November 3, 2026
Host
Welcome to the podcast. Today we're discussing a fascinating paper on how fruit flies control their spontaneous flight turns. The paper is titled 'Descending control and regulation of spontaneous flight turns in Drosophila' by Ivo Ros, Jaison Omoto, and Michael Dickinson. Can you give us a quick overview?
Guest
Absolutely. This paper investigates the neural circuits that generate and regulate body saccades—those rapid turns flies make during flight. These turns are crucial for foraging, but flies suppress them during long-distance dispersal or when tracking odors. The study identifies key neurons and a regulatory mechanism.
Host
Why are these saccades so important for flies?
Guest
Well, saccades help flies scan their environment quickly. They also reduce motion blur by restricting it to brief moments. And there's a hypothesis that comparing sensory input before and after a saccade helps flies estimate wind direction. So they're really multifunctional.
Host
Interesting. And the paper focuses on descending neurons, right?
Guest
Yes. Descending neurons carry commands from the brain to the ventral nerve cord, which controls the wings. The authors identified two pairs of descending neurons, DNae014 and DNb01, that seem to drive saccades. They also found an inhibitory neuron, VES041, that suppresses saccades.
Host
Let's get into the methods. How did they study these neurons?
Guest
They used a combination of techniques. First, they created sparse split-GAL4 lines to target specific neurons. Then they used 2-photon calcium imaging to record neural activity while tracking wing motion. They also used optogenetics to activate or silence neurons, and genetic ablation to remove them. Finally, they used connectomics to map the connections.
Host
That's a lot. Let's start with the split-GAL4 lines. Why were they necessary?
Guest
Previous studies used broad driver lines that affected many neurons, which made it hard to interpret results. For example, a prior study implicated the AX neuron, now called DNae014, but another study found no effect when silencing cells in that line. So the authors built a sparse split-GAL4 line specifically for DNae014. They also used a sparse line for DNb01.
Host
How did they build the split-GAL4 line for DNae014?
Guest
They combined two hemidrivers: VT025718-GAL4.AD and R56G08-GAL4.DBD. This resulted in expression only in DNae014, with minimal off-target cells. They confirmed the specificity with confocal imaging.
Host
And for DNb01, they used an existing line?
Guest
Yes, SS02383, which is also very sparse. It does target some ocellar neurons, but the main cells are DNb01.
Host
Now, 2-photon imaging. How did they record neural activity?
Guest
They tethered flies in a flight simulator and used a 2-photon microscope to image calcium signals from the neurons. They expressed GCaMP7f, a calcium indicator, and tdTomato for normalization. They imaged the arbors of the left and right neurons in the superior posterior slope of the brain.
Host
What kind of analysis did they do on the imaging data?
Guest
They corrected for brain motion using cross-correlation. They defined regions of interest based on pixel variance and covariance with turning behavior. They normalized fluorescence to tdTomato to account for z-motion. They also subtracted background and normalized to the maximum response for each fly.
Host
That sounds complex. Let's move to the results. Figure 1 shows the correlation between neural activity and saccades. What did they find?
Guest
They found that the difference in activity between left and right DNae014 cells correlated very strongly with the difference in wingstroke amplitude. The R-L ΔF/F explained about 89% of the variance in L-R wingstroke amplitude. For DNb01, it was about 90%. So the activity of these neurons is a remarkably good predictor of saccades.
Host
And they saw a flip-flop relationship for DNae014?
Guest
Yes. When they plotted left versus right DNae014 activity, it was highly non-linear and inverse. When one side is active, the other is suppressed. This is analogous to a flip-flop circuit in electronics, which can hold a binary state. It's a common motif for binary decisions, like turning left or right.
Host
What about DNb01?
Guest
DNb01 also showed an inverse relationship, but it was more linear. The authors note that this difference might not be functionally relevant because calcium signals are non-linear themselves. But both neurons correlate with ipsilaterally directed saccades.
Host
And they saw a one-to-one correspondence between neural activity and wing motion?
Guest
Yes. Although they couldn't record both neurons simultaneously, every rapid change in wing motion corresponded to a transient in the activity of either neuron when recorded separately. This suggests they are co-active during saccades.
Host
Let's talk about Figure 2. They presented visual looming stimuli. What happened?
Guest
They presented expanding disks at different azimuthal positions. Both DNae014 and DNb01 responded more strongly on the side opposite the loom. So a loom on the left activated the right neurons. This makes sense for an avoidance response. DNae014 had larger responses, but both showed similar directionality.
Host
Did they see any wing motion responses?
Guest
Yes, the wingstroke amplitude also changed in response to the loom, consistent with a turn away from the stimulus. The neural responses preceded the wing motion.
Host
Now Figure 3. They activated the neurons unilaterally. What did they do?
Guest
For DNae014, they used 2-photon excitation of CsChrimson to activate either the left or right cell. Activation of the left cell caused a left saccade, and the right cell caused a right saccade. This confirmed the correlation data.
Host
And for DNb01, they used a different method?
Guest
Yes, because DNb01 branches in both hemispheres, they couldn't use focal 2-photon activation. Instead, they used the SPARC system to randomly express CsChrimson in one or both cells. They then scored expression patterns after behavior. Unilateral activation of DNb01 also caused syndirectional saccades, but they were smaller and slower than those from DNae014.
Host
So both neurons can drive turns, but with different dynamics?
Guest
Exactly. This suggests they drive different components of the saccade motor program. DNae014 might be involved in the fast, large-amplitude component, while DNb01 might contribute to a slower, smaller component.
Host
Figure 4 is about ablation. What did they do?
Guest
They genetically ablated either DNae014 or DNb01 using the pro-apoptotic genes reaper and hid. Then they measured saccade frequency in two flight arenas: a rigid tether and a magnotether.
Host
What were the results?
Guest
In the rigid tether, control flies had a saccade rate of 0.48 Hz. Ablating DNae014 reduced it to 0.09 Hz, and ablating DNb01 reduced it to 0.12 Hz. Both were highly significant. In the magnotether, control was 0.37 Hz, and both ablations reduced it to about 0.20 Hz.
Host
So ablation didn't eliminate saccades completely?
Guest
Right. The remaining saccades were slower and smaller. They also found that the maximum inter-saccade interval increased from about 13 seconds in controls to 33 seconds in ablated flies. And the ablated flies had longer sequences of straight flight interspersed with bursts of saccades.
Host
What does that tell us?
Guest
It suggests that each neuron can produce partial saccades on its own, but they normally work together to produce full saccades. The altered temporal pattern also indicates that these neurons contribute to the timing of spontaneous saccades, not just relaying commands.
Host
Now Figure 5. They used connectomics. What did they find?
Guest
They searched the FlyWire and hemibrain connectomes for neurons that provide input to both DNae014 and DNb01. They identified a set of input neurons called SGUIs—saccade-generating unit input neurons. There were 8 types, including both excitatory and inhibitory neurons.
Host
And they found a special neuron, VES041?
Guest
Yes. VES041 stood out because it provides direct monosynaptic input to all four saccade command neurons—both DNae014 and DNb01 on both sides. It's predicted to be GABAergic, so inhibitory. This means a single pair of VES041 cells could inhibit the entire saccade network.
Host
That's a powerful way to suppress turns. Figure 6 shows what happens when they activate VES041.
Guest
They created split-GAL4 lines for VES041 and expressed CsChrimson. When they activated VES041 with red light, saccade rate dropped dramatically. In the magnotether, it went from 0.52 to 0.17 Hz in one line, and from 0.71 to 0.09 Hz in another. Control flies showed a slight increase, probably due to light arousal.
Host
So VES041 is a potent inhibitor of saccades. What does that mean for the fly's behavior?
Guest
It suggests that VES041 promotes straight flight. The authors hypothesize that it regulates the transition between local search, where saccades are frequent, and long-distance dispersal, where straight flight is favored. It might also be involved in suppressing saccades during upwind odor tracking.
Host
Let's discuss the overall hypothesis. How do these neurons work together?
Guest
The model is that DNae014 and DNb01 form two saccade-generating units, one for left turns and one for right. Each unit consists of one excitatory and one inhibitory neuron. They project to the wing neuropil and drive the motor program for saccades. VES041 inhibits both units, suppressing saccades.
Host
Why have both an excitatory and an inhibitory neuron in each unit?
Guest
That's a good question. The authors suggest that the two neurons drive different components of the motor program. DNae014 is cholinergic and excitatory, while DNb01 is glutamatergic and inhibitory. They might activate different sets of premotor neurons or muscles to produce the full saccade. The exact roles are still being worked out.
Host
I see. Now, I have a challenge for you. The paper shows that ablating either neuron reduces saccade rate but doesn't eliminate saccades. Could it be that these neurons are not actually command neurons but just modulators?
Guest
That's a fair point. The authors acknowledge that the remaining saccades could be generated by an alternative pathway. However, the fact that unilateral activation reliably evokes saccades, and that the activity correlates so tightly with spontaneous saccades, strongly supports a command role. The partial loss-of-function might be because the two neurons work together, and removing one leaves the other to produce weaker saccades. They couldn't do a double ablation, but VES041 activation, which inhibits all four neurons, suppresses saccades almost completely. So the evidence points to them being key command neurons.
Host
That makes sense. What about the evolutionary angle?
Guest
The authors note that these neurons are likely part of an ancient network. They are homologous to flip-flop neurons in silkmoths that control pheromone-triggered turns. This suggests a conserved mechanism for steering across insects.
Host
Fascinating. Let's wrap up. What are the main takeaways?
Guest
The paper identifies a core circuit for spontaneous flight turns in Drosophila. It shows that two pairs of descending neurons, DNae014 and DNb01, form saccade-generating units. A pair of inhibitory interneurons, VES041, suppresses these units to promote straight flight. This circuit likely regulates the switch between local search and dispersal, which is ecologically important.
Host
Thank you for this detailed discussion. It's amazing how much we can learn from a tiny fly.
Guest
Indeed. And there's still much to discover about how sensory cues regulate VES041 and how the motor program is executed in the ventral nerve cord.
Host
That's all for today. Thanks for listening.