A Genetically Modied Anti-Plasmodium Bacterium Is Harmless to the Stingless Bee Partamona Helleri

Paratransgenesis consists of genetically engineering an insect symbiont to control vector-borne diseases. Biosafety assessments are a prerequisite for the use of genetically modied organisms (GMOs). Assessments rely on the measurement of the possible impacts of GMOs on different organisms, including benecial organisms, such as pollinators. The bacterium Serratia AS1 has been genetically modied to express anti-Plasmodium effector proteins and does not impose a tness cost on mosquitoes that carry it. In the present study, we assessed the impact of this bacterium on the native bee Partamona helleri (Meliponini), an ecologically important species in Brazil. Serratia eGFP AS1 (recombinant strain) or a wild strain of Serratia marcescens were suspended in a sucrose solution and fed to bees, followed by measurements of survival, feeding rate, and behavior (walking and ying). These bacteria did not change any of the variables measured at 24, 72, and 144 h after ingestion. Recombinant and wild bacteria were detected in the digestive tract up to 144 h after ingestion, but their numbers decreased with time. The recombinant strain was detected in the midgut at 24 h and in the hindgut at 72 h and 144 h after ingestion. As reported for mosquitoes, Serratia eGFP AS1 is a safe candidate for combating vector-borne diseases, as it did not compromise the native and ecologically relevant bee, P. helleri.


Introduction
Malaria is one of the deadliest infectious diseases worldwide. Plasmodium parasites, the causative agents of malaria, are transmitted to humans through the bite of infected female Anopheles mosquitoes. Control of malaria is based primarily on reducing vector populations with insecticides and using antimalarial drugs [1]. These tools have been ineffective due to the development of mosquito insecticide resistance and parasite drug resistance [1]. The development of new tools to combat this disease is of high priority.
Paratransgenesis, the genetic manipulation of insect symbiotic microorganisms to block pathogen transmission, is a promising strategy for controlling insect-borne diseases. Its effectiveness is enhanced by the fact that bacteria share the same compartment, the midgut, with the pathogens transmitted by the insects and because bacterial numbers increase dramatically following a blood meal [2,3]. The facultative aerobic and gram-negative rod-shaped bacteria of the genus Serratia (Enterobacteriaceae) are common components of the midgut microbiota. This genus is a symbiont of many arthropods, such as mosquitoes, bees, sand ies, ticks, and aphids [4][5][6][7][8][9].
The use of any genetically modi ed organism (GMO) for biological control should impose minimal tness cost to its insect carrier [2].
Moreover, a thorough risk assessment to the environment is required prior to introduction in the eld [17]. These assessments include investigating the transfer routes of GMOs, which can be vertical (from mother to offspring), transstadial (between developmental stages), and horizontal (from one individual to another, without being parental), as well as from the effects of the GMO on behavior, survival, and reproduction of potential hosts [10,11].
The horizontal transfer of the GMO between organisms can occur through the sharing of common resources [10,18,19], for example, by water contaminated with the GMO [12]. Normative institutions, such as CTNBio (National Technical Commission of Biosafety, Ministry of Science, Technology and Innovation, Brazil), also require the assessment of the interaction of paratransgenic individuals with the environment, which includes a particular concern regarding the possible harmful effects of GMO on non-target organisms [17], including pollinators [20]. Since bees, as pollinators, play a signi cant role in maintaining biodiversity [21,22], these organisms are widely recognized for integrating risk assessment protocols [23][24][25][26].
Forager bees perform out-colony tasks, including the search for food resources (i.e., water, ber, resin, nectar, and pollen), which are in direct contact with the external environment [27]. Therefore, risk assessments are preferably carried out on foragers and include assessments of lethality and, to a lesser extent, sublethal effects [28,29]. Such assessment studies are mostly carried out with the honey bee Apis mellifera as a model organism, but this species is exotic in Neotropical environments such as South America [29,30]. In this sense, stingless bees (Meliponini) are more representative of Neotropical ecosystems [22,29] as pollinators of native and cultivated plants [31]. Therefore, stingless bees should be considered in studies to measure the potential risks of GMOs.
The ability of bees to withstand environmental stressors is linked to the gut microbiota [32,33]. In addition, the gut microbiota can in uence the behavior, metabolism, growth, and development of hosts [34,35]. The microbiota is highly conserved among several species of stingless bees [36]. The stingless bee Partamona helleri (Meliponini) has a wide range of dominant bacterial genera (approximately 33), and the genus Serratia has also been found in this species [8]. Certain Serratia species can also be pathogenic to bees, as was observed for the S. marcescens sicaria strain (Ss1) in honeybee adults [37].
This work evaluated the risk to adult P. helleri workers of ingestion of the genetically modi ed Serratia eGFP AS1, carrying anti-Plasmodium effectors. We investigated survival, food ingestion (i.e., feeding rate), and walking and ight activities of adults as it relates to the evaluation of safety to the environment.

Bees
Foragers from ve colonies of P. helleri, 25 or more days old [27] were obtained from the Central Apiary at the Universidade Federal de Viçosa (UFV), Viçosa -MG (20º 45 '14' 'S; 42º 52' 55″) under license ID75536 (ICMBio -SISBIO, Ministério do Meio Ambiente, Brazil). The foragers were captured with a glass bottle at the hive entrance and immediately transported to the Insect Molecular Biology Laboratory (UFV) without controlling the temperature or luminosity. The bees were anesthetized with carbon dioxide for 5 s and then transferred to 500 mL round transparent plastic pots. The foraging bees fasted for 1 h in an incubator (28 ± 1°C, 70 ± 5% relative humidity (RH), in the dark) until the bioassays began. This fasting period is necessary to stimulate the ingestion of food provided in oral exposure tests [38].

Exposure to the bacteria
The bacteria (recombinant and wild) stocks were thawed and cultured for 24 h in liquid LB medium at 28°C and diluted in autoclaved sucrose solution (50% v/v) to a nal concentration of 10 8 cells/mL, which was veri ed with the nal optical density (OD) 600 nm of 0.1. Different sucrose solutions were prepared to expose the bees to four separate treatments: 1) sugar only (control), 2) sterile LB with sucrose (1:1), 3) Serratia eGFP AS1 (recombinant) with sucrose, and 4) wild S. marcescens with sucrose. A total of 100 bees per treatment (20 bees from the same colony in a plastic pot × 5 colonies or replicates = 100 bees/treatment; n = 400 bees). Two independent experiments were conducted. The rst assessed survival and food consumption over 144 h (n = 400 bees). The second assessed the behavior at 72 h and tested for the presence of bacteria in the digestive tract (n = 400 bees) at 24, 72, and 144 h.
Each sucrose solution was offered to 20 bees in a 2 mL microtube drilled at the bottom and inserted through a hole in a 500 mL plastic pot [38]. After 24 h exposure, the sucrose solution was changed, and the bees received 50% sterile sucrose solution (v/v) for the next 120 h.

Survival and food consumption
Survival was monitored every 24 h for 144 h. Individuals were considered dead when they did not move after stimulation with forceps, and dead bees were discarded [38].
Food consumption was measured by weighing the microtubes on an analytical scale. The weights of the microtubes with the sucrose solution were recorded before feeding and then again before switching. The food, which contained the four different diets, was offered for 24 h. The tubes with the remaining food were then weighed. After 24 h, the microtubes were replaced with microtubes containing sterile food, and the weights of the microtubes were measured 48, 72, 96, 120, and 144 h after the beginning of the assay. Plastic containers without bees, but with microtubes containing sucrose (50% v/v), were kept under experimental conditions to estimate the losses by evaporation. These values have been used to correct food consumption [8,38,39].

Detection of bacteria in the digestive tract and ovary
The presence of Serratia eGFP AS1 and S. marcescens (wild) in the digestive tract of bees was assessed. Fifteen bees (3 bees per colony, totaling 5 colonies) were used for each treatment (i.e., different foods) for 24, 72, and 144 h (n = 180 bees). Anesthetized bees were kept on ice, sterilized for three min with 70% ethanol, and washed three times with sterile phosphate buffered saline (PBS; 0.1M, pH 7.2). The digestive tract (i.e., midgut and hindgut) was dissected, transferred separately to a microtube (1.5 mL), homogenized in 500 µL of sterile PBS, and serially diluted from 10 0 to 10 4 . From this homogenate, the micro-drop plating technique was completed using a 20 µL aliquot, which was plated in triplicate, collected for each sample, and transferred to plates with agar and LB medium containing 100 µg/mL kanamycin [40]. The plates were then incubated at 28°C for 24 h. The bacteria were identi ed according to their colony phenotype (Supp. Figure 1, 2). Colony GFP uorescence was detected using a UV transilluminator (High-Performance 2UVTM Transilluminator, λ = 365 nm).
The second set of bees from the four experimental groups was anesthetized and dissected in PBS.

Flight
The same foragers evaluated in the walking bioassay were evaluated using a ight bioassay. The assay consisted of releasing 5 bees at the bottom of a wooden tower (105 cm high × 35 cm long × 35 cm wide) covered laterally with the organza fabric. The assay was carried out in a dark room with a white lamp (60 W), 5 cm above the top of the tower. The bees were kept for 1 min at the base of the tower in a Petri dish for acclimatization and then released. The proportion of individuals that reached the top of the tower was counted [42]. Two groups of 5 individuals were analyzed for each treatment from the ve different colonies (n = 200). The average proportions of the two groups in each colony were used for statistical analyses.

Statistical analysis
Survival data were used to obtain survival curves using Kaplan-Meier estimators and were rst analyzed using the log-rank test. Subsequently, survival curves were pairwise compared using Bonferroni's method. The values of food consumption were transformed to ln and then subjected to the generalized least squares (GLS) model under different structures of variance-covariance, owing to repeated measures over time. The model was chosen based on parsimony, veri cation of residual quantile plots, and the lowest Bayesian information criterion (BIC). Back-transformed estimates were then used in graphical plots to represent consumption over time and between treatments. CFU data were rank-transformed and submitted to a linear mixed-effects model with the colony as a random effect. Subsequently, Tukey's pairwise comparisons were carried out using adjusted P values according to the Westfall method [43]. The variables of walking behavior (time at rest, medium activity time, fast activity time, and walked distance accumulated) were subjected to analysis of variance (ANOVA) with colonies considered as repetitions. For ight behavior data, a generalized linear model (GLM) was tted with a quasibinomial distribution; adequate distribution for proportion data when there was overdispersion (high residual deviance), and colonies were also used as repetitions. The residues were checked in all models to verify the adequacy of the distributions. All data were analyzed using R software [44] with a signi cance level of 5%.
Both Serratia eGFP AS1 and S. marcescens (wild) were detected in the homogenate obtained from the digestive tract (midgut + hindgut) dissected at 24, 72, and 144 h after the ingestion of the bacteria (Fig. 3, Supp. Figure 1). The bacterial colony-forming units (CFUs) were higher at 24 h than at 72 h (z = -3.57, p = 0.001) and 144 h (z = -2.21, p = 0.027), but were similar between 72 h and 144 h (z = 1.36, p = 0.17) for the recombinant strain. The CFUs were higher at 24 h than at 72 h (z = -3.65, p < 0.001) and 144 h (z = -4.40, p < 0.001), but were similar between 72 h and 144 h (z = -0.76, p = 0.45) for the wild strain. Serratia eGFP AS1 was detected in the midgut of bees at 24 h and in the hindgut at 72 h and 144 h after exposure (Fig. 4). However, this modi ed bacterium was not detected in the ovaries at any of the analyzed times (Supp. Figure 2).
There was no signi cant difference between treatments in any of the variables associated with walking activity (Supp. Table 1, Supp. Videos).

Discussion
The present study is the rst to investigate the possible effects of digestive tract colonization of a native bee by a genetically modi ed bacterium (Serratia eGFP AS1) developed for the control of vector-borne diseases. Our results demonstrated that ingestion of the modi ed bacterium did not affect the tness or behavior of P. helleri foragers. Bees colonized by Serratia eGFP AS1 had similar survival rates, ingested the same amount of food, and had similar walking and ying activities compared to individuals that ingested a non-recombinant bacterium or sterile sucrose controls. Our results corroborate the studies using mosquitoes (i.e., Culex pipiens, A. gambie, and A. stephensi) colonized by Serratia eGFP AS1, for which no negative effects on survival, feeding behavior, or fertility were detected [10,11].
In general, the genus Serratia is non-pathogenic and is constitutively found in the guts of many arthropods, including mosquitoes, bees, sand ies, ticks, beetles, and aphids [4][5][6][7][8][9]45]. In our experiments, ingestion of the two tested bacterial strains was not harmful to the bees.
However, certain S. marcescens strains are pathogenic. For instance, strain Ss1 is pathogenic when in high abundance in honeybees [33,37], whereas RPWL1 can be pathogenic to the beetle Rhynchophorus ferrugineus [46]. Conversely, a genetically modi ed Serratia symbiotica did not affect the tness or survival of aphids, suggesting that it is an important future paratransgenic tool for the control of agricultural pests [9].
Serratia eGFP AS1 and wild S. marscencens were recovered from the digestive tract (midgut + hindgut) of P. helleri at 24, 72, and 144 h after bacterial ingestion, however there was a reduction in bacterial numbers over time. This reduction may be related to the speci city of the gut microbiota [34], as native strains easily outperform non-native strains [47]. However, how this occurs is still not well understood and needs to be further studied. Using uorescence, Serratia eGFP AS1 was detected only in the midgut 24 h after ingestion, and only in the hindgut at 72 h and 144 h after ingestion. This change in the organ of colonization may be due to the protection conferred by gut bacteria naturally present in Meliponini bees [8,36,48].
In mosquitoes, Serratia eGFP AS1 persists for at least three consecutive generations and colonizes organs other than the midgut, including the ovaries and male accessory glands. In addition, this bacterium rapidly spreads among mosquito populations in the laboratory [10]. In bees, possible contamination with the bacteria occurs when they leave the colony for foraging, a function performed at the end of the bee's life [27]. Once contaminated with bacteria, foragers return to the colony where they can contaminate other individuals. However, as Serratia eGFP AS1 was not detected in the ovaries of the bees, it is unlikely that this bacterium could spread through bee populations via vertical transfer. This study highlights the importance of studying the effects of GMOs on non-target organisms, including pollinators, to aid in decision making for the release of GMOs into the environment.

Compliance with Ethical Standards
Disclosure of potential con icts of interest The authors declare that they have no con ict of interest.

Research involving Human Participants and/or Animals
No approval of research ethics committees was required to accomplish the goals of this study because experimental work was conducted with an unregulated invertebrate species.