The dynamics of the midgut microbiome in Aedes aegypti during digestion reveal putative symbionts

Abstract

Blood-feeding is crucial for the reproductive cycle of the mosquito Aedes aegypti, as well as for the transmission of arboviruses to hosts. It is postulated that blood meals may influence the mosquito microbiome but shifts in microbial diversity and function during digestion remain elusive. We used whole-genome shotgun metagenomics to monitor the midgut microbiome in 60 individual females of A. aegypti throughout digestion, after 12, 24, and 48 h following blood or sugar meals. Additionally, ten individual larvae were sequenced, showing microbiomes dominated by Microbacterium sp. The high metagenomic coverage allowed for microbial assignments at the species taxonomic level, also providing functional profiling. Females in the post-digestive period and larvae displayed low microbiome diversities. A striking proliferation of Enterobacterales was observed during digestion in blood-fed mosquitoes. The compositional shift was concomitant with enrichment in genes associated with carbohydrate and protein metabolism, as well as virulence factors for antimicrobial resistance and scavenging. The bacterium Elizabethkingia anophelis (Flavobacteriales), a known human pathogen, was the dominant species at the end of blood digestion. Phylogenomics suggests that its association with hematophagous mosquitoes occurred several times. We consider evidence of mutually beneficial host-microbe interactions raised from this association, potentially pivotal for the mosquito's resistance to arbovirus infection. After digestion, the observed shifts in blood-fed females' midguts shifted to a sugar-fed-like microbial profile. This study provides insights into how the microbiome of A. aegypti is modulated to fulfil digestive roles following blood meals, emphasizing proliferation of potential symbionts in response to the dynamic midgut environment.

Keywords: Elizabethkingia anophelis; Enterobacterales; digestion; hematophagy; whole-genome sequencing metagenomics.

Fig. 1.

The microbiome composition of female adults and larval stages. A) Compositional difference at the phylum level. The bins in the bar chart are scaled as percentages. B) Species-level variation of the top 50 phylotypes across adults and larvae. Bubbles display the relative abundance of microbial species on a quadratic scale based on the normalized number of assigned reads for each adult midgut or larval sample. Different species of Microbacterium sp. were collapsed to the taxonomic level of the genus and displayed in a separate scale (top-left) to enable the comparison between different taxa in the experimental groups.

Fig. 2.

Diversity profiles of mosquitoes' microbiome in different developmental stages and throughout 48 h of the digestion of different diets in adults. A) Boxplots showing the distribution of diversity indices and entropies. The global and pairwise significances are assessed with ANOVA and Wilcoxon's tests, respectively. B) Tridimensional scatterplot of NMDS using Bray–Curtis dissimilarities (stress = 0.07, model fit = 99%) of sugar and blood-fed groups collected 12, 24, and 48 h after feeding. Ellipses represent confidence intervals of the distances calculated from each group's centroids. ADONIS and ANOSIM tests support the contribution of each dependent variable (type of diet and time elapsed after feeding) in the distribution of microbial species. The different sizes of spheres are a function of the tridimensional perspective. C) Shared and unique microbial species between blood or sugar-fed adult mosquitoes. The intersections (connected dots) are coded by type of diet (B.12 h = Blood 12 h post-feeding; B.24 h = Blood 24 h post-feeding; B.48 h = Blood 48 h post-feeding; S.12 h = Sugar 12 h post-feeding; S.24 h = Sugar 24 h post-feeding; S.48 h = Sugar 48 h post-feeding), as are their correspondent bins in the histogram. The intersections sizes, representing the respective number of shared species between given groups, are displayed in the histogram. Bins and dots of Species shared between different diets are black.

Fig. 3.

Co-occurrence networks representing the interactions of the microbial community in adult mosquitoes fed with sugar (A) and blood (B). Graph vertices display microbial species, and edges represent their co-occurrences calculated with the Pearson correlation coefficient (r > 0.75; P < 0.05). Clusters marked with special characters (*, ○, and #) are similar in species composition.

Fig. 4.

The landscape of dominant taxa in relation to the microbial composition in the midgut of adult mosquitoes. A) Occurrence of 22 microbial species predictive of digestive states (diet and digestion time). The width of the ribbons indicates the relative abundance in the linear scale. Ribbons link microbial species to experimental groups (represented by the external circles; sugar or blood-fed) and are coded by the taxonomic order. B) Density estimates displaying the occurrence of reads identified in the taxonomic orders Flavobacteriales (top) or Enterobacterales (bottom) in the x-axis, with the mean of reads attributed to other microbial taxonomic orders in the y-axis, transformed to the log scale.

Fig. 5.

Functional classification of microbial reads in the midgut of adult females fed with blood and sugar. The heatmap was generated using z-score transformed values acquired in the metabolic pathway enrichment using SEED pathways functional classes in the metagenomes analyzed. Functional pathways are organized by hierarchical clustering. The columns represent individual metagenomes and are grouped according to its experimental group (per-group n = 10).

Fig. 6.

Functional diversity of the midgut microbiome in adult mosquitoes under different feeding regimes and digestion times. A) Biplot showing a PCoA of Bray–Curtis dissimilarities (PC1 = 58.2% vs. PC2 = 18.2%) displaying SEED pathways responsible for the sample ordination, coded by different diets and shapes, which represent hours post-feeding. B) Fuzzy set ordination (FSO) using a generalized linear model (GLM) to display the correlation between the functional dissimilarity matrix and the relative abundance of Enterobacterales. C) Differential occurrence of Enterobacterales species classified as predictors for the blood diet in log-relative scale and their individual correlations with the functional dissimilarities. The groups are composed of type of diet and time post-feeding (B.12 h = Blood 12 h post-feeding; B.24 h = Blood 24 h post-feeding; B.48 h = Blood 48 h post-feeding; S.12 h = Sugar 12 h post-feeding; S.24 h = Sugar 24 h post-feeding; S.48 h = Sugar 48 h post-feeding). D) Scatter plot using GLM to display the correlation between the principal explanatory coordinate (PC1) and the relative abundance of Enterobacterales. E) Boxplots showing the distribution of observed SEED pathways and their variability indices (Evar). Medians are indicated by the trend line, and the global and pairwise significances are assessed with ANOVA and Wilcoxon's tests, respectively.

Fig. 7.

Phylogenomic analyses of E. anophelis. A) ML tree including 690 Elizabethkingia genomes and genomes and Chryseobacterium glacei as an outgroup, highlighting two main clades within the E. anophelis species: Clade 1 and Clade 2. Branches with a mosquito icon denote strains of E anophelis associated with mosquito species. The MAG recovered in this study is highlighted. Below the circular tree, a condensed tree shows relationships within the genus Elizabethkingia. B) Details of relationships among E. anophelis genomes specifically associated with the host A. aegypti (genomes are indicated with an asterisk). The genome assembled in this study groups with clinical samples from humans within Clade 1 (tree at the top). The genome isolated from an Australian lineage of A. aegypti (tree at the bottom (51)); groups with samples from other mosquitoes within Clade 2 and is recovered as a sister group to samples isolated from humans.