An automated system for measuring the electrical properties of single cells can operate for hours without human input. Developed by the Precision Biosystems Laboratory at Georgia Institute of Technology and Massachusetts Institute of Technology, the “PatcherBot” combines machine vision, micromanipulators and a pipette-cleaning module to autonomously perform the patch-clamp technique, one of the more laborious tasks in biological research. By greatly increasing laboratory throughput, the device could speed up any investigation involving single-cell electrical recordings, such as drug development and neuronal connection profiling (J. Neural Eng. 10.1088/1741-2552/ab1834).

Some cellular properties are amenable to study by high-resolution, high-throughput techniques like fluorescence microscopy or photoacoustic microscopy. When it comes to the electrophysiology of single cells, however, researchers rely on the far more time-consuming patch-clamp method. In this technique, a glass pipette filled with an electrolyte solution is pressed against the membrane of the cell under observation. By applying a gentle suction to the cell, a seal is achieved and the voltage and current across the cell membrane can be measured.

“The user sucks on the end of a plastic tube coupled to the pipette to get the cell membrane to adhere to the tip,” says first author Ilya Kolb. “This is a very delicate process, which has led to the whole patch-clamp electrophysiology process being widely considered to be an artisanal skill in biological research.”

The difficulty of the operation means that even an experienced practitioner can typically characterize on the order of just ten cells per day. Partial automation of the technique has increased the efficiency of the process somewhat, but full automation has until now been stymied by two factors.

The first challenge is that tissues and cell cultures tend to deform as they are invaded by the pipette, turning the cell into a moving target. A human operator can deal with this complication, but automatic systems have so far lacked the necessary adaptability.

The team overcame this hurdle by equipping a commercial platform with a computer-vision system which operates on images acquired by a motorized microscope. When the user selects the cells to be studied, PatcherBot records their coordinates and appearance. The measurement sequence is decided automatically, and if a target cell has not drifted too far by the time the robot arrives at its last-known position, the image-recognition algorithm lets PatcherBot spot the cell in its new location.

The second challenge is that each procedure leaves a residue of cellular material, yet the surface of the pipette must be completely clean for the patch-clamp process to work properly. “The conventional wisdom is that pipettes can only be used once, so the main reason the technique has been manual for almost 40 years is that a trained user still had to replace used pipettes for fresh ones between every attempt,” says Kolb.

To solve this problem, the researchers fitted the robot with a chamber in which pipettes can be cleaned and rinsed automatically between procedures. With a detergent chosen specifically for its ability to remove proteins from glass, Kolb and colleagues found that a single pipette could be used dozens of times without affecting its performance.

Automating the whole process by emulating the actions of a laboratory worker in this way already makes the patch-clamp technique far less labour-intensive, but the robotic approach has even greater potential. Whereas a human operator must focus on just one cell at a time, PatcherBot can perform many patch-clamp procedures simultaneously, and is limited only by the logistical challenge of coordinating the movements of multiple manipulators and sharing microscope time between them. For short-period measurements, the researchers think four to six is the optimum number of manipulators that could be deployed at the same time.

Although none of the technologies underpinning PatcherBot are breakthroughs on their own, few laboratories have the necessary expertise to combine them into a single system. Kolb and colleagues therefore hope to disseminate the technique by licensing it to a commercial company. “This robot could enable labs and companies to screen drugs faster, and with an approximately tenfold reduction in manpower,” says Kolb. “Ultimately we hope that the broad adoption of PatcherBot in the pharmaceutical industry will speed up drug development and reduce side effects.”