← All projects Instrumentation · Neural circuits · Nature Methods

FlyMAD: dialling courtship arousal up and down

A live-tracking system that aims a laser at a freely walking fruit fly so specific neurons can be switched on and off mid-behaviour. I then used it to study how courtship arousal is controlled.

Where
Institute of Molecular Pathology (Vienna) & HHMI Janelia Research Campus
When
2011 – 2016
Role
PhD researcher & Visiting Scientist (with Barry Dickson); first author
Tools
Real-time tracking, steerable lasers, opto- & thermogenetics, computer vision

The question

When I started, behaviour research was in its technical infancy: behaviour was scored by hand, and nobody could switch specific neurons on and off with precise timing in an animal that was free to move. Without that, you can’t ask what a neuron does at the moment a behaviour happens.

The build

  • Worked with software engineers to build real-time tracking that steers an optical beam path onto a moving fly
  • Used the same beam path for high-resolution imaging of the fly and for laser light that activates neurons with opto- and thermogenetic tools
  • Measured the resulting behaviour automatically with computer vision instead of scoring it by hand
  • Made neuronal manipulation fast enough to switch a behaviour on and off within the same experiment
Dual-wavelength FlyMAD. A tracking camera finds the fly; real-time tracking steers a galvanometer mirror so a high-resolution camera follows it, and an 808 nm infrared laser (thermogenetics) or a 635 nm red laser (optogenetics) is delivered along the same light path.
Dual-wavelength FlyMAD. A tracking camera finds the fly; real-time tracking steers a galvanometer mirror so a high-resolution camera follows it, and an 808 nm infrared laser (thermogenetics) or a 635 nm red laser (optogenetics) is delivered along the same light path.
Behaviour measured automatically, frame by frame: wing angle during courtship song, and distance to the target.
Behaviour measured automatically, frame by frame: wing angle during courtship song, and distance to the target.

The discovery

Using FlyMAD, I found a neuronal trigger for courtship arousal. I then found that activating excitatory and inhibitory neurons in concert with one another allowed me to dial the level of courtship arousal up and down.

These findings are unpublished.

The courtship circuit, simplified: P1 neurons commit the fly to courtship, mAL neurons cancel it.
The courtship circuit, simplified: P1 neurons commit the fly to courtship, mAL neurons cancel it.
mAL (green) and P1 (magenta) neurons in the fly brain.
mAL (green) and P1 (magenta) neurons in the fly brain.

Outcome

The instrument was published as Bath, Stowers, Hörmann, Poehlmann, Dickson & Straw (2014), “FlyMAD: rapid thermogenetic control of neuronal activity in freely walking Drosophila”, Nature Methods 11: 756–762. Related circuit work appeared in McKellar et al. (2019), Current Biology 29: 426–434. Presented as invited talks at Cold Spring Harbor (2013) and the Max Planck Department of Circuits, Computation & Models (2014), and as a poster at the Ascona Neural Circuits Meeting (2015).

My contribution

First author on FlyMAD. I led the biological design and the experiments, co-developed the instrument with the engineering team, and carried out the circuit experiments.

Videos

How behaviour was measured: classical computer vision, with no training data and no neural networks.
FlyMAD in action, seen from outside the rig.
The view from the cockpit.
Experiment playback: tracking, wing angle and distance as neurons are switched on with the laser.
Wing extension tracked automatically in every frame.
3D reconstruction of the courtship neurons.

Gallery