A GRADIENT OF PORE SIZES:
Courtesy of Michael J. Sailor
imparts a rainbow of colors to a porous silicon chip, one of a variety of new biosensor technologies in development around the world. The different colors correspond to different sized pores, ranging from a few nanometers to hundreds of nanometers in diameter. These pores help the device discriminate and detect proteins and other molecules based on their size.
One morning in March 1995, the deadly nerve gas sarin wafted through the Tokyo subway system. Six years later, spores of lethal
It is the evident ease with which these three attacks were carried out that has security personnel around...
The basic elements of a sensor include a probe or capture reagent that hooks whatever the device is supposed to detect, and a platform, generally an electronic or optical system that notes when something gets detected. The concept of biosensors covers a lot of technological ground. Traditionally, these devices have been limited to using biological molecules such as antibodies, DNA, or proteins as probes. Some new sensor technologies, however, can detect their quarry in the absence of any biological capture molecules.
Even biological probes are changing. For a long time, bio-sensor developers focused on antibodies, which "are great if you know exactly what you are looking for," says Paton. Bioterrorists could keep changing the bioagents, though, diminishing the value of antibody use. To get more versatility in biosensor receptors, several scientists now use nucleic acids, not for capture through hybridization, but rather via the lock-and-key approach of antibodies. "Nucleic acids, like proteins, can fold into very complex, three-dimensional shapes," explains Ronald R. Breaker, associate professor of molecular, cellular, and developmental biology at Yale University.
Breaker demonstrated the potential of this approach with an array of RNA receptors in which each array element can sense a different target.1 "We have to build each RNA receptor from scratch," he says. "We have only made dozens of them, but theoretically we could literally build a sensor array that sees thousands of target molecules simultaneously."
Breaker serves on the scientific advisory board of Archemix in Cambridge, Mass., with Andrew Ellington, professor of biochemistry at the University of Texas at Austin, who also uses nucleic acid receptors. Ellington points out that these receptors can be generated using high-throughput methods and selected in vivo for target specificity. In addition, Ellington says, "Since these chemicals are synthesized, they can be modified to fit different platforms."
EYEBALL THE BIOAGENT
Other capture reagents work, too. Richard A. Durst, professor of chemistry at Cornell University, uses gangliosides, multifunctional lipids found in the cell membrane. In a recent experiment, Durst made liposomes with GT1b gangliosides, the natural receptor for botulinum toxin, and filled the spheres with red dye.2 Then, he put a line of antibotulinum toxin antibodies across a nitrocellulose strip, dipped it in a test solution, and probed the strip with the labeled vesicles. The toxin tethers the liposomes to the strip, producing a red line that can be seen by eye in just 15 to 20 minutes (see image, this page). "People are heading for improving the speed at which sensors will work, because that is the name of the game," says Durst.
©2003 Springer-Verlag Heidelberg
This test strip-based assay has three components: ganglioside-studded, dye-impregnated liposomes; anti-toxin antibodies immobilized on a nitrocellulose test strip; and the material to be tested. The liposomes and material are mixed and allowed to migrate along the strip by capillary action. If a particular toxin – in this case, botulinum toxin – is present, it will tether the dye-containg spheres to the antibodies, producing a visible band in the "analytical zone." (Reprinted from Anal Bioanal Chem, 378:68–75, 2004)
Baeumner uses microfluidics to put liposome-based detection systems like Durst's in devices the size of a cell phone. The detection part of Baeumner's miniaturized liposome system already works for many of the potential bioagents on the CDC's list,3 and she hopes to optimize the rest in the next couple years. "Using microfluidics speeds up reactions," Baeumner says, "but it is also limited to smaller sample volumes. We are trying to increase that in some cases."
Other scientists want things even smaller. Michael J. Sailor, professor of chemistry and biochemistry at the University of California, San Diego, bores pores in silicon with an electrochemical process and then soaks the silicon in a concoction of chemicals. Toxic gases can condense in the pores and combine with the absorbed chemicals to change the refractive index of the silicon, which changes its color. Sometimes the color change can be seen with the naked eye, Sailor says, but with an optical interferometer, he can measure much subtler changes.
Sailor has used this technology to build a sensor package about the size of two cigarette packs that plugs into a laptop. He also has found that a tiny chip of the silicon measuring just 100 micrometers (the size of a grain of sand) also works.4 Sailor can place these tiny sensors in water or on a surface and use their visible color change to indicate the presence of different agents.
THE POWER OF LIGHT
Advanced optics provide more than opportunities to examine samples by eye. Daniel Lim, professor of microbiology at the University of South Florida, attaches antibodies to the outside of an optical fiber and uses them to capture specific agents.5 He then applies a fluorophore-labeled second antibody, creating a sandwich structure: antibody-agent-antibody. A laser directed down the fiber reveals the agent by exciting the fluorophore.
George Anderson of the Naval Research Laboratory and Lim have used this approach to detect several bioagents with variable success. They can detect cholera toxin below lethal levels, but not the more potent botulinum toxin, of which just 0.1 μg is lethal for humans.
David R. Walt, Robinson Professor of Chemistry at Tufts University in Medford, Mass., also is adopting fiber optics. He builds biosensors from bundles of 10,000 to 60,000 optical fibers, each of which bears at its terminus a well capable of capturing a tiny addressable microsphere.6 Each sphere contains a unique color and capture reagent, in this case single-stranded DNA molecules specific for a particular agent. To use this biosensor, Walt loads the fibers with beads, maps the position and identity of each pixel in the array, and then incubates the collection with fluorescently labeled target DNA. A laser directed down the fiber bundle makes the captured targets fluoresce, and the receptor map identifies the agents (see image, p. 38).
Bill Whitten, a senior research scientist at Oak Ridge National Laboratory in Tennessee, employs an entirely different approach: He examines microorganisms one by one with ion-trap mass spectrometry. "We are trying to show that this technique can identify bacteria, and maybe even harmful bacteria," he says.
Even though Whitten's approach examines one particle at a time, it may need to look at lots of data to determine the type of bacteria it is sensing. "If you look at several hundred," Whitten says, "you can make statistical inferences about what kind of bug it is." In other words, the system needs to average the results to identify a specific type of bacteria. Fortunately then, it is incredibly fast, churning out spectra in about 100 msec. "It's like a popcorn popper popping as the particles come in," Whitten says. "The spectra come up on the screen too fast to look at by eye."
Significant problems remain. The approach can successfully distinguish bacteria from other particles, such as those that comprise smoke, but it cannot yet discriminate one species of bacteria from another. Even if the technique could be optimized to identify individual species, however, the system could be confounded by the presence of bacteria purposefully grown on an unusual substance that would produce aberrant data. In other words, a bioterrorist could make (and seemingly easily) bacteria that Whitten's high-speed technique might not recognize. Consequently, Whitten says, "We're still hoping that we can get more sophisticated and come up with better results, but that's a ways off."
TOWARD A UNIVERSAL SENSOR
This biosensor uses a fiber-optic waveguide as its substrate. An immobilized antibody captures its cognate antigen – a toxin or bacterium, for instance – from a test solution, after which a second, fluorophore-tagged antibody to the antigen is added. Presence of the antigen creates a sandwich: antibody-antigen-antibody, which can be detected by laser excitation down the fiberoptic cable. (Reprinted from Proc IEEE, 91:902–7, 2003)
Each of these approaches has demonstrated some degree of success in a controlled laboratory environment. The question, though, is how will they fare in the real world. "It is reasonably simple at the lab bench to demonstrate these sensing concepts with biological threats or mimics in carefully controlled samples, but that all changes in the field," says David Cullen, reader in biophysics and biosensors at Cranfield University in the United Kingdom.
In the real world, real engineering problems arise. A device can see only what exceeds its detection limit, and that has design implications: Is it better to monitor continuously or to collect sample and periodically test it? Additionally, says Cullen, "If the system stays in the field for weeks or months, then maintaining the biological components in the system proves difficult." The literature supports this point, Cullen adds, because most of the published studies report bench-top studies for proof of concept, and very few describe robust analytical results from the field.
The real world also demands a variety of sensors, although the perfect biosensor, a so-called universal sensor, could detect any potentially dangerous agent. The proceedings from the Consensus Conference on the Role of Biosensors in the Detection of Agents of Bioterrorism, held by the US Army Medical Research and Material Command's Telemedicine and Advanced Technology Research Center in September 2003, stated: "A 'universal' biosensor applicable to all bioagents would be the ultimate goal."
Ibis Therapeutics, a division of Isis Pharmaceuticals of Carlsbad, Calif., is working toward that goal. Ibis' Triangulation Identification for Genetic Evaluation of Risks (TIGER) project is developing a biosensor that looks for unique DNA sequences in samples, weighs them with mass spectrometry, and compares the resulting weights against a database of organisms. "This sensor will eventually detect any of the many hundreds of species of bacteria or hundreds of species of viruses, fungi, or protozoa," says company president David Ecker. So far, he says a prototype can identify virtually any bacterium, but still has a long way to go with other agents.
Some scientists, though, do not reflect Ecker's optimism. "To provide accurate results, the mass spectroscopic-based systems available today tend to be too large and require too much power to be field portable," says Sailor. "Getting these high-resolution instruments small enough to be handheld and battery powered is still science fiction," he adds. Jeff Newman, a consultant at Cranfield University's Institute of BioScience and Technology in the United Kingdom, concurs. "They've been watching too much Star Trek."
Some biosensors are already on the job, though, such as the anthrax-detection units installed at several USPS centers. Four companies collaborated to put together the system: Northrop Grumman, Los Angeles; Cepheid, Sunnyvale, Calif.; Sceptor Industries, Kansas City, Mo.; and Smiths Detection, Englewood, Md. The system collects air samples as envelopes go through an automated mail sorter. The sample goes into a Cepheid cartridge, using reagents made in collaboration with Applied Biosystems of Foster City, Calif. Inside the cartridge, a combination of PCR and TaqMan-probe chemistry searches a sample for DNA from
Still, no company provides a biosensor that could detect every bioterror attack. No smoke detector-like device exists that recognizes a wide variety of bioagents and responds in seconds. As a result, many problems remain in the lab, and even more challenges lie between the lab and the market.