Microbiome reduction prevents lipid accumulation during early diapause in the northern house mosquito, Culex pipiens pipiens

Abstract

The mosquito microbiome is critical to multiple facets of their biology, including larval development and disease transmission. For mosquitoes that reside in temperate regions, periods of diapause are critical to overwintering survival, but how the microbiome impacts this state is unknown. In this study, we compared the midgut microbial communities of diapausing and non-diapausing Culex pipiens and assessed how a reduced midgut microbiome influences diapause preparation. High community variability was found within and between non-diapausing and diapausing individuals, but no specific diapause-based microbiome was noted. Emergence of adult, diapausing mosquitoes under sterile conditions generated low bacterial load (LBL) lines with nearly a 1000-fold reduction in bacteria levels. This reduction in bacterial content resulted in significantly lower survival of diapausing females after two weeks, indicating acquisition of the microbiome in adult females is critical for survival throughout diapause. LBL diapausing females had high carbohydrate levels, but did not accumulate lipid reserves, suggesting an inability to process ingested sugars necessary for lipid accumulation. Expression patterns of select genes associated with mosquito lipid metabolism during diapause showed no significant differences between LBL and control lines, suggesting transcriptional changes may not underlie impaired lipid accumulation. Overall, a diverse, adult-acquired microbiome is critical for diapause in C. pipiens to process sugar reserves and accumulate lipids that are necessary to survive prolonged overwintering.

Graphical Abstract

Introduction

Microbiome studies have revealed great diversity in microbial communities of mosquitoes that vary by species, environment, sex, life-stage and diet (Dillon et al., 2004; Gusmao et al., 2007; Minard et al., 2013; Rani et al., 2009; Zouache et al. 2011). The use of axenic and gnotobiotic mosquitoes has indicated bacterial symbionts have the ability to influence various aspects of mosquito life history including disease transmission, development, digestion, metabolism, immunity and other physiological processes (Boissière et al., 2012; Chu and Mazmanian, 2014; Coon et al., 2016; Díaz-Nieto et al., 2016; Nieuwdorp et al., 2014; Scully et al., 2014; Shin et al., 2011; Strand, 2018). The relationship between the microbiome and mosquito diapause, a period of suspended development featuring drastic changes in mosquito biology (Denlinger and Armbruster, 2014; Denlinger and Armbruster, 2016), has not been studied and, thus, remains unclear.

Diapause is a hormonally driven developmental arrest that is initiated by specific environmental cues, including shorter photoperiod and colder temperatures (Sim and Denlinger, 2013; Denlinger and Armbruster, 2014; Spielman, 1974). This state is integral to the success of many pest and vector species, permitting the establishment of endemic populations in regions that would remain uninhabited due to cold, dry conditions during the winter or hot, dry conditions during the summer (Denlinger and Armbruster, 2014; Diniz et al., 2017; Lehmann, 2010; Musolin and Numata, 2003). In mosquitoes, diapause can occur at any developmental stage except pupae (Denlinger and Armbruster, 2014), and for adult females, this is characterized by increased storage of nutrient reserves by excessive feeding on sugar sources, decreased host seeking, and arrested ovarian development (Denlinger and Armbruster, 2014). Despite the importance of diapause for all insect systems, only few studies have examined microbiota changes during this developmental period and none of these studies have assessed the impact of bacterial associates in relation to dormancy-induced changes (Correal, 2016; Dittmer and Brucker, 2021; Ferguson et al., 2018; Ludwick et al., 2018; Lui et al., 2016; Wilches et al., 2018).

The microbiome of the field cricket, Gryllus veletis, has been shown to change concurrently with physiological changes while overwintering (Ferguson et al., 2018) and copepods have been shown to have distinct microbiomes for diapausing and non-diapausing states, though these communities were variable in nature (Almada, 2015). Additionally, when compared to age matched reproductive individuals (3 days post eclosion), diapause destined cabbage beetles, Colaphellus bowringi, harbor specific bacterial phyla and high community diversity (Lui et al., 2016). In the diapausing parasitic wasp, Nasonia vitripennis, Proteobacteria was the most abundant phylum, and a decrease in species richness and bacterial titer was noted as diapause progressed (Dittmer and Brucker, 2021). In vertebrates, the culturable microbiota of the 13-lined ground squirrel, Ictidomys tridecemlineatus, is reduced in both quantity and diversity in individuals undergoing dormancy in comparison to their summer counterparts (Carey and Assadi-Porter, 2017). Non-culture based techniques support this reduction as 13-lined ground squirrels under extended dormancy have the lower microbiome diversity, while individuals sampled two weeks after dormancy termination and subsequent refeeding had a more diverse microbiome. In fact, summer, early and late diapause microbiota community composition for 13-lined ground squirrels were identified by PCA as distinct groups (Carey and Assadi-Porter, 2017). With the exception of a few limited studies, it is unknown whether similar trends in microbiome diversity and compositions exist in invertebrates during the initiation and over the course of dormancy (Almada, 2015; Dittmer and Brucker, 2021; Ferguson et al., 2018; Heijtz et al., 2011).

No studies have directly examined if there is a diapause-specific microbiome and if specific bacteria influence the physiological changes associated with mosquito diapause (Lui et al., 2016; Mushegian et al., 2018). Studies that have directly examined mosquito dormancy-microbiome dynamics are lacking as axenic lines are difficult to generate, yield significant reductions in fitness early in larval development, and require treatment with toxic antibiotic cocktails that can interfere with other biological processes (Coon et al., 2014), with the exception of a recently developed system using transgenic bacteria and nutritional supplementation (Correa et al. 2018; Romoli et al. 2021). Our study examined diapause-microbiome interactions in the northern house mosquito, Culex pipiens pipiens. In this species, fourth instar larvae and pupae detect environmental changes such as short daylength and low temperatures which reliably signal impending adverse winter conditions (Eldridge, 1966; Sanburg and Larsen, 1973; Spielman and Wong, 1973). These cues initiate the diapause program which ultimately alters the behavior and physiology of the adult females, permitting overwintering success (Bowen, 1992; Mitchell, 1983; Mitchell, and Briegel, 1989). Our study examines diapause-microbiome interactions through the characterization of the midgut microbial community composition of adult female Culex pipiens pipiens using 16S rRNA sequencing. Diapausing mosquitoes with reduced culturable bacteria were then generated to assess if microbiome acquisition is critical to diapause in C. pipiens. The highlight from this study is that acquisition of bacteria occurs immediately after adult female emergence and these bacteria are essential for rapid lipid accumulation, a critical component in diapause preparation. The reduction of the microbiome in diapause-destined females led to lower lipid accumulation and reserves with females dying in a few weeks rather than surviving months.

Methods

Mosquito husbandry

The Culex pipiens colony, originally collected from Columbus, Ohio, USA (September 2000, Buckeye strain), were conventionally reared at 25°C at 15 h:9 h L:D, 70–75% relative humidity (RH). Egg rafts were obtained from 4–5 week old females fed a diet of chicken blood (Pel-Freez, Rogers, Arizona) via blood feeder (Hemotek, Blackburn, United Kingdom). Upon emergence, larvae were held at a density of 250 individuals in 18cm x 25 cm x 5 cm containers and fed a diet of ground fish food (Tetramin, Melle, Germany). Diapausing (D) individuals were held at a 9 h:15h, L:D cycle while nondiapausing (ND) C. pipiens were held at a 15 h:9 h L:D cycle. All were held at 18°C, ≈70–75% RH and were provided 10% sucrose and DI water ad libitum. All mosquitoes used for this study were aged 14–17 days post-ecdysis.

Gut microbiome sample collection

Only females (14–17 days old) were used in this study, and they were collected by aspiration then cold anesthetized at 4°C for 5 minutes. Anesthetized females were rinsed with sterilizing solution (10% ethanol, 5% sodium hypochlorite, 85% sterile DI water) for five minutes to eliminate surface bacteria. Dissections of adult females were performed in sterile 1x PBS. Midguts were removed using sterile instruments and pooled in groups of 5 into 1.5 ml tubes of 70% ethanol. Ten samples were collected for ND and D mosquitoes.

Diapause verification

Diapause state was confirmed by measuring primary ovarian follicle length, as size differences alone cannot sufficiently be used to distinguish between D and ND females (Spielman and Wong, 1973). Ovary dissections were performed in 1.0% PBS and stained with methylene blue for visualization. Dino-Lite Edge and Dinocapture software (AnMo Electronics Company, Hsinchu City, Taiwan) was used to measure length. Twenty ovaries from 10 individuals were dissected for each group with three primary ovarian follicles measured for each sample (measured three times). Diapause status was assessed via primary ovarian follicle length as described in Spielman and Wong (1973). The presence of secondary follicles was also assessed and were only present in non-diapausing individuals. A one-way ANOVA was performed followed by Tukey HSD to determine the statistical differences between the groups using RStudio v1.2.1335 (R Studio Teams, 2020) and R v4.0.3. This version of R was used to perform all subsequent analyses.

DNA extraction and library preparation

A DNEasy Kit (QIAGEN, Valencia, CA, USA) was used to prepare the DNA for the microbiome studies. PCR amplification of DNA was conducted similarly to previous studies (Dahlhausen et al., 2018; Yan et al., 2015.) Bacterial specific primers 27F (5’-AGAGTTTGATCMTGGCTCAG) and 1492R (5’-GGTTACCTTGTTACGACTT) to ensure sufficient DNA quality for downstream sequencing and once DNA quality was assessed, each sample was amplified in triplicate using the specific primers 515F (5’-GTGCCAGCMGCCGCGGTAA) and 806R (5’-GGACTACHVHHHTWTCTAAT) which target the 16S rRNA V4 hypervariable region (Caporaso et al., 2012). The three replicates were then pooled in equimolar concentrations for each sample to enrich the evenness of read representation across samples. The Illumina MiSeq platform was used to generate 25M overlapping, paired 251bp reads for downstream analyses. Sequencing was performed at the Center of Microbiome Science at Ohio State University.

Data analysis of 16S rRNA sequences

Illumina sequences were processed using MOTHUR (V1.38.1) on the Ohio Supercomputer Center Cluster to remove low quality reads (<27 Phred score), those less than 270bp long and to assign operational taxonomic units (OTUs) (cluster.split, method = furthest) (Schloss et al. 2009). To ensure sufficient coverage, sequences were rarefied to 9,500 reads/sample and used in the abundance-based coverage estimator (ACE). Multiple refraction curves were generated and collated within QIIME to estimate species richness in each sample type by conservatively approximating the number of OTUs present. The generated ACE rarefaction curves confirmed that each sample type achieved sufficient sampling depth to continue with downstream analyses. Beta diversity analysis was performed using the 9,500 sample reads to generate distance matrices using both Bray-Curtis (phylogenetically-independent) and PD-whole/UniF (phylogenetically-dependent) methods. The created matrices were then employed in Multi Response Permutation Procedure (MRPP) and Nonmetric Multidimensional Scaling (NMS) analyses. These were used to determine whether statistical significance between the two groups microbial composition existed. Power analysis was performed with a Wald’s test (Mattellio et al., 2016) and indicated that additional replicates are unlikely to indicate differences between the ND and D microbiomes. Indicator species analyses were performed as previously described (Dufrêne and Legendre, 1997). Files were uploaded to NCBI’s sequence read archive (Accession: PRJNA742018).

Low bacterial load C. pipiens through pupal surface sterilization

To generate mosquitoes with reduced bacterial load, diapausing and non-diapausing larvae were reared in groups of 180 individuals in the manner as described above. We did not focus on the complete elimination of the microbiome, rather the goal was a significant reduction of the microbiome in adult females, as the presence of many different species of viable bacteria is sufficient for mosquito growth and development (Coon et al., 2016; Díaz-Nieto et al., 2016; Valzania et al., 2018a; Valzania et al., 2018b). Specifically, pupae were collected, and a series of five-minute rinses was used to reduce surface bacteria. The reconstituted and LBL pupae were subjected to the following rinse protocol: sterile deionized (DI) water, 2% sodium hypochlorite, sterile deionized (DI) water, where untreated mosquitoes were rinsed 3x with DI water. Pupae were then relocated into a sterile crystallization dish (50×35 KIMEX®, Kimble) with sterilized (LBL), or unsterilized (Untreated, Reconstituted) colony water. LBL mosquitoes were placed into autoclave-sterilized metal cages (Bioquip, Rancho Dominguez, California), with ad libitum sterile sugar and water and placed in a sterilized incubator (Percival Incubator Inc., Perry, Iowa). To ensure the sugar and water would last two weeks without replacement, three dental rolls were placed in a large crystallization disk and submerged in either sugar or water (≈100ml) until only the tips remained uncovered. Untreated and reconstituted females were provided identical provisions except all components remained unsterilized. All cages were held within environmental conditions identical to those necessary to generate D pupae as described previously. All mosquitoes remained sealed within the incubators to prevent bacterial contamination until collection at two weeks of age for experiments.

To determine the success of the LBL and reconstituted treatment protocols mosquitoes were surface sterilized for 5 minutes in 70% ethanol, rinsed with sterile water, then crushed with a sterile plastic mortar in sterile, and serially diluted (1:1000) in PBS solution. One hundred μl was spread across a Luria-Bertani plate (Fisher Scientific) using a sterilized homemade glass cell spreader. The plates were incubated at room temperature and colony forming units were counted at 48 hours post plating. Each treatment consisted of 5–7 biological replicates with three technical replicates per sample. A Kruskal-Wallis ANOVA followed by a Wilcox pairwise comparison was performed to determine treatment effect. A robust yet incomplete bacterial depletion of the “sterile” treatment group (LBL) in comparison to the reconstituted and control groups prompted our subsequent studies examining survival and physiological differences of reduced microbiome on diapause. Importantly, these treatments were conducted on multiple independent groups over the course of a year, suggesting LBL establishes consistently. All boxplots were made in R using ggplot, as such, the lower and upper hinges represent the 25th and 75th quartile respectively. The whiskers depict no more than 1.5x the inner-quartile range while dots represent outliers and hold true for all boxplots in this study.

Lipid quantification

Lipid quantification and all subsequent assays focused on LBL, reconstituted and untreated groups reared under diapausing conditions. A chloroform/methanol extraction protocol was modified from Bligh and Dyer (1959) and Van Handel (1985) based on recent mosquito studies (Benoit and Denlinger, 2007; Benoit et al., 2010; Hagan et al., 2018) to determine lipid accumulation. Briefly, females from each treatment group were collected at two weeks of age and frozen at −7°C. Total mass was measured upon death using a Cahn Electrobalance (C.Y. Scientific, LLC) prior to placement in a Blue M drying oven (Electric Company). Mass was recorded daily until the recorded weight remained unchanged for two consecutive days. A Kruskal-Wallis ANOVA followed by a Wilcox pairwise comparison was performed to determine whether mass differed significantly between the groups. To reduce variability each sample consisted of two mosquitoes. Samples were homogenized with a Benchmark BeadBlaster 24 (Sayreville, New Jersey) in chloroform/methanol solution. Lipid levels were determined using the vanillin-spectrophotometric assay (Van Handel, 1985) with eight biological samples per group and three technical replicates per sample. A Synergy H1 Hybrid Reader (Gen5 2.01 software) was used to assess the colorimetric status of each sample and the resulting data was compared using a one-way ANOVA followed by a Tukey HSD.

Total carbohydrate and glycogen quantification

Whole mosquitoes were homogenized as before in STE buffer. A proportion of each sample was used in the modified total carbohydrate and glycogen assays (Benoit and Denlinger, 2007; Benoit et al., 2010; Hagan et al, 2018). For the glycogen assay, methanol was added to a sample aliquot, vortexed for 2 minutes, and centrifuged at 4°C at 5,000g for fifteen minutes. The supernatant was discarded, the remaining pellet was washed with methanol, and centrifuged at 4°C, 5,000g for five minutes. The remaining supernatant was discarded, and the pellet was resuspended in 200μl of 25% ethanol. The 40μl aliquot used for the total carbohydrate assay was brought up to 200μl with 25% ethanol. Anthrone was then added to both the glycogen (n=12, 3 technical replicates) and total carbohydrate (n=12, 3 technical replicates) samples and heated at 100°C for ten minutes and a colorimetric assessment was accomplished with a Synergy H1 Hybrid Reader (Gen5 2.01 software). Absorbance values were then compared to a standard curve to determine the carbohydrates and glycogen levels in each sample. Results were compared using a one-way ANOVA followed by a Tukey HSD to determine group differences.

qPCR analysis of genes associated with lipid metabolism

Individual mosquitoes were homogenized and extracted using 1 ml of chilled (4°C) TRIzol^® reagent and cleaned-up using GeneJET RNA Cleanup and Concentration Micro Kit (Thermoscientific, Waltham, Massachusetts). RevertAid First Strand cDNA Synthesis Kit (Thermofisher Scientific) was used to synthesize complementary DNA with each reaction consisting of 1ng of RNA, primers and Master mix (50 ng Oligo(dT)₁₅ primers dNTPs, 5mM MgCl₂ and M-MuLV RNase H⁺ reverse transcriptase) and 1ng of RNA. qPCR reactions were performed using KiCqStart SYBR Green qPCR ReadyMix (Sigma Aldrich, St. Louis, MO, USA), 300 nM forward and reverse primers (Table S2), cDNA diluted to 1:20, and 9.5μl nuclease free water. Primer3 (Untergasser et al., 2012) was used to design primers. Eco Real-time PCR System (Illumina) was used to analyze each qPCR run, with reaction protocol that included polymerase activation for 3 min at 95°C preceding 40 cycles of denaturation (10 sec at 95°C), and annealing/extension (30 sec at 55°C). A melt curve analysis was utilized from 55°C to 95°C increasing 0.5°C increments every 15 seconds, each sample ran in duplicate and the average C_q value determined. Expression was determined by fitting sample C_q values onto a standard curve A master sample which was utilized for the standard curve was created by mixing the same quantity of each sample type (LBL, Reconstituted, Untreated) prior to dilution (1:10, 1:20, 1:200, 1:1,000, 1:2,000) as described previously (Larinov et al., 2005). Ten biological replicates with three technical replicates were performed per group. The expression of each gene was assessed using a one-way ANOVA.

Results

Microbiome characterization indicates minimal differences in the bacterial content between ND and D females

OTU composition was not significantly different in terms of species richness or beta diversity between D and ND mosquitoes using MRPP (df=15, p=0.258) and NMS (R²<0.45). Overall, 136 OTUs were identified with 84 being shared between both groups, 42 exclusively 275 associated with diapause, while only 10 OTUs were ND specific. Most samples in this study were dominated by one or two taxa at the family level which composed roughly 90% of the total reads.

The bacterial families, Acetobacteraceae, Anaplasmataceae, and Halomonadaceae were indicative of diapause whereas Flavobacteriaceae, Streptococcaceae best represented nondiapause. Despite the presence of indicator species, high inter-individual variability of midgut microbial communities within each sample type (D and ND) contributed to a lack of significant differences between ND and D mosquitoes (Figure 1). The most abundant phyla in diapausing samples included Proteobacteria (90.7%), with a total of 15 phyla representing less than 1% of the remaining reads. ND samples displayed a similar trend with Proteobacteria representing 81.3% of the total reads with 14 phyla making up less than 1% of the total microbial composition. Through the comparison of all of our samples we were able to identify a “core” microbiome, which are sequences that are present in every sample, regardless of diapause state. Eighteen OTU’s were identified as part of the core microbiome including sequences associated with the phyla Actinobacteria, Bacteroidetes, Firmicutes and Proteobacteria. Of those identified, Proteobacteria was the most heavily represented phyla with 14/18 of the identified “core” families residing within it. The six most abundant familial OTUs were Flavobacteriaceae, Acetobacteraceae, Anaplasmataceae, Aeromonadaceae, Enterobacteriaceae, and Halomonadaceae (Table 1.). Additionally, most but not all of the identified bacterial reads were gram negative, with an average of 2.64% the reads representing gram positive bacteria (0.2%–19.7%).

Figure 1. Midgut family composition of diapausing and nondiapausing C. pipiens.

Relative abundances of bacterial families associated with each sample (N=10 per group). Midgut microbiome familial distribution for diapausing (D) and nondiapausing (ND) C. pipiens. Families within a single phylum are assigned shades of the same color. No significant differences in diversity or composition were identified (MRPP, D=0.258; NMS, R^2=0.415, 0.288, 0.201)

Table 1.

Phylum-level classification of bacterial communities in diapausing and nondiapausing Culex pipiens.

		Diapausing		Nondiapausing
Pylum	Order	#OTUs	Relative Abundance (%)	#OTUs	Relative Abundance (%)
Actinobacteria	Actinobacteria	14	0.61	10	1.32
Bacteriodetes	Bacteroidia	4	1.2	3	0.046
	Flavobacteriia	2	5	1	14.4
Firmicutes	Bacilli	6	0.63	6	1.65
Proteobacteria	Alphaproteobacteria	19	49.14	15	18.04
	Betaproteobacteria	9	0.23	8	0.31
	Gammaproteobacteria	18	41.3	17	63.48
Other taxa		51	1.87	28	0.72

Functional assays confirm that microbiome acquisition is critical for mosquito diapause preparation.

Colony forming unit and primary ovarian follicle length assessments indicate that the treatments were successful and therefore differences observed in subsequent functional assays were not for a lack of diapause status. All functional assays were performed on all three different treatment groups; low bacterial load, reconstituted, and untreated (control). These procedures yielded females with drastically different culturable bacterial levels, where the LBL lines had a nearly 1000-fold reduction that was partially recovered in the reconstituted females (Kruskal-Wallis ANOVA (Kruskal-Wallis chi-squared = 24.18, df = 5, p-value = 0.0002005) followed by a Wilcox pairwise comparison (D=diapause, N=nondiapause, D_Untreated-D_Reconstituted: p=0.260, D_Untreated-D_LBL: p=0.020, D_Reconstituted-D_LBL: p=0.020, N_Untreated-N_Reconstituted: p=0.232, N_Untreated-N_LBL: p=0.020, N_Reconstituted-N_LBL: p=0.022, D_Untreated- N_Untreated: p=1, D_Reconstituted-N_Reconstituted: p=0.823, D_LBL-N_LBL: p=0.804) (Figure 2A). Importantly, the ovarian follicle length was reduced in all D mosquito groups compared to ND mosquitoes, independent of treatment status (Figure 2B, Table S1).

Figure 2.

Comparison of CFU counts and ovarian follicle lengths by treatment group. A. Colony forming units (CFU) (df=5, D=diapause, N=nondiapause). B. Primary ovarian follicle length of diapausing and ND LBL, reconstituted, and untreated mosquitoes (df=5 N=5–7, 3 technical replicates per sample, Table S1). All diapausing groups retained ovarian follicles under 0.08mm suggesting that all successfully entered diapause whereas all ND groups had primary ovarian follicles above 0.1mm suggesting that treatment does not impact diapause or ND status in C. pipiens.

Survivorship was assessed for each treatment group at two weeks of age (Figure 3). LBL mosquitoes had the lowest survivorship when compared to both the untreated and reconstituted groups (Figure 3). Survival of reconstituted mosquitoes was significantly lower than control mosquitoes (ANOVA- Tukey HSD, N=7, p=0.006, df=2, f=26.5) but was drastically increased when compared to the LBL mosquitoes (p=0.001, df= 2, Figure 3). Dry mass was impacted by microbiome status, where LBL groups were smaller than both reconstituted (Kruskal-Wallis, Wilcox pairwise comparison (p=0.028, df=2) and control groups (Figure 4A, p=2.3e-05, df=2). Even though the reconstituted group was significantly smaller than the control group (p=0.029, df=2) representing a middle ground, possibly due to delayed bacterial colonization.

Figure 3. Survival Assessment.

Survival was assessed for groups of 75 untreated, LBL and reconstituted mosquitoes (df=2, N=7). Reconstituted mosquitoes displayed significantly lower survivorship than untreated mosquitoes (p=0.006) though significantly lower than LBL mosquitoes (p=0.001).

Figure 4.

Mass and nutritional reserves for untreated, LBL and reconstituted mosquitoes. A. Dry mass (g) of individual mosquitoes in each treatment, normalized to mean untreated weight for visualization, LBL mosquitoes weighed significantly less than their untreated (p=0.0004) and reconstituted (p=0.05) counterparts (df=2, N=144). B. Percent lipid in relation to total dry mass. LBL mosquitoes accumulated significantly less lipids than reconstituted (p=0.009) or untreated mosquitoes (p=0.008) (df=2, N=8). C. Glycogen in relation to dry mass. No significant differences were noted between the groups. D. Total carbohydrates in relation to dry mass. LBL mosquitoes had significantly more total carbohydrates present than mosquitoes from the untreated (p=0.02) or reconstituted groups (p=0.04) (df=2, N=8).

The reduction in dry mass suggested that nutrient reserve levels of diapausing mosquitoes may be shifted when the microbiome is reduced. When controlled for mass, LBL mosquitoes had a significant reduction in lipid reserve levels, which was nearly 50% lower compared to control or reconstituted mosquitoes (Figure 4B, ANOVA-Tukey HSD, N=7, p=0.008, p=0.009, df=2, f=7.3). Glycogen levels were not different between samples, but LBL mosquitoes had higher levels of total carbohydrates (Figure 4C–4D, ANOVA-Tukey HSD, p=0.04, p=0.02, df=2, f=4.881). This suggests an inability to convert ingested carbohydrates into fat reserves, which is a critical aspect of diapause preparation in C. pipiens (Bowen, 1992; Edman et al., 1992; Magnarelli, 1980; Nayar, 1978).

Expression of several genes have been linked to lipid accumulation during diapause in C. pipiens. These include fatty acid synthase-1 (Robich and Denlinger, 2005), an enzyme involved in the synthesis of fatty acid from aceytl-coA and NAPDH, fatty acid binding protein (Sim and Denlinger 2009), which is responsible for transporting fatty acids and, Dihydrolipoyl acetyltransferase which links glycolysis and lipid metabolism to the citric acid cycle and has been shown to increase longevity in the yeast, Saccharomyces cerevisiae, and though it has not been directly linked to diapause in C. pipiens it could be involved in the observed phenotypes in the study (Easlon et al., 2008). No significant differences in transcript expression were noted for the genes investigated in the LBL mosquitoes compared to control and reconstituted groups (Figure 5). This result suggests that the underlying mechanism for impaired survival and reduced lipid accumulation in LBL mosquitoes is not due to transcriptional changes of these diapause associated genes.

Figure 5.

Expression of genes important for nutrient mobilization via qPCR. A. Fatty acid synthase-1 (df=2, N=5, 3 technical replicates, f=0.525). B. Fatty acid binding protein (df=2, N=5, 3 technical replicates, f=0.865). C. Dihydrolipoamide S-acetyltransferase (df=2, N=5, 3 technical replicates, f=0.504). None of the genes investigated were significantly different between groups possibly due to variable success of LBL and reconstitution treatments.

Discussion

The prominence of mosquitoes as disease vectors, has prompted a focus on cataloging the mosquito microbiota of different sexes, species, localities, developmental stages, and in relation to feeding and disease status (Coon et al., 2014; Hegde et al., 2014; Hughs et al., 2014; Osei-Poku et al., 2012; Segata et al., 2016; Wang et al., 2011). Mosquitoes reared in a field setting tend to harbor a more diverse microbiota than their lab reared counterparts including a higher proportion of gram-positive bacteria (Boissière et al., 2012). The mosquito microbiota is generally acquired by larvae from the aquatic environment in which they are reared (Coon et al., 2014). This is supported by the fact that emerging first instar larvae do not have extracellular bacteria in their digestive tract and the microbiome is composed of bacteria similar to the water in which they were reared (Coon et al., 2014). Much of the bacteria is expelled from fourth instar larvae during pupation, leaving the adult digestive tract fairly sterile upon emergence (Moll, 2001). However, Wolbachia and some Asaia species do appear to form stable associations with mosquitoes throughout development into the adult stage and are transmitted maternally (Damiani et al., 2008; Favia et al., 2007; Jiggins, 2017; McMeniman et al., 2009). Neither Wolbachia nor Asaia were found in our samples, which is unsurprising as our studies focused on the gut microbiome where these bacteria tend to reside in lower numbers when found. This lack of Wolbachia and Asaia suggests that the bacteria acquired in adult females during this study were predominantly from the environment after adult emergence. Based on our ability to eliminate most of bacteria within adults by surface sterilizing the pupae and ensuring emergence under sterile conditions, it is most likely that C. pipiens adult females acquire most of their microbiome when ingesting non-sterile water following adult emergence.

The microbiome has been implicated as important to mosquito physiology including development and vector capabilities, even though there is high inter-individual bacteria variation in most mosquito groups (Coon et al., 2014; Osei-Poku et al., 2012; Valzania et al., 2018a; Valzania et al., 2018b). In concordance with previous studies, this study identified high inter-individual bacterial variation which partially contributed to the non-significant difference between diapausing and non-diapausing C. pipiens bacterial community composition. This supports a general trend for mosquitoes where the presence of a wide variety of bacterial species are sufficient as a microbiome that can support mosquito biological processes. The notion that a specific microbiome is not required for mosquito physiological processes is likely driven by functional overlap in microbial species or possibly a conserved host response to bacterial presence (Coon et al., 2016; Degli Esposti and Martinez Romero 2017; Valzania et al., 2018a; Valzania et al., 2018b). The identified phyla in this study, which include Proteobacteria, Actinobacteria, Firmicutes and Bacteroidetes, are commonly found in mosquitoes with Proteobacteria generally dominating the community (Boissière et al., 2012; Duguma et al., 2015; Gimonneau et al., 2014; Lindh et al., 2005; Minard et al., 2013; Pidiyar et al., 2004; Rani et al., 2009; Zouache et al., 2010). Gram-negative bacterial species are especially dominant for lab reared mosquitoes with few gram-positive species, whereas field caught mosquitoes may harbor an even smaller portion of gram-positive bacteria (Boissière et al., 2012). Bacterial diversity is low within an individual midgut yet is high when comparing diversity between individuals of the same species (Boissière et al., 2012) and is supported by our study which showed high variation between samples, both the ND and D females.

In some systems, diapause has been shown to reduce microbial diversity to a core subset, as seen in both stink bugs (Nezara viridula) and marine copepods (Arias-Cordero et al., 2012; Datta et al., 2018; Medina et al., 2018). Few studies to date have investigated the role of the microbiome during diapause and other overwintering strategies of insects (Almada, 2015; Dittmer and Brucker, 2021; Lui et al., 2016). In the cabbage beetle, a positive correlation was found between diapause, Proteobacteria and Firmicutes while revealing a negative correlation with Bacteroidetes (Lui et al., 2016) while in diapausing wasps Proteobacteria was once again found to be dominant, while bacterial titers and species richness decreased as diapause duration increased (Dittmer and Brucker, 2021). Diapausing and ND C. pipiens in our study both displayed a high level of microbial composition diversity. These states were sampled early in diapause (2 weeks) so it remains possible that community diversity could change as diapause progresses from weeks to months. Changes could possibly occur due to physiological changes such as those involved to suppress water loss within the female as diapause progresses or more simply, due to introduction of new bacteria through water intake, as diapausing females still continue to imbibe some water throughout the course of diapause (Benoit and Denlinger, 2007).

Despite a lack of a diapause-specific microbiome, our study identified that a viable microbiome plays an important role in diapause preparation and survival of dormancy-destined females. In specific, failure to establish the microbiome leads to a substantial reduction in survival after two weeks, which is likely due to C. pipiens inability to convert carbohydrates into lipid reserves, a critical aspect of entering and surviving diapause (Robich and Denlinger 2005, Zhou and Miesfeld 2009). Putative links between diapause and the microbiota have been made, specifically expressional changes in the insulin-like peptide signaling pathway (Mushegian and Tougeron, 2012). Insulin-like peptide (ILP) is integral in the C. pipiens diapause program as environmental cues increase its expression allowing it to bind its receptor, leading to downstream expressional changes of Forkhead transcription factor (FOXO) and juvenile hormone (JH) (Sim and Denlinger, 2013). These genes are responsible for induction of diapause in insects, nematodes and fish (Sim and Denlinger, 2013; Woll and Podrabsky, 2017) and show expressional changes in response to bacteria (Dionne et al., 2006; Lee et al., 2019; Shin et al., 2011; Zheng et al., 2017).

The genes previously linked to increased lipid accumulation (Robich and Denlinger 2005, Sim and Denlinger 2009), investigated in our study were not differentially expressed. A transcriptome of adult A. aegypti only revealed 170 differentially expressed genes associated with the axenic state (Hyde et al., 2020), none of which overlap the genes investigated in the current study. Several were metabolic in nature however, leading the authors to suggest that the microbiota play a minor role in adult mosquito nutrient metabolism (Hyde et al., 2020). Additionally, the microbiota does not greatly impact adult mosquito fecundity or longevity suggesting that the microbiota, outside of diapause may not impact adult mosquitoes as greatly as the aquatic life stages (Romoli et al., 2021). Therefore, it is possible that the bacteria do not impact these host genes but instead may be involved directly in metabolism of ingested carbohydrates early during diapause preparation and allow for lipid accumulation or provide other factors, such as specific vitamins or other micronutrients, that support lipid accumulation in C. pipiens. We did not assess if lipid accumulation is altered in ND mosquitoes when the microbiome is eliminated, but there is nearly a 60-fold increase in lipids in D females compared to ND counterparts which indicates that this process is not as drastic or critical for ND individuals (Rozsypal et al., 2021).

Based on previous studies in Drosophila and humans, there is support that specific metabolite are produced by microbiome components, such as short chained fatty acids, that are critical to lipid accumulation and likely trigger specific peptide hormones involved in fat metabolic processes (Everard and Cani, 2014; Shen et al., 2013; Wong, 2016). In mammals, the microbiota produces short chain fatty acids including acetate, butyrate and propionate (Wong, 2016). These are sensed by intestinal cells, leading to the excretion of signaling molecules which modulate host metabolism and food intake (Kasubuchi et al., 2015; Psichas et al., 2015). Similarly, in fruit flies, metabolites produced by the microbiota signal to enterocytes, which in the intestine play a role in digestive, absorptive and innate immune functions (Wong, 2016). Additionally, bacteria are known to alter hypoxia-induced transcription factor signaling in the digestive system which can also greatly affect host physiology (Valzania et al., 2018a; Valzania et al., 2018b). Therefore, it is likely that the microbiome is providing signals in mosquitoes which may be even more important during diapause preparation in which a notable metabolic shift occurs within the host mosquito. This is supported by previous studies in mosquito larvae, where lipid accumulation is perturbed by bacterial elimination (Valzania et al., 2018a; Valzania et al., 2018b). In axenic larvae, neutral lipids accumulate in the nutrient-absorbing enterocytes and lipid reserves do not increase in the fat body (Valzania et al., 2018b), it is unlikely a similar mechanism could be occurring in the current study since carbohydrate processing is most likely arrested in the gut based on our observations.

Several factors important in determining whether an organism will enter diapause are influenced by the presence of bacteria in mechanisms similar to those noted when examining axenic mosquito larvae growth defects (Valzania et al., 2018a; Valzania et al., 2018b). We did not note any expressional changes in genes associated with diapause preparation metabolism (Kang et al., 2016; Robich and Denlinger, 2005), suggesting that the early diapause transcriptional response is not altered even though the key goal of lipid accumulation is not occurring. Future studies will be necessary to confirm the specific mechanisms underlying this dysfunctional lipid accumulation in diapausing reduced microbiome C. pipiens.

Conclusion

The midgut microbiome of C. pipiens is highly variable and does not change significantly in response to diapause status. Despite lack of variation, LBL mosquitoes with their microbiome depleted were unable to obtain the diapause phenotype characters of increased lipid reserve, a larger body size and increased longevity, but still displayed arrestment of oocyte development, a hallmark of C. pipiens dormancy. The fact that diapausing LBL females died within weeks, rather than months, suggests that the microbiome is important for survival of overwintering adults. This study provides evidence that despite not having a single bacterial species or even specific phyla associated with the diapause state in C. pipiens, the midgut microbiota is essential for successful diapause preparation and will likely prevent females from surviving until diapause termination in spring.