Reporter Flaviviruses as Tools to Demonstrate Homologous and Heterologous Superinfection Exclusion

Binjari virus (BinJV) is a lineage II or dual-host affiliated insect-specific flavivirus previously demonstrated as replication-deficient in vertebrate cells. Previous studies have shown that BinJV is tolerant to exchanging its structural proteins (prM-E) with pathogenic flaviviruses, making it a safe backbone for flavivirus vaccines. Here, we report generation by circular polymerase extension reaction of BinJV expressing zsGreen or mCherry fluorescent protein. Recovered BinJV reporter viruses grew to high titres (107−8 FFU/mL) in Aedes albopictus C6/36 cells assayed using immunoplaque assays (iPA). We also demonstrate that BinJV reporters could be semi-quantified live in vitro using a fluorescence microplate reader with an observed linear correlation between quantified fluorescence of BinJV reporter virus-infected C6/36 cells and iPA-quantitated virus titres. The utility of the BinJV reporter viruses was then examined in homologous and heterologous superinfection exclusion assays. We demonstrate that primary infection of C6/36 cells with BinJVzsGreen completely inhibits a secondary infection with homologous BinJVmCherry or heterologous ZIKVmCherry using fluorescence microscopy and virus quantitation by iPA. Finally, BinJVzsGreen infections were examined in vivo by microinjection of Aedes aegypti with BinJVzsGreen. At seven days post-infection, a strong fluorescence in the vicinity of salivary glands was detected in frozen sections. This is the first report on the construction of reporter viruses for lineage II insect-specific flaviviruses and establishes a tractable system for exploring flavivirus superinfection exclusion in vitro and in vivo.


Introduction
The Flavivirus genus (family Flaviviridae) contains several medically important viruses, including yellow fever (YFV), dengue (DENV), and Zika (ZIKV), which are responsible for millions of global infections annually, resulting in a range of disease outcomes in humans [1,2]. These pathogenic flaviviruses exist in transmission cycles that primarily involve horizontal transmission between mosquitoes and amplifying vertebrate hosts.
In addition to vector-borne flaviviruses that can infect vertebrates (vertebrate-infecting flaviviruses, VIFs) and cause disease, several insect-specific flaviviruses (ISFs), named for their insect-restricted host range, have been isolated and characterised (reviewed by [7]). They belong to two phylogenetically distinct groupings: lineage I or classical ISFs (cISF), in which all members form a monophyletic clade, and a lineage II or dual-host-affiliated ISF (dISF) clade of flaviviruses that also cannot replicate in mammalian cells but are closely related phylogenetically to VIFs [7].
In addition to using ISFs for safe flavivirus vaccine backbones, there has been a considerable examination of exploiting ISFs and other insect-specific viruses (ISVs) in biocontrol strategies to control vector-borne disease [18,19]. This is due to the observation that primary or co-infections of ISVs have been shown to restrict or modulate secondary flavivirus infections in vitro [20][21][22][23][24] and in vivo in mosquitoes [23,25].
Fluorescent imaging of viruses encoding fluorescent reporter groups (i.e., zsGreen, mCherry) enables examination of virus replication or virus protein localisation in living organisms. Furthermore, it provides a chance to investigate their real-time dynamics without needing expensive imaging reagents [26].
The first infectious flavivirus reporter virus was engineered for Japanese encephalitis virus (JEV) in 2003 and contained GFP and luciferase in the 3 UTR [27]. Unfortunately, early reporter viruses were plagued with genetic instability and spontaneous loss of the insert (Reviewed by [28]). A YFV reporter virus described in 2007 [29] improved stability by fusing the first 101-aa capsid sequence with GFP, followed by a foot-and-mouth disease virus 2A ribosome stuttering peptide (FMDV 2A), codon-optimised capsid gene, and the rest of YFV polyprotein [29]. Since these early studies, many reporter viruses for the VIF flaviviruses have been constructed [28]. In contrast, only one report has demonstrated reporter viruses' construction for the Lineage I ISF Niénokoué virus [30].
Here, we generated BinJV reporter viruses expressing the fluorescent proteins zs-Green or mCherry inserted between partial and complete capsid gene sequences using our established Circular Polymerase Extension Reaction (CPER) method [31,32]. The CPER method allows de novo generation of infectious DNA from viral cDNA fragments with 20 to 40-nt complementary overlapping ends and a linker fragment containing insect expression promoter.
We examine the replication kinetics of recovered BinJV reporters compared to WT BinJV in vitro in Aedes albopictus cells and in vivo via intrathoracic injection in Aedes aegypti. The BinJV reporter viruses were then utilised to explore the homologous exclusion phenomenon or heterologous exclusion with the ZIKV reporter virus [33]. FCS and 1% GlutaMAX (Gibco). For viral infections, cells were maintained in an FCS reduced media (2% FCS) with 10,000 U/mL of penicillin and streptomycin (Gibco).

Construction and Amplification of Reporter and Linker Fragments
The complete 96 nt BinJV 5 UTR previously characterised [35], the first 153 bp of the BinJV capsid gene, the zsGreen or mCherry fluorescent protein gene, a foot-and-mouth disease virus (FMDV) 2A ribosome stuttering peptide sequence (LLNFDLLKLAGDVESNPG↓P), the full-length codon optimised BinJV capsid gene, and the 5 end of BinJV prM gene were ordered as a gBlocks Gene Fragments (IDT, Singapore), blunt-end cloned into the pUC19 vector, and confirmed by Sanger sequencing. These plasmids were then used in PCR as templates with primers Reporter_F 5 -GCCTATAAATACAGCCCGCAACGATCTGGTAAACA GTATATTTTGCGTGTGCGTTTCAAA-3 and Reporter_R 5 -TGCGTAAAGTGGCAGCCAT CGCTCCT-3 to yield 1412 and 1427bp fragments (F1) for zsGreen and mCherry reporter genes, respectively. An insect-optimised OpIE2-CA promoter-UTR linker fragment was amplified from the OpIE2-CA-UTR linker plasmid [11] with the following primer pair OpIE2_Link_F 5 -TGGATTGGGGATTGAGAAACATCAAGTATCTGGGTCGGCATGGCA TCTCCACCTCCTCGC-3 and OpIE2_Link_R 5 -TTTGAAACGCACACGCAAAATATACT GTTTACCAGATCGTTGCGGGCTGTATTTATAGGC-3 . The OpIE2-CA promoter-UTR linker fragment contains the last 20 nucleotides of BinJV 3 UT and the hepatitis delta virus ribozyme (HDVr) to generate authentic 3 -OH RNA. The linker also contains an SV40 polyA signal for transcription termination, a modified baculovirus promoter (OpIE2-CA) [11] for in vivo transcription of viral RNA by cellular RNA polymerase II, and the first 20 nucleotides of BinJV 5 UTR.

CPER Assembly, Recovery of Reporter Viruses and RT-PCR of Viral RNA
The purified reporter fragment, two BinJV viral dsDNA fragments, and linker fragment containing overlapping sequences of 26-60 nucleotides were pooled in equimolar amounts (0.1 pM each) in a 50 µL reaction containing dNTPs (200 µM each) and 2µL of Prime Star GXL DNA polymerase in a GXL reaction buffer. The following cycling conditions were used: Stage 1: 98 • C (30 s), Stage 2: 12 cycles of 98 • C (10 s), 55 • C (20 s), and 68 • C (10 min). Infectious cDNA from CPER assembly was transfected into C6/36 cells using Effectene transfection reagent (QIAGEN, Hilden, Germany) according to the manufacturer's instructions. To generate P 1 stocks and confirm the reporter genes' retention in the BinJV recombinants, cell culture fluid (P 0 ) was harvested on day ten post-transfection and blind-passaged once onto C6/36 cells. At seven days postinfection, viral RNA was isolated, and cDNA was prepared as above. PCR was performed with PrimeStar GXL Polymerase (Takara, Japan) and with the following primer

Growth Kinetics and Immuno-Plaque Assay (iPA)
Growth kinetics of wild-type and reporter BinJV were assessed in C6/36 cells. Cells were seeded in six-well plates at 10 6 cells per well and inoculated with BinJV WT , BinJV mCherry or BinJV zsGreen at a multiplicity of infection (MOI) of 0.1 for one hour. After removal of the inoculum, cells were washed three times and replaced with 2 mL of RPMI (2% FCS). Infected cells were incubated for seven days at 28 • C with culture fluid harvested at zero, three, five, and seven days post-infection. Viral titres of the harvested supernatant were quantitated with a foci-forming immunoplaque assay [34,36]. . After blocking, plates were incubated with 50 µL/well of primary antibody for one hour. BinJV WT and reporter viruses were titred using a primary mouse anti-NS1 flavivirus antibody 4G4 [37] (1:100). Zika was titred using a primary human anti-ZIKV E protein antibody Z67 (1:5000) [38,39]. Recovered P 1 ZIKV mCherry used for super-exclusion experiments were titred with mouse 4G4 (1:100). Cells were subsequently incubated with 50 µL/well of goat anti-human IRDye 800CW secondary antibody (1:2000) or goat anti-mouse IRDye 800CW secondary antibody (1:2000) (LI-COR). All antibodies were diluted with Clear Milk blocking buffer (Pierce) in phosphatebuffered saline containing 0.05% Tween 20 (PBS-T). Primary and secondary incubations were performed at room temperature for one hour. After each incubation, plates were washed 5 × 5 min with PBS-T. Plates were then scanned using an Odyssey CLx Imaging System (42 µm; medium; 3.0 mm), foci-forming units (FFU) were counted using the Image Studio Lite software (v 5.2.5) (LI-COR, Lincoln, NE, USA), and titres were calculated as FFU per mL (FFU/mL).

Superinfection Exclusion Assays
C6/36 cells were seeded in triplicates in 24-well black, glass-bottom plates (Eppendorf) and infected at~90% confluence with BinJV zsGreen at a MOI = 1 or "mock-infected" with media only and incubated at 28 • C. Cells were observed daily for five days under green channel of an inverted epifluorescence microscope. Five days post-infection demonstrated the time when green fluorescence was predominant in cell monolayers. After five days, the growth medium was removed, and cells were washed three times with RPMI and infected with either BinJV mCherry at an MOI of 1, or ZIKV mCherry at an MOI of 1, or mock-infected. Finally, plates were incubated for another five days, with images of cells taken under green and red fluorescent light with the camera of the inverted epifluorescence microscope. In parallel, samples were taken immediately after infection with either homologous (BinJV mCherry ) or heterologous (ZIKV mCherry ) viruses for 0-time points and every day for five days. Culture supernatant from each experiment was taken daily. with 100 µL of RPMI culture medium with 2% FCS. Plates were scanned daily using the Varioskan™ LUX (Thermo Scientific) microplate reader. The bottom of the wells were analysed with the SkanIt™ Software v6.0.1 (ThermoFisher Scientific, Waltham, MA, USA) under the following conditions: 29 points per well, measurement time: 100 ms, excitation bandwidth: 5 nm and automatic dynamic range. For measurement of zsGreen fluorescence in cells, the excitation wavelength (λ ex ) was 488 nm, and the emission wavelength (λ em ) was 506 nm. For quantification of mCherry fluorescence in cells, the λ ex was 585 nm and the λ em 610 nm. For linear regression analysis, plates were read at five days post-infection.

Microinjection of Aedes aegypti
Wildtype Aedes aegypti (Cairns strain) used for infection have been maintained since 2015 in the QIMR Berghofer insectary at 28 • C with 70% relative humidity and 12:12 hr light cycling with dawn and dusk fading. Adults were maintained in 30 cm cages (Bug-Dorm, MegaView Science Education Services Co., Ltd., Taichung, Taiwan). For colony maintenance, adults were fed defibrinated sheep blood (Serum Australis, Manila, NSW, Australia) weekly and continuously provided with 10% sucrose solution.

Histological Analysis and Fluorescence Microscopy
For frozen sectioning, mosquitoes were embedded in Optimal Cutting Temperature Compound (OCT) and frozen in an ethanol and dry ice slurry. Mid-sagittal sections (12 µM) were cut using a cryostat microtome. Sections were stained in 0.14 µg/mL 4 ,6-diamidino-2-phenylindole (DAPI) for 5 min and washed in tris-buffered saline-0.025% Tween 20. Microscopy was performed using an Aperio ScanScope fluorescent microscope using filters for DAPI and FITC with exposure times of 0.2 s and 0.125 s, respectively. In preliminary experiments, mosquito specimens were also fixed in 4% paraformaldehyde, 0.5% Triton X overnight, transferred to 70% ethanol, embedded in paraffin, and sectioned using standard procedures [40]. CPER-transfected and BinJV or ZIKV reporter virus-infected C6/36 cells were visualised and imaged using an inverted Nikon ECLIPSE Ts2 microscope.

Statistical Analysis
A two-way ANOVA test with Dunnett's correction was used for growth kinetics to compare viruses at each time point to WT BinJV. A general linear analysis was used for Figure 2D,E. All data were analysed and visualised using GraphPad Prism software v9.0.0 (San Diego, CA, USA). The level of statistical significance was set at 95% (p ≤ 0.05).

Construction and Generation of BinJV Expressing zsGreen and mCherry Viruses
The reporter gene insertion strategy, which permits the stable expression of fluorescent proteins in insect cells infected with flaviviruses [33,41,42], was employed to characterise BinJV properties. To this end, the CPER platform [8,16,31] was utilised to insert sequences of zsGreen or mCherry genes between the first 153 nucleotides of the BinJV capsid gene, containing cis-acting RNA elements essential for virus RNA replication [43][44][45] and the full-length BinJV capsid gene, with codon optimisation to prevent recombination and potential loss of reporter genes during virus replication ( Figure 1A) [33]. Both reporters also contained an FMDV 2A ribosomal skipping sequence at the C-terminus to ensure the release of reporter proteins from the downstream viral polyprotein ( Figure 1A). PCR amplification and electrophoresis of the two reporter gene fragments (F1) along with two Viruses 2022, 14, 1501 6 of 16 additional fragments encompassing the BinJV genome (F2 and F3) and the insect optimised OpIE2-CA promoter-UTR linker fragment showed correct sizes of the five CPER fragments ( Figure 1B).
characterise BinJV properties. To this end, the CPER platform [8,16,31] was utilised to in-sert sequences of zsGreen or mCherry genes between the first 153 nucleotides of the BinJV capsid gene, containing cis-acting RNA elements essential for virus RNA replication [43][44][45] and the full-length BinJV capsid gene, with codon optimisation to prevent recombination and potential loss of reporter genes during virus replication ( Figure 1A) [33]. Both reporters also contained an FMDV 2A ribosomal skipping sequence at the C-terminus to ensure the release of reporter proteins from the downstream viral polyprotein ( Figure 1A). PCR amplification and electrophoresis of the two reporter gene fragments (F1) along with two additional fragments encompassing the BinJV genome (F2 and F3) and the insect optimised OpIE2-CA promoter-UTR linker fragment showed correct sizes of the five CPER fragments ( Figure 1B). Amplicons were pooled in equimolar amounts and circularised using CPER assembly using GXL DNA polymerase. The CPER was then transfected into C6/36 cells and visualised daily from two days post-transfection over nine days with an inverted epifluorescence microscope ( Figure 1C). We observed the signal of zsGreen positive cells in the green channel at four days post-transfection and the mCherry signal in the red channel at five days post-transfection. Both transfected cell lines showed increased signal from day six to day ten when the supernatant containing the infectious virus was harvested. No autofluorescence of mock-transfected (media only) cells was observed.
The harvested culture supernatant containing passage 0 (P0) BinJV reporter viruses was then passaged in C6/36 cells for seven days. The resultant passage 1 (P1) stock was Amplicons were pooled in equimolar amounts and circularised using CPER assembly using GXL DNA polymerase. The CPER was then transfected into C6/36 cells and visualised daily from two days post-transfection over nine days with an inverted epifluorescence microscope ( Figure 1C). We observed the signal of zsGreen positive cells in the green channel at four days post-transfection and the mCherry signal in the red channel at five days post-transfection. Both transfected cell lines showed increased signal from day six to day ten when the supernatant containing the infectious virus was harvested. No autofluorescence of mock-transfected (media only) cells was observed.
The harvested culture supernatant containing passage 0 (P 0 ) BinJV reporter viruses was then passaged in C6/36 cells for seven days. The resultant passage 1 (P 1 ) stock was quantitated using a pan flavivirus antibody 4G4 by iPA. The titres were 1.86 × 10 6 FFU/mL for BinJV zsGreen virus and 1.14 × 10 6 FFU/mL for the BinJV mCherry virus. The P 1 BinJV reporter viruses were also examined for the retention of the reporter gene insert by RT-PCR, with the reporter genes shown to be retained in the P 1 viruses ( Figure 1D).

Growth Kinetics and In-Cell Fluorescence Quantification of BinJV Reporter Viruses
To examine the kinetics of the recovered BinJV reporters compared to CPER generated WT BinJV [35], Aedes albopictus C6/36 cells were infected at an MOI of 0.1 with P 1 virus stocks. Culture supernatants were harvested at days zero, three, five, and seven after infection and titred by iPA using the 4G4 antibody (Figure 2A).

Primary Infection of Mosquito Cells with BinJVzsGreen Inhibits Subsequent Infection with BinJVmCherry
To test the utility of BinJV reporters for superinfection exclusion experiments, we first examined the ability of infection with one BinJV to prevent the following infection with For BinJV zsGreen, the virus steadily increased to mean final titres of 4.27 × 10 8 FFU/mL at seven days from 2.48 × 10 8 FFU/mL and 2.81 × 10 8 FFU/mL at three and five days, respectively (Figure 2A). For BinJV mCherry , replication peaked at three days post-infection for a mean titre of 5.57 × 10 7 FFU/mL and declined to 4.37 × 10 7 FFU/mL on day five with final titres of 6.17 × 10 7 FFU/mL at seven days. Conversely, the WT BinJV grew to consistently higher titres than both reporter viruses, with mean titres of 3.79 × 10 8 , 1 × 10 9 and 2.35 × 10 9 FFU/mL at three, five, and seven days, respectively. Comparative statistical analysis between BinJV reporters and WT virus titres was undertaken with a two-way ANOVA with Dunnett's correction. Both reporters were significantly different on days five (BinJV zsGreen ; p < 0.05, BinJV mCherry ; p < 0.0001) and seven (BinJV zsGreen ; p < 0.001, BinJV mCherry ; p < 0.0001), the BinJV zsGreen virus was the closest in replication kinetics to WT BinJV and was no different in titres at day three. To examine if the BinJV reporters could be quantified live and non-invasively using fluorescence microplate readers, we infected C6/36 cells with serially diluted culture supernatant from the growth kinetics experiment (Figure 2A) in black optical-bottom plates. The fluorescence signals of mCherry and zsGreen were read daily for seven days. Signal was not detected in any reporter virusinfected C6/36 cells until day three, with signal increasing incrementally from day 3 to 5 (data not shown). We noted no cross-reactivity or contamination for either reporter virus, and auto-fluorescence was not present in both mock-infected and BinJV WT-infected C6/36 cells. BinJV reporters gave reliable fluorescence signals ( Figure 2B, C) above background fluorescence in mock-infected cells (~0 Fluorescence Units for mCherry and~0.2 Fluorescence Units for zsGreen). They showed reduced signal in cells infected with serially diluted virus samples. The signal could be reliably detected in cells infected with dilutions 10 −1 to 10 −3 for BinJV mCherry and 10 −1 to 10 −4 for BinJV zsGreen .
Overall, the BinJV mCherry signal units range was the highest in C6/36 cells compared to BinJV zsGreen . To examine if it was possible to generate a standard curve for relative fluorescence units and iPA titres, a simple linear relationship between the relative fluorescence units and viral titres generated from iPA-matched samples were plotted and analysed ( Figure 2D,E). Three mock-infected wells were also added to calibrate the baseline at 0 FFL/mL. Generally, there was a strong (R 2 = 0.855) linear relationship between quantitated fluorescence and iPA titres for the BinJV mCherry ( Figure 2D) and a more weak but positive general trend (R 2 = 0.539) for the BinJV zsGreen reporter virus ( Figure 2E). The slope for both linear equations was significantly non-zero (BinJV zsGreen ; p = 0.0003, BinJV mCherry ; <0.0001). This indicates that instead of expensive antibody reagents and scanning equipment used for iPA assay, fluorescence microplate readers may provide a reliable semi-quantifiable alternative to reporter virus titration.

Primary Infection of Mosquito Cells with BinJV zsGreen Inhibits Subsequent Infection with BinJV mCherry
To test the utility of BinJV reporters for superinfection exclusion experiments, we first examined the ability of infection with one BinJV to prevent the following infection with another BinJV. The format for superinfection exclusion assays involved a primary infection of C6/36 cells with BinJV zsGreen virus at an MOI = 1 for five days. Mock-infected cells were used as a control. After five days, cells were washed and infected with BinJV mCherry at a MOI = 1 for another five days.
Infection with BinJV zsGreen alone continued throughout all 10 days of the experiment ( Figure 3A). Infection with BinJV mCherry without prior infection with BinJV zsGreen also showed a strong increase in mCherry signal over the five days of infection ( Figure 3B). In stark contrast, cells previously infected with BinJV zsGreen were completely protected from a secondary infection with BinJV mCherry (Figure 3C), clearly demonstrating a strong interference. No red or green signals were observed in mock controls ( Figure 3D). Infection with BinJVzsGreen alone continued throughout all 10 days of the experiment ( Figure 3A). Infection with BinJVmCherry without prior infection with BinJVzsGreen also showed a strong increase in mCherry signal over the five days of infection ( Figure 3B). In stark contrast, cells previously infected with BinJVzsGreen were completely protected from a secondary infection with BinJVmCherry ( Figure 3C), clearly demonstrating a strong interference. No red or green signals were observed in mock controls ( Figure 3D).

Primary Infection of Mosquito Cells with BinJV Completely Inhibits Subsequent Infection with ZIKV
Given the previous reports of primary ISF infection leading to restriction of a secondary infection of VIF in mosquito bodies or cells [22][23][24][25], we wanted to examine the utility of our BinJV reporters in the potential exclusion of ZIKV.
We used a ZIKVmCherry reporter virus generated based on a published design [33]. Primary and secondary infections were undertaken with the same conditions as in Figure  3, except for the secondary infections with ZIKVmCherry (MOI = 1). As per the previous experiment, cells were monitored daily using inverted epifluorescence microscopy on green and red channels. Here, primary infections of BinJVzsGreen and mock (media only) secondary infections (Figure 4A), and mock primary infection and secondary infection with ZIK-VmCherry (MOI = 1) ( Figure 4B) were undertaken as controls. C6/36 cells previously infected with BinJVzsGreen virus completely inhibited a secondary infection with ZIKVmCherry virus ( Figure 4C). This was in contrast to mock-infected cells infected with ZIKVmCherry, in which

Primary Infection of Mosquito Cells with BinJV Completely Inhibits Subsequent Infection with ZIKV
Given the previous reports of primary ISF infection leading to restriction of a secondary infection of VIF in mosquito bodies or cells [22][23][24][25], we wanted to examine the utility of our BinJV reporters in the potential exclusion of ZIKV.
We used a ZIKV mCherry reporter virus generated based on a published design [33]. Primary and secondary infections were undertaken with the same conditions as in Figure 3, except for the secondary infections with ZIKV mCherry (MOI = 1). As per the previous experiment, cells were monitored daily using inverted epifluorescence microscopy on green and red channels. Here, primary infections of BinJV zsGreen and mock (media only) secondary infections ( Figure 4A), and mock primary infection and secondary infection with ZIKV mCherry (MOI = 1) ( Figure 4B) were undertaken as controls. C6/36 cells previously infected with BinJV zsGreen virus completely inhibited a secondary infection with ZIKV mCherry virus ( Figure 4C). This was in contrast to mock-infected cells infected with ZIKV mCherry , in which a strong increase in mCherry signal was observed from day three to day five after infection ( Figure 4B). No red or green signals were observed in mock controls ( Figure 4D).

OR PEER REVIEW
10 of 15 a strong increase in mCherry signal was observed from day three to day five after infection ( Figure 4B). No red or green signals were observed in mock controls ( Figure 4D). To validate the epifluorescence imaging results, we harvested supernatants of infected cells for experiments 4B and 4C and infected C6/36 cells quantitating viral titres of culture supernatants with iPA shown in Figure 4B (ZIKVmCherry only) and Figure 4C (Primary infection with BinJVzsGreen and secondary infection with ZIKVmCherry). To detect only ZIKV in the BinJV/ZIKV co-infection experiment, a ZIKV-specific anti-E antibody (Z67) was used to titrate the virus on mammalian Vero76 cells ( Figure 4E) and mosquito C6/36 cells ( Figure 4F). The results confirmed the lack of productive infection with ZIKVmCherry To validate the epifluorescence imaging results, we harvested supernatants of infected cells for experiments 4B and 4C and infected C6/36 cells quantitating viral titres of culture supernatants with iPA shown in Figure 4B (ZIKV mCherry only) and Figure 4C  in the BinJV/ZIKV co-infection experiment, a ZIKV-specific anti-E antibody (Z67) was used to titrate the virus on mammalian Vero76 cells ( Figure 4E) and mosquito C6/36 cells ( Figure 4F). The results confirmed the lack of productive infection with ZIKV mCherry virus in the culture fluid of cells first infected with BinJV zsGreen followed by infection with ZIKV mCherry and the increasing over time viral titres in the culture fluid of cells infected only with ZIKV mCherry .

BinJV zsGreen Infection Can Be Visualised In Vivo
Given the success of the visualisation of BinJV reporters in vitro we sought to examine the suitability of the BinJV zsGreen in a preliminary study of intrathoracic injection of adult Ae. aegypti mosquitoes. Seven-to ten-day-old mosquitoes were injected with ≈372 FFU of virus per mosquito, and mosquitoes were frozen at seven days post-infection.
We conducted standard frozen sectioning to avoid harsh processing steps used for paraffin embedding. During the frozen sectioning of mosquitoes, the morphology of specimens was disrupted. Despite this, a robust zsGreen signal was detected in most mosquitoes at seven days post-infection (representative mosquito shown in Figure 5A), with the majority of zsGreen signal localised to small areas near the salivary glands ( Figure 5B).

BinJVzsGreen Infection Can Be Visualised In Vivo
Given the success of the visualisation of BinJV reporters in vitro we sought to examine the suitability of the BinJVzsGreen in a preliminary study of intrathoracic injection of adult Ae. aegypti mosquitoes. Seven-to ten-day-old mosquitoes were injected with ≈372 FFU of virus per mosquito, and mosquitoes were frozen at seven days post-infection.
We conducted standard frozen sectioning to avoid harsh processing steps used for paraffin embedding. During the frozen sectioning of mosquitoes, the morphology of specimens was disrupted. Despite this, a robust zsGreen signal was detected in most mosquitoes at seven days post-infection (representative mosquito shown in Figure 5A), with the majority of zsGreen signal localised to small areas near the salivary glands ( Figure 5B). Preliminary attempts to examine the zsGreen signal in mosquito samples that were paraffin fixed showed that the zsGreen signal was lost during the procedure with few exceptions (data not shown). While further optimisation is required, these data indicate that BinJVzsGreen reporter can be visualised in vivo.

Discussion
Here, we report the construction of lineage II ISF BinJV reporter viruses encoding fluorescent proteins zsGreen or mCherry. BinJV reporter viruses provide a powerful tool to study viral infection and pathogenesis of the insect-restricted flaviviruses. Using fluorescence scanning, we demonstrate that BinJV reporter viruses can be semi-quantitated in vitro. We also show that visualisation of BinJVzsGreen is possible in vivo in mosquito tissues.
We show that the BinJVzsGreen reporter viruses induced superinfection exclusion of homologous infections (inhibition of the other BinJV reporter virus) and a heterologous infection, from a ZIKVmCherry reported virus. This demonstrates that infection of Ae. albopictus cells with BinJV completely inhibited subsequent infection with a pathogenic VIF, providing strong support for the potential application of insect-specific flaviviruses as biological control agents to inhibit pathogenic virus infections in mosquitoes.
Initial studies on the construction of reporter flaviviruses used various approaches for the insertion of reporter genes, primarily internal ribosome entry site (IRES)-based Preliminary attempts to examine the zsGreen signal in mosquito samples that were paraffin fixed showed that the zsGreen signal was lost during the procedure with few exceptions (data not shown). While further optimisation is required, these data indicate that BinJV zsGreen reporter can be visualised in vivo.

Discussion
Here, we report the construction of lineage II ISF BinJV reporter viruses encoding fluorescent proteins zsGreen or mCherry. BinJV reporter viruses provide a powerful tool to study viral infection and pathogenesis of the insect-restricted flaviviruses. Using fluorescence scanning, we demonstrate that BinJV reporter viruses can be semi-quantitated in vitro. We also show that visualisation of BinJV zsGreen is possible in vivo in mosquito tissues.
We show that the BinJV zsGreen reporter viruses induced superinfection exclusion of homologous infections (inhibition of the other BinJV reporter virus) and a heterologous infection, from a ZIKV mCherry reported virus. This demonstrates that infection of Ae. albopictus cells with BinJV completely inhibited subsequent infection with a pathogenic VIF, providing strong support for the potential application of insect-specific flaviviruses as biological control agents to inhibit pathogenic virus infections in mosquitoes.
Initial studies on the construction of reporter flaviviruses used various approaches for the insertion of reporter genes, primarily internal ribosome entry site (IRES)-based systems (reviewed by [28]). However, these initial construction methods typically resulted in instability of the reporter genes after serial passaging and virus attenuation due to recombination occurring at the insertion site [28]. Successive rounds of improved method design have led to the capsid duplication strategy, whereby reporter gene stabilisation is attained by introducing the reporter gene between the first 38-50 codons of the duplicated flavivirus capsid gene and a partially or fully codon-optimised complete capsid gene to prevent homology-directed recombination that can hinder efficient reporter virus generation [33,46,47]. However, studies exploiting this strategy have relied on the construction of infectious clones that require propagation in bacteria, increasing the likelihood of mutations compromising reporter gene stability and virus replication. Here, we employed the CPER method combined with the capsid duplication strategy to recover recombinant insect-specific reporter viruses. The CPER-generated viruses retained their reporter genes, evidencing the lack of homology-directed recombination between capsid sequences. Similarly, fluorescence emitted by zsGreen and mCherry proteins were visualised in CPERtransfected cells as early as four days after transfection, indicating successful expression of reporters by these viruses in vitro.
To our knowledge, the semi-quantification of reporter viruses in vitro using microplate readers described here is a reliable quantification strategy for virus titration. It allows the detection and quantification of BinJV without expensive secondary labelling reagents and may be amenable to other virus reporter systems. Additionally, it offers a noninvasive strategy to monitor the dynamics of viral infections in vitro. Finally, it improves the utility of BinJV reporter viruses for fast, high-throughput screening such as those used in whole genome, RNAi-based gene screening to identify viral receptors or antiviral host factors.
While neither reporter virus grew identical to the WT BinJV, the reporter viruses grew to sufficient titres for downstream applications such as superinfection exclusion assays. Superinfection exclusion is commonly observed in virus-virus interactions due to intracellular bottlenecks [48], with multiple reports of superinfection exclusion of flaviviruses in vitro [22][23][24][25]. The current mechanism of flavivirus superinfection exclusion is unknown. Competition for resources or perturbation of the host immune response may be two possibilities [48,49]. Previous studies with a similar regimen of infection with cell fusing agent virus (CFAV) (a lineage one ISF) in C6/36 cells have shown that CFAV decreases but does not entirely restrict a secondary ZIKV infection [23]. Previous studies using superinfection exclusion in DENV isolates in C6/36 cells suggest that exclusion is more significant between more closely related DENV isolates [50]. Given the close genetic relationship between BinJV and ZIKV and the complete inhibition of secondary infection with ZIKV observed here, dISFs may be more suitable for biocontrol. It is also possible that the high replication efficiency of BinJV may account for more efficient inhibition of subsequent infection with ZIKV.
Here, we have demonstrated that primary infection of C6/36 cells with BinJV zsGreen completely inhibits a secondary infection with homologous BinJV mCherry or heterologous ZIKV mCherry using fluorescence microscopy and virus quantitation by iPA. We did not perform a reciprocal experiment with primary infection with BinJV mCherry virus and secondary infection with BinJV zsGreen or an experiment with primary infection with WT BinJV and secondary infections with reporter BinJV or ZIKV viruses. However, data in the literature [22][23][24][25] suggest that these experiments would likely generate results similar to those observed in our experiments.
While a robust signal of reporter genes could be visualised in vitro in C6/36 cells, only a small region of zsGreen fluorescence could be visualised in BinJV zsGreen infected Ae. aegypti tissues. The reasons for this are currently unclear. While it is unlikely that BinJV zsGreen has a unique tissue tropism for the salivary gland tissues, it has been previously shown for certain flaviviruses that virus replication is observed preferentially in the proximal lateral and medial lobes of the salivary glands [51,52]. Electron microscopy has also identified the presence of crystalline flavivirus arrays in the salivary glands of vector mosquitoes [53], the formation of which might cause significant zsGreen signal production. Beyond the potential biochemical differences between the salivary glands and other tissues [54], our failure to detect the zsGreen signal might be due to insufficient signal in the zsGreen channel or lower viral titres in other tissues. In naturally blood-fed mosquitoes, there is a temporal tropism of flavivirus infection, with the salivary glands and head being among the last tissues to become infected at 7-10 days [52]. Given that the salivary glands are among the last tissues infected, it is unlikely that BinJV zsGreen has lost the reporter gene as the virus has spread through multiple tissue barriers before entry to the salivary glands. Future work to examine if the lack of reporter gene expression in other tissues is a technical limitation of the system could include detection of BinJV WT or BinJV zsGreen by immunofluorescent staining for a viral protein and cross-referencing it with direct reporter protein fluorescence. In addition, tissue-specific RT-qPCR assays for a BinJV gene or reporter genes could be performed to reveal if these reporter viruses faithfully reproduce their WT BinJV counterpart in vivo.
There has been a recent push toward discovering and characterising ISVs in vector mosquitoes [55][56][57]. However, the interactions between ISV/ISFs and pathogenic arboviruses in the mosquitoes remain poorly defined. We anticipate that the reporter viruses described here will facilitate future superinfection exclusion studies, allowing the investigation of the utility of ISFs as potential biocontrol of pathogenic flaviviruses.