In Vivo Imaging of Neural Activity in Unanesthetized Drosophila Adult Flies

Recent advancements in genetically encoded activity reporters have revolutionized neuroscience by providing real-time, highly specific insights into neuronal activity. These reporters, such as calcium or voltage sensors^{1,2,3,4,5,6,7,8,9}, allow researchers to track the electrical and chemical signals within neurons with unprecedented temporal precision and spatial resolution. By enabling the monitoring of specific, single neurons to neural networks across different brain regions, these tools have deepened our understanding of how neuronal firing patterns relate to cognitive processes like learning, memory, or decision-making^{10,11,12,13,14}.

To take advantage of these genetically encoded indicators, researchers developed advanced methods to prepare their model organism for in vivo imaging. The first report of neuronal plasticity by in vivo functional imaging, often referred to as a "memory trace", was discovered in the antennal lobe of the honeybees following appetitive olfactory conditioning¹⁵. This discovery inspired research in the genetically versatile model organism, Drosophila melanogaster, leading to the development of innovative in vivo imaging techniques^10,16,17,18.

Drosophila serves as an excellent model for understanding learning and memory due to its robust ability to form olfactory memories via classical conditioning, where an odor (the conditioned stimulus-CS) becomes associated with either an aversive or rewarding stimulus (the unconditioned stimulus-US)^19,20,21,22. In Drosophila, odors are sparsely represented in the mushroom body (MB)²³, a critical brain region for associative memory that consists of approximately 2,000 Kenyon cell neurons (KCs) per hemisphere²⁴. Olfactory information is transmitted from the calyx to the MB lobes via KC axons, which bundle together to form the lobes. During olfactory conditioning, the US, such as an electric shock or a sugar reward, is conveyed to the MB lobes through distinct dopaminergic neurons (DANs): PPL1 neurons for aversive stimuli and PAM neurons for rewarding stimuli^{25,26,27,28,29,30}. Memory acquisition is thought to involve the simultaneous activation of CS and US pathways, resulting in a synergistic increase in cyclic AMP (cAMP), which encodes the memory^31,32. This process induces plasticity at the KC→MBON synapses, modifying their synaptic weight^{11,12,13,33,34,35,36,37}. DAN-mediated plasticity ultimately drives the mushroom body output neurons (MBONs) to elicit either an approach or avoidance behavior toward the learned odor³⁴. The current model suggests that the information flow from KCs to the MBONs transforms the representation of odor from odor identity in KCs to more abstract information in the MBONs, such as the valence of an odor based on prior experience^34,38. Unlike KCs, MBONs have broadly tuned odor responses; any given odor results in a response to most MBONs³⁹. The odor tuning of the MBONs is modified by synaptic plasticity and varies significantly between individual flies, suggesting that MBONs change their response to odors based on experience. There are 34 MBONs, with the majority of the MBON network seemingly divided into two mutually antagonist classes that drive either approach of avoidance behavior³⁴. For this reason, using in vivo imaging to study memory retrieval by recording in different MBONs has been widely used.

Early methods for in vivo functional imaging include conducting surgeries to remove a section of the cuticle from the top of the head capsule, which was then covered with a small piece of plastic wrap¹⁸. While this marked the beginning of in vivo imaging in Drosophila, there have since been significant advancements in fly preparation techniques and the delivery of stimuli under the microscope to investigate processes like learning and memory. However, even today, most techniques for fly preparation for in vivo imaging involve anesthetizing the flies. This practice presents challenges, as there is extensive evidence - and an entire field of research- demonstrating how anesthesia impacts memory, as well as other cognitive and physiological processes. For instance, a study was conducted to understand the effects of different anesthetics that are very commonly used in the lab setting (carbon dioxide, isoflurane, sevoflurane, and cold shock). They found that the longer the exposure to anesthesia, there was an increase in recovery time, changes in locomotor activity, lower fecundity, and may impact water and lipid stores⁴⁰. In addition, it has been reported that giving cold shock anesthesia immediately after training can completely disrupt short-term memory²¹. From this initial study, it is now known that in Drosophila, there is unconsolidated memory (anesthesia-sensitive memory) that decays within 6-8 h, and a form of consolidated memory (anesthesia-resistant memory) that is formed gradually after learning and can last around 24 h⁴¹. Based on this, anesthesia can affect cognitive processing and the activity of neurons, which eventually allows for proper memory expression.

The effects of anesthesia on memory are not confined to Drosophila. Numerous human and other mammalian studies have demonstrated similar findings. For example, a cross-sectional study done in a university hospital found that short-term memory was significantly reduced in patients 24 h after general anesthesia⁴². Another study done in mice found that general anesthesia robustly reduced spiking dynamics, decorrelated cellular ensembles, and altered spine dynamics in the CA1 region of the hippocampus, an important brain area involved in learning and memory in mammals⁴³. To overcome these difficulties, the goal of this study is to provide a novel protocol that eliminates the use of anesthesia during fly preparation for in vivo imaging, leaving the fly's body completely free. This approach allows more accurate neuronal activity recordings, improves fly health for longer recordings, and eliminates possible confounding effects of anesthesia compared to other approaches. This protocol is designed to record activity from neurons in Drosophila, specifically looking at how neurons exhibit plasticity during and after the learning experience. However, the dissection protocol is versatile and can be adapted to various applications in live imaging research.

The details of the equipment and reagents used for this study are listed in the Table of Materials.

1. Transgenic fruit flies, Drosophila melanogaster

1. Cross female virgins and male flies (raised at 25 °C in 60% relative humidity on a 12 h light/dark cycle) carrying the desired Gal4 and UAS constructs, respectively, to produce flies in which specific neurons of interest express the desired genetically encoded activity reporter.
2. Age the progeny of the above cross until they are in the range of 2-5 days post-eclosion. Both males and females can be used, as the imaging chambers are custom-made to fit both sexes.

2. Building an aspirator (5-min workflow)

1. Cut 6 mm silicon tubing to about 8 inches long using a razor blade. Take a p200 pipette tip and cut about half of it using a razor blade for the mouthpiece (Figure 1A). Insert the cut tip into the tubing. Cut a p1000 pipette tip enough so that a fly can move through it.
2. Take a small piece of cotton and cover the silicon tubing. This is to prevent the fly from aspirating into the experimenter's mouth. Place the cut p1000 tip onto the cotton-covered tubing.
3. Cover the p1000 pipette tip with a new p200 pipette tip. To determine the appropriate length for cutting the p200 pipette tip, first, remove the p200 pipette tip and aspirate a fly into the p1000 pipette tip. Reattach the p200 tip. Gently blow and tap the fly to move it as far as it can go into the p200 tip.
   1. Once it can no longer advance further, use a dissecting microscope to mark the point where the fly's head ends, and the thorax begins, using a razor blade. Aspirate the fly back into the p1000 pipette tip and cut the p200 pipette tip at the marked position.

3. Building the metal tubing proboscis aspirator (5 - 10-min workflow)

1. Cut the 22X gauge hypodermic metal tubing to the desired size (~10 cm) using Dremel tools and a diamond saw blade. Buff ends of metal tubing with a Dremel 420 cut-off wheel to create a smooth and clean opening. This opening will accommodate the proboscis of the fly.
2. Wrap tubing around a 15 mL centrifuge tube to get the desired curved shape (Figure 1A). The curve should create a hook-like shape that will allow the fly's body to be in line with the chamber.
3. Cut a piece (~7 cm long) of the 12TW gauge hypodermic metal tubbing using Dremel tools (this tubing should fit into a 2 µL pipette tip). Take a 2 µL pipette tip and cut the end using a razor blade to fit the 22X gauge metal tubing. Fit the bigger metal tubing into the other end of the 2 µL pipette tip.
4. Mix a small amount of resin and a hardener of epoxy glue together. Apply the mixture where the small metal tubing meets the pipette tip and where the bigger metal tubing connects to the pipette tip. Allow the adhesive to fully cure overnight before connecting the assembly to a holder on a micromanipulator. Adjust the angle as necessary.

4. Building a shock and odor delivery pipette for learning and memory experiments under the microscope (10-min workflow)

1. Using a Dremel diamond tool, cut 1 mL in 1/100 glass pipette at the 3 mL mark. Cut a small rectangular piece of acrylic sheet with dimensions 24.5 mm x 8mm and a thickness of 1/8 inch.
   NOTE: Alternatively, this piece can be 3D printed to accommodate the shock grid better (the stl and obj files can be found in Supplementary File 1 and Supplementary File 2).
2. Cut the copper shock grid to fit onto the rectangular acrylic piece from the previous step. Solder two electrical wires on opposite ends of the copper grid. Place the copper grid into the rectangular acrylic piece and curve it so that the fly's abdomen and legs fit onto the shock grid (Figure 1I).
3. Tape the copper grid to the rectangular plastic using a piece of electrical tape. Using a hot glue gun, securely attach the glass pipette to the shock grid, ensuring it is straight and properly centered. Using a digital multimeter, verify the copper grid is working to effectively deliver a shock to the fly's legs.

5. Assembly of platinum resistor for glue melting (Myristic acid) (20-min workflow)

1. Using wire strippers, cut 200 mm of wires to the desired length and strip one end of both wires. Cut about 60 mm of the platinum wire (this length should be enough to create a loop that will be used to melt the glue).
2. Using a soldering gun, solder one end of the platinum wire to the electrical wire. Pass a heat-shrink tubing through the electrical wire and use a Bunsen burner to shrink the tubing onto the wire to cover the soldered parts.
3. Cut two 20 mm pieces of the 2.55 mm ID rubber tubing and pass them through the platinum wire (keep the platinum wire straight to facilitate this process). Use the soldering gun to solder the other end of the platinum wire to the other electrical wire. Use a new heat shrink tubing to cover this new soldered part.
4. Remove the plunger of a BD tb 1 mL syringe and pass the wires from the previous steps through the barrel. The platinum should be sticking out of the syringe barrel end. There should be no exposed wires, and the small rubber tubing should be covering the platinum wire except to the very end.
5. Using a wire stripper, strip a small amount of both the other ends of the electrical wires. Using a soldering gun, solder each wire to a banana plug until it is securely connected.
6. Use epoxy glue to glue the top and base of the syringe barrel to glue the wires and barrel together; this will avoid tension to the wires inside the syringe and will ensure longer life to the resistor. Let the glue cure overnight.
7. Connect the finished resistor to the DC power supply.

6. Building of recording chamber - (2-day workflow)

1. Following Supplementary File 3, laser cut a black shim sheet (thickness 0.0125 in). The size of the opening in the middle of the chamber can be modified to fit the size of the flies (the size in the file generally fits females and males).
2. Take a glass microscope slide that will serve as the base of the chamber. Mix resin and epoxy together. Take the black acrylic chamber and, using epoxy glue, attach magnets of neodymium 3 mm x 2 mm onto all four corners. It is recommended to glue one magnet at a time and hold it while the glue cures with a paper clip.
3. Place two new magnets onto each of the magnets glued on the chamber previously. Using epoxy, glue the last attached magnet to a glass slide and use paper clips to hold it in place while curating.
   NOTE: After the glue is curated, the chamber is complete, and more magnets can be added to adjust the height of the chamber for experiments. Supplementary File 4, Supplementary File 5, and Supplementary File 6 show the assembled recording chamber.

7. In vivo dissection protocol without anesthesia - (5 - 7-min workflow)

1. Remove the p200 pipette tip from the aspirator. Insert the aspirator into the vial and aspirate a single fly into the p1000 pipette tip. Replace the p200 pipette tip back onto the aspirator. Gently blow and flick the aspirator so that the fly is immobilized at the top of the p200 pipette tip, headfirst.
2. Place the dissection chamber onto the manipulator holder. Connect the vacuum to the fly-holding tubing and adjust the flow level as needed (~500 mL/min). Use a fast connector to easily and quickly connect-disconnect the vacuum (Figure 1A).
3. Move the vacuum metal tubing to center it in the field of view when looking through the microscope. Confirm the chamber is positioned correctly so that it will go directly on top of the metal tubing when ready to glue the fly. Gently, aspirate the fly's proboscis into the vacuum holder (Figure 1A).
4. Once the fly is in the correct position, move the manipulator so that the fly's head is aligned with the chamber opening.
5. Turn on the DC power supply. The current should be set to range between 0.40-0.50 Amps. Higher currents may cause the myristic acid to melt to the point where its viscosity becomes too low, increasing the risk of the glue spreading over the fly's head.
6. Using the platinum resistance, glue the sides of the eyes and thorax to the chamber using melted myristic acid (Figure 1B). To prevent getting glue on the cuticle of the head, use as little glue as possible, especially when gluing the sides of the eyes.
7. Once the eyes and thorax are glued, disconnect the vacuum tubing and use the manipulator to take the recording chamber off the vacuum tubing. Turn the chamber upside down and use the platinum resistance to glue the proboscis from below (Figure 1C). Gluing the eyes from the bottom is optional and can help keep the fly in place during dissection, but it is not necessary.
8. Once everything is glued, turn off the DC power supply, turn the chamber upright, and attach it to the glass side base. Cut a small piece of tape using scissors and apply it right in front and back the fly's head (Figure 1D).
9. Turn the chamber so that the head is facing the experimenter at 90. Using a dissecting needle, make a vertical cut along the sides of the eyes (Figure 1E). Turn the chamber so that it is horizontally placed. Using the dissecting needle, cut across the cuticle (Figure 1E).
   NOTE: Make sure the cuticle "window" just created can be lifted using the dissecting needle. If it lifts without any resistance, put the dissecting needle back onto its cover. Saline is not used in steps 13 and 14 to prevent light refraction, which could obstruct a clear view during the cuts.
10. Add 100 µL of saline to the fly's head top (Figure 1F). Using sharp forceps, carefully remove the window (Figure 1G). Using the sharp forceps, carefully remove any remaining fat or trachea. The brain should be ready for imaging now (Figure 1H-1J).

8. Visualization of learning-induced plasticity through aversive olfactory conditioning (20-min workflow)

1. Using a confocal microscope equipped with a laser and a water immersion objective, place the prepared fly on the stage of the microscope. Optionally, use a camera with a macro lens to focus on the fly, making sure that the fly is healthy and moving.
2. Adjust the position of the shock grid/odor pipette using a micromanipulator so the fly is sitting on the shock grid.
   NOTE: The fly should still be able to move freely as the shock grid is just touching the legs and slightly the abdomen (the fly should be able to remove its legs and abdomen from the grid if retracted). The odor delivery pipette should be directed toward the antennae of the fly.
3. Using the coarse Z adjustment knob, scan through the Z axis of the brain and locate the brain region of interest. Adjust the frame size to 512 x 512 px. The scanning frequency should be adjusted to the desired speed - with a minimum of 2 Hz. Finally, set the duration of the recording.
4. Start recording the neuron of interest using a custom-made or a commercial odor delivery system, and initiate a "pre-training" odor delivery simultaneously (Figure 2C). Set the recording to 2 min. The pre-training odor presentation starts by recording responses to a 5s first odor pulse (CS-) followed by a 5 s second odor pulse (CS+) with a 30 s inter stimuli interval.
5. After collecting pre-training responses, wait 5 min before initiating a training protocol in the odor delivery system. For this example, only the CS+ was presented during the training. The CS+ odor is presented for 1 min initiated at 30 s after triggering the odor delivery system. The shock stimulator is programmed to simultaneously deliver 12, 90 V shocks while the odor is presented.
6. After training, wait 5 min or 15 min before obtaining post-training responses. Record post-training responses in the same way pre-training responses were recorded in step 6. Save all files in the appropriate format for subsequent image analysis.

9. Image analysis

1. To analyze recordings, draw a region of interest (ROI) around the area of the neuron. This should be as precise as possible to limit the influence of background fluorescence.
2. Use the measurement tool to extract mean fluorescence intensity data within the ROI for each frame of the recording. Export this data as a CSV file.
3. Open the data in MATLAB, or any data analysis program to calculate ΔF/Fo values for each odor trace. Combine the data to get the average ΔF/Fo overall individual fly recordings and create a graph of the mean odor response traces pre and post-training (Figure 2E,F).

Here, in vivo imaging of adult Drosophila without anesthesia is utilized to investigate aversive olfactory conditioning inducing plasticity in a well-characterized Mushroom Body Output neuron (MBON γ2α'1)12. The calcium indicator GCaMP6f2, along with the red fluorescent protein TdTom, are expressed in an MB output neuron, whose dendrites innervate the compartment of the MB γ and α' lobes and is genetically targeted to the MBON γ2α'1 using the split-Gal4 line MB077C34 (Figure 2A,B).

To observe plasticity post aversive olfactory conditioning, the fly is placed under the confocal microscope, with its abdomen and legs aligned to the middle of the shock grid and the odor delivery pipette pointed towards the fly's antenna. First, calcium odor responses were recorded to a 5 s pulse of 4-methyl cyclohexanol (MCH) followed by a 5 s pulse of 3-octanol (OCT), separated by a 30 s inter-stimulus interval. Five minutes later, flies were trained by the simultaneous presentation of a 1 min OCT (CS+) pulse along with 12, 90 V shocks. This associative conditioning schedule produces robust behavioral aversive memory. Finally, 5 min and 15 min after conditioning, calcium responses to MCH and OCT were recorded as prior to training. MBON γ2α'1 responds robustly by calcium transients to both MCH and OCT before conditioning (Figure 2D-F). Nevertheless, matching previous findings12, 5 min post-conditioning calcium responses to CS+ (OCT) are completely depressed (Figure 2D-F). Additionally, responses to the CS- (MCH), were significantly potentiated. Fifteen min post-training showed similar results, as the calcium response to the CS+ was significantly depressed and CS- was potentiated. These results indicate that the memory trace observed in both odors lasts for at least 15 min (Figure 2D-F). These responses were quantified using the average odor response during the 5 s pulse and compared within the fly. These changes in odor responses can also be visualized by a pseudo-color representation of the change in GCaMP6 fluorescence pre and post-aversive training (Figure 2E). In contrast, anesthetizing the fly during preparation shows reduced plasticity to both CS+ and CS-. Only partial depression to CS+ is observed in anesthetized animals, as post-training responses are still present. In addition, CS- responses are not significantly different after training (Figure 2G,H). To better observe this, we calculated the induced plasticity as the ratio of CS+/CS- and showed that non-anesthetized flies show significantly increased plasticity when compared to anesthetized animals (Figure 2I). These representative results indicate that this protocol is suitable for studying learning and memory under the microscope and ensures the fly's health during imaging from neurons for a long period of time. This protocol is versatile enough and can be applied to other types of in vivo imaging and other neurons of interest.

Figure 1: In vivo imaging preparation of Drosophila without the use of anesthesia. (A) Top panel: Diagram illustrating the steps involved in aspirator preparation. Bottom panel: Diagram showing the chamber held in a micromanipulator during fly positioning. The fly's proboscis is placed at the opening of metal tubing connected to a vacuum at 500 mL/min using an aspirator. The chamber is subsequently positioned above the fly. (B) The sides of the eyes and thorax are glued with myristic acid using a custom-made platinum resistor. (C) The fly's proboscis is removed from the metal tubing, and the chamber is inverted to allow the proboscis to be glued in place, thereby minimizing movement during recordings. (D) The chamber is attached to a glass slide base. Tape is applied to the top of the chamber, covering the antennae, and to the back side to prevent saline from leaking. (E) Two vertical cuts are made alongside the cuticle near the compound eyes, followed by a third cut across the cuticle near the thorax, creating a "window" to the brain. (F) Between 100-200 µL of saline is added to the head of the fly at the site of the cuts. (G) The cuticle window is lifted, pulled, and detached using sharp, fine forceps to expose the brain. (H) Fat and tracheal tissue are removed to enable clear visualization of brain cells under the microscope. (I) The prepared fly is placed under the microscope and positioned with its body aligned on top of the shock grid. (J) Pseudo-color image showing the expression pattern of MBON γ2α′1. The region of interest is outlined with a dashed line surrounding the dendritic arbor. Scale bar: 50 µm. Please click here to view a larger version of this figure.

Figure 2: Learning-induced plasticity in MBON γ2α′1 following aversive olfactory conditioning. (A) Expression pattern of the MB077C-splitGal4 driver expressing GCaMP. Brains were dissected and stained with anti-GFP and counterstained with anti-nc82. Scale bar: 50 µm. (B) Pseudo-color image of MBON γ2α′1 highlighting the cell body, extending process, and dendritic arbor. The dendritic region represents the area from which calcium responses were collected. Scale bar: 50 µm. (C) Schematic of the aversive training protocol. Flies were exposed to 5-s pulses of MCH and OCT with 30 s of clean air in between. After a 5-min interval, flies were trained to associate OCT with twelve 90 V electric shocks. Recordings of calcium responses to MCH and OCT were performed 5 min and 15 min post-training. (D) Odor response traces to MCH (top panel) and OCT (bottom panel) are shown for pre-training, and at 5 min and 15 min post-training. Thick black bars indicate the timing of odor presentation. (E) Pseudo-color images of the dendritic arbor of MBON γ2α′1 pre-training, and 5 min and 15 min post-training, in response to both MCH and OCT. Potentiation of fluorescence is observed for the CS- (MCH), while fluorescence is abolished for the CS+ (OCT). Scale bar: 20 µm. (F) The mean odor response to MCH is significantly potentiated at 5 min and 15 min post-training, whereas responses to OCT (CS+) are significantly suppressed. Non-parametric Wilcoxon paired test, N = 12. (G) Odor response traces to MCH (top panel) and OCT (bottom panel) are shown for anesthetized flies pre- and 5 min post-training. Thick black bars indicate odor delivery. (H) In anesthetized animals, mean odor responses to MCH do not show significant potentiation post-training, and responses to OCT exhibit only partial suppression compared to non-anesthetized animals. (I) The training-induced plasticity, calculated as the CS+/CS- ratio, is significantly lower in anesthetized flies compared to those prepared without anesthesia. Non-parametric Wilcoxon paired test, N = 8-12. Please click here to view a larger version of this figure.

Supplementary File 1: Grid_holder (STL). STL file of the custom-designed component used to hold the electric grid. Please click here to download this File.

Supplementary File 2: Grid_holder (OBJ). OBJ file of the custom-designed component used to hold the electric grid. Please click here to download this File.

Supplementary File 3: Chamber_LaserCut. Adobe Illustrator file used for laser cutting the top component of the recording chambers for both male and female flies. Please click here to download this File.

Supplementary File 4: Recording_chamber (STL). STL file of the custom-designed recording chamber. Please click here to download this File.

Supplementary File 5: Recording_chamber (OBJ). OBJ file of the custom-designed recording chamber. Please click here to download this File.

Supplementary File 6: Recording_chamber_animation. MP4 video showing an animation of the assembled recording chamber. Please click here to download this File.

Understanding the neural circuity behind learning and memory is a central aim in the field of neuroscience. The genetic toolkit available in Drosophila, combined with the simplicity and versatility of behavioral testing, makes it an ideal model organism for studying these processes. Published studies from various labs have described different protocols for preparing flies for functional imaging without the use of anesthesia^44,45. However, these protocols vary widely and often rely on highly specialized, custom-made tools with limited guidance on how to build or use them. In this protocol, a method for performing in vivo functional imaging in flies that avoids the confounding effects of general anesthesia is presented. Furthermore, this approach allows for the capture of real-time neuronal activity within the same fly, enabling direct comparisons of neural responses before and after exposure to various stimuli.

When completing this protocol, critical steps include making sure the fly's proboscis is properly placed into the vacuum tubing. This step ensures that subsequent gluing will occur in the right spots, and it makes sure the head is at the correct angle to visualize the neurons of interest. Minimal glue application is essential to maintain fly health; using the smallest amount necessary to secure the fly in place is ideal. This protocol also allows some modifications to match specific needs, such as gluing only the sides of the eyes so the entire body is free. Similarly, a perfusion chamber can be attached to the top of the recording chamber to facilitate ringer perfusion or drug delivery. Finally, a rotating floating treadmill can be placed below the fly as previously reported, allowing the analysis of locomotion directionality^46,47.

This protocol does have some limitations, including the stress the flies are subjected to during the microsurgery to prepare the flies for live imaging. Handling the small flies, applying glue sparingly, and carefully opening the head capsule without damaging any brain tissue or body requires substantial expertise, which can only be acquired through meticulous practice and repetition. While this method is particularly beneficial for studying cognitive processes at the cellular level, in vivo imaging still has some limitations when compared to electrophysiological recordings. Additionally, while our protocol avoids the use of anesthesia prior to imaging, the fly is still constrained to a physical chamber, which can always alter cellular responses. Nonetheless, our methodology provides the closest way to preserve the fly's health for optimal in vivo imaging. Compared to existing methods that use anesthesia and constrain the fly, our method aims to keep the fly as freely moving as possible while being able to image from cells with minimal movement.