 0.5 to 1 (data not shown). This is similar to the plating efficiency for other arboviruses, like Dengue, West Nile and St Louis encephalitis viruses [45], when comparing titration on C6/36 and vertebrate cells.

At different times after infection, cell lysates were prepared and equal amounts of sample were separated on 12% SDS-PAGE. Separated proteins were transferred to Hybond-C Extra membranes (Amersham). The membranes were incubated with rabbit anti-BUN N protein antibody (1∶2000 dilution) and mouse anti-tubulin antibody (Sigma; 1∶10000) as a loading control, followed by incubation with anti-rabbit horseradish peroxidase-coupled antibody (Cell Signaling Technology). Immunocomplexes on the membranes were detected by SuperSignal West Pico Substrate (Pierce) according to manufacturer's instructions.

Mosquito cells were cultured on glass coverslips and infected with rBUNGceGFP or rBUNdelNSs-GcGFP at MOI of 1 PFU/cell. At various times post infection, cells were fixed with 4% formaldehyde and washed with PBS. The cells were incubated with rabbit anti-BUN N (1∶200) and mouse anti-tubulin (Calbiochem; 1∶100) antibodies, followed by incubation with Texas Red-conjugated anti-rabbit (Cell Signaling Technologies; 1∶200) and CY5-conjugated anti-mouse (Sigma; 1∶400) antibodies. Nuclei were stained with DAPI. Prepared slides were examined with a Zeiss LSM confocal microscope.

Laboratory-bred Ae. aegypti (Paea strain) mosquitoes were reared as previously described [46]. Female mosquitoes were selected and exposed to blood-meal containing approx. 10⁸ PFU/ml of wtBUNV or rBUNdelNSs2 as described previously. At various times post infection, mosquitoes were anesthetized and either homogenized whole or the midguts and salivary glands were dissected. Organs and whole mosquitoes were homogenized in 100 µl of Leibowitz 15 medium (Gibco). Twenty-five microliters of each sample were used for titration by plaque assay on Vero E6 cells.

dsRNA approx. 1000 bp long was prepared using the Megascript RNAi kit (Ambion). Virus specific dsRNA were prepared from linearized plasmids containing full-length cDNAs of the BUNV genome segments pT7riboBUNS(+) and (−), pT7riboBUNM(+) and (−) and pT7riboBUNL (+) and (−) [13]. Renilla-specific dsRNA were prepared from plasmid phRL-CMV by adding a T7 promoter sequence at each end by PCR as described previously [40]. Purified dsRNA was stored in aliquots at −80°C. A total of 3×10⁵ cells per well were cultured in 24 well plates. 100 ng of Renilla dsRNA or 100 ng of a 1∶1∶1 mixture of S-, M-, and L-segment specific dsRNA was used for transfection. One microliter of Lipofectamine-2000 (Invitrogen) was used per 100 ng and the transfection mixes were prepared in the final volume of 100 µl as per the manufacturer's protocol; this was then applied onto the cells with 400 µl fresh complete L15 medium. After 5 hours incubation at 28°C, 500 µl of fresh medium was added. The cells were infected with wtBUNV or rBUNdelNSs2 24 h post-transfection, supernatants were collected at various times thereafter for assay of released virus by plaque titration on VeroE6 cells.

In mosquito cells no shut-off of host cell transcription or translation has been observed. Therefore, we first confirmed that BUNV NSs protein is expressed in these cells. Ae. albopictus C6/36 cells were infected at MOI of 10 pfu/cell and the cells were harvested for protein analysis by Western blotting at different time points (Figure 1). Although BUNV N and NSs proteins are translated from the same mRNA, NSs was produced, though slightly later in the course of infection than the nucleoprotein. This is in line with previous observations [8] based on radiolabelling of infected cells.

We next compared virus replication in Ae. albopictus C6/36, C7-10 and U4.4 cell lines. Cells were infected at an MOI of 1 PFU/cell and the samples were collected at various times post infection. wtBUNV was able to replicate in all three cell lines but with different kinetics (Figure 2A). Growth was most efficient in C6/36 cells, with maximum titres of released virus exceeding 10⁸ pfu/ml, 10-fold and 100-fold higher than in C7-10 or U4.4 cells, respectively. rBUNdelNSs2 grew more slowly than wtBUNV in C6/36 cells and yielded maximum titres about 100-fold less, confirming previous results [16], [17]. In C7-10 cells, the mutant virus showed similar kinetics to wt BUNV. In marked contrast, no increase in titre of rBUNdelNSs2 in culture medium of U4.4 cells was detected. (The detection of some rBUNdelNSs2 virus in supernatant medium (Figure 2A) represents residual virus that remained after removal of the inoculum and replacement with fresh medium; these cells did not adhere firmly enough to the culture vessel to permit extensive washing).

These data were supported by Western blot analysis to detect accumulation of the viral N protein (Figure 2B). In C6/36 cells infected with wtBUNV, N could be seen as early as 16 h post-infection (hpi), whereas in cells infected with the NSs deletion, mutant N was not detected until 36 hpi. In the C7-10 cell line, no significant difference between the two viruses was observed in terms of accumulation of N protein, though growth was slower than in C6/36 cells, with N protein first detectable at 48 hpi. In U4.4 cells infected with wtBUNV, N was detectable as early as 42 hpi whereas no N protein could be detected in cells infected with rBUNdelNSs2. These results suggest that BUNV NSs protein is not essential for growth in either C6/36 or C7-10 cell lines, but is required for successful replication in U4.4 cells.

To analyse how BUNV spreads in mosquito cells, Ae. albopictus C6/36, C7-10 and U4.4 cells were infected with recombinant BUNV expressing green fluorescent protein [41], either rBUNGc-eGFP or rBUNdelNSs-GcGFP at an MOI of 3 PFU/cell. Three different stages of infection, as proposed by Lopez-Montero et al. (2009), were observed in all the Ae. albopictus cell lines infected with rBUNGc-eGFP. These phases were defined by changes in cell morphology due to virus replication, most notably that infected cells produced projections extending towards neighbouring cells (Figure 3A). During the early phase of infection with the wild type version of the eGFP-expressing BUNV, the levels of viral proteins were relatively low and the cells resembled uninfected cells in shape (Figure 3B). However, as infection progressed and the levels of viral proteins increased, cells transitioned into the acute phase. This phase was characterised by formation of filopodia that were most abundant between the 24 hpi and 48 hpi; over this time period, cells maintained physical contact via the projections. During the acute phase, the levels of eGFP-tagged Gc increased and the glycoprotein was also found in the filopodia (Figure 3B). The late phase of the infection was characterised when the cellular filopodia started to disappear (from 48 hpi), which was also manifested by elevated levels of Gc glycoprotein detected in the cells. Later on in the infectious cycle, the cells returned to their normal form (Figure 3B).

These results were consistent for infection of C6/36 and C7-10 cells. Infection in U4.4 cells was slower, and fewer cells were infected in comparison to C6/36 and C7-10 cells. Also, U4.4 cells produced fewer filopodia (Figure 3B). Comparison of the levels of viral proteins in infected cells revealed that BUNV replication in U4.4 cells was less intense than in C6/36 and C7-10 cell lines, as relatively lower amounts of N and Gc proteins were produced. These observations corresponded with the differences in viral titres observed in the previous experiments (Figure 2), where U4.4 cells produced less infectious particles than the two other cell lines.

The course of infection with the NSs deletion recombinant, rBUNdelNSs-GcGFP, was similar to rBUNGc-eGFP in terms of levels of viral protein expression, but the infection started later in C6/36 and C7-10 cells. Investigation of cell morphology showed that fewer filopodia were produced and they were less pronounced in rBUNdelNSs-GcGFP infected cells, suggesting involvement of NSs in this process. Minimal disruption of the normal cell morphology was observed; those changes that were seen were attributed to cell movement and consecutive attachment to the surface, as similar changes were observed in uninfected cells (Figure 3B). These data suggested that NSs protein contributed to the efficiency of viral replication, but was a non-essential protein.

When U4.4 cells were infected with rBUNdelNSs-GcGFP, a few cells (less than 5%) were observed to harbour replicating virus in that synthesis of tagged Gc protein could be observed by its autofluorescence (Figure 3B). This suggests that infection by rBUNdelNSs-GcGFP is abortive and few, if any, new infectious particles were produced, in line with the titration data shown in Figure 2.

In order to rule out the possibility that changes in morphology were due to alterations of Gc glycoprotein caused by fusion of the GFP sequences, another fluorescent Bunyamwera virus was used, with eGFP fused in-frame into NSm [46]. Infection of the C6/36, C7-10 and U4.4 cells with rBUNM-NSm-EGFP, but not with rBUNdelNSs-NSm-EGFP, resulted in similar marked morphological changes as shown by rBUNGc-eGFP (data not shown).

It has been previously documented that the outcome of infection of C6/36 cells with wtBUNV is the establishment of a persistent infection [6]–[8]. To determine whether persistent infections could also be established in C7-10 and U4.4 cells, and if the NSs protein participates in this process, cell monolayers were initially infected with wtBUNV or rBUNdelNSs2 at an MOI of 0.1 PFU per cell. After four days, supernatant fluids were collected for titration of released virus, and the infected cells were passaged (split ratio of 1∶5) and grown until confluent. Thereafter, they were regularly passaged and maintained for about 7 months (25 passages). All lines initially infected with wtBUNV or rBUNdelNSs2 continued to shed infectious virus at each passage, though the titres fluctuated widely (Figure 4A). Similarly, N protein was detected by Western blotting in all passaged cells, though the levels varied. Thus wtBUNV could establish persistent infections in all three cell lines, and rBUNdelNSs2 in C6/36 and C7-10 cells. As expected, since U4.4 cells did not show evidence of productive infection by rBUNdelNSs2, no indication of a persistent infection was obtained. Thus the NSs protein was not essential for establishment of persistent infection in C6/36 or C7-10 cells.

In previous studies of C6/36 cells persistently infected with BUNV, the generation of viruses displaying different plaque phenotypes when titrated in mammalian cells was observed [6], [7]. Similarly, when titrating the two viruses released from all persistently infected cell lines described above in Vero cells, we observed the appearance of large, small and cloudy plaques from different passages (data not shown).

As mentioned above C7-10 and C6/36 cells are reported to have defective Dicer 2-based RNAi responses [36]–[39], whereas the U4.4 cell line encodes a fully functional Dicer 2 gene [36], [40]. To determine whether the cells could mount an RNAi response against BUNV infection, we transfected C6/36, C7-10 and U4.4 cells with long virus-specific dsRNA, or Renilla luciferase-specific dsRNA as a control, and then infected the cells with either wtBUNV or rBUNdelNSs2 at a MOI of 5 PFU per cell. Supernatant fluids were collected at various times post-infection and assayed for the presence of infectious virus by plaque formation. Each infection was done in triplicate and the experiment was repeated twice. As shown in Figure 5, no effect on virus growth was seen in C6/36 or C7/10 cells infected with either virus. However, in U4.4 cells transfected with virus-specific dsRNA and then infected with wt BUNV, virus replication was inhibited, and no increase in virus titre was observed. In contrast, transfection of Renilla luciferase-specific dsRNA had no effect on wtBUNV growth in any cell line. These results are consistent with U4.4 cells having a functional dsRNA-mediated interference system. Unfortunately, because rBUNdelNSs2 virus does not replicate productively in U4.4 cells it was not possible to determine whether BUNV NSs protein could be involved in evasion of an RNAi response in mosquito cells.

Although Ae. albopictus cell lines are widely used to investigate arbovirus replication, the genome sequence of Ae. albopictus has yet to be determined, thus limiting detailed investigation of molecular details of virus-host interaction. To date three mosquito genome-sequencing projects have been completed, Anopheles gambiae, Aedes aegypti and Culex quinquefasciatus [47]–[49], and hence we examined whether cell lines from other mosquito species would be useful to monitor BUNV replication. In preliminary experiments, no evidence for BUNV growth in the Sua4 cell-line derived from An. gambiae Suakoko strain [50] was obtained (data not shown). However, the Ae cell line [51] derived from Ae. aegypti was shown capable of supporting wtBUNV replication (Figure 6). The virus grew to titres approaching 10⁷ PFU/ml, and accumulation of BUNV N protein from 48 hpi was detected by Western blotting. In contrast, rBUNdelNSs2 appeared unable to replicate in these cells, as N protein was not detected by Western blotting (Figure 6B) and no increase in titre of infectious virus in the supernatant medium was measured. (Again as these cells did not adhere firmly to the culture vessel, extensive washing to remove the virus inoculum was not possible, and only residual infecting virus was detected).

To investigate if BUNV was capable of establishing persistent infection in the Ae cell line, cells were infected with wtBUNV or rBUNdelNSs2 at an MOI of 0.1 PFU per cell. Supernatants were collected and infected cells were then passaged using a fifth of gathered cells as it was done with Ae. albopictus cell lines. Cells were maintained for 7 passages. Analysis for the presence of infectious virus in the supernatants showed the Ae cell line to be persistently infected with wtBUNV (Figure 6C). However the levels of infectious virus production remained low and there was no significant difference between consecutive passages as was seen in Ae. albopictus cell lines. Infection with rBUNdelNSs2 showed a different pattern. Infectious virus particles were detected after the first passage, but no virus was detected in the supernatants for the next two passages. However, from passage 4, low titres of virus were detected (Figure 6C). This pattern was reproducibly observed in two further independent infections of Ae cells with rBUNdelNSs2, and also when another Ae. Aegypti cell line, A20 [51] was infected with rBUNdelNSs2: in all cases no virus could be detected by plaque assay of supernatants from the second and third passages, but virus was detected at subsequent passages at low levels (data not shown).

Having demonstrated that BUNV could replicate in Ae. aegypti cell cultures, we next studied the infection of Ae. aegypti mosquitoes. It has been reported previously that wtBUNV is capable of replicating in laboratory raised Ae. aegypti mosquitoes and of being transmitted via mosquito saliva to mice [52]. By comparing infection with wtBUNV and rBUNdelNSs2, we investigated the role of NSs in an insect vector. Female Ae. aegypti (Paea strain) mosquitoes were allowed to feed on a blood-meal containing wtBUNV or rBUNdelNSs2 (approx. 10⁸ PFU/ml in the blood meal). After feeding, only engorged mosquitoes were kept for the course of the experiment. Three engorged females were collected from each infection group immediately after feeding to determine the amount of virus ingested. Based on the titration results, it was estimated that each mosquito imbibed two to three microliters of blood (data not shown). These blood-meal volumes were similar to those previously published for various mosquito species, including Ae. aegypti [53]–[56]. The survival rates following the blood-meal were calculated by recording the number of dead mosquitoes daily. Survival rates for each virus were above 98%, which suggests that neither wtBUNV nor rBUNdelNSs2 had any detrimental effects on mosquito viability.

At various times post feeding, 8 to 10 female mosquitoes were collected, homogenised and the levels of infectious virus they contained were titrated by plaque assay. Infection rates were calculated by dividing the number of infected female mosquitoes by the total number of engorged mosquitoes tested at a given day post-feeding (Figure 7). For wtBUNV, >70% of mosquitoes contained infectious virus at all time points. In contrast, at day 1 after feeding, only 20% of mosquitoes fed rBUNdelNSs2 showed evidence of infectious virus. However, by 5 days post-feeding, virus was detected in 60% of mosquitoes, and this rose to 80% by day 9 (Figure 7). Thus the lack of NSs seemed to delay the progress of infection.

To confirm that lack of NSs results in delayed BUNV replication, in another experiment, midguts and salivary glands were dissected from engorged mosquitoes and examined for presence of virus. At days 1 to 13 post-feeding, nine mosquitoes, and at days 15 to 21 post-feeding, six mosquitoes, were collected, and pools of three isolated organs were made. Infectious virus could be detected in the midgut by 2 days post-infection for both viruses. By 4 days post-feeding, midgut infection rates (calculated as percentage of virus positive midguts among engorged mosquitoes tested) for wtBUNV reached about 80%, and stayed at high levels for the duration of the experiment (Figure 8A). In comparison infection rates by rBUNdelNSs2 were significantly lower, and more variable. The titres of virus in midguts were also different between the viruses, with wtBUNV titres being about 100-fold higher than those of rBUNdelNSs2 (Figure 8B).

In order to be successfully transmitted by a mosquito vector to a vertebrate host, following replication in the midgut the virus must disseminate to secondary tissues like muscles, haemolymph, fat body and eventually the salivary glands. If the virus is found only in the midgut, infection is regarded as non-disseminated. Therefore the transmission potential of BUNV in infected mosquitoes was estimated by calculating disseminated infection rates to salivary glands. Both viruses managed to disseminate to salivary glands successfully, though with different kinetics. wtBUNV was detected in salivary glands by 3 days after feeding whereas the NSs-deletion mutant was not detected in salivary glands until 6 days post-feeding (Figure 7C). The titres of wtBUNV were generally higher in salivary glands than those of rBUNdelNSs2 (Figure 7D). These data suggest that the virus lacking NSs had more difficulty in overcoming the cellular defences in the midgut, but when rBUNdelNSs2 did manage to overcome the midgut escape barrier it was capable of spreading throughout the rest of the body, including to salivary glands.