6

Detection and Measurement

of Biological Agents

The previous chapter was devoted to an analysis of what the committee feels is the
most probable course of events in a terrorist attack involving a biological agent--a covert
attack that, after a period of hours to weeks, will result in victims widely distributed in
time and location. Because the biological agents being discussed in this report do not
immediately produce effects, the first indication of an attack with a biological agent may
be the recognition of an unusual distribution or number of cases of disease, long after the
initial aerosol or solution has been dispersed or degraded. An important part of this
detective work is laboratory analysis of clinical samples, most often blood from a sick
patient. The previous chapter alluded to the possibility of new developments in such
diagnostic testing that might significantly decrease the time needed to arrive at a definitive
diagnosis. The present chapter examines those developments in more detail.

The chapter also examines the application and utility of these developments in the
detection of biological agents in the environment. There will be no fire and rescue teams
responding to a 911 call in an incident involving covert release of a biological agent, and
thus little use for the sort of rapid detection devices that are so important in responding to
chemical releases. Public health surveillance systems and the rapid analysis of
information from those systems may in time provide an indication of when and where the
biological agent was released, but unless there is a continuing source of agent, testing the
release site at that point will probably be useful for forensic purposes only (testing may
also be helpful in guiding clean-up after an attack with spore-forming agents like anthrax
that can survive in the environment for years). This is far different from the battlefield
scenario of military units facing an enemy with an arsenal of identified biological
weapons. Monitoring the environment for those agents and providing these at-risk troops
with the means to rapidly identify contaminated air, water, food, and equipment would
literally be vitally important. To the extent that similar high-risk situations can be
identified in the civilian environment (the President's State-of-the-Union Address? The
Superbowl? A soggy package labeled "anthrax"?), there may be a civilian need for
monitoring and detector technology as well. For this reason, although the committee does
not believe these situations will be frequent enough to merit a major investment for
civilian use, the chapter concludes by summarizing current R&D efforts on environmental
detection by military and other laboratories.

DETECTION OF BIOLOGICAL AGENTS IN

CLINICAL SAMPLES (PATIENT DIAGNOSTICS)

The classical approach to microbial detection involves the use of differential metabolic
assays (monitored colormetrically) to determine species type in the case of most bacteria,
or the use of cell culture and electron microscopy to diagnose viruses and some bacteria
that are intracellular parasites. Samples taken from the environment, such as soil and
water, and most clinical samples must be cultured in order to obtain sufficient numbers of
various cell types for reliable identification. The time required for microbial outgrowth is
typically 4—48 hrs (or two weeks for certain cases, such as Mycobacterium tuberculosis).
Furthermore, bacterial culture suffers from an inherent drawback: cells that are viable may
not be culturable, because they possess unanticipated nutritional requirements as a result
of genetic mutation. The following few pages lay out some general approaches being
taken to eliminate these drawbacks of the traditional methods and provide some examples
of efforts to apply them to detection of potential biological weapons. Biodetection is a
very large and active field which merits a study all by itself, and for that reason the rest of
the chapter is deliberately confined to technologies and research that has focused on the
agents of central concern to this report. The interested reader is referred to any or all of
the following general reviews: Turner et al. (1987), Janata (1989), Wolfbeis (1991),
Taylor and Schultz (1995), Van Emon et al. (1996), Rogers et al. (1995), Kress-Rogers
(1997). Boyle and Laughlin (1995) provide a history of the U.S. military biodetection
program, and Boiarski et al. (1995) described a large number of biodetection technologies
being explored by the U.S. military at that time.

In summarizing the current review, it is convenient to consider detection of biological

agents as a two-stage process involving: (1) a probe, and (2) a transducer. Probe
technology deals with how the assay or detection device recognizes the particular target
microbe. Transducer technology deals with how the assay or detection device
communicates the activity of the probe to the observer. Together, probe and transduction
systems determine specificity, sensitivity, and time required to make an identification.

Probe Technologies

Probe technologies include those based on: nucleic acids, antibody/antigen binding,
and ligand/receptor interactions.

Nucleic acid-based probes capitalize on the extreme selectivity of DNA and RNA
recognition. Nucleic acid probes, engineered single strands of RNA or DNA, bind
specifically to strands of complementary nucleic acids from pathogens. These probes and
their binding can be detected directly or by tagging with an easily detected molecule that
provides a signal. The design of the probe can be highly specific if there is a good fit to a
pathogen-unique region of the target nucleic acid, or it can provide more generic
identification if there is a fit with a region of nucleic acids conserved among several
related pathogens. The sensitivity of these hybridization assays for bacteria is between
1,000 and 10,000 colony-forming units; improved sensitivity is an important area of
research. Since the reaction is in real time, the time-consuming part of the method relates
to sample preparation and the time required to detect the signal.
 The main advantages of nucleic acid-based methods are universality (all living
organisms have DNA and/or RNA), specificity (every type of organism has some unique
sections of DNA or RNA), sensitivity (with amplification, very small amounts can be
detected), adaptability (base sequences common to several microbes, or even a whole
class of microbes, can be used as probes), and multiplex capabilities for a host of different
microbes (a sample can be probed for many different sequences simultaneously).
Disadvantages of this technology include difficulty in isolation and "clean-up" of DNA
samples, degradation of the nucleic acid probes, and interference from related sequences
or products. These are important obstacles to be overcome, even after specific and
accessible target sequences are identified and probes constructed.

Some commercial products are already available for applications unrelated to

biological weapons. Parke-Davis, for example, markets an RNA-based device to study
HIV RNA: protein interactions. A dozen or more biotechnology companies are pursuing
production and a variety of applications of "DNA chips," microarrays of 100 to 100,000
or more DNA or oligonucleotide probes immobilized on glass or nylon substrates
(Marshall and Hodgson, 1998; Ramsey, 1998). Santa Clara-based Affymetrix, for
example, has developed a dime-sized GeneChipTM using arrays of 100,000 or more
fluorescence-tagged hybridization probes and scanning confocal optical readout to search
for mutations of genes known to be involved in specific human diseases. The readout
instrumentation is expensive and the chips themselves have a shelf life of only a few
months, but the speed and thoroughness of the search may have appeal for pharmaceutical
and biotechnology companies. Roche Molecular Systems recently announced its intention
to purchase GeneChip arrays for use in planned diagnostic kits for application to HIV
drug resistance and cancer staging, and a collaboration among Affymetrix, Lawrence
Livermore National Laboratory, and the U.S. Army Medical Research Institute for
Infectious Diseases (USAMRIID) is attempting to adapt the technique to detect biological
weapons as characteristic sequences are identified. A large array incorporating many more
common pathogens as well might encourage everyday use in large medical labs and
eliminate the bottleneck to rapid diagnosis identified in the previous chapter--the need for
a suspicious clinician to order an assay for a very rare disease.

A similar but smaller microarray of gel-immobilized, fluorescence-labeled nucleic

acids is being developed by Argonne National Laboratory (Yershov et al., 1996). One
application seeks to develop a "bacillus microchip" that will detect B. anthracis, indicate
whether it is alive or dead (DNA matches, but no RNA matches), and distinguish it from
other related bacteria, such as B. thuringiensis, B. subtilis, and B. cereus (Mirzabekov,
1998).

A second application of the Argonne/3M array takes advantage of that latitude by

employing RNA probes from the highly conserved 16S ribosome to provide a tentative
taxonomic assignment to unknown bacterial pathogens, including novel or bioengineered
organisms (Risatti et al., 1994; Stahl, 1998). This strategy will not work for all
bioengineered organisms--identification by taxonomic markers must be supplemented by
identification based on markers of pathology, however, if we are to successfully cope with
harmless microbes provided with genes from pathogenic organisms.

Antibody-based probes (immunosensors) offer another highly specific probe

technology, since antibodies recognize very specific sites or cellular components
(epitopes). Antibodies specific for any microbe can be made if the microbe can be
obtained in pure culture. These must be screened for binding characteristics, that is,
binding affinity, on- and off-rates, and epitope recognized. The production of monoclonal
antibodies requires significantly more time and effort in the development of hybridoma
cell lines with appropriate characteristics. It is, therefore, desirable to provide for breaking
the antibody-antigen bond after a positive test and reusing the antibody in additional tests.
The binding of the target (antigen) to the antibody can be monitored directly with a
transduction method, such as luminescence or electrochemical signal, or can be monitored
in a sandwich assay in which a second antibody labeled with a fluorescent dye binds to
another epitope on the captured cell or to the probe antibody. Indirect methods monitor
the bound epitope by its competition with a standard epitope labeled with a fluorescent
dye. While this indirect format is more sensitive, the antibody must bind very strongly to
the antigen target.

Fluorescence-based fiber optic immunosensors have demonstrated the detection of 104

microbial cells/ml, and immunoelectrochemical sensors have demonstrated 103 cells/ml.
Problems include nonspecific binding, degradation of the antibodies over time,
reproducibility of the antibodies, and whether the target can be produced in pure culture to
provide a monoclonal antibody. There is also a problem with cross-reactivity, that is,
closely related organisms frequently cannot be distinguished by immunochemical
techniques. In addition, some viruses possess hypervariable coat proteins, and a
monoclonal antibody raised against a particular coat protein of a virus may be totally
useless for detection of the same virus after it has been propagated for several generations.
Nevertheless, some of the most sensitive sensors are based on antibody probes, and a
recent variation called immunoPCR that tags the antibody with a short strand of DNA
takes advantage of PCR amplification of the antigen-antibody complex to increase
sensitivity still further (Joerger et al., 1995). E. I. DuPont Co. and USAMRIID are
attempting to apply this technology to simultaneous detection of multiple threat agents.

Ligand-based probes were developed on the principle that every cell has cell-surface
proteins that bind other specific molecules. Ligands may be small or large, specific to a
particular microbial serotype or common to related groups, and bind with varying degrees
of affinity.

Until the recent development of combinatorial chemistry methods, ligand-based probes

directed at specific receptors had been dyes that are structural analogs for ligands of
microbial receptors and used in classical microbiological screening tests. More recently,
scientists at Utah State University (Powers and Ellis, 1998) have capitalized on the fact
that pathogenic bacteria, and only pathogenic bacteria (at least the >40 bacterial pathogens
they have tested to date), bind hemin to produce a bacterial pathogen detector that, while
not identifying the pathogen by genus and species, will detect as few as 100 pathogens in
a sample containing ten million or more nonpathogenic bacteria. Researchers at the
University of Alabama Birmingham (Turnbough and Kearney, 1998) have screened a
library of random 7-amino acid peptides to identify a peptide ligand that binds very
strongly and specifically to the spore coat of the nonpathogenic bacterium Bacillus
subtilis. A similar strategy is to be employed to find a tight-binding ligand for spores of B.
anthracis and other biological agents. Other ligands include microbial adhesins and
oligosaccharides. The Utah State University researchers are now using a variety of
combinatorial libraries to find ways to "capture" the toxins produced by B. anthracis, C.
botulinum, S. aureus, and numerous other pathogenic microbes (Powers and Ellis, 1998).
Two potential advantages of this approach are that several toxins may operate by similar
mechanisms and therefore may be detectable with the same ligand and that a toxin-based
probe will be useful even if inventive weaponeers find a way to deliver a known toxin
with a bioengineered organism or a common and ordinarily harmless microbe. Difficulties
encountered in developing ligand/receptor probes are interference and competition with
natural ligands, as well as the fact that receptor sites are under gene regulation that may
alter the expression state in various environmental conditions.

Transducer Technologies

Transducer technologies include: electrochemical, piezoelectric, color-imetric, and

optical systems. The transducer system must acquire signals that are unique to the probe
system and generate low noise signals that can be further processed without degradation
to provide a human observer with an indication of probe system activity.

Electrochemical transducers utilize enzymes to generate an electrochemical signal,

either amperometric or potentiometric (amperometric sensors are more sensitive).
Commercial examples include sensors for glucose, lactose, and a host of cell products. A
Navy-funded R&D effort at Northwestern University is the only example of this approach
in our inventory. The principal investigator hopes to immobilize redox-active
oligonucleotides on a film in such a way that only sequence-specific hybridization can
carry current through the film. Stability is affected by usage and nature of the probe.
Response and recovery times are primarily dependent on the rates of diffusion from target
to probe reaction sites and from product to electrode. Measurements of 103 microbes/ml
have been demonstrated in 1—3 minutes.

Piezoelectric transducers rely on the use of certain crystals that produce an electric
charge when subjected to pressure. Subjecting those crystals to an electric current causes
them to vibrate at a frequency that is dependent upon their dimensions, including their
mass. Coating the surface of the crystal with, for example, antibody or nucleic acid probes
will alter that frequency, and more importantly, antigen binding or nucleic acid
hybridization will cause still more frequency change. There are some problems with
reproducing surface coatings, and the sensitivity is typically 105—106 cells. Specificity is
derived from the probe material. Piezoelectricity is the basis for several chemical agent
detectors using surface acoustic wave (SAW) technology, and it has been a popular
approach to biodetection in the recent past (Guilbault and Schmid, 1991; Guilbault, Hock,
and Schmid, 1992). Our inventory of active research shows only two such entries,
however: a NASA-funded contract at Southern University to develop a liquid-phase
crystal immunosensor, and a Naval Research Laboratory effort to develop an antibody-
based force amplified biological sensor (FABS). Both are still in the proof-of-principle
stage, the former using E. coli for prototype development, and the latter MS2 virus and B.
globigii.

Light absorption, or colorimetry, has also been used for transduction. A binding event
causes a color change that can be observed by the naked eye and/or quantified by
spectroscopic measurements. For example, colloidial gold bound to agent-specific
antibodies produces a red spot when "collected" by antigens in the sample. The "litmus
test" being developed by Charych and colleagues at the Lawrence Berkeley National
Laboratory is another colorimetric assay. Ligands that bind to specific viruses and toxins
are incorporated into a polymerized bilayer assembly that changes color when the agent
binds (Charych et al., 1996). This quick and simple test has a sensitivity of 108 virus
particles and 20 ppm for toxins. Although the sensitivity of colorimetric methods in
general is significantly less than that achieved with fluorescence, such methods are useful
where the agent is likely to be present in high concentrations.

Optical transduction is employed in the majority of the biodetectors listed in Appendix

B. Although a variety of methods based on light scattering and absorbance have been
explored in other settings, nearly all the optical examples in our inventory involve
fluorescence and other luminescence spectroscopies. Fluorescence approaches involve
excitation of the molecules of a material with light, usually in the ultraviolet (UV) portion
of the spectrum. The excited component spontaneously reverts to its unexcited state, a
process accompanied by emission of light at different wavelengths. These emission
wavelengths are dependent upon both the exciting wavelength and the molecules being
irradiated, so it is possible to use the resulting emission spectrum to identify the irradiated
material. Many biological materials, for example tryptophan, are naturally fluorescent.
Due to a number of factors, including the presence of common substances like tryptophan,
the luminescence characteristics of many biological and environmental substances
overlap--often making identification difficult, if not impossible. However, a variety of
methods have been developed to separate individual contributions and the background. Of
particular importance are wavelength and phase modulation, as well as time-correlation
and line-shape fitting methods. A related indirect approach involves introducing a special
fluorophore (a fluorescing chemical with a distinctive emission spectrum) into the sample
or the probe molecule prior to irradiation. Ultimately, background and scattering limit the
sensitivity and overlapping substances limit specificity. Regardless, optical methods offer
the highest sensitivity and selectivity and have been the only methods used for research
requiring single-molecule detection.

Two variants of fluorescence being utilized in DoD research on bioagent detection are
up-converting phosphor technology (UPT) and the fiber optic evanescent wave guide
(FOWG). The former, whose development is being funded through DARPA
(Wollenberger et al., 1997; Wright et al., 1997; Cooper, 1998), uses a number of rare earth
compounds that, in crystal form, have the unique property of emitting a photon of visible
light in response to absorbing 2 or 3 photons of lower-energy infrared light of the proper
wavelength. Coating the crystals with antibody provides for a highly identifiable signal,
since no naturally occurring substances upconvert. Nine spectrally unique phosphors have
been synthesized to date, making it possible to simultaneously probe with as many as 9
antibodies. More phosphors are under development, although it seems likely that the
multiplexing limit will probably be closer to 9 than to 100. The Analyate 2000 fiber-optic
evanescent waveguide biodetector developed by the Naval Research Laboratory (Cao et
al., 1995; Anderson et al., 1996) also uses antibody probes, some of which are bound to a
glass optical fiber immersed in a capillary tube containing an aqueous solution of the
sample. Other antibodies, tagged with a fluorescent dye, are added to the sample, where
they bind to the target antigen. The antigen-labeled antibody complex then binds to the
immobilized antibody. Light from a near infrared diode laser travels through the fiber,
which contains it almost completely. The very small amount of light escaping, the
evanescent wave, excites the fluorescent tag, whose emission is sent back up the fiber and
detected via photodiode.

Hybrid Technologies

There are some detection devices in which there is no clear division of probe and
transducer. Methods based on physical properties and separation are good examples: mass
spectrometry and gas or liquid chromatography. Mass spectrometry (MS) is a major
analytical technique in which materials to be analyzed are converted into gaseous ions or
otherwise characteristic fragments. The fragments are then separated on the basis of their
mass-to-charge ratio. A display of this separation constitutes the mass spectrogram. MS is
an extremely sensitive, selective, and rapid technique. Quantities of chemicals as small as
10—18 moles can be detected within milliseconds in highly purified samples, and MS has
demonstrated detection of 106 cells. In a field environment, or whenever samples are
heterogeneous, the constituents must be separated before they can be reliably identified, a
task accomplished in a variety of ways, including gas chromatography (GC), high
performance liquid chromatography (HPLC), or the use of two mass analyzers (one to
perform the separation and a second to produce the mass spectrum of the resulting
analytes).

Another separation-based detector system specifically for viruses is being developed

by the Army's Edgewood Research, Development and Engineering Center (ERDEC).
Based primarily on sedimentation rate with ultrafiltration, proven technologies, the device
uses an ultracentrifuge and a series of passes through an ultrafilter to separate viruses
from the fine solids onto which they are typically adsorbed and from other nonviral
background materials. The final stage of the detection process involves electrospray
aerosolation of the filtrate, differential mobility analysis, and a condensation nucleus
counter to quantify the viruses present. ERDEC recently licensed a commercial partner
(EnViron) to continue development and field testing of the device, which they claim will
detect and identify all viruses within an hour, with sensitivity as low as 1000 virus
particles even in air or liquid with very high levels of contaminating dust, bacteria,
protein, pollen, and fungi. The system accepts both air and liquid samples, including
blood, without pretreatment (Wick et al., 1997, 1998).

DETECTION OF BIOLOGICAL AGENTS IN THE ENVIRONMENT

Real-time detection and measurement of biological agents in the environment is

daunting because of the number of potential agents to be distinguished, the complex
nature of the agents themselves, and the myriad of similar microorganisms that are a
constant presence in our environment and the minute quantities of pathogen that can
initiate infection. Few, if any, civilian agencies at any level currently have even a
rudimentary capability in this area. A number of military units, most notably the Army's
Technical Escort Unit, the U.S. Marine Corps Chemical Biological Incident Response
Force, and the Army Chemical Corps, presently have some first-generation technology
available.

For example, the Biological Integrated Detection System (BIDS) continuously samples
ambient air and determines the background distribution of aerosol particles. Aerosol
particles with diameters in the 2 to 10 micron range are concentrated and analyzed for
biological activity, as indicated by the presence of adenosine 5´-triphosphate (ATP). Flow
cytometry then separates and concentrates bacterial cells, and antibody-based tests are
conducted for specific agents. At present, the system includes tests for the bacteria
responsible for anthrax and plague, botulinum toxin A, and staphylococcal enterotoxin B.

Much less expensive point detectors are available as prototype "One Step Hand-Held
Assay" devices. These instruments are currently produced by the Navy Medical Research
Institute (NMRI) at Bethesda, Maryland, (similar devices have recently become
commercially available through Environmental Technologies Corporation) and are based
on antigen capture chromatography. Eight different devices are used to assay liquid
samples for the presence of Y. pestis, F. tularensis, B. anthracis, V. cholerae, SEB, ricin,
botulinum toxins, and Brucella species, respectively. A color change provides a positive
or negative indication within 15 minutes. The sensitivity of these assays varies from an
order of magnitude below a fatal dose (ricin) to more than an order of magnitude above
the infectious dose (anthrax). These devices are strictly screening assays, and the analyses
are subject to error from the introduction of other contaminants. Therefore, positive
results need to be confirmed with standard microbiology assays, conventional
immunoassays, or genome detection via polymerase chain reaction (PCR) technology.
Both NMRI and USAMRIID at Ft. Detrick, Maryland, have deployable field laboratories
that can perform these additional confirmatory assays (and assays for 15 to 20 other
potential agents). However, the confirmatory assays do not yield results as quickly.
Detectors with higher sensitivity than those presently available will be needed to detect
biological aerosols at minimally hazardous concentrations.

Potential Advances

Implicit in the three-stage approach to agent identification by the BIDS is the

realization that in some circumstances one need only know that there are more particles in
air than normal to take some important action, such as put on a respirator. In other
circumstances, one might need more information about the nature of the particles (are
they biological, and if so, are they living?) to take action. In still other circumstances
(forensics or treaty verification), one needs to be able to identify a specific bacterium or
virus.

The perceived need for faster, surer results for timely detection of hazardous
biologicals in the environment has spawned a large and growing number of research
programs. Biological detection is the largest single category in the committee's inventory
of relevant technologies (Appendix B). Space does not allow discussion of each, but most
of the devices are variations on a small number of approaches that were described in the
previous section on patient diagnostics. As in the case of chemical detectors, the
underlying approach largely determines the sensitivity, selectivity, versatility, and
reliability. Application to detection of biological agents in the environment differs from
patient diagnostics primarily in the increased need for portability, ease of use by
nonscientists, speed, and methods for collecting and preparing the sample. The following
pages first describe the main approaches to sampling the environment for biological
agents. We then consider some current research and development on new and better
devices for detecting, identifying, and quantifying biowarfare agents and how they might
meet needs of civilian medical personnel in domestic terrorism scenarios. We conclude
with recommendations for prioritizing R&D in this area.

Sampling

Sampling has to do with how the material that is to be tested is brought to the detector,
whether it comes from air, liquid, solid objects, surfaces, or from human tissue. There are
several issues that make sampling for biological agents challenging. The first issue is that
the sampling is normally targeted at living organisms; therefore, the technology must not
"harm" the sample. Secondly, because most detector devices require a liquid sample,
collection of airborne microbes must be extracted from an aerosol or particulate for and
concentrated in a liquid. Third, the target microbe is generally only one component of a
complex matrix of biological elements and chemical compounds that may affect the
detection process, so the sample must often be purified to some extent. Last, the sample
must be highly concentrated for a rapid analysis. Four general types of sampling devices
designed to accomplish one or more of these objectives are: (1) viable particle-size
impactors, (2) virtual impactors, (3) cyclone samplers, and (4) bubblers/impingers. Each
of these technologies is described below.

Viable Particle-Size Impactors. The viable particle-size impactors usually have

multiple stages. Each stage contains a number of precision-drilled orifices that are
appropriate for the size of the particles to be collected in that stage, and orifice sizes
decrease with each succeeding impactor state. Particles in the air enter the instrument and
are directed towards the collection surface by the jet orifices. Any particle not collected by
that stage follows the stream of air around the edge of the collection surface to the next
stage. The collection plate is typically a petri dish with agar or other suitable growth
medium (Boiarski et al., 1995).

Virtual Impactors. A virtual impactor is similar to a viable particle-size impactor, but

uses a collection probe instead of a flat plate as its impaction surface. Air flows through
the collection probe and the collected particles are transported to other portions of the
collector for additional concentration. By controlling the flow in the impactor, it is
possible to adjust the cutoff size to the particles collected. By passing the collection probe
airflow into successive virtual impactors, the particles can be concentrated to many times
the original air concentration before collection. The final stage can then impact the
particle stream into a liquid, resulting in a highly concentrated liquid sample (Boiarski et
al., 1995).

Cyclone Samplers. A cyclone is an inertial device that is commonly used in industrial

applications for removing particles from large air flows. A particle-laden air stream enters
the cyclone body and forms an outer spiral moving downward towards the bottom of the
cyclone. Larger particles are collected on the outer wall due to centrifugal force. Smaller
particles follow the airstream that forms the inner spiral and leave the cyclone through the
exit tube. Application of a water spray to the outer walls of a cyclone facilitates particle
collection and preservation. (Boiarski et al., 1995).

Bubblers/Impingers. Most bubblers or impingers operate by drawing aerosols through

a current inlet tube and jet. Usually the jet is submerged into the liquid contained in the
sampler. As the air passes through the liquid, the aerosol particles are captured by the
liquid surface at the base of the jet. In order to collect the smallest particles possible, the
jet is typically made with a small critical orifice causing the flow to become sonic. Other
designs have a fitted jet so that tiny air bubbles are formed in the liquid as air leaves the
jet. (Boiarski et al., 1995).

Two very important sampling issues must be addressed, regardless of the technology
employed. First, the environment in which the target microbe exists can significantly
affect the physiology of the microbe and with that the efficacy of the detection procedure.
Bacillus anthracis, the causative agent in anthrax, provides a simple example: in the
environment it exists as a hard, oval, inactive spore highly resistant to sunlight, heat, and
disinfectants, but in tissue, including blood, it germinates into a rod-shaped vegetative
bacillus actively proliferating and producing its characteristic toxins. Detection strategies
appropriate for one form of the organism may be entirely ineffective in the presence of the
other form. Less dramatic but equally important, components of the matrix in which a
microbe exists contribute significantly to the microbe's growth state and gene expression
in a way that is just beginning to be explored for most organisms. Detection strategies
focused on a specific structure or gene product can thus vary wildly, if sampling
conditions are not clearly specified.

The second overarching sampling issue is especially important in attempts to detect

microbes in very low concentrations: the process is a statistical problem, and due
consideration must be given to the variables that affect any statistical conclusion, namely
the size, number, randomness, and independence of the samples.

Stand-off Detection

If there is advanced warning of an "event," then diagnostic capability requirements also

include not only "point" detection (in which the detector directly samples the
contaminated environment), but also real-time "stand-off" detection (detection is
accomplished from a distance) Because most of the agents under consideration in this
document are considered attractive as weapons in part because they can be delivered as
aerosols, DoD is developing "stand-off" monitors aimed at detecting particles of a
biological nature in distant clouds. The simplest of these optical devices merely looks for
unexplainable increases in the thermal emissions from a given direction, but the more
sophisticated uses ultraviolet laser-induced fluorescence to identify the presence of
tryptophan. Current prototypes are a large improvement over earlier stand-off systems, but
they cannot yet consistently identify specific organisms because of the similarity of their
emission spectra. Advanced signal processing techniques may improve identification.

Sensitivity to infectious dose level is probably not important for early warning, since
an aerosol cloud intended to kill or incapacitate even one individual will certainly involve
concentrations far in excess of the infectious dose (later decisions about clean-up and
reoccupation of contaminated areas may need that level of sensitivity, but speed will be
less of an issue, and respiratory protection will allow use of more sensitive point
detectors). Specificity also may not be critical in the use of stand-off detectors. For
example, we may just need to be alerted to the presence of live biologicals. This is also
true for the control of contaminated environments, determination of decon efficacy, and
dynamic threat assessment (real-time assessment of a threat, including remediation).
 Stand-off detection offers safe, real-time determination of microbial contamination.
Significant advances have been made with the use of lasers for the detection of
aerosolized agents by light-scattering characteristics, infrared and Raman spectroscopy,
and fluorescence, but these same methods can also be used to determine total microbial
contamination on objects (Powers and Ellis, 1998) and in situations where effective
sampling is impossible for reasons other than distance. The efficacy of these devices is
somewhat limited by the range at which the determination is desired (typically several
kilometers for military systems). Longer distances, of course, are more difficult, and the
necessity for prior intelligence, subsequent deployment, and then line-of-sight use of the
technology would seem to limit its utility in urban bioterrorism scenarios. Applications of
true stand-off detection would seem to be limited to monitoring predetermined, high-risk
sites or large public gathering places, such as stadiums, for aerosol clouds.

An alternative approach for long distance detection is the small model airplane-like
unmanned aerial vehicles (UAV) being developed by the Naval Research Laboratory and
Research International (Foch, 1998). These vehicles, ranging in size from a few inches to
a foot in size, may eventually carry on-board sensors and down-link data to ground-based
control. There is a weight and size limitation on the sensors that can be carried on-board,
but prototype vehicles have been successfully demonstrated in cities and inside buildings
as well as in outdoor terrain. Furthermore, they are reusable and easily transported. In the
event of biological agents being released in a building, such vehicles could locate "hot
zones" and monitor decon efficacy, reducing human exposure and risk.

Point Detection

Point detection refers to testing a sample that has been taken directly from the
environment suspected of harboring the target agent. Needs in this regard include not only
investigation of suspected sources of contamination but also monitoring the air/water
systems in buildings for general pathogen contamination or contamination by specific
biological agents. A number of the embryonic microbe detectors described above in the
section on patient diagnostics are being examined for utility in environmental detection as
well.

In point detection as in stand-off detection, many situations will demand neither

exceptional sensitivity nor exceptional specificity. Assessment of the total microbial
content may be sufficient to determine contamination and alert personnel to danger. For
example, if there is already a suspicion that a terrorist attack is likely, then a sharp and
unexplainable rise in total microbial count probably should be sufficient to trigger
protective action, regardless of whether the specific pathogen can be identified. Total
microbial count might also be sufficient for the assessment of decontamination efficacy.
In other situations, perhaps a detection situation when information is not available on
"background" microbial levels, knowledge of total pathologic organisms present may be
sufficient to guide short-term actions by rescue and medical personnel, even if the specific
pathogen is not identified. More precise identification would be important for forensic
uses of course, and for optimal treatment of many agents (e.g., broad-spectrum antibiotics
might be prescribed as soon as the agent is identified as bacterial, even if the species is
unknown, but this practice contributes to the development of resistant strains; the few
antiviral drugs available have thus far proven to be virus-specific).

Most of the current R&D on detection of biological weapons employs nucleic acid- or
antibody-based probes combined with optical, most often fluorescence, transduction, or it
involves adapting separation-based technology like mass spectrometry.

Regardless of the transducer technology employed with nucleic acid probes,

"amplification" is generally required to detect the very low number of microbes that
suffice to infect humans. A distinguishing feature of nucleic acids is the possibility of
rapidly multiplying ("amplifying") distinctive nucleotide sequences in samples too small
to be analyzed by other methods. This is accomplished by enzymatic [polymerase chain
reaction (PCR), ligase chain reaction (LCR), Q-beta replicase] or nonenzymatic methods,
such as Chiron Corporation's HIV RNA assay using a covalently branched DNA
structure. All of these methods separate a piece of the normally double stranded DNA into
constituent single strands, each of which, given the necessary amino acids, assembles a
complementary strand, the net result of this "cycle" being a doubling of the number of
target DNA strings.

The sensitivity of detection of nucleic acids can thus be greatly improved by nucleic
acid amplification. The polymerase chain reaction (PCR) takes time, and a major aim of
current research is to shorten the time to approach real-time amplification. Idaho
Technology's LightCycler, one of the fastest presently on the market (Wittwer et al.,
1997), can carry out 30 cycles in 6 minutes by using tiny glass capillary tubes for the
sample and high-velocity hot and cold air. RNA can be converted to cDNA by reverse
transcriptase (RT) and thus amplified by PCR. The time required for conversion to cDNA
is also a subject of active research. Several new amplification methods do not require heat
cycling. These include Transcription-based Amplification System (TAS) (Kwoh et al.,
1989), Self-Sustained Sequence Replication (Guatelli et al., 1996), and Strand
Displacement Amplification (SDA)(Walker et al., 1992).

In general, however, degradation of the nucleic acid probes and interference from
related sequences or products from the microbial environment significantly limit the
current application of this technology beyond well-equipped and experienced laboratories.
A single microbial cell can be detected in the laboratory from highly purified DNA by
these methods, but environmental samples have regularly failed to achieve this, usually
having a detection limit of 105 microbes. PCR detected 100 percent of spiked samples in
one study (Candrian , 1995), but only 15 percent of naturally infected samples.
Considerable effort is being made at Lawrence Livermore National Laboratory (LLNL) to
solve these problems and combine nucleic acid-based assays with antibody-based tests in
an automated field-deployable system (Mariella, 1998; Belgrader, 1998). Miniaturized
PCR units with significantly reduced cycling times have also been developed by a
partnership of USAMRIID, LLNL and the California biotech company Cepheid, Inc.
(Ibrahim et al., 1998; Belgrader et al., 1998; Northrup et al., 1998). The long-term goal of
this work is a hand-held instrument featuring disposable cartridges containing all
necessary reagents, reaction chambers, waste chambers, and microfluidics to extract,
concentrate, amplify, and analyze nucleic acids. Concurrent efforts at sequencing the
genes of possible biological warfare agents and identifying organism-unique probes are
under way at Army (USAMRIID) and Navy (NMRI) laboratories (Farchaus et al., 1998;
Higgens et al., 1998), LANL (Keim et al., 1997), LLNL (Andersen et al. 1996), the
University of Texas-Houston (Hoffmaster and Koehler, 1997), and Duke University
(Harrell et al., 1995), so piggybacking onto a commercial market that Cepheid estimates
at over $1 billion seems feasible.

The one-step hand-held tickets described above that are produced at NMRI and more
recently by Environmental Technologies Corporation are an example of immunoassay
technology combined with chromatographic transduction. The sensitivity of these simple
devices is much lower than that achieved in clinical laboratories, but they are inexpensive
and easy to use. For those reasons, they are probably the most logical choice for Hazmat
teams and other emergency responders seeking to test the contents of a suspicious
package for the presence of the dozen of so agents on the military threat list. The Analyte
2000 is another well developed (but not yet commercially available) immunosensor. The
Naval Research Laboratory developed this device, which combines antibody probes with
a fiber-optic waveguide transduction system. Other work is focusing on miniaturizing and
automating the testing process, incorporating the requisite antibodies with optimal
sampling and transducer technology, and producing antibodies against specific biological
agents and strains. Scientists at the University of Texas, Austin (Daugherty et al., 1998;
Georgiou and Iverson, 1998) are taking the last of these areas one step further, reducing
the size of anthrax antibodies to that fragment of the light chain actually binding the
antigen, identifying the relevant amino acids at the binding site, and making systematic
substitutions to achieve higher affinity and selectivity.

The previously described device under development at Utah State University for
detection of total pathogenic microbes (including spores) is an example of a ligand-based
probe with fluorescence-based transduction. A hand-held unit simply using fluorescence
to determine total viable microbes requires no physical contact with the samples and no
specialized expertise to use, but it can provide detection in seconds with a sensitivity of
~100 cells. In this respect, it is useful for determining contamination on objects and from
environments where it is difficult to obtain samples.

Three substantial R&D efforts are currently under way that focus on mass
spectrometry (MS) for identifying biological agents. DoD is close to fielding a truck-
portable Chemical Biological Mass Spectrometer (CBMS) and already has research under
way at Oak Ridge National Laboratory (ORNL) for a second-generation unit that is
lighter, faster, and more sensitive (Wayne Griest, personal communication to FJ Manning,
1/23/98). Although very expensive compared to most portable chemical or biological
detectors and dependent on a rapid and efficient separation system, the name underlines
an important advantage of this approach--the potential for a single instrument that will
detect both chemical and biological agents, industrial and naturally occurring as well as
military. Unlike many of the current test systems and detectors, such an MS-based
detector could be used in a whole gamut of Hazmat situations rather than as confirmation
of a hypothesis about a possible agent. The instrument's versatility would be limited only
by the size of the existing library of mass spectra.

DARPA is sponsoring a collaboration of Johns Hopkins University, the University of

Maryland, and USAMRIID to develop a portable, fully automatic MS system and a
library of bioagent "signatures" (Cotter, 1998; Fenselau, 1997, 1998; Bryden et al., 1998).
The Department of Energy's Chemical Biological Nonproliferation Program is sponsoring
a similar developmental effort at ORNL, where researchers are attempting to lev-erage
hardware and software engineering currently under way in connection with the second
generation CBMS to produce a man-portable, real-time system capable of identifying
airborne bacteria or volatile organics as well as characteristic proteins of biowarfare
viruses, toxins, and bacteria (McLuckey, 1998; McLuckey at al., 1998; Stephenson et al.,
1998).

Although MS has the potential to identify infective agents and recent advances have
significantly reduced the size of the device, libraries of unique signatures of agents have
not been determined. In addition, it is not clear that these signatures can be distinguished
in a natural environment containing signatures of large amounts of other microbes,
especially at concentrations near infectious-dose levels.

Other detectors being developed at Sandia National Laboratory are based on

miniaturizing standard laboratory separation techniques, such as capillary zone
electrophoreses, size exclusion chromatography, and reverse phase and affinity
electrochromatography coupled with fluorescence (Vitko and Kottenstette, 1998; Dulay et
al., 1995; Ramsey et al., 1995). The challenge with these technologies is to achieve high
sensitivity in the presence of large amounts of interfering substances. Interfering
substances may have the same physical parameter that is being used for selectivity, such
as, charge, size, mass, which can cause wrong results, even though the results are highly
reproducible. For that reason, the investigators propose to use as many as four of these
techniques in parallel. Only when a sample is positive on all methods would the result be
considered unequivocal.

R&D NEEDS R&D NEEDS R&D NEEDS

 The type of detection technology that is needed depends upon the scenario, and, as is the
case with chemical agent detectors, it is likely that no one detector will meet all civilian
needs. As with R&D needs in other parts of this report, detector technology needs were
evaluated with three scenarios in mind: (1) general monitoring in the high-risk
environment, (2) an "event" (most likely a suspicious package in the case of biological
agents, but possibly an explosion of some sort), and (3) a "covert" release (patient
diagnostics). The first two scenarios call for some ability to detect biological agents in the
environment (air, water, food, etc.), while the third calls for methods that will detect and
identify pathogens in fluids or tissues from patients who exhibit signs or symptoms, or
who are known to have been exposed to a pathogen.

The committee does not see routine monitoring in the manner of smoke detectors (i.e.,
without some independent reason to suspect an attack) as either feasible in the foreseeable
future or worthy of a high-priority effort to develop that capacity, but there may be times
and places where pre-incident intelligence may suggest temporary deployment of existing
military monitoring systems.

Given the delayed effects of the biological agents, it is also difficult to envision many
situations that would demand highly sensitive biological detection by first responders. The
ability to determine total viable microbes present, total pathogenic microbes, and specific
viable pathogens will likely cover the needs presented by both overt and covert "events"
as well as provide monitoring and early warning. The ERDEC/EnViron virus detection
system might prove to be a useful complement to a ligand probe system for detecting total
pathogenic bacteria and handheld immunoassay tickets in a multistage approach
beginning with the very general and progressing to the highly specific as required. The
alternative might be a miniaturized mass spectrometer of the sort being developed at
Hopkins/Maryland or Oak Ridge to be a generic chemical and biological identifier.
Although prototypes are decreasing in size and weight, the real challenge lies in the
development of a library of unique signatures for biological agents in the presence of
large quantities of other microbial contamination and interferents, in addition to the
achievement of infectious-dose-level sensitivity.

In the area of patient diagnostics, there is a clear need for methods capable of detecting
infective dose levels (e.g., 10—100 cells or virions) of most biowarfare agents at a speed
that allows for effective therapeutic strategies to be administered (e.g., antibiotics,
vaccination, supportive therapy). Furthermore, these new methods must also be able to
detect "friendly" microbes that have acquired virulence factors by natural or genetic
engineering methods and those that have been microencapsulated to disguise their identity
(such as the detection of virulence factors or toxin production). Ideally, this technology
will be incorporated into a diagnostic system capable of identifying many more common
pathogens, assuring frequent use of the system and eliminating the need for clinicians to
make a specific request for a very seldom-used assay.
 The committee therefore has identified the following research and development needs:

6-1 In the area of patient diagnostics, the Public Health Service should encourage federal
research agencies to leverage burgeoning commercial development of faster, cheaper,
easier assays of common pathogens rather than independently developing diagnostic
technology for the less common pathogens thought to be good candidates for
bioterrorism.
6-2 In the area of environmental detection, the Public Health Service should closely
monitor military biodetection R&D efforts for inexpensive or multipurpose biodetectors
that might be appropriate for purchase or loan by civilian agencies rather than
developing threat agent-specific assays from the ground up.
6-3 Both of these leveraging efforts will require the federal government to conduct or
support:
 Basic research to identify characteristics which might be used to develop more
effective probes and/or enhance probe performance for known biowarfare agents and
especially genetically altered microbes. Understanding of microbial metabolism,
sporulation, toxin production and excretion, regulation of virulence factors, and
bacteriophage interaction are crucial in this respect. New approaches for preventive and
therapeutic strategies are also likely from this basic understanding.
 Scenario-specific testing of detection performance and comparisons under standard
conditions for characterization of the sensitivity, specificity, reliability, response
constraints, and usability (ease of use, cost, robustness, useful life, response time, and
human effort and experience required).