Incomplete bunyavirus particles contribute to within-host spread and between-host transmission

Bunyaviruses lack a specific mechanism to ensure the incorporation of a complete set of genome segments into each virion, explaining the generation of incomplete virus particles lacking one or more genome segments. Such incomplete virus particles, which may represent the majority of particles produced, are generally considered to interfere with virus infection and spread. Using the three-segmented Rift Valley fever virus as a model bunyavirus, we here show that two distinct incomplete virus particle populations that are unable to spread autonomously, are able to efficiently complement each other in both mammalian and insect cells following co-infection. We further show that incomplete virus particles are capable of co-infecting mosquitoes, resulting in the rescue of infectious virus that is able to disseminate to the mosquito salivary glands. Our findings reveal a significant role of incomplete particles in within-host spread and between-host transmission, reminiscent of the life cycle of multipartite viruses.


Introduction 30
Segmented and multipartite viruses have genomes divided over multiple segments. The classical 31 paradigm in virology states that segmented viruses package all their genome segments into a 32 single virus particle, whereas multipartite viruses package each genome segment into a distinct 33 virus particle 1 . To ensure a productive infection, it is generally accepted that multipartite viruses 34 (mainly found to infect plants and fungi) rely on co-infection of the same cell with a set of 35 complementing particles, each particle containing a different genome segment 1,2 . Alternatively, 36 complementation can occur at the tissue level, as proposed in a recent study with the 37 plant-infecting faba bean necrotic stunt virus (FBNSV, family Nanoviridae). FBNSV was shown to 38 complement its missing genome segments by export and distribution of viral mRNAs and proteins 39 across interconnected neighboring cells 3 . By contrast, it has been thought that segmented viruses 40 (mainly found to infect animals) solely rely on individual cells as units of viral replication, and thus 41 have to carry at least one copy of each genome segment within a single virus particle to ensure 42 the delivery of a complete genome 4,5 .  (Fig. 1c). Upon simultaneous inoculation with RVFV-eGFP and RVFV-mCherry2, both BSR-T7/5 102 (hamster) and C6/36 (mosquito) cells were found to be susceptible to co-infection, as clearly 103 evidenced by the co-localization of green and red fluorescent signal in a fraction of the cell 104 population (Fig. 1d-e).

106
Simultaneous infections with RVFV-eGFP and RVFV-mCherry2 allowed us to qualitatively confirm 107 cell susceptibility to co-infection (Fig. 1e). However, the fast spread of three-segmented RVFV 108 over multiple infection cycles impeded us to assess accurately how often these co-infection events 109 are actually taking place. To overcome this, we again used the T7 polymerase-based reverse 110 genetics system to generate two non-spreading incomplete RVFV particle populations lacking the or Gn, confirming that these particles are incomplete and not able to spread due to the lack of the 127 M genome segment (Fig. 2c). To quantify to what extent cells can be co-infected with the two different non-spreading fluorescent 130 virus variants, we simultaneously infected BSR-T7/5 cells with iRVFV-SL-eGFP and 131 iRVFV-SL-mCherry2 at increasing multiplicities of infection (MOIs, ranging from 0.1 to 2.5) 132 (Fig. 2d). Through direct detection of co-localized eGFP and mCherry2 expression, we confirmed 133 that these particles also could co-infect BSR-T7/5 cells (Fig. 2e). Next, using flow cytometry, we 134 5 quantified the fraction of non-infected cells, singly-infected cells and co-infected cells (Fig. 2f). 135 Mock-infected cells and cells infected with only one population of incomplete particles were the 136 basis to gate the flow cytometry data. A clear double (eGFP and mCherry2)-positive cell 137 population was identified after co-infection with both populations of incomplete particles. 138 Interestingly, at each MOI tested, the percentage of infected and co-infected cells closely 139 resembled that of a predictive mathematical model based on the assumptions that genome 140 segments are randomly packaged into virus particles and that host susceptibility is heterogeneous 141 (Fig. 2g). Moreover, the population of co-infected cells rose sharply with increasing MOI in a 142 dose-response fashion, suggesting that there is no apparent mechanism leading to the exclusion 143 of multiple particles entering the same cell in the present experimental setup. To study the potential role of incomplete particles in the bunyavirus life cycle, we needed at least 147 two distinct populations of RVFV particles having incomplete but complementing genomes.

148
Besides generating iRVFV-SL-FP particles, we also generated incomplete RVFV particles lacking 149 the S segment (iRVFV-ML), by following a similar T7 polymerase-based reverse genetics strategy 150 (Fig. 3a). Complete M and L genome segments were encoded in antigenomic-sense orientation 151 by transcription plasmids, and the N protein was provided by an expression plasmid. Due to the 152 absence of the S segment in the rescued particles, and consequently, the unavailability of 153 N protein, infections with iRVFV-ML particles did not result in any appreciable viral genome 154 replication. Accordingly, neither N nor Gn was detected in an immunofluorescence assay with 155 cells infected with iRVFV-ML (Fig. 3b). Due to the non-replicating nature of RVFV particles lacking 156 the S genome segment, the infectious titer of iRVFV-ML stocks could not be determined with a 157 conventional virus titration assay. Therefore, we confirmed the genomic composition of rescued 158 iRVFV-ML particles through quantification of the S, M and L genome segments via RT-qPCR 159 (Fig. 3c). In both batches of rescued iRVFV-ML particles, high copy numbers of only the M and L 160 genome segments were detected, whereas the S segment was not detected. Similarly, in samples 161 containing iRVFV-SL-eGFP particles, only the S and L genome segments were present at high 162 copy numbers and the M segment was not detected. All three genome segments (S, M and L) 163 were detected in samples containing the three-segmented RVFV-eGFP used as control.   (Fig. 5a). As expected, in cells exposed exclusively to 197 iRVFV-SL-eGFP, we detected abundant copies of the S and L genome segments, while the 198 M segment and the Gn glycoprotein were absent. No expression of eGFP or Gn was detected in 199 cells exclusively exposed to iRVFV-ML particles, as the S genome segment is missing in this 200 population, and thus no genome replication or transcription could take place. Contrary to cells 201 exposed to only one population of incomplete particles, cells simultaneously exposed to both  Finally, we used our smFISH-immunofluorescence method to obtain additional insights into the 206 intra-virion composition of the incomplete particle populations and the progeny virions released 207 upon co-infection with these incomplete particles (Fig. 5a). Importantly, the M genome segment 208 was not present in iRVFV-SL-eGFP particles, and the S genome segment was not present in 209 iRVFV-ML particles, confirming the absence of the respective segment in each of the incomplete  Having confirmed the ability of incomplete RVFV particles to generate a productive infection 218 in vitro upon co-infection, we hypothesized that incomplete bunyavirus particles could also play a 219 role in between-host virus transmission. To investigate this, groups of Aedes aegypti mosquitoes 220 were fed with bovine blood meals spiked with diverse virus populations. Mosquitoes were exposed 221 to a mock blood meal (group #1), a blood meal spiked with a single incomplete virus particle 222 population (group #2 to iRVFV-SL-eGFP and group #3 to iRVFV-ML), a blood meal spiked with   Using the three-segmented RVFV as a prototype bunyavirus with a non-selective genome 305 packaging strategy, we show here that incomplete bunyavirus particles can contribute significantly 306 to within-host spread and between-host transmission.

308
By creating fluorescently labeled RVFV variants, we were able to visualize infection of single cells 309 by more than one virus particle and to show that mammalian and insect cells are prone to 310 co-infection as a function of the MOI. Notably, the fractions of experimentally infected and 311 co-infected cells coincided with theoretical models that predict the absence of a major bottleneck 312 determining the probability of (co-)infection other than intrinsic heterogeneous susceptibility within 313 a host cell population. In our experiments, we exposed cells to two different RVFV variants 314 simultaneously, as this represents a common scenario during a localized infection, in which 315 non-infected cells are exposed to the burst of virus particles released by a neighboring infected 316 cell. Nevertheless, it should be noted that susceptibility to co-infection may differ when cells are      Altogether, the results of this study show that upon co-infection, incomplete bunyavirus particles 378 can initially drive efficient progeny virus generation and spread, even in the absence of 379 three-segmented RVFV infectious particles in the inoculum. In the context of natural infection with 380 a mixed bunyavirus particle population, we propose that incomplete particles, instead of 381 interfering, can substantially contribute to within-host spread and between-host transmission, 382 facilitating the dual life cycle of bunyaviruses in mammalian and insect hosts (Fig. 8a-b). Moreover, 383 we propose that this contribution, particularly important at intermediate MOIs (Fig. 8c)        for each channel were set empirically by individual examination of images. The threshold to define 588 co-localized spots was set to a maximum distance of 3-4 pixels between the centers of the spots.

589
For visualization purposes, image brightness and contrast were manually adjusted in ImageJ.  (Table 1). Virus-blood mixtures were back-titrated as 599 described above. After feeding, mosquitoes were anesthetized using a semi-permeable CO2 pad. tested by virus isolation. Briefly, 12-15 days after the blood meal, mosquitoes were anesthetized 607 using a semi-permeable CO2 pad, wings and legs were removed with forceps, and mosquitoes 608 were forced to salivate by inserting their proboscis inside a 20 µL filter tip pre-filled with 7 µL of a  File 1) and hence was selected for visualization.

648
Modeling virus spread and the relationship between MOI and co-infection 649 We generated a simple simulation model of viral spread to consider the impact of non-selective