Our strategy for investigating the co-transmission of genetically diverse baculoviruses through OBs and ODVs was based on analyzing infection foci produced following viral entry through the oral route. F1 midgut analysis showed that the fraction of mixed infection foci expressing both GFP and Cherry reporters was an order of magnitude lower than expected by sheer chance (5.3% versus 65.7%), a result that did not vary appreciably depending on whether larva were inoculated with OBs carrying complete or polyhedrin-defective genomes. This coinfection rate was low compared to the results obtained previously using midgut cells of another species 24. Our experiments with complete (polyhedrin positive) viruses did not provide information about whether the two variants were packaged in the same OBs in the producer cells (co-occlusion), since OBs are dissolved in the midgut lumen and hence should not necessarily deliver their content to the same midgut cells. However, the assays performed with a polyhedrin-defective GFP-expressing virus and a functional Cherry virus confirmed that OBs contained a mixture of both variants, since the polyhedrin-defective virus alone would be unable of undergo oral transmission. Previous work has amply shown that OBs can be shared by different virus variants 13,15–17,19−22. Mixing occurred both in producer Sf21 cells and in vivo, since the polyhedrin-defective virus was also transmitted to the F2 generation, albeit at lower frequency than in F1, suggesting more restricted OB sharing in vivo. Still, transmission of this polyhedrin-defective variant was lower than in previous studies with Trichoplusia ni larvae17,25. The persistence of the defective virus is influenced by the composition of the OBs, the proportion of each variant in the population, and the infection dose, since transmission bottlenecks could favor extinction of the defective virus25.
Segregation of the GFP and Cherry variants into different midgut infection foci was observed in all experiments and suggests that the two variants were typically not packaged into the same ODVs, neither in cultured Sf21 cells nor in vivo. Previous reports have shown some extent of ODV co-packaging of virus variants in cell cultures 16, but our data suggest that this process occurs at low frequency. Based on this, we speculate that virus variants coinfecting the same cell often remain compartmentalized at the intracellular level. Baculovirus replication and encapsidation takes place inside nuclei in discrete structures called the virogenic stroma 8,26. Imported viral DNA molecules might initiate a different replication and encapsidation center each, limiting admixture of different genetic variants in the same ODVs. Alternatively, it is possible that co-packaging in the same ODV occurred but that productive infection of midgut cells was typically initiated by only one nucleocapsid from each ODV. In principle, such sieving could obey different processes. First, it is possible that the majority of nucleocapsids in an ODV were structurally defective and failed to undergo decapsidation and gene expression. Second, some ODV-derived nucleocapsids may traffic directly to tracheal cells or the hemolymph by budding through basal plasma membrane of midgut cells without replicating, as suggested previously 8,11,12. In either case, this would result in variant segregation during early infection, preventing coinfection-dependent virus-virus interactions such as genetic complementation and, hence, removing a large fraction of defective viruses at this stage. In line with our findings, defective viruses that have been produced by high-MOI passaging in cell cultures can be rapidly counter-selected in vivo 27, probably as a result of spatial segregation during primary infection.
Our analysis of late-stage infected larvae showed that the GFP and Cherry variants mixed unfrequently despite widespread disseminated infection. This may explain why F2 midgut foci contained even fewer coinfected cells than F1 foci, but contrasts with pervious work suggesting that coinfection of individual cells by BVs of different virus variants is an important mechanism for the maintenance of genetic diversity in baculoviruses 17,28. As a note of caution on our results, terminally infected larvae contained large numbers of dead cells. We indeed noticed high levels of green autofluorescence in these preparations, and this might have biased GFP-positive cell counts. Nevertheless, the numbers of doubly fluorescent cells were exceedingly lower than those of singly fluorescent cells, suggesting that coinfection was truly rare. This finding was apparently puzzling, considering that the MOI should be relatively high at this infection stage. However, it may be accounted for by superinfection exclusion, a mechanism displayed by most types of viruses whereby a resident virus impedes secondary infections of the same cell. In baculoviruses, superinfection exclusion is established as soon as 3 hpi, and constitutes an efficient mechanism for preventing mixed infections 29.
Cellular coinfection necessarily occurs in vivo, since it is a condition for recombination and for the maintenance of defective viral genomes, both of which are well-known processes in baculoviruses that were also observed in our experiments. In light of our results, we speculate that mixed infections are allowed, but only episodically and, maybe, specifically in certain cell types or at certain infection stages, such as for instance in some OB-producing cells. Relaxation of superinfection exclusion mechanisms might help promote these changes. In contrast, mixed infection avoidance might be less efficient in cell cultures, resulting in faster production of recombinants and defective genomes compared to natural infections 30.
Overall, our findings suggest that the natural baculovirus infection cycle imposes certain barriers to interactions between different genetic variants, including genetic complementation and defective interference. In the social evolution field, it has amply established that systematic mixing of genetically unrelated individuals makes cooperation unlikely, since under this scenario, selection favors cheater genotypes (including defective interfering viruses), which benefit from cooperators without reciprocating 31–33. Spatial structure, such as the foci segregation we found in the midgut, increases genetic relatedness among potentially interacting viruses, making cooperation evolutionarily more stable. Different forms of spatial segregation have been reported in widely different viruses undergoing collective transmission, such as in enteroviruses transmitted through vesicles 34, HIV-1 spreading through viral synapses 35, and tomato mosaic virus spreading through plasmodesmata 36. This suggests that collective infectious units do not favor cooperation between different genetic variants of a virus (e.g. genetic complementation). In contrast, the fitness advantage of spreading collectively could reside in other processes, such increasing environmental stability, accelerating early infection stages, or evading cellular innate immune responses 2,37,38.