All animal procedures were carried according to protocols approved by the Institutional Animal Care and Use Committee (IACUC) at the American University of Beirut (AUB), and all methods were carried out in accordance with relevant IACUC guidelines and regulations. Adults were continuously supplied with cotton pads soaked in a 10% sucrose solution and had access to water cups containing clean tap water. Larvae were fed on yeast for the first 24 hours, then on fish pellet food till pupation. Pupae were collected with a plastic pipette and placed in water cups inside plastic cages. A local strain of Aedes albopictus mosquitoes (originally captured from Sarba in the suburbs of Beirut, Lebanon) was maintained in the insectary at 28°C and 75% humidity using a 12:12 light/dark photocycle. This strain has been kept in the insectary for more than 10 years. Feeding was allowed on anesthetized mice, and eggs were collected on filter paper four days after the blood meal. Eggs were dried for two weeks before inducing hatching by immersion in aged tap water. A local strain of Culex pipiens mosquitoes (Makhoul strain captured from the AUB neighborhood, Beirut, Lebanon) was used in this study. This strain has been kept in the insectary since 2014. Egg rafts were collected once every generation and were allowed to hatch in tap water. D. melanogaster W¹¹¹⁸ strain and Relish^E20 were used in these experiments. Stocks were reared in 50 mL vials containing standard cornmeal agar food, prepared according to the D. melanogaster Bloomington Stock Center recipe. The main stocks were kept at 18°C, with the humidity set at 45%, under a 12:12 light/dark photocycle.

The bacterial strains used in this study were Serratia marcescens pGEN222, ampicillin-resistant and GFP labelled Escherichia coli, and the isolated Aeromonas hydrophila strain. Bacterial strains were cultured overnight in Luria-Bertani (LB) broth at 33°C with shaking at 180 rpm. The overnight cultures were centrifuged at 5,000 × g for 10 minutes, and the bacterial pellets were washed once in sterile phosphate-buffered saline (PBS, pH 7.4). The washed pellets were resuspended in PBS to obtain an optical density (OD₆₀₀) of 1. Serial 10-fold dilutions were then prepared, and 100 µL from appropriate dilutions were plated onto LB agar plates containing ampicillin, and colony-forming units (CFU) were counted to calculate the CFU/mL in the original suspension. All experiments were performed using the same protocol for both bacterial species.

For infection experiments, mosquitoes and Drosophila were aged between 5 and 7 days old. Flies were starved for 2 hours before their cups were supplemented with cotton pads soaked in 10% sucrose (for controls), or a bacterial suspension OD600 = 15. Flies were allowed to feed continuously until the guts were dissected for immunohistochemistry and Real-Time PCR, 24 hours after the start of treatment.

Female mosquitoes and Drosophila were starved for 2 hours before their cups were supplemented with cotton pads soaked in 3% sucrose for controls, or in a bacterial suspension (OD600 = 15) in 3% sucrose. Dead insects were counted at different time intervals. Each infection was done in triplicate with 15–20 females for each mosquito species and Drosophila per experiment, and the rates of survivals were plotted as function of time. For statistical analysis of the survival data, the log-rank (Mantel-Cox) test was performed.

Flies were cold anesthetized by placing the cups on ice, then transferred one at a time onto a glass slide in a drop of 1X PBS. The isolation of midguts was performed under a light stereomicroscope. Using fine forceps, the animal head was cut, and the mosquito abdomen was pulled from the posterior end until the midgut detaches. The isolated midguts were then placed in 1.5 ml Eppendorf tubes containing 1X PBS and kept on ice.

Isolated guts were fixed for 30 minutes using a 4% Paraformaldehyde (VWR, USA) solution in 1X PBS. This was followed by three 15-minute washes in PBS-Triton 0.1% to allow permeabilization of the guts. Blocking was then performed for 30 minutes by adding a solution of 1X PBS -Triton 0.1%-BSA 2%. After blocking, the primary rabbit α -PH3 antibodies (ABCAM, UK) were added (1:800 in 1X PBS-Triton 0.1%-BSA 2%) overnight at 4°C. Following another three 15-minute washes in PBS-Triton 0.1%, the secondary antibodies Alexa Fluor^® 594 (ABCAM, UK) were added (1:800 in PBS-Triton 0.1%-BSA 2%) for 3 hours at room temperature. After secondary antibodies removal, DAPI stain was applied at a concentration of 1:10000 for 2 minutes. Three final washes in PBS-Triton 0.1% were performed before mounting the guts on microscope slides in anti-fade medium (Immu-Mount, Thermo Scientific). Finally, coverslips were sealed with colorless nail varnish.

The slides prepared were observed under an inverted fluorescence microscope (Zeiss Axiovert 200, Source: AttoArc2 HBO 100 W) for counting proliferating cells, and an upright fluorescence microscope (Leica DM6 B) for image acquisition. Cell counts were analyzed using the Graphpad Prism software, and an unpaired t test was performed.

Mosquitoes were starved for 2 hours before being placed on sucrose alone (control) or sucrose supplemented with a suspension of ampicillin-resistant E. coli GFP (OD600 = 50) or A. hydrophila (OD600 = 15), for 24 hours. The bacterial suspension was then replaced by sucrose and the hemolymph from anesthetized mosquitoes (after clipping their proboscis) was collected into 1xPBS, 24 hours later. For mosquitoes co-infected with both A. hydrophila (OD600 = 15) and E. coli GFP (OD600 = 50), they were starved for 2 hours before being placed on sucrose supplemented with both bacteria: A. albopictus for 24 hours and C. pipiens for 36 hours. Finally, the mosquitoes were placed on sucrose solutions for another 12 hours. Dilutions in sterile LB of approximately 5 μl of hemolymph were plated on LB plates supplemented with ampicillin (100 μg/mL). The colonies were counted to estimate the approximate CFUs per mosquito.

Dissected guts were directly placed and homogenized in TRIzol^®. RNA was extracted using chloroform and precipitated with isopropanol according to the manufacturer’s instructions (Invitrogen). The extracted RNAs were quantified using a nanodrop spectrophotometer (Thermo), and 500 ng were retrotranscribed into cDNA (iScript Biorad) for each sample. Real-time PCR was performed in the presence of SYBR green (Qiagen) on 1/20 dilutions of the RT reactions, using a BIO RAD thermocycler (CFX 96 Real-timeSystem, C1000). Ct values for target genes were normalized to Rsp7 and compared to controls using the delta Ct method. A minimum of three independent experiments were averaged and unpaired t tests were performed. Primers were designed using the Primer3 online software.

A. albopictus :

C. pipiens:

D. melanogaster:

Genomic DNA was extracted from the bacterial isolate cultured on Luria Bertani agar using the Quick-DNA™ Fungal/Bacterial Miniprep kit (Zymo Research, Irvine, CA). Cells were lysed by mechanical disruption in ZR BashingBead™ lysis tubes. DNA was purified using spin-column technology and the Genomic DNA Clean & Concentrator™ kit (Zymo Research) according to the manufacturer’s protocols. NanoPhotometer N60 spectrophotometer (Implen, Munich, Germany) was used for the quantification of purified DNA.

DNA library preparation was performed using the Illumina DNA prep kit (Illumina, San Diego, CA), according to Illumina’s protocol: DNA underwent an initial tagmentation step followed by post-tagmentation cleanup. Then, tagmented DNA was amplified with addition of DNA/RNA unique dual (UD) indexes (Illumina) followed by library cleanup and elution. The concentration of DNA libraries were measured using the Qubit dsDNA High Sensitivity assay kit (Invitrogen, Waltham, MA) on a Qubit 4 fluorometer (Invitrogen). DNA libraries were pooled, denatured and diluted to 12 pM. PhiX Control v3 (Illumina) was added then the pooled library was sequenced using a MiSeq V2 Reagent Kit (500 cycles) on an Illumina MiSeq platform (Illumina) for 250×2 cycles and achieved a 50x coverage.

DNA library preparation was performed using the rapid barcoding kit (V14) (ONT, Oxford, UK) according to ONT’s protocol. Briefly, bacterial genomic DNA underwent barcoding, pooling then cleanup. Then, the resulting DNA libraries were eluted and their concentrations were measured on a Qubit 4 fluorometer (Invitrogen). Finally, the DNA libraries were loaded on an R10.4.1 flow cell (ONT) and sequenced on an Mk1B device (ONT).

Raw short and long reads were used to reconstruct the genome using Unicycler assembler (hybrid assembly) (PMID: 28594827). QUAST (PMID: 23422339) was used to assess the quality of the assembled genome and to extract the percentage of GC-content and the total length of the genome. To confirm the isolate species, KmerFinder tool (PMID: 24172157) was applied on the FASTA assembled genome. Antimicrobial resistance genes were identified using the Comprehensive Antibiotic Resistance Database (CARD) (PMID: 36263822). Genome annotation was performed using PROKKA annotation tool (PMID: 24642063). Circos plot was generated by converting the PROKKA GenBank file to XML format and plotted using cgview tool (PMID: 15479716). The genome data for A. hydrophila has these references (Bioproject number: PRJNA613441 - BioSample: SAMN48681286 - SRA: SRS25118285) and can be accessed via the following link: https://www.ncbi.nlm.nih.gov/biosample/48681286.

An unusual increase in larval death was observed in the laboratory reared C. pipiens colony. This coincided with the introduction of new larvae/water to the insectary. We suspected that this mortality could be due to a bacterial infection. For this reason, we took the guts of C. pipiens dying larvae and plated them on LB plates. Interestingly, one phenotypically distinct cultivable bacterium was highly predominant. 16S rRNA gene sequencing revealed that the isolated bacterium was Aeromonas hydrophila. This Gram-negative bacterial strain was resistant to several antibiotics including ampicillin ( Table 1 ). To determine whether A. hydrophila was responsible for C. pipiens mortality, adult female mosquitoes were starved before being allowed to feed on either sucrose solution (control), sucrose supplemented with Serratia marcescens, a Gram-negative bacteria used as a model to infect insects (Nehme et al., 2007), or sucrose supplemented with A. hydrophila (OD600 = 15 which corresponded approximately to 2.4 x 10¹⁵ CFU/mL for S. marcescens and to 1.05x10¹² CFU/mL for A. hydrophila). Mosquito survival was monitored and the results showed that feeding on A. hydrophila caused high mortality in C. pipiens, where approximately 50% of the mosquitoes were killed 30 hours post-infection ( Figure 1A ). This mortality was significantly higher than that caused by S. marcescens. To check whether A. hydrophila is also pathogenic to other disease vectors, the same experiment was performed on Aedes albopictus mosquitoes. The survival analysis showed that 50% of A. albopictus fed on A. hydrophila died only 24 hours post-infection, which is faster than the killing effect observed in C. pipiens ( Figures 1A, B ). In addition, as observed in C. pipiens, A. hydrophila killed A. albopictus at a higher rate than S. marcescens. These results indicate that A. hydrophila is a new mosquito pathogen.

We have previously shown that dividing cells exist in the midguts of both C. pipiens and A. albopictus, and that the number of these cells significantly increases following chemical or bacterial damage (Janeh et al., 2017, 2019). Taking into consideration that A. hydrophila causes death upon oral ingestion, we speculated that this bacterium is able to cause gut damage and induce cell division. To explore this possibility, adult females of the two mosquito species were starved before feeding on either sucrose (control), or sucrose supplemented with A. hydrophila (OD600 = 15). Mosquito guts were dissected 24 hours post-feeding and stained using anti-phospho-histone H3 protein antibodies (anti-PH3), a specific marker of mitotic cells (Taracena et al., 2018). In both C. pipiens and A. albopictus midguts, an increase in the number of dividing cells was observed after feeding on A. hydrophila ( Figures 2B–D ), in comparison to the control groups ( Figures 2A, C ). The quantification of the proliferating cells revealed statistically significant differences, as shown in ( Figures 2E, F ). Indeed, in A. hydrophila fed C. pipiens mosquitoes, an average of 16.58 ± 2.711 PH3 positive cells per midgut (n=19) was observed, as compared to the midguts of sucrose-fed mosquitoes, which showed an average of 3.100 ± 0.4286 (n=20) (E). Similarly, in A. albopictus,12.20 ± 1.976 dividing cells per midgut (n=20) were detected, when compared to controls which showed 2.000 ± 0.3297 dividing cells per midgut (n=24) (F). This implies that A. hydrophila damages the midgut, and that increased cell proliferation aims to repair the damage.

To further validate that A. hydrophila inflicts gut damage, female mosquitoes were allowed to feed on sucrose supplemented with either ampicillin-resistant, GFP labelled, E. coli or A. hydrophila, for 24 hours. Bacterial suspensions were replaced by sucrose solutions for another 24 hours to ensure that no bacteria were left in the proboscis of the mosquitoes. Then, hemolymph was collected from anesthetized mosquitoes and dilutions were plated on LB plates supplemented with ampicillin. No CFUs were detected in the hemolymph when E. coli was administrated to the two mosquito species. In contrast, in A. hydrophila fed mosquitoes, high numbers of CFUs were present in the hemolymph ( Figures 3A, B ; Supplementary Figure 1 ). This indicates that A. hydrophila was able to cross the gut barrier, resulting in leaky gut syndrome. Finally, when C. pipiens and A. albopictus were co-fed both bacteria simultaneously, E. coli was detected in the hemolymph of both mosquito species ( Figures 3C, D ). This finding provides further evidence that A. hydrophila disrupts gut integrity, allowing a non-pathogenic bacterium to reach the hemolymph.

Given the pathogenicity of A. hydrophila and its damaging effects on the mosquito’s guts, we checked whether the ingestion of these bacteria triggers the production of local AMPs in the guts of both species. For this reason, mosquito guts were dissected 24 hours after feeding on either sucrose alone or sucrose supplemented with A. hydrophila. Then, qRT-PCR was performed to assess the levels of Cecropin A1 and Cecropin A2, for A. albopictus and C. pipiens, respectively. The results showed that Cecropin genes were upregulated in both mosquito guts. CecA2 levels in C. pipiens were approximately 42 folds higher than in the non-infected control guts ( Figure 4A ). For A. albopictus, CecA1 levels were approximately 8 folds higher compared to the control guts ( Figure 4B ). This proves that the mosquito’s immune system is locally triggered by A. hydrophila.

Typically, Gram-negative bacterial infections induce the production of several mosquito AMPs via the Imd pathway (Zhang et al., 2017). Since A. hydrophila is a Gram-negative bacterium that promotes the upregulation of AMPs in mosquitoes ( Figure 4 ), we speculated that this induction is mainly controlled by the Imd pathway. Taking advantage of the availability of genetic tools and mutant strains in D. melanogaster, we wanted to test this hypothesis. For this purpose, D. melanogaster flies were fed on sucrose supplemented with A. hydrophila, and qRT-PCR results showed a significant increase in the levels of Diptericin transcripts in wild-type infected flies, as compared to the non-infected control group. However, feeding Rel mutant flies on A. hydrophila did not trigger any upregulation of Diptericin transcription in the guts ( Figure 5A ). These results highlight the importance of the Imd pathway in the gut immune response to A. hydrophila and confirm that AMPs are locally induced in the guts in an Imd-dependent manner.

In parallel, we checked whether feeding on A. hydrophila kills wild-type and Imd pathway mutant D. melanogaster. Wild-type and Rel mutant females were fed on either sucrose alone (control), or sucrose supplemented with A. hydrophila, and their survival was monitored. A. hydrophila caused the death of Rel mutants but had no killing effect on wild-type flies ( Figure 5B ). The fact that A. hydrophila exerts a killing effect solely on Imd mutant flies confirms that the Imd/Relish pathway is required for D. melanogaster survival to this bacterial infection.

In order to gain insights into the possible mechanisms by which the isolated A. hydrophila strain exerts its pathogenicity, we decided to perform complete genome sequencing. The data showed that this strain possesses a repertoire of key virulence-encoding genes, such as the cytotoxic/enterotoxic genes hemolysins, aerolysin, toxin RTX-I, and exotoxin A ( Table 1 ; Figure 6 ). In addition to toxin-encoding genes, type 3 secretion system (T3SS) secretin and a gene encoding chitoporin - a protein facilitating the breakdown of chitin (Suginta et al., 2013) - were also found in the A. hydrophila genome. Moreover, sequencing revealed a large number of antibiotic resistance genes, including β-lactamase genes (blaOXA-18, beta-lactamase, and a metallo-β-lactamase type 2), tetracycline and phenicols resistance genes, and several multidrug efflux pump genes (mdtA, mdtL etc.) ( Table 1 ; Figure 6 ). Other key genes involved in pathogenicity and antibiotic resistance are listed in ( Table 1 ). Altogether, these findings support the hypothesis that the isolated A. hydrophila strain is a mosquito pathogen, while highlighting its multidrug resistance potential.