Thomas Knöpfel (right) with some of his coauthors. Credit: RIKEN Brain Science Institute Laboratory for Neuronal Circuit Dynamics

German scientist Thomas Knöpfel moved from Europe to Japan because of a jellyfish. As a young scientist in the early 1990s, he had labored to understand neural circuits using chemical dyes that stain cell membranes and change color in response to voltage. These dyes, which stain cells indiscriminately, showed that neurons were firing but could not be used to track specific groups of neurons. What was missing was a way to label and visualize voltage changes in specific populations of living neurons.

After other researchers successfully cloned the jellyfish gene gfp into mammalian cells, Knöpfel knew his next step should be to combine this fluorescent protein with a protein voltage sensor, but he lacked one essential ingredient: funding. No one had made a genetically encoded probe that worked in neurons before, and grants were not available for such risky projects. “I had no preliminary data,” he says, “only an idea.”

The opportunity to realize his dream of creating a genetically encoded voltage sensor came when the RIKEN Institute recruited him to establish a laboratory at its new Brain Science Institute in Wako City, near Tokyo. The position promised scientific freedom, and Knöpfel would be able to pursue several highly expensive techniques, including probe development and in vivo imaging. Knöpfel believes he got the job more because of his reputation than because of his proposal for creating voltage sensors. “My suspicion is that they were looking for foreigners to join this enterprise; maybe they thought that after one or two years I'd give up and do some standard things.”

After 12 years in Japan working on voltage-sensor proteins, Knöpfel knows little Japanese but has learned to communicate with his Japanese colleagues. “Much better than speaking Japanese is understanding Japanese body language,” he explains. Few Japanese researchers would directly disagree with a senior colleague, for example; instead, the way a person bends his head signals enthusiasm or reluctance.

Knöpfel arrived at RIKEN in 1998, but the voltage-sensor proteins did not follow quickly. He began by engineering a gene encoding a protein that combined a fluorescent unit with ones whose shapes changed in response to voltage, but these proteins did not go to the membrane where they belonged. He coauthored a paper, along with other notable neuroscientists, acknowledging that none of the first-generation voltage sensors produced by anyone were good enough. But Knöpfel kept at it and generated a construct with fewer subunits that was properly incorporated into the membrane, though the signals it produced were still weak.

By that time though, many researchers had turned to a genetic probe that relies on a different strategy to image neural activity. Calcium sensors emit signals in response to calcium ions, an indirect effect of voltage changes. Whereas Knöpfel has done considerable work using these probes, he believes direct voltage measurements can be used to ask more subtle questions. For example, calcium sensors can detect action potentials, but voltage sensors can detect 'synaptic potentials', smaller inputs from multiple cells. “While only action potential really matters as the output of a neuron, we need to look at synaptic potentials to understand why there is an action potential or not,” Knöpfel says.

Desire to study these phenomena spurred Knöpfel to continue to improve voltage sensors. Fluorescence measurements in animals face multiple sources of noise, and Knöpfel and colleagues needed to make sure that changes in fluorescence really reflected changes in voltage. For this, the researchers coupled the voltage-sensing domain to two fluorescent domains so that a change in voltage would increase the signal from one but decrease the signal from the other domain. Researchers could home in on true signal by looking for change that occurred in opposite directions. The researchers now show that these improved voltage-sensor proteins reveal expected activity in brains of anesthetized mice subjected to standard stimuli, in this case, a tug on their whiskers. But the technique can be applied well beyond whiskers, says Knöpfel, and can even be used to record signals from transgenic neurons directly through the skull.

Now, he says, he can begin the work he came to RIKEN to do: unraveling neural communications involving small signals across multiple cells. “Neurons are speaking to each other,” he says. “The action potential shows they say something, but you want to understand why.” In particular, Knöpfel hopes to find the origin of brain waves, very fast oscillations that occur across many neurons. Previous tools, he says, saw large signals from small numbers of cells. But smaller effects from larger assemblies of cells could be just as important. “How could you understand society by just asking your neighbor?” he asks.