August 2008

Picking Up the Scent

A next-generation artificial nose developed by Tufts neuroscientists uses DNA to detect odors, and possible applications run the gamut from medical to commercial to defense

By Jacqueline Mitchell

The gadget surely doesn’t look like a scientific breakthrough. The three-pound, handheld, odor-detecting device, called the ScenTraK, more closely resembles a supermarket price scanner from 20 years ago. But inside, a familiar molecule plays a revolutionary role.

John Kauer (left) and Joel White with a prototype of the sensor they built using DNA. “People think of DNA in one way,” says White. “We lose sight that it has other properties.” Photo: Webb Chappell

Researchers at Tufts have pioneered the use of DNA molecules to detect millions—maybe billions—of odors, and this novel use of deoxyribonucleic acid could have almost as many applications, from ferreting out contaminants in our food supply to uncovering chemical weapons.

The newest version of ScenTraK represents a quantum leap in electronic smelling technology, which lags far behind instruments that can see and hear for us. The innovation comes decades after Tufts researchers John Kauer, professor of neuroscience, and Joel White, research assistant professor of neuroscience, began their quest to build a better nose.

The team’s body of work—much of it funded by the U.S. Defense Advanced Research Projects Agency (DARPA), the Office of Naval Research and Homeland Security—has military, industrial, commercial, even medical applications, and could help scientists better understand what our brains are really doing when we smell fresh-cut grass or smoke.

Smelling Like a Tiger Salamander

Kauer has been studying olfaction for 35 years, 25 of them at Tufts. Much of his research has focused on the tiger salamander, an amphibian with large, robust brain cells that is trainable enough to signal when it detects an odor. By the late 1980s, Kauer’s team was able to observe the salamander’s brain in real-time using voltage-sensitive dyes that indicated which neurons fired in response to specific smells.

“We puffed the odorant on the salamander’s nose, and we could see the brain doing its thing,” he says.

Olfaction, it turns out, is more complicated than taste. Kauer likens the two senses to notes played on a piano. In taste, one key plays one note, with specific receptors responding to five or six tastes. Olfaction is more like a chord; odor molecules stimulate an array of smell receptors over time, and the brain, by mechanisms not yet fully understood, makes sense of the patterns of signals.

Kauer and White have been working to build an electronic model of biological olfaction, in part to shed light on what happens in an animal’s brain when it smells something. “It’s fair to say that we don’t really understand higher levels of brain activity,” notes Kauer.

In early versions of the electronic nose, airborne odors passed over a square of silk screen treated with a mixture of a reactive polymer (a large molecule comprising a chain of smaller ones) and a fluorescent dye. If some property of the odor—its molecular shape, polarity or charge—interacted with the polymer, the fluorescent dye would glow in response. The trouble was, for each odor they wanted to detect, the researchers had to find, mainly through trial and error, the specific polymer that could serve as a sensor.

Over 15 years, the Tufts team and researchers elsewhere discovered 20 to 30 polymers capable of detecting a handful of odors such as mold and DNT, a chemical cousin to the TNT found in landmines. NASA scientists working to engineer ammonia detectors for the space shuttle, for example, tested 80 to 100 polymers before finding the one that suited their needs.

A New Use for DNA

“It was a very plodding way to work,” Kauer notes. Emulating biological olfaction, in which about a thousand different receptors allow animals to distinguish among the world’s many smells, would require a far more extensive library of sensors. The researchers would need a more malleable molecule to work with. Three years ago, Joel White thought DNA would fit the bill.

“Some people would call it a hare-brained idea,” White admits.

“People think of DNA in one way. We lose sight that it has other properties.” Like the other smell-detecting molecules Kauer and White used, DNA is a polymer, a big molecule made up of a chain of small molecules chemically bonded together.

DNA takes the shape of a double helix—imagine a ladder, twisted. Each rung of the ladder contains two complementary molecules, called base pairs. James Watson and Francis Crick’s description of DNA’s double helix in the journal Nature in 1953 explained how genetic information is stored, copied and combined, changing the course of biomedical research forever. Today, it’s a very well-studied molecule.

“All the labs around us in this building right now have freezers full of it,” says White.

“We can call the physiology department and ask them to synthesize DNA strands 20 bases long,” much like a ladder with 20 rungs, adds Kauer. “And they can make gallons of identical molecules if we need them to.” DNA is easy to come by; it’s stable and dissolves in water, a bonus because most other potential polymer sniffers dissolve in less-friendly solvents.

But it is DNA’s almost endless capacity to shuffle and re-shuffle itself into new combinations—the same characteristic that allows for the diversity of life on earth—that makes it useful for Kauer and White’s purposes. Take a strand of DNA 20 base pairs long. Because four possible molecules make up those ladder rungs, the rules of probability dictate there are 420 possible configurations of a strand of DNA 20 base pairs long. That’s about a trillion unique sequences, each a potential smell sensor.

Single-strand Eureka

The big question was, would it work? Would dry DNA react with airborne molecules? The scientists turned to the literature, but “there hadn’t been a single report,” says Kauer.

It was up to the Tufts team to find the answer. Building on their earlier designs, the scientists combined strands of DNA dissolved in water with a fluorescent dye and dried the mixture to a square of silk screen.

This time, however, the team developed an array that allowed them to test dozens of different DNA sequences at once. As in the earlier versions of the artificial nose, the polymer-coated silk screen sits inside the ScenTraK, which draws air and ambient odors over the sensor array.

The first attempts with double-stranded DNA yielded disappointing results. But when the researchers ran experiments with single strands of DNA, from which one side of the ladder had been removed, they hit the jackpot.

As they reported in the online journal PLoS Biology in January 2008, of the 29 sequences of DNA the research team tested, 10 groups responded to a given set of odors. In these tests, the device detected DNT, the landmine explosive, at concentrations as low as 280 parts per billion.

“That’s like dropping teaspoons of dye into an Olympic-size swimming pool and noticing the color of the water change,” says White. By repeating these tests with different sets of odors, the researchers can quickly screen thousands of DNA sequences and build libraries of sensor molecules.

White presented the DNA theory at the annual meeting of Association for Chemoreception Science in 2004. For Alan Gelperin, a neuroscientist at the Monell Chemical Senses Center in Philadelphia and adjunct professor of neuroscience at the University of Pennsylvania School of Medicine, the idea was a revelation.

“Sensor technology has been the most important limiting factor in electronic olfaction,” says Gelperin, who once designed an artificial nose capable of discerning different kinds of oranges at a supermarket checkout. Today, with colleague Charlie Johnson, a physicist at UPenn, Gelperin is working on a version of the artificial nose that uses semiconducting carbon nanotubules coated with DNA—rather than the ScenTraK’s fluorescent dyes—to reveal the presence of odors.

Gelperin and Johnson’s machine provides electronic data, not optical data like Kauer and White’s, but “our work all started from observations and measurement from John and Joel’s lab,” he says.

Sniffing Chemical Cocktails

Gelperin adds that because DNA molecules are naturally coated with a thin layer of water molecules, the DNA-based sniffing devices have an advantage in real-world settings, where humidity is often an unpredictable factor.

But the real beauty of the DNA sensor, which puts it head and shoulders above earlier iterations of the artificial nose, is that you don’t even have to know the chemical composition of the odor you want to detect. “If you can see which sequences respond to an odor, you can make a sensor that can discriminate that odor from another,” says Kauer.

Earlier versions required researchers to match the odor to the polymer that would signal its presence. Now, says Kauer, all you need is a sample of the odor you wish to detect. “Pass it over the array, pick the sequences that react and tailor the device to detect the odor in question,” he explains.

That makes the machine perfect for detecting particularly nasty chemical cocktails in the event of industrial or mining accidents, chemical spills or chemical warfare. In 2002, Kauer and White launched their start-up company, CogniScent, to develop the e-nose for these kinds of commercial and military applications. Housed in incubator space at the Cummings School of Veterinary Medicine in Grafton, Mass., CogniScent is mainly supported by federal grants and has eight full-time employees working to bring the artificial nose to market.

“Every fire engine, police car and EMT should have one of these,” Kauer argues. U.S. Customs and Homeland Security officials could use the device to inspect containers and incoming vehicles.

The device could be tailored for use in the food and beverage industry, ensuring high quality products and detecting possible contaminants. Coffee producers could identify different bean types; meat processors could sniff for bacteria. Physicians might also use the device to discern odors that are the hallmarks of certain diseases, such as the ketones associated with diabetes and the amines linked to liver disease. The possibilities are almost endless.

“We’ve tested only a very small quantity of DNA, and yet we have already found sequences that respond,” says White.

“Compared to maybe thousands of plastic polymers, there are hundreds of billions of DNA sequences to test,” adds Kauer. “The area to explore is huge.”

This story first appeared in the Spring 2008 Tufts Medicine magazine. Jacqueline Mitchell may be reached at

Article Tools

emailE-mail printPrint