Non-structural protein NS3/NS3a is required for propagation of bluetongue virus in Culicoides sonorensis

Bluetongue virus (BTV) causes non-contagious haemorrhagic disease in ruminants and is transmitted by Culicoides spp. biting midges. BTV encodes four non-structural proteins of which NS3/NS3a is functional in virus release. NS3/NS3a is not essential for in vitro virus replication. However, deletion of NS3/NS3a leads to delayed virus release from mammalian cells and largely reduces virus release from insect cells. NS3/NS3a knockout BTV in sheep causes no viremia, but induces sterile immunity and is therefore proposed to be a Disabled Infectious Single Animal (DISA) vaccine candidate. In the absence of viremia, uptake of this vaccine strain by blood-feeding midges would be highly unlikely. Nevertheless, unintended replication of vaccine strains within vectors, and subsequent recombination or re-assortment resulting in virulent phenotypes and transmission is a safety concern of modified-live vaccines. The role of NS3/NS3a in replication and dissemination of BTV1, expressing VP2 of serotype 2 within colonized Culicoides sonorensis midges was investigated. Virus strains were generated using reverse genetics and their growth was examined in vitro. A laboratory colony of C. sonorensis, a known competent BTV vector, was fed or injected with BTV with or without expressing NS3/NS3a and replication in the midge was examined using RT PCR. Crossing of the midgut infection barrier was examined by separate testing of midge heads and bodies. Although the parental NS3/NS3a expressing strain was not able to replicate and disseminate within C. sonorensis after oral feeding, this virus was able to replicate efficiently when the midgut infection barrier was bypassed by intrathoracic injection, whereas the NS3/NS3a knockout mutant was unable to replicate. This demonstrates that NS3/NS3a is required for viral replication within Culicoides. The lack of viremia and the inability to replicate within the vector, clearly demonstrate the inability of NS3/NS3a knockout DISA vaccine strains to be transmitted by midges.


Background
Arthropod-borne viruses (arboviruses) have a significant social and economic impact on both human and animal health. A majority of all emerging and re-emerging infectious diseases are vector-borne or zoonotic [1,2]. Combating arboviral diseases requires an interdisciplinary approach, possibly including vector control, surveillance programs and outbreak containment, with prevention of disease by vaccination likely the most promising [3]. Bluetongue virus (BTV, family Reoviridae, genus Orbivirus) [4] and Schmallenberg virus (family Bunyaviridae) [5] are examples of emerging arboviruses in countries with a moderate climate, which are transmitted by Culicoides (Diptera: Ceratopogonidae) biting midges [6][7][8]. BTV infection results in a haemorrhagic disease of domestic and wild ruminants called Bluetongue (BT). In cattle, BT is typically subclinical, but infection in sheep can result in severe disease with high mortality [9]. Due to the wide host range, prolonged viremia with often less-severe disease symptoms, and virus spread by midges, outbreak control is difficult using control measures other than vaccination [3].
BTV is a non-enveloped virus with a complex triplelayered capsid containing the ten-segmented double stranded (ds) RNA genome and the replication complex. In addition to seven structural proteins (VP1-7), BTV encodes four non-structural (NS) proteins [10][11][12]. The virus release mechanism differs for mammalian and insect cells, with insect cells showing non-lytic release, whereas cell lysis is the prominent release mechanism in mammalian cells [13]. Virus particles can leave the cell by budding, acquiring a temporary envelope, but also via disruption of the cell membrane [13,14].
NS3 and N-terminal truncated NS3a are viroporin-like membrane proteins [15,16], functional in virus release [17,18]. NS3/NS3a is expressed in larger amounts in insect cells and is hypothesized to be mainly important for the non-lytic release from these cells [19][20][21][22][23]. NS3/NS3a comprises a long N-terminal domain, two transmembrane domains with a short extracellular domain in between, and a shorter C-terminal cytoplasmic domain [24]. Interaction with the p11 cellular calpactin complex subunit and recruitment of the ESCRT-I TsgI protein [25][26][27], highlights involvement in membrane trafficking/modification and virus release. Recently, we showed that expression of NS3/NS3a is not essential for BTV replication in vitro. However, release of NS3/NS3a knockout mutants from mammalian cells is significantly delayed and release from insect cells is strongly reduced [28].
By inducing an out of frame deletion in BTV Seg-10, encoding NS3/NS3a, using reverse genetics, knockout BTV viruses were generated. A BTV vaccine strain with such a Seg-10 deletion has been described to be a very promising vaccine candidate, named the disabled infectious single animal (DISA) BT vaccine. Sheep have been vaccinated with this virus, and no clinical signs after vaccination were induced. Vaccine virus replicated only locally and sterile protection to virulent BTV infection was induced [29][30][31]. Since vaccination did not result in detectable viremia of BT DISA vaccine virus, oral uptake of the vaccine by insect feeding is highly unlikely.
Oral infection of midges with BTV and subsequent virus transmission to the ruminant host is complex and differs even between individuals of one Culicoides species. This leads to variable proportions of individuals within a midge population being susceptible to oral virus infection or capable of virus transmission. Vector arthropods present several 'barriers' which could prevent infection, dissemination, or transmission of the arbovirus to the susceptible host. For the Culicoides vector, a midgut infection barrier (MIB), a midgut escape barrier (MEB) and a dissemination barrier have been identified (Fig. 1) [32][33][34]. Since NS3/NS3a has a prominent role in virus release in insect cells in vitro, we now considered whether propagation of DISA vaccine virus, not expressing NS3/ NS3a, would be significantly diminished or blocked in BTV-competent Culicoides biting midges in vivo. This blockade would improve the safety of replicating BT DISA vaccine by minimizing the risk on uncontrolled vaccine spread. Therefore, we here investigated the propagation of a DISA vaccine based on BTV1, expressing VP2 of serotype 2, after oral infection and intrathoracic injection of in Culicoides sonorensis (formerly Culicoides variipennis sonorensis) [35], a species known to vector several BTV serotypes including U.S. BTV2 [36].

In vitro virus release assay
Monolayers of KC cells (5 × 10 6 ) or BSR cells (5 × 10 5 ) in 2 cm 2 wells were infected with an MOI of 0.01. Virus was adsorbed for 1.5 h at 27°C or 37°C for KC and BSR cells respectively. Unattached virus was removed by washing with PBS, and fresh medium was added. This time point was set as 0 h post-infection (hpi). Incubation was continued and cells and culture media were harvested at indicated time points. Cells were lysed to be able to study the intracellular BTV fraction, by freeze thawing at −80°C. Virus titers were determined and growth experiments were independently repeated.

Insect feeding and sampling
Colonized 3-4 day old female C. sonorensis midges from the Arthropod-Borne Animal Diseases Research Unit, Manhattan, KS, USA [43] were offered a blood meal, consisting of 1:1 (v/v) defibrinated sheep blood and virus stock at highest titer available, in an artificial feeder using a parafilm membrane [44]. Midges were allowed to feed for 2 h, removed from the blood source, anesthetized for 10-15 s with CO 2 , removed from the feeding cage, and sorted as to blood-feeding status on a CO 2 fly pad (Diamed Lab Supplies, Inc., Mississauga, Ontario, CA). Engorged females were put in cardboard cages with cotton-plugged vials containing 10 % sucrose and held at 26°C. At 0, 7, 10 (BTV11 control only) and 14 days post feeding (dpf), 50 midges were anesthetized with CO 2 and heads were separated from bodies using ultra-fine tweezers (EMS Hatfield, PA, USA) and a dissecting microscope (SMZ 1500; Nikon Instruments, Melville, NY, USA). Heads and bodies were separately placed in 100 μl RNAlater (Qiagen, Germantown, MD, USA), and stored at 4°C.

Insect inoculation
Colonized 3-4 day old female C. sonorensis midges were injected intrathoracically with 46 nl of virus stock using the Nanoject II microinjector (Drummond Scientific, Broomall, PA, USA) under a dissecting microscope (Nikon Instruments). This volume was based on hemocoel injection capacity of the midges, highest rate of consistency, complete fluid retention, and maximum post inoculation survival. Total virus injected per midge, based on highest starting virus stock titers possible, was 10 1.5 TCID 50 of DISA 2 and 10 2.1 TCID 50 of BTV1[VP2]2 virus. Inoculated midges were placed in cardboard cages with cottonplugged vials containing 10 % sucrose and held at 26°C. At 0, 7, and 10 days post inoculation, 25 midges were anesthetized, decapitated and stored in RNA later (Qiagen) at 4°C as described above. Time 0 samples were taken from 1 to 4 h post-injection. This variation was due to the timeintensive nature of microinjecting large numbers of midges needed to ensure adequate numbers of surviving midges for each time point.

RNA isolation and PCR
PBS (400 μl) and one 5 mm stainless steel ball (Qiagen) were added to midge bodies and heads in RNA later in micronic tubes. Tubes were shaken for 3 min at 50 Hz in a tissue lyser (85600, Qiagen). After centrifugation, 200 μl of supernatant was used for RNA isolation using the MagNApure 96 DNA and viral NA Small Volume kit (Roche) by the MagNApure isolation robot (Roche) according to the manufacturer's protocol [45]. The realtime RT-PCR test for Seg-1 was performed using primer F-pan-S1 (5′-TTAAAATGCAATGGTCGCAATC-3′), primer R-pan-S1 (5′-TCCGGATCAAGTTCACTCC-3′) and probe P-pan-S1 (5′-6-FAM-CCGTGCAAGGTGC-MGB-3') [46] according to the all-in-one method, including the pre-denaturation step, as described for the pan BTV Seg-10 PCR test [45]. Crossing point (Cp) values of each PCR were calculated, and negative results were arbitrarily set as 45 to allow these to be included in the analysis. Due to the maximum of 45 cycles, the highest Cp value that could still be calculated was 40. Statistical differences in Cp values were calculated using a one way ANOVA and subsequent Tukey's multiple comparison test, with p < 0.001 indicating significant differences between days or groups.

NS3/NS3a is required for release from Culicoides cells in vitro
Similar to previous results with NS3/NS3a knockout mutant viruses [28], DISA 2 did not induce CPE in BSR cells. Immunostaining of VP7 indicated BTV replication, but NS3/NS3a expression could not be detected. In contrast, BTV1[VP2]2 parental virus induced CPE and was immunostained with both VP7 and NS3 MAbs (Fig. 2a). Sequencing of Seg-10 of DISA 2 confirmed the ΔC deletion, and thus the absence of NS3/NS3a expression by DISA 2. No additional insertions or deletions in Seg-10 of DISA 2 were identified (not shown). In agreement to previous results [28], release of DISA 2 from BSR cells was slightly delayed compared to the parent virus, with clear increase examined at 48 h post infection, compared to 16 h post infection for BTV1[VP2]2 (Fig. 2b). The virus titer in the cells was higher for BTV1[VP2]2 (10 7.2 TCID 50 ml −1 ) compared to DISA 2 (10 4.9 TCID 50 ml −1 ) at the end of the experiment at 72 h post infection. Release of virus was also reduced for DISA 2 (10 3.1 TCID 50 ml −1 ) compared to the parental virus (10 6.5 TCID 50 ml −1 ). Growth in KC cells was highly attenuated, with ten-fold lower virus titer in both the cell fraction and the released fraction for DISA 2 at the end of the experiment. DISA 2 virus release was diminished, with 10 2.1 TCID 50 ml −1 at 0 dpi and still only 10 2.6 TCID 50 ml −1 at 120 dpi, the end of the experiment (Fig. 2c).
C. sonorensis midges can be infected orally with BTV11, but not with BTV1[VP2]2 BTV11-fed midges were harvested at day 0, 7, 10 and 14 dpf. For each day of sampling, heads and bodies of 25 midges were harvested and were tested individually. Viral RNA was isolated for PCR testing and semiquantitated by Cp values (Fig. 3). Directly after feeding (day 0), all bodies were PCR positive (mean Cp: 25.3, range 27-23.9), whereas heads were negative (13 out of 25) or had very high Cp values (mean Cp:39.5). As blood meals Similarly, 50 BTV1[VP2]2 fed midges were harvested at day 0, 7 and 14, and tested by PCR. At day 0, directly after feeding, bodies showed Cp values (mean Cp: 26.1, range 27-23.5) not significantly different from that of bodies of BTV11 fed midges. At day 7, 40 out of 50 bodies were PCR negative, and at day 14 post feeding only 6 out of 50 bodies had a Cp value of 40. All heads were PCR negative at day 0, 7 and 10, except for one head at day 0 and 7 (Cp:40). These results indicated that BTV1[VP2]2 was unable to infect the midgut epithelium and was unable to escape the midgut barrier in C. sonorensis. Feeding with DISA 2 virus was also performed, but harvested heads and bodies were not further processed because of the negative results of the BTV1[VP2]2 parental strain.
NS3/NS3a is required for propagation of artificially disseminated BTV in C. sonorensis Intrathoracic injection of virus into the hemocoel of midges can lead to replication of virus unable to naturally escape the midgut following oral uptake [32,47]. Therefore, to study virus propagation in the absence of a midgut barrier, midges were injected with the parent strain BTV1[VP2]2 and its NS3/NS3a knockout derivative DISA 2. Groups of 25 injected midges were harvested at day 0, 7, and 10 post injection. Viral RNA in heads and bodies was individually tested and semi-quantitated by PCR (Fig. 4). On
We compared a reverse genetics derived parent strain BTV1[VP2]2 (BTV1 expressing VP2 of serotype 2) with the same virus containing a deletion in Seg-10, which abrogates NS3/NS3a expression (DISA 2). Virus growth of DISA 2 in mammalian cells (BSR) is inhibited by the NS3/NS3a knockout mutation, and is delayed (Fig. 2b). Both release and growth of DISA 2 in Culicoides cells is strongly attenuated by the NS3/NS3a knockout mutation (Fig. 2c). BTV1[VP2]2 and DISA 2 grow to lower titers in BSR cells, compared to similar viruses with other serotypes, as already observed in earlier studies [30]. So apparently, serotyping with VP2 from serotype 2 influences in vitro growth in a negative manner, likely by disturbing infection.
Initially, we compared oral infection of C. sonorensis with BTV1[VP2]2 and an American BTV11 isolate. This BTV11 isolate was included to confirm 'normal' feeding behaviour of this specific midge colony and was fed with a very high dose of 10 8.2 TCID 50 ml −1 , since this dose led to virus replication in numerous previous experiments (not shown). Unfortunately, the BTV1[VP2]2 strain was not able to escape the midgut barrier as evidenced by the lack of viral RNA in the heads. The results of day 0 post feeding suggests a similar starting dose in the midguts of both groups by PCR testing. However, the blood meal for BTV1[VP2]2 contained 100 times less infectious virus than that of BTV11 as determined by titration. This might be explained by the presence of non-infectious particles in the BTV1[VP2]2 virus stock, although this needs further investigation. The lower dose could explain the negative results for BTV1[VP2]2, but a similar dose of about 10 6 TCID 50 ml −1 of BTV1, as used in our study for BTV1[VP2]2,has been used already successfully in several vector competence studies using C. sonorensis [32,33,48,49]. Therefore, it is highly unlikely that the used dose is too low to be able to infect the midges and it is not expected that a higher dose would have led to BTV1[VP2]2 replication. In further support of this conclusion, intrathoracic injection of only 46 nl of the same virus stock resulted to similar Cp values at day 0, proficient for virus replication, while it is estimated that fully engorged C. sonorensis midges ingest approximately 50-100 nl of blood (personal communication B. Drolet, U.S. Department of Agriculture). This volume correlates to 50-100 TCID 50 of virus per blood meal when feeding with viremic blood containing 10 6 TCID 50 of BTV ml −1 . Only a very few Culicoides species have been shown as transmission-competent vectors for various serotypes of BTV [32,[50][51][52][53]. Vector competence for a specific arbovirus is affected by multiple factors during the virus-vector interaction, including viral genetics, vector genetics, gut microbiota, physiological barriers, salivary components, environmental temperature, and vector innate immunity [32,33,[54][55][56][57]. Physiological barriers include the MIB, MEB, salivary gland infection barrier and salivary gland escape barrier. For bite transmission of virus after ingestion of BTV1[VP2]2 or DISA2 strain, viruses must exit the midgut (with or without infection), enter the hemocoel, disseminate, and then infect and escape the salivary glands. Based on the PCR results of bodies and heads of orally infected midges, BTV1[VP2]2could not overcome the MIB or MEB, since no virus replication in the bodies and no dissemination to heads was observed. This strongly indicates that, C. sonorensis is not competent for this recombinant virus. No propagation of parental NS3/NS3a expressing strains in the midge vector could make the DISA vaccine even safer regarding vaccine virus spread in the field. Intrathoracic inoculation was performed to circumvent the midgut barriers and to study the ability of disseminated BTV1[VP2]2 and DISA 2 to propagate within Culicoides.
Replication of BTV is likely after intrathoracic injection, even for viruses that do not replicate in the midge after oral uptake [32]. Indeed, in contrast to oral uptake, BTV1[VP2]2 propagated in injected midges. As expected, DISA 2 virus was detected in the heads directly after injection due to dissemination in the hemocoel. The amount of DISA 2 virus RNA was significantly less in heads (p < 0.001) (higher Cp), compared to BTV1[VP2]2 at day 0. This might be due to possible variations in exact sampling times post inoculation for the day 0 time point (1-4 hpi) or differences in amount of virus delivered in the 46 nl injection (10 1.5 vs. 10 2.1 ). Less time between injection and decapitation could lead to incomplete dissemination of virus to the head via the hemolymph. Significantly higher DISA 2 RNA levels (lower Cp value) were measured in the heads at day 7 compared to day 0, which is likely due to full dissemination of injected virus at this time point. Indeed, the Cp value at day 10 was higher again compared to day 7, suggesting degradation of viral RNA. A few individual bodies and heads of DISA 2 fed midges showed a Cp value lower than the average at day 7 and 14. This might indicate a low level of replication in these midges. However, compared to BTV1[VP2]2 fed midges, Cp values are still high, and spread via the salivary glands to the ruminant host is highly unlikely.

Conclusions
In conclusion, NS3/NS3a knockout strongly reduces virus release and propagation in the midge after injection, once the midgut barrier has been passed. This is likely due to the strongly reduced virus release, also seen Culicoides cells. It is also possible that the delayed release of DISA 2 from the cell results in faster virus clearance by intracellular innate immunity such as RNAi [58,59], Toll-Imd, Jak-STAT, NfkB, and autophagy [60].
The virus backbone used in DISA vaccines is based on non-virulent BTV and is safe with respect to fever and clinical signs [61]. Currently, the DISA principle is based on the absence of viremia in the vertebrate host, thereby preventing uptake by insect vectors [29]. Here, the safety of the DISA principle is supported by the absence of DISA 2 propagation in C. sonorensis.
Live-modified vaccines for orbiviruses potentially imply several risks, such as reversion to virulence, reassortment events and uncontrolled spread of vaccine virus. An elegant method to combat these major disadvantages is to target functions that are specific and essential for virus propagation in the insect vector, and for virus transmission between host and vector. Most likely, similar NS3/NS3a deletion mutants for other orbiviruses will have similar characteristics [62], whereas targeting functions essential for vector transmission in other arboviral pathogens will also result in safe vaccine candidates.