Environmental yeasts differentially impact the development and oviposition behavior of the Asian tiger mosquito Aedes albopictus

Abstract

Background

While the Asian tiger mosquito (Aedes albopictus), a known vector of many arboviruses, establishes symbiotic associations with environmentally acquired yeasts, their impact on mosquito biology remains poorly investigated. To better understand these associations, we hypothesized that waterborne yeasts colonizing the larval gut differentially support mosquito development based on their capacity to produce riboflavin or recycle nitrogen waste into proteins by secreting uricase, as B vitamins and amino acids are crucial for mosquito development. To address this hypothesis, we used axenic and gnotobiotic insects to gauge the specific impact of different environmental yeasts on Ae. albopictus development and survival. We then evaluated whether the observed variations across yeast species could be linked to differential uricolytic activities and varying quantities of riboflavin and proteins in insecta. Finally, given that mosquito oviposition site selection favors conditions that enhance offspring performance, we tested whether yeasts that promote faster development mediate oviposition site selection by gravid females.

Results

Differences in mosquito development times were observed based on the environmental yeast used. Yeasts like Rhodotorula mucilaginosa and Aureobasidium pullulans promoted rapid development and were associated with improved survival. Conversely, yeasts such as Torulaspora delbrueckii and Martiniozyma asiatica, which led to slower development, produced smaller adults. Notably, R. mucilaginosa, which promoted the fastest development, provided high riboflavin intakes and enhance nitrogenous waste recycling and protein synthesis through strong uricolytic-ureolytic activity. Behavioral experiments indicated that yeasts promoting rapid development “attract gravid females.

Conclusions

Our findings highlight that a set of environmental yeasts present in natural larval breeding sites can be associated with improved mosquito development and survival by enhancing nutritional intake, thereby attracting gravid females. Variations in mosquito development time are likely linked to the differential levels of riboflavin production and nitrogenous waste recycling capacities among yeast species. This study opens new perspectives on the trophic interactions between mosquitoes and their mycobiota, emphasizing the importance of nitrogen-containing molecules such as essential amino acids, proteins, or vitamins provided by the mycobiota.

Video Abstract

Background

With nearly one million described species and up to 2.2 billion estimated ones, insects represent more than 80% of the animal biodiversity on Earth [1]. This class plays crucial roles in almost every continental ecosystem. Most notably, insects pollinate more than 80% of flowering plants [2], play a key role in natural nutrient cycles [3], and constitute a major source of food for many species [4]. Beside these beneficial roles, insects such as crop pests and disease vectors can also be a burden for human activities and health [5, 6]. Insects have succeeded in colonizing most continental ecosystems on Earth [7]. This success can be tied to symbiotic associations with microorganisms that provide complementary functions otherwise lacking. These associations help insects inhabit ecologically challenging niches (e.g., feeding on nutrient-imbalanced diets such as phloem, wood, and blood) or invade new ecological niches [8].

Most insects rely on microorganisms that synthesize essential metabolites they are unable to produce themselves and that are lacking in their diet [9,10,11]. This associated microbial community, called microbiota, include several fungi forming a mycobiota [12]. Insect mycobiota is often dominated by yeasts, which mostly establish commensal or mutualistic relationships with their host [13,14,15]. Based on the host degree of dependence, their association can be classified as obligate (or primary) or facultative (or secondary) [12]. Like bacteria, obligate yeast symbionts colonize specialized organs such as the mycetomes of certain Coleoptera (anobiid beetles) and Hemiptera (planthoppers, aphids, cicadas) species [16,17,18,19] or the mycangia of certain Coleoptera species (ambrosia and stag beetles) [20, 21]. Facultative yeast symbionts predominantly colonize the digestive tract [22,23,24,25] but also other tissues such as the crop [26, 27], ovaries [28], and Malpighian tubules [29, 30]. While yeast symbionts are known to provide insect protection against pathogens and parasites [31,32,33], they are also essential for the optimal development of their host by supplying essential nutrients [9, 12]. Among these, B vitamins are the most consistently supplied nutrients by obligate symbioses across insect taxa [8]. The supply of B vitamins by facultative symbionts, such as living bacteria and yeasts, has been shown to be a key factor in mosquito development [34,35,36]. By supplying riboflavin (vitamin B2), bacteria and yeasts stimulate the respiration of mosquito’s intestinal cells by inducing the biosynthesis of flavin adenine dinucleotide (FAD) and flavin mononucleotide (FMN), two essential cofactors of enzymes involved in the respiratory metabolism [34, 35]. They generate gut hypoxia (i.e., oxygen levels below 5%) that activates in turn hypoxia-induced transcription factors (HIFs) and stimulate a signal transduction cascade leading to the accumulation of neutral lipids in the fat body and molting [35]. Production of riboflavin highly varies across yeast species ranging from weak to strong [37]. This differential capacity to produce riboflavin could partly explain yeast-mediated variations in mosquito development time [38, 39]. In addition, it has been suggested that yeast species with the highest amounts of proteins lead to better accumulation of energy reserves in mosquitoes, favoring their larval development [39]. The presence of uricolytic-ureolytic yeasts in their mycobiota may help mosquitoes to recycle their major nitrogen waste (i.e., uric acid) into amino acids and proteins, similar to what has been previously reported in other insects, such as cochineal and sand flies [29, 30]. While Wang et al. [35] emphasized the necessity of riboflavin supply from gut bacteria to counteract its instability and support mosquito development, no study has explored how the quantitative production of riboflavin by waterborne yeasts colonizing the larval gut, along with their ability or inability to recycle nitrogen waste into amino acids and proteins, influences mosquito development and consequently female oviposition behavior. Moreover, to our knowledge, no research has investigated the potential role of gut microbiota in recycling nitrogen waste in mosquitoes.

Given that insect oviposition-site selection is strongly oriented toward maximizing the performance of their offspring [40], gravid female mosquitoes from Aedes species preferentially select breeding site waters colonized by conspecific larvae not starved and not infected with deleterious parasites [41, 42]. Gravid female mosquitoes are able to modify their oviposition behavior in response to olfactive, gustative, or tactile cues that are directly or indirectly linked with the presence of microbial communities [40, 43]. Therefore, an increasing number of studies highlight the role of waterborne microorganisms and their emitted volatile organic compounds (VOCs) in mediating the oviposition responses of Aedes species [44,45,46]. The response of gravid females to the presence of deleterious microorganisms within breeding sites can lead to lower number of eggs laid in water whenever microbial emitted cues are repellent or deterrent. While the presence of microorganisms such as yeasts that promote larval development through improved nutritional intake can lead to a higher number of eggs laid whenever the cues are attractive or stimulant [40, 43]. Clements [47] defined a repellent as a volatile compound that causes gravid females to make oriented flights away from its source, whereas a deterrent is a waterborne substance that reduces egg production. Conversely, an attractant is a volatile compound that induces gravid females to make oriented flights toward its source, while a stimulant is a waterborne substance that increases egg production. Based on these definitions, microbial oviposition attractants and repellents should act as long- to short-range cues, guiding gravid females toward or away from the source. In contrast, microbial oviposition stimulants and deterrents function more as contact cues, assisting gravid mosquitoes in making a final decision regarding egg production [43].

Following these previous observations, we hypothesized that waterborne yeasts colonizing the larval gut could differentially impact mosquito development depending on their capacity to produce riboflavin and/or recycle nitrogen waste into amino acids and proteins, thereby influencing oviposition behavior. We specifically address this question in the Asian tiger mosquito Aedes albopictus, a vector of more than 20 arboviruses and one of the most invasive mosquito species worldwide [48]. This mosquito species harbors a diverse and environmentally acquired fungal community, which includes up to 84% of yeasts [14, 49]. In this study, yeast species simultaneously isolated from larvae and water samples collected within breeding sites naturally colonized by Ae. albopictus larvae were evaluated for their ability and efficiency in restoring the development of axenic larvae. We assessed whether the observed variations in mosquito development time could be correlated with differences in riboflavin production and extracellular uricolytic activity. To confirm the importance of yeast nitrogen waste recycling in mosquito development, we performed a urease inhibition assay on larvae specifically associated with a uricolytic/ureolytic (Rhodotorula mucilaginosa) and a non-uricolytic/ureolytic (Martiniozyma asiatica) yeast species. Urease is the only enzyme in the yeast recycling pathway that is absent from the Asian tiger mosquito genome, preventing it from recycling its own waste. Finally, we tested whether waterborne yeasts colonizing larvae and promoting their development mediate oviposition site selection by gravid females through the emission of specific VOCs.

Material and methods

Study site and sampling procedure

Field sampling was performed in September–October 2019 within a community garden composed of 75 parcels and localized in Saint Priest (45°42′59.7″N, 4°56′17.214″E), Metropolis of Lyon, France. Third-/fourth-instar larvae and water samples were collected with sterile plastic pipettes and then transferred into 50-mL conical sterile tubes (Greiner Bio-One). Samples were collected from eight different breeding sites naturally colonized by Ae. albopictus larvae such as watering cans, buckets, or containers (Table S1). Water samples were systematically agitated before sampling to mix organic detritus. For each breeding site, four larvae and two water samples (one from the upper surface and another one at a depth of 10–30 cm) representing around 50 mL were collected (Table S1). All samples were properly stored in an icebox and brought to the laboratory to be directly processed.

Yeast isolation and morphotype-based characterization

Yeast isolation was performed using the selective Dichloran-rose bengal chloramphenicol (DRBC) agar medium (Biokar Diagnostics). A 100-μL aliquot of each homogenized water sample was plated onto a DRBC plate in triplicate. For larvae, a pool of four individuals was incubated 40 min at − 20 °C, surface-sterilized in 3% sodium hypochlorite solution for 1 min, rinsed in sterile water, surface-sterilized in 70% ethanol for 2.5 min, and rinsed three times in sterile water. The last wash water was carefully removed and systematically plated onto selective DRBC and nonselective brain heart infusion (BHI) plates to control the efficiency of the surface-sterilization treatment. Larvae were then homogenized in 400 μL of sterile 1 × PBS (Gibco) using sterile 1.5-mL microtube pestles. After a brief vortex agitation, the homogenate was plated onto four DRBC plates (100 μL/plate). All plates were incubated at 22 °C, checked daily for 2 weeks and a last time 3 weeks after plating. Characterization of yeast strains was first based on the identification of morphotypes according to different criteria, including color, shape, or size of colonies. Individual colonies corresponding to each morphotype from each sample were selected and streaked onto CYM agar plates (maltose 10 g.L⁻¹, glucose 20 g.L⁻¹, yeast extract 2 g.L⁻¹, tryptone 2 g.L⁻¹, MgSO₄ 0.5 g.L⁻¹, KH₂PO₄ 4.6 g.L⁻¹, 20 g.L⁻¹ agar), a nonselective culture medium that supports the growth of most yeast strains. To ensure the purity of colony-forming strains before their molecular identification and storage in 15% glycerol at − 80 °C, the newly formed colonies were then streaked again onto fresh CYM agar plates.

Yeast identification and extracellular uricolytic activity measurement

Genomic DNA from each morphotype of the yeast isolates was extracted according to the method described by Liu et al. [50]. The internal transcribed spacer (ITS) or the D1/D2 region of the large subunit (LSU) ribosomal DNA (rDNA) were amplified from purified genomic yeast DNA using the ITS1 F (5′-CTT GGT CAT TTA GAG GAA GTA A- 3′)/ITS4R (5′-TCC TCC GCT TAT TGA TAT GC- 3′; [51, 52]) or NL1 F (5′-GCA TAT CAA TAA GCG GAG GAA AAG- 3)/NL4R (5′-GGT CCG TGT TTC AAG ACG G- 3′; [53]) primers pairs, respectively. A 25-μL PCR reaction mix was prepared with 2.5 μL of 10 × polymerase buffer (Invitrogen), 0.75 μL of MgCl₂ (50-mM, Invitrogen), 0.3 μL of BSA (20 mg.mL⁻¹, New England Biolabs), 2.5 μL of dNTPs (2 mM each, Thermo Fisher Scientific), 1 μL of each primer (20 μM, Invitrogen), and 0.1 μL of Taq DNA polymerase (Invitrogen) and 60 ng of DNA. Cycling conditions, performed on a T100 thermal cycler (Bio-Rad), consisted of 3 min at 94 °C followed by 35 cycles of 45 s at 94 °C, 45 s at 55 °C, and 1 min at 72 °C, with a final extension of 10 min at 72 °C. Control reactions without nucleic acid were systematically run in parallel. PCR products were purified, and both strands were sequenced by Microsynth France SAS (Vaulx-en-Velin) using the Sanger method. Forward and reverse sequences were manually curated and assembled. Curated sequences were compared to reference sequences available in the NCBI GenBank (https://blast.ncbi.nlm.nih.gov/Blast.cgi) and UNITE databases (https://unite.ut.ee) using BLASTn. DNA sequences are available in GenBank under accession numbers PQ037192-PQ037194 and PQ038096-PQ038114. Among all species isolated from the collected samples, only yeasts simultaneously associated with larvae and breeding-site waters were selected to evaluate their capacity to produce riboflavin, secrete uricase, and their subsequent impact on mosquito development.

Evaluation of yeast extracellular uricolytic activity and riboflavin production

The yeast extracellular uricolytic activity was assessed using drop-test assays on YPU agar medium (10 g.L⁻¹ yeast extract, 10 g.L⁻¹ peptone, 4 g.L⁻¹ uric acid, 20 g.L⁻¹ agar). For each yeast species, an overnight culture of 5 mL in CYM medium (28 °C at 180 rpm) was centrifugated at 3000 g for 5 min. The resulting cell pellet was washed twice with sterile water, resuspended in 3-mL sterile water, and diluted to obtain an optic density of 1. Serial tenfold dilutions from 10⁻¹ to 10⁻⁴ were spotted on YPU agar plates. After incubation at 28 °C for 48 h, uric acid utilization was confirmed by the formation of a degradation halo [30]. The assay was performed three times for each tested species. To maximize riboflavin production, each yeast species was cultivated in 3 mL of minimal medium adjusted to pH 6.8, with an initial concentration of 1.10⁶ cells.mL⁻¹ [54]. The medium contained 40 g.L⁻¹ glucose, 5 g.L⁻¹ (NH₄)₂SO₄, 2 g.L⁻¹ KH₂PO₄, 1 g.L⁻¹ K₂HPO₄, 0.5 g.L⁻¹ MgSO₄, 0.01 g.L⁻¹ FeSO₄, 0.01 g.L⁻¹ ZnSO₄, 0.005 g.L⁻¹ MnSO₄, and 0.005 g.L⁻¹ CuSO₄. The cultures were incubated under constant agitation (180 rpm) for 24 h at 28 °C in the dark. Riboflavin measurements were performed using 200-μL aliquots of the culture supernatants obtained after pelleting the cells by centrifugation at 10,000 g for 10 min. As previously described [55], the riboflavin concentration produced by each yeast species was measured using a SpectraMax iD3 microplate reader (Molecular Devices) with excitation and emission wavelengths set to 450 and 525 nm, respectively. Riboflavin concentrations were proportional to their relative fluorescence units and were quantified using an external riboflavin standard (Sigma-Aldrich) with an 8-point calibration curve (0, 0.1, 0.25, 0.5, 1, 2.5, 5, 10 μg.mL⁻¹). The assay was performed in triplicate for each yeast species tested. Yeast concentrations were estimated by cell counting under a Leica DM500 microscope (× 40) using a Thoma hemocytometer counting chamber.

Mosquito colony rearing

The laboratory mosquito colony AeAlbSP was established from Ae. albopictus larvae collected from breeding sites within the community garden of Saint Priest. After hatching, mosquito larvae were reared at 28 °C under a 16:8 h light:dark (L:D) photoperiod in plastic trays containing dechlorinated water. They were daily fed a crushed blend of 75% tropical-fish flakes (TetraMin, Tetra) and 25% yeast tablets (Biover) until pupation. Adults were raised in large BugDorm cages (32.5 × 32.5 × 32.5 cm) inside climatic chambers (Panasonic MLR- 352) maintained at 28 °C, 80% relative humidity, and under a 16:8 h L:D photoperiod. They were fed ad libitum a 10% sucrose solution. Adult females were fed on defibrinated sheep blood (Thermo Fisher Scientific) once a week through a membrane feeding system (Hemotek Ltd) using pig intestine as membrane. Eggs were collected on blotting paper and stored for up to 3 months under dry conditions at 28 °C.

Preparation of sterile diet and axenic larvae

A mixture of 5% w/v crushed tropical-fish flakes (TetraMin) and 1.5% w/v agar was prepared in distilled water. The solution was autoclaved, and 20 mL was poured into sterile Petri dishes and allowed to set. Agar plugs (~ 0.6 g) were excised from agar plates using 15-mL conical sterile tubes (Greiner Bio-One). In parallel, 10% w/v crushed tropical-fish flakes was also prepared in distilled water. The solution was autoclaved prior to being aliquoted (1 mL) and frozen (− 20 °C) until use. Ae. albopictus larvae free from non-intracellular microbiota (so-called thereafter axenic larvae) were generated using an egg sterilization protocol adapted from Correa et al. [56] and Raquin et al. [57]. Briefly, eggs fixed on a blotting paper were gently unhooked using a toothbrush and transferred into a 50-mL conical sterile tube with filter cap (Greiner Bio-One). Eggs were washed once with 50 mL of sterile water (Gibco) and then surface-sterilized in 50 mL of 70% ethanol for 5 min, followed by a 5 min wash in 50 mL of 3% sodium hypochlorite solution and an additional 5 min wash in 50 mL of 70% ethanol. The surface-sterilized eggs were then rinsed three times with 50-mL sterile water (Gibco) and immerged in 45-mL sterile water (Gibco). After adding 50-μL sterile 10% liquid fish-food, the tube was closed (with a filter cap), and then eggs were hatched under vacuum (− 20 mm Hg) overnight at 28 °C. For an optimal growth of axenic mosquitoes, a single plug of 5% w/v of crushed tropical-fish flakes and 10 first-instar (L1) larvae were added to each well of a sterile CellStar 6-well plate (Greiner Bio-One) filled with 6 mL of sterile water prior to their incubation at 28 °C in the dark.

For each well of each plate, a subset of randomly picked larvae was used to test the absence of living bacterial/fungal cells through a culturing approach. Mosquitoes were individually transferred into 1.5-mL Eppendorf tubes containing 150-μL sterile 1 × PBS (Gibco) and then crushed with sterile pestles. Then 50 μL of each homogenate was plated onto CYM and BHI agar plates and incubated at least 48 h at 28 °C. Additionally, the water of several plate wells was used to extract DNA following a published protocol [58] in the aim to amplify the fungal ITS and the bacterial 16S regions of rDNA genes as previously described [59]. Negative results were confirmed by an absence of bacterial/fungal growth and the absence of PCR amplification. Moreover, at the end of the experiment, the absence of microbial contamination was also confirmed by observing the water of each well under the microscope (× 40).

Generation of gnotobiotic and conventional larvae

Gnotobiotic larvae (i.e., larvae with controlled microbiota) were mono-associated to a selection of yeasts isolated from both larvae and breeding-site water. For each yeast species, 5 mL of sterile liquid CYM medium was inoculated with a single colony and incubated overnight at 28 °C under orbital agitation (180 rpm). This 5 mL of overnight culture was then used to inoculate 45 mL of sterile liquid CYM medium. After an 8-h incubation (28 °C, 180 rpm), a total of 2.10⁷ living cells were harvested by centrifugation (3000 g for 5 min). The resulting cell pellet was then washed twice with sterile water (Gibco) and resuspended in 1-mL sterile water. A volume of 100 μL of this yeast cell suspension (corresponding to 2.10⁶ living cells) was added to each well of six-well plates containing 6 mL of sterile water, a single plug of sterile fish-food, and 10 newly hatched axenic L1 larvae. In parallel, a “heat-killed (HK) condition” was also generated for each yeast species by preheating the 2.10⁷ cell inoculum (40 min at 100 °C) and adding 100 μL to each well, as previously described. Culture assays on CYM medium were performed to confirm that no viable yeasts remained after heat treatment. To generate conventional larvae (i.e., larvae associated with complex microbiota), the newly hatched axenic larvae were placed into plate wells containing a single plug and 6 mL of diluted (10⁻⁹) water collected from plastic trays containing rearing larvae of the mosquito laboratory colony. As for axenic, conventional and gnotobiotic larvae were incubated at 28 °C in the dark.

Evaluation of development time, survival, and adult size

Since all gnotobiotic larvae could not be generated at the same time, two distinct experiments each including axenic, conventional, and a set of gnotobiotic larvae were performed. For each experiment, a total of 180 L1 larvae (3 independent replicates of 6-well plates containing 10 L1 larvae per well each) were monitored for each infection status (i.e., conventional, axenic, and gnotobiotic larvae associated with living or dead yeasts). The presence of pupae was recorded daily at fixed hours for a period of 21 days after hatching. Pupae were individually transferred in 2-mL sterile tubes with ~ 300 μL of rearing water until emergence. Adults were then sexed and kept in their individual tubes prior to be stored together with the water of emergence at − 20 °C. Over the course of the experiment, the impact of each modality of infection on development times and survival rates was assessed. Development time represents the time (in days) needed to reach adult emergence, while survival rate corresponds to the proportion of larvae that reached the adult stage. Wing length was measured as a proxy of mosquito size [60]. To that end, adults stored at − 20 °C were thawed (n = 20 females and 20 males per infection status), and both wings were dissected under a Leica M80 stereomicroscope. Wings were fixed with 50 μL of Euparal (Carl Roth) onto 10-well epoxy-coated glass slides (Labelians). Slides were photographed at 20 × magnification with Leica MC170 HD camera. Images were analyzed with ImageJ (version 2.1.0/1.53c). Wing length was measured from the intersection of the second and third veins to the intersection of the seventh vein with the wing border as previously described [61].

Quantification of uricolytic activity and uric acid, soluble protein, and riboflavin content in larvae

Uricase activity as well as uric acid, soluble protein, and riboflavin content was quantified in second- (L2) and fourth-instar (L4) larvae retrieved from four contrasted infection status (i.e., axenic, conventional, and two gnotobiotic larvae displaying the highest differences in terms of development time). These measurements were performed on 10 pools of five larvae for each instar and each condition. Larvae reared in CellStar 6-well plates, as previously described, were dried and transferred into sterile 2-mL screw-cap tubes containing three 3-mm glass beads and then stored at − 80 °C. Larvae, protected from direct light to avoid riboflavin degradation, were freeze-dried for 8 h at − 50 °C. After dry-mass estimation, larvae were homogenized for 30 s in 200-μL cold sterile 1 × PBS (Gibco) using a Mini-Beadbeater- 8 (BioSpec). Regarding the quantification of uricase activity, uric acid and soluble protein content, supernatants obtained after centrifugation of larval homogenates (5 min at 1000 g and 4 °C) were transferred in 1.5-mL black Eppendorf tubes and stored at 4 °C. Uricase activity and uric acid concentrations were measured using the Amplex Red Uric Acid/Uricase Assay Kit (Invitrogen) according to the manufacturer’s instructions using non-diluted and 1/50 diluted supernatants for the L2 and L4 larvae, respectively. Following a reaction incubation of 30 min at 37 °C, the fluorescence intensity (excitation 530 nm, emission 570 nm) was measured using a SpectraMax iD3 microplate reader (Molecular Devices). Measurements of soluble protein concentrations were performed on the same supernatants using the Qubit Protein BR Assay (Invitrogen) according to the manufacturer’s instructions. For riboflavin measurements, supernatants obtained after centrifugation (5 in at 10,000 g and 4 °C) were transferred in sterile 1.5-mL black Eppendorf tubes and stored at 4 °C. Concentrations of the molecule were measured on non-diluted supernatants using the B₂ vitamin (riboflavin) Elisa kit (Abbexa) according to the manufacturer’s instructions. Following a reaction incubation of 15 min at 37 °C, the absorbance intensity was measured at 450 nm using a SpectraMax iD3 microplate reader. For uric acid and riboflavin concentrations, as well as uricase activities, standards of the quantified compound were used to correlate absorbance and concentrations as recommended by the manufacturer.

Uric acid recycling pathway in R. mucilaginosa and the impact of its inhibition on larval development

To confirm the presence of the uric acid recycling pathway in R. mucilaginosa, a genomic analysis was conducted using all publicly available, assembled, and annotated genomes, including the R. mucilaginosa strains ATCC 58901 and J31 from the JGI MycoCosm portal (https://mycocosm.jgi.doe.gov), as well as the reference strain GDMCC2.30 from NCBI GenBank (accession number PRJNA1034680). Specific primers were designed to amplify the genes encoding all enzymes involved in this recycling pathway (i.e., uricase, HIU hydrolase, OHCU decarboxylase, allantoinase, allantoicase, and urease; Table S2). Genomic DNA was extracted following the protocol described by Duval et al. [62]. PCR amplifications were performed on a T100 thermal cycler (Bio-Rad) in a 25-μL reaction volume, using the 5 × Hot FIREpol Blend Master Mix ready to load with 12.5 mM MgCl₂ (Solis BioDyne), 1 μM of each primer, and 20 ng of DNA. The cycling conditions included an initial denaturation at 94 °C for 12 min, followed by 35 cycles of 45 s at 94 °C, 45 s at 60 °C, and 45 s at 72 °C, with a final extension step at 72 °C for 10 min. PCR products were purified, and both strands were sequenced by Microsynth France SAS using the Sanger method. The forward and reverse sequences were manually curated and assembled. These curated sequences were compared to reference sequences using BLASTn (NCBI GenBank). The amplified DNA sequences and their percentage of homology to reference sequences are provided in Table S2.

To confirm the involvement of the yeast uric acid recycling pathway in larval development, a urease inhibition assay was performed. Urease was targeted because it is the only enzyme in this recycling pathway absent from the Asian tiger mosquito genome. This urease inhibition assay was conducted in parallel on gnotobiotic larvae associated with R. mucilaginosa (the yeast possessing the complete uric acid recycling pathway) and gnotobiotic larvae associated with M. asiatica (a yeast lacking the pathway). A 20-mM stock solution of acetohydroxamic acid (AHA; Sigma-Aldrich), an irreversible inhibitor of urease [63], was prepared in distilled water and sterilized by filtration (0.22 µm). Then, 750 μL of this inhibitor stock solution was added to each well of a CellStar 6-well plate containing 6 mL of sterile water, a single plug of 5% sterile fish-food, 10 newly hatched axenic L1 larvae, and 2.10⁶ living yeast cells. The impact of this inhibition on development time, survival rate, and sex ratio of gnotobiotic mosquitoes (n = 60 per infection status) was assessed as previously described. Prior to this experiment, we confirmed that the inhibitor concentration used 2.5 mM effectively inhibited R. mucilaginosa urease activity without affecting yeast growth or axenic larval development (Fig. S1; Table S4).

Oviposition bioassays

The impact of different yeast species at two cell concentrations (i.e., 3.10² and 3.10⁵ cells.mL⁻¹) on the oviposition behavior of gravid females was evaluated using two distinct and complementary laboratory bioassays (Fig. 1). If the lower value corresponds to the mean yeast concentration found in environmental breeding-site waters, the higher one corresponds to the yeast concentration used to generate gnotobiotic larvae. As described in detail below, oviposition-site selection bioassays (hereafter referred to as oviposition preference) consisted of small-cage three-choice oviposition tests, in which a single gravid female was allowed to choose between three different double-beaker system (DBS), referred to as oviposition sites (Fig. 1A): the control, containing only sterile water and CYM medium, and the two others containing water and medium, each inoculated with one of the two tested yeast concentrations (Fig. 1B). These oviposition preference bioassays were used to assess short-range attractiveness or repellency of volatile compounds emitted by yeasts but only because they were systematically paired with oviposition stimulation bioassays conducted in conical tubes, where a single gravid female was specifically exposed to either sterile water or water inoculated with one of the two tested yeast concentrations to enable the evaluation of stimulant or deterrent effects of yeast-derived waterborne substances (Fig. 1C).

Fig. 1

Experimental setups for mosquito oviposition preference and stimulation bioassays. A Schematic diagram of the double-beaker system (DBS), referred to as oviposition site, providing an oviposition substrate containing waterborne substances (sterile water, either inoculated with yeasts or not) and olfactory stimuli (volatile organic compounds emitted by either sterile CYM medium or yeast cultures) to a single gravid female during small-cage three-choice oviposition tests. B Schematic diagram of oviposition preference bioassays: small cage setup used to assess the impact of each yeast species and its cell concentration on oviposition-site selection by a single gravid female, based on emitted stimulant/deterrent waterborne substances and attractant/repellent volatiles during three-choice oviposition tests (n = 24). The control DBS offered sterile water as oviposition substrate and sterile CYM medium as stimulus to gravid females. C Schematic diagram of oviposition stimulation bioassays: no-choice tube setup offering a single oviposition substrate containing waterborne substances (sterile water or sterile water inoculated with yeasts at a specific cell concentration) that could stimulate or inhibit the gravid female from laying eggs on the water surface and onto the blotting paper being, which was slightly folded over one of its ends to create a resting area (n = 25)

It is important to note that, on their own, oviposition preference bioassays based on the total number of eggs laid in each tested condition compared to the control do not allow differentiation between stimulant or deterrent waterborne substances and attractant or repellent volatile compounds derived from yeasts. This is because these bioassays simultaneously provide oviposition substrates containing waterborne substances (sterile water or sterile water inoculated with yeasts) and olfactory stimuli (volatile organic compounds emitted by either sterile CYM medium or yeast cultures) to gravid females. Only by comparing the total number of eggs laid in control and yeast-inoculated conditions across both bioassays (preference vs stimulation) can the presence of attractant or repellent volatiles derived from yeasts be confirmed, thus enabling the evaluation of the short-range attractiveness or repellency of volatile compounds emitted by yeasts during oviposition preference bioassays. To enhance the high egg-laying success of Ae. albopictus females, we implemented several recommendations necessary for evaluating oviposition substrate preferences [64]. An equal number of male mosquitoes were maintained in the preparation cages to ensure appropriate mating. Young females (< 15 days old) were individually placed in cages or 50-mL conical tubes 3 days after their blood meal to ensure that the majority of most mosquitoes laid eggs, as the peak period for egg-laying of Ae. albopictus occurs around 4 days post-blood meal. This approach helped to prevent pseudo-outcomes by excluding mosquitoes that do not lay eggs or those that skip oviposition. Therefore, for these experiments, 120 males and 120 females were raised together in standard BugDorm cages (17.5 × 17.5 × 17.5 cm) placed inside climatic chambers since their emergence, and, after copulation, 10-day old females were fed with defibrinated sheep blood. Blood-fed females were collected with a manual aspirator and transferred for 2 days in new standard rearing cages provided with 10% sucrose solution to ensure blood digestion.

Oviposition preference was then carried out in standard BugDorm cages placed inside the insectary at 28 °C, 65–70% relative humidity, and 16:8 h L:D photoperiod. A specific device was developed to evaluate the attractiveness/repellency of yeasts (Fig. 1). Oviposition substrates containing waterborne substances (25 mL of sterile water inoculated or not with yeast cells) were offered in 100-mL sterile borosilicate glass beakers (5-cm diameter and 7-cm high) to females for laying eggs (Fig. 1A). As yeasts need carbon sources to produce attractive or repellent VOCs, a sterile 25-mL beaker containing 10 mL of yeast-added CYM medium (closed with carded cotton to avoid contamination) were inserted within the 100-ml glass beaker (Fig. 1A). For yeast-free oviposition substrate used as control, the 25-mL beaker was filled with 10 mL of sterile CYM medium. Individual gravid mosquitoes were introduced into standard cages (n = 24) provided with 10% sucrose solution and three oviposition beakers (Fig. 1B) corresponding to three different oviposition sites (i.e., yeast-free oviposition site, oviposition site containing a yeast concentration of 3.10² cells.mL⁻¹, and oviposition site with a yeast concentration of 3.10⁵ cells.mL⁻¹). Beakers with different oviposition substrates were placed at a random spot in each rearing cages. Eggs laid on the water surface and onto the blotting paper for treatments and controls were counted after 4-day exposure period. To test oviposition stimulation, individual gravid mosquitoes were placed into 50-mL conical sterile tubes (n = 25) containing a piece of blotting paper being slightly folded over one of its ends to create a resting area for the female mosquito. The tubes were filled with 20 mL of oviposition substrate containing waterborne substances (i.e., yeast-free sterile water, sterile water inoculated with a yeast concentration of 3.10² cells.mL⁻¹ or of 3.10⁵ cells.mL⁻¹) and provided with 10% sucrose solution (Fig. 1C). Previously, the eggs laid on the water surface and onto the blotting paper for treatments and controls were counted after 4-day exposure period.

Yeast volatolome analysis

Volatile organic compounds emitted by yeasts species that significantly impacted larval development and oviposition behavior of gravid mosquito females, as well as those emitted by the non-inoculated CYM medium (control), were analyzed using an established protocol [65]. Briefly, each yeast species was individually cultured in 75 mL of CYM liquid medium at 28 °C under continuous agitation (130 rpm) for 16 h to reach the exponential growth phase. Subsequently, 250-mL Erlenmeyer flasks containing either 100 mL of non-inoculated CYM medium or 100 mL of CYM medium inoculated at the cell concentration influencing the oviposition behavior were incubated under the same conditions for 18 h. Yeast VOCs, as well as compounds emitted by the CYM medium, were extracted by headspace-solid-phase micro-extraction (HS-SPME) using 65-μm PMS/DVB StableFlex™ fibers (Supelco). On the day of the analysis, each 250-mL Erlenmeyer flask containing the 100 mL of yeast culture (or non-inoculated CYM medium) was sealed with two layers of parafilm and let incubate under constant agitation during 15 min at room temperature. Cleaned fibers, previously submitted to a blank GC run, were then introduced into the headspace of each 250-mL Erlenmeyer flask and let opened during 30 min with constant agitation. After this incubation period, volatile compounds adsorbed on the fiber were immediately analyzed by GC–MS using a Hewlett-Packard 6890 N gas chromatography system coupled to a HP 5973 mass spectrometer (Agilent Technologies) and an Agilent DB- 5MS column (60-m long, 0.25 mm in diameter, 0.25-μm film thickness) with helium as carrier gas at a constant flow rate of 2.3 mL.min⁻¹ (Agilent). The oven temperature program was a 10 °C.min⁻¹ ramp from 50 to 280 °C. The ionization source was an electron impact at 70 eV, and mass spectra were recorded between 35 and 600 m/z at 1.29 scans.s⁻¹. Emitted VOCs were manually integrated using the MassHunter software (Agilent) and aligned according to their retention time into a matrix. Nonspecific yeast compounds were removed from the analysis. Key compounds were identified using NIST, Wiley, and aro.CRNS libraries, along with Kovats index calculation.

Statistical analyses

All statistical analyses were performed using R software version 4.2.0 and packages ade4, car, MASS, emmeans, DHARMa, pscl, coxme, lem4, and glmmTMB [66]. The effect of infection status and sex on development time was evaluated using a mixed Cox’s proportional hazards regression model with plate well as a random variable. The impact of infection status on survival rate and adult sex ratio was tested using a generalized linear model (GLM) with negative binomial distribution and probit as link function. The effect of infection status on adult size and metabolism parameters was assessed using linear models (LM). In the case of adult size, two separate models were generated based on the sex of the insect. The impact of yeast species and cell concentration on oviposition preferences was analyzed using a zero-inflated generalized linear mixed model (GLMM) with negative binomial distribution. The impact of yeast species and cell concentration on oviposition stimulation was analyzed using a generalized linear mixed model (GLMM) with negative binomial distribution.

For each model, the statistical significance of fixed effects was evaluated using type II ANOVA or Wald χ². Subsequent Tukey-HSD post hoc tests were conducted to test pairwise differences between groups. All analysis codes and parameters were deposited in the following site: https://doiorg.publicaciones.saludcastillayleon.es/10.5281/zenodo.12758792.

Results

Environmental yeasts differentially impact Ae. albopictus development

A total of 22 yeast species were isolated and identified from the collected samples (i.e., larvae and/or breeding-site waters). Most of them were budding yeasts, also called “true yeasts.” Indeed, 77.3% of the isolated species belonged to the phylum Ascomycota and the subphylum Saccharomycotina (Table 1). In addition, Basidiomycetous yeasts represented 13.6% of the isolated species.

Table 1 Taxonomic identification of yeasts isolated from larvae and water samples collected from Ae. albopictus breeding sites. Field sampling was performed in September–October 2019 within a community garden localized in Saint Priest (Metropolis of Lyon, France). The 10 waterborne yeasts colonizing the gut of Ae. albopictus larvae were selected to evaluate their capacity to secrete uricase and impact mosquito development

A total of 10 out of 22 yeast species were shared between larvae and breeding-site waters (i.e., Aureobasidium pullulans, Debaryomyces hansenii, Hanseniaspora uvarum, M. asiatica, Metschnikowia pulcherrima, Meyerozyma guilliermondii, Pichia kluyveri, R. mucilaginosa, Torulaspora delbrueckii, and Wickerhamomyces anomalus). Those yeast species were then selected because the major goal of this study was to evaluate the impact of waterborne yeasts colonizing the larval gut on the development of their host. Their ability to secrete uricase, produce riboflavin, and restore axenic larval development was then evaluated. Drop-test assays carried on YPU agar medium showed that 60% (6 out of 10) of the yeast species presented an extracellular uricolytic activity (Table 1, Fig. S2). Under culture conditions designed to promote riboflavin production, cellular riboflavin production was shown to vary across yeast species, ranging from 1.75 ± 0.08 to 0.07 ± 0.01 × 10⁻⁶ ng.cell⁻¹ (Fig. 2).

Fig. 2

Riboflavin production by environmental yeasts isolated from Ae. albopictus larvae and breeding-site waters. Yeasts were grown in minimal medium under constant agitation for 24 h in the dark at 28 °C. Riboflavin measurements were performed on the culture supernatant after pelleting the cells, using a fluorometer (450-nm excitation, 525-nm emission). The cell counts were estimated using a cell counter

The results we obtained revealed that infection status had a significant impact on the development time of larvae to adult emergence (Table 2). Regardless of the sex, axenic larvae exhibited a significant delay in development time to adult emergence compared to conventional larvae (Fig. 3A, Table S3). On average, the axenic adult mosquitoes emerged more than 7 days after conventional ones. Regardless of their infection status, male mosquitoes emerged first (7.88 ± 1.03 to 15.18 ± 2.26 days for males vs. 8.66 ± 1.11 to 16.82 ± 2.69 days for females; see Table S4). While all yeast species tested were able to restore the development of axenic larvae, important variations in development time were observed (Fig. 3A). Indeed, adult emergence was reached between 8 and 11 days (on average 8.66 ± 0.85 days for males and 9.70 ± 0.98 days for females) for R. mucilaginosa-associated individuals while it took from 13 to 19 days (on average 14.73 ± 1.72 days for males and 16.36 ± 2.48 days for females) for M. asiatica-associated individuals (Table S4). Culture assays from surface-sterilized L4 larvae confirmed that viable yeasts were colonizing all gnotobiotic individuals (Fig. S3), whereas no viable yeasts were detected in axenic and heat-killed conditions. We also demonstrated that yeasts need to be alive to restore an optimal mosquito development as heat-killed yeasts failed to restore the development of axenic larvae (Fig. 3B). Variations in mosquito development could not be attributed to a better ability of certain yeast species to colonize larvae since no differences in yeast concentrations within larvae were recorded between the best (i.e., R. mucilaginosa) and the worst (i.e., M. asiatica) development-recovering yeasts (Fig. S3). Strikingly, the two yeast species that promote faster development (i.e., R. mucilaginosa and A. pullulans) secrete uricase and produce high levels of riboflavin. Conversely, those resulting in slower development times (i.e., T. delbrueckii and M. asiatica) did not show any extracellular uricolytic activity and reduced levels of riboflavin (Table 1, Fig. 2). Impact of environmental yeast species on Ae. albopictusmosquito performance, survivorship, metabolism, and oviposition behavior

Table 2 Impact of environmental yeast species on Ae. albopictusmosquito performance, survivorship, metabolism, and oviposition behavior

Fig. 3

Developmental times of Ae. albopictus immature stages depending on their infection status. A Time in days from first-instar larvae to adult emergence shown for conventional (green), axenic (red), and gnotobiotic (blue) larvae associated with single living yeast species. B Time in days from first-instar larvae to adult emergence shown for conventional (green), axenic (red), and gnotobiotic (gray) larvae associated with single heat-killed yeast species. Studied individual mosquitoes are represented by dots. Abbreviations (Uri +) and (Uri −) indicate whether yeasts are able to secrete or not uricase, respectively. The average riboflavin production by yeast cells (10⁻⁶ ng.cell.⁻¹) is also provided for each species. Statistically significant differences between groups were identified with Tukey post hoc tests, and columns labelled with different letters are significantly different with a p-value < 0.05

Larval exposure to yeasts impacts individual survival and drives variations in certain adult traits

We showed that the infection status significantly impacted larval survival (Table 2). While the percentage of larvae reaching the adult stage did not differ significantly between axenic, conventional, and most gnotobiotic states (Fig. 4A), larvae associated with the two best development-recovering yeasts (i.e., R. mucilaginosa and A. pullulans) showed higher survival rates (on average 83.35 ± 20.36 and 83.76 ± 16.82% survival, respectively; see Table S4) compared to axenic and conventional larvae (on average 63.94 ± 16.45 and 66.81 ± 20.49% survival, respectively). Interestingly, this mortality decrease could not be observed when these two yeasts were heat-killed (Fig. 3B).

Fig. 4

Percentage of Ae. albopictus larvae reaching the adult stage depending on their infection status. A Larvae reaching the adult stage for conventional (green), axenic (red), and gnotobiotic (blue) larvae associated with single living yeast species. B Larvae reaching the adult stage for conventional (green), axenic (red), and gnotobiotic (gray) larvae associated with single heat-killed yeast species. Since all gnotobiotic, larvae (10 different yeast species) could not be generated at the same time, 2 distinct experiments including axenic, conventional, and 5 gnotobiotic larvae were performed. For each experiment, a total of 180 first-instar larvae were monitored for each infection status, and percentage of Ae. albopictus larvae reaching the adult stage was measured for each plate well. Abbreviations (Uri +) and (Uri −) indicate whether yeasts are able to secrete or not uricase, respectively. The average riboflavin production by yeast cells (10⁻⁶ ng.cell.⁻¹) is also provided for each species. Statistically significant differences between groups were identified with Tukey post hoc tests, and columns labelled with different letters are significantly different with a p-value < 0.05

To determine whether infection status during larval development resulted in adult phenotypic trait variations, we monitored the percentage of larvae reaching the adult stage, the adult sex ratio, and their wing lengths as a proxy of insect sizes. The infection status had little to no effect on mosquito sex ratio (Table 2) as the percentage of males ranged on average from 42.43 ± 21.70 to 61.73 ± 19.84% of the emerged adults (Fig. 5A). Conversely, it significantly impacted the adult sizes for both sexes (Table 2). Indeed, conventional males and females displayed larger wing lengths than their axenic cohorts (Fig. 5B, C).

Fig. 5

Biometric assessment of Ae. albopictus adult mosquitoes. A Male/female ratio obtained per each plate well for conventional, axenic, and gnotobiotic (associated with single yeast species) mosquitoes. Since all gnotobiotic larvae (10 different yeast species) could not be generated at the same time, 2 distinct experiments including axenic, conventional, and 5 gnotobiotic larvae were performed. For each experiment, a total of 180 first-instar larvae were monitored for each infection status (conventional, axenic, and gnotobiotic larvae). B Female wing lengths. C Male wing lengths. Points represent wing lengths of individual mosquitoes (n = 20 females and 20 males per infection status). Median wing length and interquartile ranges of all individuals are signified by the black dot and bars, respectively. Abbreviations (Uri +) and (Uri −) indicate whether yeasts are able to secrete or not uricase, respectively. The average riboflavin production by yeast cells (10⁻⁶ ng.cell⁻¹) is also provided for each species. Statistically significant differences between groups were identified with Tukey post hoc tests, and columns labelled with different letters are significantly different with a p-value < 0.05. ns, not significant

While most mosquitoes raised in the presence of living yeasts did not show any significant size difference compared to conventional ones, males emerging from larvae associated with the two worst development-recovering yeasts (T. delbrueckii and M. asiatica) appeared to be as small as axenic individuals. Regarding females, only larvae associated with M. asiatica developed into adult individuals that were significantly smaller than conventional ones (Fig. 5B, C). Females systematically displayed larger wing lengths than males independently from their infection status (on average from 1.46 ± 0.07 to 1.58 ± 0.06 mm for females and from 1.18 ± 0.04 to 1.29 ± 0.05 mm for males respectively; Table S4).

Larvae associated with the best development-recovering yeast exhibit higher riboflavin and protein concentrations

To assess whether variations in mosquito development time could be linked to the ability of environmental yeasts to produce riboflavin or to recycle nitrogenous waste through the secretion of uricase in insecta, riboflavin, uric acid, and soluble protein concentrations, as well as uricase activities, were measured in L2 and L4 larvae from four contrasted infection status (i.e., conventional, axenic, and gnotobiotic larvae associated with the best or the worst development-recovering yeast). Regardless of larval instars (L2 or L4), larvae associated with the more efficient development-recovering yeast (i.e., R. mucilaginosa) showed significantly higher riboflavin concentrations (from 2.32 to 2.42 times more) compared to individual associated with the worst (i.e., M. asiatica) development-recovering yeast (Fig. 6). Riboflavin concentrations increased during larval development, as L4 larvae showed higher quantities of riboflavin compared to L2 larvae. On average, this increase ranged from 3.67 ± 1.12 to 14.61 ± 2.27 ng.mg⁻¹ of dry weight for larvae associated with R. mucilaginosa and from 1.58 ± 0.53 to 6.03 ± 2.40 ng.mg⁻¹ of dry weight for individuals reared with M. asiatica (see Table S4). Interestingly, for both developmental stages (L2 and L4), no significant difference in riboflavin concentrations was detected either between conventional larvae and gnotobiotic larvae associated with R. mucilaginosa or between axenic larvae and gnotobiotic larvae associated with M. asiatica, respectively (Fig. 6).

Fig. 6

Quantification of riboflavin in Ae. albopictus larvae according to their developmental stage and infection status. Riboflavin concentrations were measured in second- and fourth-instar larvae retrieved from four contrasted infection statuses (axenic, conventional, and two gnotobiotic larvae displaying the highest differences in terms of development time). These measurements were performed on 10 pools of 5 s-instar larvae and 10 pools of 5 fourth-instar larvae. Statistically significant differences between groups were identified with Tukey post hoc tests, and columns labelled with different letters are significantly different with a p-value < 0.05

As expected, a strong positive linear correlation (Spearman’s correlation, Rs > 0.97, p-value < 0.001; Fig. 7A) was observed between the uricase activity (the enzyme) and the quantity of uric acid (its substrate). Regardless of the larval instar, uric acid concentrations and uricase activities measured in insecta were significantly higher (from 1.56 to 2.96 times higher) for larvae associated with R. mucilaginosa than for individuals reared with M. asiatica (Table 2; Fig. 7B, C). It is important to mention here that unlike R. mucilaginosa, the yeast species M. asiatica does not secrete uricase (Table 1; Fig. S2). Similar to what was observed for riboflavin, uric acid concentrations and uricase activities increased during larval development. No significant difference in uric acid concentrations and in uricase activities was observed between gnotobiotic larvae associated with M. asiatica and their axenic cohorts as well as between gnotobiotic larvae associated with R. mucilaginosa and conventional individuals (Fig. 7B, C). However, due to a strong variability of these parameters observed for conventional larvae, uric acid concentrations measured at the fourth instar were significantly lower compared to gnotobiotic larvae associated with R. mucilaginosa (Fig. 7B).

Fig. 7

Quantifying uric acid metabolism in Ae. albopictus larvae by stage and infection status. A Positive Spearman correlation between uricase activity levels and uric acid concentrations measured in second- and fourth-instar larvae from four different infection statuses, B uric acid concentrations, C uricase activity levels, and D soluble protein contents measured in second- and fourth-instar larvae retrieved from four contrasted infection statuses (axenic, conventional, and two gnotobiotic larvae displaying the highest differences in terms of development time). These measurements were performed on 10 pools of 5 s-instar larvae and 10 pools of 5 fourth-instar larvae. Statistically significant differences between groups were identified with Tukey post hoc tests, and columns labelled with different letters are significantly different with a p-value < 0.05

Same patterns were observed for soluble proteins at the second instar as larvae associated with R. mucilaginosa showed significantly higher protein concentrations (2.39 times more) compared to insects associated with M. asiatica (Fig. 7D). Moreover, no significant difference in protein concentrations was observed between gnotobiotic larvae associated with M. asiatica and axenic larvae, as well as between gnotobiotic larvae associated with R. mucilaginosa and conventional individuals (Fig. 7D). Soluble protein concentrations in L4 larvae reached similar values (on average from 39.89 ± 14.31 to 48.31 ± 17.76 µg.mg⁻¹ of dry weight) independently of their infection status (see Table S4).

Based on a genomic analysis of publicly available, assembled, and annotated R. mucilaginosa genomes, along with PCR amplifications, we demonstrated that all R. mucilaginosa strains, including the one used in this study, possess the complete set of genes encoding the uricolytic/ureolytic enzymes required for amino acids synthesis through nitrogen waste recycling (Fig. 8, Table S2).

Fig. 8

Uric acid recycling pathway present in R. mucilaginosa strains, including the one isolated from both Ae. albopictus larvae and water-breeding sites. An analysis of all publicly available, assembled, and annotated R. mucilaginosa genomes (strains ATCC 58901 and J31 from the JGI MycoCosm portal, as well as the reference strain GDMCC2.30 from NCBI GenBank, accession number PRJNA1034680) confirmed that R. mucilaginosa possess the complete set of genes encoding the uricolytic/ureolytic enzymes required for amino acids synthesis through nitrogen waste recycling. Some of these genes (blue arrows) were identified as expressed in an available transcriptome [67], and their presence in the genome of the strain used in this study was confirmed by PCR (red stars)

The presence of this recycling pathway in the genome of the R. mucilaginosa strain used is consistent with the observed results, which showed higher concentrations of uric acid and soluble proteins, along with increased uricase activity levels, in R. mucilaginosa-associated larvae, despite the yeast’s ability to degrade uric acid. Indeed, during larval development, proteins undergo autolysis and are catabolized into uric acid. As a result, the higher protein content in larvae, attributed to the yeast’s ability to recycle their primary nitrogen waste (i.e., uric acid) into amino acids, might have led to increased uric acid production and uricase activity levels.

Using a urease inhibition assay, we confirmed that the uricolytic/ureolytic yeast R. mucilaginosa aids larvae in recycling their primary nitrogen waste (uric acid) into amino acids and proteins, leading to improved development. Indeed, inhibiting the uric acid recycling pathway of R. mucilaginosa with the urease inhibitor acetohydroxamic acid (AHA) caused a significant developmental delay in treated gnotobiotic larvae compared to untreated ones (Fig. 9; Table S3). Adult emergence was delayed by an average of 1.78 days for females and 2 days for males (Table S4). Since urease is the final enzyme in the yeast’s recycling pathway and is absent from the Asian tiger mosquito genome, its inhibition disrupts the nitrogen waste recycling. In contrast, the AHA inhibitor had no impact on the development time of gnotobiotic larvae associated with M. asiatica, which lacks the pathway (Fig. 9; Tables S3, S4). No significant impact on survival rate (df = 1; χ² = 0.010; p-value = 0.921) and sex ratio (df = 1; χ² = 2.826; p-value = 0.093) was observed in the presence or absence of the AHA inhibitor for both infection status.

Fig. 9

Impact of acetohydroxamic acid-mediated urease inhibition on the development time of gnotobiotic Ae. albopictus larvae associated with uricolytic/ureolytic (R. mucilaginosa) and non-uricolytic/ureolytic (M. asiatica) yeast species. To confirm the involvement of the yeast uric acid recycling pathway in larval development, we inhibited it by targeting urease, the only enzyme in the yeast recycling pathway absent from the Asian tiger mosquito genome. The impact of this inhibition on the development time of gnotobiotic mosquitoes (n = 60 per infection status) was assessed using an inhibitor concentration of 2.5 mM. Prior to this experiment, we confirmed that this inhibitor concentration effectively inhibited R. mucilaginosa urease activity without affecting yeast growth or axenic larval development. Statistically significant differences between groups were identified with the Tukey post hoc tests. Columns labelled with an asterisk (*) are significantly different with a p-value < 0.05. ns, not significant

Environmental yeasts facilitating mosquito development mediate Ae. albopictus oviposition

The two best (i.e., R. mucilaginosa, A. pullulans), worst (i.e., T. delbrueckii, M. asiatica), and intermediate (i.e., H. uvarum, M. pulcherrima) development-recovering yeasts were selected, and we evaluated their impact on the oviposition behavior of gravid females through two distinct and complementary bioassays (i.e., oviposition preference and oviposition stimulation). During the oviposition preference bioassays conducted in cages, individual female mosquitoes could choose to lay eggs in one of three beakers corresponding to three different oviposition sites (i.e., a yeast-free site, a site containing a yeast concentration of 3.10² cells.mL⁻¹, and a site with a yeast concentration of 3.10⁵ cells.mL⁻¹). Each site contained an oviposition substrate with potentially stimulant or deterrent waterborne substances and provided an olfactory stimulus that could act as short-range attractant or repellent. The total number of eggs laid by each female per cage (n = 24), regardless of the selected oviposition site, was significantly influenced by the yeast species (Table 2). When R. mucilaginosa and A. pullulans were present, individual females laid more eggs per cage compared to all other conditions, with averages of 42.11 ± 15.51 and 42.68 ± 27.77 eggs, respectively (Fig. 10A; Table S4). Conversely, T. delbrueckii caused a significant reduction in the total number of eggs laid per cage (10.43 ± 14.08 eggs; Table S4). Furthermore, the percentage of females that did not lay eggs increased to 23% and 29% in the presence of M. asiatica and T. delbrueckii, respectively (Fig. 10A).

Fig. 10

Impact of yeast species and their cell concentrations on oviposition behavior of Ae. albopictus gravid females. The impact of six yeast species (R. mucilaginosa, A. pullulans, H. uvarum, M. pulcherrima, T. delbrueckii, and M. asiatica) at two cell concentrations (3.10² and 3.10⁵ cells.mL⁻¹) on the oviposition behavior of gravid females was evaluated using two distinct and complementary laboratory bioassays. Oviposition preference bioassays (corresponding to A & B) consisted of small-cage three-choice oviposition tests, in which each single gravid female (n = 24) was allowed to choose between three oviposition sites (double-beaker systems) providing oviposition substrates containing waterborne substances (sterile water or sterile water inoculated with yeasts) and olfactory stimuli (volatile organic compounds emitted by sterile CYM medium or yeast cultures). A Total number of eggs laid by each female per cage, regardless of the selected oviposition site. The pie charts represent the percentages of females that have laid eggs (shown in gray) or have not laid eggs (shown in black). B Number of eggs laid by each female in each oviposition site during three-choice oviposition tests. These oviposition preference bioassays were used to assess short-range attractiveness or repellency of volatile compounds emitted by yeasts but only because they were systematically paired with oviposition stimulation bioassays conducted in 50-mL conical tubes (C), where a single gravid female (n = 25) was specifically exposed to either sterile water or water inoculated with one of the two tested yeast concentrations to enable the evaluation of stimulant or deterrent effects of yeast-derived waterborne substances. C Number of eggs laid by each female according to each type of oviposition substrate. Statistically significant differences between groups were identified with the Tukey post hoc tests. Columns labelled with different letters or with an asterisk (*) are significantly different with a p-value < 0.05. Columns labelled with a dot (•) are almost significantly different with a p-value < 0.1

Interestingly, all female mosquitoes laid eggs when R. mucilaginosa was present (Fig. 10A). The oviposition preference bioassays also showed that both yeast species and their cell concentrations significantly influenced the selection of oviposition sites (Table 2). Indeed, female mosquitoes laid more eggs (on average, 19.11 ± 18.65 and 22.41 ± 23.44 eggs per female; see Table S4) in oviposition sites containing the highest cell concentration (3.10⁵ cells.mL⁻¹) for the yeast species R. mucilaginosa and A. pullulans, respectively (Fig. 10B). Conversely, during oviposition stimulation bioassays, exposing gravid females to waterborne substances derived from the presence of these two yeasts at the same highest cell concentrations did not stimulate oviposition (Fig. 10C). Taken together, these results suggest that yeasts that promote rapid development attract gravid females at short-range distances. A significant oviposition stimulation (≈1.5 times more eggs compared to the condition without yeast; see Table S4) was only observed with high cell concentrations of H. uvarum and M. asiatica (Fig. 10C). Based on this observation, the reduced number of eggs laid (on average, 8.55 ± 12.78 at 3.10⁵ cells.mL⁻¹) and the increased proportion of females (23%) that did not lay eggs in the presence of M. asiatica during oviposition preference bioassays strongly suggested a short-range repellency of this yeast species (Fig. 10A, B; Table S4).

In contrast, the waterborne substances derived from the presence of T. delbrueckii at high cell concentration during oviposition stimulation bioassays acted as deterrents (Fig. 10C). Therefore, the significant reduction in eggs laid (on average, 6.95 ± 12.33 at 3.10⁵ cells.mL⁻¹) and the highest proportion of gravid females (29%) that did not lay eggs in the presence of T. delbrueckii during oviposition preference bioassays highlighted a combination of short-range repellency and deterrence of this yeast species (Fig. 10B).

The analysis of VOCs emitted by yeasts that significantly impacted the oviposition behavior of gravid mosquito females at high cell concentration (i.e., 3.10⁵ cells.mL⁻¹), due to short-range attractiveness (i.e., R. mucilaginosa and A. pullulans) or repellency (i.e., M. asiatica and T. delbrueckii), revealed distinct VOC profiles between these two yeast categories (Fig. 11A; Table S5). Repellent yeasts emitted a greater diversity of VOCs compared to attractive yeasts, with at least four specific volatiles identified, including the known ovipositional repellent ethyl octanoate [68]. Only two volatile compounds, 3-methyl- 1-butanol and 2-methyl- 1-butanol, characterized in literature as mosquito attractants [65, 69] were shared by attractive yeasts. These two compounds were also emitted by repellent yeasts, but at significantly higher concentrations, which are known to be repellent [70, 71] (Fig. 11B, C).

Fig. 11

Volatile organic compounds emitted by yeast species significantly affecting larval development and oviposition behavior of gravid mosquito females. Each yeast culture was prepared in CYM medium at a final volume of 100 mL and an initial cell concentration of 3.10⁵ cells.mL.⁻¹ (i.e., concentration shown to significantly impact both larval development and oviposition behavior). Cultures were incubated under constant agitation (130 rpm) for 18 h prior VOC analysis. A Heatmap illustrating the relative abundance of emitted VOCs detected by HS-SPME/GC–MS from five independent replicates of each yeast culture. B, C Relative abundances of 3-methyl- 1-butanol and 2-methyl- 1-butanol, two microbial VOCs shared by both attractive (R. mucilaginosa and A. pullulans) and repellent (M. asiatica and T. delbrueckii) yeast species. The median is indicated by a horizontal black line. Statistically significant differences between yeasts, labelled with different letters, were determined using Dunn’s test with Bonferroni-Holm correction (p-value < 0.05)

Discussion

Mosquito larvae, including Ae. albopictus, develop in aquatic habitats, while adults live in terrestrial habitats. This separation of ecological niches allows larvae and adults to avoid exploiting the same resources and minimize intraspecific competition [72]. While terrestrial Ae. albopictus adults of both sexes consume various forms of plant sugars, especially flower nectar [73], larvae primarily feed on organic matter present in water, such as microorganisms or plant debris, using filtering, grazing, and shredding behaviors [74]. While these developmental stages inhabit different environments, they are not completely independent of each other as the quality of larval development can significantly impact various adult life history traits, such as body size and lifespan. Factors influencing larval development include competition, temperature, and the nutritional status of larvae [75, 76]. Insects consuming nutrient-deficient diets commonly harbor heritable microorganisms that serve as nutritional symbionts [77]. However, to our knowledge, no heritable nutrient-provisioning symbionts were identified in mosquitoes [78]. Therefore, Ae. albopictus larvae exposed to resource limitation strongly depend on their microbiota, primarily acquired from the environment as there is a continuous flow of bacteria and fungi between larval gut and breeding-site water, for their development [79, 80]. To unravel the mechanisms behind this interaction between larvae and their microbiota, as well as the role of the microbiota in mosquito biology, methods enabling the generation of axenic and gnotobiotic larvae were developed [36, 56, 81]. However, to date, axenic rearing from larvae to adults has mostly been reported for the closely related yellow fever mosquito Aedes aegypti, as axenic Ae. albopictus L1 larvae show very slow growth rates and do not reach pupation [49]. The fact that axenic larvae or those supplemented with heat-killed microorganisms took longer to develop and produced smaller adults compared to gnotobiotic and conventional ones indicates that living microorganisms are mandatory for an optimal larval growth [34, 56]. Such studies based on axenic/gnotobiotic models highlight the key role of bacteria in supplying essential nutrients like B vitamins, and especially riboflavin and folate, underscoring the nutritional significance of the microbiota in mosquito larval development [35, 36]. In addition, larval exposure to different bacterial species results in varying benefits on development (e.g., time development and pupation rate) and subsequent adult traits (e.g., body size and vector competence) [82, 83]. In consequence, one may expect that mosquitoes have develop strategies to ensure associations with the most beneficial microorganisms in their environment [58, 84, 85].

Most studies on the microbiota during larval development focused on bacteria, even though several studies have highlighted the presence of an important fungal diversity in Ae. albopictus mosquitoes at both larval and adult stages [14, 49, 80, 86]. While yeasts or yeast-like fungi are recognized as significant constituents of its mycobiota, reaching up to 84% in certain populations [14, 49], the influence of environmental yeasts naturally occurring in water-breeding sites on Ae. albopictus development remains unexplored. To fill this gap, we employed an axenic/gnotobiotic methodology to assess how waterborne yeasts colonizing the larval gut affect development and subsequent adult traits. As different yeasts produce variable amounts of riboflavin [37], and showed variations in their abilities to recycle uric acid into essential amino acids [30], we then evaluated whether those factors could be correlated with the larval development. A recent study highlighted that Ae. albopictus mosquitoes are more attracted to oviposition sites with higher density of conspecific larvae associated with their microbiota [42]. Consequently, we also investigated the link between habitat attractiveness to gravid females and its composition, specifically focusing on water colonized with yeast species that promote mosquito development.

Culturable yeasts in breeding-site waters and larvae are dominated by Ascomycota species

Despite a growing amount of information reporting the composition and diversity of yeasts associated with Ae. albopictus at the adult stage [14, 86, 87], as well as their implication in mosquito fructose nutrition [24, 27, 88], in comparison, there is limited data available on yeasts associated with larvae and their water habitats. Our study based on the culturable subset of the yeast communities, similarly to previous and more exhaustive findings [80], supports that Ae. albopictus larvae and their aquatic habitats are predominantly colonized by Ascomycota species, as evidenced by the predominance (> 75%) of isolated yeasts belonging to the Saccharomycotina subphylum. This result can be explained by the ubiquity of Ascomycota in freshwater ecosystems [89], including mosquito breeding sites [58, 90]. We showed that the yeast communities associated with larvae largely overlapped those of the breeding-site water. Such similarity can be explained by the filter-feeding behavior of mosquito larvae which presumably captures a large portion of yeasts from their surrounding environment [74]. Among the isolated yeasts, R. mucilaginosa, D. hansenii, P. kluyveri, and M. guilliermondii are classified as aquatic yeasts due to their remarkable ability to survive in fresh or marine water [91]. Furthermore, they have already been found to be associated with larvae from Aedes, Culex, and Anopheles species [38, 92]. Overall, our results support the idea that the larval habitat is the major factor shaping the mycobiota of larvae, as previously suggested for bacterial and fungal communities associated with the breeding sites of Ae. albopictus [80] and Ae. aegypti [58]. However, we noted the presence of two nectar-generalist species, A. pullulans and H. uvarum [93], both in larvae and breeding-site waters. As these yeasts are usually associated with flowers, this observation suggests an external route of breeding-site colonization for those. Interestingly, A. pullulans appears to be one of the most abundant and widespread yeast species among Ae. albopictus females, as it was detected in 90.5% of the mosquito specimens collected in France, Madagascar, and Vietnam [14]. Consequently, the presence of these yeasts probably results from their release during egg-laying, defecation, or after adult death, similar to what was demonstrated for adult-associated bacteria [94].

Given that mycobiota composition in mosquito larvae can also be influenced by interkingdom interactions [49] and host selection [58], only yeast species isolated from both larvae and water-breeding sites were further tested for their ability to restore the development of axenic larvae.

Exposure to different environmental yeasts results in varying degrees of benefits on mosquito development and subsequent adult traits

Similar to what was observed by Correa et al. [56] concerning the closely related species Ae. aegypti, axenic Ae. albopictus larvae exhibit a growth delay of more than 7 days compared to conventional larvae. Gnotobiotic larvae developing most rapidly (i.e., larvae associated with R. mucilaginosa and A. pullulans) have lower mortality rates (from ~ 17%) than axenic (~ 36%) or conventional larvae (~ 33%). The higher survival rate in these gnotobiotic modalities could be explained both by better nutritional intake, which promotes better resistance to environmental stress, and/or the absence of antagonistic interactions between microorganisms [56, 82]. Moreover, longer larval development times and pupation time, typically observed in cases of nutritional deficiencies, are often associated with high mortality risks such as predation or environmental condition instability [39]. Contrary to what has been observed for Culex pipiens mosquitoes, where the presence of the yeast Cryptococcus gattii prevents high pupation rates due to strong nutritional deficiencies [38], all environmental yeasts tested in the present study enable a significant reduction in the development time of axenic larvae and the production of adults. The impact of sex on development was due to sexual dimorphism in Ae. albopictus, with females developing slower and being larger than males [95]. Our results also demonstrate that the differential impact of yeasts during larval development triggers metabolic changes that affect adult traits, as reflected by differences in the size of emerging adults. Indeed, yeasts leading to longer development times (T. delbrueckii and M. asiatica) generate smaller adult mosquitoes. Such reduced body size may impact mosquito physiology and therefore its fitness and vector capacity [83]. For example, in Ae. aegypti and Ae. albopictus, smaller individuals exhibit higher rates of dengue virus infection and dissemination [96]. The observed differences in development time were shown to be connected to the presence of living yeasts supplying nutrients, rather than variations in the larvae’s ability to be colonized by the yeasts.

The supply of B vitamins by gut microbiota was demonstrated to be essential for larval development [34,35,36], as well as for fecundity, longevity, and vector competence of adult Ae. aegypti [78]. Regarding larval development, the absence of microbiota that can supply riboflavin, a highly photosensitive B vitamin decaying within 6–12 h, causes larvae to arrest their development at the first stage if they are placed under standard conditions and not maintained in darkness [35, 49, 85, 97]. Our results showed that the observed variations in mosquito development time are positively correlated with levels of riboflavin production by yeasts. The best development-recovering yeast, R. mucilaginosa, supplies twice as much riboflavin throughout larval development, comparable to the amounts provided by the complex microbiota of conventional larvae. Conversely, the worst development-recovering yeast, M. asiatica, produces low amounts of riboflavin, with quantities not exceeding those found in the larval food. The increase in riboflavin concentrations observed between stages L2 and L4 might suggest a continuous production of riboflavin by yeast cells, whose presence increased over time in the plate wells and likely within the larvae’s gut as well. This is because animals, including mosquitoes, are unable to store riboflavin [78]. However, other B vitamins supplied by microbiota could be essential for optimal mosquito development. For instance, the absence of folate reduces the proportion of Ae. aegypti larvae that successfully pupate and emerge as adults [36]. At this stage, we did not investigate whether environmental yeasts may also supply the host with other B vitamins.

Larvae must reach a critical mass before transitioning to the next stage of development and undergoing metamorphosis [98]. Moreover, yeast species with the highest amounts of proteins were suggested to lead to a better accumulation of energy reserves in mosquitoes, favoring their larval development [39]. Taken together, these observations suggest that protein and amino acid metabolism are crucial for mosquito juvenile development with regard to larvae-associated microbiota. Furthermore, transcriptional differences observed among axenic and conventional Ae. aegypti larvae unveiled an alteration in protein and amino acid metabolism in axenic larvae. This alteration involved the repression of genes encoding peptidases and the overexpression of genes encoding amino acid transporters [99]. Additionally, it has been demonstrated in Drosophila melanogaster that some members of the mycobiota are able to extract amino acids from nutrient-poor diets, which can be provided to their host, subsequently promoting its development [100]. Interestingly, an increasing number of studies highlighted the important role of microbiota (bacteria and yeasts) in recycling uric acid, the major nitrogen waste, in several insects such as cockroach, cochineal, sand fly, planthopper, and ant [29, 30, 101,102,103]. This breakdown of uric acid by microorganisms releases valuable nitrogenous compounds, including essential amino acids, which can be absorbed by insects and reused for various physiological processes, including protein synthesis [104]. In insects, including mosquitoes, uric acid tends to accumulate in the fat bodies before being stored in the Malpighian tubules prior its elimination [105]. Uricolytic yeasts, which possess the uricase enzyme that catabolizes uric acid, were found associated with the midgut and/or Malpighian tubules of certain insects such cochineal and sand fly [29, 30]. Interestingly, our results show that yeast associations resulting in longer development times (i.e., T. delbrueckii and M. asiatica) are not uricolytic, whereas the best development-recovering yeasts (i.e., R. mucilaginosa and A. pullulans) exhibit uricase extracellular activities. Axenic larvae exhibit uricase activity, which is attributed to the uricolytic pathway present in Ae. albopictus. This pathway converts uric acid to urea and glyoxylic acid through a three-step process involving the urate oxidase, allantoinase, and allantoicase enzymes [106].

However, Ae. albopictus cannot recycle this form of nitrogen waste into amino acids on its own, as it lacks urease, the enzyme that converts urea into ammonium [106]. Instead, it must rely on its microbiota to complete this recycling process. Larvae associated with the uricolytic/ureolytic yeast R. mucilaginosa exhibited significantly higher uric acid content, uricase activity, and protein levels compared to both axenic larvae and gnotobiotic larvae associated with the non-uricolytic/ureolytic yeast M. asiatica. Similar to M. guilliermondii, a yeast species involved in uric acid metabolism in sandflies [30], R. mucilaginosa encodes the full set of enzymes required for amino acid synthesis through nitrogen waste recycling. The presence of this recycling pathway aligns with our experimental findings, which revealed elevated levels of soluble proteins, uric acid, and uricase activity in larvae associated with R. mucilaginosa, despite the yeast’s ability to degrade uric acid. Indeed, during larval development, protein autolysis generates uric acid as a catabolic byproduct [107]. Consequently, the ability of R. mucilaginosa to recycle uric acid into amino acids increases protein availability in the larvae, which in turn promotes higher uric acid production and greater uricase activity. The role of yeast-driven nitrogen waste recycling in supporting larval development was further evidenced through the uricase inhibition bioassay, which caused a significant delay in adult emergence among gnotobiotic larvae associated with R. mucilaginosa when treated with the inhibitor, compared to untreated ones. Therefore, as observed by Vera-Ponce et al. [29] for cochineal, our results suggest that R. mucilaginosa is involved in nitrogenous waste recycling and subsequently releases essential amino acids that support Ae. albopictus protein synthesis and development. Since proteins from muscles, intestines, cuticle, and other tissues are autolyzed into amino acids and then catabolized into uric acid during molts [107], our results imply that the larval developmental rate depends on the efficiency with which the larval microbiota recycles uric acid. The lack of difference in protein levels observed at the L4 stage could be attributed to the completion of larval development, during which proteins remain stable, a phenomenon observed in the black soldier fly [108].

Yeasts in aquatic habitats that promote mosquito development positively influence oviposition

In insects, the selection of egg-laying sites by females is a crucial step that determines the fate of their progeny and is therefore expected to be under strong selective pressures [109]. Given that mosquito immature stages are gregarious and unable to move from one habitat to another, meaning they cannot escape unsuitable habitats, female preferences in oviposition sites are oriented toward maximizing the performance of their offspring, as natural selection might filter out progeny from mothers who make incorrect decisions [40]. Gravid female mosquitoes from Aedes species have been shown to preferentially select breeding sites colonized by conspecific larvae, and consequently their associated microbiota, provided they are not starved or infected with deleterious parasites [41, 42]. Since gravid female mosquitoes rely on olfactory and gustatory signals to select oviposition sites, including CO₂ and VOCs emitted by microorganisms in water [40, 110], an increasing number of studies have pointed out the role of waterborne bacteria and their emitted VOCs in mediating the oviposition responses of Aedes species [44,45,46, 111, 112]. In contrast to bacteria, yeast’s ability to attract gravid female mosquitoes has been poorly investigated [40]. To our knowledge, it has only been shown that the presence of Candida pseudoglaebosa in the oviposition-site water attracts gravid females for oviposition and subsequently colonizes them [41]. In the present study, we tested whether the presence of yeast species that promote rapid Ae. albopictus development would attract or stimulate more gravid females to lay eggs. To that end, we compared the impact of the two best (R. mucilaginosa, A. pullulans), worst (T. delbrueckii, M. asiatica), and intermediate (H. uvarum, M. pulcherrima) development-recovering yeasts on the oviposition behavior of Ae. albopictus females.

Since yeasts produce VOCs only in the presence of carbon sources [113, 114], to assess the attractiveness or repellency of environmental yeasts at short-range distances, we devised a unique egg-count cage bioassay using a double-beaker system (DBS) that maintains yeast culture sterility, ensuring the integrity of emitted VOCs, and allows female mosquitoes to detect these VOCs and taste the water. Although such a bioassay provides valuable information on the oviposition response of Ae. albopictus mosquitoes, it does not allow us to separate olfactory from contact stimuli responses. Hence, the positive oviposition responses observed may be due to yeast-emitted volatile attractants, yeast-derived waterborne substances acting as stimulants, or a combination of both. To differentiate between the attractant/stimulant and repellent/deterrent effects of environmental yeasts, oviposition stimulation bioassays were systematically conducted in parallel, wherein females were exposed to water containing a specific yeast at a given concentration and thus stimulant or deterrent yeast-derived waterborne substances. This study provides the first evidence of the sole effect of stimulation by microorganisms [40]. Similar to observations made with bacteria [44], yeast concentrations strongly influence mosquito responses regarding attractiveness/stimulation and repellence/deterrence. Interestingly, the most significant impacts on female oviposition behavior were obtained at high cell density (3.10⁵ cell.mL⁻¹), which corresponds to the same concentration used to generate gnotobiotic larvae and that allowed observing differences in development. Our results demonstrate that both best development-recovering yeasts, R. mucilaginosa and A. pullulans, attract gravid females at short-range distances but do not stimulate oviposition at a high concentration. However, it is important to mention here that the adaptive bet-hedging trait of Ae. albopictus females, leading them to oviposit in multiple containers [115], tends to reduce the statistical significance of certain observations. Oviposition stimulation was only observed for two yeasts, H. uvarum and M. asiatica, at high cell concentration. Similar yeast oviposition stimulation in Diptera was observed only in Drosophila suzukii, where gravid females were shown to lay more eggs in blueberries inoculated with H. uvarum, as it enhances larval development [116]. Based on this observation, the negative impact of the presence of M. asiatica on oviposition responses of Ae. albopictus can only be attributed to a short-range repellency of this yeast species. In contrast, the presence of T. delbrueckii, one of the worst development-recovering yeasts, acts as both a deterrent and a repellent at high cell concentrations, preventing oviposition to a greater extent during choice experiments. Interestingly, the VOC profiles emitted by attractive yeasts (R. mucilaginosa and A. pullulans) are different from those of repellent ones (M. asiatica and T. delbrueckii). Attractive yeasts emit a lower number of VOCs compared to repellent ones. The two main volatile compounds (3-methyl- 1-butanol and 2-methyl- 1-butanol), which are shared by both attractive yeasts, have been demonstrated to be highly attractant microbial compounds for host-seeking mosquitoes [65, 70, 71], nectar-seeking mosquitoes [117], and gravid females searching for oviposition sites [69]. These two compounds are also emitted by both repellent yeasts, but at significantly higher concentrations, which are known to turns repellent [70, 71]. In addition, to these two compounds at high concentration, repellent yeasts produced specific compounds for some of with were demonstrated to induce an ovipositional repellency such as the ethyl octanoate [68]. According to the work of Aldridge et al. [113], our study shows that repellent yeasts emitted more complex VOC profiles, characterized by the presence of specific repellent substances and attractive volatiles that become repellent at high concentrations.

Conclusions

Taken together, our data highlight that gravid Ae. albopictus females rely on VOCs emitted by yeasts to select oviposition sites that enhance offspring performance. Waterborne yeasts that subsequently colonize larval gut and promote rapid development and better survival through improved nutritional intake attract gravid females, whereas yeasts with poor development-recovery profiles tend to act as deterrents, thereby preventing oviposition. The observed variations in mosquito development time could be linked to different levels of riboflavin production and nitrogenous waste recycling capacities of these environmental yeast species. Therefore, this study opens new perspectives for the investigation of trophic interactions between mosquitoes and their mycobiota. It underlines the importance of the supply of nitrogen molecules such as essential amino acids, proteins, or vitamins by mycobiota. In the future, metabolomic and fluxomic analyses using 15 N-labelled uric acid as a nitrogen source should be conducted on axenic and gnotobiotic mosquitoes, whether exposed to starved conditions or not, to further understand the functional role and significance of uricolytic/ureolytic yeasts in mosquito waste nitrogen recycling. More broadly, such analyses would provide insights into the ecological role of nitrogen metabolic interactions between mycobiota and mosquitoes. A deeper comprehension of these crucial mechanisms and their responsiveness to environmental factors need more attention, given their potential as candidates for the development of alternative and targeted mosquito control strategies. For example, a thorough characterization (including identification and concentration estimation) and utilization of the VOCs emitted by waterborne yeasts colonizing the larval gut and attracting gravid females could enhance the efficacy of oviposition traps.