Deep-prior ODEs augment fluorescence imaging with chemical sensors.

Area of Science:

• Cellular biology
• Biophysics
• Computational imaging

Background:

• Designing fluorescent sensors is crucial for studying biological signaling.
• Sensor binding kinetics are often overlooked, leading to interpretation artifacts in fluorescence measurements.
• Accurate reconstruction of chemical messenger concentrations is essential for understanding cellular processes.

Purpose of the Study:

• To develop a method for reconstructing spatiotemporal concentrations of chemical messengers, considering sensor binding kinetics.
• To improve the interpretation of fluorescence measurements from biological sensors.
• To validate the method using GCaMP calcium sensors and analyze calcium distribution in neurons.

Main Methods:

• Developed a method to fit fluorescence data constrained by chemical reactions.
• Integrated a deep neural network prior to enhance data fitting.
• Applied the method to GCaMP calcium sensors to recover messenger concentrations.
• Analyzed the spatiotemporal distribution of calcium in single neurons.

Main Results:

• Recovered concentrations showed a common temporal waveform irrespective of sensor kinetics.
• Ignoring binding kinetics and assuming equilibrium introduced artifacts in concentration measurements.
• The method successfully revealed distinct spatiotemporal calcium events in single neurons.
• The computational approach improved the accuracy of biological signaling studies.

Conclusions:

• Incorporating physical constraints, specifically binding kinetics, is vital for accurate computational imaging of cellular signals.
• The proposed method enhances the utility of current chemical sensors.
• This work provides a more robust approach to analyzing biological signaling dynamics.