Background

The microbiome profoundly influences many phenotypes in a host. In mosquitoes, much of the focus in this area has centered on how bacterial microbiota play an important role in mosquito biology, particular in relation to vector competence or how bacteria can be exploited for vector control [1,2,3,4]. Many of these studies have examined how the bacterial microbiome influences mosquito traits important for vectorial capacity, including growth, reproduction, and blood meal digestion [5,6,7,8,9]. While these studies provide convincing evidence that microbes can influence traits important for vectorial capacity of mosquitoes [9, 10], the role of the fungi on mosquito biology is understudied and less well understood.

Several studies have characterized the fungal microbiome in different mosquito species using culture-dependant and -independent methods [11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27]. In general, these studies indicate the majority of fungal taxa that colonize mosquitoes are within the Ascomycota and Basidiomycota phyla [16, 19, 22, 28,29,30,31]. Shotgun metagenomic sequencing of Cx. pipiens, Culisetra incidens, and Olchelerotatus sierrensis uncovered a diverse array of fungal taxa in mosquitoes, but only two fungal genera, Cladosporium and Chromocliesta, were present in multiple mosquitoes [13]. Amplicon sequencing of bacterial and fungal microbiomes of Ae. aegypti found fewer eukaryotic taxa compared to bacterial, although the majority of eukaryotic reads in mosquitoes were designated to gregarine parasites, rather than fungal species [18]. While our appreciation of the fungal community is expanding, we have a poor understanding of its functional relavance or interactions with other members of the microbiome.

Fungal community composition and abundance appear to be influenced by several factors, similar to their bacterial counterparts [28]. Aspects that appear to affect fungal microbiota include habitat, host species, diet, and pathogen infection [16, 22, 23, 30, 31]. For instance, in the tree hole mosquitoes Ae. triseriatus and Ae. japonicus, both blood feeding and La Cross virus infection were shown to reduce fungal richness [17]. Like the bacterial microbiome, mycobiome community structure varies between mosquito species and habitats [16,17,18,19, 27] and fungal diversity is seen between mosquito tissues [19, 22, 30]. While it is evident that mosquitoes possess diverse fungal taxa, sequence based assessment of the fungal microbome can be challenging due to inadvertent amplification of the host. To overcome these challenges, methods to selectively amplify the fungal sequences at the expense of host sequence have been accomplished [11].

Fungi can influence mosquito phenotypes that have important ramifications for vectorial capacity. For instance, the presence of a common mosquito-associated Ascomycete fungus Penicillium chrysogenum in the midgut of An. gambiae enhances the mosquito’s susceptibility to Plasmodium infection [30]. Similarly, Talaromyces fungus increased Ae. aegypti permissiveness to dengue virus infection [31], while Beauverua bassiana reduces vectorial capacity of Ae. albopictus to Zika virus [32]. Other studies have examined the effect of yeast on mosquito development and survival, which are traits that could influence vectorial capacity. Supplementation of Saccharomyces cerevisiae or native yeast strains supported the development of Cx. pipiens [22], although there was a strain-specific effect on the overall growth and development [12]. Recent advances in rearing approaches have enabled mono-association infections to be undertaken whereby a single (or group) of microbe(s) is inoculated in to germ-free L1 larvae to enable mosquito growth and development [7, 23, 33, 34]. While studies using mono-axenic rearing approaches have focused on the influence of the bacterial microbiome on their ontogeny [7, 23, 35,36,37,38], the ability of fungal isolates native to mosquito fungi have not been evaluated using this innovative mosquito rearing approach.

To address these gaps in our knowledge regarding fungal-host association in mosquitoes, we used high-throughput sequencing to examine the fungal microbiome of Ae. aegypti, Ae. albopictus and Cx. quinquefasciatus mosquitoes caught in the field or reared in the lab. Using gnotobiotic infection approaches, we reared these mosquitoes mono-axenically with fungal isolates to examine colonization and effects on mosquito development. Our results provide insights into the role of the environment on the composition and abundance of the fungal microbiome, microbe-microbe interactions in mosquitoes, and the influence of native fungal isolates on mosquito life history traits.

Material and methods

Mosquito samples and high-throughput sequencing. We used the DNA from Ae. aegypti, Ae. albopictus, and Cx. quinquefasciatus samples either collected from the field or reared in the lab for high-throughput sequencing to examine the fungal microbiome [35]. The field collection of mosquitoes were followed as described previously[35]. Briefly, all the field collected mosquitoes were trapped using Biogents Sentinel (BG) or Harris County gravid (G) traps, which selectively collect host-seeking or gravid female mosquitoes, respectively [39,40,41]. We have analysed 11 each of lab and field Ae. aegypti, 9 each field and 10 lab Ae. albopictus, and 11 each of lab and field Cx. quinquefasciatus.To characterize the fungal microbiome of these mosquito species, the internal transcribed sequence (ITS) was sequenced. The region spanning ITS2 was sequenced according to the Illumina metagenomic sequencing protocol. Libraries were prepared following the amplicon protocol which includes the use of indexes from the Nextera XT Index Kit v2 (Illumina). Library preparation was done according to Illumina amplicon protocol (Illumina) (Additional file 7: Table S1, ITS primers) [42]. Libraries were sequenced on the MiSeq System with the MiSeq Reagent Kit v3 (Illumina, Catalog No. MS-102–3003). All MiSeq runs were performed with a run configuration of 2 × 251 cycles for PNA blocker PCR samples (see next section) and 1 × 501 cycles for all other samples. To enable the calculation of error-rate metrics and to increase nucleotide base diversity for more accurate base-calling, all libraries were spiked with 5% PhiX Control v3 (Illumina, Catalog No. FC-110–3001). The NCBI Genbank accession number for the raw sequencing data reported here is PRJNA999749.

PNA blocker PCR with microbiome samples. To block host amplification, PNA blocker was designed and synthesised (PNA Bio, USA). The PCR was performed with 1 µM of each primer (Additional file 7: Table S1), 2 µM PNA,1X KAPA master mix (NEB) and 50 ng of template DNA. The PCR conditions were as follows: 3 min at 95 °C for initial denaturation; 30 cycles of 30 s at 95 °C, 30 s at 70 °C, 30 s at 55 °C, 30 s at 72 °C, 5 min at 72 °C, then 30 s at 70 °C clamping step for PNA. The product was digested with SphI which cuts the fungal ITS amplicon but not the region in mosquitoes (Additional file 1: Fig. S1). The PCR products were purified and sequenced as described above.

Bioinformatic analysis. To identify the presence of known fungi, sequences were analyzed using the CLC Genomics Workbench 12.0.3 Microbial Genomics Module. Reads containing nucleotides below the quality threshold of 0.05 (using the modified Richard Mott algorithm) and those with two or more unknown nucleotides were excluded and finally the sequencing adapters were trimmed out. Reference based OTU picking was performed using the UNITE v7.2 Database [43]. Sequences present in more than one copy but not clustered in the database were placed into de novo OTUs (97% similarity) and aligned against the reference database with an 80% similarity threshold to assign the “closest” taxonomical name where possible. Chimeras were removed from the dataset if the absolute crossover cost was three using a kmer size of six. Additionally, OTU’s were reclassified using BLASTn 2.7.1 + [44] against the nt nucleotide collection database. The blast results were used for taxonomic categorization of the origin of ITS sequences between those from the host, metazoan, and fungi. Alpha diversity was measured using Shannon entropy (OTU level), rarefaction sampling without replacement, and with 100,000 replicates at each point.

Isolation and identification of fungal isolates from mosquitoes. Homogenates of five adult female mosquitoes were from Ae. albopictus (Galveston strain) and Cx. quinquefasciatus (Galveston strain) were plated on Brain Heart Infusion (BHI) agar (BD Difco), Yeast Peptone Dextrose (YPD) agar (BD Difco), malt extract agar (BD Difco), Yeast Malt agar (BD Difco), and Sabouraud Dextrose Broth (BD Difco). Colonies were purified by streaking a colony on a fresh agar plate and incubated at 30 °C for 2 days and transferred to 22–25 °C for 4–5 days until colonies to appeared before proceeding with culturing in the respective media. Five colonies from each growth media type were screened based on the colony characteristics (Additional file 8: Table S2). Genomic DNA was isolated and PCR used to amplify ITS as the way to identify the isolated fungi. The PCR was completed using 1 × reaction buffer (NEB), 200 µM dNTPs, 1 µM of each primer (Additional file 7: Table S1), and 1U of Taq DNA polymerase (NEB). The PCR conditions were an initial denaturation of 1 min, 30 s at 95 °C, then 35 cycles of 30 s 95 °C, 30 s at 55 °C, 30 s 72 °C and a final extension of 5 min at 72 °C. The PCR products were separated on agarose gels before Sanger sequencing with ITS3 and ITS4 primers. Sequences were analysed using the BLASTtn NCBI database.

In vitro growth analysis of fungal isolates. The growth of Rhodotorula mucilaginosa, Candida oleophila, S. cereviciae and Lachancea thermotolerance were undertaken by culturing in liquid YPD medium at 28 °C. Overnight cultures of fungal isolates were diluted 1:100 in YPD medium and were grown at 28 °C for 48 h. The growth was assessed by recording OD at 600 nm at 0, 2, 4, 8, 24 and 48 h (Additional file 2: Fig. S2). The assay was done in five replicates and repeated twice.

Mosquito mono-association infection with fungi. Mono-association (MA) rearing was used to assess the colonization of fungi in absence of a natural microbiome. Axenic L1 larvae were generated as described previously [7, 35]. The 45 axenic larvae (N = 15 per flask) were infected with 1 × 10^7 cfu/ml fungi R. mucilogenosa, C. oleophila, L. thermotolerance, S. cerevisiase and C. neteri bacteria. Fungi R. mucilogenosa, C. oleophila, L. thermotolerance are the culturable fungi present in the lab colonies of Ae. albopictus and Cx. quinquefasciatus mosquitoes and C. neteri is the abundant culturable bacteria found in the laboratory Ae. aegypti mosquito colony. All the procedures related to mono-association infection of mosquitoes were undertaken in a sterile environment and sterility was verified by plating larval water on LB agar plates. The mono-associated larvae were fed with sterile fish food at the concentration of 20 ugm/ml. The axenic L1 larvae without microbes have slow growth rates and do not reach pupation. For the mono-associated infections, larvae were maintained in the T75 flask till they reached pupae stage and the pupae were transferred to a container to eclose into adults. The adults were maintained on sterile 10% sucrose solution untill they were harvested for CFU quantification. The infection in the T75 flasks were maintained till day 16 by this time most of the larvae had pupated. To quantify their fungal or bacterial symbionts loads, we surface sterilized L4 larvae with 70% ethanol for 3 min and 2 times 1X PBS for 5 min. Larvae were then homogenized and plated on YPD agar for fungi and LB agar for bacteria. After incubation for 2 days, colonies were counted. Five larvae from each flask (total N = 15) were tested for CFU analysis. Both bacterial and fungal quantification were done from the same larval and adult sample. Time to pupation and the percentage of L1 larvae to reach adult stage were recorded to determine the effect of fungi on mosquito growth and development. Time to pupation was recorded as the day when pupae were collected from the flask post infection. The number of adults emerged from each flasks (N = 15 larvae per flasks) were recorded and the percentage of L1 that emerged as adults was calculated. To assess the interkingdom interactions between native microbiome and fungi, Ae. aegypti mosquitoes were also infected with R. mucilogenosa, C. oleophila, L. thermotolerance, S. cerevisiase either in mono-association or in conventional rearing settings. The bacteria C. neteri was used as a control for inter-microbial interactions which we described in our previous study [35]. To assess the interkingdom interactions between fungi and bacterial microbiome, we did the fungi and bacteria infection of mosquito with and without native microbiome. All the procedures relating in in the interkingdom interactions study were followed as did for the mono-association infection.

Fungal qPCR analysis. We used qPCR to determine the fungal load in Ae. aegypti, Ae. albopictus, and Cx. quinquefasciatus using 18S rRNA primers and probes [45]. PCRs consisted of 50–100 ng of DNA, 1 µM of. each primer (Additional file 7: Table S1), 225 nM of the TaqMan probe (Additional file 7: Table S1) 1% formamide, 1X Platinum Quantitative PCR SuperMix-UDG w⁄ROX (Invitrogen Corp.) and molecular biology grade water. We used the following PCR conditions: 3 min at 50 °C for UNG treatment, 10 min at 95 °C for Taq activation, 15 s at 95 °C for denaturation, and 1 min at 65 °C for annealing and extension for 40 cycles. We used host S7 or actin gene specific primers as endogenous control. The relative fungal copies were compared to host genome copies.

Statistical analysis. All statistical analysis of the CFU and mosquito growth analysis data were done using GraphPad Prism software. First, we performed normality test to assess the normal distribuction of the data. Here, we performed D'Agostino & Pearson test and Shapiro–Wilk tests respectively to assess. If our data sets passed both of these tests, we then assumed Gaussian distribution and equal SD and further analysed data by ordinary one-way ANOVA (Tukey’s multiple comparision test). We also performed Brown-Forsythe and Welch ANOVA test to assess the homogeneity of variance. The prevalence data were analysed by a Fisher exact test with 2 × 2 matrix where number of infected and uninfected for each treatment was compared with every other treatment for each mosquito species. P-value 0.05 was considered significant.

Results

Fungal microbiome sequencing and analysis

We sequenced the ITS2 region from field-collected and lab-reared mosquitoes to characterize their fungal microbiome. Across all samples, we obtained 9,310,520 reads and recorded, on average, 155,175 reads per mosquito sample. However, similar to other high throughput sequencing (HTS) studies characterizing the fungal microbiota in eukaryotic hosts [11, 46, 47], our attempts were hampered due to the amplification of host or metazoan sequences. This was most pronounced for Ae. aegypti where about 99% of the reads were nonfungal derived (Fig. 1), while Cx. quinquefasciatus and Ae. albopictus had an average 21% and 8% fungal reads, respectively. To block nonselective amplification in Ae. aegypti samples, we employed a PCR clamping approach using a PNA blocking probe. While we saw evidence of suppression of host ITS amplification in PCR-based assays (Additional file 1: Fig. S1) and a large reduction of host ITS reads (38% reduced to 0%), this did not result in a substantial increase in fungal reads (Fig. 1; a change from 0 to 1%). PNA blockers have been previously used to exclude Anopheles 18S rRNA reads when sequencing [11] but we saw little difference in the fungal reads, mainly due to an increase in amplification of metazoan sequences as a percentage of the overall reads in the PNA blocker treatment (Fig. 1). We speculated that these Ae. aegypti lacked significant fungal communities and therefore we saw non-specific amplication of host DNA in this sample as there was a lack of fungi ITS template to amplify. To further address this we completed qPCR to estimate total fungal density in lab-reared mosquitoes using universal fungal primers. Here we saw significantly reduced fungal loads in Ae. aegypti compared to the other two mosquito species (Additional file 3: Fig. S3; ANOVA with Kruskal–Wallis test, P < 0.0001). Given the evidence for reduced fungal loads in Ae. aegypti, our attention then focused on examining the fungal microbiome of Cx. quinquefasciatus and Ae. albopictus mosquitoes (Additional file 9: Table S3). Despite the fungal reads comprising a relatively small proportion of the overall reads in Cx. quinquefasciatus and Ae. albopictus, rarefaction curve analysis indicated that our sampling depth was sufficient to observe the majority of fungal OTUs in the majority of indivudal mosquitoes (Additional file 4: Fig. S4).

Fig. 1
figure 1

Average of percentage of reads from ITS2 sequencing: Average of percentage of ITS2 sequencing reads from Ae. aegypti, Ae. albopictus and Cx. quinquefasciatus. Ae. aegypti samples were sequencing again with the addition of a PNA blocker targeting the host ITS sequence. To generate average reads per species 11 laboratory reared and 22 field collected samples from Ae. albopictus and Cx. quinquefasciatus were analysed to generate average reads per species. For Ae. aegypti 22 field collected samples were assessed while 16 field collected samples were amplified with the PNA blocker

Full size image

Fungal richness, diversity, and community structure.

We examined the species richness of the fungal microbiome in Cx. quinquefasciatus and Ae. albopictus by evaluating the difference between field-collected mosquitoes caught in either the gravid (G) or BG traps. When comparing within each species, we saw no significant difference in the Shannon diversity between traps (BG or gravid traps, Additional file 5: Fig. S5A; Tukey’s multiple comparison test, P > 0.05) for either species nor did we see significant differences between traps for beta diversity estimates (Additional file 5: Fig. S5B and S5C; Bray–Curtis dissimilarity, best stress value = 0.23 for both species; adonis2, P > 0.05 in both cases). As such, we combined these mosquitoes for further analyses and considered them “field-collected”. When comparing between mosquito species, we found that the field-collected Ae. albopictus had significantly elevated Shannon diversity compared to Cx. quinquefasciatus (Fig. 2A; Tukey’s multiple comparison test, P < 0.05), but no difference was seen between species in lab-reared mosquitoes. Similarly, there was no significant difference in Shannon diversity when comparing within a species between environments (i.e. field vs lab; Fig. 2A). This was also true for the number of OTUs with no difference within a species but Ae. albopictus had significantly more OTUs compared to Cx. quinquefasciatus regardless of environment (Fig. 2B; Tukey’s multiple comparison test, P < 0.05). We then examined the community structure of the fungal microbiome using Bray–Curtis NMDS analysis. Overall, the fungal microbiome clustered distinctly with both species and environment identified as significant factors.. This was predominantly driven by the field samples which, when analyzed separately, were significantly different between each species (Fig. 2C; Bray–Curtis dissimilarity, stress = 0.18; adonis2, P = 0.0009, R2 = 0.065), but when mosquito species were reared in a common lab environment the fungal microbiomes were similar (Fig. 2D; Bray–Curtis dissimilarity, stress = 0.12; adonis2, P = 0.0559, R2 = 0.088). When comparing field-caught and lab-reared mosquitoes, both Cx. quinquesfaciatus (Fig. 2E; Bray–Curtis dissimilarity, stress = 0.24; adonis2, P = 0.0029, R2 = 0.072) and Ae. albopictus (Fig. 2F; Bray–Curtis dissimilarity, stress = 0.20; adonis2, P = 0.0009, R2 = 0.095) had distinct microbiomes, indicating environmental factors contributing to the diversity of fungal communities.

Fig. 2
figure 2

Alpha and Beta diversity analysis of fungal microbiome. Shannon entropy measuring abundance of fungal microbiome in Ae. albopictus and Cx. quinquefasciatus A. Number of operational taxonomic units represents species richness of fungal microbiome in Ae. albopictus and Cx. quinquefasciatus B. Non-metric Multi-dimensional Scaling (NMDS) plots showing Bray–Curtis dissimilarities of fungal OTUs (C-F): the fungal community structure in the field collected samples C and laboratory reared mosquitoes D were compared between the two species. The fungal community structure between different environments (lab v field) was compared within Cx. quinquefasciatus E and Ae. albopictus F. Numbers inside the graph indicates the p-value between groups. The field samples includes mosquitoes were collected in G and BG traps

Full size image

Next, we examined the taxa present in each mosquito species. There were 244 fungal OTUs in mosquitoes, of which 76 and 97 were present above a 0.1% threshold in Cx. quinquefasciatus and Ae. albopictus, respectively (Additional file 9: Table S3). While the majority of taxa were unidentified (Fig. 3 and Additional file 6: S6), of the known OTUs, most were classified within the Ascomycota and Basidiomycota phyla, (Additional file 6: Fig S6), which was similar to other studies [16, 18, 48]. Saccharomycetaceae were the most abundant in Ae. albopictus while the Malasseziaceae where dominant in Cx. quinquefasciatus (Fig. 3A and Additional file 6: S6). Unsurprisingly, considering the beta diversity analysis, the microbiomes of the lab-reared mosquitoes were comparable, however when examining the diversity between individuals, there was variation (Additional file 6: Fig S6), which is also a feature of the bacterial microbiome [35]. In many cases, OTUs that were dominant in one individual were absent or at low abundances from others (Additional file 5). 

Fig. 3
figure 3

Relative abundance of fungal taxa. The relative abundance of fungal OTUs at family level with 0.01% cut-off between Ae. albopictus and Cx. quinquefasciatus field and laboratory samples

Full size image

Fungal isolates colonize and supports mosquito growth in mono-association

Microbes are required for mosquito growth and development [7, 37]. Eukaryotic microbes such as the model yeast, S. cerevisiae, are known to promote larval growth [23], however it is not clear how symbiotic fungi affect mosquito growth and development. We cultured and identified symbiotic fungi from Ae. albopictus and Cx. quinquefasciatus. To determine if these native fungal taxa colonize mosquitoes and supported growth of their hosts, we reared mosquitoes in a mono-association using four fungal species. Three of these species, C. oleophila, R. mucilagenosa, and L. thermotolerans were native mosquito isolates while the model yeast S. cerevisiae was used as a positive control. The growth of mosquitoes infected with fungi was also compared to a native bacterial isolate, Cedecae neteri, which is a common bacterium present in our lab-reared Ae. aegypti and complements growth of mosquitoes in a mono-association [35]. When colonizing germ-free mosquitoes, fungi were more effective at colonizing Ae. aegypti and Cx. quinquefasciatus (Fig. 4A and C, circles, Fisher’s exact test, P > 0.05) having high prevalance rates in adults while prevalance was reduced for all microbes in Ae. albopictus (Fig. 4B, circles, Fisher’s exact test, P < 0.05). Intrigingly, colonization rates of 100% were observed in both larvae and adults of Ae. aegypti for all microbes (Fig. 4A, circles). Additionally, the native fungal densities were comparable to that of the symbiotic bacteria C. neteri (Fig. 4A, Dunn’s multiple comparition test, P < 0.05). Both C. oleophila and R. mucilaginosa poorly infected adult Ae. albopictus despite infecting larvae (Fig. 4B, Dunn’s multiple comparition test, P < 0.05). Similar to Ae. aegypti, the native fungal infection prevalence was 100% in larvae while there was no significant difference in the infection prevalence of microbes in adults (Fig. 4C, circles, Fisher’s exact test, P > 0.05) although variable infection densities were observed in both life stages (Fig. 4C, Dunn’s multiple comparisonn test, P < 0.05).

Fig. 4
figure 4

Fungal colonization of axenic mosquitoes. The scattered plot shows CFUs/mosquito of Ae. aegypti A, Ae. albopictus B and Cx. quinquefasciatus C larvae and adults. The CFU data were analysed by Kruskal–Wallis Test with a Dunn’s multiple comparisons test. The circle above each scattered plot shows prevalence of infection for that treatment. Prevalence data were analysed by Fisher exact test. Letters above each scattered plot and prevalence circle indicate significance between the treatments. For all statistical analysis P < 0.05 was considered significant. Sample size was N ≥ 10 for larvae and N ≥ 5 for adults – each dot on the graph represents an individual mosquito. The dotted horizontal line inidicates threshold detection limit

Full size image

Mosquito development assay

Given bacterial microbiota can influence development we also determined the life history traits associated with mono-association infection. In Ae. albopictus, mosquitoes infected with L. thermotolerans had reduced times to pupation compared to the other native fungal microbes, while there was variability in pupation times in Ae. aegypti but no differences in Cx. quinquefasciatus between microbes (Fig. 5A-C). We also measured the percentage of L1 larvae that reached adulthood in these mono-associations. In general, Ae. albopictus had higher rates of mosquitoes reaching adulthood for all microbes, while the percentage of Culex mosquitoes emerging as adults was below 40% for all fungal taxa (Fig. 5D-F). In Ae. aegypti mosquitoes, R. mucilaginosa infections had significantly different effects compared to the other two native fungi, while in Ae. albopictus its effects were only significantly different from S. cerevisiae (Fig. 5D &E, Tukey’s multiple comparision test, P < 0.05).

Fig. 5
figure 5

Life history traits in mono-association infections. Time to pupation of each species in mono-axenic associations A–C. Data were analysed by one-way ANOVA with Dunn’s multiple comparision test. Growth was determined by percentage of L1 larvae to reach adulthood D–E. Data were analysed by one-way ANOVA with Tukey’s multiple comparision test. None of the axenic larvae pupated and hence, the percentage to adulthood are zero for all axenic controls

Full size image

Fungal infection in presence and absence of native bacterial microbiome

We have previously shown that colonization of symbiotic bacteria is influenced by members of the native bacterial microbiome [35, 49]. Given the ability of fungi to infected Ae. aegypti in a mono-association but the lack of fungal reads in field-collected mosquitoes, we speculated that bacteria may inhibit fungal infection. To determine if cross kingdom interactions influenced fungal colonization, we infected fungi into conventionally reared or axenic Ae. aegypti, which either possessed or lacked their native bacterial microbiome, respectively. Strikingly, we did not recover any fungal CFUs in either larvae or adults when the mosquitoes were grown conventionally in the presence of a native microbiome, however in stark comparison, fungal isolates were able to effectively colonize germ-free mosquitoes (Fig. 6, Mann Whitney Test, P < 0.05). Intringuingly, the reduced colonization capacity of fungi of conventionally reared mosquitoes was seen in both larvae (Fig. 6A, Mann Whitney Test, P < 0.05) and adults (Fig. 6B, Mann Whitney Test, P < 0.05). In agreement with our previous study [38], the positive control, C. neteri also was more effective at colonizing germ-free mosquitoes compared to their conspecfic’s that possessed a conventional microbiome, however this effect here was more subtle compared to the almost complete blockage of fungi seen when mosquitoes had bacterial microbiota.

Fig. 6
figure 6

Fungal colonization in presence or absence of a native microbiome. R. mucilaginosa, C. oliophila, L. thermotolerans were incolulated into conventionally (C) reared Ae. aegypti mosquitoes that possessed their native microbiota or axenic germ-free mosquitoes to create a mono-association (MA). CFUs were quantified in A L2-L3 larvae and B three to four day old adults. The bacterium C. neteri was used as a positive control. A contamination control was undertaken by rearing axenic larvae without infection. These mosquitoes did not develop confirming sterility. The CFU/mosquito data were analysed by unpaired t test and prevalence data by a Fishers exact test. Asterisks (*) indicates significance, while ns denotes non-significant

Full size image

Discussion

We characterised the fungal microbiome of Ae. aegypti, Ae. albopictus and Cx. quinquesfaciatus collected from different environments. Sufficient fungal reads were obtained from Cx. quinquesfaciatus and Ae. albopictus to evaluate their fungal microbiomes. In these species, we found the fungal composition varied substantially between species and environments. These findings were similar to other reports whereby environment has been shown to be a major determinant of fungal microbiome composition [16, 18, 19]. At the individual level, there was variability in the composition of fungal taxa within mosquitoes. Of the known taxa, Malassezia, Saccharomcetales, and to a lesser extent, Candida were fungi that were frequently seen in either species and other studies have identified these genera in mosquitoes suggesting they may commonly infect these vectors [13, 16, 29, 50, 51].

Strikingly, our sequencing data suggest that the fungal microbiome of Ae. aegpyti is dramatically reduced as we only observed a small fraction of fungal reads in these mosquitoes. Initially we speculated that the low number of fungal reads was due to preferential amplification of the host, and as such we used blocking PNA oligonucleotides to suppress host reads, in a similar fashion to other studies [11, 46, 47]. Despite our blocking primer reducing host ITS reads, there was no significant increase in the number of fungal reads, but rather an increase in off target host reads, indicating that these field caught mosquitoes lacked fungi at an amplifiable level. Supporting this finding, qPCR analysis of lab-reared Ae. aegypti found significantly reduced fungal densities compared to Ae. albopictus and Cx. quinquesfaciatus. Together these data indicate that these Ae. aegypti mosquitoes have a reduced fungal microbiome. Further studies are required to determine if this is consistent across other lab-reared or field collected Ae. aegypti mosquitoes.

Little is known about the capacity of members of the fungal microbiome to colonize their mosquito host. Although our sequencing data indicate Ae. aegypti lacked a robust fungal microbiome, specific taxa were able to colonize when infected into germ-free mosquitoes. The ability of germ-free mosquitoes to harbour fungi suggests that the reduced fungal load that we saw in Ae. aegypti by sequencing or qPCR was not due to an incompatibility between the fungal species and the mosquito, but rather due to microbial incompatibility. To empirically test this, we compared infection of fungal taxa in germ-free compared to conventially reared mosquitoes and found fungi infected the mosquitoes in absence of native microbiome. While the microbiome can be composed of a variety of microbes, we speculated that bacterial microbiota were interfering with fungal infections. We have previously identified several bacterial co-occurrence interactions in these mosquitoes and experimentally validated inter-bacterial interactions in co-infection studies [35, 49, 52]. However, fungal-bacterial co-occurance has not been exclusively investigated. Several other studies identified fungal and bacterial communities co-existing from individual mosquitoes, but these were not in Ae. aegypti [13,14,15]. More generally, the influence of bacteria-fungi interactions on colonization has been observed in diverse microbial systems including the soil microbiome, and the microbiota of livestock and humans [53,54,55,56], so further investigations of these interactions in mosquitoes are warranted.

Several studies have shown that the bacterial microbiome is required for mosquito growth and development [7, 38, 57]. Other eukaryotic microbes can also facilitate development including the model yeast S. cerevisiae and insect cells [23, 58]. Here we show that native fungal species that associate with mosquitoes also have the ability to support mosquito growth and development. We did observed developmental variation between fungal microbes and between mosquito species, however, S. cerevisiae had similar developmental rates compared to previous studies [23, 58]. Interestingly, we saw variability between replicates in terms of S. cerevisiae infections. These replicate experiements (Fig. 4A [S. cerevisiae had high prevalance and density] and Fig. 6 [lack of S. cerevisiae infection]) were performed on the same mosquito lines but reared at different institutions. Our most recent analysis of microbiome from these mosquito lines reared at these different insectaries revealed they possessed significantly different microbiomes [59] and given our findings regarding fungal-bacterial interactions, it is tempting to speculate that differences in the native microbiota were responsible for the variation in S. cerevisiae colonization. These findings will be important to confirm given that S. cerevisiae is being investigated for novel vector control strategies [60].

Conclusions

In summary, here we showed that Ae. albopictus and Cx. quinquefasciatus harbor fungal taxa as part of their microbiome, but, Ae. aegypti appear to lack a robust mycobiome. The lack of fungal taxa in Ae. aegypti appears to be due to cross kingdom microbial interactions. Despite this, when the bacterial microbiome is removed, fungi can infected these mosquitoes and support their growth. Together, our findings have shed a light on an understudied aspect of the mosquito microbiome and shown that native fungal symbionts influence mosquito biology.