Biological Warfare Canaries

Once a year a select group of scientists and engineers gathers at Dugway Proving Ground, a salt flat in the Utah desert 65 km from the nearest traffic light. They are there by invitation of the U.S. Defense Department's Joint Program Office for Biological Defense to compete in an unusual showdown: to field-test their systems for detecting biological warfare agents.

This is no idle contest; this is serious business, critical to national security. As the recent contentious negotiations over the 29-year-old Biological and Toxins Weapons Convention have shown, concern over the possibility of a biological attack is mounting worldwide.

Shortly before the trial gets under way, the competitors make final adjustments to their devices and then retire to trailers nearly a kilometer away. Moments later, an aerosol cloud containing spores of Bacillus globigii is released into the air about 90 meters from the devices. This harmless bacterium simulates a real bioagent such as Bacillus anthracis, the germ that causes anthrax.

As the aerosol cloud passes, the devices collect and prepare air samples, and then feed them through diagnostic tests, looking for the bioagents by using highly specific molecular interactions [see photo]. High-tech versions of the proverbial coalminer's canary, these devices are designed to determine the type and concentration of the agent within minutes, enough time to let soldiers on the battlefield don protective gear. Data are sent to a remote sensing post by wireless modem.

At the trial's conclusion, the biodetectors are rated on how well and how quickly they identified the surrogate agent, as well as on their ruggedness, power consumption, weight, size, reliability, and safety.

A model aircraft plays the role of delivering a biological or chemical weapon [reddye on ground] at Dugway Proving Ground in Utah's Great Salt Lake Desert [above]. After a surrogate bioagent is released, various detectors try to identify it within a matter of minutes.Douglas C. Pizac/AP

The ultimate goal of the Dugway trials, and of more controlled experiments in the laboratory, is to develop portable, fully automatic, remote sensing systems that can detect a variety of known and novel biological agents before troops on the battlefield are exposed. The dream solution, though still many years off, is a wristwatch-sized biodetector capable of rapid detection, rapid diagnostics, and, potentially, rapid treatment.

What are biological weapons?

Biological weapons include bacteria, viruses, and toxins that are spread deliberately in air, food, or water to cause disease or death to humans, animals, or plants. Bacteria and viruses work by entering the body, multiplying, and then overcoming the immune system. Examples include Bacillus anthracis, Yersinia pestis (which causes plague), and Variola major (smallpox). Biotoxins are the poisons given off by living entities, such as botulinum toxin, produced by the bacterium Clostridium botulinum, and ricin, which can be isolated from castor oil seeds. According to Plague Wars, Tom Mangold and Jeff Goldberg's history of biological warfare [see To Probe Further], Western counterproliferation agencies recognize 23 bacteria, 43 viruses, and 14 toxins as potential threats.

A few grams of a dried bioagent such as anthrax could infect thousands of people. Symptoms of infection would follow within a week and deaths in the days thereafter. Certain biological weapons could also cause destabilizing epidemics. Smallpox, for example, can be easily transmitted from one infected individual to another. Anthrax, in contrast, is deadly when inhaled but is noncommunicable.

In the past, only naturally occurring organisms and toxins were considered real threats. Recent advances in genetic engineering, though, have paved the way to designer bioagents. Scientists practicing such "black biology" have already created drug-resistant strains of anthrax, plague, and tularemia (a highly infectious disease that causes skin ulcers and pneumonia). Biowarfare agents could also be bred to be far more virulent and long-lived.

To an aggressor, the appeal of such weapons is that they attack populations, leaving infrastructure intact; they are effective in very small amounts; they can be produced at low cost in a short period of time; and protection and detection are difficult.

In wargames of a fictional attack on Oklahoma City, it was predicted that an infectious agent such as smallpox could spread to three million people throughout the continental United States within 12 weeks of an attack. Thankfully, the world has yet to see such a full-blown assault. But for the reasons above, there is increasing concern that bioweapons will become the preferred weapon of mass destruction.

To be sure, fabricating devices to disperse biological agents is not trivial. Typically, one needs to create an aerosol cloud containing just the right particle size--1­5 µm is the most lethal when inhaled. What's more, some agents are quite fragile and die quickly in sunlight; others, though, are more robust. Efforts in the former Soviet Union and Iraq succeeded in generating vast quantities of plague and anthrax agents, as well as the means to deliver them--by aircraft equipped with spray tanks, cluster bombs, and missiles with multiple warheads.

In short, it is far easier to make a biological weapon than to create an effective system of defense.

Diplomatic solutions

For much of the last century, many countries have tried to develop biological weapons. U.S. intelligence has identified 24 nations suspected of developing or harboring biological weapons today. The United States itself ran an extensive biowarfare R and D program, before it was abruptly halted by President Richard Nixon in 1969.

Worldwide pressure to halt such weapons culminated in the Biological and Toxins Weapons Convention (BWC) of 1972. The treaty bans developing, producing, stockpiling, or acquiring biological weapons or their delivery systems. The BWC has no provision for enforcement, instead leaving each country to police itself.

Next month, the fifth review conference of the BWC will get under way in Geneva. Nearly all 143 countries that have now signed the treaty agree that giving teeth to the convention will mean creating a verification regime that would allow inspection of suspected biological weapons facilities.

To their great consternation, however, the current Bush administration, at the urging of the pharmaceutical industry, has resisted any regime proposed to date, because it would enable foreign governments to inspect, among other things, the research labs of U.S. drug companies.

Current state of the art

While banning bioweapons, the BWC allows development of countermeasures. The United States has led the world in developing technologies to counter bioweapons. Most of the emphasis has been on detecting attacks on the battlefield, either at or near their source (known in the military as point detection) or from a distance (standoff detection). Spotting random bioterrorist acts, though, is considerably harder [see sidebar, "Are We Prepared for a Bioterrorist Attack?"].

Existing detection systems have tended to be fairly large and not terribly accurate, and to require humans to operate them. Typical is the U.S. Army's mobile lab, which is housed in a Humvee and manned by four technicians. An alternative is the helicopter-mounted unit made by Orlando, Fla.­based Schwartz Electro-Optics. From a distance of 50 km, it can distinguish naturally occurring clouds from the cigar-shaped plume emitted from a moving object, the telltale sign of a biowarfare attack. It cannot identify specific agents, however.

Detecting DNA: The palm-sized module [left] at the core of Cepheid's biodetector repeatedly heats and cools the bioagent, creating millions of copies of its DNA. Fluorescent probe DNA then binds with bioagent DNA. Detection occurs when LEDs [above] cause the probes to fluoresce; the glow is captured by photodiode detectors. One module can detect four species; 16 modules fit into the portable detector.CEPHEID

The future: biodetectors

Although these human-operated mobile units remain important, many of the laboratory tests used to identify bioagents have become robust and reproducible enough to be carried out robotically.

What researchers involved in developing ways to detect biological attacks hope to achieve are fast, reliable, and portable devices. Starting with a raw air sample, the device would prepare the sample for testing (for example, cracking open spores of anthrax to extract the DNA within), and then determine what kind of organism or poison is present and in what concentration--all within about 30 minutes.

The detectors now under development fall into three broad classes:

• Biochemical systems, which detect a DNA sequence or protein unique to the bioagent through its interaction with a test molecule.
• Biological tissue-based systems, in which a bioagent or biotoxin affects live mammalian cells, causing them to undergo some measurable response.
• Chemical mass spectrometry systems, which work by breaking down a sample into its component amino acids and then comparing their weights to those of known bioagents and other molecules.

DNA detection-on-a-chip

One class of detectors relies on comparing the DNA taken from microorganisms in a sample with the DNA of known biowarfare agents. To that end, researchers have in recent years been sequencing the DNA of an array of potential agents, using the now-common gene-sequencing methods for studying more ordinary microbes.

When a biodetector collects a sample in the field, the sample may contain only tens or hundreds of bioagent molecules, too few for most detection methods. To speed up identification, multiple copies of the DNA are made, using the so-called polymerase chain reaction (PCR).

PCR starts by splitting the agent's double-stranded DNA in two, by heating it to about 95 degrees C. The temperature is then lowered to the point where the single strands will bind together. This time, though, a new strand of DNA is synthesized across each of the original strands, from nucleotides in the solution.

When this two-step heating cycle is repeated n times, it produces 2n copies of target sequence, at least in theory. So from a single starting molecule, 20 cycles yields over a million copies, 30 cycles yields a billion, and 40 cycles yields a trillion. Typically, 25 to 40 cycles are needed, each lasting about a minute.

While the PCR is busy copying the DNA, the resulting strands are mixed with fluorescent DNA probes. The gene sequence of each probe exactly complements that of a specific bioagent's DNA, so that it will bind with that bioagent, but not with closely related benign microbes' DNA. The probe is equipped with a fluorescent marker; under ultraviolet light, samples in which the probes have bound with the bioagent will glow.

By using different fluorescent markers for different DNA sequences, several PCR reactions can be run in parallel, thus enabling detection of more than one agent at a time. PCR-based systems have the advantage of being extremely sensitive; from a raw sample containing as few as 10 organisms, they can identify the pathogen in less than 30 minutes.

The table lists the features of the various types of detectors now being worked on.

This lab-on-a-chip system contains tiny channels, valves, and chambers through which milliliters of sample are pumped. After concentrating the sample to a microliter volume, it cracks open, or lyses, the cells of the microorganism ultrasonically, extracts the DNA within, and then performs PCR. Small thin-film heaters rapidly heat the DNA, reducing the cycle time to 25 seconds and enabling the machine to be battery operated

According to Kurt Petersen, Cepheid's president, "In DNA detection, everyone's been concentrating on sample amplification and detection, while ignoring sample preparation, which can be just as complicated." That's because bioagents come in many forms. Bacterial spores, for example, are quite hardy and call for fairly harsh handling to extract their DNA. More fragile samples require a lighter touch.

In the field, though, one lacks the luxury of such individualized preparation. Cepheid has therefore incorporated an automated sample preparation scheme into a briefcase-sized detector called GeneXpert. Set to ship to the U.S. Army next month, this new system is said to reduce a procedure that can take six hours in the lab to just 30 minutes.

Measuring DNA with magnets

Meanwhile, scientists at the Naval Research Laboratory (NRL) in Washington, D.C., are developing a DNA detector that relies on magnetic fields, rather than optics and fluorescence. The NRL's device, called the Bead Array Counter, or BARC, comprises an array of magnetic field microsensors [see figure].

According to Lloyd Whitman, head of the lab's surface nanoscience and sensor technology section, each wire-like magnetic sensor, or group of sensors, is coated with single-stranded DNA probes specific for a gene from a bioagent. Once a strand of bioagent DNA in the sample binds with a probe, the resulting double strand binds a single magnetic microbead. (The 2.8-µm beads are off-the-shelf products used for biochemical separations.) When a magnetic bead is present above a sensor, the sensor's resistance decreases by a small but detectable amount. The more beads, the larger the change in resistance.

The sensors can detect single microbeads, Whitman said, and so in theory could detect a single DNA molecule. To date, though, the BARC system has only been tested with samples of at least a few thousand bioagent molecules.

Mimicking the immune system

Mammals' immune systems are amazingly specialized. Among their myriad functions, they churn out proteins called antibodies in response to certain foreign matter, including bioagents. An antibody generated from a particular bioagent will bind to it and nothing else--a phenomenon biochemical detectors can exploit.

Antibody tests have the added advantage of being able to detect both microorganisms and biological toxins, which carry no DNA. At their simplest, such systems resemble a home pregnancy test: antibodies are fixed to a strip of cellulose on a plastic backing. The sample is applied, and the reaction between antibody and bioagent causes two colored lines to appear on the strip, meaning the bioagent has been detected. The appearance of one line, in contrast, means no agent is present. Thomas O'Brien, vice president at Tetracore in Gaithersburg, Md., claims that the company's strips, used in conjunction with an automatic strip reader, are "presently the only [biowarfare] detection system on the commercial market."

A more sophisticated test, being developed at Lawrence Livermore National Laboratory, in Livermore, Calif., uses fluorescent antibodies to bind to bacterial cells. The sample passes through a portable flow cytometer, a common laboratory device that counts cells by measuring their fluorescence or other properties as they move in a liquid. The Livermore cytometer is gated to detect particles that are both fluorescent and of the same size as the bacteria.

Tracking probes with colored beads

Flow cytometry systems can potentially be multiplexed to identify multiple agents in the same sample. To do that, of course, means keeping track of the various probes. Luminex Corp., of Austin, Texas, has one solution: it has created an array of 100 color-coded microspheres, each uniquely colored with two fluorescent dyes. When used in the Lawrence Livermore flow cytometer, each colored microsphere is coated with an antibody to a particular biowarfare agent. The antibodies bind the bioagent to the microsphere, and a second fluorescent antibody then binds to the bioagent, to signal that the microsphere and bioagent are now joined.

With the bioagent thus sandwiched between the antibodies, the microspheres pass in single file through the cytometer's detector, where a succession of lasers illuminates the microspheres' dyes and the fluorescent antibodies. Each color-coded microsphere corresponds to a particular bioagent, so when the cytometer detects a microsphere that is also fluorescing, it knows that bioagent is present. To date, the system has been able to pick out 20 types of agent in a single sample.

The Naval Research Laboratory and Research International in Woodinville, Wash., have jointly developed another antibody-based system. Called the Raptor, it uses short polystyrene optical fibers instead of microspheres. Like the Luminex system, the detection process uses two types of antibody: one to bind the bioagent to the optical fiber and a fluorescent type to identify it. When a diode laser excites each fiber, the fluorescence from the antibodies is captured and collimated by a molded lens and then focused onto a photodiode using a ball lens, chosen for its light-gathering power and short focal length.

Listening to heart cells

A drawback with the DNA and antibody tests, though, is that they require prior knowledge of the bioagent. Alan Rudolph, director of the Tissue-Based Biosensors program at the U.S. Defense Advanced Research Projects Agency, one of the main funders of biodetector R and D, has noted that the recent emergence of the West Nile virus in New York City, which took several weeks to identify after claiming its first victims, "highlights the lag times involved in identifying unknown agents." Work is therefore going into developing detection schemes that require no such prior knowledge.

Many toxins, for example, trigger reactions in living cells. More like a scream of pain than a highly specific response, these reactions can still be measured and differentiated. That is the basis of tissue-based biosensors, which consist of live mammalian cells, such as heart cells, that are cultured in the lab and then seeded into a cartridge containing a microelectrode array.

When a biotoxin is introduced, the tiny ion channels on the cell's surface open or close, creating millivolt changes in the cell membrane that the electrodes detect. Greg Kovacs and his group at Stanford University, in California, have developed and field-tested a handheld tissue-based detector.

Handheld biodetector, developed by Greg Kovacs and his group at Stanford University, measures the millivolt signals given off by live mammalian heart cells [sealed in the green cartridge] in the presence of a biotoxin.Derek Debusschere, Standford University

Eventually, Kovacs said, added computational power may allow the measured responses to be compared to known libraries of responses, the better to identify specific agents.

Biosensors may also be seeded with B cells, a part of the immune system that carry antibodies on their surfaces. When the B cell binds with its pathogen, it triggers specific genes in the cell to turn on. To flag the event, these genes are replaced with a gene from a jellyfish that enables them to glow. When a bioagent binds with the genetically engineered B cell, the cell starts producing a fluorescent protein.

In tests with a detector designed by the Biosensor Technologies Group at the Massachusetts Institute of Technology's Lincoln Laboratory, in Cambridge, the whole process, from when the sample hits the B cells to when the cells start to glow, took less than 15 seconds. Each channel in the detector contains groups of cells specific for a particular bioagent.

Compared with the other classes of detectors, tissue-based biosensors are still in the early stages of development. Just keeping the fragile cells alive on the battlefield, for example, is tricky. Kovacs has field-tested a carrying case that will keep cells alive for several days, but for troops stationed far from the nearest biology lab, that may not be long enough.

Biodetection without bioreagents

Given the difficulties of preserving live tissue and other fragile biological reagents, researchers agree that a detection scheme that does not rely on them at all would provide the ideal solution--hence the interest in adapting mass spectrometry for detecting the proteins from biological agents.

Like PCR, mass spectrometry starts by heating the sample, which contains fragments of bioagent protein, to between 150 degrees and 250 degrees C; this turns the liquid sample into a gas. The gas is bombarded with an electron beam to give a charge to the fragments. The positively charged fragments are then accelerated through an electric field, which sorts them by mass and charge; the mass-to-charge ratios are used to calculate the fragments' molecular weights.

The fragments are then passed through a second mass spectrometer, which reduces them further to their component amino acids. Each protein fragment has a unique ratio of amino acids, against which the sample can be compared. Such a system can differentiate mixtures of closely related bacterial species and can also detect as few as 100 bacteria in a sample.

One challenge has been miniaturizing the refrigerator-sized mass spectrometers without compromising their effectiveness. A remaining bottleneck is in interpreting the data, said Douglas Stahl, director of Biomedical Informatics at City of Hope National Medical Center, in Duarte, Calif., who has worked on systems to automatically analyze mass spectrometry data.

The fingerprint of a specific biomarker can be compared to a library of previously collected spectra. Eventually, Stahl said, "a computer should be able to run the machine [and] interpret the data through a pattern-matching algorithm."

Technology and diplomacy

While many of the technologies discussed above show promise and while dozens of research groups in the United States and elsewhere chip away at refinements, it will take much more time and effort to produce a truly reliable battlefield detector.

In the meantime, next month's review of the Biological and Toxins Weapons Convention presents an excellent opportunity to close loopholes in the historic treaty. In this author's opinion, such arms control efforts are not only desirable but necessary as a deterrent. To discount them, and to place one's faith solely in technological solutions, is simply naive.

Jean Kumagai, Editor