Project Objectives

• Model the whole pathway, not just the nerve cell. The team’s current chip models pain-sensing neurons on their own. This project extends it to four connected compartments: skin cells, where pain signals begin; clusters of sensory neurons, which carry the signal; Schwann cells, which insulate nerve fibres; and spinal cord neurons, which receive the signal and pass it on. All of the cells are human.
• Find the best design. The team will test up to ten versions of the chip layout against the same criteria: ease of handling, how long each cell type survives, and the quality of the electrical recordings. The best design will be published and made available to order from the manufacturer.
• Automate the experiments. A fluid-handling system will deliver nutrients and test compounds without manual pipetting, which can itself disturb the delicate signals being measured. The finished system will run five separate experiments in parallel on a single chip, unattended, for up to two weeks.
• Identify electrical signatures of nerve damage. Some common medicines, including certain chemotherapy drugs, can damage nerves in patients. By exposing the neurons to these drugs, the team will look for characteristic changes in the nerve signals, and then test whether candidate protective compounds prevent those changes.
• Prove the system works in other hands. Thorsten Hornemann’s group at the University of Zurich will use it with nerve cells from patients with HSAN1, an inherited disorder that affects the sensory nerves. Aziz Moqrich’s group at the Institute of Developmental Biology of Marseille will use it to test 20 candidate pain drugs at several doses each.
• Make it open. The chip designs, fluid-handling blueprints, software, protocols, and data will all be published openly, so that other labs can build and run the system themselves.

3Rs Impact

• A human-based alternative to early-stage pain drug testing in rodents. This is especially relevant to studies that surgically injure a nerve to produce chronic pain and then measure the animal’s sensitivity through its behaviour.
• Direct measurement instead of inference. The chip records electrical activity from hundreds of individual nerve fibres at once, rather than relying on how an animal reacts to a stimulus.
• Human biology from the start. The Marseille team’s candidate drugs are based on TAFA4, a protein made naturally by a type of sensory neuron. Most of what is known about it comes from rodent studies, but comparisons of gene activity suggest that rodents and humans differ, which is why the team wants human data.
• A concrete animal saving. Based on the Marseille group’s earlier work, the team estimates that a comparable study in mice would need around 280 animals to test the 20 compounds at a single dose, and around 1,400 to test five doses.
• Lower barriers for other labs. The chips are made by a commercial manufacturer, the designs will be freely available, and the team will record and publish the full cost of each data point during the drug screen. This addresses one of the most common reasons labs give for not switching methods.
• Training. The lab already hosts visiting scientists for up to a year to learn the technology. It also plans to adapt its engineering lab course so that students work with pain-sensing neurons and drug testing.

Background

Chronic pain affects between 20 and 30% of people, and treating it remains a serious problem, not least because of the risks of long-term opioid use. Much of what is known about pain comes from rodent studies. These depend on behaviour: an animal can’t describe what it feels, so researchers infer pain from reactions such as how quickly it withdraws a paw. Many studies also require injuring a nerve to produce chronic pain in the first place. Rodent research has taught us a great deal about the molecular basis of pain, but those findings have rarely led to new drugs for patients.

Nerve cells grown from human stem cells offer a different starting point. The International Association for the Study of Pain has recognised them as a valuable research tool, partly because cells derived from a patient carry that person’s genetic background. Neurons made from patients with inherited pain disorders are more excitable than those from healthy donors. In one published case, testing a drug on a patient’s own stem-cell-derived neurons pointed to a treatment that went on to relieve that patient’s pain. But pain-sensing neurons on their own are only one part of the system. In the body, pain involves the skin, the insulated nerve fibres, and the spinal cord working together.

Janos Vörös’s lab at ETH Zurich builds small, precisely arranged networks of human nerve cells on chips carrying thousands of tiny electrodes. Microscopic channels guide each nerve fibre past the electrodes, so its signals can be recorded individually. The group has shown that its pain-sensing neurons reproduce a signature behaviour of real pain fibres: their signals slow down when stimulated repeatedly. It has also grown the insulating myelin layer around the fibres, and has kept skin cells, sensory neurons, and spinal cord neurons alive together on a single chip for more than two months.

This project brings those elements together into one model of the pain pathway, automates it for drug testing, and hands it to partner labs to show that it works outside ETH. The team is realistic about the limits. It expects that the model won’t reproduce every aspect of pain in the body, and plans to focus on the features that it reproduces reliably. One practical challenge is already known: the skin cells grow quickly and can overfill their compartment, which currently limits how long each chip can be used.

The lab has already made the shift in its own work. In 2023 it used 26 rats and 263 rat foetuses for its neuroscience research. Since February 2024, after moving to methods based on human stem cells, it has used no animals at all.