should be proportional to the amount of analyte being recognized. Turner said it took another 12 years to compress the laboratory-sized glucose biosensor to a pocket-sized meter for selfmonitoring of blood glucose by diabetic patients. The first product was a pen-style device launched in 1987 by MediSense, a company started in Boston by three entrepreneurial Harvard University graduates. The technology behind this device was A. Maureen Rouhi and head of the Institute of BioScience & based on ferrocene or ferrocene derivaC&EN Washington Technology at Cranfield University, Bed- tives mediating electron transport between enzymes and electrodes. It was defordshire, England. veloped by Turner's group at Cranfield in According to Turner, the pocket-sized From the ACS meeting glucose biosensor—and the biosensor collaboration with researchers at the Uniarticipants at a biosensors sympo- field as a whole—had its beginnings in versity of Oxford. The phenomenal success of the Medisium at the American Chemical So- 1962 when the physiologist Leland C. ciety's national meeting last month Clark Jr. described an experiment in Sense device is clear from the exponential in San Francisco have reason to be baffled. which he used a dialysis membrane to fix growth of sales, reaching $175 million in From the research perspective, the the enzyme glucose oxidase onto an ox- 1996, said Turner. It's also been reflected field of biosensors is vibrant. Researchers ygen electrode. When glucose is present, in the entry into the market of competing continue to probe basic questions in mo- oxygen is consumed as the enzyme oxi- devices from other companies, including lecular recognition and to explore new dizes the sugar, and the decrease in oxy- Boeliringer-Mannheim and Bayer. Perhaps areas of application. gen concentration correlates with glu- the most emphatic affirmation of the MediSense device has come from Abbott LabOn the other hand, the high expecta- cose concentration. tions of commercial success from biosensThirteen years later, the idea of an en- oratories. In March 1996, Abbott paid ing technology have not yet been realized. zyme electrode was successfully translat- $876 million to acquire MediSense. With And with the gap between the research ed into a laboratory glucose analyzer by that, the leader in biosensors for blood glulab and the marketplace still far from being Yellow Springs Instruments, based in cose self-testing systems for people with firmly bridged, the mood at the sympo- Yellow Springs, Ohio. The analyzer em- diabetes became a wholly owned subsidiary of the world's leadsium ranged from exuberance about the ing diagnostics company. exciting research to doubt about the fuTurner estimates the ture of the technology in the marketplace. market for home blood The symposium on biosensing and bioglucose testing was worth sensors attracted about 250 papers coverabout $1.7 billion in 1996. ing the whole spectrum of activities in Much of it went to prodbiosensor research, development, and ucts based on the convencommercialization. These were presenttional technology of reed in sessions organized by eight ACS diflectance photometry— visions, the ACS Biotechnology Secretaribasically paper strips that at, and the ACS Committee on Science. change color. Only about At the applied end of the research $425 million went to biospectrum, sessions organized by the Divisensors such as the Medision of Small Chemical Businesses highSense device. lighted commercial applications. At the But because of the risfundamental end of the spectrum, sessions organized by the Division of Organ- Pocket-sized blood glucose monitor is the singular success ing incidence of diabetes, Turner says, the gluic Chemistry focused on sensing recep- story in the biosensor field. cose home diagnostics tors that have little biological connection, being largely products of chemical bodies the essence of a biosensor: an an- market is growing at a rapid pace of about ingenuity. In between these extremes, alytical device incorporating a biological 13% per year worldwide. And Turner various sessions over five days offered a or biologically derived sensing element sees good prospects for glucose biosensors increasing their market share. He diverse sampling of work in emerging ar- that's associated with a transducer. eas of application as well as a peek into The sensing element could be an en- predicts they will displace the paper the future of analysis—massive analytical zyme, as is found in the glucose analyzer, strip technology within four years. That's power through sensing arrays married to or it could be an antibody, a nucleic acid, the good news. What's not-so-good news is that the tomicroprocessors. or even whole microorganisms. Its funcThe singular success story in the bio- tion is to recognize—to act as a receptor tal biosensor market worldwide is modest—only $508 million in 1996—accordsensor field is the pocket-sized glucose for—the analyte of interest. analyzer. The story was revisited by AnThe transducer converts the interac- ing to Turner. And at $425 million, pockthony P. F. Turner, one of the world's tion between the analyte and the recep- et-sized glucose biosensors dominated that leading figures in sensors and diagnostics tor to some form of signal. The signal market last year. Coming in at a distant
Biosensors Send Mixed Signals
Despite remarkable breadth of biosensor research, commercial success has been limited
P
MAY 12, 1997 C&EN 41
science/technology second was a laboratory instrument aimed at the research market, with 1996 sales of only about $19 million. These sales figures and uneven performance fall short of what Turner said were "fantastic predictions in the 1980s that biosensors would quickly take over the analytical world and generate a multi-billion-dollar business." Some observers of the biosensor industry believe the technology will succeed only in niche markets. For example, Howard H. Weetall, a research biologist at the National Institute of Standards & Technology, Gaithersburg, Md., suggested that biosensors, except in some very specific areas, will not be competitive with the kinds of high-volume analyzers now used. "I don't see an immunosensor, for example, competing in a clinical laboratory with an automated analyzer that is precise, accurate, fast, and inexpensive to operate," Weetall explained. Weetall's conservative estimate of the impact of biosensors in the analytical marketplace is corroborated by market research by Frost & Sullivan, a consulting group based in Mountain View, Calif., that has been tracking the biosensor field. Their most recent survey of markets worldwide shows essentially flat (3 to 7%) or even declining growth rates for sales of biosensors in most areas of application through 2003. Only in the medical research area did the survey forecast continued growth in sales at a robust 10 to 14% through 2003. "When I looked at the Frost & Sullivan study, I was pretty shocked and disturbed," said Noe Salazar, vice president for sensor systems at Systems & Processes Engineering Corp. (SPEC), Austin, Texas. "They couldn't be right, [because] there is a lot of development work in many companies. SPEC alone currently has several very promising biosensor programs in progress. All this indicates an imminent explosion of biosensors in other areas." Salazar believes biosensors are ready to take ofl" in the military, industrial, and environmental arenas in just a few years. That view is not shared by the market analysts: "Foster & Sullivan basically said, 'Every company we talked to said the same thing. They're all excited. The only problem is we heard this five years ago and nothing has changed. So we don't think it's going to happen,' " Salazar recalled. Although glucose analyzers have dominated the biosensor market, other application areas are feasible, as evidenced by numerous applied research papers at the San Francisco symposium. Biosensor 42 MAY 12, 1997 C&EN
Turner: problems breaking into markets
technology could be applied to food, environment, defense, and industrial processing. According to Turner, in the environmental field alone, there is already a market worth about $5 million per year for one biosensor to measure biological oxygen demand (BOD, a measure of the extent of pollution in aqueous wastes). But Turner said there are problems in breaking into these markets. In the environmental field, the market is driven by regulation: A lot of companies measure the environment only because they have to. Furthermore, biosensors can't comply with what legislation prescribes. Turner explained: "When analytical parameters are enforced by law, the method by which they have to be measured is laid down in the legislation. You're told how you must make the test, as well as what limits you have to comply with. Legislation is conservative; methods take a long time to get updated." For example, Turner explained that the BOD test currently enshrined in most legislation around the world has to do with the typical length of time water used to be kept in a holding tank before being discharged in England—five days. So the BOD test takes five days. The BOD sensor, on the other hand, produces a similar result in 20 minutes. "The results are not exactly the same," stressed Turner. "It's a scientific impossibility to deliver in 20 minutes what you get in five days." So the BOD biosensors can't be used to comply with legislation. But companies are using them to check their compliance on a regular basis. If the BOD legislation changes, then there
'te':?/'-
would be a boom in these biosensors, Turner predicted. Closely linked to environmental applications are the needs of military and defense establishments to rapidly monitor chemical and biological warfare agents, for which biosensors are a real option. Biosensors also are being explored for detection of explosives and unexploded ordnance (C&EN, April 21, page 9). Biosensors such as the pocket-sized glucose analyzers have many advantages. They're small, compact, and easy to use. They work in complex matrices and give detailed analytical information very simply. These advantages prompt people to wonder: What else can be measured with a pocket-sized device? Bacteria in food come to mind, and indeed, researchers have looked at opportunities for biosensors in food applications. However, finding a place for biosensors in the food industry is likely to be difficult. This area is "a lot trickier to penetrate," noted Turner. Although it's a huge market, it's very diverse, highly competitive, and very margin conscious. "Nobody wants to pay even one-tenth of a cent more than they have to pay for anything," he said. The key would be to look for generic targets for biosensors, such as microorganisms, sugars, indicators of food deterioration, residues, and contaminants. There are other areas, however, that biosensors could quickly penetrate. For example, NIST's Weetall was upbeat about what he calls "genosensors." These are sensor arrays containing oligomers of different sequences that can be used to detect DNA or RNA sequences or mutations. At the moment, there is no other way to do this kind of analysis that is not highly labor intensive, he explained. Any biosensors developed for this purpose won't have to face competition from anything available that's inexpensive, accurate, and precise. The fiiture, however, won't be in designing more single-analyte probes that can be stuck into a matrix to yield analytical information. Turner believes the future lies in immense analytical power married to microprocessors. The whole system will be encapsulated in a form that interfaces with the environment, the body, the food, the processing line, whatever. "Microprocessors are really smart, but they have to be fed with information. If they are given senses—the power to acquire information on their own—I can't imagine what the impact of that is going to be," said Turner. Turner's biosensor program in Cran-
Chemists trap benzyne in a cage Perhaps the most imaginative chemistry at a biosensors symposium during the American Chemical Society's national meeting last month in San Francisco was that described by chemistry professor Donald J. Cram of the University of California, Los Angeles. For his pioneering work in host-guest chemistry, from which emanated principles of molecular recognition and enzyme-substrate interactions, Cram shared the Nobel Prize in Chemistry in 1987. Speaking at a session devoted to chemosensors, he described the latest work from his group—stabilization of o-benzyne in solution by incarceration in a container molecule and subsequent spectroscopic and chemical studies of the trapped reWarmuth Cram active species. o-Benzyne is highly reactive and short-lived. By generating it inside a zyne was benzocyclobutenedione inside molecules," he explained. So it was for container molecule, Cram and postdoc- a container molecule. Warmuth, soon to o-benzyne: Warmuth not only could be an assistant professor take its nuclear magnetic resonance at Kansas State Universi- spectrum in solution but also could exty, Manhattan, prepared amine its interaction with its container. Despite being trapped, o-benzyne is the host-guest complex from a well-known con- not completely subdued. Cram and Wartainer molecule—a hemi- muth observed that it undergoes a Dielscarcerand—by heating Alder reaction with the surrounding the empty container with host shell, forming covalent bonds with benzocyclobutenedione. the inner phase of the container. When the product, called The kinetic behavior of the reaction is a hemicarceplex, was il- similar to that of intramolecular reacluminated with long- tions. However, the reactants are sepawavelength UV light, the rate species, not covalently bonded. So trapped bicyclic dione the reaction is really a bimolecular reacwas transformed to ben- tion following first-order kinetics. For zocyclopropenone. Fur- this new reaction type—which occurs ther illumination with only if one reactant is completely enshort-wavelength UV closed and retained by the inner phase light ejected CO, leaving of the second reactant—Warmuth and Light hitting the benzocyclopropenone guest in the hemicarceplex (left) triggers separation of CO (middle), o-benzyne in the cage. Cram have proposed the term "innerwhich exits the host and leaves o-benzyne alone in the As Cram has shown molecular" reaction. cavity (right). For a talk on chemosensors, Cram before, container molecules can tame reactive warned his audience that his presentatoral fellow Ralf Warmuth were able to species—which Cram calls "tender" tion was going to be "a little bit rich on stabilize it enough so that for the first molecules, because they tend to do the 'chemo' part and a little poor on the time the reactive species could be char- many things. Inside containers, they are 'sensor' part." But who knows what an"protected from doing bad things, such alytical devices or reagents can come acterized in solution. The precursor to incarcerated o-ben- as dimerizing or doing things to other out of his findings.
field is targeting to place 1 million sensors on a square centimeter chip within the next decade. "That's a crazy man-on-themoon project," he said, "because who wants to measure a million things?" But he noted that people asked a similar question of computers: "We had no idea what we could do with the computing power, but now even cars and cameras rely on it." If the future will be about multiple sensor arrays to measure "many many
more things than we know what to do with at the moment," that means going beyond the "crude idea of taking a bit of an animal or bacterium and sticking it on a sensor," Turner explained. "We've got to be more sophisticated than that. "We'll have to understand what these molecules are doing, we'll have to learn to mimic these molecules, we'll need to use synthetic chemistry to copy structures and build new structures, we'll
need to generate millions of new receptors, and we'll use microprocessing power to unravel signals from devices." One group that's already taking definite steps toward Turner's vision of enormous analytical power married to computers is that of chemistry professor David R. Walt at Tufts University, Medford, Mass. Walt's team is exploring a radically different approach to sensing, using high-density fiber-optic arrays. Tiny wells MAY 12, 1997 C&EN 43
seience/teehnol
.#
10 μΓΠ
15
20
Tufts researchers (from left) Michael, Taylor, Walt, and Schultz have produced a sensor array of microscopic beads in microwells as shown above in an atomic force microscopy image.
are fabricated by etching the cores of im aging optical fibers. These wells can be separately probed with light because each well is connected to its own optical channel. The sizes and shapes of the wells can be tailored to hold single mi croscopic beads of various sizes. "I was really excited to present this idea," said Walt. "Because it is a dramatic departure from the conventional way people make biosensors." In traditional sensor arrays, the chem istry at each position is exactly known and is identical for every sensor made in a particular manufacturing process. In the approach being developed by Walt's team, it is not necessary to locate the chemistry at each specific position in the array even though every sensor array produced will have a unique distribution of chemistries. Each microwell becomes a sensor through the microscopic bead it holds. Each bead contains chemistry that emits an optical signal when a specific analyte is detected, as well as encoding dyes that tell the analyst what chemistry is incor porated in a particular bead. Hence, all the information can be obtained through optical signaling and imaging software. Walt and his team—graduate students Karri L. Michael and Laura C. Taylor and postdoctoral researcher Sandra L. Schultz— have produced arrays of up to 6,000 mi crowells in a unitary fiber-optic bundle that measures about a third of a millimeter in diameter. "You're talking of potentially 6,000 assays in a tiny tip," noted Turner. "And you don't have to worry where you're putting the assay because you can find it afterward." The sensor array works by optical "in terrogation" at two wavelengths. Light at the analytical wavelength excites a fluo rescent molecule associated with the sensing chemistry. Light at a different 44 MAY 12, 1997 C&EN
wavelength causes the encoding dyes to fluoresce. When the array is exposed to a sam ple, every analyte for which detecting chemistry is present in the beads will re act. When the fiber is illuminated at the analytical wavelength, every bead where chemistry has taken place will change its light emission and will be revealed by the imaging software. Because the beads represent different chemistries, interrogation moves to the encoding wavelengths to determine what chemistry is taking place in each well. The encoding wavelengths excite only the encoding dyes. When the dyes send their signals back, the detector dis plays a two-dimensional picture of the ar ray. "Essentially we get a bead-by-bead re port of whether a reaction has occurred in a well and what analytical chemistry has taken place in that well," said Walt. Walt explained that the real break through presented by this approach is moving the burden from manufacturing to signal processing. "You no longer have to make everything the same," he said. "You know the chemistries in the array. You just don't know where each is so you have to find them after the sensor is made. It takes just a few seconds to figure out what's where." What's needed now is to transfer the chemistries that already exist to detect and measure a host of analytes onto mi croscopic beads. "Imagine having thou sands of diagnostic tests in a compact device for everything on the [Environ mental Protection Agency's] priority pol lutant list, or every known clinical diag nostic test," said Walt. "The challenge is
to convert all these chemistries to a bead format." The goal of a million sensors on a chip assumes the availability of as many receptors, not all of which will come from natural sources. Sessions organized by the Division of Organic Chemistry fo cused on chemosensors, which use nonbiological receptors for recognition. With so many analytical needs waiting to be met, it hardly matters whether the sensing unit is biotically or abiotically de rived. But there is a determined push to develop abiotic receptors for practical reasons. Whether based on a catalytic reaction (enzyme binding followed by catalysis) or just a binding reaction (antibody-antigen binding), biosensors are unstable for appli cations in settings that are destructive to proteins, noted Anthony W. Czarnik, se nior director for chemistry at Irori Quan tum Microchemistry, La Jolla, Calif. Using the example of glucose, Czar nik explained: "The most important ap plication for glucose measurement would be in the blood of diabetics. Un der these conditions, proteins can be ex pected both to denature over time and to be degraded by serum proteins designed to metabolize proteins. After all, living or ganisms have evolved to view biological macromolecules as foodstuffs." Metal-ion sensing is one area in which Czarnik believes chemosensors will have a major impact. Metals are good targets for abiotic receptors because highaffinity antibodies cannot always be easi ly raised against them. And because met al ions have very important structural and regulatory roles in cellular processes,
1
in protein design and architecture. Nature has evolved a remarkable ability to selectively recognize metal HoNions through protein architecture, she explained. If the metalrecognizing peptide framework can be attached covalently to an environmentally sensitive fluorophore it is of interest to measure them in living whose photophysical properties are systems in real time, under conditions changed by the metal-binding event, one where enzyme and antibodies easily can craft a signaling unit that will either switch off or switch on when binding could be destroyed. For example, Czarnik noted that calci- occurs. um is an important signal for cellular One metal-ion sensor Imperiali deevents. Because of fluorescent chemo- scribed is based on the well-known zinc sensors—such as those developed by finger domains. Tucked into these doRoger Y. Tsien, professor of pharmacolo- mains are hydrophobic zinc-binding sites gy and of chemistry and biochemistry at that usually are lined with two imidazole the University of California, San Diego— ligands (from histidine) and two thiolate intracellular concentrations of calcium ligands (from cysteine). A cluster of three now can be monitored routinely in real other hydrophobic residues folds into time. The chemosensors survive in the the domain during metal binding. cell for a long period because they are In the absence of zinc, these strucabiotic and are not decomposed by cellu- tures are very disordered. When zinc is lar enzymes, Czarnik explained. present, they are highly ordered. FurtherIn San Francisco, Czarnik described more, the domains bind zinc with very some of the approaches he and his high affinity—in the pico- and nanomolar former research team at Ohio State Uni- ranges. They also have extreme selectiviversity, Columbus, have been using to ty, binding preferentially to zinc over the design artificial receptors for sensing closely related cobalt by about five ormetal ions in aqueous environments. ders of magnitude. One approach, based on rigid immobiliTo convert these domains to a zinc-sigzation of polyamine ligands onto a fluo- naling unit, Imperiali and graduate student rophore framework, has yielded a fluo- Grant Walkup replaced one member of rescent chemosensor for two transition- the hydrophobic cluster with a nonstandmetal ions. ard amino acid that incorporates a special Czarnik and graduate student Juyoung Yoon prepared the receptor from 1,8bis(bromomethyl)anthracene and tris(3aminopropyl)amine. It selectively senses only two metal ions—Hg(H) and Cu(H)— in water from among 20 metal ions tested. Its fluorescence signal diminishes when it chelates either metal. It may be argued that chemosensors such as those described by Czarnik are outside the sphere of biosensors. But other research described in the sessions on chemosensors indicates that the line separating what's biological and what's nonbiological easily can become blurred. An example is the work described by chemistry professor Barbara Imperiali of California Institute of Technology on the use of peptides in chemosensors for metals. Although a self-proclaimed neophyte in the area of chemosensors, Imperiali entered the field with extensive experience Imperiali: using inspiration from nature
Polyamine and fluorophore make chemosensor for Hg(ll) and Cud I)
fluorophore, the 5-(dimethylamino)-lnaphthalenesulfonyl, or dansyl, group. The group's fluorescence depends on whether it is in a polar or a hydrophobic environment. Stitching the dansyl group into the zinc finger domain in such a way that its environment changes when a metal binds yielded a polypeptide that fluoresces strongly only in the presence of zinc—a zinc sensor. Polypeptide metal sensors such as those being explored by Imperiali's group are readily accessible through the well-developed technology of solid-phase peptide synthesis. The technology allows versatility. The peptides can be as l o n g or as short—as desired. Because biosensors based on intact proteins are prone to proteolytic degradation and denaturation, Imperiali's team has been focusing on short polypeptides. A major advantage is that the design is not limited to naturally occurring amino acids used in metal binding. "My group has made a business in the past few years of synthesizing different and intriguing amino acids for metal-ion coordination," Imperiali said. Those efforts have yielded bipyridinyl and phenanthrolyl amino acids that can be used to tune the metal-binding properties of peptide motifs. For example, they have fashioned a copper-recognizing finger that incorporates two histidines and a phenanthroline. The finger binds copper extremely tightly even in the presence of large amounts of zinc. Solid-phase biosynthesis also offers the opportunity to bring in combinatorial chemistry to help generate diversity in receptor design. "We can't in all cases perfectly predict the sorts of metal-ionbinding properties that we might want to use," she said, explaining the reason combinatorial tools are important. "In some cases, we're going to have to go blind." To illustrate the power of the approach, Imperiali described results her group obtained recently with a small library of resin-bound peptides incorporating the dansyl fluorophore. When copper was added to the tethered peptides in separate test tubes, irradiation of the resin, even with a "low-tech" iUuminator, revealed dramatically which test tubes contained peptides that bind copper. Clearly for Imperiali, nature provides the starting point. What's nice about the peptide platforms "is that at the end of the day you have a viable product," she said. "We're using inspiration from nature, but we're letting synthesis carry us the rest of the way to allow more versatility than even nature provides."^ MAY 12, 1997 C&EN 45
