Over the last 50 years, the number of species of wild bees and other pollinating insects has declined [reviewed in: (1)]. For some species, the number of individuals has extremely decreased, other species have become extinct [reviewed in: (1)]. Furthermore, the number of managed honey bee colonies has declined in North America and in many European countries (2). In China and Argentina, on the other hand, the number of managed honey bee colonies has increased (3–5). There are alarming reports of high colony losses in managed honey bees (colony collapse disorder CCD) from several areas of the world (6). Several factors including CCD are suspected to cause the decline of bees. One of them is habitat loss and the resulting lack of plants as food source. Furthermore, diseases caused by bacterial and viral infections, and parasites e.g., the parasitic mite Varroa destructor contribute to the decline of bees [reviewed in: (1)]. Also the exposure to pesticides was suggested as driver of the bee decline worldwide (7–9).

The large-scale use of herbicides resulting in fewer attractive flowers being available and insecticides used to protect crops from pests, can have direct adverse and toxic effects on beneficial insects and pollinators such as bees [reviewed in: (10); reviewed in: (1)]. Bees can be exposed to pesticides directly during spray application (droplets, dust, and drift) and via contaminated nectar, honey and pollen (11). In most cases, pollen and nectar are contaminated not only with one, but also with a cocktail of different pesticides (12, 13). Sanchez-Bayo and Goka (13) identified in their metastudy on pesticide residue surveys performed on pollen, honey and wax samples from European and US honey bee colonies a total of 161 different pesticides. Of these, insecticidal neonicotinoids (such as clothianidin, imidacloprid, and thiamethoxam) and organophosphates (like phosmet and chlorpyrifos) pose severe risks to bees (13, 14).

Neonicotinoids in particular are associated with the decline of bees. They are neurotoxic compounds and bind as agonists to the nicotinic acetylcholine receptor and cause hyperstimulation, paralysis and death (15). Besides the acute toxicity of neonicotinoids, their sublethal effects might also play a role in the decline of bees [reviewed in: (16)]. Sublethal effects by neonicotinoids include adverse effects on the immune system (17), impaired orientation and memory formation (18), reduced colony size and negative effects on larval development (19, 20), altered homing flight activity (21, 22), negative effects on self-grooming (23) and changes in gene expression (24–28). Negative effects of neonicotinoids in terms of prolongation of flight times and flight activities of foragers are known (22, 29). The radiofrequency identification (RFID) technology is a common method to analyze negative effects of neonicotinoids on homing flight performance of honey bees (30–34). By applying RFID technology, it was possible to show that exposure of foragers to thiamethoxam resulted in a reduced return rate and in a prolonged homing flight duration of returning bees (35).

To study sublethal effects of pesticides on gene expression in honey bees, quantitative PCR and next generation sequencing are reliable tools, and were already used to prove that the exposure to neonicotinoids alters the expression of acetylcholine receptors, vitellogenin, stress transcripts and transcripts linked to the immune system in honey bees (24). Moreover, global gene expression analysis using next generation sequencing showed negative effects of the insecticides clothianidin, imidacloprid and thiametoxam on metabolism and detoxification (26). In addition, the exposure of honey bees to fungicides such as azoxystrobin and chlorothanolin, and to the insecticide spinosad resulted in altered expression of oxidative phosphorylation transcripts (36, 37). Thus, these substances may also disturb the energy metabolism of honey bees.

In our opinion, there are two coherent explanations for the failure of pesticide exposed bees in returning to their hive. Firstly, the orientation could be impaired and thus the path home could no longer be found or detours could be flown (38). Secondly, the energy metabolism in the brain of foragers could be disturbed and thus displaying negative effects on flight behavior. To test the latter hypothesis, we conducted a series of RFID experiments in which we exposed pollen foragers to thiamethoxam- or thiacloprid-laced sugar solutions. Through this experimental design, we monitored both return rates and return durations in combination with gene expression analysis in successfully returned pollen foragers. Thus, we were able to link return times of individual pollen foragers with expression patters, for the first time. In order to understand the underlying molecular mechanisms, we measured transcript levels of genes involved in energy metabolism, endocrine regulation, detoxification and oxidative stress (such as oxidative phosphorylation genes) in pooled and individual bee brain samples. To investigate possible differences between honey bee colonies, the experiments were carried out on foragers originating from different bee colonies. In addition, a standardized laboratory-based exposure was performed independently from an RFID homing flight test to confirm the results obtained under controlled conditions, aiming at developing a functional laboratory-based bioassay. As a further point, in the study presented here, we investigated the effects of group feeding and individual feeding on data variability in gene expression and homing flight ability of treated bees, as the dosing of the treatment might vary due to unequal trophallaxis among caged bees (35, 39). A graphic description of the performed experiments can be seen in Figure 1. In a first pilot study (experiment I), an RFID experiment was conducted with pollen foragers exposed to thiamethoxam. The return rate and the return time were measured. Successfully returning pollen forager bees were sampled 30 h after release and gene expression analyzed in pooled brain samples. At that time, we knew from previous studies that transcripts of endocrine regulation, memory formation, and the detoxification system were altered by neonicotinoids (24, 27). Therefore, we examined the expression of the following transcripts: hbg-3, ilp-1, vitellogenin, pka, creb, cyp9q1, cyp9q2, and cyp9q3. The goal was to see possible differences in expression patterns between controls and thiamethoxam exposed foragers. Since this was the case, we decided to conduct experiment II and III. Beside the possible negative effects of neonicotinoids on memory formation, we thought that a dysregulation of the transition of nurse bees to foragers might also have negative effects on flight behavior (40). Therefore, we were interested in the expression of transcripts linked to endocrine regulation. In a second “worst-case” study (experiment II), RFID homing flight experiments were conducted with pollen foragers exposed to thiamethoxam and thiacloprid. Again, the return rate and the return time were assessed. Returning foragers were collected directly upon their return, ranked based on their return time and differences in gene expression of fast returning controls and slow returning bees exposed to neonicotinoids were analyzed. Since we had evidence that exposure of bees to pesticides negatively affects energy metabolism (36, 37), we analyzed in experiment II transcripts of oxidative phosphorylation (cox5a, cox5b, cox6c, and cox17) in addition to transcripts of endocrine regulation (buffy, hbg-3, ilp-1, and vitellogenin). Since we could only isolate a limited amount of RNA from one brain, we could only analyze a limited number of transcripts. In a third experiment, a standardized, laboratory exposure was performed independently of an RFID test (experiment III) to confirm previous data. Therefore, pollen foragers from three honey bee colonies were exposed to thiamethoxam and thiacloprid and the expression of different transcripts was analyzed immediately after treatment. To verify gene expression data from experiment II, transcripts of oxidative phosphorylation (cox5a, cox5b, cox6c, and cox17) and endocrine regulation (hbg-3, ilp-1, and vitellogenin) were again analyzed. In addition, transcripts of the detoxification system (cyp9q1, cyp9q2, and cyp9q3) and a general stress marker (catalase) were analyzed. We were once more able to show effects on transcripts of oxidative phosphorylation genes; however, responses varied between different honey bee colonies. Overall, our results suggest that neonicotinoid exposure of forager bees disturbs their energy metabolism, and this disturbance might cause a reduced return rate or a prolonged flight time in neonicotinoid exposed bees in the RFID homing flight test.

Chemicals: Thiacloprid and thiamethoxam (purities of all >99%) were purchased from Sigma–Aldrich (Buchs, Switzerland). Stock solutions for each compound were prepared in acetone and diluted into 20% sucrose-solution to a final exposure concentration.

The study presented here consists of three experiments that build on each other. The exact sequence of the three experiments is shown in Figure 1.

In the study presented here, we aimed at finding answers to the following questions: Is there a measurable difference in gene expression of forager bees exposed to neonicotinoids and non-treated control bees, which were previously tested in RFID homing flight experiments? Is there a correlation between homing flight duration and gene expression in the brain of foragers after exposure to neonicotinoids? Can we identify possible biomarker transcripts that indicate altered flight behavior after exposure to neonicotinoids? Are there differences in the response to neonicotinoid exposure between different bee colonies? Does the type of feeding, individual or group feeding, influence the data quality/scatter? To answer the first question, we analyzed gene expression in the brains of returned unexposed and exposed foragers in the performed RFID experiment. We detected differences in the expression of various transcripts, such as vitellogenin, ilp-1, creb, cyp9q1, and cyp9q2 (Supplementary Figures 1–3), although most significant effects showed no dose dependence. In addition, the individual RFID-runs showed different expression patterns. An altered expression of vitellogenin and creb in the brain of foragers after neonicotinoid exposure has already been shown in another study (24). Effects of neonicotinoids on endocrine regulation, and thus on the expression of endocrine-relevant transcripts such as ilp-1 after neonicotinoid exposure, have also been shown (27). In these studies, most detected effects did not show dose dependence, as observed in this study. With these initial experiments, we were able to confirm published data on effects of neonicotinoids on the gene expression of bees (24, 27) and demonstrate that foragers from the RFID experiment are useful study objects to perform gene expression analyzes. The prolongation of the flight time of foragers after neonicotinoid exposure may result, among other things, from the fact that the orientation of the bees and the memory formation are disturbed and/or that the energy metabolism is disturbed and therefore too little energy is available for flying. In the literature, both theories are confirmed: On one hand, several studies show negative effects of neonicotinoids on the orientation of honey bees supporting the first theory. The exposure of foragers to thiamethoxam resulted in impaired orientation and longer flight duration (48). Exposure of foragers to imidacloprid, clothianidin, and thiacloprid resulted in reduced return rates and fewer direct returns from the food source to the hive (38). Exposure of foragers to clothianidin, imidacloprid and thiamethoxam led to altered expression of creb and pka, two transcripts linked to memory formation (24). Furthermore, some studies have shown that exposure of bees to pesticides impair the energy metabolism and that neonicotinoids and other pesticides disturbed mitochondrial function (22, 36, 37, 49). Exposure of bees to nicotine leads to an upregulation of proteins of the energy metabolism [oxidative phosphorylation, proteins of complexes I, III and IV; (50)]. This upregulation serves to meet the increased energy demand through detoxification of nicotine by cytochrome P450 enzymes (50). Exposure of honey bees to pyraclostrobin has negative effects on mitochondrial activity, leading to inhibition of oxidative phosphorylation and altered mitochondrial membrane potential. This is particularly critical for the energy supply of foragers (49). Exposure of honey bees to different fungicides such as azoxystrobin, chlorothanolin and folpet leads to altered expression of transcripts such as cox5a, cox5b, cox6c, cox17, and ndufb7. These transcripts encode proteins of complex I and IV of oxidative phosphorylation (36). Spinosad, an insecticide used in organic farming, also leads to altered expression of oxidative phosphorylation transcripts in the brain of foragers (37). In order to investigate the relationship between prolonged flight time and impaired energy metabolism after neonicotinoid exposure of sampled bees, the expression of oxidative phosphorylation transcripts in the brain of bees used in the RFID homing flight test was investigated. Gene expression analysis was performed on the brain of individual bees to establish a direct correlation between expression pattern and flight time. In contrast to the pilot study, foragers were frozen directly after returning to the bee hive to analyze gene expression pattern at the time of return to the bee hive and not as in the pilot study only several hours later. To establish a “worst-case” scenario, the fast-returning controls and the very slow-returning thiamethoxam-exposed foragers were selected for gene expression analysis (Supplementary Figure 4). In slowly returning thiamethoxam-exposed foragers, the expression of cox5a was significantly increased and the expression of cox17 was significantly decreased (Figure 5A). There was a weak, but statistically significant correlation between up-regulation of cox5a expression and prolongation of homing flight duration (Figure 5B). The change of expression of transcripts linked to oxidative phosphorylation after exposure to pesticides is in accordance to already published studies (36, 37). These data support the hypothesis that exposure of foragers to neonicotinoids has negative effects on energy metabolism and therefore might cause prolonged return time in these bees. The expression of cox5a may be used in future studies as biomarker of disturbed homing flight activity. However, it must be mentioned that in this study, fast-returning unexposed control bees were compared with slow-returning exposed bees. This approach assumes that all bees have the same energy metabolism before exposure and therefore the same flight ability. This assumption is certainly oversimplified and in follow-up experiments, gene expression in slow-flying non-exposed bees should therefore also be analyzed. In addition, the comparison between individual feeding and group feeding showed that data generated from group feeding fluctuated more than data generated from individual feeding (Figure 4). Due to this, no significant effects were detected in foragers of the group feeding approach (Figure 5). When several bees are fed in a cage (group feeding), single bees take up the food and then distributes it to the other bees in the cage. This phenomenon is known as trophallaxis. Therefore, the food containing the tested substance offered to the bees is not evenly distributed among all bees within the cage. Hence, when honey bees are exposed to pesticides via sucrose solution during group feeding, different exposure concentrations may occur in individual bees (39). This finding is important in relation to existing OECD guidelines. For the determination of acute oral toxicity, the OECD 213 guideline suggests feeding in groups of 10 bees. Based on the findings of the present study, this should be reconsidered with respect to data quality and reproducibility. To support the hypothesis of disturbed energy metabolism after neonicotinoid exposure, to confirm the obtained data and to analyze differences between different honey bee colonies the third experiment was conducted. Foragers from three different colonies were exposed to thiamethoxam and thiacloprid and then the expression of different transcripts was analyzed. When comparing experiment II to III, it must be noted that in experiment II, the bees flew in addition to neonicotinoid exposure (two possible stressors) while the bees of experiment III were exposed only to neonicotinoids (one stressor). Foragers from colony 1 and 3 showed a significant upregulation of cox5a, cox5b, cox6c, and cox17 after thiamethoxam exposure. Thiacloprid induced cox5a, cox5b, cox6c, and cox17 only in foragers from colony 1. Foragers from colony 2 showed a completely different response to exposure to thiamethoxam and thiacloprid (Figure 8). As the mitochondrial activity changes during the transition from nurse bees to foragers and the capacity of oxidative phosphorylation in the head of foragers is lower compared to nurse bees (51), effects of neonicotinoids on the expression of transcripts of the oxidative phosphorylation may have tremendous effects on energy metabolism in foragers. This experiment demonstrated that exposure to neonicotinoids can lead to altered expression of oxidative phosphorylation transcripts. In particular, the expression of cox5a was altered. However, it also clearly shows that transcriptional changes vary greatly when comparing different honey bee colonies. Based on the data shown here, cox5a may be a potential biomarker for altered homing flight duration. However, further studies are needed to confirm this.

In the study presented here, first, it was shown that exposure of foragers to neonicotinoids has a negative impact on return rates. Secondly, it could be shown that neonicotinoid exposure has an influence or sublethal effect on gene expression and that a different expression pattern were identified especially in relation to transcripts of energy metabolism between fast and slow returning foragers. There was a correlation between flight duration and cox5a expression. The altered expression of cox5a could be partially confirmed in the laboratory study. Based on the data presented here, a laboratory test can be developed for the future in which sublethal effects (for example flight behavior) can be predicted based on gene expression patterns. Furthermore, for the risk assessment of plant protection products for bees, such a laboratory test could be established as a first test approach triggering higher tier studies, e.g., a homing test (OECD guidance document No. 332) in a risk assessment scheme to evaluate sublethal effect on honey bees. Before such a laboratory test can be developed, more transcription data must be available. Therefore, studies applying next generation sequencing are needed to identify the exact gene expression patterns, as it is difficult to make assumptions based on expressional changes of only a few transcripts. Third, it has been shown that feeding mode has a or significant effect on the variability of the data. This should be considered in the development of future guidelines. Another important point that should be considered in future guidelines is the fact that it has been clearly shown that different colonies respond differently to the same exposure. A larger study with more colonies should be performed to confirm this hypothesis.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding.

Homing flight experiments were done by DG, LJ, and ME. Gene expression analysis was done by VC. J-DC supervised the work and was involved in the experimental design. DG, LJ, and VC contributed all in the same way to write the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.