Elizabethkingia anophelis MSU001 Isolated from Anopheles stephensi: Molecular Characterization and Comparative Genome Analysis

Elizabethkingia anophelis MSU001, isolated from Anopheles stephensi in the laboratory, was characterized by matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-ToF/MS), biochemical testing, and genome sequencing. Average nucleotide identity analysis revealed 99% identity with the type species E. anophelis R26. Phylogenetic placement showed that it formed a clade with other mosquito-associated strains and departed from a clade of clinical isolates. Comparative genome analyses further showed that it shared at least 98.6% of genes with mosquito-associated isolates (except E. anophelis As1), while it shared at most 88.8% of common genes with clinical isolates. Metabolites from MSU001 significantly inhibited growth of E. coli but not the mosquito gut symbionts Serratia marcescens and Asaia sp. W12. Insect-associated E. anophelis carried unique glycoside hydrolase (GH) and auxiliary activities (AAs) encoding genes distinct from those of clinical isolates, indicating their potential role in reshaping chitin structure and other components involved in larval development or formation of the peritrophic matrix. Like other Elizabethkingia, MSU001 also carried abundant genes encoding two-component system proteins (51), transcription factor proteins (188), and DNA-binding proteins (13). E. anophelis MSU001 contains a repertoire of antibiotic resistance genes and several virulence factors. Its potential for opportunistic infections in humans should be further evaluated prior to implementation as a paratransgenesis agent (by transgenesis of a symbiont of the vector).

Recent studies have shown that clinical human specimens including wound swabs, sputum, urine, body fluids, and blood frequently reveal the presence of E. anophelis [18,19].Infections with E. anophelis pose a significant risk to individuals who are already ill, immunocompromised, or at age extremes [4,18,20].Its causative diseases include neonatal meningitis, catheter-related bacteremia, and many others, leading to high mortality rates, ranging from 18% to 70% [6,20].Moreover, a recent outbreak in the Upper Midwest region of the United States, specifically in Wisconsin, Illinois, and Michigan between 2015 and 2016, was attributed to E. anophelis [21].In the Chicago metropolitan area, 14 people were sickened by Elizabethkingia in a ventilator-capable skilled nursing facility between 2021 and 2023 [22].Several outbreaks have also been documented in Asia (Singapore, Taiwan, Hong Kong, and Mainland China), Europe, and Africa [11,20,21,23].Elizabethkingia infections can apparently be acquired through both community and nosocomial settings, via exposure to contaminated surfaces of medical devices and equipment (such as hemodialysis and mechanical ventilation), water bodies and faucets, and the contaminated hands of healthcare workers [6].Multiple transmission routes of Elizabethkingia to humans have been proposed [1,6].An outbreak of Elizabethkingia infections has been linked to mosquitoes in the Central African Republic, while E. anophelis was further demonstrated to be transmitted from mosquitoes to mammalian hosts through mosquito bites [24,25].However, the occurrence of several winter outbreaks may diminish the significance of this transmission route [21,22].The above observations suggest that clinically important E. anophelis may have emerged from different lineages compared to mosquito-associated ones.
Several genomes of mosquito-associated E. anophelis strains have been sequenced, yet comprehensive genome analyses and systematic comparisons with clinically important strains have rarely been reported [11,[26][27][28].E. anophelis MSU001, a predominant bacterial member in the mosquito midgut, infected multiple mosquito species and was present in larval and adult life stages [9,17].Therefore, it has great potential for the biocontrol of mosquito-borne disease.Moreover, it can be used as a model organism for studying microbe-mosquito interactions, due to its amenability for genetic manipulation [9,17].In this study, we characterized a newly isolated strain and sequenced its genome to better understand its symbiotic traits.Furthermore, comparative genome analyses permitted investigation of its virulence factors and drug resistance, antecedent to applications as a paratransgenesis agent.

Materials and Methods
2.1.Culture E. anophelis strain MSU001, the primary strain of focus in this study, was isolated from the dissected midguts of adult, female Anopheles stephensi Liston mosquitoes (Johns Hopkins strain) fed with 10% sucrose on the 7th day after adult emergence.It was held at a colony in an insectary at Michigan State University, using mosquito colonization methods and sterile techniques, as described elsewhere [9,17].E. anophelis strain MSU001, E. coli JM109, and Serratia marcescens strain ano1 were grown in Luria-Bertani (LB) broth while shaking at 200 rpm at 30 • C [15].Trypticase soy broth (TSB) medium was used for the culture of Asaia sp.W12 under the same conditions [15].After MSU001 was cultured for 48 h, the spent broth was centrifuged at 4000 rpm for 15 min, filtered through a 2 µm filter, and heated at 80 • C for 10 min.To assess the effects of the spent medium on the growth of the tested bacteria including E. coli, Serratia marcescens ano1, and Asaia sp.W12, we added 100 µL of spent broth (prepared above) to 1.9 mL of bacterial suspension.After being cultured at 28 • C without shaking for 24 h, cell formation units (CFUs) were assayed by plating 100 µL of the above culture on their respective solid agars.For solid LB medium, Bacto agar (Difco, Detroit, MI, USA) was added at a final concentration of 20 g/liter and supplemented with erythromycin (Em) (100 µg/mL) for transposon selection.Previous studies showed that arginine is a critical amino acid that supports E. anophelis growth in M9 medium [9].An arginine utilization-deficient mutant (strain SCH873) was obtained by transposon-directed (pHimarEm1) mutagenesis (Chen, unpublished).Strain SCH814 (as the wild-type control) had been previously created by conjugatively transferring a transposon carrying expression cassette PompA + nluc [9].Both strains were used for metabolism experiments.For biochemical characterization of E. anophelis MSU001, we inoculated 150 µL of the bacterial suspension into a Biolog GEN III microplate and then incubated it at 30 • C. The color change was determined by following the manufacturer's recommendation.

MALDI-ToF MS Analyses
E. anophelis strains were streaked onto separate sheep blood agar plates and incubated at a temperature of 35.5 • C. Individual colonies were chosen for identification through VITEK MS, a MALDI-TOF/MS system manufactured by BioMérieux in the USA.A small portion of a colony was applied to a target plate and then immediately covered with 1 µL of α-cyano-4-hydroxycinnamic acid matrix solution.After drying, the target plate was inserted into a VITEK mass spectrometer instrument.The resulting spectra were recorded in linear mode within a mass range of 2 to 20 kDa.The subsequent spectra were analyzed by comparing the characteristics of the obtained spectrum with the typical spectrum of each known species.The primary spectrum for MSU001 was compared to the VITEK MS MS-ID database (version 2.0) for identification.

Genome Sequencing, Assembly, and Annotation
Next generation sequencing (NGS) libraries were prepared using an Illumina TruSeq Nano DNA Library Preparation Kit.Completed libraries were evaluated using a combination of Qubit dsDNA HS, Caliper LabChipGX HS DNA, and Kapa Illumina Library Quantification qPCR assays.Libraries were combined in a single pool for multiplexed sequencing, loaded on one standard MiSeq flow cell (v2), and sequencing was performed in a 2 × 250 bp paired-end format using a v2, 500 cycle reagent cartridge.NGS libraries were sequenced by Illumina Miseq paired-end sequencing technology at the Research Technology Support Facility (RTSF) at Michigan State University.The reads were assembled using CLC Genomics Workbench (version 10).Gene annotation was carried out using National Center for Biotechnology Information (NCBI) Prokaryotic Genome Automatic Annotation Pipeline (PGAAP 3.3) [29].Initial prediction and annotation of coding sequences (CDS) and tRNA/rRNA gene prediction were carried out via Glimmer 3 through the Rapid Annotation using Subsystem Technology server (RAST) [30].

Bioinformatics
The selected genome sequences (Table 1) were downloaded from NCBI and annotated using Prokaryotic Genome Annotation Pipeline (PGAP) (version 6.5).The average GC contents, coding sequences, predicted genes, and genome size were predicted by PGAP.The functional categorization and classification of predicted CDS of MSU001 were performed on the RAST server-based SEED viewer [31].The multi-drug resistance genes were predicted in the CARD database [31].Prophages and clustered regularly interspaced short palindromic repeats (CRISPR) were predicted using CRISPRfinder [32].For genomic similarity assessment, average nucleotide identity (ANI) and digital DNA-DNA hybridization (dDDH) values were computed using the web tools OrthoANIu and GGDC 2.0, respectively [33,34].For quantification and classification of regulatory system proteins, the web tool P2RP was used [35].The pan genome, core genome, and specific genes of MSU001 were analyzed by comparison with 16 representative Elizabethkingia genomes using EDGAR 3.2 [36].Sizes of pan genomes and core genomes were estimated using the core/pan development feature [37].Carbohydrate active enzyme families, including enzymes of glycan assembly (glycosyltransferases, GT) and deconstruction (glycoside hydrolases, GH, polysaccharide lyases, PL, carbohydrate esterases, CE), were semi-manually annotated using the Carbohydrate Active Enzyme (CAZy) database curation pipelines [38].The metabolism pathways were predicted using antiSMASH (https://antismash.secondarymetabolites.org,accessed on 23 October 2023), RAST, gutSMASH (https://gutsmash.bioinformatics.nl,accessed on 23 October 2023), and previous metabolomics data.A phylogenetic tree of the 18 Elizabethkingia genomes was constructed based on the complete core genome.For all 2307 gene sets of the core genome, a multiple alignment was constructed using MUSCLE [37].Subsequently, all alignments were concatenated and used as input for the neighbor joining method, as implemented in PHYLIP [39] and the approximate maximum likelihood method of Fasttree 2.1 [40].The resulting phylogenies were basically identical.In total, 41,526 CDS were used, with 783,693 amino acid residues per genome, and 14,106,474 in total.

Biochemical Characterization and Identification by MALDI-ToF/MS
E. anophelis MSU001 recovered from A. stephensi grew well in 5% sheep blood agar, without obvious hemolytic activity (Figure 1A) after 24 h incubation.It was nonmotile when cultured on motility test media (Figure 1B).It was oxidase positive and catalase positive.MSU001 cells were straight rods (Figure 1C,D) and had a diameter of 0.3 µm and length of 13.0 µm (Figure 1C).Carbon source (see Table S1), nitrogen source utilization, and osmotic tolerance were characterized by incubating cells in Biolog GEN III microplates at 37 • C overnight (Table S1).Our results showed that E. anophelis MSU001 tolerated up to 4% NaCl, but growth was inhibited at 8% NaCl.

Biochemical Characterization and Identification by MALDI-ToF/MS
E. anophelis MSU001 recovered from A. stephensi grew well in 5% sheep blood agar, without obvious hemolytic activity (Figure 1A) after 24 h incubation.It was nonmotile when cultured on motility test media (Figure 1B).It was oxidase positive and catalase positive.MSU001 cells were straight rods (Figure 1C,D) and had a diameter of 0.3 µm and length of 13.0 µm (Figure 1C).Carbon source (see Table S1), nitrogen source utilization, and osmotic tolerance were characterized by incubating cells in Biolog GEN III microplates at 37 °C overnight (Table S1).Our results showed that E. anophelis MSU001 tolerated up to 4% NaCl, but growth was inhibited at 8% NaCl.It metabolized several carbon sources, including the carbohydrates D-maltose, D-trehalose, D-cellobiose, D-gentibiose, D- sucrose, D-turanose, D-melibiose, D-glucose, D-mannose, D-fructose, D-fucose, D-mannitol, and D-glycerol.Moreover, it utilized D-serine, L-alanine, L-aspartic acid, L-glutamic acid, and L-histidine.The above observations indicated that E. anophelis MSU001 was capable of surviving in diverse environments.
The MALDI-TOF/MS system initially identified the strain as Elizabethkingia meningosepticum (Figure S1).However, analysis of the 16s rDNA sequence revealed a striking 99.93% similarity with E. anophelis Ag1 and E. anophelis R26, while only sharing an 80.37% similarity with E. meningosepticum strain NCTC10016 (ATCC 13253).This discrepancy can be attributed to the limitations of the default MALDI-ToF MS databases inaccurately classifying various members of the Flavobacteriaceae, particularly closely related strains within the Chryseobacterium and Elizabethkingia genera [41].The MALDI-TOF/MS system initially identified the strain as Elizabethkingia meningosepticum (Figure S1).However, analysis of the 16s rDNA sequence revealed a striking 99.93% similarity with E. anophelis Ag1 and E. anophelis R26, while only sharing an 80.37% similarity with E. meningosepticum strain NCTC10016 (ATCC 13253).This discrepancy can be attributed to the limitations of the default MALDI-ToF MS databases inaccurately classifying various members of the Flavobacteriaceae, particularly closely related strains within the Chryseobacterium and Elizabethkingia genera [41].

Genomic Features of E. anophelis MSU001
E. anophelis MSU001 had a genome size of 4.05 Mb and an average GC content of 35.4% (Table 1).The MSU001 genome encompassed 3857 coding sequences and 3753 genes.MSU001 possessed the second highest number of coding sequences (3857).The 17 selected Elizabethkingia genomes (comprising fourteen E. anophelis, two E. meningoseptica, and one E. miricola) exhibited similar general features (Table 1).These strains were isolated from diverse sources, such as mosquitoes, aquatic animals, plants, and humans in clinical settings.The genome sizes ranged from 3.59 to 4.42 Mb, with the GC content ranging between 35% and 36%.Among the mosquito-isolated E. anophelis strains (n = 6), the average genome size was 4.00 Mb.The genome size of E. anophelis MSU001 closely resembled those isolated from A. gambiae and A. sinensis, except for being slightly larger than E. anophelis As1.However, there was no statistically significant difference (p > 0.05, Student's t-test) compared to the average genome size of 4.2 Mb (n = 5) observed in E. anophelis strains isolated from human clinical samples.The distribution of coding sequences among specific subsystems was predicted using SEED subsystems by RAST analysis (Supplemental Figure S2).This revealed 27 subsystems consisting of 87 categories.The major subsystems included "Amino acids and derivatives" (265 coding sequences), "Carbohydrates" (133 coding sequences), "Cofactors, vitamins, prosthetic groups, pigments" (131 coding sequences), and "Protein metabolism" (124 coding sequences).Notable subsystems also encompassed "Virulence, disease, and defense" (32 coding sequences) and several invasive genetic elements such as "Phages, prophages, transposable elements, plasmids" (24 coding sequences) (Figure S2).CRISPRs may alter the genome and modulate gene functions to serve as an adaptive immune system.MSU001 showed the presence of one CRISPR, while the other mosquito-associated isolates lacked any.Of the remaining E. anophelis isolates, CRISPRs were only seen in LDVH-AR107, 296-96, and SUE (each of which showed the presence of two CRISPRs).CRISPRs were otherwise only seen in E. meningoseptica strains (Table 1).

Gene Repertoire and Phylogenetic Interference of E. anophelis MSU001
MSU001 showed a high ANI (>99%) with other strains of E. anophelis including R26 (type species), Ag1, AR4_6, AR6_8, and As1 (Table S2).The ANI value was greater than 97% for all other selected E. anophelis strains, indicating that MSU001 is indeed a strain of E. anophelis.However, ANI values were lower in comparison with E. meningoseptica (<81%) and E. miricola (<93%).Additionally, DDH values were calculated and were consistent with the analysis by ANI (Table S2).The phylogeny of selected E. anophelis strains is shown in Figure 2. E. anophelis MSU001 from A. stephensi was phylogenetically close to isolates from other mosquitoes (strain Ag1, R26, AR4-6, AR4-8 and As-1).The clinical strains were divided into three clusters and separated from the clade formed by mosquito isolates (Figure 2).

Metabolites Involved in Symbiosis
Several important metabolites such as sphingolipids (SLs) and inositol were detected in the extracts from the midguts of mosquitoes which were fed with both sugar and blood meals in a previous study [42].Genes involved in the biosynthesis of SLs and inositol were detected in E. anophelis genomes, highlighting that E. anophelis may contribute to the above process.Although SLs are not commonly found as components of bacterial membranes, they have been uniquely identified in certain groups of microbes such as Bacteroides and Sphingomonads [43].Interestingly, the putative sphingolipid synthesis genes were identified in all selected Elizabethkingia genomes, suggesting their potential involvement in symbiotic relationships, affecting cytotoxicity, colonization of the host, biofilm formation, and modulation of host inflammation [44].Furthermore, inositol, an important nutritional and signaling factor, was found to be involved in metabolic pathways [45].These pathways may participate in regulating the stress response, such as cold tolerance, in the hosts.
The growth of SCH873 in M9 medium was impaired, compared to the WT (SCH814) (Figure 5A, left panel).When a 20-diluted LB broth was added into M9 medium, the growth of SCH873 was promoted, while the cell density was much lower than that in SCH814 (Figure 5A, right panel).At 7 days post-infection in adult mosquitos, the cell density of WT Elizabethkingia cells was around 15.8-fold higher than that of arginine utilization

Metabolites Involved in Symbiosis
Several important metabolites such as sphingolipids (SLs) and inositol were detected in the extracts from the midguts of mosquitoes which were fed with both sugar and blood meals in a previous study [42].Genes involved in the biosynthesis of SLs and inositol were detected in E. anophelis genomes, highlighting that E. anophelis may contribute to the above process.Although SLs are not commonly found as components of bacterial membranes, they have been uniquely identified in certain groups of microbes such as Bacteroides and Sphingomonads [43].Interestingly, the putative sphingolipid synthesis genes were identified in all selected Elizabethkingia genomes, suggesting their potential involvement in symbiotic relationships, affecting cytotoxicity, colonization of the host, biofilm formation, and modulation of host inflammation [44].Furthermore, inositol, an important nutritional and signaling factor, was found to be involved in metabolic pathways [45].These pathways may participate in regulating the stress response, such as cold tolerance, in the hosts.
The growth of SCH873 in M9 medium was impaired, compared to the WT (SCH814) (Figure 5A, left panel).When a 20-diluted LB broth was added into M9 medium, the growth of SCH873 was promoted, while the cell density was much lower than that in SCH814 (Figure 5A, right panel).At 7 days post-infection in adult mosquitos, the cell density of WT Elizabethkingia cells was around 15.8-fold higher than that of arginine utilization mutants in A. stephensi, indicating that Elizabethkingia cells might need to interact with either mosquito host or other microbes to obtain arginine for growth (Figure 5B).To assess the effects of E. anophelis metabolites on the growth of other common mosquito gut symbionts (Asaia sp.W12 and Serratia marcescens), the number of colonies that grew from cultures with added metabolites was compared to control groups (Figure 5).In cultures of E. coli (a representative for non-symbionts), the metabolites significantly hindered colony formation, resulting in less than half the number of viable colonies compared to the control group and indicating a reduction in growth by approximately 58%.The growth inhibition of Asaia sp.W12 and Serratia marcescens with metabolites was less pronounced, with approximately 26% and 17% reductions in growth (Figure 5C), respectively.These findings suggest that E. anophelis metabolites have inhibitory effects on the growth of common mosquito gut symbionts, highlighting the potential role of E. anophelis in modulating the microbial community within the mosquito gut.
tures with added metabolites was compared to control groups (Figure 5).In cultures of E. coli (a representative for non-symbionts), the metabolites significantly hindered colony formation, resulting in less than half the number of viable colonies compared to the control group and indicating a reduction in growth by approximately 58%.The growth inhibition of Asaia sp.W12 and Serratia marcescens with metabolites was less pronounced, with approximately 26% and 17% reductions in growth (Figure 5C), respectively.These findings suggest that E. anophelis metabolites have inhibitory effects on the growth of common mosquito gut symbionts, highlighting the potential role of E. anophelis in modulating the microbial community within the mosquito gut.

Regulatory System Proteins
The genome of E. anophelis MSU001 possessed genes encoding 51 two-component system proteins, 188 transcription factor proteins, and 13 other DNA-binding proteins, resulting in a total count of 252 regulatory proteins (Table 2).This count was the highest among the mosquito-associated E. anophelis isolates, except for As1, which displayed

Regulatory System Proteins
The genome of E. anophelis MSU001 possessed genes encoding 51 two-component system proteins, 188 transcription factor proteins, and 13 other DNA-binding proteins, resulting in a total count of 252 regulatory proteins (Table 2).This count was the highest among the mosquito-associated E. anophelis isolates, except for As1, which displayed reduced protein counts in all categories, totaling 215 proteins (Table 2).The other mosquitoassociated isolates shared similar counts of two-component system proteins and transcription factor proteins.The main variation among these isolates was observed in the number of DNA-binding proteins, with Ag1, AR4-6, and AR6-8 lacking only one fewer ODP (another DNA-binding protein), and R26 lacking two (Table 2).

Carbohydrate Active Enzymes
A total of 124 CAZyme-encoding genes were predicted in E. anophelis MSU001, consisting of approximately 3% of the bacterial genome (Tables S3 and S4).Notably, CBM12 (carbohydrate-binding module family 12) and AA10 (auxiliary activity family 10, lytic polysaccharide monooxygenases) were exclusive to mosquito-associated E. anophelis strains, highlighting their importance in establishing a symbiotic relationship with insects.The overall predicted CAZyme repertoires in mosquito-associated E. anophelis were comparable, featuring 61 glycoside hydrolases (GHs).In contrast, E. anophelis As1 exhibited a slightly lower count of 56 GHs (Table S3).This collective decrease in GHs among mosquito isolates, ranging from 61 to 67, contrasted with clinical species, suggesting a distinct evolutionary route.Compared to the clinically important strains, decreased copy numbers of GH3, GH29, and GT4 were detected in insect-associated Elizabethkingia strains (Table S3), showing that while these specific CAZyme genes may be involved in pathogenesis in humans, they may not be relevant for insect symbiosis.Both E. anophelis and E. miricola species harbored single copies of GH1 (β-glycosidase), which is absent in E. meningoseptica.Conversely, GH30, present in E. meningoseptica, was only detected in selected clinical E. anophelis strains and was absent in E. miricola.Additionally, E. anophelis lacked GH33 (sialidase), a characteristic found in E. meningoseptica and some E. miricola strains.Genes encoding GH5 (subfamily 46) and CBM6 (β-glucan binding), consistently observed in E. anophelis, were not found in E. meningoseptica.

Pathogenesis Potential Revealed by Virulence Factors and MDR Analysis
Using the VFDB protein Set B database, a comparative analysis of selected Elizabethkingia isolates was conducted to identify homologs of virulence factors (VFs) (Table 3).Ten VFs of interest were discovered, namely C8J 1080, DnaK, EF-Tu, eno, htpB, katG, mps1-1, mps1-2, pgIC, and RmIA.These VFs play diverse roles in cellular functions such as mitotic regulation, capsule formation, stress response (involving heat shock proteins, catalase, and hydratase), ion transport proteins, secretion systems, and defense or invasion mechanisms during pathogenesis.Among the selected VFs, genes encoding DnaK, EF-Tu, mps1-1, mps1-2, and RmIA were present in all E. anophelis isolates.Eno and htpB were found in all mosquito-associated isolates, while their presence in clinically isolated human samples varied.PgIC was observed in all mosquito-associated isolates but was completely absent in human Elizabethkingia strains.Both mosquito-and human-associated E. anophelis strains shared the presence of C8J 1080 and katG, which were not identified in other animal-associated strains (Table 3).
The antimicrobial resistance profile of E. anophelis was determined using the broth microdilution method.The strain exhibited resistance to 13 out of the 16 tested antibiotics, including aminoglycosides, tetracycline, nitrofuran, and all β-lactam antibiotics, such as cephalosporins, monobactams, and extended-spectrum penams/β-lactamase inhibitors.However, it showed susceptibility to trimethoprim/sulfamethoxazole (sulfonamide) and ciprofloxacin (quinolone), and intermediate susceptibility to tigecycline (Table 4).In addition, the prediction of antibiotic resistance genes in E. anophelis MSU001 revealed its multidrug resistance traits (Table S4).Notably, Elizabethkingia species are known for their high resistance to β-lactam drugs, due to the production of β-lactamases (Table S4), which hydrolyze these antibiotics.In the case of MSU001, it carried at least five different β-lactamase genes (BlaB, CME-1, GOB-9, IND-7, and TLA-1) that may confer broad resistance to penams, cephalosporins, and carbapenems.It is interesting that the presence of IND-7, which encodes for a class B carbapenem-hydrolyzing β-lactamase, was unique to the MSU001 strain.Mosquito-associated E. anophelis strains carried GOB-9 (encoding a class B β-lactamase) and TLA-1, which were only found in a few clinical Elizabethkingia isolates.Furthermore, it is noteworthy that GOB-9 was absent in E. miricola and E. meningoseptica.Genes encoding BlaB (inducible class C cephalosporinase) and CME-1 (class A β-lactamase) were present in most selected Elizabethkingia species (Table S4).However, mosquito-associated E. anophelis lacked several β-lactamase genes found in other selected Elizabethkingia strains, indicating unique evolutionary routes for these mosquito-associated strains.

Discussion
Studies have shown that a substantial portion of the colonizing bacteria found within adult mosquito hosts are acquired in aquatic habitats during larval life stages [9,16,17].Elizabethkingia species are common mosquito symbionts dispersed in natural water bodies (dams, wetlands, and rivers), but do not normally predominate in these environments (composing 6.25 × 10 −6 to 8.21 × 10 −6 of the total bacterial community) [46,47].However, Elizabethkingia species populate mosquito midguts and can spread to other organs and tissues, including the salivary glands, reproductive organs (ovary or testicles), crop, and alimentary canal of mosquitoes at various development stages [47].The complex interactions between arthropod hosts and their associated microbes warrant a holistic analysis of these communities and the environments that foster them [47].Bacteria need to overcome digestion, microbial competition, and a multitude of other stress factors (e.g., iron and oxidative stress, larval metamorphosis, temperature, pH) associated with mosquito physiology [9,17].The ability to thrive in dynamic environments within a host emphasizes the importance of bacterial adaptability and likely highlights a deeper symbiotic relationship underlying microbial persistence [47].By conducting an analysis of the genomic and molecular mechanisms behind Elizabethkingia colonization, we hoped to enhance our understanding of microbe-host interactions.
Correctly identifying Elizabethkingia species has proven to be a challenge with varying success, further complicated by prior nomenclature changes and various method limitations [41].Current classification of Flavobacteriaceae members relies heavily on MALDI-ToF mass spectrometry, but despite its wide utility in bacterial identification, it struggles to accurately classify members from Chryseobacterium and Elizabethkingia genera [19,41,46].Furthermore, standard databases are limited to only a few Elizabethkingia isolates, often falsely defaulting to E. meningoseptica or E. miricola [41].This was evidenced by our own study, as well as others, where MALDI-ToF frequently misidentified E. anophelis as E. meningosepticum [41,46,48].The use of 16S rRNA sequences has been shown to be limited in its taxonomic utility as well [48].The fact that misidentification via conventional methodologies is so prevalent in the literature may indicate E. anophelis is an underrepresented pathogen responsible for more disease in humans than previously attributed [46].These limitations highlight the need for updating standard MALDI-ToF databases, as well as for thorough, enhanced identification methodologies that utilize a combination of widely adopted bacterial identification methods like 16s rDNA sequencing in conjunction with biochemical testing [41,46,48].Moreover, whole genomic sequence analysis and aver-age nucleotide identity as a complementary method may be used to correctly identify E. anophelis [46,49].
Genome size and GC content were similar among most E. anophelis strains.MSU001 exhibited characteristics of an open pan-genome, likely relating to its diverse habitats, spanning both aquatic and terrestrial environments, as well as the many different human, animal, and plant hosts that it may colonize [46].However, the core genome analysis demonstrated that strains from mosquitoes shared more conserved genes than those from clinical specimens.Furthermore, the phylogenetic placement of mosquito-associated E. anophelis species formed different clades from clinical isolates.They were also distinct from E. meningoseptica and E. miricola clades.Collectively, these results indicate that E. anophelis MSU001 and other mosquito isolates likely evolved in different routes to adapt to mosquito hosts compared to clinical strains.
Another notable finding was the presence of Elizabethkingia genes involved in sphingolipid biosynthesis.Sphingolipids are a ubiquitous component in eukaryotic cell membranes that have been shown to play critical roles in cell signal transduction, regulation of apoptosis, adhesion and uptake, and inflammation in the host [50].Several pathogens can actively synthesize or hydrolyze these molecules to hijack host cell responses and orchestrate favorable immune responses [50].Furthermore, certain sphingolipids like sphingosine have also been shown to possess a possible antibacterial effect [50].Bacteria employ diverse mechanisms to facilitate host interactions and survival in their environments.The production of various secondary metabolites by Elizabethkingia likely conferred advantages over other members of the microbial community, allowing it to disturb the bacterial consortium and outcompete or even inhibit its competitors [50].
Chitin is one of the most abundant polysaccharides, forming important structures in the insect exoskeleton and gut linings [51].Due to the vital role of chitin in development and defense against pathogen invasion, insects need to frequently reshape its structure and components [51].Microbial symbionts may be involved in chitin degradation and its synthesis [52].In this study, we observed that the modules of CBM12 associated with chitinase and AA10 were uniquely found in mosquito-associated E. anophelis (except As1).These CAZymes possibly contribute to the binding and lysing of chitin [52].For example, upon a mosquito's bite, the ingested blood meal triggers the midgut epithelium to release various factors including chitin microfibrils (3-13%) and protein complexes, which form a peritrophic matrix (PM) [53].The PM effectively creates a barrier between the blood bolus and the midgut epithelial cells, serving as a protective shield against abrasive particles and microbial infections [53].After the red blood cells have been thoroughly digested, the PM needs to be dismantled to release the nutrients.Microbial chitinase secreted by gut microbiota may facilitate this process [52][53][54].Moreover, microbial chitinases may contribute to the reshaping of chitin components during mosquito molting, supported by the presence of E. anophelis in various mosquito body sites [51,52].The majority of predicted CAZymes in Elizabethkingia species appear to be involved in utilizing simple sugars rather than degrading complex plant polysaccharides, which is consistent with their living niches (e.g., within mosquitoes or humans) [46][47][48].Our results also indicated that pathogenic E. anophelis possibly requires additional copies of GH3, GH29, and GT4 to participate in pathogenesis.Furthermore, E. anophelis and E. miricola have different sets of CAZymes involved in sugar metabolism.Therefore, future characterization of their physiological functions is warranted.
Despite their different sources, Elizabethkingia bacteria exhibited comparable numbers of response regulators, phosphotransferase proteins, histidine kinases, one-component systems, transcriptional regulators, sigma factors, and other DNA-binding proteins (Table 2).These regulatory proteins play critical roles in maintaining bacterial metabolism and function, explaining their consistent presence across Elizabethkingia species (Table 2).The numbers of regulatory protein genes between mosquito-associated and clinical E. anophelis genomes varied and were not statistically different.The retainment of similar complicated regulatory systems may indicate an adaptability of this organism to diverse host envi-ronments [46].E. anophelis living in the adult female mosquito midgut may experience similar stress conditions to those where bacteria invade the bloodstream of mammalian hosts [9,16,17].For example, mosquito-associated bacteria are exposed to iron-depleting conditions and relatively lower temperatures prior to blood meals [13,17]; conversely, they encounter iron-rich environments during and after blood meals [13].Similar processes may occur prior to entry into the bloodstream or after the lysis of the erythrocytes during a bacteremia event [25,28].Furthermore, the evasion of immune cells and resistance to temperature variations during the above processes are expected to be similar [55].
The emerging pathogenicity of Elizabethkingia is likely attributed to its large genome, ecological and metabolic plasticity, a multitude of virulence factor genes present in its genetic repertoire, and broad antibiotic resistance [46,48].Among the diverse virulence factors, we discovered that PgIC was only present in mosquito-associated isolates.PglC plays a vital role in the N-linked protein glycosylation pathway in Campylobacter jejuni [56].This pathway primes proteins for nucleophilic attack by the polyprenol acceptor within the cellular membranes, which may play important roles in epithelial cell adherence, invasion, and colonization of the host during the infection course [56,57].Antimicrobial susceptibility patterns vary across strains and in the case of clinical isolates, provide an additional layer of difficulty in the selection of appropriate therapeutics [23,58,59].While β-lactamase synthesis remains the most employed defense among Gram-negative bacteria to withstand antibiotics, other resistance mechanisms include the alteration of target drug sites and the implementation of efflux pumps to eliminate the drug from the cell [59].The presence of specific β-lactamase genes varies across different host-associated strains, suggesting that these genes confer certain advantages within Elizabethkingia and their respective evolutionary routes [20,23].Those virulence factors that aid in transmission promote adhesion, motility, and biofilm formation, while other factors mediate host interactions and allow for extended persistence within hostile environments [58,60].Further research into variations in genomic features between mosquito-associated and clinically significant strains of Elizabethkingia is warranted.

Figure 1 .
Figure 1.Growth features and microscopic observation of E. anophelis MSU001.(A) Hemolytic activity on sheep blood agar; (B) motility test; (C) scan electron microscopy; (D) demonstration of bacterial morphology by electron microscopy with negative stain.

Figure 1 .
Figure 1.Growth features and microscopic observation of E. anophelis MSU001.(A) Hemolytic activity on sheep blood agar; (B) motility test; (C) scan electron microscopy; (D) demonstration of bacterial morphology by electron microscopy with negative stain.

Figure 2 .
Figure 2. Phylogenetic placement of E. anophelis MSU001.The tree was constructed with 18 genomes with a core of 2307 genes per genome, 41,526 in total.The core had 783,693 amino acid residues/bp per genome, 14,106,474 in total.The horizontal bar represents 0.05 substitutions per site.

Figure 2 .
Figure 2. Phylogenetic placement of E. anophelis MSU001.The tree was constructed with 18 genomes with a core of 2307 genes per genome, 41,526 in total.The core had 783,693 amino acid residues/bp per genome, 14,106,474 in total.The horizontal bar represents 0.05 substitutions per site.

Figure 2 .
Figure 2. Phylogenetic placement of E. anophelis MSU001.The tree was constructed with 18 genomes with a core of 2307 genes per genome, 41,526 in total.The core had 783,693 amino acid residues/bp per genome, 14,106,474 in total.The horizontal bar represents 0.05 substitutions per site.

Figure 4 .
Figure 4. Venn diagram illustrating the distribution of shared and specific clusters of orthologous groups in the selected Elizabethkingia genomes.(A) Venn diagram of shared and unique genes in the selected mosquito-associated Elizabethkiniga.(B) Venn diagram of shared and unique genes in MSU001 and the clinically important Elizabethkiniga.The unique and shared genomes among the compared genomes were determined using the BLAST score ratio approach of EDGAR 3.2 with a cutoff of 30%.

Figure 4 .
Figure 4. Venn diagram illustrating the distribution of shared and specific clusters of orthologous groups in the selected Elizabethkingia genomes.(A) Venn diagram of shared and unique genes in the selected mosquito-associated Elizabethkiniga.(B) Venn diagram of shared and unique genes in MSU001 and the clinically important Elizabethkiniga.The unique and shared genomes among the compared genomes were determined using the BLAST score ratio approach of EDGAR 3.2 with a cutoff of 30%.

Figure 5 .
Figure 5. Inhibitory effects of Elizabethkingia metabolites on selected bacteria.* Statistically significant difference (p < 0.05).(A) Growth comparison between wild type strain for arginine utilization (SCH814) and arginine metabolism mutant (SCH873) in the M9 medium and M9 medium supplemented with 20-fold diluted LB medium.(B) Comparison between growth of SCH814 and SCH873 in mosquitoes.(C) The effects of spent media on the growth of Asaia sp.W12, Serratia marcescens and E. coli.The spent broth from E. anophelis MSU001 (48-h incubation) was added E. coli, Serratia marcescens ano1 and Asaia sp.W12, statically cultured at 28 °C for 24 h and plated on their respective solid agar media for CFU calculation.

Figure 5 .
Figure 5. Inhibitory effects of Elizabethkingia metabolites on selected bacteria.* Statistically significant difference (p < 0.05).(A) Growth comparison between wild type strain for arginine utilization (SCH814) and arginine metabolism mutant (SCH873) in the M9 medium and M9 medium supplemented with 20-fold diluted LB medium.(B) Comparison between growth of SCH814 and SCH873 in mosquitoes.(C) The effects of spent media on the growth of Asaia sp.W12, Serratia marcescens and E. coli.The spent broth from E. anophelis MSU001 (48-h incubation) was added E. coli, Serratia marcescens ano1 and Asaia sp.W12, statically cultured at 28 • C for 24 h and plated on their respective solid agar media for CFU calculation.

Table 1 .
Genomic features in selected Elizabethkingia species.

Table 2 .
Predicted regulatory proteins in the selected Elizabethkingia species *.
[35]e regulatory proteins were predicted by the web tool P2RP[35].TOC, two-component systems; TF, transcription factors; ODP, other DNA-binding proteins; RR, response regulators; PP, phosphotransferase proteins; HK, histidine kinases; OCS, one-component systems; TR, transcriptional regulators; SF, sigma factors.The numbers in this table are the gene copies encoding the regulatory proteins.