15The mosquito microbiota impacts the physiology of its host and is essential for normal larval 16 development, thereby influencing transmission of vector-borne pathogens. Germ-free 17 mosquitoes generated with current methods show larval stunting and developmental deficits. 18Therefore, functional studies of the mosquito microbiota have so far mostly been limited to 19 antibiotic treatments of emerging adults. In this study, we developed a novel approach to 20 produce germ-free Aedes aegypti mosquitoes. It is based on reversible colonisation with 21 bacteria genetically modified to allow complete decolonisation at the end of larval 22Overall, this study provides a new tool to study the microbiota in insects and deepens our 29 knowledge on the metabolic contribution of bacteria to mosquito development. 30We developed a transient bacterial colonisation method to support larval development while 56 producing fully developed germ-free adults. Using this approach, we were able to support 57 development within a standard timeline and without affecting adult size, showing that these 58 germ-free individuals represent a valid tool for mosquito microbiota studies. We further used 59 transient colonisation during early larval development to characterise the role of bacteria in 60 larval development using a transcriptomic approach and diet complementation. Our data 61 indicate that bacteria contribute to folate biosynthesis and energy storage and that folic acid 62 supports larval development in the absence of a microbiota. 63 64 Results 65 66
Production of germ-free adult mosquitoes via transient bacterial colonisation 67After microbiological sterilisation of mosquito eggs, hatchlings provided with autoclaved 68 standard rearing diet are able to survive for weeks but remain first-instar larvae. Their 69 development can be rescued by bacterial colonisation, for instance with Escherichia coli 4,10 . 70 Therefore, we hypothesised that transient bacterial colonisation would be an efficient way to 71 support larval development and produce germ-free adults. To test this, we used HA416, an E. 72 coli HS-derived strain that is auxotrophic for two non-standard amino-acids required for 73 peptidoglycan synthesis (meso-diaminopimelic acid, m-DAP, and D-alanine, D-Ala) and that 74 was previously used to transiently colonise the gut of germ-free mice 12 . We supplemented the 75 food of gnotobiotic larvae with m-DAP and D-Ala to support the proliferation of auxotrophic 76 E. coli during larval development only ( Figure 1A). These supplements were toxic to first instar 77 larvae when provided at the concentrations previously used in vitro, but 4-fold lower 78 concentrations supported growth of auxotrophic E. coli (Supplementary Figure S1) without 79 causing toxicity in larvae. The auxotrophic E. coli strain rescued the development of axenic 80