A TRANSPARENT WHEEL-BASED PLATFORM FOR LOCOMOTION-ON-DEMAND AND MULTI-VIEW BODY AND FACIAL KINEMATICS IN HEAD-FIXED MICE

This paper presents a low-cost, modular behavioral platform for head-fixed mice that combines a transparent running wheel with air-stream stimulation and multi-view videography to enable precise, stimulus-evoked locomotion-on-demand while simultaneously capturing ventral paw, body, and facial kinematics compatible with neural recording techniques.

The Gist

Imagine you want to study how a mouse's brain controls its legs, face, and eyes all at the same time. The problem is, mice are tiny, fast, and hard to watch while they run. Usually, scientists have to choose: either they watch the mouse run on a treadmill (but can't see its belly or paws clearly), or they freeze the mouse's head to take brain pictures (but then the mouse can't run naturally).

Paranjape and his team have built a "magic treadmill" that solves this puzzle.

Here is a simple breakdown of what they did, using some everyday analogies:

1. The "Glass Slipper" Treadmill

Most running wheels for mice are solid plastic or metal. If you put a camera under them, you can't see the mouse's feet because the wheel blocks the view.

• The Innovation: They built a transparent running wheel (like a clear glass slipper).
• Why it matters: Because the wheel is see-through, they can place cameras underneath it. It's like having a "bottom-up" view of a car driving on a clear road, allowing them to film the mouse's paws stepping in perfect detail while the mouse is running.

2. The "Gentle Breeze" Starter

Usually, to get a mouse to run in a lab, scientists have to starve it slightly and give it water as a reward (like a dog waiting for a treat). This takes weeks of training and makes the mouse stressed.

• The Innovation: Instead of a treat, they use a tiny puff of air blown at the mouse's back.
• The Analogy: Think of it like a gentle breeze from a fan that says, "Hey, time to run!" The mouse has an instinct to run away from the air.
• Why it matters: The mouse starts running immediately without needing weeks of training or being hungry. It's like a "locomotion-on-demand" button.

3. The "Swiss Army Knife" of Cameras

The setup doesn't just watch the feet. It's a multi-angle surveillance system.

• The Setup: They have cameras looking at the mouse's belly (paws), its face, and even its eyes.
• The Analogy: Imagine a sports broadcast where you have cameras on the field, in the locker room, and a drone in the sky, all showing the same game at the exact same time.
• The Result: They can see the mouse's legs moving, its whiskers twitching, its face scrunching up, and its pupils changing size, all while the mouse is running.

4. The "Sync-Beep" (The Secret Sauce)

The biggest headache in science is making sure all these different cameras and computers agree on when things happen. If Camera A says "Air started at 1:00" and Camera B says "Air started at 1:01," the data is messy.

• The Innovation: They put a red LED light in the view of every single camera.
• The Analogy: Imagine a conductor waving a baton. Every time the air puff starts, the red light flashes. Because the light is visible in every video, the scientists can look at the footage later and say, "Ah, the light flashed here, so the air started here."
• Why it matters: It perfectly syncs the video of the paws, the face, and the brain data, even if the cameras are running at slightly different speeds.

5. What Did They Find?

They tested this on a few mice and found:

• It works: The air puff makes the mice run instantly and consistently.
• It's coordinated: When the air hits, the mouse doesn't just scramble; it runs with a smooth, rhythmic gait.
• It's expressive: The mice's faces and bodies move in a specific, predictable way when they start running.

Why Should You Care?

This isn't just about watching mice run. This setup is a universal translator for the brain.

• For Neuroscientists: It allows them to film the brain (using special microscopes) while the mouse is doing complex, natural movements.
• For Disease Research: If a mouse has a disease that affects movement (like Parkinson's or aging), this system can spot tiny, subtle changes in how the mouse walks or blinks long before the mouse stops moving entirely.
• For Efficiency: It's cheap, easy to build (using 3D printers and common electronics), and doesn't require starving the animals.

In short: They built a high-tech, see-through gym where mice can run on command, allowing scientists to finally see the "whole picture" of how the brain controls movement, from the tip of the nose to the tip of the tail.

Technical Summary

1. Problem Statement

Understanding how the brain generates coordinated movement requires experimental paradigms that simultaneously capture precise, multi-modal behavioral data (whole-body kinematics, facial dynamics, and autonomic signals) while remaining compatible with high-resolution neural recording techniques (e.g., two-photon imaging, electrophysiology).

• Limitations of Current Systems: Existing head-fixed setups often rely on opaque treadmills or conveyor belts, which obstruct ventral views of the paws. Many systems capture only limited behavioral readouts (e.g., speed or whiskers) via single-camera streams.
• Training Constraints: Traditional reward-based locomotion tasks require extensive water deprivation and weeks of training, limiting throughput and experimental efficiency.
• Synchronization Challenges: Integrating multiple video streams with different sampling rates and independent hardware clocks with neural or stimulus data often leads to temporal misalignment.

2. Methodology

The authors developed a low-cost, modular, open-source behavioral platform called the AIR (Air-Induced Running) Wheel system.

Hardware Design

• Transparent Running Wheel: A non-motorized, transparent hamster wheel mounted on a vertical aluminum rod. This design allows unobstructed ventral (bottom) and side views of the animal's paws and body during locomotion.
• Head-Fixation: A custom 3D-printed headplate (PLA) is secured to the mouse skull, allowing stable fixation relative to the wheel and cameras.
• Stimulus Delivery: An air-stream system delivers brief puffs of air (~10 PSI) to the mouse's lower back to induce locomotion. The airflow is controlled by a solenoid valve driven by an Arduino microcontroller.
• Sensors & Acquisition:
  • Rotary Encoder: A quadrature encoder (1000 CPR) mounted on the wheel shaft measures wheel position and speed at 5 kHz via a National Instruments DAQ card.
  • Multi-View Videography: Three cameras are used:
    1. Ventral View: Captures paw movements (1440x1080, 60Hz).
    2. Facial View: Captures facial dynamics (1440x1080, 60Hz).
    3. Pupil View: Captures eye/pupil dynamics (320x240, 60Hz).
  • Illumination: Infrared (IR) lights are used for all cameras to allow recording in total darkness.
• Synchronization Mechanism: A critical innovation is the use of a red LED driven by the same TTL signal as the air solenoid. The LED is placed within the field of view of all cameras, providing a visual timestamp for air-on/air-off events directly in the video streams.

Experimental Paradigm

• Stimulus-Induced Locomotion: Mice are trained to run on the wheel in response to air puffs. The air is delivered until the mouse runs a predefined distance (e.g., 25 cm), at which point the air stops.
• Training: Animals undergo habituation to head fixation and task training (2–14 days), significantly shorter than reward-based protocols.
• Subjects: The study utilized one representative adult male C57BL/6J-Tg mouse (GCaMP6s) and a cross-animal reproducibility cohort of three wild-type mice.

Data Analysis Pipeline

• Kinematics:
  • Optical Flow: Dense optical flow (Dual TV-L1) and motion-energy metrics were computed to quantify gross movement speed and intensity.
  • Pose Estimation: DeepLabCut was used to track specific body landmarks (fore/hind paws, tail base, nose) with high confidence (>0.9 likelihood).
• Synchronization: Video streams were aligned to the DAQ-recorded air-control signal by detecting the LED transitions in the video, correcting for minor frame-rate discrepancies between independent cameras.

3. Key Contributions

1. Transparent Wheel Platform: The first implementation of a transparent running wheel combined with air-induced locomotion, enabling direct ventral imaging of paw kinematics previously obscured in opaque systems.
2. Stimulus-on-Demand: A robust, reward-free paradigm that induces reliable locomotion within minutes of air delivery, eliminating the need for prolonged water deprivation and extensive training.
3. Multi-View Synchronization: A novel, low-cost LED-based synchronization method that aligns independent video streams (paws, face, pupil) with high temporal precision (sub-millisecond error) relative to stimulus delivery, despite varying camera clocks.
4. Open-Source Ecosystem: The authors provide full schematics, 3D printing files, assembly guides, and analysis code (MATLAB/Python) to facilitate replication.

4. Results

• Reliable Locomotion: Air delivery consistently triggered structured locomotion. In the representative animal, 34 trials showed a significant increase in running speed during "air-on" epochs (mean 5.09 cm/s) compared to "air-off" epochs (mean 2.93 cm/s) ($p < 0.001$).
• Kinematic Precision:
  • Paw Movement: Optical flow and DeepLabCut analyses revealed robust, stimulus-locked increases in paw speed and motion energy. All tracked body parts (paws, tail, nose) showed coordinated movement patterns.
  • Temporal Dynamics: Running speed increased rapidly (within 1–2 seconds) after air onset and decayed gradually after air offset, demonstrating stable trial structure.
• Facial and Autonomic Dynamics:
  • Facial Motion: Significant increases in facial motion (optical flow speed and motion energy) were observed during locomotion, confirming that air-induced running engages the whole body, including the face.
  • Pupillometry: No significant change in pupil area was detected during air-on epochs. The authors attribute this to the dark experimental conditions (IR illumination) causing maximal pupil dilation, establishing a baseline for future studies under different lighting.
• Cross-Animal Reproducibility: The paradigm was successfully replicated in a separate cohort of three mice, showing consistent speed modulation and kinematic patterns across animals.
• Synchronization Accuracy: LED-based event detection confirmed that air-on/off transitions were aligned across all video streams with errors well below one video frame (<0.01 frames).

5. Significance

This study presents a versatile, scalable framework for studying the neural basis of motor control and behavioral state transitions.

• Neural Integration: The system is explicitly designed for integration with neural recording modalities (two-photon imaging, electrophysiology), allowing researchers to correlate stimulus-evoked locomotion with neural activity.
• Efficiency: By replacing reward-based conditioning with air-induced locomotion, the system drastically reduces training time and animal stress, increasing experimental throughput.
• Comprehensive Phenotyping: The ability to simultaneously capture ventral paw kinematics, facial expressions, and eye dynamics in a single trial provides a holistic view of motor output and arousal.
• Future Applications: The platform is well-suited for investigating motor deficits in neurodegenerative disease models, studying sensorimotor integration, and exploring the neural circuits underlying state transitions (immobility to locomotion).

In summary, Paranjape et al. have successfully bridged the gap between high-fidelity behavioral measurement and neural recording by creating a transparent, modular, and synchronized platform that overcomes the limitations of traditional head-fixed locomotion assays.