Aedes aegypti (Aag2)-derived clonal mosquito cell lines reveal the impact of pre-existing persistent infection with the insect-specific bunyavirus Phasi Charoen-like virus on arbovirus replication

Background Aedes aegypti is a vector mosquito of major public health importance, transmitting arthropod-borne viruses (arboviruses) such as chikungunya, dengue, yellow fever and Zika viruses. Wild mosquito populations are persistently infected at high prevalence with insect-specific viruses that do not replicate in vertebrate hosts. In experimental settings, acute infections with insect-specific viruses have been shown to modulate arbovirus infection and transmission in Ae. aegypti and other vector mosquitoes. However, the impact of persistent insect-specific virus infections that more closely mimic the situation in nature has not been investigated extensively. Cell lines are useful models for studying virus-host interactions, however the available Ae. aegypti cell lines are poorly defined and heterogenous cultures. Methodology/Principle Findings We generated single cell-derived clonal cell lines from the commonly used Ae. aegypti cell line Aag2. Two of the fourteen Aag2-derived clonal cell lines generated harboured markedly and consistently reduced levels of the insect-specific bunyavirus Phasi Charoen-like virus (PCLV) known to persistently infect Aag2 cells. In contrast to studies with acute insect-specific virus infections in cell culture and in vivo, we found that pre-existing persistent PCLV infection had no major impact on the replication of the flaviviruses dengue virus and Zika virus, the alphavirus Sindbis virus, or the rhabdovirus vesicular stomatitis virus. We also performed a detailed characterisation of the morphology, transfection efficiency and immune status of our Aag2-derived clonal cell lines, and have made a clone that we term Aag2-AF5 available to the research community as a well-defined cell culture model for arbovirus-vector interaction studies. Conclusions/Significance Our findings highlight the need for further in vivo studies that more closely recapitulate natural arbovirus transmission settings in which arboviruses encounter mosquitoes harbouring persistent rather than acute insect-specific virus infections. Furthermore, we provide the well-characterised Aag2-derived clonal cell line as a valuable resource to the arbovirus research community. AUTHOR SUMMARY Mosquito-borne viruses usually only infect humans through the bite of a mosquito that carries the virus. Viruses transmitted by the ‘yellow fever mosquito’ Aedes aegypti, including dengue virus, Zika virus, yellow fever virus and chikungunya virus, are causing an ever-increasing number of human disease cases globally. Mosquito-borne viruses have to infect and replicate inside the mosquito before they are transmitted to humans, and the presence of other infectious agents can change the efficiency of virus transmission. Mosquitoes are known to be infected with ‘insect-specific viruses’ that only infect mosquitoes and cannot cause human disease. We have shown here that in laboratory cell cultures derived from the Aedes aegypti mosquito, pre-existing infection with an insect-specific virus called Phasi Charoen-like virus does not affect the infection and growth of the mosquito-borne viruses dengue virus, Zika virus, Sindbis virus or vesicular stomatitis virus. Compared to previous research, our research is more reflective of conditions that mosquito-borne viruses encounter in nature, and our results provide important new insights into whether and how insect-specific viruses affect mosquito-borne virus transmission. Ultimately, this information could inform ongoing research into whether insect-specific viruses could be used to prevent the transmission of mosquito-borne viruses to reduce global disease burdens.

Arthropod-borne viruses (arboviruses) are a major public health concern worldwide, 73 with many considered emerging or re-emerging pathogens [1]. Significant taxons to which 74 arboviruses belong include the positive-sense single-stranded RNA (+ssRNA) families 75 Flaviviridae (genus Flavivirus) and Togaviridae (genus Alphavirus), and the negative-sense 76 single-stranded RNA (-ssRNA) order Bunyavirales and family Rhabdoviridae (genus 77 Vesiculovirus). Many arboviral taxons also include related insect-specific viruses that can 78 infect vector insects but not vertebrate hosts [2,3]. Arboviruses transmitted by the vector 79 mosquito Aedes aegypti are of particular concern to human health, as this mosquito species 80 thrives in urban environments and is highly anthropophilic, feeding primarily on humans [4].  concern. For example, mosquitoes harbouring the obligate intracellular bacteria Wolbachia 93 spp. are less able to transmit DENV and other arboviruses [10][11][12][13] and are being released in 94 endemic settings to test their impact on human disease burdens [14]. Similarly, insect-specific 95 viruses have also been proposed as potential biocontrol agents to reduce arbovirus 96 transmission [2,15].

140
The Aag2 cell line was originally generated in the 1960's by Peleg from whole 141 homogenised embryos, and has been referred to as 'Aag2' since the 1990's when Lan and 142 Fallon adapted the culture for growth in E-5 medium [52]. Cells within the culture exhibit 143 differing morphologies (Fig 1A), and it has been suggested that the varying morphologies of 144 mosquito cells in culture may be indicative of the presence of a diversity of embryonic and 145 differentiated cell types [53,54]

272
To analyse the genomic integration of PCLV, total DNA or total RNA was extracted 273 from 1-3 x 10 6 cells using the Quick-DNA or Quick-RNA Miniprep Kits (Zymo Research, Irvine,

274
CA USA) respectively, as per manufacturer's instructions. RNA samples were spiked with 275 NDV prior to isolation. Nucleic acids were treated with DNase for 40 min at 37˚C using the 276 DNA-free DNA Removal Kit (ThermoFisher Scientific) or with RNase A (Sigma-Aldrich) for 1 277 h at 37˚C as per manufacturers' instructions. Nucleases were removed by re-purifying the 278 nucleic acids as described above. cDNA was generated from RNA using the iScript cDNA   (Fig 1Ciii). These 398 clustered and floating morphologies are also observed in parental Aag2 cells (Fig 1A). Across 399 all single-cell clones generated, these three morphologies ('parental Aag2-like', 'clustered',

400
'rounded') were represented at similar levels, with slightly more 'clustered' cell lines observed 401 ( Fig 1D). However, all of the clonal cell lines reverted back to the parental Aag2 morphology 402 over time (Fig 1E), with this parental Aag2 morphology being stably maintained over many

422
We observed no gross differences in the growth kinetics of any of the

442
To rule out the possibility that fragments derived from persistent PCLV infection are 443 integrated into the specific version of the parental Aag2 cell line growing in our lab, we 444 designed primers to amplify short (50-150-nt) fragments covering each genome segment in 445 1,000-nt intervals. As a positive control we used RNA purified from parental Aag2 cells that 446 had been subjected to a reverse transcription reaction (Fig 2A)

451
DNA forms within the cell. For each PCLV-specific primer pair, and the NDV and Rps7 452 controls, RNase treatment eliminated the PCR signal, while DNase treatment did not (Fig 2A).

453
This confirms the purity of the RNA samples. No PCLV signal was detected when PCR was 454 performed on RNA that had not been subjected to reverse transcription, or on RNA isolated 455 from the mammalian Madin-Darby canine kidney (MDCK) cell line, which should not contain 456 genomic integrations of insect-specific virus sequences (Fig 2A). We then repeated the 457 experiment using genomic DNA isolated from parental Aag2 cells, and detected no evidence 19 of 49 of DNA sequences derived from PCLV (Fig 2A). Importantly, Rps7 was amplified when 459 genomic DNA was treated with RNase, but not DNase, confirming the purity of the DNA 460 samples (Fig 2A). Although we cannot exclude the possibility that the primers we designed 461 missed smaller fragments of integrated PCLV sequence, our data suggest that PCLV

487
RNA serving as a template control (Fig 2B). In contrast, clones Aag2-AF10 and Aag2-AF12 488 did not contain detectable levels of the PCLV S segment in this assay, while PCLV RNA was 489 clearly detectable to varying degrees in parental Aag2 cells and in the other clones (Fig 2B).

490
To verify this result, we performed strand-specific RT-PCR to amplify genome and antigenome 491 sequences from each of the three PCLV genome segments, with Rps7 serving as a template 492 control. In this experiment we did detect PCLV M segment RNA in the Aag2-AF10 clone at 493 lower levels than the parental Aag2 cell line (Fig 2C). Although any amplification of the L and 494 S segments were below the limit of detection, the presence of both genome and antigenome 495 sequences for the M segment indicates that the virus must be replicating its RNA and therefore 496 the L (RdRp) and S (nucleocapsid) segments, which are both required for genome replication, 497 must also be present. No PCLV RNA was detected in the Aag2-AF12 clone in this assay.

498
Finally, we measured PCLV L segment RNA by RT-qPCR and detected low levels of

499
PCLV RNA in the Aag2-AF10 clone, with PCLV RNA consistently maintained at lower levels 500 relative to the parental Aag2 cell line over multiple cell passages (Fig 2D). We did not detect 501 PCLV RNA at early passages in the Aag2-AF12 clone, though very low levels of PCLV were 502 detected at later passages ( Fig 2D). Overall, our data indicate that clones Aag2-AF10 and

503
Aag2-AF12 harbour markedly reduced levels of PCLV infection that are maintained at 504 consistently low levels over multiple cell passages.

514
confirming that RNAi is active in the parental Aag2 cell line (Fig 3). In contrast, the luciferase 515 dsRNA did not significantly reduce reporter activity in RNAi-defective C6/36 cells (Fig 3). The

562
We next tested the flavivirus ZIKV, which replicated with a peak in titres three days 563 post-infection in the parental Aag2 cell line at MOI 2 ( Fig 4C). While ZIKV replicated with faster 564 kinetics (1, 2, 3 days post-infection) and to a more than one-log higher titre at its peak in clone 565 Aag2-AF12, the growth kinetics and peak titres were similar to the parental Aag2 cell line in 566 clone Aag2-AF10 (Fig 4D). Therefore, although clone Aag2-AF12 appears to be more 567 permissive to ZIKV replication, this is not linked to PCLV levels, which are also reduced in 568 clone Aag2-AF10. Again, ZIKV replication kinetics varied somewhat across the other clones, 569 but fell close to within half a log from the parental Aag2 cell line at their peak.

570
Replication kinetics of both DENV-2 and ZIKV in clone Aag2-AF5 were comparable to 571 the parental Aag2 cell line (Fig 4B and 4D).  As a contrast to the +ssRNA arboviruses tested, we next tested the -ssRNA 585 rhabdovirus VSV. In a one-step growth curve (MOI 2), VSV replication peaked 12 hpi in 586 parental Aag2 cells ( Fig 4G). As for the other viruses tested, peak titres of VSV in clones Aag2-

591
Overall, we therefore conclude that pre-existing persistent infection with PCLV does 592 not notably alter the replication of a diverse range of +ssRNA and -ssRNA arboviruses, since 593 replication kinetics and peak titres of DENV-2, ZIKV, SINV and VSV were not markedly 594 different from the parental Aag2 cell line in clones Aag2-AF10 and Aag2-AF12, which harbour 595 drastically reduced levels of persistent PCLV infection.

597
Next, we further characterised clone Aag2-AF5 with the goal of providing a better-598 defined Aag2-derived cell line for the research community. Clone Aag2-AF5 was selected 599 because, of all the isolated clones, arboviral infectivity in Aag2-AF5 cells was most similar to 600 the parental Aag2 cells (Fig 4). Furthermore, this clone formed a more uniform monolayer and 601 was more resilient and easier to handle in culture than parental Aag2 cells.

602
First, we tested the transfection efficiency of this clone by transient transfection with a 603 constitutive GFP expression plasmid (Fig 5A). A similar proportion of Aag2-AF5 cells (56%)

604
were detectably GFP-positive compared to the parental Aag2 cell line (47%) (Fig 5B); 605 differences were not statistically significant. However, the GFP signal was brighter in Aag2-

606
AF5 cells (Fig 5A). This higher level of transgene expression in Aag2-AF5 cells was confirmed 607 by transient transfection with a constitutively active firefly luciferase reporter plasmid (Fig 5C). Aag2-AF5 with all stimuli (Fig 6). There was however some variability in the relative levels of 631 antimicrobial peptide induction in clone Aag2-AF5 compared to the parental cell line for 632 different gene/stimulus combinations. Thus, all tested antimicrobial peptide genes were less 633 inducible in clone Aag2-AF5 with E. coli stimulation (Fig 6A), and CecD was also less inducible 634 in clone Aag2-AF5 for all stimuli tested (Fig 6Aiii, 6Biii and 6Ciii), though some of these 26 of 49 differences were not significant. In contrast, DefD and CecB were more inducible in clone 636 Aag2-AF5 during stimulation with Gram-positive bacteria compared to the parental Aag2 cell 637 line (Fig 6Bi, 6Bii, 6Ci and 6Cii), with some of these differences again being non-significant.

638
Therefore, while there may be subtle differences in the immune sensitivity of clone Aag2-AF5 639 in terms of antimicrobial peptide production compared to parental Aag2 cells, there is no gross

679
We believe that our Aag2-derived clone Aag2-AF5 represents a useful standardised Ae.

743
We therefore provide new insights that may have important implications for the use of 744 insect-specific viruses as biocontrol agents to reduce the transmission of arboviruses.

745
Furthermore, clone Aag2-AF5 represents a valuable new clonal and better-defined cell line to 746 provide a more standardised system for studying arbovirus-vector interactions in cell culture.