Speed dependent descending control of freezing behavior in Drosophila melanogaster

Abstract

The most fundamental choice an animal has to make when it detects a threat is whether to freeze, reducing its chances of being noticed, or to flee to safety. Here we show that Drosophila melanogaster exposed to looming stimuli in a confined arena either freeze or flee. The probability of freezing versus fleeing is modulated by the fly's walking speed at the time of threat, demonstrating that freeze/flee decisions depend on behavioral state. We describe a pair of descending neurons crucially implicated in freezing. Genetic silencing of DNp09 descending neurons disrupts freezing yet does not prevent fleeing. Optogenetic activation of both DNp09 neurons induces running and freezing in a state-dependent manner. Our findings establish walking speed as a key factor in defensive response choices and reveal a pair of descending neurons as a critical component in the circuitry mediating selection and execution of freezing or fleeing behaviors.

Fig. 1

- Flies jump in response to repeated looming. a Schematic of behavioral assay. b Schematic of visual stimuli, 300 flies were tested in each condition. c Number of jumps detected per fly during the 5 min stimulation period. Center line, median; box limits, upper (75) and lower (25) quartiles; whiskers, 1.5x interquartile range. d Jump timing within a 1 s window around looming stimuli. Dashed lines indicate looming onset and offset. Top, size of looming disk (visual angle). e Proportion of flies jumping throughout the 10 min session. Dashed lines indicate stimulus presentations. *** denotes p < 0.001

Fig. 2

Flies freeze upon looming stimulation. a Speed raster for a random subset of 50 flies ordered by average speed during stimulation (ascending). Each row corresponds to one fly and each vertical line to 500 ms bins. Bar on top indicates the 5 min stimulation period. Color bar values refer to mm s⁻¹. b Proportion of freezing flies (n = 300 for each condition). Dashed lines represent stimulus presentations. c Startle behavior during freezing. Average (±s.e.m.) number of pixels changing around the fly in a 1 sec window around looming stimuli, including only trials where flies were freezing before and after the stimulus (n = 1434). Dashed lines indicate looming onset and offset. Top, size of looming disk (visual angle). d Distribution of individual flies by time spent freezing during the stimulation period. e Proportion of freezing flies to a shorter stimulation (5 looming presentations, n = 100). Dashed lines represent looming presentations. f Distribution of individual flies by time spent freezing during the stimulation period for the shorter stimulation experiment. *** denotes p < 0.001

Fig. 3

Flies flee to looming stimuli. a Average (±s.e.m.) fly speed including only time periods classified as walking. Dashed lines indicate stimulus presentations (in a, b looming n = 280 flies, control n = 298 flies). b Change in walking speed caused by stimulation (baseline period subtracted from stimulation period). c Example trajectories of walking trials. Black section corresponds to looming window and blue sections 500 ms before and after looming. d, e Distribution of path orientations before (d) and after (e) looming. Bar height indicates counts. Stimulus source (screen) was located at 90°. In d–f, only looming events where flies were walking before and after the stimulus were included (looming n = 1574 trials, control n = 2881 trials). f Looming-triggered speed profile. Average (±s.e.m.) speed in a 1 s window around looming for all walking trials. Blue numbers represent four stages of the response to looming: 1- pre-looming, 2- pause, 3- run, 4- post-looming. Top, size of looming disk (visual angle). g Looming-triggered speed profile of walking trials from the looming condition separated into responses that included a pause (dark blue, n = 806) and responses that did not (light blue, n = 767). h Change in walking speed caused by stimulus presentation (pre-looming period subtracted from post-looming period). i Fraction of flies performing the described behaviors for each of the 20 looming presentations. *** denotes, p < 0.001, ns not significant. Box plot elements: center line, median; box limits, upper (75) and lower (25) quartiles; whiskers, 1.5× interquartile range

Fig. 4

Response to looming is modulated by walking speed. a Looming-triggered speed profile (average ± s.e.m.) including only transition trials where flies were not freezing before the stimulus (n = 2501 loomings), separated into events that led to freezing (light gray, n = 346) and events that did not (dark gray, n = 2155). b Distribution of fly speed 500 ms before loomings where flies were not freezing (blue bars, n = 2501 loomings) overlaid with the probability of freezing after the stimulus for each speed interval (black dots and trace). X-error bars shows standard deviation (s.d.) for speed sampled within each interval and Y-error bars show 95% confidence intervals (CI). a, b Show data from the experiment in Figs. 1–3, whereas panels c–e show data from closed-loop experiments. In c–e, dark green corresponds to high speed (n = 56) group and light green to low speed group (n = 60). c Average + s.e.m. speed during baseline period for high and low speed groups. d Proportion of freezing flies. Dashed line indicates onset of stimulation. e Percent time spent freezing during stimulation period. *** denotes p < 0.001, ns not significant. Box plot elements: center line, median; box limits, upper (75) and lower (25) quartiles; whiskers, 1.5× interquartile range

Fig. 5

Silencing DNp09 neurons disrupts freezing but not running. a DNp09 morphology. DNp09 split-GAL4 driving membrane-bound GFP (green) combined with anti-synaptotagmin-HA staining (purple). In b–h, red corresponds to DNp09 > Kir2.1 flies, dark gray to DNp09/ + flies and light gray to Kir2.1/ + flies. In b–d, n = 60 flies for each condition. b Proportion of freezing flies. Dashed lines represent stimulus presentations. c Percent time spent freezing during stimulation period. d Average (±s.e.m.) fly speed including only time periods classified as walking. Dashed lines indicate stimulus presentations. e Looming-triggered speed profile (average ± s.e.m.) of walking trials that include a pause (DNp09 > Kir2.1 n = 234, DNp09/ + n = 253, Kir2.1/ + n = 259). f Average ( + s.e.m.) speed during baseline period (n = 60 for all conditions). In g–h, n = 50 flies for each group. g Proportion of freezing flies for low speed closed-loop looming stimulation. Dashed line indicates onset of stimulus presentations. h Percent time spent freezing during stimulation period for closed-loop experiment. ** denotes p < 0.01, *** denotes p < 0.001, ns not significant. Box plot elements: center line, median; box limits, upper (75) and lower (25) quartiles; whiskers, 1.5x interquartile range

Fig. 6

Activation of DNp09 descending neurons leads to freezing. a Schematic of experimental set-up for CsChrimson stimulation and stimulation protocol. Test flies supplemented with retinal (n = 80) and control flies raised in normal food (n = 72). b Fraction of flies freezing. Dashed lines indicate light stimulation. c Fraction of flies freezing aligned on light activation (retinal n = 800, control n = 720 stimulation events). d Stimulus-triggered speed profile (average ± s.e.m.) aligned on light presentation for stimulations that induced freezing (dark blue, n = 492), stimulations that did not (light blue, n = 308) and control (black, n = 720). e Probability ( + 95% CI) of jumping at light offset for control and test flies (black and red bars). Probability of jumping at light offset for stimulation events of test flies that induced freezing and events that did not (dark and light blue bars). f Probability of freezing to red light stimulation as a function of pre-stimulation speed (r² = 0.87, p = 0.007). X-error bars show s.d. for speed sampled within each interval and Y-error bars show 95% CI. g Probability of freezing for DNp09 > CsChrimson flies (n = 50 for each condition) tested at different speeds (number of stimulation events for very high, high and low were 403, 674 and 593, respectively). *** denotes p < 0.001