Abstract
Symbiotic bacteria are pervasive in mosquitoes and their presence can influence development, reproduction, and immunity of their host. It is evident that environmental and host genetic factors contribute in shaping the microbiome of mosquitoes, but we have a poor understanding regarding how bacterial genetics affects colonization of the mosquito gut. While CRISPR/Cas9 gene editing is a powerful tool to modify bacterial genomes this approach has yet to be applied to insect symbionts. To demonstrate that gene editing can be completed in non-model bacterial species isolated from insects and to investigate the role of bacterial genes in gut colonization, we mutated the outer membrane protein A (ompA) gene of an Enterobacter symbiont using the CRISPR/Cas9 system. The ΔompA mutant had an impaired ability to form biofilms and poorly infected Ae. aegypti when reared in a mono-association under gnotobiotic conditions. In adults, the mutant had a significantly reduced infection prevalence compared to the wild type or complement strains, while no differences in prevalence were seen in larvae, suggesting bacterial genetic factors are important for adult gut colonization. Integration of genes (antibiotic resistance and fluorescent markers) into the symbiont genome demonstrated this technology can be exploited to develop novel symbiotic control strategies to interfere with arboviral pathogens such Chikungunya, Zika and Yellow fever viruses transmitted by Aedes mosquitoes. Our results shed insights onto the role of ompA gene in host-microbe interactions in Ae. aegypti and confirm that CRISPR/Cas9 gene editing can be employed for genetic manipulation of non-model gut microbes.
Importance CRISPR/Cas9 gene editing approaches have revolutionized several biological fields, however despite their applicability for altering bacterial genomes, few studies use this technology in microbes that associated with eukaryotic hosts. Here we use this editing approach to knockout a gene encoding a membrane protein in an Enterobacter isolated from Aedes mosquitoes and show this gene is essential for biofilm formation and promotes bacteria colonization of the gut. A reduced bacterial load of the mutant compared to the wild type or complement strains, was seen in both larval and adult mosquitoes, however this was most evident in adults, likely due differences in the mode of acquisition of microbes at each life stage. Our work extends CRISPR/Cas9 genetic manipulation into a new bacterial species, and in conjunction with other studies, suggests that members within Enterobacteriaceae are amenable to genome engineering by this approach. This study will facilitate the development of novel microbial-based approaches to mitigate mosquito-borne disease.
Introduction
Mosquitoes harbor a community of microbes within their guts. In general, the gut-associated microbiome of mosquitoes tends to have low species richness but can differ greatly between individuals and habitats1-8. Importantly, these microbes can modulate many host phenotypes, several of which can influence vectorial capacity9-11. As such, it is imperative that we understand how the microbiome is acquired and maintained within mosquito vectors. While environmental factors unquestionably influences mosquito microbiome composition and abundance2-4,8, and studies are elucidating the role of microbial interactions5,7,12,13 and host genetic factors14-18 in shaping the microbiome, we have a poor understanding regarding bacterial factors that influence colonization of the mosquito gut.
In other invertebrates, several bacterial genes have been implicated in gut colonization. For example, a genome wide screen exploiting transposon-sequencing found a suite of genes from the bacterium Snodgrasselia involved in colonization of the honey bee gut19. These bacterial genes were classified into the broad categories of extracellular interactions, metabolism and stress response19. Knock out of a purine biosynthesis gene in Burkholderia impaired biofilm formation and reduced bacterial colonization rates in a bean bug. Biofilm formation was also shown to play a role in virulence of pathogenic Pseudomonas in artificial infections of Drosophila, with strains that lacked the capacity to form biofilms being more virulence to the host, while a hyperbiofilm strain was less virulent than the WT strain20. In other blood feeding invertebrates, bacterial genetics also appears critical for host colonization. Knockout of the type II secretion system in Aeromonas veronii reduced infection in Hirudo verbena leeches21. In Tsetse flies, the outer-membrane protein A (ompA) gene of Sodalis glossinidius is essential for symbiotic interactions22. Sodalis mutants lacking the ompA gene poorly colonized the fly gut compared to the wild type (WT) Sodalis22 and the mutant strain also had a reduced capacity to form biofilms23. Heterologous expression of the ompA gene from pathogenic Escherichia coli in Sodalis mutants induced mortality in the fly implicating this gene as a virulence factor in pathogenic bacteria22. Taken together, these studies suggest that bacterial genetic factors are critical for host colonization of invertebrates and that biofilm formation facilitates symbiotic associations in insects.
In mosquitoes, few studies have investigated how bacterial genetics affect gut colonization. However, evidence from experimental evolution studies suggests bacterial genetics plays a critical role. In two separate studies, Enterobacter was selected for increased persistence in the gut of Anopheles gambiae mosquitoes, the major malaria vector in sub-Saharan Africa, by repeatedly infecting mosquitoes with strains that persisted in the gut for longer periods of time24,25. Transcriptomics comparisons of effective and ineffective colonizers in liquid media identified 41 genes that were differentially expressed between these two strains25, further implicating the importance of bacterial genetics in mosquito infection, however the role of these genes in colonization of the mosquito gut has not been resolved. In a separate study, in vitro screening of a transposon mutant library of Enterobacter identified a waaL gene mutant that was insensitive to oxidative stress26. The waaL gene encodes an O antigen ligase which is needed for attachment of the O antigen to lipopolysaccharide and the mutant was found to have lower rates of colonization of the midguts of Anopheles mosquitoes26.
Gene knockouts approaches in bacteria provide compelling evidence of the role of bacterial genes in host-microbe interactions21-23,26-28. In general, most studies use transposon mutagenesis for gene knockout, which requires screening of the mutant library. As in vivo screening is burdensome and in some cases not feasible, a targeted gene knockout approach is highly desirable to investigate the functionality of bacterial genes in host-microbe interactions. In the past few years, the CRISPR/Cas9 gene editing system has been employed to modify bacterial genomes29-31. While much of the work has been done in model bacterial species32-36, editing approaches have expanded into non-model bacterial systems37-42. Despite this expansion, few studies have used this technology in host-associated microbes43. In the vector biology field, gene knockout approaches can be used to interrogate the role of bacterial genes responsible for host-microbe interactions, while the ability to integrate genes into the bacterial symbiont genome has great potential for applied paratransgenic control strategies11,44-46. Previously, manipulation of non-model symbionts that associate with insect vectors have has been accomplished by plasmid transformation47-55 or stable transformation of the genome using transposons or integrative plasmids56-61, but the use of CRISPR/Cas9 gene editing in symbionts has yet to be accomplished. For paratransgenic strategies, stable site-specific integration of transgenes into the symbiont genome is critical, and as such, the application of CRISPR/Cas9 gene editing technology to non-model bacteria that associate with insect vectors will stimulate research in this field.
We therefore undertook studies to develop CRISPR/Cas9 genome editing approaches in an Enterobacter species isolated from Aedes aegypti mosquitoes, the major vector of arboviruses such as dengue, Zika and Yellow fever viruses. We used the Scarless Cas9 Assisted Recombineering (no-SCAR) method to disrupt the ompA gene of the non-model Enterobacter species35. The no-SCAR approach is a single step genome editing system that does not require a selectable marker35. After characterization of the mutant in vitro, we examined the role of the ompA gene in host-microbe interactions by re-infecting bacteria into mosquito in a mono-association. To demonstrate that the CRISPR/Cas9 gene-editing system could be useful for applied symbiotic control approaches we inserted genes conferring antibiotic resistance or a fluorescent protein into the bacterial genome and re-infected the altered strains back into mosquitoes. Our result sheds insights into the role of the ompA gene in host-microbe interactions in Ae. aegypti and confirm that CRISPR/Cas9 gene editing can be a powerful tool for genetic manipulation of native gut-associated microbes of mosquitoes.
Results
Enterobacter biofilm formation in Ae. aegypti guts
Over the course of conducting mono-axenic infections in Ae. aegypti mosquitoes with an Enterobacter symbiont, we repeatedly observed a conglomeration of bacterial cells in the gut that was indicative of a biofilm (Figure 1). This formation of bacteria has a similar appearance to biofilms observed in the guts of other insects20,23. No bacteria were observed in Ae. aegypti when infections were conducted with E. coli that was not adapted to the mosquito gut environment (Figure 1), although as seen previously, infection with E. coli enabled mosquito development62-64. We therefore sort out to examine the role of bacterial genetics in biofilm formation and host colonization of this gut-associated bacterium of Aedes mosquitoes. While several genes have been implicated in biofilm formation20,23,65, we chose to knockout the ompA gene of Enterobacter given that this gene has been demonstrated to influence biofilm formation and gut colonization of Sodalis22,23, an Enterobacteriaceae symbiont of Tsetse flies, which is phylogenetically related to Enterobacter. The CRISRP/Cas9 genome editing system was employed to edit the symbionts genome.
Genome editing in non-model Enterobacter bacteria isolated from mosquitoes
To edit the Enterobacter isolate that resides within the gut of Aedes mosquitoes, we employed the no-SCAR gene editing approach that had been developed in E. coli35. To optimize the approach in our hands, we performed initial experiments in E. coli to delete a ∼1 kb region of the ompA gene (Figure 2A). As the no-SCAR approach exploits the λ-Red recombineering system to repair double stranded breaks, we supplied cells with a double stranded DNA template that had regions of homology flanking the gRNA site (250 bp for each arm). Using this approach, we successfully deleted a 1001 bp fragment of the ompA gene. Of the colonies we screened, we saw an editing at a frequency of 6.25% (N = 48) (Figure 2A). Given our successful editing in E. coli, we employed this technique in the non-model Enterobacter. However, we altered our editing procedure to delete a 598 bp fragment from the Enterobacter ompA gene. This was done to attain a higher frequency of mutants66 and accommodate a different PAM site in the ompA gene of Enterobacter. Using a donor template designed for the Enterobacter ompA gene that had similar length flanking homology arms as the previous experiment done in E. coli, we obtained mutant knockouts at a rate of 32% (N = 50) (Figure 2B). For both bacterial species, Sanger sequencing across the integration site indicated the deletion occurred at the expected loci in the bacterial genome (Figure 2C; Supplementary text 1 and 2).
Characterization of the Enterobacter ompA mutant
We quantified the growth rates of the ΔompA mutant in comparison to the WT Enterobacter and the ΔompA/ompA complement in liquid LB media. We saw no significant difference between the WT, the ΔompA mutant or the ΔompA/ompA complement (Figure 3A). To examine the stability of the deletion, we subcultured the ΔompA mutant on LB media for 10 generations and performed PCR to amplify across the deletion. At alternative generations PCR analysis indicated the deletion was present indicating genomic stability at this site (Figure 3B).
Previously, ompA has been shown to be important in biofilm formation as Sodalis deletion mutants were unable to form biofilms23. As such we characterized in vitro biofilm formation using the crystal violet (CV) biofilm assay. After visual inspection, it was clear the ΔompA mutant had distinctly less biofilm deposition compared to either the WT or the ΔompA/ompA complement (Figure 3C), and after quantification and normalization to account for any difference in growth between the strains, biofilm formation was confirmed to be significantly different between the ΔompA mutant and the WT (Figure 3D; Tukey’s multiple comparisons test, P < 0.0001) or ΔompA/ompA complement (Tukey’s multiple comparisons test, P < 0.0001), while the was no significant differences between the WT and the ΔompA/ompA complement (Tukey’s multiple comparisons test P = 0.2).
The role of ompA gene in mosquito infection
To examine the importance of the ompA gene on bacterial colonization of mosquitoes, we infected Ae. aegypti mosquitoes in a mono-association under gnotobiotic conditions6. This infection method was used to avoid other gut-associated microbes influencing host colonization rates7 and it also assisted in quantification of introduced bacteria by measuring colony forming units (CFUs). In larvae we saw a significant reduction in bacterial titer in the mutant compared to both the WT (Kruskal-Wallis test; P < 0.01) and the ΔompA/ompA complement (Kruskal-Wallis test; P < 0.05) (Figure 4A). Similarly, in adults, there was a significant reduction in bacterial infection in the ΔompA mutant compare to either the WT or ΔompA/ompA complement (Kruskal-Wallis test; P < 0.001) (Figure 4B). While no significant changes were seen in the prevalence of infection (number of mosquitoes infected) in the larval stage (Figure 4C, Fisher’s exact test; WT compared to ΔompA P = 0.24 and ΔompA compared to ΔompA/ompA P = 0.24), in adults, the prevalence of infection was significantly different (Figure 4D, Fisher’s exact test; WT compared to ΔompA P < 0.0001 and ΔompA compared to ΔompA/ompA P < 0.0001), with only 45% of adults infected by the ΔompA mutant compared to 95% and 88% by the WT and ΔompA/ompA complement, respectively. We also examined the growth rates of mosquitoes administered with the WT, ΔompA mutant and ΔompA/ompA complement. No significant differences were seen in the time to pupation (Figure 5A) or percentage of first instar larvae that reached adulthood (Figure 5B) between any of the strains.
Integration of genes into the Enterobacter chromosome
We undertook experiments to demonstrate the CRISPR/Cas9 gene-editing approaches can be used to integrate genes into the chromosome of non-model bacteria that associate with mosquitoes. We created two independent transgenic strains that had either, a gene encoding mCherry fluorescence or a gene encoding resistance to the antibiotic gentamicin, inserted into the bacterial chromosome. These genes were integrated into the genome using the same gRNA that was used for deletional mutagenesis, and as such, these insertions also disrupted the ompA gene. Sequencing across the integration site indicated the insertion of these genes occurred within the ompA gene and thereby disrupted its function (Figure 6A and 6D). Continual subculturing was undertaken for both strains and molecular analysis indicated the stability of these lines for ten generations (Figure 6B and 6E). To demonstrate the integrated genes were functional, we observed expression of mCherry fluorescence and successfully cultured the strain containing gentamicin resistance on plates containing the antibiotic (Figure 6C and 6F). Finally, we infected these transgenic strains into mosquitoes to demonstrate that these strains were able to colonize the mosquito gut and functionality of the integrated gene was confirmed by observing fluorescence or by rearing the Enterobacter ompA∷gentamicin strain in mosquitoes administered sugar supplemented with gentamicin. Fluorescent bacteria were observed in the gut of mosquitoes while no signal was seen in controls (WT Enterobacter infected mosquitoes) (Figure 5G). The Enterobacter ompA∷gentamicin was successfully rescued from mosquitoes reared on gentamicin and was seen to stably infect mosquitoes over time at a density of 1×104 CFUs/mosquito. Consistent with our previous finding (Figure 4B), the WT bacteria initially infected mosquitoes at higher titers (T test; day 0 P < 0.001). However, at 4 days post infection (dpi), the total bacterial load of culturable microbes in mosquitoes supplemented with WT Enterobacter was significantly reduced when reared on sugar supplemented with antibiotic (T test; day 4 P < 0.05), and no CFUs were recovered after at 6 dpi (T test; day 6 P < 0.001) (Figure 6H).
Discussion
We harnessed the CRISPR/Cas9 gene editing system to create knockout mutants in an Enterobacter gut symbiont of Ae. aegypti mosquitoes enabling us to examine the role of bacterial genetics, specifically the ompA gene, in biofilm formation and gut colonization. A deletion of the ompA gene of Enterobacter decreased bacterial colonization of the mosquito host at both the larval and adult stages after infection in a mono-association. Strikingly, we found this effect was most pronounced in adult mosquitoes with more than half of the mosquitoes not possessing any culturable mutants, while there was no difference in prevalence of infection between the mutant and WT bacteria in larvae. The reduced prevalence of mutant bacteria in adults likely reflects differences in microbial colonization of each mosquito life stage. Larvae are continually subjected to bacteria in the larval water habitat while adults only have a short time frame to acquire bacteria immediately after eclosion, when they are thought to imbibe a small amount of larval water which seeds the gut with microbiota67. Our data shows greater variation in colonization of the adult stage between the mutant and WT strains, indicating that the ompA gene, and potentially bacterial factors in general, may be critical for colonization of the adult gut. These findings are also consistent with other sequence-based studies, that indicate adult stages have greater variability in species composition of their microbiota, while the microbiome of immature stages reflects the larval water habitat1-8.
Overall, our findings are similar to studies done in Tsetse flies whereby an ompA mutant of Sodalis, an Enterobacteriaceae symbiont, has impaired biofilm formation and reduced colonization rates22,23. These studies, in conjunction with our work, suggests that the ompA gene is imperative for symbiotic associations within dipterans. It also suggests that biofilm formation may be a strategy employed by bacteria to colonize the gut of insects. In pathogenic infections in mammals, biofilms enable bacteria to colonize new niches, promote infection and are associated with virulence68,69. Although less is known regarding the importance of biofilm formation in insects, in an artificial Pseudomonas-Drosophila infection model, biofilm formation was associated with virulence and host survival20. In a natural symbiotic association between bean bugs and Burkholderia, disruption of a purine biosynthesis gene in the bacterium also reduce biofilm formation and colonization of the insect65. In mosquitoes, gut biofilm formation could also have implications for vector competence as Chromobacterium, which was isolated from Aedes mosquitoes, produced molecules that inhibited dengue virus only when grown in vitro as a biofilm and not when grown in a planktonic state70. However, it is unknown if biofilm formation occurred in vivo70. Our data provide evidence that biofilms occur within the gut of mosquitoes and facilitate host colonization.
While we have shown that the ompA gene of Enterobacter is important for host colonization, we see no evidence that deletion of this gene alters mosquito development or growth rates. This is in contrast to the Riptortus-Burkholderia symbiosis whereby mutation of the purT gene in Burkholderia resulted in reduced growth rates and reduction in body weight of the host compared to insects that were infected with the WT bacterium65. The difference in our study to the findings in the Riptortus-Burkholderia symbiosis could be related to different requirements of the bean bug compared to the mosquito host as well as the different genes mutated in the symbionts. Our findings are consistent with another study in mosquitoes whereby an ompA mutant of E. coli was reported to not influence growth when a genomic mutant library was screened in Ae. aegypti in a mono-association71. Using a similar gnotobiotic system that exploits the ability to sterilize mosquito eggs and rescue development of the insect by supplementation, several recent reports describe approaches to create bacteria-free mosquitoes62,63. Here, we reared mosquitoes in a mono-association where they were only subjected to Enterobacter. However, more than half the adult mosquitoes inoculated with the ΔompA mutant were not infected by bacteria, but nevertheless, had similar development and growth rates compared to mosquito possessing WT bacteria. It would be intriguing to determine if these uninfected mosquitoes were bacteria-free. If so, the use of mutant bacteria that rescue development but cannot colonize mosquitoes may provide a simple means to create sterile bacteria-free mosquitoes.
CRISPR/Cas9 gene editing has revolutionized genetic approaches in model bacteria32-35, and while the use of this approach is expanding to other non-model bacteria37-42, to our knowledge, there are no examples of editing undertaken in symbiotic microbes. Here we demonstrate that editing approaches functional in E. coli can be applied to phylogenetically related symbiotic bacteria that associate within the guts of mosquitoes. Our overall goal was to delete the ompA gene in Enterobacter and therefore we altered our editing strategy when carrying out experiments in the non-model Enterobacter compared to our initial attempts in E. coli, and as such it is difficult to compare rates of editing between bacterial species. Nevertheless, the number of mutant colonies was considerably greater in Enterobacter compared to E. coli indicating the editing approach is applicable to this symbiotic Enterobacteriaceae bacterium. Similar to our findings, a two-plasmid CRISPR/Cas9 system, exploiting the use of λ-Red recombineering, developed in E. coli was also functional in the Enterobacteriaceae, Taumella citrea, without the need for modification36. Taken together, these findings suggest that CRISPR/Cas9-based approaches may be applicable to a wide range of Enterobacteriaceae. Similar to the original study in E. coli35, we observed counterselection escapees in our editing experiments. The no-SCAR approach uses a counterselectoin method to edit bacteria and cells that possess both the Cas9 and the gRNA should not be viable as a result of double stranded breaks in their genome. As the genetic basis of counterselection escape remains unclear and our primary objective was to create a mutant strain, we did not pursue the mechanism behind these escapees further.
Previous integration attempts in E. coli using the no-SCAR approach inserted a small fragment (80 bp) into the bacterial genome35. It was hypothesized however, that this approach could be used to integrate larger sized fragments over 1 kb into the genome based on the efficiency of integration and counterselection escape rates, although screening would be required35. Here we demonstrate this is indeed feasible as we inserted 1.3 kb and 1.5 kb fragments into the Enterobacter genome, and importantly, these genes, driven by the AmTr promoter72, were functional both in vitro and in vivo. Our work expands the list of bacteria amenable to CRISPR/Cas9 based genome editing approaches and provides an elegant tool to investigate specific bacterial genes that influence host-microbe interactions in mosquito vectors.
The application of CRISPR/Cas9 genome editing to gut-associated bacteria of mosquitoes has significant applied potential. Paratransgenesis strategies are being evaluated in a range of medical and agricultural systems to mitigate pathogen transmission from insect vectors, however, most approaches engineer symbionts by plasmid transformation47-55 and where genome integration has been accomplished in symbionts56-61, it has been done with technologies that did not allow for site specific integration. Here, we demonstrate site-specific integration of transgenes into the bacterial symbiont genome. Paratransgenic approaches suitable for use in the field will need to stably integrate genes into the bacterial genome in a manner that does not compromise bacterial fitness. As such, the use of CRISPR/Cas9 to engineer specific sites in the bacterial genome, such as intergenic regions within the genome, will undoubtedly be beneficial for these applied approaches.
In summary, we have demonstrated that the CRISPR/Cas9 gene editing system can be applied to non-model symbiotic bacteria that associate with eukaryotic hosts to interrogate the role of bacterial genes in host-microbe associations. We created knockout and knockin mutants by deleting and disrupting the ompA gene of Enterobacter. The knockout mutant displayed a reduced ability to form biofilms and colonize the gut of Ae. aegypti mosquitoes in a mono-association, demonstrating bacterial genetic factors are important determinants that influence colonization of mosquito guts. Aedes mosquitoes are becoming powerful systems to investigate the genetics of host-microbe interactions given the scientific community has simple and efficient approaches to alter both the microbes (this study) and mosquito host genome73,74 at their disposal, as well as methods to create mono-associated bacterial lines64. Finally, rapid, efficient, and site specific gene editing approaches for gut bacteria that associate with mosquitoes will facilitate the development of novel paratransgenic approaches to control arthropod-borne disease.
Materials and Methods
Bacterial and mosquito strains
E. coli BL21(DE3) (NEB) and an Enterobacter strain previous isolated from a lab-reared colony of Ae. albopictus (Galveston) mosquitoes7 were used in this study. Cultures were grown in liquid LB media at 37°C with the appropriate antibiotic unless stated otherwise. Mosquitoes were reared in the UTMB insectary under conventional conditions or in mono-associations (described below).
CRISPR gene editing
Editing the ompA gene of E. coli and Enterobacter (Supplementary Table 1) was complete as described in Reisch and Prather35. The photospacer sequence for the ompA gene was designed using the CHOPCHOP75, and cloned into pKDsgRNA-ack plasmid35 directly upstream of gRNA scaffold using REPLACR mutagenesis protocol76 (Supplementary Table 2). The plasmids were acquired from Addgene (Supplementary Table 1; Addgene plasmid 62655 and 62654). The resulting plasmids pKDsgRNA-Ec-ompA and pKDsgRNA-Ent-ompA were Sanger sequenced to confirm insertion of photospacer sequence. These plasmids were then transformed into either E. coli or Enterobacter containing the pCas9-CR4 plasmid. Transformants were selected at 30°C on LB agar plate containing spectinomycin (50 µg/mL), chloramphenicol (34 µg/mL), and with or without anhydrotetracycline (aTC) at 100ng/mL. Colonies from the –aTC plate were grown overnight in LB broth with the appropriate antibiotic at 30°C. A 1:100 diluted overnight culture was (grown until 0.4 OD600) supplemented with 1.2% arabinose to induce the expression of λ-Red recombinase. Cells were then transformed with 1-1.5 µg of double stranded donor DNA that flanked the PAM site for homologous recombination. Donor DNA was created by either PCR amplification or by gene synthesis (Genewiz). Regardless of the method of construction, each donor had flanking regions of 250 bp homologous to the target DNA. The resulting colonies were screened for mutations by colony PCR with primers flanking the integration site and positive clones were Sanger sequenced. Positive colonies were grown in LB broth and genomic DNA was isolated. For further validation, the flanking regions of deletion or insertions were amplified and the PCR product Sanger sequenced.
Stability of insertion
The stability of the knockout ΔompA mutant and the knockin ompA∷gentamicin and ompA∷mCherry strains was assessed in LB medium. The ompA∷mCherry and knockout ΔompA mutant cultures were grown for 10 passages in LB broth. At each passage 40 µl of culture was transferred into 4ml fresh LB medium. The ompA∷gentamicin strain was grown with or without gentamicin (50 µg/mL). Genomic DNA was isolated from the 0, 2, 4, 6, 8 and 10th subculture and PCR that amplified across the integration site was performed.
Complementation of ompA mutant
Functional rescue of the ompA mutation was achieved by complementing the mutant with the WT gene. The WT ompA gene was amplified from Enterobacter genomic DNA and cloned into the pRAM-mCherry vector7 and thereby creating pRAM-mCherry-Ent-OmpA. The Sanger sequence-verified plasmid was transformed into the ΔompA mutant, thereby generating the ΔompA/ompA complement strain. Colonies that acquired the plasmid were selected on LB plates containing kanamycin (50 µg/mL).
In vitro characterization of Enterobacter strains
To assess the impact of the gene deletion on bacterial growth the WT, ΔompA mutant and ΔompA/ompA complement were grown in LB broth and the density of bacteria (OD600) was quantified by spectrophotometer. A 1:100 dilution of an overnight culture was inoculated into a 5 ml LB broth in 50 ml tube and incubated at 37°C for 24 hrs. At 2, 4, 6, 8, 10, 12 and 24 hours growth was recorded at OD600. The biofilm assay was performed as described previously77,78. Briefly, biofilm formation by Enterobacter strains was quantified on polystyrene microtiter plates after 72 h of incubation at 37°C by CV staining. Three independent experiments were performed, and the data were represented as CV OD570 after normalizing by CFUs.
Mosquito infections
Mono-association in Ae. aegypti mosquitoes were done using gnotobiotic infection procedure64, with slight modifications7. Briefly, mosquito eggs were sterilized for 5 min in 70% ethanol, 3 min 3% bleach+0.01% Coverage Plus NPD (Steris Corp.), 5 min in 70% ethanol then rinsed three times in sterile water. Eggs were vacuumed hatched for 30-45 min and left overnight at room temperature to hatch any remaining eggs. Exactly twenty L1 larvae were transferred to T175 flask containing 60 ml of sterile water and fed on alternative days with 60 µl of fish food (1 µg/µl). Larvae were inoculated with 1×107/ml of either the WT Enterobacter, the ΔompA mutant or the ΔompA/ompA complement. The WT and ΔompA strains were transformed with the pRAM-mCherry plasmid that conferred resistance to kanamycin (but did not possess a functional ompA gene). L4 larvae were collected, washed three times with 1X PBS, and then homogenized in 500 µl of 1X PBS and 50 µl of homogenate was plated on LB agar containing 50 µg/mL kanamycin. Similarly, adult mosquitoes were collected 3-4 days post emergence and bacterial infection was quantified in the same manner as larvae. In order to assess the growth of the mosquitoes, time to pupation and growth rate were observed. Time to pupation was determined by quantifying the number of pupae each day post hatching, while survival to adulthood was calculated by quantifying the number of L1 larvae that reached adulthood. The experiment was repeated three times.
Knock-in mutants were administered to adult Ae. aegypti in a sugar meal. Three to four day old mosquitoes were fed with 1×107 of WT or the ΔompA∷gentamicin strain for three days in 10% sucrose solution. After three days, mosquitoes were either administered sugar supplemented with gentamicin (50 µg/mL) or sugar without antibiotic. CFUs were determined at days 0, 2, 4, and 6 dpi by plating homogenized mosquitoes (N=10) on LB agar. Similarly, the ΔompA∷mCherry and WT Enterobacter were fed to mosquitoes and midguts were dissected to assess the colonization of bacteria in the tissue. For visualization of bacteria, midguts were fixed in 1% paraformaldehyde (PFA) in 1X PBS for 30 minutes and permeabilized with 0.01% Triton X-100 in 1X PBS for 20 min. The tissues were stained with 1:250 diluted Phalloidin (Sigma) for 20 minutes and samples were washed twice with 1X PBS for 10 minutes. Finally, midguts were then stained with 1:500 diluted DAPI (Invitrogen) for 10 min. Samples were transferred to slides and mounted with ProLong™ Gold Antifade (Invitrogen). The slides were observed under Revolve FL (ECHOLAB).
Competing interests
The authors declare no competing interests.
Acknowledgements
We would like to thank the UTMB insectary core for providing mosquitoes. GLH was supported by NIH grants (R21AI138074, R21AI124452 and R21AI129507), a University of Texas Rising Star award, the Western Gulf Center of Excellence for Vector-borne Diseases (CDC grant CK17-005), the Robert J. and Helen Kleberg Foundation and the Gulf Coast Consortia. This work was also supported by a James W. McLaughlin postdoctoral fellowship at the University of Texas Medical Branch to SH and a Dutsadi Piphat Scholarship and Thailand Research Chair Grant from NSTDA (p-15-50004) to PN. We also wish to thank Dr Ulrike Munderloh (University of Minnesota) for the kind gift of the pRAM plasmid.