Role of plants in the transmission of Asaia sp., which potentially inhibit the Plasmodium sporogenic cycle in Anopheles mosquitoes

Biological control against malaria and its transmission is currently a considerable challenge. Plant-associated bacteria of the genus Asaia are frequently found in nectarivorous arthropods, they thought to have a natural indirect action on the development of plasmodium in mosquitoes. However, virtually nothing is known about its natural cycle. Here, we show the role of nectar-producing plants in the hosting and dissemination of Asaia. We isolated Asaia strains from wild mosquitoes and flowers in Senegal and demonstrated the transmission of the bacteria from infected mosquitoes to sterile flowers and then to 26.6% of noninfected mosquitoes through nectar feeding. Thus, nectar-producing plants may naturally acquire Asaia and then colonize Anopheles mosquitoes through food-borne contamination. Finally, Asaia may play an indirect role in the reduction in the vectorial capacity of Anopheles mosquitoes in a natural environment (due to Plasmodium-antagonistic capacities of Asaia) and be used in the development of tools for Asaia-based paratransgenetic malaria control.

In the context of malaria pre-elimination, the emergence of resistance in both vectors and parasites threatens the progress made in recent decades. The search for alternative methods is becoming a priority for the WHO and the scientific community. The vector environment has very often been neglected during the implementation of control strategies.
The present study focuses on the role of nectar-producing plants in reducing the vectorial capacity of malaria-transmitting mosquitoes. We found that flower nectar is home to viable strains of Asaia, known for their natural antagonism to Plasmodium falciparum, and we isolated Asaia strains from Ipomoea pes-caprae. Moreover, we found that bacterial exchanges between plants and mosquitoes are common in the natural environment and thus discovered the natural Asaia/Anopheles/flower nectar cycle. Through these exchanges, plants, as the habitat for Plasmodium-suppressing Asaia bacteria, may contribute to the reduction in the vectorial capacity of Anopheles (which has already been empirically reported). It was recently reported that an Asaia-based paratransgenetic malaria control strategy may also rely on the Asaia natural cycle as a delivery tool.
Particular attention should be paid to the environment in which mosquitoes are founding the development of malaria control strategies.
Malaria continues to be one of the most common causes of death on earth. In addition, its transmission is only partially controlled due to the risk of the emergence of insecticide-resistant mosquitoes. This information justifies the search for alternative and nonpolluting solutions to control mosquito-borne infections, especially since, with the risk of global warming, the area affected by malaria may increase. In this context, the search for biological control agents against malaria and its transmission is a considerable challenge. The simultaneous presence of Asaia and Plasmodium falciparum has not yet been reported in mosquitoes suggesting antagonism 1,2 . Thus, identifying the source of Asaia contamination in Anopheles may help to determine how to apply biological control to limit Anopheles infestation by P. falciparum. Recently, a paratransgenetic strategy was proposed based on Asaia bacteria that conditionally expressed an anti-plasmodial protein only when a blood meal was present 3 .

Results
Antibiogram and selective medium preparation. The resistance status of the previously isolated A. aff. bogorensis GD01 strain 2 to different antibiotics was assessed ( Table 1). The data obtained were then used to prepare an antibiotic-containing Columbia-based selective solid medium for the isolation of Asaia strains from naturally infected flowers and to obtain an Asaia-free mosquito line from a laboratory-reared colony of An. coluzzii.
In summary, A. aff. bogorensis GD01 was resistant to most families of the antibiotics tested (Table 1), except piperacillin-tazobactam, doxycycline, and carbapenems (doripenem and imipenem). Among them, gentamycin was used for the Columbia-based selective solid medium, and doxycycline was chosen to obtain our Asaia-free mosquito line.
natural Asaia spp. infection in wild flowering plants. Flowers were collected from 11 endemic and widely distributed plant species (Table 2) across the study area, including those mainly encountered inside and around the villages of Dielmo and Simong. Among the collected flowers, only those of Datura metel 25% (2/8) and I. pes-caprae 25% (28/114) were found to be contaminated with Asaia spp. I. pes-caprae were collected at three different locations (Table 2); this was the most targeted plant due to its wide distribution in the study area and its proximity to mosquito breeding sites.
The colonies cultivated from all the infected specimens of I. pes-caprae that were morphologically similar to those of A. aff. bogorensis GD01, previously isolated by our team from An. gambiae s.l. collected from Dielmo (Senegal), were confirmed by qPCR 2 . The qPCR molecular screening showed that two colonies out of a total of 20 morphologically similar colonies isolated on Asaia-selective antibiotic-containing solid medium tested positive for A. aff. bogorensis. The subsequent phylogenetic analysis based on the sequences of almost the entire 16 S rRNA gene confirmed that the isolated strains were A. bogorensis, with slight differences from A. aff. bogorensis GD01, which was previously isolated from An. gambiae s.l. in the same village ( Fig. 1), forming a clade within the A. bogorensis cluster.
www.nature.com/scientificreports www.nature.com/scientificreports/ natural prevalence of Asaia spp. in a laboratory mosquito strain. A subsample of 158 An. coluzzii (130 females and 28 males) of the laboratory strain were randomly selected and screened to determine the prevalence of natural infection by A. aff. bogorensis in our experimental study population. The results showed a global carriage rate of A. aff. bogorensis up to 61% (96/158), which was significantly higher among the female group (74%,96/130) than among the male group (46.4%,13/28)(χ 2 = 6.864, df = 1, p = 0.008). The naturally infected An. coluzzii laboratory strain was used as the Asaia-positive line in the following experiments. experimental infection model. qPCR screening of newly opened Pseuderanthemum reticulatum flowers used in the experimental infection revealed the absence of natural Asaia infection. The plants were then isolated and protected from contact with arthropods. A batch of 100 An. coluzzii individuals, randomly selected from a naturally Asaia-infected laboratory colony, were exposed to P. reticulatum to confirm the effectiveness of   www.nature.com/scientificreports www.nature.com/scientificreports/ transmission. During the entire experimental period, the only source of food for mosquitoes was the plant nectar, to which they were exposed for 3 consecutive days. Screening for Asaia infection on 36 randomly selected open flowers revealed the presence of Asaia spp. in 31/36 (86.1%) of the flowers. This shows that mosquito nectar feeding could be a significant route of bacterial contamination of plant flowers.
PCR screening of a randomly selected subsample of 56 anopheles (30 females and 26 males) out of the 200 An. coluzzii treated with antibiotics for 3 days to establish an Asaia-free lineage revealed no infection, thus confirming the Asaia-free status of the batch. The remaining females were reared to obtain the first generation (F1) of the lineage used for the following experiments. Samples of the Asaia-free lineage, selected randomly and in a timely manner, were regularly tested to verify whether reinfection with Asaia occurred. The lineage remained pure and uncontaminated throughout the experiments.
The subsequent exposure of two replicates of 265 An. coluzzii specimens belonging to the Asaia-free laboratory colony to P. reticulatum plants previously infected by the Asaia-positive lineage revealed a contamination rate of 27% (71/265) ( Table 3). The infection rate was significantly lower in the female group (16.0%, 24/153) than in the male group (42.0%, 47/112) (χ 2 = 21.45, df = 1, p < 0.05). These results have enabled us to propose a transmission cycle of Asaia spp. among mosquitoes in the natural environment (Fig. 2).

Discussion
Arthropods maintain a complex relationship with the elements present in their natural environment. The impact of these elements on the transmission of vector-borne diseases has long been under-researched. However, nectar-producing plants play a very important role in the transmission of pathogens hosted by nectarivorous arthropods. Nectar is one of the main sources of food for insects and helps to increase their survival in the natural environment. As in larval breeding sites, where some bacteria belonging to the microbial flora can be ingested by mosquito larvae 13,24 , bacteria present in plant nectar can also pass into adult mosquitoes during nectar feeding. Bacterial species, such as those belonging to the Asaia genus, represent great potential for use in biological malaria control, especially in the current context of the emerging resistance of P. falciparum to artemisinin [25][26][27] and of the main vectorial species of malaria to insecticides [28][29][30][31] . Asaia bacteria belong to the Acetobacteraceae   www.nature.com/scientificreports www.nature.com/scientificreports/ family and are often isolated from fermented foods or plants 32,33 . They have also been described as symbionts of insects that often feed on natural sources of sugar 34,35 .
Asaia spp. could be modified by trangenesis to obtain a strong antagonistic effect against P. falciparum 9,36 and may be able to protect mosquitoes against plasmodial infection. This antagonistic effect was first highlighted by Favia et al. 9,37 . Furthermore, the Asaia bacterium has been found to be stably associated with the microbiota of larvae and adults of An. stephensi, An. maculipennis and An. gambiae 37,38 . This bacterium has also been found to be stably associated with the microbiota of the laboratory strain of An. coluzzii, and naturally infect An. funestus 2 . Online searches in the metagenomic 16 S databases have also shown the presence of Asaia in different nectarivorous insects, such as butterflies and bees, in which the most likely propagation pathways identified were the venereal route, paternal transmission 39,40 and food ingestion 41 .
Asaia, a genus of acetic acid bacteria was first discovered and isolated from plants, namely, Bauhinia purpurea and Plumbago auriculata flowers 42 . Currently, only 8 species of Asaia have been described: Asaia astilbis 43 , Asaia bogorensis 42 , Asaia krungthepensis 18 , Asaia lannensis 44 , Asaia platycodi 43 , Asaia prunellae 43 , Asaia siamensis 16 , and Asaia spathodeae 45 . To our knowledge, no studies have been able to highlight the role of plants in the contamination of arthropods by Asaia spp. In the study of tularemia, scientists have shown through an experimental model that nectar may play a key role in the spread of the disease. Indeed, Francisella tularensis has the ability to survive in nectar, thus facilitating its ingestion by vectors, which are then infected through the nectar feeding route, thus ensuring the transmission of the bacteria 14 . Asaia strains were found in the nectar of some plants in this study, suggesting that infected plants could be the missing link to the maintenance of mosquito infection in nature. Our hypothesis is reinforced by previous reports on the ability of bacteria to survive in plant nectar 14  Together, the high prevalence of Asaia spp. in plants in the direct environment of mosquitoes (I. pes-caprae growing close to mosquito breeding sites and D. metel found in the vicinity of human dwellings, which also serving as adult mosquito resting sites) and the transmission of Asaia from infected to noninfected An. coluzzii mosquitoes via flower nectar provide enough evidence to support our hypothesis. Based on this evidence, we suggest the transmission cycle for Asaia spp. in the natural environment (Fig. 2). In addition to mosquitoes, the carriage of Asaia spp. is widespread in nectarivorous insects in nature. Indeed, Asaia spp. have already been isolated or identified in Anopheles mosquitoes 46   During the diet or oral (potentially co-feeding) transmission phase (B), mosquitoes are contaminated or contaminate the plants they feed on. Then, during the copulation or sexual transmission phase (C), male or female mosquitoes can contaminate their partners. The fertilized females will search for a blood meal to ensure the maturation of their eggs (D), and the males will continue to feed on nectar from plants, such as I. pes-caprae, to satisfy their sugar needs and maintain the oral transmission cycle (E). After a blood meal, the females will lay eggs (F), some of which will be contaminated by Asaia spp. as well as larvae (known as the transstadial transmission phase).The infected emerging adults will initiate and maintain the transmission cycle.
In the context of the spread of insecticide resistance and the increased need for innovative approaches to control vectors, the development of genetically modified mosquitoes (GMMs) that are resistant to pests has emerged as the most promising alternative to insecticide interventions 49 . However, the GMM approach has one main limitation: it modifies the fitness and competitiveness of mosquitoes in relation to natural mosquito populations 9 . Given the recent evidence of the stable natural infection of malaria vectors by several bacteria that may impede Plasmodium development, Asaia spp. have become one of the most promising potential paratransgenic weapons against malaria 9 . Asaia spp. are easily cultivable and can be genetically manipulated to deliver anti-pathogen effects on molecules that are then reinserted into their insect host 36 . A prominent model of the paratransgenetic malaria control strategy based on Asaia bacteria has been recently proposed 3 . Asaia bacteria have been genetically modified to release scorpine into the intestinal lumen of An. stephensi, leading to the elimination of P. berghei.
We have noticed that in laboratory colonies of An. coluzzii, male mosquitoes are more likely than females to be infected by Asaia. We therefore speculate that the females (probably fertilized) were in search of a blood meal for maturing their eggs and, therefore, consume less Asaia-infected nectar. However, this hypothesis was not tested in the present study.
In fine, this study is an important steppingstone toward the successful use of any Asaia-based vector control approach. Given the strong antimalarial capacities of these bacteria, the involvement of plants as a key link between nectarivorous insects, including several hematophagous arthropods that transmit vector-borne diseases, in the natural transmission cycle of Asaia bacteria offers an unprecedented opportunity for the direct or indirect use of Asaia in an alternative biological and/or transgenic approach to malaria control.

Material and Methods
Description of the study area and plant collection. Entire flowers were manually collected from endemic plants growing on the banks of the Nema River, next to Dielmo (13°43′23.43′′N, 16°24′46.27′′W) and Simong (13°37′55.63′′N, 16°23′08.24′′W) (Fig. 3). Dielmo is located in the Fatick region of Senegal, in an area of Sudanese savannah approximately 174 miles from Dakar and 9.32 miles north of the Gambian Republic. Since 1990, the area near Dielmo has hosted one of the oldest cohort studies investigating the epidemiological relationships between the malaria parasite and its vectors and human hosts 50 . Previous studies have already laid out the geographical and epidemiological characteristics as well as the dynamics of malaria in Dielmo 50,51 .
Members of the An. gambiae s.l. and An. funestus group are the main vectors of malaria in this area 52,53 . The flowers were collected from 27 to 30 October 2016. They were cut at the stalk level, placed in bags (one bag per plant), stored at 4 °C in the Dielmo laboratory and transported to Dakar at the same temperature. In Dakar, the samples were immediately frozen at −80 °C until they were processed to assess their molecular biology and for bacterial cultures.

Molecular analyses. DNA extraction.
The petals and sepals of each flower were removed, and then the peduncle was cut off. The remaining flower parts, including nectar and pistil, were immersed in an Eppendorf tube containing 600 μl of 1X phosphate buffered saline (PBS) and then crushed. After centrifugation, 200 μl of the supernatant was collected for DNA extraction, which was performed according to the 2% CTAB method 54 supplemented with β-mercaptoethanol at a concentration of 0.2%. The mixture was left to digest overnight. The same DNA extraction procedure was performed for the mosquitoes. Before DNA extraction, individual specimens were sacrificed at −80 °C, washed with 70% ethanol to remove superficial environmental bacteria and then www.nature.com/scientificreports www.nature.com/scientificreports/ crushed. Then, genomic DNA was immediately extracted from the whole body of each individual specimen as previously described 54 .
Asaia bogorensis molecular detection. Taqman ® qPCR methods with custom designed primers and probe sets were used to screen samples for the presence of Asaia spp. by targeting the rpoB gene (Fwd: 5′-GACGCCAAGGACCTGATCTA-3′; Rev: 5′-ATAGGCCAGGATTTCGTCCT-3′; Probe: 6-FAM-GGTCACGACCCTGCTCTATG-TAMRA) 2 . The targeted region was amplified in a total reaction volume of 20 μl, containing 10 μl of ROCHE ® master mix (Roche Diagnostics, Indianapolis, IN, USA), 0.5 μl of each primer, 0.5 μl of the probe, 3 μl of distilled water, 0.5 µl of UDG, and 5.0 μl of test DNA. The conditions of the amplification with the CFX96 Touch detection system (Bio-Rad, Marnes-la-Coquette, France) were as follows: 2 minutes at 50 °C for UDG action followed by an initial denaturation of 5 minutes at 95 °C, then 39 cycles of 5 seconds at 95 °C and 30 seconds at 60 °C. A sample was considered positive when the Ct value was less than 35 cycles. The Anopheles negative control was prepared from a crush of certified Asaia-free An. coluzzii, and the positive control was prepared from An. coluzzii to which we added a suspension of A. aff. bogorensis GD01.
16 S rRNA gene sequencing and phylogenetic tree construction. The molecular phylogenetic evolutionary history was inferred from a nearly complete 16 S rDNA sequence (1249 bp), which was sequenced using universal 16 S rDNA primers fD1 and rp2 55 , as described by Drancourt et al. 56 .
Initially, we used a Peltier PTC-200 model thermal cycler (MJ Research Inc., Watertown, MA, USA) to perform the PCR amplifications. The reactions were carried out using HotStar Taq polymerase (Qiagen, Hilden, Germany) following the manufacturer's instructions. For each assay, we included negative and positive controls for the validation of the run. The success of amplification was confirmed by electrophoresis on a 1.5% agarose gel.
The purification of the PCR products was performed using NucleoFast 96 PCR plates (Macherey-Nagel EURL, Hoerdt, France) according to the manufacturer's instructions. The products of the amplification were then sequenced using a Big Dye Terminator Cycle Sequencing Kit (Perkin Elmer Applied Biosystems, Foster City, CA) with an ABI automated sequencer (Applied Biosystems).
The amplified sequences were assembled and corrected using ChromasPro software (ChromasPro 1.7, Technelysium Pty Ltd., Tewantin, Australia) and then compared with reference sequences available in Genbank using the BLASTN server (http://blast.ncbi.nlm.nih.gov/Blast.cgi) to identify closely related species and/or strains.
The taxonomic relationships of the new strains were inferred against the existing isolates. The reference sequences retrieved from the Genbank database, together with the Senegalese strains, were aligned using the ClustalW multisequence alignment program 57 in BioEdit software 58 . A maximum likelihood phylogenetic tree was reconstructed using TOPALi v2.5 based on the Hasegawa-Kishino-Yano (HKY85) substitution model 59 , which includes the proportion of invariable sites and the gamma distribution. The robustness of the individual branches was estimated by bootstrapping with 100 replicates 60 . establishment of an Asaia-free An. coluzzii strain. Sensitivity of A. aff. bogorensis GD01 to antibiotics. The sensitivity of the A. aff. Bogorensis GD01 strain, previously isolated from An. gambiae from Dielmo village 2 , to antibiotics was determined. A pure suspension of A. aff. bogorensis GD01 at a concentration equal to the 0.5 McFarland standard was inoculated onto Mueller-Hinton agar medium 61,62 . Antibiotic (amoxicillin, amoxicillin/clavulanic acid, piperacillin/tazobactam, doripenem, imipenem, aztreonam, ceftriaxone, ceftazidime, cefpirome, gentamicin, doxycycline, erythromycin, trimethoprim/sulfamethoxazole, ciprofloxacin, linezolid, metronidazole and rifampicin) discs were deposited onto dried agar plates and the plates were then incubated at 28 °C under 5% CO 2 atmosphere. Finally, the inhibition diameters were measured and compared with the reading charts according to the manufacturer's instructions (i2a -Siège Social, 401 Avenue du Walhalla, CS83406, 34060 Montpellier Cedex 2,France) and the recommendations of the Antibiogram Committee of the French Society of Microbiology 63 .
Isolation of strains of Asaia spp. from wild plants. Suspensions of the qPCR-positive I. pes-caprae were seeded on Columbia agar supplemented with 5% fresh sheep blood and 4 mg/l gentamicin (batch No. F140512, manufactured by Xin K. Pharm Co. Ltd). The seeded medium was then incubated at 28 °C under an atmosphere of 5% CO 2 . The colonies morphologically resembling Asaia were seeded on another Columbia agar plate supplemented with 5% fresh sheep blood but without antibiotics and incubated under the same conditions. Approximately 50 purified colonies were then collected and suspended in 200 μl of sterile 1X PBS and used for bacterial DNA extraction using the 2% CTAB method supplemented with 25 μl of proteinase K (Tritirachium album, 25 mg, ref. EUC0090-A, EUROMEDEX, 24, rue des Tuileries BP684 67460 Souffelweyersheim, France). Finally, the extracted DNA was screened for the presence of Asaia using A. bogorensis-specific qPCR.
Treatment of adult mosquitoes with antibiotics. Adult An. coluzzii colonies maintained at a temperature of 26 ± 2 °C and a relative humidity of 80 ± 10% in our laboratory, as described previously [50][51][52][53] , were fed a mixture of sterile water containing doxycycline (batch No. B113304 manufactured by SERB Laboratories) at a concentration of 8 mg/l and 10% sucrose solution for 4 days. qPCR was performed on the treated mosquitoes to confirm their Asaia-free status.
Insectary rearing of the Asaia-free mosquito strain. The Asaia-free mosquito line, fed sugar solution prepared with autoclaved water, was maintained at a temperature of 26 ± 2 °C and a relative humidity of 80 ± 10%. When feeding the mosquitoes, a volume of 50 ml of autoclaved water was filtered with a 0.2 µm filter and then mixed with sterile sucrose solution at a concentration of 10%. This mixture was offered daily to mosquitoes and changed daily to prevent mosquito reinfection. The physical conditions of breeding were the same as those of the other mosquitoes in the insectarium. The control of the absence of Asaia in this mosquito line was assessed weekly and before each experiment using our Asaia-specific custom-designed qPCR system.

Design of the experimental infection through transmission of A.aff. bogorensis.
The two An. coluzzii colonies (Asaia-infected and Asaia-free), maintained separately in the insectarium of the VITROME laboratory of the Institut de Recherche pour le Développement (Dakar, Senegal), were used for the experimental infection of A. aff. bogorensis transmission. The common ornamental plant P. reticulatum (Acanthaceae) was selected as the source of nectar during the experimental infection study after the confirmation of its natural Asaia-free status and due to its easy procurement from urban horticulturists in Dakar, where our laboratory is based. The plants were grown in pots within a closed room with no access to flying arthropods to prevent the contamination of newly opened flowers with Asaia. Randomly selected flowers from each plant were screened for the presence of Asaia spp. and only plants that were negative were used for the subsequent experimental infection.
Overall, 310 mosquitoes of the naturally Asaia-infected laboratory colonies (123 males and 187 females) were brought into contact with an uninfected P. reticulatum bearing newly opened flowers. The plant in its pot was introduced into a closed 100 × 50 × 50 cm mosquito cages for 3 days. During the experimental infection, the average temperature of the standard thermo-hygronomic device was 26 ± 2 °C, and the relative humidity was 80 ± 10%. At the end of the third day, the mosquitoes were sacrificed at −80 °C and then screened with randomly selected flowers using our qPCR method to confirm that the mosquitoes and the flowers of the exposed plants were infected with Asaia. The following day, an unfed batch of mosquitoes from the Asaia-free lines was allowed to feed for 3 days on the Asaia-infected plants under the same conditions described above. After exposure, the mosquitoes were sacrificed, stored individually in Eppendorf tubes for DNA extraction and then examined for Asaia infection. This experiment was repeated twice with the same number of anopheles for both lineages and the same plant species under identical physical conditions.

Statistical analyses.
This study was carried out over two consecutive years. Two groups of mosquioes were used each time. This first group was the An. coluzzii strain naturally infected by A. aff. bogorensis and the second was as An. coluzzii Asaia-free group obtained after antibiotic treatment and then exposed to infected P. reticulatum. Within each group, we analyzed the differences in infection rates between the two sexes. Statistical analyses were performed with Epi Info software version 7.0.8.8.0 (Centers for Disease Control and Prevention, Atlanta, GA, USA). The differences were analyzed using the Yates corrected χ2 test with one degree of freedom and a 95% confidence interval. The difference was considered significant when the bilateral p value < 0.05