National Academies Press: OpenBook

Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases (2005)

Chapter: 7 Structure-Based Identification for Detect-to-Warn Applications

« Previous: 6 Nucleic Acid Sequence-Based Identification for Detect-to-Warn Applications
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

7
Structure-Based Identification for Detect-to-Warn Applications

Structure-based sensors—sometimes referred to as affinity-based sensors—represent a marriage of biology, biotechnology, physics, and instrumentation technology. Their function involves the following sequence of discrete steps:

  • Direct binding of the target to a specific molecular recognition element,

  • Transduction of the binding event into a measurable signal,

  • Evaluation of that signal to determine the amount of target bound, and

  • Use of that result to determine the amount of target present in the sampled environment.

The molecular recognition event is typically a specific interaction that is reversible, analogous to the interaction between a lock and a key, although in many cases the binding would more accurately be described as induced fit, during which the recognition element changes shape upon binding. This mode can be exploited in the sensor; detection of binding could rely on observation of that shape change. There are several factors that influence detection in signature-based sensors. These factors include the affinity of the target for the molecular recognition elements, nonspecific binding of extraneous material at the binding site, and the sensitivity of detection. Understanding how all these components influence the response of the sensing system is critical, especially when low (e.g., attomolar) detection limits are required. To achieve this goal, one should use high-affinity molecular recognition elements, reduce nonspecific binding by appropriate selection of the material that comes in contact with the sensor system, and employ high-sensitivity detection methods. Table 7.1 lists some transduction methods used in biosensors. The committee recognizes that many technologies involved in sensor signal transduction, such as optical fibers and waveguides, are rapidly evolving to enable the development of small, low cost, sensitive sensor systems. However, a review and evaluation of all sensor system transduction methods is beyond the scope of this report. Some specific examples of structure-based sensor systems that have been investigated for pathogen detection are given later in this chapter.

Table 7.2 lists some molecular recognition systems that could potentially be used for biothreat detection. DNA hybridization is another common molecular recognition approach used for biothreat detection. As it is covered in the previous chapter, it will not be repeated here.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

TABLE 7.1 Transduction Modes in Biosensors

Transduction Mode and Device

Observed Output

Optical

 

Fiber-optic and planar devices utilizing absorption, fluorescence, scattering, polarization, reflectivity, and/or interference of light

Changes in wavelength, intensity, emission profile, reflectivity, fringe patterns, polarization state, and refractive index.

Electrochemical

 

Potentiometric devices (e.g., ion-selective electrodes), amperometric devices, and conductometric devices

Changes in voltage, current, impedance and/or resistance.

Gravimetric

 

Acoustic wave devices, magnetic acoustic resonator sensors

Changes in mass and surface viscosity through shifts in frequency or phase of resonant vibrations.

Thermal

 

Thermistor devices

Changes in temperature through shifts in electrical output.

Magnetic

 

Magnetic field detectors

Changes in magnetic properties of paramagnetic particle reporters.

In many ways, structure-based biosensors mirror the highly effective in vivo processes that enable living organisms to respond appropriately to their environment. For example, cells respond rapidly and specifically to other cells, bacteria, viruses, hormones and other molecules and do so in proportion to the concentration of those signaling agents. In these signal transduction systems, the cell produces and displays its molecular recognition elements on its surface, embedded in its membrane. Each such element binds a specific target, usually to an extent that reflects the amount that is present. The binding activates a "reporter" function—usually a conformational change in the membrane-embedded molecular recognition molecule itself—that is then either detected directly or leads to a change in the molecular balance in the cell. Some structure-based sensors are modeled after biological systems, but are simpler, retaining only the specific binding components of the biological system. In general, they are easily replicated independent of the organisms (see Table 7.2 for some examples). The most common types of structure-based biosensors are immunosensors, which employ antibodies or antibody fragments as the recognition elements.1 Antibodies are proteins that are generated within organisms to bind molecules (antigens) that the organism recognizes as foreign. Thus, they will bind to the surfaces of potentially dangerous viruses, cells, or nonbiological chemicals. Given that vertebrates produce in excess of 1011 different antibodies, it is highly likely that one or more antibodies can be found to bind any given target.

Antibodies have historically been produced by inoculating animals (often rabbits) with the target analyte of interest and isolating the antibodies from the serum or the specific cells that generate them. This is a relatively costly and laborious process, and methods have recently been developed for generating antibodies in vitro, without the inoculation of vertebrates. For example, methods have been developed for generating antibodies on the surface of a bacteriophage,2 and a library of 109 human antibody fragments has been generated on the surface of yeast.3 Once these libraries of antibodies are

1  

B. Hock. 1997. Antibodies for immunosensors: A review. Analytica Chimica Acta 347:177-186.

P.B. Luppa, L.J. Sokoll, and D.W. Chan. 2001. Immunosensors: Principles and applications to clinical chemistry. Clin. Chem. Acta 314:1-26.

2  

I. Benhar. 2001. Biotechnological applications of phage and cell display. Biotechnology Advances 19:1-33.

3  

M.J. Feldhaus, R.W. Siegel, L.K. Opresko, J.R. Coleman, T.M. Feldhaus, Y.A. Yeung, J.R. Cochran, P. Heinzelman, D. Colby, J. Swers, C. Graff, H.S. Wiley, and K.D. Wittrup. 2003. Flow-cytometric isolation of human antibodies from a nonimmune Saccharomyces cerevisiae surface display library. Nature Biotechnology 21:163-170.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

TABLE 7.2 Molecular Recognition Systems for Biosensing

Molecular Recognition Element

Target Inhibitor

Comments

Single-stranded DNA

Complementary sequence of DNA

DNA hybridization is the basis for DNA biochip arrays and DNA amplification methods such as polymerase chain reaction (PCR), which are used for trace detection.

Antibody (a protein)

Proteins, carbohydrates, small organic molecules, etc.

Basis for immunoassays and immunosensors. Whole antibodies and parts of antibodies can now be developed in vitro.

Peptide (small part of a protein)

Proteins, carbohydrates, small organic molecules, etc.

Analogous to antibodies but much smaller and developed in vitro.

Enzyme (a protein)

Substrate (such biochemicals as urea, glucose, acetic acid).

Catalyzes the conversion of the substrate to a detectable product.

Lectin (a protein)

Carbohydrate

Lectins bind to polysaccharides on cell surfaces. Lectins typically bind to at least several types of organisms. This approach is expected to be most useful for sample preparation and general biodetection rather than for specific pathogen identification.

Receptor (a protein)

Proteins, carbohydrates, small molecules

In nature, receptors are often embedded in the membranes of cells. Ligand binding to a receptor causes a conformational change in the receptor that triggers detectable intracellular events.

Aptamer (a nucleic acid sequence)

Proteins, small organic molecules, etc.

Recognition is analogous to ligand-receptor binding, in contrast to sequence-specific hybridization between complementary strands of DNA.

Small molecules

Proteins, cells, etc.

Recognition is analogous to the interaction between an antibody and small antigen molecule; however, the small molecule is used as the molecular recognition element, and a biomolecule such as a protein on the surface of a cell is the target.

Imprinted polymers

Proteins, small organic molecules, whole cells, etc.

Under development but not yet proven for biodetection.

generated, high-throughput selection processes can be used to select the cells that contain antibodies that selectively bind the antigens of interest, and the selected cells can then be used to rapidly generate large quantities of the antibodies for sensor development. There is potential for these types of high-throughput in vitro methods to generate low-cost molecular recognition reagents for biothreat detection.

In addition to antibodies, a wide variety of other molecular recognition elements can be used for biosensing,4 some of which are summarized in Table 7.2. Many of these molecular recognition elements are proteins (e.g., enzymes, lectins, receptors), but some other types of molecular recognition elements under development may have properties (e.g., greater temperature and chemical stability) that make them better suited for use in environmental biosensors than are proteins. In the sections below, the committee describes the key features required for structure-based detection of biothreats, discusses some structure-based biosensor systems that have been investigated for biothreat detection, and highlights the areas that are promising and/or need development in order to achieve reliable operation in detect-to-warn situations, which will require rapid, reliable, and sensitive detection.

4  

S.S. Iqbal, M.W. Mayo, J.G. Bruno, B.V. Bronk, C.A. Batt, and J.P. Chambers. 2000. A review of molecular recognition technologies for detection of biological threat agents. Biosensors and Bioelectronics 15:549-578.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

THE STRUCTURE-BASED BIOSENSOR: BASIC ELEMENTS

Implementation of a structure-based sensor involves performing several steps in sequence:

  • Sample collection,

  • Sample concentration,

  • Binding of the target to the molecular recognition element,

  • Possible addition and removal of "reporter" groups,

  • Detection of target molecular recognition element complex,

  • Analysis of the output signal, and

  • Renewal of the sensor surface for repeated monitoring.

Each of these steps imposes its own unique constraints on the system, as discussed below.

Sample Collection

As in the case of all other sensors, the first step in identifying a foreign agent involves the collection of the sample. Much of the discussion of collection methods in Chapter 4 applies directly to structure-based biosensors. It is critical, however, that the process of collection not alter the structure of that portion of the target that is to be bound by the molecular recognition element. Thus, cell surface proteins, or protein toxins in particular, must not be denatured in the collection process.

For some sensors, particularly those that are function-based (see Chapter 9), there is concern that the collection process might damage or kill the cellular target. This is not likely to be a problem with structure-based sensors. The molecular-level cell surface structures to be bound typically appear repeatedly on the surface of the cellular target and, individually, cover a very small area. Thus even if the cell surface were disrupted, the surface structures would remain intact, albeit on many particles (rather than one), and would retain their ability to bind the molecular recognition elements.

The breaking of the cells could introduce other factors, however. One possibility is that this could release large numbers of molecules from inside the cell that are common to a variety of organisms and may resemble surface molecules and compete with them for the binding sites. Other interfering components that could be released upon breaking of cells are proteases. Proteases are enzymes that will interfere with the detection of proteinacious components, and they could also hydrolyze protein-based reagents. On the other hand, fragmentation could simplify the molecular recognition element's access to the binding site or even expose unique groups that were inaccessible in whole, intact cells.

Collection techniques that kill cellular targets would not typically bring about structure-based detection. Dead cells generally exhibit the same surface molecules as live cells. This actually confers an advantage for structure-based sensors compared with those that depend on target cell function, in that there would be more targets to bind and detect. The presence of dead cells would usually indicate the presence of live cells elsewhere in the sample or physical environment from which the sample was taken, so their detection is often of value.

Sample Concentration

Sample concentration is a significant consideration for structure-based recognition. Currently deployed immunosensors typically require 104 binding events or more for detection. Therefore, to attain a detection threshold of 10 to 100 agent-containing particles per liter of air (ACPLA) or better, these sensors must collect and concentrate thousands of liters of air. For example, once triggered, the Joint Biological Point Detection System (JBPDS) collects air at 800 liters per minute for 2.5 minutes before conducting immunoassays. This time would have to be shortened in order to achieve a total detect-to-warn time (from sample collection to answer) of 1 minute or less. Fortunately, improvements in structure-based assays have the potential to improve the level of detection (LOD) to 100 (or even fewer) binding events. If actualized, this will greatly reduce the demand on the collector/concentrator prior to detection.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

Sample purification also must be considered, especially if there is a desire to obtain low detection limits in samples with a high background concentration of particles. Sample components that nonspecifically bind to a sensor surface can effectively block binding sites. In some sensor configurations, nonspecific binding of matrix components also contributes a background signal, which can result in a false positive if the matrix composition changes over time.

Binding of Target to the Molecular Recognition Element

To meet the constraints for detect-to-warn systems that report in 1 to 2 minutes, the binding of the molecular recognition elements, the target, and other required reagents must occur as quickly as possible. If there are multiple binding steps required for detection, the time constraints for each binding step are even more severe.

There have been many reports investigating the kinetics of binding targets to molecular recognition elements immobilized onto biosensor surfaces.5 These results show that under flowing conditions and with sufficient target concentration, detectable one-step binding can be achieved in a few seconds.6,7,8 However, the results to date indicate that 1-minute analysis time is a challenge and will require a sensor design that minimizes the number of binding and processing steps and enhances mass transport of the target to the sensor surface.

For example, it has been shown that at a protein concentration of 1 microgram per milliliter (about 7 nanomoles for a protein of 150,000 daltons), protein binding to an antibody-coated sensor can be detected within 20 seconds when the sample is flowing over the sensor surface.9 However, several minutes were required to achieve a response that was 50 percent of the steady-state response. The response would be faster if the target concentration were higher, if the rate of target transport to the sensor surface were increased, or if the target were smaller in size (to increase its rate of diffusion to the sensor surface). The time required to obtain a detectable signal can be several minutes or longer if there is no flow, if the target concentration is further decreased, and/or if the target diffuses slowly in the sample fluid and artificial mixing cannot be achieved.10

5  

A. Sadana. 1998. An analysis of analyte-receptor binding kinetics for biosensor applications: Influence of the fractal dimension on the binding rate coefficient. Biosensors and Bioelectronics 13:1127-1140.

Y.Y. Yu, B.J. Van Wie, A.R. Koch, D.F. Moffett, and W.C. Davis. 1998. Real-time analysis of immunogen complex reaction kinetics using surface plasmon resonance. Analytical Biochemistry 263:158-168.

B. Goldstein, D. Coombs, X. He, A.R. Pineda, and C. Wofsy. 1999. The influence of transport on the kinetics of binding to surface receptors: Application to cells and BIAcore. J. Mol. Recognit. 12:293-299.

R.A. Vijayendran, F.S. Ligler, and D.E. Leckband. 1999. A computational reaction-diffusion model for the analysis of transport-limited kinetics. Anal. Chem. 71:5405-5412.

A. Ramakrishnan and A. Sadana. 2000. An evaluation of cellular analyte-receptor binding kinetics utilizing biosensors: A fractal analysis. J. Colloid. Interface Sci. 224:219-230.

H.P Jennissen and T. Zumbink. 2001. Mass transport-free protein adsorption kinetics in biosensor systems. FASEB Journal 15:A531.

A. Sadana. 2001. A kinetic study of analyte-receptor binding and dissociation, and dissociation alone, for biosensor applications: A fractal analysis. Analytical Biochemistry 291:34-47.

K.E. Sapsford, Z. Liron, Y.S. Shubin, and F.S. Ligler. 2001. Kinetics of antigen binding to arrays of antibodies in different sized spots. Anal. Chem. 73:5518-5524.

Y.S.N. Day, C.L. Baird, R.L. Rich, and D.G. Myszka. 2002. Direct comparison of binding equilibrium, thermodynamic, and rate constants determined by surface- and solution-based biophysical methods. Protein Sci. 11:1017-1025.

C.C. Fong, M.S. Wong, W.F. Fong, and M. Yang. 2002. Effect of hydrogel matrix on binding kinetics of protein-protein interactions on sensor surface. Analytica Chimica Acta 456:201-208.

P. Gomes and D. Andreu. 2002. Direct kinetic assay of interactions between small peptides and immobilized antibodies using a surface plasmon resonance biosensor. Journal of Immunological Methods 259:217-230.

6  

A. Sadana and A. Ramakrishnan. 2002. A kinetic study of analyte-receptor binding and dissociation for biosensor applications: A fractal analysis for cholera toxin and peptide-protein interactions. Sens. Actuator B-Chem. 85:61-72.

7  

M. Abrantes, M.T. Magone, L.F. Boyd, and P. Shuck. 2001. Adaptation of a surface plasmon resonance biosensor with microfluidics for use with small sample volumes and long contact times. Anal. Chem. 73:2828-2835.

8  

Sapsford et al., 2001. See note 5 above.

9  

Sapsford et al., 2001. See note 5 above.

10  

Vijayendran et al., 1999. See note 5 above.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

The relationship between target concentration and analysis time has implications for biodetection in detect-to-warn situations, because to rapidly detect trace concentrations, it is necessary to transport the target to the biosensor surface rapidly and therefore to increase the target concentration near the sensor surface. In addition, even for the example above (1 microgram per milliliter protein target), in which a signal was detectable in a few seconds (and within the 1 minute time constraint), a much larger detection signal could be obtained by locally increasing the target concentration at the sensor surface. A larger signal would be desirable to increase the signal-to-noise ratio and therefore decrease the number of false positives and also to improve the sensitivity to low concentration. In fact, the committee expects that detect-to-warn applications will require the detection of concentrations at least as low as 1 nanogram per milliliter (about 10 pM, or 10-12 moles per liter) for toxins and concentrations on the order of attomoles (10-18 moles per liter) for other biothreats (for examples, see Boxes 7.1 and 7.2).

One method of avoiding transport-limited kinetics is to increase the flow rate. However, this is not always feasible since it requires larger sample volumes. Oscillatory flow has been shown to perform just as well as increasing the flow rate in some situations, resulting in detectable signals in a few seconds.11 Some other methods that could be used to transport the analyte to the sensor surface include ultrasonic focusing, electrophoretic focusing, and centrifugation. All of these methods have inherent limitations that depend upon factors such as the molecule size and surface charge, so the method of choice depends upon the analyte to be detected and the composition of the sample matrix.

Specific Detection and False Alarms

Specificity is also a critical issue, especially when the identification of biothreats is desired. The choice of biomarkers for detection will determine the potential selectivity of a structure-based bioagent detector. For example, if one develops antibodies or aptamers for cell surface epitopes that are present on all Bacillus species, even a perfectly operating sensor will respond to all Bacillus species and not specifically to Bacillus anthracis. However, if one develops a structural recognition element for a virulence marker protein, with proper sample preparation, the virulence marker protein will be detected, even if the virulence is engineered into a completely different organism. Therefore, careful selection of biomarkers and the use of multiple biomarkers are important considerations to enable specific bioagent detection. Research to determine the appropriate biomarkers for bioagents of interest and develop structural recognition elements for those biomarkers is critical for selective bioagent detection.

Another important consideration for sensor specificity is the binding of untargeted substances to the sensor surfaces. Binding of nontarget substances will result in false positive responses and is likely to be caused by two major factors: (1) the difficulty in achieving absolute specificity on the part of the molecular recognition elements and (2) the ubiquitous, nonspecific binding of extraneous material in the sample to the molecular recognition element or to the surfaces of the sensor itself. Both of these undesirable effects can be mitigated with improved design. In the former case, for example, it is known that extremely specific proteins can be designed, as shown by the existence of the exceptionally high specificity of proteolytic enzymes involved in blood clotting or proenzyme activation. In the latter case, engineering of the sample or the surfaces of the sensor, or adjustment of the pH, ionic strength, or other conditions of the assay, would likely be of great value.

Lack of specificity could also be mitigated if the system is designed to respond not to a single binding event but to two or more that arise independently. In one such scheme, two or more molecular recognition elements would be used, each designed to bind to its own distinct target epitope (binding structure on the target). A positive response would be recorded only when all recognition elements are bound simultaneously. Since each binding event is independent of the other, and they occur in parallel,

11  

Abrantes et al., 2001. See note 7 above.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

Box 7.1
Notional Structure-Based Detection and Identification System: Reliable Level of Detection at 10,000 Targets Bound to the Sensor Surface

Below, the committee outlines a detection system that includes system components that are only slightly beyond what has already been demonstrated in a few laboratory systems. This notional system highlights the fact that a dramatic improvement in system-level detection limit beyond what is now achievable is desirable for detection in 2 minutes or less. This notional system assumes that an approximately 1 femtomolar concentration of analyte (3 × 104 analyte molecules in 50 microliters) is required for reliable detection. This detection limit has been demonstrated using several structure-based detection systems currently available.


Sample Collection

  • Collect the sample for 400 seconds from the air into an aqueous solution using a two-stage, precollection fractionator with an overall capture efficiency of 50 percent and a collection rate of 90 liters of air per minute into 50 microliters of aqueous volume.a An air sample with 100 target structures per liter would be concentrated to 30,000 spores in a 50 microliter solution.b

  • Bind molecules to the sensor surface over 20 seconds via an active transport mechanism (e.g., pressure, electrophoretic transport, ultrasonic focusing) that moves the analyte to the sensor surface. About 10,000 analyte molecules bind to the sensor surface.

  • Wash the sensor surface for 10 seconds.

Reporter Group (Optimal)

  • Deliver reporter to sensor surface over 20 seconds.

Detection

  • Detect the presence of the 10,000 analyte molecules within 2 seconds.

Regeneration

  • Wash the sensor surface for 10 seconds.

Total analysis time is 7 minutes, 42 seconds with all options, and 7 minutes, 2 seconds for direct detection without washing or the need for a reporter. This would not meet the needs of a detect-to-warn system. In both cases, most of the time is dedicated to collecting and concentrating the sample. However, if multiple targets are present on the organism, it may be possible to dramatically shorten the analysis time by lysing the organism to generate multiple, separate targets for detection. The initial steps might then become:

  • Collect for 4 seconds, resulting in 300 spores in 50 microliters.

  • Lyse organisms for 10 seconds to generate 100 targets from each spore (30,000 total targets).

This possibility would dramatically shorten the time for the collection of 30,000 targets from 6 minutes, 40 seconds (without lysis) to 14 seconds (with lysis) and would reduce the total analysis time to 1 minute, 16 seconds. This is well within the range of detect-to-warn requirements. Similarly, in some detection schemes and with some organisms, multiple targets can be detected on each organism without lysis, resulting in greater levels of sensitivity.

a  

Collection of aerosol directly into a liquid can be replaced by collection of dry aerosol onto a surface such as a sample tube, followed by addition of liquid for the detection assay (e.g., see B cell example in the "Modified Cell-Based Systems" section of this chapter).

b  

Many structure-based detection systems report detection limits ranging from picomolar to nanomolar concentrations, which would require increasing the collection time by three to six orders of magnitude in this example. The resulting collection times required to achieve a detection limit of 100 ACPLA would therefore range from about 4,000 to 4,500 days, which is certainly much longer than needed for detect-to-warn (and many other) applications.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

Box 7.2
Notional Structure-Based Detection and Identification System: Reliable Level of Detection at 100 Targets Bound to the Sensor Surface

Below, the committee outlines a detection system that is based on a level of detection that has been demonstrated in the best laboratory systems (e.g., see CANARY B cell example in the text). This notional system assumes that approximately 300 targets (or cells/spores/viruses) must be collected in 50 microliters for reliable, rapid detection. This corresponds to a molar target concentration of about 10 × 10-18 M (10 aM).

  • Collect the sample over 4 seconds from the air into a Joule aqueous solution, using a two-stage, precollection fractionator with an overall capture efficiency of 50 percent and a collection rate of 90 liters of air per minute;a 100 spores/liter of air would produce 300 spores in 50 μL solution.

  • Wait 20 seconds for an active transport mechanism (e.g., pressure, electrophoretic transport) to move the analyte to the sensor surface; about 100 analyte molecules bind to the sensor surface.

Optional steps depend upon assay format and transduction method:

  • Wash the sensor surface for 10 seconds.

  • Deliver reporter to sensor surface over 20 seconds.

  • Wash the sensor surface for 10 seconds.

  • Wait 2 seconds to detect the presence of the 100 analyte molecules.

The total analysis time is 1 minute, 6 seconds with all options and would be 26 seconds for direct detection without washing or the need for a reporter. This would therefore meet the needs for a detect-to-warn system. Analysis of only 10 ACPLA could be achieved by increasing collection time by 36 seconds, resulting in a total analysis time that is still less than 2 minutes for this example. Also, note that the total analysis time doubles to 76 seconds with the addition of binding and washing steps for adding one reporter group.

a  

Collection of aerosol directly into a liquid can be replaced by collection of dry aerosol onto a surface such as a sample tube, followed by addition of 50 microliters liquid for the detection assay (e.g., see CANARY B cell example).

this scheme would not slow the rate of response. The reduction in false positives could, however, be dramatic. Two binding sites that individually produce false positives once in 103 events would, together, give a false alarm only once in 106 events. Three elements would be expected to have a false alarm once in 109 events. This is especially important for continuously operating detect-to-warn systems. Even a low false alarm rate of 10-3 would result in an alarm about every 20 hours on average, if the system is cycled once every minute. A false alarm rate of 10-6 would result in an alarm only about once every 2 years.

This advantage of parallel sensor design to minimize false alarms will only be realized if the false alarms are not correlated to one another. For example, if the nonspecific binding of matrix material systematically produces false alarms in all sensor elements (e.g., all SPR sensors that will detect nonspecifically bound proteins), the false alarm rate due to this factor would not be improved. (However, a control sensor surface without the selective chemistry can be used to normalize the sensor signal and minimize the effects of matrix materials.) Of somewhat less value would be the deployment of several different types of structure-based sensors using independent binding and detection schemes.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

Addition and Removal of Reporter Groups

In many sensor systems, binding is detected only through the presence of a separate reporter group—for example, a fluorescent, magnetic, or other type of tag. In other systems, e.g., optical techniques such as surface plasmon resonance, target binding is detected directly, so that the analyte itself is also the reporter. The use of reporter groups adds some complexity to the system, because separate steps are typically required for their binding and also for washing to remove nonspecifically bound and unbound reporters from the sensor surface. Any additional binding steps are governed by the same transport and reaction kinetics described above for the binding of the molecular recognition element.

In general, surface-sensitive detection methods are desirable, because these methods will only detect the reporters at or near the sensor surface and not the unbound reporters in the bulk solution. Surface-sensitive methods have the potential to provide a more rapid response than bulk detection methods, because detection can be measured without the requirement that the unbound reporter be washed from the sensing surface. When bulk detection methods are used, the sensor detects the reporter whether or not it is bound to the target, necessitating the additional wash. Some common surface-sensitive detection methods include surface plasmon resonance devices, optical detection systems utilizing evanescent waves for optical excitation from optical fibers or waveguides, and acoustic wave devices. Some magnetic detection schemes ignore unbound reporters.12

Even more desirable is a method in which only specifically bound reporters (and not the nonspecifically bound reporters) are detected. In this case, the false positive rate will decrease, and the limit of detection may also improve. This could occur, for example, if the reporter reacts with the target to generate a detectable signal and reaction only occurs upon specific binding of the reporter with the target.

Wash steps not only take time but also can lead to the dissociation of the target molecular recognition element complex and result in a reduced signal. It is therefore important that the target molecular recognition complex be stable. This is often the case for structure-based molecular recognition systems. For example, while antigen binding occurs in seconds, antigen removal with gentle washing occurs in many minutes to hours. This is because for high-affinity binding, the rate of binding is many orders of magnitude greater than the rate of dissociation.13 Other molecular recognition elements can be designed with similar properties.

Detection of Target Molecular Recognition Element Complex

A variety of signal transduction methods (Table 7.1, and also see Box 8.2 in Chapter 8) have been proposed for the detection and reporting of the target molecular recognition element (reporter) complex.14

12  

Y.R. Chemla, H. L. Grossman, Y. Poon, R. McDermott, R. Stevens, M.D. Alper, and J. Clarke. 2000. Ultrasensitive magnetic biosensor for homogeneous immunoassay. Proc. Natl. Acad. Sci. 97:14268-14272.

13  

Hock, 1997. See note 1 above.

Vijayendran et al., 1999. See note 5 above.

14  

A. Guiseppi-Elie and A.M. Wilson. 1995. Electroconductive polymer thin films with bioactive moieties for biosensor applications. Polym. Mater. Sci. Eng. 72:404-405.

F.W. Scheller, F.F. Bier, and D. Pfeiffer. 1995. Biosensors: Principles and applications. Technisches Messen 62:213-219.

J.H. Kim. 1997. Research trends in biosensors. Hwahak Sekye 37:23-32.

J. Rishpon and D. Ivnitski. 1997. An amptronic enzyme-channeling immunosensor. Biosensors and Bioelectronics 12:195-204.

D.R. Baselet, G.U. Lee, M. Natesan, S.W. Metzger, P.E. Sheehan, and R.J. Colton. 1998. A biosensor based on magnetoresistance technology. Biosensors and Bioelectronics 13:731-739.

P.M. Fratamico, T.P. Strobaugh, M.B. Medina, and A.G. Gehring. 1998. Detection of Escherichia coli O157: H7 using a surface plasmon resonance biosensor. Biotechnol. Tech. 12:571-576.

J.C. Pyun, H. Beutel, J.U. Meyer, and H.H. Ruf. 1998. Development of a biosensor for E-coli based on flexural plate wave transducer. Biosensors and Bioelectronics 13:839-845.

I. Abdel-Hamid, D. Ivnitski, P. Atanasov, and E. Wilkins. 1999. Flow-through immunfiltration assay system for rapid detection of E-coli O157:H7. Biosensors and Bioelectionics 14:309-316.

D. Ivnitski and I. Abdel-Hamid. 1999. Biosensors for detection of pathogenic bacteria. Biosensors and Bioelectronics 14:599-624.

Chemla et al., 2000. See note 12 above.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

Many of these transduction methods are extremely fast, and the analysis of the signal should also be very fast. Thus, so long as transport and binding are fast, detection can be achieved in less than 1 minute. However, the system needs to be sufficiently automated so that cleaning and preparation of the instrument and loading of the next sample are not too time consuming.

Renewal of the Sensor Surface for Continuous Monitoring

For continuous use of a sensor, the rate of molecular recognition element (MRE) target must be tuned to regenerate the sensor surface at the end of an analysis. This is sometimes achieved by changing the wash conditions to disrupt the interaction between the target and the molecular recognition element. The stringent washing must remove bound target and nonspecifically bound materials but leave the molecular recognition element unaffected. Renewal of the sensor surface is typically only possible for a limited number of cycles, after which time the sensor surface must be replaced due to degradation.15 Therefore, molecular recognition elements that can withstand harsh washing procedures are desirable to enable repeated renewal of the sensor surface.

CONSUMABLES CONSIDERATIONS FOR DETECT-TO-WARN APPLICATIONS

Both the cost and amounts of consumables must be minimized for detect-to-warn applications, which require near-continuous operation. For example, if an analysis is done every 2 minutes, 720 assays will be completed each day, and 256,320 assays will be done each year. Therefore, even a consumables cost of only 4 cents per assay will add up to over $10,000 per sensor per year. Current assay costs are between one and two orders of magnitude higher than this cost. However, the committee expects that it will be possible to decrease the consumables cost per assay to 4 cents per assay or less by decreasing the size of the sensor systems and total liquid volumes to 100 microliters or less per sample and by the development of novel, low-cost reagents and methods for selective binding and detection.

The consumables costs include all reagents required for detection and system cleaning, and also the cost of generating and maintaining the selective sensing surface itself. It is known that repeated use of a structure-based sensing surface will require repeated cleaning (and therefore consumables) to remove nonspecific and specifically bound materials. In addition, degradation of a sensing surface typically occurs after repeated use, so that methods will have to be developed for periodic automated replacement of the sensing surface (or complete replacement of the sensor). For example, a sensor that can be reused 100 times would be replaced 2,500 times a year, and a sensor that can be reused 1,000 times would be replaced 250 times per year.

   

E. Howe and G. Harding. 2002. A comparison of protocols for the optimization of detection of bacteria using a surface acoustic wave biosensor. Biosensors and Biolectronics 15:641-649.

W.M. Mullett, E.P.C. Lai, and Y.M. Yeung. 2000. Surface plasmon resonance-based immunoassays. Methods 22:77-91.

S.T. Pathirana, J. Barbaree, B.A. Chin, M.G. Hartell, W.C. Neely, and V. Vodyanoy. 2000. Rapid and sensitive biosensor for Salmonella. Biosensors and Bioelectronics 15:135-141.

C. Aston. 2001. Biological warfare canaries. IEEE Spectrum (October):35-40.

P. Ertl and S.R. Mikkelsen. 2001. Electrochemical biosensor array for the identification of microorganisms based on lectin-lipopolysaccharide recognition. Anal. Chem. 73:4241-4248.

A.P. Ferreira, M.M. Werneck, and R.M. Ribeiro. 2001. Development of an evanescent-field fibre optic sensor for Escherichia coli O157: H7. Biosensors and Bioelectronics 16:399-408.

J.B. Delehanty and F.S. Ligler. 2002. A microarray immunoassay for simultaneous detection of proteins and bacteria. Anal. Chem. 74:5681-5687.

F.L. Dickert and O. Hayden. 2002. Bioimprinting of polymers and sol-gel phases: Selective detection of yeasts with imprinted polymers. Anal. Chem. 74:1302-1306.

C. Ercole, M. Del Gallo, M. Pantalone, S. Sartucci, L. Mosiello, C. Laconi, and A. Lepidi. 2002. A biosensor for Escherichia coli based on a potentiometric alternating biosensing (PAB) transducer. Sens. Actuator B-Chem. 83:48-52.

Z.Z. Li, F.C. Gong, G.L. Shen, and R.Q. Yu. 2002. Bacteria-modified amperometric immunosensor for a Brucella melitensis antibody assay. Analytical Sciences 18:625-630.

F.S. Ligler, M. Breimer, J.P. Golden, D.A. Nivens, J.P. Dodson, T.M. Green, D.P. Haders, and O.A. Sadik. 2002. Integrating waveguide biosensor. Anal. Chem. 74:713-719.

15  

M.A. Gonzalez-Martinez, R. Puchades, and A. Maquieira. 1999. On-line immunoanalysis for environmental pollutants: From batch assays to automated sensors. Trends in Anal. Chem. 18:204-218.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

In addition to the cost of consumables, there are practical engineering and deployment challenges related to the use of consumables. Both structure-based and sequence-based detection approaches require a water environment for specific binding. If only 100 microliters of aqueous solution are used per analysis (e.g., 50 microliter volume for aerosol collection plus 50 microliters for additional processing and washing), 2 liters of liquid would be used and accumulated each month, and 25 liters of liquid would be used each year. Therefore, current approaches that use a total liquid volume on the order of 1 milliliter per analysis are not suitable for detect-to-warn applications. With significant investments in structure-based sensor development, the committee believes that novel approaches that use small volumes of liquid per analysis are technically achievable for detect-to-warn applications (see Box 7.1 regarding needs for aerosol collection into small volumes).

It is also conceivable that specific structure-based detection could be conducted on a sensing surface that includes a water environment (e.g., hydrogel or liquid droplets) containing all reagents for the structure-based assays. The sensor surface could then directly accumulate aerosol particles for each analysis, and a fresh surface would be used for each analysis (analogous to moving tape matrix-assisted laser desorption/ionization/mass spectrometry (MALDI-MS) systems under development). While such an approach may be feasible, there are many science and engineering challenges to realizing such a system. While structure-based detection is capable of rapid detection for detect-to-warn applications, future research is needed to develop novel detection concepts that minimize reagent volumes, minimize disposables, and decrease the cost per assay.

NOTIONAL STRUCTURE-BASED DETECTION SYSTEMS

In considering potential approaches for detect-to-warn applications, it will be useful to compare systems and approaches described throughout the rest of this chapter to the two notional systems described in Boxes 7.1 and 7.2. These estimate the analysis time for two different detection limits: (1) one femtomolar (about 10,000 targets bound to the sensor surface, see Box 7.1) and (2) 10 attomolar (about 100 targets bound to the sensor surface, see Box 7.2). These situations were selected because 104 molecules bound to the sensor surface (Box 7.1) is a detection level that is reported for many commercially available structure-based detection systems and other systems under development. A detection level of 100 molecules bound to the sensor surface (Box 7.2) has been reported for a few systems under development and is therefore expected to be reasonably attainable in the near future.

In both examples, an overall system-level detection limit of 100 ACPLA in air was used, and an aerosol sampling rate that is consistent with the notional examples presented in Chapter 6 was assumed (90 liters per minute collection into 50 μL liquid with 50 percent efficiency). These notional examples highlight several important points:

  • For detect-to-warn applications (detection in 2 minutes or less), a very sensitive detection limit (tens of attomolar concentration, or about 100 targets bound to the sensor) is required in order to shorten the aerosol collection time.

  • Detection time can be decreased dramatically by decreasing the number of processing steps.

  • Transport of the targets (and also detector molecules if they are used) close to the sensing surface is an important consideration to allow binding and washing steps that each require only 10 to 20 seconds or less.

  • These notional systems have been carefully prepared and are presented in terms of specific numerical levels of target number, target concentration, sample volume, concentration efficiency, and time. No meaningful discussions of various recognition elements and signal transduction systems can take place unless these numbers are available for each system to be evaluated.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

DETAILED CONSIDERATIONS: MOLECULAR RECOGNITION ELEMENTS

The heart of the biosensor is the molecular recognition element that must, with high affinity and selectivity, bind a target that might often be found in low concentrations in a complex mix of similar structures. In discussing the alternative molecular recognition elements listed in Table 7.2, a number of critical factors must be evaluated, including specificity, affinity (both rates of association and dissociation), stability, and manufacturability (e.g., cost, time for production, storage requirements, and lifetime).

Antibodies

The classical molecular recognition elements, antibodies, are produced by most vertebrates. Roughly 1011 different antibodies are produced by each organism and can be found circulating in its blood stream. Well-established techniques exist for isolating large quantities of a single antibody that binds a specifically defined target (antigen).16 The screening of blood from suitably immunized animals allows the collection of polyclonal antibodies, a population of a number of different antibodies, each of which can bind the target, albeit with its own characteristic binding constant (which can vary over several orders of magnitude). Other techniques allow for the selection of a single, monoclonal antibody against the target.

Antibody binding constants are quite variable depending on the target and the method used to raise the antibodies. This provides valuable flexibility in sensor design. Some applications would benefit from low-specificity antibodies that could bind any of a family of targets. In other cases, high specificity is required. Precedent exists for this: Some proteolytic enzymes—for example, digestive enzymes—have very little substrate discrimination, hydrolyzing the peptide bond between amino acids regardless of the nature of those or neighboring amino acids. Other enzymes, often with similar structures and mechanisms of action, can be exceptionally specific—for example, those that activate the blood clotting system or cleave the polyprotein precursors of viruses such as HIV. Some antigens appear to present significant difficulties in the development of high-affinity antibodies, although techniques do exist to mitigate these problems. Naturally occurring antibodies can be used, but recently developed techniques of "antibody evolution" allow for the selection of antibodies with increasingly high affinities.

Techniques have recently been developed for generating antibodies in vitro, without the inoculation of vertebrates. For example, methods have been developed for generating antibodies on the surface of bacteriophages (viruses that infect bacteria),17 and a library of 109 human antibody fragments has been produced on the surface of yeast.18 Each yeast cell virus produces a single antibody, and once these libraries of antibodies are generated, high-throughput selection processes can be used to select the individual cells that produce the antibody that selectively binds the antigen of interest, and the selected cells can be used to rapidly generate large quantities of the antibodies for sensor development.

Because antibodies are proteins, questions about storage and stability must be addressed. Small variations in temperature, pH, and ionic strength can lead to denaturation (the unfolding of the three-dimensional structure of the antibody) and the loss of its ability to bind its target antigen. In addition, the ubiquitous presence of proteolytic enzymes, which degrade proteins, can severely limit antibody lifetimes.

16  

Hock, 1997. See note 1 above.

L.M. Houdebine. 2002. Antibody manufacture in transgenic animals and comparisons with other systems. Curr. Opin. Biotechnol. 13:625-629.

C.R. Suri, M. Raje, and G.C. Mishra. 2002. Immunosensors for pesticide analysis: Antibody production and sensor development. Crit. Rev. Biotechnol. 22:15-32.

T.J. Torphy. 2002. Pharmaceutical biotechnology: Monoclonal antibodies, boundless potential, daunting challenges. Curr. Opin. Biotechnol. 13:589-592.

17  

S.S. Sidhu. 2000. Phage display in pharmaceutical biotechnology. Curr. Opin. Biotechnol. 11:610-616.

I. Benhar. 2001. Biotechnological applications of phage and cell display. Biotechnology Advances 19:1-33.

R.H. Hoess. 2001. Protein design and phage display. Chemical Reviews 101:3205-3218.

B. Zhou, P. Wirsching, and K.D. Janda. 2002. Human antibodies against Bacillus: A model study for detection of a protection against anthrax and the bioterrorist threat. Proc. Natl. Acad. Sci. 99:5241-5246.

18  

M.J. Feldhaus, R.W. Siegel, L.K. Opresko, J.R. Coleman, T.M. Feldhaus, Y.A. Yeung, J.R. Cochran, P. Heinzelman, D. Colby, J. Swers, C. Graff, H.S. Wiley, and K.D. Wittrup. 2003. Flow-cytometric isolation of human antibodies from a nonimmune Saccharomyces cerevisiae surface display library. Nature Biotechnology 21:163-170.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

The Defence Science and Technology Laboratory (Dstl) in the United Kingdom reports the use of antibodies with better than 90 percent maintenance of activity over 30 hours in their optical evanescent biosensors for ricin.19 In other systems—for example the antibodies for Staphylococcus enterotoxin B proteins—the antibodies were reported to lose half of their activity after only 3 hours.20 In addition, the washing conditions required to remove bound antigens and regenerate an antibody sensor surface are known to degrade many antibodies over time.21 Therefore, for detect-to-warn applications that require continuous monitoring, the development of robust antibody fragments for molecular recognition and the development of alternative molecular recognition elements that are more rugged and have longer lifetimes than antibodies are desirable to decrease the cost of repeatedly renewing the sensing surfaces and the complexity of the final sensor system.

Aptamers

Aptamers are oligomers of RNA or DNA that spontaneously fold into specific three-dimensional shapes that can bind defined targets. The specificity of the shape of the binding site arises from the base sequence of the aptamer, which determines the base-pairing pattern of the oligomer. The prototypical, proof-of-principle example of how this base pairing and three-dimensional folding within a single nucleic acid molecule can create specific binding sites is the complex and distinctive L-shaped structure of transfer RNA (tRNA). This molecule has regions of base pairing and regions that are single-stranded. Both are critical to its role in protein synthesis. Molecules of tRNA with 20 distinct binding sites are produced, each having its own exceptional specificity for binding an acylating enzyme and amino acid.

Aptamers are easily synthesized through the use of automated machines that can produce oligomers of desired sequence and length, and automated methods have been developed to select aptamers with high affinity to targets of interest.22 Typically, aptamers have about 40 bases. With 4 different bases available, 440 possible sequences can be generated. In practice, libraries of 1015 members are made for screening against targets. The nucleotide sequences of those that bind well are determined, and large numbers of variants of those "leads" are produced to optimize binding. The process is repeated, yielding molecules with greater specificity and binding strength at each cycle. This systematic evolution of aptamers can be highly successful, as has been shown for the development of RNA aptamers that bind to the protein bacteriophage T4 DNA polymerase.23 Known sequences of DNA are typically added at the ends of the 30 to 40 base aptamers to allow PCR amplification for the production of many copies of a selected aptamer. Aptamers have also been modified to resist nuclease digestion and thereby exhibit extended stability. A variety of technical modifications such as the cross-linking of two aptamers that bind to different parts of a target are being investigated.24

Aptamer technology provides additional opportunities in terms of the design of molecular recognition elements that could have affinities and selectivities that complement those of the more traditional antibody reagents. In some cases DNA aptamers have been reported to have higher affinities than antibodies.25 Aptamers also are reported to have greater stability and a longer shelf life than proteins

19  

Peter Biggins, Dstl. Presentation to the committee on June 12, 2002.

20  

Biggins, 2002. See note 19 above.

21  

Gonzalez-Martinez et al., 1999. See note 15 above.

22  

B.E. Eaton, L. Gold, B.J. Hicke, N. Janjic, F.M. Jucker, D.P. Sebesta, T.M. Tarasow, M.C. Willis, and D.A. Zichi. 1997. Post-SELEX combinatorial optimization of aptamers. Bioorg. Med. Chem. 5:1087-1096.

J.C. Cox and A.D. Ellington. 2001. Automated selection of anti-protein aptamers. Bioorg. Med. Chem. 9:2525-2531.

L.J. Sooter, T. Reidel, E.A. Davidson, M. Levy, J.C. Cox, and A.D. Ellington. 2001. Toward automated nucleic acid enzyme selection. Biol. Chem. 382:1327-1334.

23  

C. Tuerk and L. Gold. 1990. Systematic evolution of ligands by exponential enrichment: RNA ligands to bacteriophage-T4 DNA-polymerase. Science 249:505-510.

24  

E.N. Brody, M.C. Willis, J.D. Smith, S. Jayasena, D. Zichi, and L. Gold. 1999. The use of aptamers in large arrays for molecular diagnostics. Molecular Diagnostics 4:381-388.

25  

L. Gold, Somalogic. Presentation to the committee on June 13, 2002.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

such as antibodies. In addition, as with proteins and peptides, aptamers can easily be attached to surfaces. They can also be renatured to their proper, active shape after nonhydrolytic denaturation.26

Aptamers have been reported to detect targets at concentrations as low as 20 femtomoles in a sample of blood containing large numbers of complex proteins, carbohydrates, lipids, and whole cells.27 Although this is a low detection limit, the committee estimates that the system-level detection limit and time constraints for detect-to-warn applications will require a detection limit on the order of only 300 organisms per 50 microliter sample volume, which corresponds to a target concentration of only tens of attomoles per liter (see Box 7.2). If 1,000 binding targets are expected per organism, and these binding sites can each be detected, then it is possible that a detection limit of 20 femtomoles will provide the sensitivity required for detect-to-warn applications. Aptamer-based sensor systems will need to be tested for actual biothreat detection to determine the detection limits for agents of interest.

Aptamer-target binding can be rapid, on a time scale that is likely faster than that for antibody systems because of the smaller mass of the aptamer. As with antigen-antibody systems, the rate of detection depends upon the concentration of the target, mass transfer of the target to the sensor surface, and the number of processing steps required for detection. Aptamer binding to targets, while normally noncovalent (as is antibody-target binding), can be designed to be covalent and thus more resistant to the harsh washes that are advantageous in minimizing nonspecific binding events. This has been demonstrated by substituting bromodeoxyuridine for uridine in the aptamer sequence.28 Ultraviolet irradiation after target binding creates bromodeoxyuridine-free radicals, which covalently bind with electron-rich tyrosines in the protein target. After binding and washing, the proteins are chemically labeled with dyes for optical detection. While this approach should be useful for decreasing the number of false positives, there are remaining challenges, including detection within 1 to 2 minutes (since there are several processing steps) and developing methods for renewing the sensor surface to allow continuous monitoring.

Peptides

Antibodies are large proteins that bind the target molecules, but only a small fraction of their surface is dedicated to the binding site. This is the case with most proteins that bind small molecules, although it is also true that the binding site is formed by amino acids from the full length of the protein chain. Efforts have been successful in cutting antibody molecules into smaller pieces and isolating and using only the so-called variable regions that are involved in binding. Alternatively, bottom-up approaches involve the design, synthesis, and study of short peptides, looking for those that are long enough to fold and create a specific binding site.

One such approach involves the use of bacteriophage libraries. Through the use of combinatorial synthesis techniques, unique DNA base sequences, each coding for a different 7 to 12 amino acid peptide, can be incorporated into the pIII tail fiber gene of up to 109 to 1011 bacteriophage M13.29 As a result, the peptide sequences to be screened are displayed on the tail fibers of these bacterial viruses. This phage library is then mixed with target—for example, spores from Bacillus anthracis. The phage that do not bind the target are washed away. Those that do bind the target are analyzed to determine the amino acid sequence of the peptides that can bind the target. Similar libraries have also been expressed on the surface of other organisms such as yeast.30

26  

Gold, 2002. See note 25 above.

27  

Gold, 2002. See note 25 above.

28  

H. Petach and L. Gold. 2002. Dimensionality is the issue: Use of photoaptamers in protein microarrays. Curr. Opin. Biotechnol. 13:309-314.

29  

R.H. Hoess. 2001. Protein design and phage display. Chemical Reviews 101:3205-3218.

30  

Y.A. Yeung and K.D. Wittrup. 2002. Quantitative screening of yeast surface-displayed polypeptide libraries by magnetic bead capture. Biotechnology Progress 18:212-220.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

A seven amino acid peptide that binds tightly to Bacillus anthracis has been identified31 and shown to have excellent species specificity, not binding even to closely related Bacillus strains. The peptide has an N-terminal sequence of asparagine, histidine, phenylalanine, leucine, followed by a tripeptide of variable but high proline amino acid content. The peptide has been shown to bind to the SpsC protein of the spore.32 Similar library construction and selection techniques could be employed to identify short peptide molecular recognition elements for other targets or other surface features of B. anthracis.

Small Molecules

The ligand-receptor complex that functions in cellular signal transduction, enzyme catalysis, and other biological systems usually involves a small molecule in which interactions with a large protein or cell surface feature. Specific synthetic small molecules have been designed for years to bind proteins and enzymes, inactivating them in drug discovery or disease therapy applications. It is clear, therefore, that the class of small-molecule molecular recognition elements can be explored in the search for high-affinity, high-specificity elements to bind large molecules or surface features on target organisms. This is being accomplished most effectively through the use of combinatorial libraries, with appropriate selection techniques, as pioneered by several groups.33 Positional scanning libraries of millions of substrates, combined with high-throughput, fluorescence-based assays34 are now widely applied by pharmaceutical companies in drug discovery. Synthesis can be performed on solid supports or by automated methods.35 Small-molecule prospecting libraries cover substantial diversity space and are developed through novel synthetic methods for the rapid generation of scaffolds from which diverse functionality can be displayed.36

Protein Receptors and Other Cell Surface Features

Vertebrates can detect (smell) a wide variety of odors at very low levels and distinguish among many that are very similar. This is achieved via cell surface structure-based sensors—receptors embedded in the cell membrane that bind an odorant molecule and signal the brain for identification and quantitation. In most cases, a given odorant excites a set of receptors, and identification is achieved through the ability of the brain to associate a particular pattern of receptors activated and the extent to which each is activated with the particular target odorant. Fish, for example, have been shown to respond to the presence of certain chemicals at picomolar concentrations,37 and a fish odorant receptor has been isolated and shown to be specific for basic amino acids.38

31  

C. Turnbough, University of Alabama. Presentation to the committee on June 13, 2002.

32  

C. Turnbough, 2002. See note 31 above.

33  

Houdebine, 2002. See note 16 above.

G. Liu and J.A. Ellman. 1995. Combinatorial asymmetric catalyst development: General solid phase synthesis strategy for the preparation of 2-pyrrolidinemethanol ligans. J. Org. Chem. 60:7712-7713.

34  

J.L. Harris, B.J. Backes, F. Leonetti, S. Mahrus, J. Ellman, and C. Craik. 2000. Rapid and general profiling of protease specificity by using combinatorial fluorogenic substrate libraries. Proc. Natl. Acad. Sci. 97:7754-7759.

D.J. Maly, L. Huang, and J.A. Ellman. 2002. Combinatorial strategies for targeting protein families: Application to the proteases. ChemBioChem 3:17-37.

35  

M.R. Spaller, M.T. Burger, M. Fardis, and P.A. Bartlett. 1997. Synthetic strategies in combinatorial chemistry. Current Opinion in Chemical Biology 1:47-53.

Houdebine, 2002. See note 16 above.

36  

Spaller et al., 1997. See note 35 above.

P.A. Bartlett and G.F. Joyce. 1999. Combinatorial chemistry: The search continues—editorial overview. Current Opinion in Chemical Biology 3:253-255.

M.R. Spaller, W.T. Thielemann, P.E. Brennan, and P.A. Bartlett. 2002. Combinatorial synthetic design. Solution and polymer-supported synthesis of heterocycles via intermolecular Aza-Diels-Alder and iminoalcohol cyclizations. J. Comb. Chem. 4:516-522.

37  

A.L. Barth, N.J. Justice, and J. Ngai. 1996. Asynchronous onset of odorant receptor expression in the developing zebrafish olfactory system. Neuron 16:23.

38  

D.J. Speca, D.M. Lin, P.W. Sorenson, E.Y. Isacoff, J. Nagai, and A.H. Dittman. 1999. Functional identification of a goldfish odorant receptor. Neuron 23:487-498.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

Other naturally occurring cell surface features—e.g., glycoproteins, glycolipids, and proteoglycans found embedded in membranes—serve as binding sites for a variety of molecules, many of which are potential biological agents. Cholera and botulinum toxin, for example, attack cells by first binding to membrane gangliosides. These glycolipids, with a specific sequence and arrangement of sugars, extend from the membrane surface and exhibit a high degree of specificity toward their targets. Viruses also are known to bind to cell surface receptors. These various cell surface receptors could thus be used, perhaps in an engineered form, as molecular recognition elements for their natural pathogenic targets. In general, knowledge of the mechanism of action of pathogenic agents can lead to the identification of their cell surface targets and their development as molecular recognition elements.

Imprinted Polymers

Polymerization of certain monomers in the presence of target structures has been shown to create binding sites in the polymer that are specific for that target. High degrees of specificity have been reported for some small molecules,39 although it could be argued that the sophisticated electronic, polar, and nonpolar interactions that normally increase binding constants in molecular recognition systems are likely to be lacking. While imprinted polymers developed to date are known to suffer from severe nonspecific binding problems, they do have increased stability over time and may therefore prove useful in the future as a molecular recognition material. Imprinted polymers are under development for the detection of whole cells,40 but there is not yet convincing evidence that this approach is sensitive or selective enough for pathogen detection applications.

DETAILED CONSIDERATIONS: NOTIONAL DETECTION SYSTEMS

Once the target has bound to the molecular recognition element, that binding must be detected and quantified. A complete review of all potential transduction methods and sensor systems is beyond the scope of this report. Below are some representative structure-based biosensor systems that have been investigated for biothreat detection. Also included within each example are discussions about the current limitations of each of these systems or approaches for detect-to-warn applications.

Immunoassay Tickets

The most common structure-based sensors are immunoassay tickets. Several different types are under development and commercially available for the detection of a wide range of bacterial agents and toxins. These handheld, disposable sensors are analogous to widely used pregnancy test kits and are easy to use. A liquid sample is manually added to the test strip (ticket), and other reagents are added as required. As target molecules in the sample wick through the ticket, they bind to immobilized antibodies and detection molecules in a "sandwich" format. The appearance of a colored pattern on the ticket indicates a positive result. Some examples include the handheld immunochromatographic assays (HHAs), BTATM Test Strips, and the sensitive membrane antigen rapid test (SMART) system.41

While these devices are easy to use, they are disposable and the cost per assay is on the order of $1 or more, making them unsuitable for continuous monitoring. Even if the disposables are minimized and these systems are reconfigured into an automated format for aerosol monitoring, the detection limit for these systems is currently too high for detect-to-warn applications. At least 10,000 targets bound to the sensor surface are required for detection, and the analysis time is 15 minutes or longer. As summarized in Box 7.1, even for a best-case notional detection system with this detection limit, more than 6 minutes will be required to collect enough sample to enable binding of 10,000 targets to a sensing surface. If a

39  

K. Haupt and K. Mosbach. 2000. Molecularly imprinted polymers and their use in biomimetic sensors. Chemical Reviews 100:2495-2504.

40  

Dickert and Hayden, 2002. See note 14 above.

41  

A.A. Fatah, J.A. Barrett, and T.F. Moshier. 2001. Introduction to Biological Agent Detection Equipment for Emergency First Responders, NIJ Guide 101-00. Rockville, Md.: National Institute of Justice.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

lysis step can be designed to release 100 targets per bioagent, then this detection limit might be suitable. In many cases where multiple targets per bioagent are not available, more sensitive detection methods are needed for detect-to-warn applications. As shown in the notional example in Box 7.2, a structure-based sensor with a detection level of 100 cells bound to the sensor surface (or a target concentration of about 10 attomoles in 50 microliters of solution) is required for a detect-to-warn detection system with a response time of less than 2 minutes.

Direct Binding Assays

It is clear that when considering the rapid response requirement of less than 2 minutes for a detect-to-warn system, assays that minimize the number of binding and washing steps are desirable. Although discussion of the notional example in Box 7.2 indicates that multiple rapid binding steps could theoretically be conducted within a total analysis time of 2 minutes, a sensor system will require fewer reagents, have simpler fluidics, and have a shorter response time if only one direct binding step is used for detection. A variety of transduction methods have been proposed that require only a single binding event for detection. Below are several direct binding assays that have been investigated for biodetection and a discussion of further work that is needed to realize structure-based detect-to-warn sensor systems.

Surface Plasmon Resonance

Optical sensors based on changes in evanescent electromagnetic fields at the surface of a thin film of a noble metal such as gold or silver have been widely investigated for direct binding assays. In one configuration, the evanescent field is established by layering a low index of refraction coupling layer between a prism and a high index of refraction resonant layer. These sensors have been used to detect a variety of binding events including analyte-surface binding of small molecules, ligand receptor binding, protein adsorption, antibody-antigen reactions, and DNA and RNA hybridization.42 Surface Plasmon Resonance (SPR) instruments are typically used for measuring binding constants of analytes in known solutions and are commercially available from several vendors (e.g., BIAcore, Texas Instruments, and Spreeta).

Detection is based on SPR spectroscopy, which measures alterations in the optical evanescent waves that result from changes in the refractive index near a surface following a binding event at that surface. The change can be detected by a shift in the angle of incidence to maintain resonance, by a wavelength shift, or by imaging. One challenge in using this approach for biodetection is that any nonspecific binding will contribute to the refractive index near the surface and result in a background signal that will affect the detection limit and potentially result in false positives. Methods are therefore needed to minimize nonspecific binding, enhance the signal from only the analyte, or allow stringent washing to remove nonspecifically bound molecules (and nonspecifically bound analytes).

The approach is sensitive to refractive index changes near the sensor surface, with sensitivity as high as one part in 105 to 106 at the sensor surface, corresponding to a mass sensitivity of 10-12 grams per square millimeter.43 If one considers the binding of protein targets (150,000 molecular weight) onto a 1 square millimeter sensor surface, this detection limit corresponds to the binding of more than 106 targets onto the sensor surface. This detection limit is about two orders magnitude higher than immunoassay

42  

B. Persson, K. Stenhag, P. Nilsson, A. Larsson, M. Uhlen, and P. Nygren. 1997. Analysis of oligonucleotide probe affinities using surface plasmon resonance: A means for mutational scanning. Anal. Biochem. 246:34-44.

D.G. Myszka, M.D. Jonsen, and B.J. Graves. 1998. Equilibrium analysis of high affinity interactions using BIACore. Analytical Biochemistry 265:326-330.

J.M. McDonnell. 2001. Surface plasmon resonance: Towards an understanding of the mechanisms of biological molecular recognition. Current Opinion in Chemical Biology 5:572-577.

M.J. Cannon, D.G. Myszka, J.D. Bagnato, D.H. Alpers, F.G. West, and C.B. Grissom. 2002. Equilibrium and kinetic analyses of the interactions between vitamin B-12 binding proteins and cobalamins by surface plasmon resonance. Analytical Biochemistry 305:1-9.

43  

A.R. Mendelsohn and R. Brent. 1999. Protein biochemistry: Protein interaction methods: Toward an endgame. Science 284:1948-1950.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

tickets and the notional example considered in Box 7.1 and therefore is also not sensitive enough for detect-to-warn applications. While the binding of whole cells onto a sensor surface would result in a larger signal per binding event than the binding of single proteins44 (because this technique is most sensitive to material within 50 nanometers of the surface), most of the volume of the cells would not be detected, and the signal enhancement due to whole cell binding rather than proteins would not provide the 104 improvement in sensitivity required to detect only 100 targets (see Box 7.2).

Response times for SPR sensors have been reported to be between 10-1 and 103 seconds for high concentration samples. Real-time sensing has, however, been reported to be severely mass-transport-limited because of slow diffusion rates, leading to response times on the order of 103 to 104 seconds for analytes at concentrations between 10-6 and 10-7 moles. If diffusion dominates mass transport, the time required for analyte surfaces to reach half saturation coverage scales as the inverse square of bulk concentration. Therefore, mass transport is even more challenging for detect-to-warn applications, which require extremely low detection levels (estimated to be about 10 × 10-18 moles, see Box 7.2). Rapid, sensitive detection will require the development of sensor systems that include methods for minimizing mass transport times.

In addition to the need for enhanced mass transport approaches, the sensitivity of current SPR systems must be improved for detect-to-warn applications. Modifications of SPR systems are under development to increase sensitivity; there are signs that the extremely low detection limits required for detect-to-warn applications might be achievable in the future. Lithographically patterned, nanometer-size triangular silver particles (100 nanometers wide, 50 nanometers high) on a surface exhibit extremely large molar extinction coefficients (3 × 1011 per mole per centimeter), and the localized surface plasmon resonance (LSPR) spectrum is sensitive to nanoparticle-size shape and local (<30 nanometers) external dielectric environment.45 In a model system, biotin was attached to the nanotriangles and exposed to 100 nanomolar streptavidin solution. A 27-nanometer red shift in peak extinction wavelength was observed. The limit of detection was reported to be in the low picomolar to high femtomolar range. While this is still not sensitive enough for detect-to-warn applications (a detection limit of about 10 attomoles is required), the committee projects that optimization will lead to detection of a few molecules in times on the order of seconds. The instrument is simple, small, light, robust, and low cost, so if the performance expectations are realized, this approach has potential for detect-to-warn applications.

Flow Cytometry

Cell biologists have for many years used flow cytometers—devices that identify, count, and sort cells on the basis of preselected properties.46 In a sensor application, target cells are provided with fluorescently labeled molecular recognition elements (e.g., antibodies), and a stream of individual cells is passed through the detector for analysis.47 Since flow cytometry analyzes single cells, the detection of 300 cells per 50 microliters, as described in the notional example in Box 7.2, is routinely achieved using this approach. The time required to analyze a sample volume of 50 microliters using a standard benchtop flow cytometer is between 10 and 60 seconds. Therefore, if specific labeling of the cells can be achieved in about 1 minute or less, then a total analysis time of less than 2 minutes is possible using flow cytometry (this conclusion assumes an aerosol collection time of 4 seconds and sample volume as described in Box 7.2). Direct detection of toxins or viruses is more challenging, since they are often too small for direct analysis using standard flow cytometers and the required reagents for each analysis. One approach for

44  

P.M. Fratamico, T.P. Strobaugh, M.B. Medina, and A.G. Gehring. 1998. Detection of Escherichia coli O157: H7 using a surface plasmon resonance biosensor. Biotechnol. Tech. 12:571-576.

45  

A.J. Haes and R.P. Van Duyne. 2002. A nanoscale optical biosensor: Sensitivity and selectivity of an approach based on the localized surface plasmon resonance spectroscopy of triangular silver nanoparticles. J. Am. Chem. Soc. 124:10596-10604.

46  

T.G.S. Kicker. 1999. Clinical analyzers: Advances in automated cell counting. Anal. Chem. 71:363R-365R.

47  

T. Krupa. 2002. Optical technologies in the fight against bioterrorism. Optics and Photonics News. February 23.

S.A. Sincock, H. Kulaga, M. Cain, P. Anderson, and P.J. Stopa. 1999. Applications of flow cytometry for the detection and characterization of biological aerosols. Field Analytical Chemistry and Technology 3:291-306.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

the detection of these targets, which will be discussed below, is to bind them to microbeads and then analyze the microbeads using flow cytometry.

One advantage of a flow cytometry detection approach is that the analysis is done in solution, and it does not include a fixed sensor surface, which has a limited lifetime due to fouling and degradation. Disadvantages of flow cytometers for detection, however, are their large size, complexity, and high cost. Commercially available flow cytometers are benchtop systems that currently cost more than $30,000 per system, although portable, miniaturized microfluidic flow cytometry systems (e.g., those available at http://www.micronics.net) are being developed. While this research has resulted in some miniature flow cytometry system components,48 completely autonomous systems that are suitable for unattended monitoring are not currently available. When they are, they will be inherently complex, with many required fluidic manipulations and lasers that must be precisely aligned to optically interrogate single particles or cells one at a time. In the near term, flow cytometry is therefore well-suited for detection at centralized locations, but not for distributed use in many locations.

Multiplexed detection based upon direct flow cytometry of cells can potentially be achieved by monitoring variations in the individual properties of cells such as the optical scattering properties, electrical impedance, and fluorescence due to specific labeling with fluorescent dyes. The use of multiple dyes to enable the detection of multiple bioagents in a mixture complicates the optical detection system requirements (e.g., numbers and types of light sources and detectors needed). Therefore, new methods that enable multiplexed biothreat detection using flow cyometry concepts within simple microfluidic systems may be useful for detect-to-warn applications. Flow cytometry concepts of the future may be very different from current flow cytometers and may include transduction methods that are amenable to miniaturization and multiplexed analysis (see Table 7.1).

In addition to the detection of directly labeled cells, flow cytometry has been demonstrated for the multiplexed detection of biothreats using sandwich assays on color-coded beads. In this approach, microspheres carrying antibodies that bind bioagent targets are mixed with a target-containing sample and a second, fluorescently labeled antibody that also binds to the target. The optical properties of the beads themselves are used to code for up to 100 different antibody surfaces for binding specific bioagents. When bioagents and fluorescent dye molecules bind to a bead, it lights up and is measured by the flow cytometer. The flow cytometer is also used to measure the bead color and thereby identify the bioagent. This bead suspension array analysis is therefore analogous to a planar microarray, except that the specific binding elements are monitored by bead type rather than spatial location on a microarray. Benchtop bead suspension array systems are commercially available (e.g., Luminex Corp., Austin, Texas), and a Luminex LX-100 bead suspension array system has been incorporated into a fairly large but field-portable autonomous pathogen detection system (APDS) for deployment at locations where the public is at high risk49 in order to provide detect-to-treat information. Methods are also under development for enabling unattended operation of bead-based assays using flow cytometers.50

Bead suspension arrays were recently demonstrated for the simultaneous detection of four different bioagent simulants.51 The detection limits obtained with a total analysis time of about 1 hour (starting with a liquid sample) were about 5 × 104 cfu per milliliter for Erwinia herbicola, 1.5 × 104 cfu per milliliter for Bacillus globigii, 4.2 × 107 cfu per milliliter for MS2 (an RNA bacteriophage that is a simulant for smallpox virus), and 1 nanogram per milliliter (about 109 targets per milliliter) for ovalbumin (a protein that is a

48  

D.P. Schrum, C.T. Culbertson, S.C. Jaconson, and J.M. Ramsey. 1999. Microchip flow cytometry using electrokinetic focusing. Anal. Chem. 71:4173-4177.

M.A. McClain, C.T. Culbertson, S.C. Jacobson, and J.M. Ramsey. 2001. Flow cytometry of Escherchia coli in microfluidic devices. Anal. Chem. 73:5334-5338.

49  

M.T. McBride, S. Gammon, M. Pitesky, T.W. O'Brien, T. Smith, J. Aldrich, R.G. Langlois, B. Colston, and K.S. Venkateswaran. 2003. Multiplexed liquid arrays for simultaneous detection of simulants of biological warfare agents. Anal. Chem. 75:1924-1930.

50  

J.W. Grate, C.J. Bruckner-Lea, A. Jarrell, and D. Chandler. 2003. Automated sample preparation methods for suspension arrays using renewable surface separations with multiplexed flow cytometry fluorescence detection. Analytica Chimica Acta 478(1):85-98.

51  

McBride et al., 2003. See note 49 above.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

simulant for protein toxins). In all cases, a 100 microliter sample was used for each analysis. These detection limits were comparable to the gold standard enzyme-linked immunoassays (ELISA) that were conducted in parallel. Shorter analysis times of 30 minutes were reported to provide reasonable results, but they compromised the detection limit.

The number of cells required for each analysis above at a detection limit of 1,000 to 5,000 cfu in 100 microliters is about 10-fold higher than the number of cells considered in the notional example in Box 7.2. However, a 10-fold increase in aerosol sampling time could be used to increase the number of organisms collected, which would therefore increase the aerosol collection time from 4 seconds to 40 seconds in that example and still leave some time for analysis within the 2-minute time frame. Therefore, the detection limits for cells or spores using current bead suspension array systems appear to be compatible with detect-to-warn applications, and the detection limit for the toxin stimulant, 1 nanogram per milliliter, is comparable to a lethal dose for many toxins. However, the detection limit for the virus simulant is several orders of magnitude higher than needed for detect-to-warn applications (see Box 7.2), and in all cases the total analysis time to achieve the reported detection limits was about 1 hour. Therefore, novel approaches are needed to significantly decrease the time required for detection. Given that it is known that antibody-antigen binding kinetics are rapid when the binding partners are close to one another (see the section "Rapid Detection" in this chapter), methods that have the potential of decreasing the time for mass transport of the targets and detection tags to the bead or cell surfaces are needed.

Some challenges for detect-to-warn applications using flow cytometry are, therefore, (1) to dramatically decrease the time required for each of the binding and washing steps by enhancing mass transport to the cells or microspheres; (2) to decrease analysis time by minimizing the number of binding and washing steps; (3) to develop new flow cytometry-type approaches that are low cost (in terms of both instrumentation and consumables), amenable to miniaturization and field use, and suitable for multiplexed detection; and (4) to develop approaches to improve detection limits, especially for toxins and viruses.

Target Binding That Changes Detectable Properties of Smart Sensor Surfaces

Another general type of sensor system that is attractive for rapid detection, as required for detect-to-warn applications, is a sensor that is designed so that its surface changes in a detectable way only upon specific binding of the target to the sensor surface. This type of smart sensor surface could have advantages, including rapid one-step binding and detection, minimal false positives, sensitive detection if there is a built-in amplification scheme, and the potential for reversible (continuous) operation if the bound targets can be removed from the sensor.

Colorimetric Detection

One example of this general approach is the use of poly(diacetylene) liposomes that are engineered to contain bioagent binding sites52 for the colorimetric detection of bioagents. The liposomes are optically monitored in solution or after attachment to a planar surface. Binding of targets disrupts the membrane and causes a change in color from blue to red, which can be detected by eye or with a dedicated spectrometer. The device integrates over time, since each new binding event sequentially contributes to the color change. While this one-step approach is rapid (with a response time less than 2 minutes) and simple, the detection limit for cholera toxin was found to be 20 micrograms per milliliter (230 nanomoles). Dramatic improvements in detection limit are therefore required for detect-to-warn applications.

52  

A. Reichert, J.O. Nagy, W. Spevak, and D. Charych. 1995. Polydiacetylene liposomes functionalized with sialic-acid bind and colorimetrically detect influenza virus. Journal of the American Chemical Society 117(2):829-830.

J.J. Pan and D. Charych. 1997. Molecular recognition and colorimetric detection of cholera toxin by polymerized lipsomes. Langmuir 13:1365-1367.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

Fluorescence Detection

Smart sensor surfaces that exhibit a fluorescence change upon specific binding of targets are also under development. For example, several groups are developing one-step detection approaches that involve fluorescently labeled aptamers that change conformation upon target binding. The sensing surface is engineered so that the conformational change in the aptamer results in a change in fluorescence intensity and/or fluorescence spectra of the sensor surface.53 In one model biosensor system tested for thrombin detection, the detection limit was 5 nanomoles (0.7 attomole thrombin in 140 picoliters) and the analysis time was 10 minutes.54 Given that single molecule fluorescence detection has been demonstrated in benchtop fluorescence detection systems with properly engineered smart sensor surfaces, it is conceivable that approaches based upon fluorescence detection will be able to detect 100 bound molecules (notional example in Box 7.2) in the near future.

One challenge is the development of fluorescent optical tags that are stable over time. Some promising optical tags include fluorescent nanocrystals (quantum dots) and green fluorescent protein, which includes a fluorosphere that is protected within the interior of a protein. Other challenges in fluorescence detection include the development of low-cost, portable optical detection systems and their components, including optical fibers and waveguides, and the development of engineered smart sensor surfaces that can reliably result in a measurable signal upon binding of fewer than 100 targets.

One-Step Signal Amplification Concepts

Smart sensor surfaces that include an engineered mechanism for one-step signal amplification would also be valuable to enable rapid, sensitive detection. For example, conducting polymers have been developed that carry multiple fluorescence groups, all of which can be quenched through the binding of a single triggering molecule.55 This approach is reported to have the potential to amplify the detection signal up to a millionfold.56 A biodetection amplification scheme based upon the conductance switching of ion channels upon target binding has also been demonstrated.57

While the value of such engineered amplification schemes has been demonstrated, challenges remain, including the development of robust smart sensor surfaces that include engineered amplification schemes to enable sensitive, rapid biothreat detection.

53  

R.A. Potyrailo, R.C. Conrad, A.D. Ellington, and G.M. Heiftje. 1998. Adapting selected nucleic acid ligands (aptamers) to biosensors. Anal. Chem. 70:3419-3425.

S. Jhaveri, M. Rajendran, and A.D. Ellington. 2000. In vitro selection of signaling aptamers. Nature Biotechnology 18:1293-1297.

S.D. Jhaveri, R. Kirby, R. Conrad, E. Maglott, M. Boswer, R.T. Kennedy, G. Glick, and A.D. Ellington. 2000. Designed signaling aptamers that transduce molecular recognition to changes in fluorescence intensity. J. Amer. Chem. Soc. 122:2469-2473.

C. Frauendorf and A. Jaschke. 2001. Detection of small organic analytes by fluorescing molecular switches. Bioorg. Med. Chem. 9:2521-2524.

N.A. Hamaguchi, A. Ellington, and M. Stanton. 2001. Aptamer beacons for the direct detection of proteins. Analytical Biochemistry 294:126-131.

J.W.J. Li, X.H. Fang, and W. Tan. 2002. Molecular aptamer beacons for real-time protein recognition. Biochem. Biophys. Res. Commun. 292:31-40.

J. Perlette, J.W. Li, X.H. Fang, S. Schuster, J. Lou, and W.H. Tan. 2002. Novel DNA probes for detection and quantification of protein molecules. Rev. Anal. Chem. 21(1):1-14.

54  

Potyrailo et al., 2000. See note 53 above.

55  

L.H. Chen, D.W. McBranch, H.L. Wang, R. Hegelson, F. Wudl, and D.G. Whitten. 1999. Highly sensitive biological and chemical sensors based on reversible fluorescence quenching in a conjugated polymer. Proc. Natl. Acad. Sci. 96:12287-12292.

S.A. Kushon, K.D. Ley, K. Bradford, R.M. Jones, D. McBranch, and D. Whitten. 2002. Detection of DNA hybridization via fluorescent polymer superquenching. Langmuir 18:7245-7249.

56  

Chen et al., 1999. See note 55 above.

57  

B.A. Cornell, V.L.B. Braach-Maksvytis, L.G. King, P.D.J. Osman, B. Raguse, L. Wieczorek, and R.J. Pace. 1997. A biosensor that uses ion-channel switches. Nature 387:580-583.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

Modified Cell-Based Systems

Modified cell-based systems that include direct binding assays are also under development. One advantage of cell-based systems is that they can already contain a built-in signal amplification system. One example is the cellular analysis and notification of antigen risks and yields (CANARY) B cell detector.58 B cells are components of the vertebrate immune system that display antibodies on their surface against virtually any foreign structure previously encountered by the organism. These cells can be cloned and engineered so that an entire population expresses a single antibody that is specific for the target of interest. Target binding of the surface antibodies is thought to lead to dimerization of the antibodies, which results in the release of Ca++ ions. In the CANARY sensor, B cells are engineered to express antibodies against a defined target and also the luminescent protein aequorin from the jellyfish Aequorea victoria. Aequorin responds to the Ca++ released upon target binding by emitting blue-green light. The cell response therefore both transduces and amplifies the molecular recognition event.

This relatively complex transduction method is quite rapid: The engineered B cells emit more than 2,000 photons within 30 seconds after bioagent binding.59 For the purpose of biosensing, mixtures of the B cells are briefly centrifuged with samples in order to rapidly bring the target and sensor surface in close proximity to one another for binding. Detection limits are on the order of 50 target cells.

The CANARY system has been successfully tested using dry aerosol collection, in which aerosol is directly impacted into a small sample tube. After aerosol collection, the B cell solution is added, the sample is centrifuged (5 seconds), and the optical response is detected (30 seconds). One advantage of this approach is that it is a one-step method, with no additional washing or reagent addition steps. Given the low detection limit and rapid response of this approach, detection of 10 to 100 ACPLAs should be achievable in seconds (see notional example in Box 7.2), and a total analysis time within 2 minutes for detect-to-warn applications should be possible. Some testing with aerosol backgrounds has been done, and no detrimental effect on the optical signal was seen in samples containing 500 cfu Y. pestis, with and without the addition of an aerosol sample from a parking garage.

The system could also be multiplexed to detect a number of targets. B cell lines are now available for the detection of Bacillus anthracis, and orthopox viruses including smallpox, Yersinia pestis (plague), Francisella tularensis (tularemia), Vibrio cholerae (cholera), Brucella spp. (brucellosis), the foot and mouth disease virus, and the Venezuelan equine encephalitis virus. Others are in development. While this approach works well for the detection of cells and large viruses, the detection of small viruses is less sensitive because they do not sediment during standard centrifugation. Toxin detection using this approach is conceptually possible but has not been demonstrated yet (at least two binding sites on the target are required for the dimerization process to occur on the surface of the B cells).

One challenge that will arise in implementing a continuous detect-to-warn system is the incidence of false positives. The CANARY system is reported to have a low false positive rate of 0.4 percent over 1,288 laboratory tests and similar false positive rates of 0.3 percent to 0.6 percent for tests using actual indoor and outdoor aerosols (determined from more than 300 samples in each case).60 Although these rates are low, if samples are analyzed every 2 minutes, this will result in false positives on average every 250 samples (0.4 percent), or 500 minutes (8.3 hours). The occurrence of successive positive samples may be one way to increase confidence that an actual positive has occurred, but this will decrease response time to the time required for the analysis of several samples in sequence. The sources and mechanisms of the false positives are not well understood. For detect-to-warn applications that require near-continuous monitoring, research is therefore needed to develop approaches to minimize false positives, by both improving individual detection systems and combining information from orthogonal sensors.

58  

J.D. Harper, MIT Lincoln Laboratory. Presentation to the committee on June 13, 2002.

59  

Harper. See note 58 above.

60  

Harper. See note 58 above.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

A portable CANARY system that includes integrated aerosol collection and detection is being assembled, with a goal of detecting 100 ACPLA in 1.7 minutes. Some practical challenges of this type of approach are: The need to maintain live cells, the cost of each assay, and the disposables that will be generated for each assay. The cells survive 2 to 3 weeks when refrigerated,61 but extended use in the field will require the development of methods for long-term storage and maintenance of the cells. The cost of this analysis is estimated to be 25 cents per assay.62 Although this is considered to be inexpensive for a liquid-based biodetection assay, continuous monitoring every 2 minutes would result in a system operation cost of $180 per day ($65,700 per year).

The cost of biodetection assays in general is a significant consideration for continuous monitoring in detect-to-warn situations. Another practical issue is the potential accumulation of disposables during continuous operation. If one disposable sample tube is used per analysis, the system would result in the accumulation of 720 tubes each day and 262,800 tubes per year. This highlights another general challenge for all detect-to-warn systems that include liquid-based analysis: Novel chemistry and engineering solutions will be required to dramatically decrease disposables.

Waveguides and Fluorescent Detection

Many bioagent detection systems under development include the use of optical waveguides for collecting light from fluorescent molecular recognition element reporter groups. An optical waveguide system in a planar microarray format has recently been demonstrated for the simultaneous detection of multiple bacterial, viral, and protein analytes.63 The detection limit of this system, developed at the Naval Research Laboratory, is similar to that of ELISA assays (e.g., 6 × 104 cfu per milliliter for Bacillus globigii and 1 to 10 nanograms per milliliter for various protein toxins).64 The sample volumes analyzed are typically about 600 microliters, and the shortest reported analysis time is 15 minutes. The analysis was done using a sandwich assay that includes the following steps: sample (antigen) binding, washing, binding of fluorescently labeled secondary antibody, and washing.

Although an order of magnitude improvement in the detection limit for cells would be desirable for detect-to-warn applications, the 1 × 104 to 10 × 104 cfu per milliliter detection limits currently achieved using this approach may be suitable if aerosol collection times are approximately 1 minute (rather than 4 seconds, as listed in the notional example in Box 7.2) and the liquid volumes are scaled up to handle the quantities required for the assay. The major shortfall of this approach for detect-to-warn applications, however, is the analysis time, which is 1015 minutes in the best case. Therefore, all of the methods previously discussed to decrease analysis time (i.e., decrease volumes and dimensions, add active mass transport methods, develop one-step assay formats) are needed here. In addition, continuous monitoring will present engineering and chemistry challenges with regard to the need for the repeated renewal and reuse of the sensor, the accumulation of disposables, and the cost per assay.

Alternative approaches to two-step sandwich assays are one-step displacement assays or one-step competitive binding assays. For small molecules, the measured detection limits for sandwich, displacement, and competitive assays are typically different but still similar to one another in magnitude (e.g., ranging from 5 to 20 nanograms per milliliter).65 However, for the detection of biothreats such as cells or spores, a sandwich assay can be at least one or two orders of magnitude more sensitive than the

61  

Harper. See note 58 above.

62  

Harper. See note 58 above.

63  

J.B. Delehanty and F.S. Ligler. 2002. A microarry immunoassay for simultaneous detection of proteins and bacteria. Anal. Chem. 74:5681-5687.

K.E. Sapsford, P.T. Charles, C.H. Patterson Jr., and F.S. Ligler. 2002. Demonstrations of four immunoassay formats using the array biosensor. Anal. Chem. 74:1061-1068.

C.R. Taitt, G.P. Anderson, B.M. Lingerfelt, M.J. Feldstein, and F.S. Ligler. 2002. Nine-analyte detection using an array-based biosensor. Anal. Chem. 74:6114-6120.

64  

Delehanty and Ligler, 2002. See note 63 above.

65  

Sapsford, 2002. See note 63 above.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

other approaches if multiple-labeled secondary antibodies can bind to the target. Therefore, new one-step binding approaches that enable sensitive detection are needed for rapid biodetection.

One recently reported improvement in the detection limit of fluorescent waveguide sensors is the use of a capillary waveguide in which the sample solution itself is the waveguide. In this case the fluorescence signal is integrated over a larger surface area without increasing the background noise. Initial experiments indicate that this capillary waveguide format (used in combination with a sandwich assay) can improve the detection limit by about two orders of magnitude over the planar waveguide format.66 The logistics of using and reusing a capillary for continuous operation in a biothreat detection system, however, may be a challenge when compared with other sensor system formats.

FINDINGS AND RECOMMENDATIONS

Finding 7-1: Theoretical considerations and experimental results support the view that structure-based sensors can be designed to respond within the 2-minute time period that is of interest for this study. The challenges for rapid detection for such detect-to-warn applications include accelerating the transport of the target to the molecular recognition element, decreasing the number of processing steps to speed detection, and improving detection sensitivity (detection of as few as 100 targets bound to the sensor surface is needed).


Recommendation 7-1: Conduct research on practical ways of accelerating mass transport to the molecular recognition sites and developing sensing platforms that reduce the number of processing steps. Conduct research on approaches to enable the required level of detection for detect-to-warn applications.


Finding 7-2: Impressive progress has been made in designing and synthesizing molecular recognition elements. A variety of classes of these molecules has been identified, many of which exhibit attractive properties, including ease of synthesis, specificity, stability, and affinity. However, few have been tested under real-world environmental operating conditions, and false positive rates must be minimized for detect-to-warn applications.


Recommendation 7-2: Research across a broad front of molecular recognition elements should be pursued. The focus of research should be on development of a greater understanding of the structural and binding characteristics of these materials and of practical ways to reduce the false positive rates for these sensors and on testing the sensor performance (sensitivity, selectivity, and false alarm rate) under the anticipated environmental operating conditions (with appropriate background organisms and aerosols).


Finding 7-3: Impressive progress has been made in the development of reporter tags and instrumentation capable of detecting them.


Recommendation 7-3: Research across a broad front of reporter tags and detection schemes should be pursued. The focus should be on simplicity of the system (absence of a wash step, for example) and simplicity of instrumentation to allow wide distribution of these systems for detect-to-warn applications.

66  

F.S. Ligler, M. Breimer, J.P. Golden, D.A. Nivens, J.P. Dodson, T.M. Green, D.P. Haders, and O.A. Sadik. 2002. Integrating waveguide biosensor. Anal. Chem. 74:713-719.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×

Finding 7-4: Research described above is being pursued in a very large number of university, government, and industrial laboratories. Unfortunately, it is often difficult to compare results and to evaluate progress because a variety of measurement units are used to report (or not report) results.


Recommendation 7-4: A standardized reporting scheme for sensor description needs to be developed. It should be based on performance requirements such as affinity, specificity, speed, false positive rate, cost, manufacturability, and other criteria.

Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 105
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 106
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 107
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 108
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 109
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 110
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 111
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 112
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 113
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 114
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 115
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 116
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 117
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 118
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 119
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 120
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 121
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 122
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 123
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 124
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 125
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 126
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 127
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 128
Suggested Citation:"7 Structure-Based Identification for Detect-to-Warn Applications." National Research Council. 2005. Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases. Washington, DC: The National Academies Press. doi: 10.17226/11207.
×
Page 129
Next: 8 Chemistry-Based Identification for Detect-to-Warn Applications »
Sensor Systems for Biological Agent Attacks: Protecting Buildings and Military Bases Get This Book
×
Buy Paperback | $68.00 Buy Ebook | $54.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

Over the last ten years, there has been growing concern about potential biological attacks on the nation’s population and its military facilities. It is now possible to detect such attacks quickly enough to permit treatment of potential victims prior to the onset of symptoms. The capability to “detect to warn”, that is in time to take action to minimize human exposure, however, is still lacking. To help achieve such a capability, the Defense Threat Reduction Agency (DTRA) asked the National Research Council (NRC) to assess the development path for “detect to warn” sensors systems. This report presents the results of this assessment including analysis of scenarios for protecting facilities, sensor requirements, and detection technologies and systems. Findings and recommendations are provided for the most probable path to achieve a detect-to-warn capability and potential technological breakthroughs that could accelerate its attainment.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!