An obligate microsporidian parasite modulates defense against opportunistic bacterial infection in the yellow fever mosquito, Aedes aegypti

ABSTRACT The ability of Aedes aegypti mosquitoes to transmit vertebrate pathogens depends on multiple factors, including the mosquitoes’ life history traits, immune response, and microbiota (i.e., the microbes associated with the mosquito throughout its life). The microsporidium Edhazardia aedis is an obligate intracellular parasite that specifically infects Ae. aegypti mosquitoes and severely affects mosquito survival and other life history traits critical for pathogen transmission. In this work, we investigated how E. aedis impacts bacterial infection with Serratia marcescens in Ae. aegypti mosquitoes. We measured development, survival, and bacterial load in both larval and adult stages of mosquitoes. In larvae, E. aedis exposure was either horizontal or vertical and S. marcescens was introduced orally. Regardless of the route of transmission, E. aedis exposure resulted in significantly higher S. marcescens loads in larvae. E. aedis exposure also significantly reduced larval survival but subsequent exposure to S. marcescens had no effect. In adult females, E. aedis exposure was only horizontal and S. marcescens was introduced orally or via intrathoracic injection. In both cases, E. aedis infection significantly increased S. marcescens bacterial loads in adult female mosquitoes. In addition, females infected with E. aedis and subsequently injected with S. marcescens suffered 100% mortality which corresponded with a rapid increase in bacterial load. These findings suggest that exposure to E. aedis can influence the establishment and/or replication of other microbes in the mosquito. This has implications for understanding the ecology of mosquito immune defense and potentially disease transmission by mosquito vector species. IMPORTANCE The microsporidium Edhazardia aedis is a parasite of the yellow fever mosquito, Aedes aegypti. This mosquito transmits multiple viruses to humans in the United States and around the world, including dengue, yellow fever, and Zika viruses. Hundreds of millions of people worldwide will become infected with one of these viruses each year. E. aedis infection significantly reduces the lifespan of Ae. aegypti and is therefore a promising novel biocontrol agent. Here, we show that when the mosquito is infected with this parasite, it is also significantly more susceptible to infection by an opportunistic bacterial pathogen, Serratia marcescens. This novel discovery suggests the mosquito’s ability to control infection by other microbes is impacted by the presence of the parasite.

In addition to blood-borne pathogens that can impact human health, there are myriad other environmental microbes that live in association with Ae. aegypti.These include bacteria and fungi that mosquitoes routinely encounter, for example, in larval development water, when nectar feeding on flowers as adults, or as a result of dam age to their cuticle (6)(7)(8).Mosquitoes ingest microbes from their habitats; some of these microbes are digested as food (9) while some colonize the digestive tract and persist (10,11).Exposure to environmental microbes impacts mosquito physiology and transmission of mosquito-borne pathogens.For example, live bacteria are critical for larval development, and mosquitoes reared with extensive dietary supplementation in place of bacteria are smaller and have reduced lifespans compared to conventionally reared mosquitoes (10,12,13).In addition, the presence of different bacterial species in mosquito digestive tracts (larval and adult) induces significant changes in lifespan and susceptibility to human pathogens like dengue virus and the malaria parasite (14)(15)(16)(17)(18)(19).In some instances, microbes are pathogenic to the mosquito and can cause immune system activation, disease symptoms, and death (18,(20)(21)(22).One such group of pathogens is the microsporidia.
The microsporidia are obligate intracellular parasites and pathogens most closely related to fungi.They are extremely diverse in their infectivity, host specificity, path ogenicity, and transmission cycles, and infect a wide range of metazoan organisms, from protists to vertebrates (23,24).There are multiple species of microsporidia that infect mosquitoes and establish pathogenic infections (25)(26)(27), and the prevalence of infection in natural populations can be high; one recent report detected microsporidia in an average of 60% of sampled individuals across seven different mosquito species (28).Notably, microsporidia infection has been shown to impact the microbiota of mosquitoes, and one microsporidian parasite of Anopheles interferes with infection by the Plasmodium parasite, the etiological agent of malaria (29,30).Among the mosquitoassociated microsporidia is Edhazardia aedis (E.aedis), which is an obligate intracellular pathogen of Ae. aegypti (31,32).E. aedis is a polymorphous microsporidium that has a complex life cycle (Fig. 1) involving two types of spores which can be transmitted horizontally from infected to healthy larvae or vertically through infected eggs laid by E. aedis-infected adult females (31,33).The infection cycle of E. aedis in Ae. aegypti (31) begins horizontally with the ingestion of uninucleate spores in the larval stage, where spores germinate in the gut and begin to infect larval cells (31).After larvae are orally infected by E. aedis uninucleate spores, binucleate spores develop and disseminate from the larval gut to other parts of the body.In adult females, E. aedis binucleate spores infect the reproductive tissues and oenocytes, resulting in the production of vertically infected eggs which subsequently develop into infected larvae.Vertically infected larvae then produce uninucleate spores in large numbers in the fat body, which are released upon larval death and ingested by healthy larvae, leading to horizontal transmission and completion of the transmission cycle (Fig. 1) (31).E. aedis infection has dramatic effects on mosquito physiology including reduced lifespan, lengthened development time, and increased expression of multiple immune system genes (34,35).
We hypothesized that E. aedis infection could potentially impact the colonization and/or infection dynamics of other microbes in Ae. aegypti.This is of interest because (i) the effects of E. aedis on other microbes could have downstream impacts on mosquito physiology or disease transmission and (ii) interactions between E. aedis and other naturally occurring pathogens may influence the virulence of one or both in Ae. aegypti.We chose to investigate this question using Serratia marcescens, a common member of the gut microbiota and opportunistic pathogen of many insects, including mosquitoes (22,36).S. marcescens is of particular interest in mosquitoes as it has been shown to impact mosquito susceptibility to multiple human pathogens (37,38), can improve the efficacy of entomopathogenic biocontrol agents (22), is highly effective at colonizing mosquito tissues (39), and has been identified as a promising candidate for vector control via paratransgenesis (39,40).
(C) Overall survival was significantly affected by E. aedis (P = 1.00 × 10 −11 ) but not Serratia (P = 0.9) infection.We were unable to test for an interaction, but the survival curves suggest no interaction given the highly similar effects of Serratia regardless of E. aedis infection status.(D) Serratia load was significantly higher in E. aedis(+)/H individuals compared to E. aedis(−) individuals, but this varied over time (E.aedis × hour interaction, P = 0.015).Serratia load was higher at 48 hours (P = 0.003) and 72 hours (P = 7.06 × 10 −4 ) but not at 6, 24, or 96 hours post-initial Serratia exposure.Each box plot represents n = 10 mosquitoes, and asterisks denote significant differences from the E. aedis(−) treatment at the corresponding time point.

DISCUSSION
Several studies have investigated the impact of the obligate parasite E. aedis on Ae. aegypti mosquitoes and have documented negative impacts on the survival and development of Ae. aegypti larvae and adult females (32,34,(41)(42)(43).Despite these advances, the question of whether E. aedis infection augments Ae. aegypti's susceptibility to other microbial threats remains unresolved.
In this study, we present the novel finding that E. aedis impacts the susceptibility of Ae. aegypti mosquitoes to gut colonization and systemic infection by the opportun istic bacterial pathogen Serratia marcescens.We hypothesized that the susceptibility of Ae. aegypti to E. aedis and Serratia individually, or the interaction between these pathogens, may differ depending on the mosquito's developmental stage.Therefore, we investigated the interplay between these pathogens in both larval and adult stages.In the adult stage, we chose to only test adult females because only females transmit human pathogens such as dengue and yellow fever viruses.We also hypothesized that the transmission route of E. aedis could impact mosquito immune defense; therefore, we included both horizontal transmission (via ingestion by larvae) and vertical transmission (inherited by offspring) in our study.The most notable effect of E. aedis infection was on S. marcescens bacterial loads.In all experiments, E. aedis(+) individuals had significantly higher numbers of S. marcescens.These higher bacterial loads generally did not impact pupation, eclosion, or survival except in the case of S. marcescens systemic infection in adults.In this case, higher Serratia loads in E. aedis(+) individuals corresponded with 100% mortality within 24 hours.
Infection by E. aedis alone significantly slowed development time and increased mortality among infected larval and adult Ae. aegypti, and for both vertically and horizontally infected individuals.These phenomena have been previously reported in multiple studies (32)(33)(34)41), and our observation of them here confirms that we successfully induced pathogenic infection by E. aedis.
Infection by S. marcescens alone had little to no impact on pupation, eclosion, or survival for up to 150 hours post-Serratia infection when introduced orally in larvae or adults.However, it caused substantial mortality when injected systemically into adults.This is consistent with the previously described role of S. marcescens as a common inhabitant of the gut (36,44,45) and as an opportunistic pathogen (22).
In co-infection, E. aedis(+) individuals had higher S. marcescens bacterial loads; this was irrespective of E. aedis infection route (horizontal or vertical), developmental stage (larvae or adult), or route of S. marcescens exposure (oral or intrathoracic injection).Notably, (22) discovered that S. marcescens proliferates in the mosquito digestive tract when the mosquito is infected by Beauveria bassiana (a fungal insect pathogen) and causes enhanced mortality by entering the insect body cavity during infection.This presents clear parallels to our study, as we also observed the proliferation of S. marces cens in individuals co-infected with E. aedis microsporidia.However, in experiments where Serratia infection was via oral ingestion, we did not observe a corresponding increase in mortality with bacterial proliferation.One hypothesis to explain this result is that the burgeoning bacterial population remains localized to the alimentary canal; however, this remains untested.( 22) also showed that infection by B. bassiana decreases transcript abundance of multiple anti-microbial peptide genes as well as Duox (which is involved in the production of reactive oxygen species) and concluded that this allows the proliferation of S. marcescens.A previous study found that infection by E. aedis did not alter Duox levels and increased transcript abundance of anti-microbial peptide genes (35), suggesting a different mechanism may explain proliferation of S. marcescens in our study than that proposed by Wei et al. (22).( 35) also showed that E. aedis infection altered expression of many other immune response genes, for example, it increased transcript abundance of many CLIP-domain serine proteases , which are involved in anti-microbial peptide (AMP) production and melanization, two key components of insect immunity (46)(47)(48)(49).It also upregulated multiple C-type lectins and Serine Protease Inhibitors, which are negative regulators of melanization and AMP activity (50,51).C-type lectins have also been shown to bind to the surface of S. marcescens and protect the bacteria from AMP activity (52), which warrants further investigation as a potential mechanism for the proliferation of S. marcescens we observe here.
While higher bacterial loads were broadly observed among E. aedis(+) individuals, there were some differences in the Serratia infection dynamics between vertically and horizontally infected larvae.Specifically, when E. aedis was vertically transmitted, there was an effect on Serratia loads immediately after exposure (at 6 hours post-Serratia infection).However, when E. aedis was horizontally transmitted, Serratia loads were the same in both E. aedis(+)/H and E. aedis(−) larvae at 6 hours post-Serratia infection, and an effect of E. aedis on Serratia loads was not observed until 48 hours post-Serratia infection.In addition, Serratia load was higher at all time points tested for vertically infected larvae.However, in horizontally infected larvae, a fluctuation in Serratia load was observed over time.These differences in the effect of E. aedis on Serratia load could be due to differences in the nature of vertical vs horizontal infection.For example, in vertical infection, larvae are infected from the time of hatching and therefore could experience more costs of infection than horizontally infected larvae which are not exposed to the parasite until the second to third instar stage.In addition, in vertical infection, spores primarily infect the larval fat body, while in horizontal infection, spores invade through the digestive tract and eventually infect the oenocytes (specialized cells that function in lipid metabolism and immune system signaling).Lastly, compared to horizontally infected, vertically infected larvae have fewer immune genes that are upregulated by infection and the genes that are impacted overlap very little with those in horizontally infected individuals (35).These differences in tissue localization and immune gene expression could differentially impact S. marcescens bacterial load, though substantial work is still needed to fully uncover the mechanism of our findings.
It is also important to note that E. aedis(+)/V larvae were approximately 1 day older at the time of Serratia infection than E. aedis(+)/H larvae.We chose this approach so that Serratia infection would coincide with stages of E. aedis infection we anticipated would be most relevant to eliciting an immune response.For horizontal infection, we intro duced Serratia during the time of initial gut invasion by E. aedis, and for vertical infection, we introduced Serratia during the time of maximum spore production prior to larval death.For this reason, E. aedis(+)/H larvae experienced one molt after initial Serratia exposure that E. aedis(+)/V larvae did not.This likely explains the low S. marcescens loads in E. aedis(+)/H larvae at 24 hours post-exposure because Ae. aegypti larvae expel >90% of bacteria during molting and metamorphosis (53,54), similar to what has been observed in other insects (55).Similarly, metamorphosis-associated expulsion of microbes likely explains the drop in S. marcescens loads observed for both E. aedis(+)/V and E. aedis(+)/H individuals assayed after pupation (96-hour time point).The difference in larval age at the time of S. marcescens exposure may also influence feeding patterns and alter the number of S. marcescens cells ingested by each larva over the 6-hour dosing period.Indeed, the median initial dose in E. aedis(+)/V larvae was 150 CFU while in E. aedis(+)/H larvae, it was approximately 30 CFU.However, the median initial dose in E. aedis(−) larvae was 30 CFU in both experiments, indicating that higher bacterial load in E. aedis(+)/V larvae at 6 hours is not merely due to larger larvae eating more S. marcescens but is rather a result of E. aedis infection.
To further explore the interaction between E. aedis and S. marcescens, we conducted an additional experiment with Ae. aegypti adult females by introducing Serratia via oral exposure or injection, which mimics natural routes of infection in healthy and injured mosquitoes, respectively.E aedis(+) adult females had higher Serratia loads compared to E. aedis(−) females regardless of whether the bacteria were introduced orally or via injection.Bacteria from the genus Serratia have been shown to impact adult female susceptibility to infection by dengue and chikungunya viruses (38,56,57) .Therefore, proliferation of the bacteria in E. aedis-infected individuals could have downstream implications for disease transmission.
As expected, survival was significantly reduced by E. aedis infection alone.Oral Serratia infection had no effect on mortality, and co-infection with E. aedis and Serratia (orally) did not increase mortality beyond that already induced by E. aedis infection.Intrathoracic injection of Serratia did induce high mortality, however, and coupled with E. aedis infection resulted in 100% mortality by 24 hours.Co-infection appeared to induce mostly additive effects on endpoint mortality, given that Serratia infection increased endpoint mortality by ~70% in both E. aedis(+) and E. aedis(−) individuals.However, early in infection (18 hours), it is notable that co-infected individuals suffer 90% mortality while mortality due to E. aedis and Serratia infection alone is approximately 20% and 25%, respectively.This suggests that co-infection may have a synergistic effect early in the infection time course, inducing rapid and severe mortality.The high mortality corresponds with bacterial loads of 10 6 CFU per mosquito, the highest we observed in any experiment, suggesting that dramatic proliferation of Serratia occurs in the hemocoel of E. aedis(+) females, causing death.It is unclear whether the mechanism underlying higher Serratia loads in E. aedis(+) adults is the same as that in larvae, however, and further study is warranted.
In conclusion, we have found that infection by the obligate microsporidian parasite E. aedis induces higher S. marcescens bacterial loads in larval and adult female Ae.aegypti mosquitoes.These findings suggest that E. aedis modulates the immune defense of Ae. aegypti.Given that S. marcescens is an opportunistic pathogen that can impact susceptibility to viral infection in adult mosquitoes, these findings have implications for mosquito fitness, vector competence, and the overall efficiency of human pathogen transmission.

Preparation of Serratia culture for Ae. aegypti infection and Serratia CFU quantification
Serratia was cultured in Luria broth supplemented with 50 µg/mL of kanamycin (LB + kan), incubated overnight (16-18 hours) at 30°C with shaking at 220 rpm.The overnight culture of Serratia was centrifuged at 5,000 × g for 3 minutes, then the pellets were washed twice with sterile 1× PBS.The pellets were resuspended in sterile 1× PBS and measured in a spectrophotometer to obtain 1.0 OD 600 which corresponded to ~10 9 CFU/mL.

Preparation of Serratia for oral infection
For larval oral infection with Serratia, a 10-fold dilution of 1.0 OD 600 S. marcecens-GFP was prepared in 1× PBS to obtain ~10 8 CFU/mL.Then 1 mL (~10 8 CFU) was added to each larval tray (50 larvae/tray).For oral infection of Ae. aegypti adult females, S. marcecens-GFP (1.0 OD 600 ) suspension was prepared in filter-sterilized 10% sucrose instead of 1× PBS.Then a 1.5 mL microcentrifuge tube was filled with the prepared suspension, and a filter paper wick was inserted.The microcentrifuge tube was introduced to each cage (50 females/cage) for 6 hours.After 6 hours, the infected sucrose meal was taken out and the cages were provided with filter-sterilized 10% sucrose ad libitum.

Serratia CFU quantification
CFUs of S. marcescens-GFP were measured after infection by homogenizing five living larvae, pupae, or adults (based on the experiment) individually in 150 µL sterile 1× PBS.Serial dilutions (10 −2 and 10 −4 ) and undiluted homogenates were cultured on LB + kan (50 µg/mL) and incubated overnight at room temperature.The resulting fluorescent colonies were counted using a NIGHTSEA fluorescent viewing system (Nuhsbaum, McHenry, IL, USA) Preparation of E. aedis spore suspension for horizontal and vertical transmis sion E. aedis spore suspension E. aedis can only be cultured in live Ae.aegypti mosquitoes.Spore suspension of E. aedis was prepared as described in reference (59).E. aedis-infected Ae. aegypti eggs were hatched in larval trays with 1 L deionized (DI) water and 1 piece of cat food.Twenty-four hours after hatching, larval density was reduced to ~100 larvae per tray with 1 L of DI water and 1 piece of cat food.All trays were incubated under standard insectary conditions.When E. aedis-infected larvae reached the late fourth instar stage (the stage at which infectious pyriform spores are most numerous), 10 infected larvae were moved to a 1.5 mL microcentrifuge tube, washed twice with DI water, and homogenized in 500 µL DI water using a pestle and mechanical homogenizer (VWR, USA) to release E. aedis spores into solution.The concentration of spores in the homogenate suspension was determined using a hemocytometer.

Horizontal transmission of E. aedis in Ae. aegypti larvae
The horizontal transmission of E. aedis spores in Ae. aegypti larvae was performed as described in reference 59.In all, 100 second to third instar heathy Ae. aegypti larvae were transferred to 100 mL cups.Approximately 5 × 10 4 E. aedis spores and 2 mL of food slurry (2 g yeast extract and 3 g liver powder in 25 mL DI water) were then introduced to each cup of 100 larvae.Twenty-four hours after the addition of E. aedis spores, larvae from each cup were moved to larval trays (100 infected larvae/tray) with 1 L of DI water and one piece of cat food and used for experiments as described in experiment 2 (see below).Horizontally infected larvae were reared to adults and used for experiments as described in experiment 3 (see below).Larvae and adults infected in this manner are referred to throughout as E. aedis(+)/H.

Vertical transmission of E. aedis spores to Ae. aegypti eggs
Eggs laid by mothers infected with E. aedis are naturally vertically infected (58).We have observed a 96% vertical transmission rate using the described protocol (59).To generate infected mothers, larvae were horizontally infected with E. aedis and allowed to pupate, then E. aedis(+)/H pupae were transferred to cages and provided with filter sterilized 10% sucrose ad libitum.Five days post-eclosion, E. aedis-infected adult females were provided a blood meal (defibrinated rabbit blood with 0.1M ATP) for 1-3 hours.Eggs vertically infected with E. aedis were collected on filter paper, dried, and stored under standard insectary conditions.Mosquitoes hatching from these vertically infected eggs are referred to throughout as E. aedis(+)/V.

Experimental design
We evaluated how prior infection by E. aedis affects S. marcescens bacterial infection in Ae. aegypti larval and adult mosquitoes.We conducted three separate experiments, each evaluating a different co-infection scenario.

Experiment 1: Ae. aegypti larvae vertically infected with E. aedis spores and subsequently orally infected with Serratia
E. aedis(+)/V Ae. aegypti larvae were reared to the early fourth instar.Simultaneously, we reared healthy larvae that were never exposed to E. aedis [E.aedis(−)].Fifty of each E. aedis(+)/V larvae and E. aedis(−) larvae were transferred to two 100 mL cups with 10-15 mL of DI water, and then one cup of each was inoculated with ~10 8 CFU of S. marcescens-GFP.Six hours post-Serratia infection, all treated larvae were washed twice with DI water and all cups were filled with 100 mL of DI water and incubated under controlled insectary conditions.24 hours post-Serratia infection, larvae from each group were moved to separate larval trays with 1 L of DI water and one piece of cat food.The trays were then incubated under controlled insectary conditions for the duration of the experiment.Throughout the experiment, dead larvae were removed from trays daily to reduce the possibility of horizontal transmission by ingestion of spores.Nonetheless, while the primary mode of infection was vertical, we cannot rule out the possibility that some horizontal infection may have occurred in this treatment.S. marcescens-GFP load was measured by homogenizing and culturing five larvae individually at 6, 24, and 48 hours post-Serratia infection, and five pupae individually at 96 hours post-Serratia infection on LB agar plates with 50 µg/mL kanamycin (LB + kan agar plates).Larval development was observed daily, and after pupation, pupae were transferred to cages with filter sterilized 10% sucrose ad libitum to monitor eclosion and survival of adults for 7 days post-Serratia infection.The workflow for experiment 1 is shown in Fig. S1.

Experiment 2: Ae. aegypti larvae horizontally infected with E. aedis spores and subsequently orally infected with Serratia
Fifty third instar E. aedis(+)/H and E. aedis(−) larvae were infected orally with S. mar cescens-GFP as described in experiment 1. S. marcescens-GFP load was measured by homogenizing and culturing five larvae individually from treated groups at 6, 24, 48, and 72 hours post-Serratia infection, and five pupae individually from treated groups at 96 hours post-Serratia infection on LB+kan agar plates.Development time and adult survival were monitored for 10 days post-Serratia infection.The workflow for experiment 2 is shown in Fig. S2.

Experiment 3: Ae. aegypti adult females horizontally infected with E. aedis spores and subsequently infected with Serratia
E. aedis(+)/H and E. aedis(−) larvae were reared to adulthood and then infected as adults with S. marcescens-GFP either orally or through thoracic injection.
For oral Serratia infection, n = 100 E. aedis(+)/H and n = 100 E. aedis(−) adult females were starved for 12 hours and then n = 50 females from each group were given a sugar meal containing 1.0 OD of S. marcescens-GFP (~10 9 CFU/mL) prepared in 10% sucrose (as described earlier in methods).Controls (n = 50 per group) were given sterile 10% sucrose.At 6, 24, 48, 72, 120, and 264 hours post-Serratia infection, five Serratia-exposed females from each E. aedis treatment group were transferred individually to 1.5 mL centrifuge tubes (one adult per tube), washed twice by DI water and homogenized in 150 µL of 1× PBS. 100 µL of homogenate suspension was then spread on LB +kan agar plates.In addition, the survival of females from all groups was observed for 11 days after Serratia infection.The workflow for experiment 3, oral infection, is shown in Fig. S3.
For intrathoracic injection, n = 50 E. aedis(+)/H and n = 50 E. aedis(−) adult females were anesthetized on ice and injected with S. marcescens-GFP by puncturing the soft tissue of the thoracic cleft of the mesothorax.Injections were performed with a borosilicate glass needle with an orifice diameter no greater than 500 μM (measured with a stage micrometer) and a Nanoject II Auto-Nanoliter Injector (Drummond).Approxi mately 69 nL of 1:500 dilution of 1.0 OD S. marcescens-GFP culture was injected per individual (equivalent to ~ 138 CFU/individual).In addition, n = 50 control females per group were injected with sterile 1× PBS.Females from each group were transferred to four separate cages (50 individuals/cage) with 10% sucrose ad libitum.At 0, 12, 18, and 96 hours post-S.marcescens-GFP infection, n = 5 females were transferred individually to 1.5 mL centrifuge tubes (one adult/tube) and homogenized in 150 μL of 1× PBS. 100 µL of homogenate was then spread on LB + kan agar plates.Survival of females from all groups was observed for 4 days after Serratia/PBS injection.The workflow for experiment 3, intrathoracic injection, is shown in Fig. S4.

Statistical analysis
Bacterial load data were analyzed using an ANOVA with E. aedis infection status, time, and replicate as predictor variables.Full models included the main effect of each variable plus an interaction between E. aedis infection status and time.In instances where the interaction was significant, we then stratified the data by time point and performed an ANOVA with E. aedis infection status and replicate as predictor variables for each time-specific model.To analyze pupation and eclosion, we used a Cox proportional hazards model with E. aedis infection status, Serratia infection status, and replicate as predictor variables.For survival, we could not conduct this analysis, either because there were no events in one or more treatment groups or because the data did not conform to the assumption of proportional hazards.In these cases, we performed log-rank tests to compare survival curves using survdiff from the survival package in R. Raw data, R code, and model outputs can be found in File S1, File S2, and File S3, respectively.

FIG 1
FIG 1 Life cycle of the microsporidia Edhazardia aedis in Ae. aegypti mosquitoes.In horizontal transmission, infective uninucleate spores of E. aedis are ingested by Ae. aegypti larvae from the aquatic environment.These spores invade the larval midgut lumen and develop into binucleate spores.Infected adults eclose and binucleate spores of E. aedis infect oenocytes of adult females.They are then transmitted through the hemocoel to the ovaries where they are vertically transmitted to offspring.When larvae hatch from infected eggs, they experience a highly virulent infection by E. aedis primarily in the fat body, which leads to the death of larvae and the release of infective uninucleate spores in the aquatic environment.These spores are ingested by susceptible larvae to complete the cycle.This image was created with BioRender (https://www.biorender.com/).

FIG 3
FIG 3 Larvae infected horizontally with E. aedis have significantly higher numbers of Serratia bacteria.Ae. aegypti larvae were either uninfected [E.aedis(−)] or horizontally exposed to E. aedis spores in larval water [E.aedis(+)/H] and then exposed orally to S. marcescens-GFP (Fig. S2).Oral exposure lasted for 6 hours at which point larvae were washed and transferred to clean water.Data were collected from a total of two replicate experiments.In each replicate, starting sample sizes were n = 50 mosquitoes per treatment group.(A) Pupation was significantly slowed by E. aedis (P = 3.39 × 10 −4 ) but was not affected by Serratia (P = 0.8669)