248 words 17 Importance: 135 words 18 Main text: 4,747 words 19 Importance 39The evolution of influenza virus, in terms of single nucleotide variants and the reassortment of 40 gene segments, has been studied in detail. However, influenza is known to generate defective 41 viral genomes (DVGs) during replication, and little is known about how these genomes evolve 42 both within hosts and at the population level. Studies in animal models have indicated that 43 prophylactically or therapeutically administered DVGs can impact patterns of disease 44 progression. However, the formation of naturally-occurring DVGs, their evolutionary dynamics, 45and their contribution to disease severity in human hosts is not well understood. Here, we 46 identify the formation of de novo DVGs in samples from human challenge studies throughout the 47 course of infection. We analyze their evolutionary trajectories, revealing the important role of 48 genetic drift in shaping DVG populations during acute infections with well-adapted viral strains. 49 of the samples analyzed and were most common in the PB2, PB1, and PA gene segments. A 73 limitation of this study, however, is that it offered only a cross-sectional view of IAV DVG 74 populations. While it has been shown that Sendai virus DVG populations expand during the first 75 12 hours of infection in a mouse model (25), the evolution of IAV DVG populations within a 76 human host over the course of an infection has not been well characterized. 77Here, we report an analysis of IAV DVG populations identified from deep sequencing 78 data taken over the course of infection during two longitudinal human challenge studies with 79 different treatment cohorts. We observe the generation of de novo DVGs in nearly all subjects, 80 primarily in the polymerase gene segments (PB2, PB1, and PA). DVG populations were highly 81 variable over time in DVG species composition as well as in DVG species relative abundance. 82Over the course of infection, individual DVG species were observed to arise, fluctuate in 83 abundance, as well as disappear from the DVG population. Overall, we found no trend towards 84 decreasing diversity of DVG populations or towards shorter DVG species during the five days 85 post challenge, likely due to the dominance of stochastic effects. Furthermore, we were unable to 86 detect an association between DVG levels and peak viral titers, potentially due to the negative 87 feedback between DVG and wild-type virus. Similarly, higher DVG levels were not associated 88 with more severe symptoms. This study helps to illustrate the stochastic dynamics of DVG 89 populations within a host during acute infection with a well-adapted viral strain, a scenario under 90 which fitness variation in the wild-type virus population is expected to be relatively small. 91
Materials and Methods 92Ethics statement. The procedures followed in the human challenge studies were in accordance 93 with the Declaration of Helsinki. The studies were approved by the institutional review boards 94 (IRBs) of Duke Univer...