This manuscript describes several protocols for administering pharmacological agents to honey bees, including simple noninvasive methods for free-flying bees, as well as more invasive variants that allow precise localized treatment of restrained bees.
Honey bees demonstrate astonishing learning abilities and advanced social behavior and communication. In addition, their brain is small, easy to visualize and to study. Therefore, bees have long been a favored model amongst neurobiologists and neuroethologists for studying the neural basis of social and natural behavior. It is important, however, that the experimental techniques used to study bees do not interfere with the behaviors being studied. Because of this, it has been necessary to develop a range of techniques for pharmacological manipulation of honey bees. In this paper we demonstrate methods for treating restrained or free-flying honey bees with a wide range of pharmacological agents. These include both noninvasive methods such as oral and topical treatments, as well as more invasive methods that allow for precise drug delivery in either systemic or localized fashion. Finally, we discuss the advantages and disadvantages of each method and describe common hurdles and how to best overcome them. We conclude with a discussion on the importance of adapting the experimental method to the biological questions rather than the other way around.
Since Karl von Frisch elucidated their dance language1, honey bees have remained a popular study species for researchers in animal behavior and neurobiology. In recent years a myriad of new disciplines have emerged at the intersection of these two fields, and several other disciplines (e.g., molecular biology, genomics, and computer science) have arisen alongside them. This has led to rapid development of new theories and models for understanding how behavior results from activity within nervous systems. Because of the unique lifestyle, rich behavioral repertoire, and ease of experimental and pharmacological manipulation, bees have remained at the forefront of this revolution.
Honey bees are being used to study basic neurobiological questions such as those underlying learning and memory2,3, decision making4, olfactory5, or visual processing6. In recent years, the honey bee has even been used as a model for studying topics generally reserved for medical research, such as the effects of addictive drugs7–11, sleep12, ageing13, or the mechanisms underlying anaesthesia14.
Unlike for the classical genetic model organisms (e.g., D. melanogaster, C. elegans, M. musculus), there are very few genetic tools available for manipulating neural functions in honey bees, although this is currently changing15. Instead, honey bee studies have primarily relied on pharmacological manipulations. This has been very successful; however, the diversity of bee research is such that a range of methods for pharmacological administration are needed. Research with honey bees addresses highly diverse questions, is studied by researchers from different disciplines and backgrounds, and uses a variety of experimental approaches. Many research questions require bees to either be free-flying, freely interacting in their colony, or both. This can make it difficult to keep track of individual experimental animals, and makes restraint or cannulation unfeasible.
To accommodate the diversity of honey bee research, a variety of drug delivery methods are needed, allowing for robust and flexible administration while ensuring that the pharmacokinetic and pharmacodynamic profiles, invasiveness of the method, and its reliability, suit the paradigm in question. Because of these diverse needs, most research groups have developed their own unique drug administration methods. So far, this has been a strength of the bee research community; it has led to the development of arrays of methods allowing for administration of the same drug in different circumstances. Our goal here is not to develop a single standardized method for pharmacological manipulations of bees, but rather to highlight methods that have proven to be particularly successful, and help researchers adopt these. We discuss the basic principles of how they work, as well as their advantages and disadvantages.
1. Drug Administration for Harnessed Bees
2. Drug Administration Methods for Free-flying Bees
A selection of representative results for the methods described above are shown, primarily to demonstrate that the methods allow pharmacological agents to reach the brain and affect honey bee behavior.
Specific effects on brain processes can be easily obtained following thorax injection.
Because pharmacological agents injected through the thorax may act on multiple targets in the body, and get diluted into the body before reaching the brain, this technique may raise possible specificity concerns. Nevertheless, it has been used widely in the literature to interfere with cognitive processes, without the necessity to use very high doses that might yield major secondary effects. For example, blockers of transcription have been administered using this technique, in order to identify phases of memory that require gene expression. Thorax injection of such molecules is compatible with survival for several days27, which means that their potential toxic action on other targets can be limited, provided the concentration is well chosen. In such conditions, selective and time-dependent effects on memory can be obtained, thus showing efficient targeting of the brain (Figure. 2).
Diffusion of molecules into the head hemolymph leads to quick, dose-dependent effects.
Ocellus injection is a way to enable a quick diffusion of molecules of interest into the whole head through the hemolymph, especially if they may have many widespread targets in the brain. This method was used to administrate allatostatins, neuropeptides that may also act as neurohormones28). As a consequence, a reduced performance was observed in an olfactory learning assay, consistent with the suggested presence of allatostatin receptors in different brain regions involved in olfactory processing and learning28. A dose-dependent curve for this effect could be established, by injecting different concentrations to independent groups run in parallel (Figure. 3).
Different ways of administration can yield to similar effects on brain function.
Emetine, a blocker of protein synthesis, is used to impair the formation of early olfactory memory long-term memory, which is typically expressed 1-2 days after conditioning. In most published studies it has been injected into the thorax29. We showed that similar effects could be obtained by administering it directly to the brain through the ocellar tract (Figure. 4): providing an adjustment of injection parameters (smaller volume, higher concentration and shorter delay before conditioning), we obtained a decrease (~20%) similar to that found in the literature using the same drug amount (10 nM) – compare with Figure 4 in Stollhoff et al., 200529.
The effects of localized injections are confined in time and space
To test the spatial and temporal properties of drugs microinjected into specific brain regions, harnessed bees were trained in an olfactory PER conditioning paradigm, and then injected bilaterally with 0.5 nl of 740 mM procaine (an anesthetic) in the mushroom body calyces or vertical lobes (saline was used as a control). When bees were successively tested for recall 1, 2, and 3 hr after injection, performance was only impaired in bees with bilateral injections into the lobes (Figure. 5). Intact neural output from the lobes, but not from the calyces, is known to be necessary for olfactory memory retrieval, so this suggests that procaine remained localized to the lobe in which it had been injected for at least 3 hr. It also shows that, when injected into the calyces, diffusion into the nearby lobes was limited over the same period, since a calycal injection of procaine did not lead to blockade of the lobes.
Behavioral phenotypes following drug administration are often context-dependent
Previous experiments have shown that after treatment with cocaine bees over-estimate the quality of a sucrose solution10,30. To see if this effect was dependent on context (here, baseline sucrose quality), free-flying honey bees were treated with volatilized cocaine. Individually marked free-flying honey bees were allowed to forage at a feeder containing 1 M sucrose solution. At the feeder, bees were gently captured in a 50 ml centrifuge tube as they were about to alight from the feeder. Bees were treated with either 100 µg of freebase cocaine or vehicle control (evaporated ethanol). After treatment, the sucrose feeder was either replaced by a 0.5 M or a 2.0 M sucrose feeder, and the rate foragers returned to the feeder was recorded. Using this paradigm, cocaine-treated bees increased their foraging effort at the 0.5 M feeder, but not at the 2.0 M feeder (Figure 6). The difference in effect seen with the two sucrose concentrations nicely demonstrates the importance of taking environmental cues into account when studying bee behavior.
Figure 1: Confocal Laser scanning Image of the Injection Site. Alexa 546-labelled dextran is injected together with the drug solution (red). To identify the neuropils a counter-staining with DAPI is added (green). In the right hemisphere the injection site was located in the vertical lobe (VL), shown as an example for a successful injection. In the left hemisphere the injection site was located dorsal of the vertical lobe in the ring neuropil, shown as an example for an unsuccessful injection. Scale bar = 100 µm, MB: Mushroom Bodies, AL: Antennal Lobes, d: dorsal, v: ventral, l: left, r: right. Please click here to view a larger version of this figure.
Figure 2: Time-dependent Effect of Actinomycin D (Transcription Blocker) on Long-term Memory, when Injected into the Thorax. At different delays following appetitive olfactory conditioning (6, 9 or 12 h), 1 μL actinomycin D (1.5 mM in PBS) was injected into the thorax. Long-term Memory (LTM) retrieval was assessed 3 d after conditioning (n = 25 – 65). Memory performance was reduced in a time-dependent fashion, as compared to that of PBS-treated controls: the effect was significant when injection too place 6 h after conditioning (χ2= 18.04, p <0.005), but not at longer delays (9 h: χ2= 0.95; 12 h: χ2= 0.47), suggesting that LTM formation requires a wave of transcription that takes place during a defined time window after conditioning. Error bars represent standard errors. Data was previously published27and is recreated here with permission. Please click here to view a larger version of this figure.
Figure 3: Dose-dependent Inhibition of Learning Performance Following Ocellar Injection of a Neuropeptide. The neuropeptide allatostatin C was injected into the head hemolymph (200 nl in PBS), through the median ocellus, 1 hr before olfactory conditioning. Independent groups of animals injected with different concentrations (or PBS for controls) were trained. Allatostatin C treatment led to a decrease in the learning performance, as assessed by the percentage of conditioned responses in the last conditioning, in a dose-dependent manner following a U-shape curve (n = 70-78). This decrease was significant at 10-6 M but not at other concentrations. Error bars represent standard errors. Data was previously published28, and is adapted here with permission. Please click here to view a larger version of this figure.
Figure 4: Blockade of 1 Day Memory Following Injection of Emetine (Translation Inhibitor) through the Ocellar Tract. The protein synthesis inhibitor emetine (50 mM in PBS, 200 nl) was injected into the brain, through the ocellar tract, 20 min before olfactory conditioning. Memory was then tested 24 h later. The treatment significantly impaired memory retention (χ2 = 7.03, p <0.01) as compared to PBS-treated controls (n = 57-70). Error bars represent standard errors. JM Devaud, unpublished data. Please click here to view a larger version of this figure.
Figure 5: Anatomical and Temporal Specificity of Microinjections. Following appetitive olfactory conditioning, procaine was injected bilaterally into either the mushroom body calyces or vertical lobes. Memory retrieval was assessed 1 hr after injection and was only affected by procaine injections into the lobes (1 hr after treatment: vs. saline: χ2 = 10.00, p <0.005; vs. procaine to calyces: χ2 = 32.92, p <0.005). The effect could still be seen 2 hr (χ2 = 6.65, p <0.01) and 3 (χ2 = 27.22, p <0.005) after injection, and was still location-specific (2 hr: χ2 = 8.60, p <0.05; 3 hr: χ2 = 17.15, p <0.0001), suggesting that only the injected area was affected by procaine. Proportions are relative to conditioning level during the last conditioning trial. Error bars represent standard errors (n = 23-28). Data was previously published31, and is recreated here with permission. Please click here to view a larger version of this figure.
Figure 6: Effects of Cocaine on Free-flying Bees. Visitation rate (number of visits by a given bee/average visits for all bees during test period) was increased following volatilized cocaine treatment at a low quality source (0.5M : t70 = 5.0710, p = 0.00003), but not at a high quality source (2M: t70 = −0.2087, p = 0.8353). The boxes represent 1st and 3rd quartiles with the midline showing the median. The whiskers extend to 1.5x the interquartile range. Outliers are not plotted as all individual data points are superimposed. Data was previously published10, and is recreated here with permission. Please click here to view a larger version of this figure.
Treatment | Can be done with free- flying bees? | Pros | Cons |
Oral treatment | Yes. | Easy, minimally invasive. | Bee digestion is not straightforward |
Topical treatment | Yes. | Easy, minimally invasive, quick. | Repeated treatments can be problematic. |
Injection into the thorax | Complicated, affects bees flying abilities | Consistent and robust. | Somewhat invasive. Potential to harm/stress bee. |
Injection into the median ocellus | Not recommended. | Consistent and robust, somewhat localized. | Somewhat invasive. Potential to harm/stress bee. |
Injection into the ocellar tract | Not recommended. | Very localized | Very invasive. Potential to harm/stress bee. |
Micro-injection into brain regions | Not recommended. | Very localized | Very invasive, hard to perform. Potential to harm/stress bee. |
Volatilized drug delivery | Yes. | Easy, minimally invasive, quick. | Does not work for all drugs. |
Table 1: Comparison of the Different Treatment Methods and Their Properties.
The methods outlined above allow simple, effective and robust treatment of either free-flying or harnessed honey bees. These methods are compatible with many experimental paradigms and biological questions (Table 1). All of the free-flying methods can easily be applied to harnessed bees. The reverse is less successful, however, since temporary restraint and invasive treatment methods can often compromise bees' flying ability.
The methods have been presented from a brain-centric perspective. This is not due to inherent limitations of the techniques, but rather because of the authors' personal interests. There is no reason why these methods cannot be used for studying other organs. However, small modifications might be needed to make the method more suitable to other organ systems. For example, while topical treatment intended to reach the brain is typically applied to the thorax, it might be better to apply this to the abdomen if the intended target is the ovaries. Similarly, injections can easily be applied to other areas than the thorax or head (e.g., abdominal organs can be targeted by injecting between the abdominal sclerites).
In terms of which compounds can be administered to bees, there really are no limits. Typically, people have administered pharmacological compounds such as signal molecules21or their antagonists32, and custom-made peptides28. However, there has been a recent increase in administering to bees compounds with applied questions in mind, such as pesticides33and anthropogenic contaminants34. Recently, compounds administered have started to include RNA molecules that interfere with gene expression directly, such as dsRNA activating the RNA interference pathway35or even microRNAs36and antagomiRs37. Not all methods work equally well for all compounds. This is perhaps best illustrated by bitter or sour compounds that make sugar water unpalatable to bees, thus preventing them from consuming it. Fragile molecules, such as RNAs or certain polypeptides, are broken down when heated during a volatilization procedure or placed in a harsh solvent like DMF. It is therefore important to understand the chemistry of what is being administered to ensure it survives the treatment procedure.
Getting a pharmacological agent into a bee is the easy part, but there are three big concerns that should never be taken lightly when performing pharmacological experiments. The first is figuring out a good dose for the experiment in question. Depending on the drug, there might already be published literature available, but for the most part, this will have to be resolved by a mixture of literature searches, informed guesswork, and dose-response curves. Depending on how complicated the experimental protocol is, it might be useful to first generate a dose-response curve in a simpler bioassay (e.g., quantifying overall movement or survival) to get a better idea of a dose-range worth trying in a more elaborate bioassay. In our laboratory, a starting dose is either found in the bee literature or by doing a mg/kg conversion based on data from the rodent literature. From this starting point, bees are treated with the starting dose, plus 2 or 3 doses 10 times larger and smaller than the starting dose (e.g., if the starting dose is 1 mg, 0.01, 0.1, 10, and 100 mg would also be used), and of course an appropriate vehicle control.
The second problem is slightly more finicky: drug specificity. Most drugs were not developed with honey bees, or any other insect, in mind. Because of this, off-target effects are common (e.g., mianserin, a vertebrate serotonin receptor antagonist38, was long thought to be an insect octopaminergic receptor antagonist, but recent findings show that in bees it is also a dopaminergic receptor antagonist39). A common solution to this problem is, rather than relying on only one drug, to repeat the same experiment with a suite of drugs known to have the target of interest in common. Basically, if several drugs are known to block a certain target, observing similar results across different drugs should give greater confidence that the drug has the expected effect, since different drugs often have unique off-target profiles.
The last issue involves ensuring that the drug is acting where it is supposed to be acting. In this regard, there will always be a trade-off between specificity and invasiveness. Systematic treatment methods are generally the least invasive, but there is no control of where in the bee body the drug is having its effect. Even for microinjection of targeted tissues drugs may travel with the hemolymph to other parts of the bee body. How this issue is addressed needs to be informed by the questions asked. For certain experiments anatomical location is irrelevant, whereas for others this is the only question of importance. The best way to address this is to start with systemic treatments and gradually narrow down to an anatomical location by using increasingly more specific methods. If the behavior being studied is particularly incompatible with invasive treatment methods, it might be worth trying to deconstruct it into simpler components before doing a whole series of experiments with very specific pharmacological treatments.
This problem of drug leakage is even more exaggerated with oral treatment of free-flying bees, where drugs can affect non-target bees. Forager honey bees collect nectar in the field to bring back to their colony. They will offload the majority of their sucrose solution in the hive upon returning rather than absorb it. In the hive it is packed in cells, dehydrated, and stored as honey. Because of this, drugs can potentially affect non-target bees. With more specific methods (such as microinjections) this problem is minimized.
With these caveats in mind, and addressed properly, neuropharmacological manipulation of honey bees can be a very powerful tool. While transgenic tools are being developed for honey bees15, because of their social lifestyle it is unlikely that transgenics will ever be an easy and reliable way to conduct these kinds of experiments. It is therefore likely that pharmacology will continue to be an important element of bee research in the future. While some bee researchers have made calls for standardized experimental methods40, in this case this would be a mistake. Part of the power of the bee system has always been the diversity of experimental approaches, and how techniques have been developed with real biological questions in mind rather than the other way around. It is nevertheless important that we ensure usage of the most appropriate method for the question at hand. If comparisons to previous studies are key, standardized protocols must be followed strictly. However, utilizing established protocol for the sake of using standardized methods must not be allowed to stand in the way of the development of novel methods that can open new experimental possibilities.
The authors have nothing to disclose.
This project was funded by ARC grant DP0986021 and NHMRC grant 585442. ABB is supported by an ARC Future Fellowship (FT140100452). JAP is supported by an iMQRES scholarship awarded by Macquarie University and by a DAAD-Doktorandenstipendium awarded by the German Academic Exchange Service. JMD is supported by CNRS and University Paul Sabatier.
Sucrose | Sigma-Aldrich | S8501 | Any supplier will do |
Sodium Chloride | Sigma-Aldrich | S7653 | |
Potassium Chloride | Sigma-Aldrich | P9333 | |
Magnesium Chloride hexahydrate | Sigma-Aldrich | M2670 | |
Calcium Chloride dihydrate | Sigma-Aldrich | C8106 | |
Dextrose monohydrate | Sigma-Aldrich | 49159 | |
Phosphate Buffer Saline (PBS) | Sigma-Aldrich | P4417 | |
Protection Wax | Dentaurum | 124-305-00 | |
HEPES | Sigma-Aldrich | H3375 | |
dimethylformamide | Sigma-Aldrich | D4551 | |
95% Ethanol | Sigma-Aldrich | 493511 | |
Glass capillary | WPI | 1B100F-3 | |
23 G NanoFil needle | WPI | NF33BV-2 | |
Very fine forsceps | Dumont | 0208-55-PO | |
Electrode puller | SRI | 2001 | |
FemtoJet Microinjector | Eppendorf | 5247 000.01 | |
Eicosane | Sigma-Aldrich | 219274 | |
manual micromanipulator | Brinkmann Instrumentenbau | MM-33 | |
electronic micromanipulator | Luigs & Neumann Feinmechanik + Elektortechnik | Junior unit XYZ | |
stereomicroscope | Leica | M80 | |
soldering iron | Weller | WESD51 | |
Dextran, Alexa Fluor 546, 10000 MW | ThermoFisher Scientific | D-22911 | |
Dextran, Alexa Fluor 568, 10000 MW | ThermoFisher Scientific | D-22912 | |
small Petri dish | Sigma-Aldrich | P5481 | |
mineral oil | Sigma-Aldrich | M5904 | |
50 mL Centrifuge tube | ThermoFisher Scientific | 339652 | |
forceps | Australian Entomological Supplies | ||
Blade holder and breaker | Australian Entomological Supplies | E130 | |
Feather double edged razor blade | ThermoFisher Scientific | 50-949-135 | |
Nichrome wire | Any supplier will do | ||
Electrical wires | Any supplier will do | ||
Model paint | Tamiya USA | Depends on colour | |
Repeating dispenser | Hamilton company | PB-600-1 | |
Glass syringe | WPI | NANOFIL | |
flourescence viewing system | Nightsea | SFR-GR | |
graticule | ProSciTech | S8014-24 | |
microcapillary with holder | Drummond | 1-000-0010 | |
Liquid silicone | Any supplier will do | ||
Thermocouple | Digitech | QM-1324 | |
Micropipette | Eppendorf |