Analytica Chimica Acta 559 (2006) 137151

Review

Biosensor technology for detecting biological warfare agents: Recent progress and future trends
J. Justin Gooding
School of Chemistry, The University of New South Wales, Sydney 2052, Australia Received 2 August 2005; received in revised form 30 November 2005; accepted 5 December 2005 Available online 24 January 2006

Abstract This review paper outlines the important issues with regards to the development of biosensors for the monitoring of biological warfare agents (BWAs) starting with the basic components of biosensors and the features of BWAs, which are compatible with detection using biological recognition molecules. The advantages and limitations of biosensors are discussed followed by the current state of the art in biosensors for detecting BWAs. Finally the developments required to enable biosensors to become more effective at providing early warning of possible biological attack and advances towards these developments are covered. 2005 Elsevier B.V. All rights reserved.
Keywords: Biosensors; Biological agents; Afnity sensors; Bioweapon detection; Detect-to-protect

Contents
1. 2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Biosensors for detecting biological agents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Nucleic acid-based biosensing technologiesdetect-to-treat biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Detection using surface featuresdetect-to-protect biosensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cutting edge of biosensor technology which may be applicable to bioweapons detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Air sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Reducing the detection limit and providing a more rapid response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Decreasing the response timethe needle in a haystack problem. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 139 139 141 146 147 147 148 149 149 149

3.

4.

1. Introduction With the increased threat of terrorist attacks, an attack using a BWA is a conceivable event. We have already seen evidence of the destructiveness of such attacks, and the associated fear they create, through episodes such as the Sarin gas and anthrax attacks by the Aum Shinrikyo cult in Tokyo in 1995, the use of Salmonella in Oregon restaurants in 1984 by the Rajneeshee cult and the anthrax letter attacks soon after September 11th.

Tel.: +61 2 9385 5384; fax: +61 2 9385 6141. E-mail address: Justin.gooding@unsw.edu.au.

One of the problems with biological attacks is actually determining whether an attack has occurred. The difculty arises because often the initial symptoms after infection from BWAs are difcult to distinguish from symptoms from infections from more benign biological agents. The solution to this detection problem is to employ molecular techniques, which can identify chemical markers from known biological agents. Two classes of such devices can be envisaged. These classes of devices can be referred to as detect-to-treat, thus allowing early identication of an infection, and detect-to-protect to provide a warning that a site is contaminated by BWAs to prevent infection [1]. The criteria for such analytical methods are quite different. A detect-

0003-2670/$ see front matter 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.aca.2005.12.020

138

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151

to-treat system must be able to identify a BWA, from a biological sample, within a few hours of infection. Hence detect-to-treat biosensors are compatible with an analytical laboratory using personnel trained to perform the analysis. A detect-to-protect system must be able to provide a warning within a couple of minutes from, most frequently, an airborne sample without user intervention. Analytical devices, which could satisfy both sets of criteria, and are possibly the only answer to the detect-to-protect criteria, are biosensors. A biosensor comprises a biorecognition molecule immobilised over a signal transducer to give a reagentless analytical device. The biorecognition molecule, such as an enzyme, antibody, sequence of DNA, peptide or even a microorganism, provides the biosensor with its selectivity for the target analyte so that the molecule of interest can be picked out by the biosensor from a matrix of many other molecules. The signal transducer determines the extent of the biorecognition event and converts it into an electronic signal, which can be outputted to the end user. Common transducers include amperometric electrodes, optical waveguides or mass sensitive piezoelectric crystals. Biosensors can be subdivided into two classes based on the type of biorecognition molecule. Catalytic biosensors employ enzymes and microorganisms as the biorecognition molecule which catalyses a reaction involving the analyte to give a product. Common analytes for catalytic biosensors are small organic molecules like glucose. Although some BWAs are small molecules, such as aatoxin, most frequently they are large macromolecules and microorganisms, which cannot usually be detected using catalytic biorecognition molecules. The other category of biosensors is afnity biosensors. Biorecognition molecules commonly used in afnity biosensors include antibodies, DNA, peptides and lectins. Afnity biosensors are characterised by a binding event between the biorecognition molecule and the analyte (the afnity reaction) often with no further reaction occurring. Hence the challenge then becomes transducing the biorecognition event. As this class of biosensor is compatible with the detection of virtually all biological agents it is this challenge that faces researchers attempting to develop portable devices for detecting toxins, microbes and viruses. Transduction of afnity biosensors has been achieved using labelled species and using label free approaches. If transduction is achieved using labelled species the principles are very similar to immunoassay with the amount of analyte detected being inferred from the amount of label which binds to the interface. The most common transducers for detecting labelled species are optical, where an optically active label is detected [24], or electrochemical, where the label is electroactive [3,5]. Label-free methods most frequently involve evanescent wave based optical methods [68] or using mass sensitive acoustic wave devices [9] which monitor molecules binding to, or desorbing from, a transducer surface. Any change in species adsorbed to this surface will give a response. In the case of both label and label free methods one of the key factors, which limits biosensor performance is non-specic binding. The problem of non-specic binding highlights the importance of interfacial design in a biosensor [10,11].

This review will outline the developments in biosensors for detecting biological warfare agents before discussing what problems require solutions for biosensors for detecting BWAs to become common. The emphasis will be on the development towards handheld analytical devices, which provide warning of BWA attacks rather than laboratory instruments where impressive advances have been made with automated PCR and mass spectrometry instruments [12]. The analytes, anthrax, tularemia, botulinum toxin, plague (aerosol version only), smallpox and hemorrhagic fever will be treated as indicative analytes of a wider class of potential bioterrorist agents. They are all category A agents according to the Center for Disease Control and Prevention (CDC) in the United States [13]. They cover the main types of biological agents from Gram-positive bacteria, which form spores (Bacillus anthracis, Anthrax; Yersinia pestis, Pneumonic plague), gram-negative (Francisella tularensis, tularemia), toxins derived from bacterial species (botulinum toxin from Clostridium botulinum) and viruses (Smallpox). For most of these analytes the current preferred methods of detection are classical taxonomic tests. In the case of the microorganism derived biological agents, microbiological tests where different growth media, different shapes and responsiveness to different stains are used to identify the species [14]. Such tests typically take about 48 h and therefore are not applicable to detect-toprotect technologies and may take too long for some detect-totreat applications. For the viral species typically scrapings from lesions are examined using microscopic techniques for any tell tale taxonomy before samples are analysed by nucleic acid techniques [13]. To develop biosensors for these analytes the important considerations are the matrix in which the analyte will be found (for example, air, water, food, person), the form the analyte will be in (spore, bacteria, virus, toxin) and the possible features of the analyte, which could be recognised with an appropriate biorecognition molecule. Identication of such features requires some knowledge of the biological species and the forms in which they are found. In the case of the bacterial species and viruses, nucleic acid techniques can be used to identify the organisms or afnity molecule can be used to detect surface sites on these organisms. An alternative for bacterial species is to monitor the toxins they release. Looking at the analytes gram-positive bacteria, gram-negative bacteria, toxins and viruses in turn will provide some insight into the sorts of recognition molecules that can be used as the basis of a biosensor. (1) Gram-positive bacteria: The organism causing anthrax is an example of such bacteria and is one of the very few pathogenic organisms, which is both aerobic and spore forming [15]. The bioterrorist threat from anthrax is most frequently as a spore, which are very resistant to degradation [13]. Upon inhalation of spores, in 75% of cases infection is lethal with the number of spores required being between about 10000 and 1000. The opportunities exist to detect either the spores or the toxins (endema toxin and lethal toxin [13]) after infection. Detecting the spores would be required if a biosensor was to be a detect-to-protect system for the presence of anthrax spores in the atmosphere whilst

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151

139

the presence of the toxins would be used for a detect-to-treat device. The problem with detecting the toxins after release is that approaches such as immunoassays are too insensitive to detect toxins in infected patients unless the levels are so high that death will be the result [16]. To detect anthrax spores there are essentially three options. Firstly to detect dipicolinic acid (DPA) which is a major constituent of all spores [17,18]. The drawback of detecting DPA is it is nonspecic, it only informs that bacterial spores are present [17]. The second option is to detect the DNA from a spore. Using polymerase chain reaction to amplify the DNA as little as one spore has been detected [16]. The third option is to detect surface features (epitopes) on the spores, which are specic to anthrax spores using afnity molecules such as antibodies. For example, the surface of spores have glycoproteins present, which differ between species and hence can be used as the epitopes to which antibodies are raised [15]. (2) Gram-negative bacteria: Francisella tularensis is an example of such bacteria which are non-spore forming. Tularemia can cause infection by inhalation of contaminated dust, contact with the mucous membrane or from insect bites [14]. As little as 25 organisms can cause infection [19]. Immunoand nucleic acid markers are the most likely recognition molecules. (3) Toxins: The Botulinum neurotoxins (BoNT) and cholera toxin (CT) are perhaps the two most important analytes of this type of biological agent. With botulism the infection can come from the bacteria Clostridium botulinum. Both the spores or the toxin can be aerosolised and aerosolised is the most likely weaponised form [20]. Similar symptoms are observed if BoNT is ingested or inhaled. The toxins are zinc endopeptidases of 150,000 Da, which disrupt nervous function and are the most toxic compounds known [13]. Being proteins antibodies can be raised to them as the basis of an immunosensor [20] or nucleic acid techniques could be used to identify the microorganism. An alternative biorecognition molecule, which has found use for both these toxins are lipids with carbohydrate heads known as gangliosides. Both CT and BoNT penetrate cell walls by binding at gangliosides on the outer cell membrane. (4) Viruses: Smallpox is a brick shaped DNA virus (variola major) 200 nm in diameter, which is aerosol infective and has approximately 30% mortality rate whilst hemorrhagic fever is an RNA virus. Most viruses can simplistically be thought of as a bundle of nucleic acids in a protein or glycoprotein bag. The detection options are to identify nucleic acid markers or to use afnity ligands raised to the protein coat of the virus. 2. Biosensors for detecting biological agents The majority of existing technologies for detecting BWAs such as those listed above have relied on either antibodies as the recognition molecule, which bind with a surface feature of the BWA, or have used nucleic acids and determined a nucleic acid sequence known to be found in the BWA being tested. A few

other types of recognition molecules such as peptides, glycolipids and aptamers have also been used [2,13,15,21] with similar transduction schemes. The most commonly used biosensing methodologies for detecting BWAs will be described in detail. 2.1. Nucleic acid-based biosensing technologiesdetect-to-treat biosensors For the detection and identication of viruses and bacteria, but not toxins, nucleic acid techniques are exceedingly attractive because they can provide unambiguous identication a characteristic sequence, are highly specic and because using polymerase chain reaction (PCR) amplication techniques the detection of femto or even attograms of nucleic acids is not uncommon [22]. The exquisite sensitivity of advanced nucleic acid techniques are demonstrated by the work of Hartley and Baeumner [16] who have demonstrated the detection of DNA from a single anthrax spore. Similarly, Versage et al. [23] have detected a single Francisella tularensis organism using real time PCR. Analyses can be performed relatively quickly when compared with microbiological techniques, but typically still take several hours. The speed of the assays have been reduced to an hour using real time PCR where the amplication reaction is followed as it progresses [24]. Furthermore, with the use of multiple target sequences exquisite selectivity as well as sensitivity can be achieved. For example, a single Francisella tularensis bacteria was differentiated from not only non-Francisella bacteria but from Francisella philomiragia [23]. Nucleic acid detection still suffers from a number of drawbacks. Firstly, the ability to amplify minute quantities of DNA means that contamination can be a big problem with the contaminant also being amplied. As a consequence, nucleic acid techniques require pure un-degraded nucleic acids, which implies signicant sample preparation and hence is contrary to the entire concept of a biosensor being a reagentless system. The sample preparation steps required to perform a nucleic acid analysis in the simplest case require: (1) The release of the DNA from the target organism in sufcient quantity to allow amplication. In essence this means the removal of the DNA from the nucleoid, which means lysing the microorganism. (2) The removal of PCR inhibitory contaminants such as RNases or DNase enzymes and other proteins. This step may be inuenced by many factors including the cell type, the stage of the cell cycle whether it is a cell or spore and the sample matrix. (3) The type of nucleic acid being amplied as RNA is far less stable than DNA. Despite these potential problems nucleic acid techniques, which employ PCR have been used for the detection of anthrax spores [16,25], Francisellas tularensis [19,23] and orthopox viruses [26,27]. Furthermore, nucleic acid detection methods employing PCR for the identication of disease agents have been shown to be reliably performed under eld conditions by enlisted personnel [22]. Recent developments in miniaturisation of PCR

140

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151

Fig. 1. The principle of the biosensor described by Baeumner et al. [30]. The biosensor consists of sulforhodamine B dye-entrapped liposomes, which contain covalently attached probe DNA on their surface and capture probes immobilised onto the surface of polyethersulfone membranes. To perform the assay the liposomes are mixed with the sample containing the potential target DNA or RNA sequence. Placing the membrane into the sample results in mixture migrating up the membrane. Capture of the target species by hybridisation, results in the liposomes also being captured and the capture zone will uoresce. Reproduced with permission from [30], Copyright Springer (2004).

thermocyclers has seen hand held PCR thermocyclers close to commercialisation [19,28], which perform well in comparison to established laboratory based technology when analysing for Francisella tularensis [19]. The inuence of toxins on gene expression monitored using a gene chip has also been proposed as an approach to detecting the presence of cholera toxin [29]. This paradigm shift in the use of DNA microarrays for detecting biological agents relies on the knowledge of gene markers, which respond when a cell is exposed to a toxin. In the case of cholera toxin cell lines, which contained ganglioside GM1, the glycolipid responsible for transfer of the cholera toxin A subunit across the cell membrane, were used. The power of the approach is that if a variety of genetic markers are known then identication of a previously unknown toxin is possible. Nucleic acid techniques can also detect articially created pathogens or unknown pathogens by detecting DNA sequences unchanged from the original organism [22]. Afnity molecules, which bind to surface sites on the known species will not possess this ability if the surface epitope used for recognition is changed by the genetic modication. The strengths and drawbacks of nucleic acid based biosensors for the detection of BWAs is exemplied by a recent paper by Bauemner et al. [30] for the detection of mRNA from germinated spores. The detection is achieved using a sandwich assay where oligonucleotides specic for a sequence indicative of anthrax are immobilised onto a polyethersulfone membrane (the capture probe), see Fig. 1. A reporter probe is also synthesized which is complementary to a different part of the target sequence. The reporter probes are conjugated with a liposome containing the uorescent dye sulforhodamine B. Placing the membrane containing the capture probe into a solution of the amplied target and the reporter probes results in a sandwich

being formed and that part of the membrane emitting light. The advantage of the biosensor is its capability of detecting as little as a femtomole of target in 4 h with no cross-reactivity with 11 other similar microorganisms tested. Furthermore, the detection of mRNA, being the nucleic acid, which is the active component of genes, indicates that the spores are viable and no cross-reactivity is observed from nonviable spores. Greater sensitivity could be achieved with longer sample preparation time down to the level of detecting mRNA from a single spore. These clear advantages are counterbalanced by the 4 h to perform the entire assay, although the actual biosensing part of the assay takes only 15 min. The rest of the assay time is involved in geminating the spores, extracting the mRNA and nally amplifying the mRNA. These sample preparation steps require considerable skill and care to perform, especially when handling RNA, which is about 100,000 times less stable than DNA. Integration of automated sample preparation with DNA amplication and detection has been achieved to produce benchtop size instruments [12] such as the Biohazard Detection System used by the US Postal service and developed in conjunction with Northrop Grumman and the US Army Soldier and Biological Chemical Command. For truly portable hand held devices, which require the user to simply contact the sample with the biosensor the entire detection procedure must be automated (handheld thermocyclers still require the preparation of the DNA sample). There are essentially three steps to this process. Firstly, the preparation of the nucleic acids for amplication, secondly the amplication and labelling of the nucleic acids if required and nally the detection, typically using surface immobilised probe sequences. The last step of this process has been the subject of considerable research effort with some impressive technologies being developed using both electrochemical [31,32] and optical transduction [33,34]. Researchers have focussed on better surface chemistry for DNA immobilisation [3538], new types of labels [3941], novel transduction methods [4249] and miniaturisation [50]. These studies have made signicant advances with regards to some known stumbling blocks to the widespread use of DNA biosensors such as sensitivity [43,51,52] and the ability to differentiate between a perfectly complementary sequence and a sequence containing a single base-pair mismatch [53,54]. Advances in performing PCR on a chip have been briey mentioned above with regards to handheld PCR thermocyclers [5557]. There has been considerably less research on sample pre-treatment, which proves to be the stumbling block for the in-eld use DNA biosensors, although some impressive advances have been made on microuidic platforms [58]. For example, Nanogen has shown the application of electric elds for lysing cells and manipulating DNA [5961]. Similarly, Taylor et al. [62] have shown sonication can be used to lyse anthrax spores. Without having the sample preparation steps automated, and more importantly, without being able to reduce the entire assay time to a few minutes, DNA based biosensors will never full the criteria of detect-to-protect devices. They will, however, prove highly useful in laboratory based detect-to-treat testing of infection and identication of suspect organisms. This view is supported in a recent article in the journal Science [1] where the head of the

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151

141

US chemical and biological national security at Lawrence Livermore National Laboratory is quoted as saying if I go from wanting an answer in hours to wanting one in 2 min, I have eliminated all kinds of technologies, like PCR. 2.2. Detection using surface featuresdetect-to-protect biosensors Biosensors, which use surface features of the BWA, rather than requiring extraction of a molecular marker indicative of the agent, are compatible with the development of detect-toprotect devices. To satisfy the ideal criteria for such devices, these biosensors must give a rapid response, must require no user intervention, must not give false positives, must have the sensitivity to detect just a few of the biological agents and must be able to sample in air despite biological molecules having evolved to operate in an aqueous environment. A representative list of afnity biosensing concepts that are motivated towards producing detect-to-protect devices are listed in Table 1 with some of their attributes. Some of the important developments are discussed in terms below. The rst commercial devices for detecting biological warfare agents such as those by BioVeris [63], Response Biomedical Corporation [64] and QTL [65] are an attempt to automate a sandwich immunoassay by immobilising capture antibodies on beads. The main differences between these three companies are in assay formats and the types of labels used. The QTL system exemplies the approach. The user has an instrument smaller than a laptop computer weighing three pounds, which is sold with self-sealing cartridges. The cartridges are swabbed onto the area of interest, placed in the meter, which gives an answer within 10 min. The main direction in the development of automated immunoassays is for more sensitive labels to reduce detection limit. BioVeris employs electrochemiluminescent labels where once a sandwich of the capture antibody, the analyte and the reporter antibody is formed on magnetic beads, the beads are trapped over an electrode and a potential applied to the electrode to stimulate emission of light from an electrochemiluminescent label. The ability to repeatedly stimulate the electrochemiluminescent label provides the potential for enhanced sensitivity with this transduction principle. QTL use a newly discovered phenomenon of uorescence superquenching. In uorescence superquenching, a uorescent polymer is employed. Association of this polymer with an energy or electron transfer quencher will cause the uorescence of the whole polymer strand to be quenched in an analogous manner to the breaking of one Christmas light bulb turning off the entire circuit of Christmas lights. The way QTL congure their immunoassay the uorescent polymers are coated around magnetic beads. The presence of the BWA typically causes the quencher to dissociate from the polymer with a concomitant increase in uorescence. The potential sensitivity of uorescence superquenching over using conventional uorophores is apparent but it is possible to envisage other energy transfer species in a biological sample, which could interfere. Despite the sensitivity of the label over conventional uorescence Table 1 shows the Response Biomedical Corporation device, which employs a conventional uorophore has a

lower detection limit than QTL for anthrax spores. The higher detection limit of the QTL system may reect interferences or the surface immobilisation chemistry employed as the surface chemistry is vital in enabling thermodynamically favourable surface binding and to limit non-specic binding. In the screening of a sample for several potential bioagents using immunoassay approaches either multiple labels could be used, say with different coloured quantum dots [66] or alternatively, protein chips with biorecognition molecules for certain bioagents are deposited at dened locations. Ligler and co-workers from the Naval Research Laboratory have worked extensively on the latter concept [6777] employing a pragmatic approach of only using reagents, which are commercially available. They have developed two different approaches. The rst [6771] has been commercialised by Research International, Monroe WA, USA, as the Analyte 2000 and then as a portable, automated version as the RAPTOR [78]. A monolayer of capture antibodies are immobilised on optical bres with the RAPTOR instrument having four channels, thus allowing four analytes to be detected. Binding of the antigen is then transduced by uorescently labelled reporter antibodies in a sandwich assay format. An important innovation in this work is the use of a specially tapered optical bres which allow the excitation of the uorophore label on the reporter antibody, via the evanescent wave generated by light propagating down the bre, as well as the collection of the uorescent light emitted by the label. The advantage of using evanescent electromagnetic radiation is that analytes can be detected in real time as only uorophores near the surface of the optical bre will be excited, thus relaxing the requirements for separation of bound from free reporter antibodies. The more recent technology of the Naval Research Laboratory uses a planar waveguide to allow arrays to be developed. The biosensor interface is prepared on glass substrates, which are modied with a silane based self-assembled monolayer to which antibodies and gangliosides for the target analytes are immobilised. Six stripes containing six different biorecognition molecules (Fig. 2) are immobilised using a microuidic block containing six channels. In the latest publication, [77] on working towards a desktop type instrument, the sample is then delivered using a similar microuidic block to deliver samples at right angles to the stripes containing the biorecognition molecules (Fig. 2B). Exposure to reporter antibodies with uorescent labels allows the presence of BWAs to be detected using evanescent wave excitation monitored with a CCD camera (Fig. 2D). The higher detection limits relative to the BioVeris immunoassay are most likely a consequence of the antibodies being immobilised on a at surface but the clear advantage is up to nine analytes have been measured at the same time [75]. Portable biosensors where the automation of sandwich immunoassays has been achieved without requiring batteries are currently on the market using the same principle as the pregnancy test kits. A typical pregnancy biosensor works by detecting the hormone Human Chronionic Gonadotrophin (hCG) in urine [79]. The biosensor has an absorbent swab to which the sample is applied and a plastic holder with two windows; the measurement window reports the results of the test and the validation window reveals whether the actual device is operating normally. Wetting

142

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151

Table 1 Representative detect-to-protect biosensors for different BWAs or BWA type analytes Analytes 1 Anthrax spores Detection limit 200 spores/mL Assay time 5 min Reagent free Yes Real samples No Comment Love wave acoustic sensor. Non-specic binding can be a problem. Commercialised by Response Biomedical, a uorescent bead immunoassay commercialised as a benchtop instrument Bead linked to the end of a cantilever using a linkage which is attacked by the BoNT B which is a zinc endopeptidase Fluorescent dyes laden liposomes containing ganglioside GT 1b as reporters. Similar to pregnancy tests Liposomes with ganglioside GM1 act as reporters in a sandwich assay where formation of the sandwich was monitored using impedance ISFET where CT binds directly to surface immobilised anti-CT antibody. Problems are reproducibility and non-specic binding Bilayer with donor labelled GM1 and acceptor labelled GM1. Binding of CT brings the donor and acceptor labelled GM1s closer together and increased F rster energy transfer o Uses redox active lipid microstructures containing ganglioside GM1 on an electrode surface. Binding of CT to GM1 reduces current Use liposomes containing polydiacetylene and ganglioside GM1. Binding of the CT to the GM1 results in a change in liposome colour Ion selective eld effect transistor (ISFET) which detects enzymatically the hydrolysis of NAD+ by cholera toxin Chemiluminescent immunometric assay to detect labelled antibodies Refs. [84]

Anthrax spores BoNT Ricin Smallpox

4000 spores 5 ng 10 ng 100000 pfu

15 min

No

Surfaces

[64]

BoNT B

5 pg/mL (8 nM)

15 min

Yes

No

[20]

BoNT

15 pg/mL

20 s

Yes

No

[81]

CT

1 fg/mL

20 s

No

No

[85]

CT

0.1 pg/mL

1 min

Yes

No

[86]

CT

12 pg/mL

Few minutes

Yes

No

[127,128]

CT

120 pg/mL

5 min

Yes

No

[9193]

CT

90 g/mL

NA

Yes

No

[90]

10

CT

160 g/mL

90 min

Yes

No

[129]

11

Anti-CT antibody

200 ng/mL

70 min

Yes

No

[130,131]

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151 Table 1 (Continued ) Analytes 12 Anthrax spores BoNT Ricin SEB Detection limit <10000 spores NA NA NA Assay time 10 min Reagent free Yes Real samples Surfaces liquids Comment Fluorescence polymer superquenching where magnetic beads are coated with uorescent polymers. Binding of analyte removed quencher. All reagents included in one use cartridge. This is the technology of QTL Pregnancy type test using ow immunosensor systems in a single cartridge. Commercialised by Biowarfare Agent Detection Devices (BADD) BioVeris has commercialised an electrochemiluminescence coupled with magnetic beads, essentially a sandwich immunoassay for bioweapons agents Optical bre based uorescent inmmunoassay. Field instruments called The Analyte 2000 and the Raptor have been developed from this technology by Research International Multianalyte arrays using sandwich assays with uorescent labels where the capture antibodies are immobilised on a wave-guide. Excitation is by evanescent wave and microuidics are used to deliver samples Uses an evanescent wave device to detect binding label free. Non-specic binding will be a problem A portable 2 channel SPR device, which has been tted to aircraft for airborne monitoring. Detection limts are decreased with a one-step and two-step amplication Simple immunoassay using quantum dots as the label. Using different size dots can do multiplexing in the same sample Sandwich assay which employed interdigitated electrodes and redox cycling. Non-specic binding becomes the limiting factor

143

Refs. [65,132]

13

Anthrax spores BoNT Ricin

1 g/mL 400 ng/mL 400 ng/mL

1525 min

Yes

Surfaces, liquids

[80]

14

BoNT Ricin SEB Smallpox

0.05 ng/mL 1 ng/mL 10 pg/mL 105 pfu/mL

30 min

No

Surfaces, milk, food

[22,63]

15

Anthrax BoNT CT Ricin F. tularensis

30 cfu/mL 110 ng/mL 0.11.0 ng/mL <0.5 ng/mL 5 104 cfu/mL

15 min

Yes

Liquids, blood

[6771]

16

Anthrax BoNT A BoNT B CT Ricin SEB Bacillus globigii F. tularensis BoNT SEB Ricin F. tularensis SEB

104 105 cfu/mL 40 ng/mL 200 ng/mL 1.6 ng/mL (antibody) 40 ng/mL (ganglioside) 8 ng/mL 3 ng/mL 2.5 104 cfu/mL 100 ng/mL <1 ng/mL 5 ng/mL 1 104.5 cfu/mL 2 ng/mL 0.56 ng/mL, one step 2.8 pg/mL, two steps

12 min

No

Nasal swabs, plasma, serum, airborne

[7277]

17

15 min

No Yes Yes No No

No

[82]

18

15 min 60 min 60 min

Urine Seawater

[8789]

19

CT Ricin SEB

10 ng/mL 30 ng/mL 3 ng/mL

2h

No

No

[66]

20

Bacteriophages Viruses

90 ng/mL 1.5 1010 viruses/mL

8 min

No

No

[133]

144 Table 1 (Continued ) Analytes 21 Anthrax Y. pestis Detection limit 1000 cfu 50 cfu

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151

Assay time 3 min

Reagent free Yes

Real samples Nasal swabs

Comment Genetically modied B lymphocytes engineered with luminescent proteins and membrane bound recognition antibodies Benchtop instrument combining immunoassay and PCR based DNA detection to verify immunoassay results. Can monitor air continuously for more than 1 week. Referred to as Automatic Pathogen Detection System (APDS)

Refs. [94]

22

Anthrax BoNT Y. pestis

1100 ng/mL

2h

No

Airborne

[96100]

All entries consider the biosensor alone rather than the entire sample collection and preparation system to allow comparison between the actual sensing components of each device. All detection limits refer to measurements in buffer to allow comparisons between systems. With regards to using a biosensor, reagent free refers when no reagents need to be added by the user to operate the biosensor but does not consider any sample preparation. Real sample refers to where the analysis of natural samples with the biosensor was reported with the types of samples analysed as listed. BoNT refers to Botulinum toxin, CT is Cholera toxin, SEB is Staphylococcal enterotoxin B, NA refers to the information not being available, and the shaded rows indicate technologies which are or will soon be commercially available.

a swab with urine results in it wicking up the swab towards the capture antibodies. The urine also solubilizes reporter antibodies, which are labelled with a coloured bead. As the hormone is smaller than the reporter antibody it wicks up the swab quicker and reaches the capture antibody rst. If the hormone is present, and is captured the reporter antibody with its coloured bead will also bind to the antibody and a coloured band will appear in the test window. The reporter antibody also ows past the upstream validation window where a coloured line reveals a negative result is not a consequence of the biosensor not working. This biosensor format has been applied to other analytes such as cryptosporidium in water and is the basis of the Biowarfare Agent Detection Devices (BADD) technology [80] for the detection of anthrax, botulinum toxin (BoNT) and ricin. In the case of anthrax the biosensor can detect as low as 1 g/mL in solution or 0.25 g off a surface in 1525 min. Slightly better performances are obtained with botulinum toxin and ricin biosensors (detection limits of 400 ng/mL or 50 ng from a surface in 510 min). The advantage of these biosensors is simplicity.

No power is required for the device as the transducer of the optical signal is the users eye. The drawbacks are the need to wet the device and the poor detection limits. The high detection limit is a consequence of the low sensitivity of the human eye with a considerable number of reporter antibodies needing to be captured to generate a discernable colour change for our eyes to see. The problem of detection limit with these devices appears to have been solved for BoNT by Durst and co-workers [81] where the reporter is a liposome containing a uorescent rhodamine dye. Incorporated into the lipid bilayer of the liposome is the ganglioside GT 1b, which specically binds to BoNT. The device has an incredibly low detection limit of 15 pg/mL. Coupled with the simplicity and power free needs of the device this is a very attractive technology. In common with the BADD technology, the requirement for wetting the sample for the detection scheme to work remains and is therefore incompatible with detecting airborne analytes. The biosensing concepts discussed above all use innovative approaches to performing a sandwich immunoassay, which

Fig. 2. Orthogonal patterning and assay method. Biotinylated capture antibodies are immobilized in stripes on a NeutrAvidin-coated slide/waveguide via avidinbiotin interactions (A). For analysis, samples are loaded onto the slide using a PDMS ow guide module with channels oriented orthogonal to the stripes of immobilized capture antibodies (B). After incubation with uorescent sample (direct assay) or after interrogation of the array with uorescent tracer antibody (sandwich assay), uorescent complexes are formed in the appropriate loci (C). The pattern of uorescent spots are imaged and digitised (D). Reproduced with permission from [77], Copyright Springer (2004).

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151

145

requires little or no user intervention to perform. The difculty with making a sandwich immunosensor reagentless arises from the need for two recognition molecules, the capture species and the reporter which must not only bind to different epitopes of the antigen but the two binding events must be separated temporally. In recent times other biosensing concepts have been developed which do not employ the sandwich assay concept, many of which are label free. Truly label free methods for detecting biological agents have been developed using optical based evanescent wave devices [82] such as surface plasmon resonance (SPR) and the resonant mirror, reectometry [7,83], acoustic wave devices such as the Love wave acoustic sensors [84] and quartz crystal microbalances [85] and ion selective eld effect transistors (ISFET) [86]. As can be seen from Table 1 incredibly low detection limits and rapid responses can be obtained with such devices. The key drawback of all these devices is that they all detect a change at the transducer surface rather than use a signal which requires actual binding between the biorecognition molecule and the analyte. Any change at the interface will cause a similar response and therefore non-specic binding becomes a problem; a problem exacerbated by the high sensitivity of the devices. This problem highlights one of the advantages of the sandwich assay format in using two different biorecognition molecules. Two biorecognition molecules provide two layers of selectivity which helps reduce the problem of non-specic binding. The problem of non-specic binding with label free methods has however been diminished signicantly by a two channel SPR device developed by Texas Instruments and using for the detection of BWAs by Naimushin et al. [8789]. One channel was modied with antibodies for Staphylococcus aureus enterotoxin B (SEB) and the other channel is used as a reference channel. The reference channel compensates for changes in refractive index due to non-specic binding, if any, and changes due to different solutions. Without any amplication steps the biosensor can detect 70 pM (2 ng/mL) SEB in 15 min and can run continuously for hours. Amplication of the SPR signal using secondary antibodies which bind to the captured SEB and using a third antibody to bind to the secondary antibody detection limits down to 100 fM (2.8 pg/mL) can be achieved in buffer [87]. These detection limits are somewhat elevated when analysis is performed in urine but 50 pM SEB can still be detected in urine [87]. The SPR system is a portable SPR and the same workers have recently reported its incorporation within an aircraft for air monitoring [89]. There are a number of new biosensing concepts which are effectively reagentless which are perhaps more robust then the completely label free methods discussed above. Lipid bilayers have been used for a number of reagentless biosensing concepts. The attractiveness of lipid bilayers is that gangliosides can easily be incorporated. Gangliosides are lipids with carbohydrate heads found on cell surfaces. Gangliosides are typically the location on the cell surface where toxins bind to gain entry into the cell [90]. Hence gangliosides have been shown to be selective for a variety of toxins [72]. Pan and Charych [90] reported on polydiacetylene liposomes containing ganglioside GM1, which is selective for cholera toxin (CT). Binding of the CT causes a conformational

Fig. 3. A schematic of the proposed mechanism for redox microstructure sensor. (A) Without binding of analyte electrons can hop from one ferrocene to another whilst if cholera binds (B) the hopping pathway is blocked. Reproduced with permission of The Royal Society of Chemistry from [93], Copyright Royal Society of Chemistry (2004).

change to the polydiacetylene backbone in the liposome with a resultant colour change from purple-blue to orange. A similar idea has been converted to an electrochemical format where lipid microstructures are adsorbed onto electrode surfaces [9193]. Apart from the normal lipids and ganglioside GM1 the lipid structures also contained ferrocene labelled lipids. Ferrocene molecules on the other side of the lipid structure from the electrode would normally be regarded as too far from the electrode to be oxidised or reduced. However, all ferrocene molecules can be interrogated electrochemically in this system due to electron hopping from ferrocene to ferrocene (Fig. 3). Binding of CT interrupted the electron-hopping pathway and a reduced current was observed. A reduced signal on binding of an analyte is usually disfavoured as false positives are more likely than with a transduction method where an enhanced signal arises from a binding event. Swanson and co-workers have used the multivalent character of CT to allow multiple ganglioside GM1 molecules to bind to the same toxin in a uorescent biosensor. The gangliosides are labelled with either a donor or an acceptor. Binding of CT brings the donor and acceptor closer together, which increases the amount of F rster Resonance Energy Transfer (FRET) and o

146

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151

Fig. 4. A microcantilever based biosensor developed by Parpura and co-workers [20] for the detection of botulinum toxin. (A) Top view of a cantilever with a bead functionalised tip. (B) Schematic of how the bead is attached to the tip and the resultant frequency response of the cantilever. Recombinant synaptobrevin 2 (red) is attached to bead surface where it binds to recombinant syntaxin 1a, which is attached to the cantilever tip. The synaptobrevin 2-syntaxin intermolecular interaction leads to the beads being tethered to the cantilever tip. The amplitude of oscillation of the tip with frequency is shown below the cartoon. BoNT-B in the presence of zinc ions (Zn2+ ) cleaves synaptobrevin 2 in such manner that the short fragment of synaptobrevin 2 remaining on the bead has no ability to interact with syntaxin 1a on the cantilever tip, leading to the detachment of the bead from the tip (double-headed vertical arrow) and a more dramatic change in amplitude with frequency (panel below cartoon). Reproduced with permission from [20], Copyright National Academy of Sciences, USA (2003).

so increased acceptor uorescence is observed upon binding of the toxin. Parpura and co-workers [20] from the University of California at Riverside have recently published an idea for detecting botulinum toxin (BoNT), which is a complete departure from the use of afnity biosensors for detecting BWAs. The beauty of the idea is to exploit the biological action of the toxin to allow its detection. BoNT B is a zinc endopeptidase enzyme which disrupt the synaptic nerves by cause a cleavage in synaptobrevin 2 if zinc is present. The biosensor monitors the change resonant oscillation microcantilever. A bead is attached to the cantilever using an afnity complex between synaptobrevin 2 attached to the bead and syntaxin 1A attached to the cantilever. BoNT causes a break in the afnity complex, thus releasing the bead and causing a change in the resonant oscillation (Fig. 4). The detection limit of the biosensor is 5 pg/mL if the assay is left for 15 min (Table 1). Unlike all the afnity biosensors discussed above where the binding reaches equilibrium this biosensor is a kinetic assay reliant on the biological activity of the toxin. Therefore, if the user was willing to wait longer then the detection limits could be further reduced or alternatively if a faster response is required this is achievable but at the cost to the detection limit. Another novel technology applicable to the detection of BWAs is CANARY (cellular analysis and notication of antigen

risks and yields). Developed by Rider et al. [94], CANARY uses B lymphocyte cells as the detector, and as the reporter, of the presence of pathogens with response times of a few seconds. The lymphocytes were genetically engineered to express both cytosolic aequorin, a calcium sensitive bioluminescent protein from the jellysh Aequoria victoria, as well as membrane bound antibodies specic for the pathogen of interest. Cross-linking of the antibodies during recognition of the pathogen results in elevated intracellular calcium with a concomitant emission of light. With regards to bacteria of interest in biological warfare, cells engineered for Yersinia pestis (plague) could detect as little as 50 cfu in 3 min whilst suffering no interference from a large excess of Francisella tularensis. Similarly, cells were also developed to detect anthrax spores from seeded nasal swabs down to a level 1000 cfu. 3. Cutting edge of biosensor technology which may be applicable to bioweapons detection None of the existing biosensors discussed above for detecting biological warfare agents satisfy the most demanding of requirements, a detect-to-protect biosensor which could be worn by a soldier or civilian in an environment where the possibility of a biological attack is prevalent. There are a number of impediments to the development of such a biosensor including

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151

147

(1) transferring an airborne sample to an aqueous environment where the biosensor can operate, (2) reducing the detection limit, and (3) obtaining more rapid responses. There are a number of advances with regards to both novel biosensing components and novel biosensors, which will enable these problems to be overcome. 3.1. Air sampling The transferral of airborne samples to an aqueous environment is necessary because the function of biorecognition molecule is optimised for aqueous environments. There are very few options currently available for performing this task. The demands on any air sample technology are quite stringent. Extremely low levels of airborne BWAs can still have a severe impact and therefore it is desirable for the sampling technique to not only transfer biological material into the aqueous phase but also concentrate it. The average person breathes roughly 6 L/min. In this large volume sample there need be only a BWA or two to do harm [95] and therefore this must be concentrated down to a few millilitres at most. The concentration of the analyte must be achieved without harming the delicate biological sample and be rapid so the response time of the biosensor is not limited by this step. There are currently two main technologies for achieving this at present. These are the wetted cyclone samplers and virtual impactors. At present neither of these types of samplers is compatible with a wearable device but are highly suitable for sampling in conned spaces. Therefore, if the biosensor is required to provide early warning of biological agents entering a building then these types of concentrators are ideal as virtually all the air entering a building can be sampled in a similar manner to the BioWatch ltering system used in some state buildings in the United States. A technology has been developed by the Lawrence Livermore National Laboratory in the United States, which employs both a virtual impactor and a cyclone sampler in series to concentrate aerosol for the multiplex detection of BWAs by immunoassay and PCR [96,97]. The benchtop device is called the autonomous pathogen detection system (APDS). The APDS is designed to sample air for BWAs over extended periods (at present a week) in domestic situations (such as airport terminals) where the public may be at risk. The virtual impactor collects particles of 110 m in size, which are concentrated in the cyclone sampler. Once an hour the aerosol collector dispenses liquid to an automated uidics module, which performs a bead based immunoassay, which requires 2 h to complete [98]. The sandwich immunoassay employs a ow cytometer and antibody modied microspheres labelled with a bar code of uorescent dyes [98100]. The bar code allows up to 100 different analytes to be monitored with the APDS currently monitoring eight including anthrax, botulinum toxin and pneumonic plague [96,97]. Detection limits are typically between 1 and 100 g/L depending on the analyte [12]. Part of each sample analysed by immunoassay is archived. If a positive signal is recorded, and the BWA contains DNA, part of the archived sample is retrieved for DNA purication and analysis by a real-time ow-through PCR detection system based on an earlier technology developed by Belgrader et al. [25]. In

this way, nucleic acid conrmation of the immunoassay result is obtained. APDS instruments have been deployed in at strategic locations where they have reliably operated for weeks or months [96]. At this point in time it is not evident if alternatives to the cyclone sampler and virtual impactor are being explored, with a view to personalised biosensors for sampling airborne pathogens carried by individuals. The low concentration of biological agents required to cause infection does create a problem as it means that a small biosensor placed in a single location could quite easily miss interacting with a BWA. An alternative to a sampler, which concentrates the air sample is to use a high surface area biosensor. The research interest in electronic textiles and other smart fabrics, which could perform additional tasks such as energy conversion or sense environmental stimuli is highly compatible with a high surface area biosensor [101104]. Thus it is possible to envisage an electronic textile where biorecognition molecules are incorporated within the textile and the electronic bres act as the transducer. A high surface area biosensor however requires the biosensor to either work in the airborne environment or for the biorecognition molecules to be in a water-like environment. 3.2. Reducing the detection limit and providing a more rapid response Any innovative biosensor, which can provide a warning of airborne analytes will be required to have very low detection limits. The detection limit of biosensors can be lowered in one of three ways. Firstly, to use biorecognition molecules or surface chemistry which have higher equilibrium constants of binding, secondly to limit non-specic binding or nally to use more sensitive transducers. Novel biorecognition molecules with high afnity for proteinaceous species include aptamers [105], peptides and antibodies developed from phage display libraries. Aptamers [106108] are sequences of DNA, which bind to a large macromolecules in a similar manner to an antibody. New aptamers can be found by starting with libraries of random oligonucleotides, which are passed down an afnity column, which contains an analogue of a surface epitope of the target biological agent. Removal of the oligonucleotides, which do not bind followed by elution of the bound ones allows a renement of the library. Cloning of the eluted oligonucleotides followed by repeated passages through the column allows further renement until only the most strongly binding aptamers with the highest afnity for the analyte are left. The selected aptamer is then amplied and a new afnity molecule is identied. This procedure for producing aptamers is called the SELEX method and can isolate new biorecognition molecules with afnities higher than natural antibodies. Similar principles have been used to identify peptide sequences for microorganisms [109,110] and new antibodies [111]. Using biorecognition molecules with higher afnity has its limitations however. The strongest known of all bioafnity reactions is avidin with biotin and it is hard to envisage a Ka higher than 1015 M1 . Recently some exciting work where binding

148

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151

of aptamers to the proteins allows amplication of the bound aptamers only which has allowed detection limits signicantly below the afnity constant to be achieved [112]. However, this currently requires user intervention through amplication of nucleic acids using PCR. To get lower detection limits still will then require, preventing any non-specic binding and using more sensitive transducers. Higher sensitivity transducers are probably the major areas of focus into biosensor research. For labelled biosensors, which employ the sandwich assay format the most common approach to increasing the sensitivity of transduction is to use brighter labels, that is label systems that require fewer labels to be present for the transducer to be able to detect them. Examples of brighter labels including uorescence superquenching [46,47], quantum dots [39,66], electrochemiluminescent molecules [63] and dye labelled liposomes [81] were discussed above and show considerable promise. The alternative to brighter labels is to use more sensitive transducers, which can detect just a few molecules that have bound without using labels. The microcantilever transducer used by Parpura and co-workers [20] is one example. Microcantilevers transducers either use the sensitivity of the characteristic frequency of resonance to changes in mass [113] or bending of the cantilever due to changes in surface stress [114,115] to monitor afnity reactions. Recently Gupta et al. [116] have used the resonance of a microcantilever to detect the binding of a single virus particle. It is proposed that the biosensing concept could be used to detect airborne viruses although the applicability of the biosensor has yet to be demonstrated for the detection of viruses in samples other than pure buffer solutions. The detection of a single virus has been demonstrated by Lieber and co-workers using a silicon nanowire eld effect transistor (FET), see Fig. 5. Nanowire or nanotube based FETs have been demonstrated by a number of workers [117] for the detection of, protein binding [118,119], enzyme activity [120] and even for detecting small molecules such as oxygen [121]. All of these biosensors work on the same transduction principle of a change in resistance through the nanotube or nanowire as a consequence of the biorecognition event. The Lieber groups recent contribution is the rst example of this highly attractive transduction technology being employed for biological agents. Furthermore, the ability of the biosensor to detect viruses has been performed with both pure and unpuried virus solutions with similar results. Recent work but this group has also shown these nanowire based FETs can work in protein samples with no evidence of non-specic binding [122]. 3.3. Decreasing the response timethe needle in a haystack problem. When the potential stumbling blocks with biosensors that detect just a few molecules are solved there is an additional challenge, which relates to detecting single particles or a few particles. How long must one wait for a response? Expressed in another way, is the absence of a positive signal evidence that there is no biological agent or is it a false negative due to the fact that the biological agent did not cross the nano-dimensional

Fig. 5. Schematic of the nano-wire based detection of a single virus showing the two nanowire devices in parallel (left) and the expected response (right). Only nanowire 2 is modied with antibodies, which are specic for the virus. Binding of the virus reduces the conductance. Reproduced with permission from [126], Copyright National Academy of Sciences, USA (2004).

space occupied by the biosensor? An array of the nanoscale biosensors could reduce this problem but still presents a related problem, which can be referred to as the needle in a haystack problem. When a biosensor is in a dened location in space, on a surface, even if the analyte is present it must still diffuse to the sensing surface for recognition to occur. Even in a small volume this could represent diffusing a long distance on the molecular scale. Hence the analyte may never be found by the biosensor in a reasonable time in a similar manner to a needle in a haystack never being found. One solution to the needle in the haystack problem is to use a highly porous, high surface area biosensor where there is intimate mixing of the sample for analysis with the biosensor. This idea is exploited by the company PamGene [123] for the detection of DNA using uorescently labelled reporter sequences. Although labels may complicate the assay the short diffusion path-lengths as a consequence of the intimate mixing of sample and transducer allows the hybridisation of target DNA in real time. The attractiveness of highly porous transducers without using conventional labels is the subject of intense research activity via porous silicon as a transducer [7,8,83,124]. Porous silicon photonic crystals can be used as mirrors where binding events cause changes in reectivity. The high porosity can potentially provide rapid responses and the mirrors that can be fabricated can have reectance peaks with subnanometer full

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151

149

width half maximum [125]. Such sharp peaks mean that even a few molecules bound can cause a large shift. 4. Conclusions Biosensors have incredible potential for achieving the very difcult task of producing detect-to-protect devices once a number of criteria are met. These criteria include having sufciently low detection limits to detect just a few biological warfare agents, being able to sample analytes from the air and from complex biological matrices, giving responses within a minute or two, giving very few false positives and being light, portable devices which require no power to operate. No single device will full all these criteria in the short term and to do so will require a long term focussed scientic and nancial effort if this exciting technology is to full its promise. In the short term however devices, which meet some of these goals have been and will be developed. Acknowledgements I would like to thank the Alexander von Humboldt Foundation fellowship support during part of the time this review was written. References
[1] K. Brown, Science 305 (2004) 1228. [2] C.L. Baird, D.G. Myszka, J. Mol. Recog. 14 (2001) 261. [3] E. Mallat, D. Barcelo, C. Barzen, G. Gauglitz, R. Abuknesha, Trends Anal. Chem. 20 (2001) 124. [4] Optical Biosensors: Present and Future, Elsevier, Amsterdam, 2002. [5] P. Skl dal, Electroanalysis 9 (1997) 737. a [6] T. Weimar, Angew. Chem. -Int. Ed. 39 (2000) 1219. [7] V.S.Y. Lin, K. Motesharei, K.P.S. Dancil, M.J. Sailor, M.R. Ghadiri, Science 278 (1997) 840. [8] K.P.S. Dancil, D.P. Greiner, M.J. Sailor, J. Am. Chem. Soc. 121 (1999) 7925. [9] C.K. OSullivan, G.G. Guilbault, Biosens. Bioelectron. 14 (1999) 663. [10] A. Akkoyun, U. Bilitewski, Biosens. Bioelectron. 17 (2002) 655. [11] J.J. Gooding, F. Mearns, W. Yang, J. Liu, Electroanalysis 15 (2003) 81. [12] J.P. Fitch, E. Raber, D.R. Imbro, Science 302 (2003) 1350. [13] L.A. Broussard, Mol. Diagn. 6 (2001) 323. [14] B. Robinson-Dunn, Arch. Path. Lab. Med. 126 (2002) 291. [15] A. Fox, G.C. Stewart, L.N. Waller, K.F. Fox, W.M. Harley, R.L. Price, J. Microbiol. Meth. 54 (2003) 143. [16] H.A. Hartley, A.J. Baeumner, Anal. Bioanal. Chem. 376 (2003) 319. [17] A.A. Hindle, E.A.H. Hall, Analyst 124 (1999) 1599. [18] B.J. White, J.A. Legako, H.J. Harmon, Sens. Actuators B 97 (2004) 277. [19] P.A. Emanuel, R. Bell, J.L. Dang, R. McClanahan, J.C. David, R.J. Burgess, J. Thompson, L. Collins, T. Hadeld, J. Clin. Microbiol. 41 (2003) 689. [20] W. Liu, V. Montana, E.R. Chapman, U. Mohideen, V. Parpura, Proc. Natl. Acad. Sci. U.S.A. 100 (2003) 13621. [21] M. May, Scientist 18 (2004) 36. [22] J.A. Higgins, M.S. Ibrahim, F.K. Knauert, G.V. Ludwig, T.M. Kijek, J.W. Ezzell, B.C. Courtney, E.A. Henchal, Sensitive and Rapid Identication of Biological Threat Agents in Food and Agricultural Security, vol. 894, 1999, 130.

[23] J.L. Versage, D.D.M. Severin, M.C. Chu, J.M. Petersen, J. Clin. Microbiol. 41 (2003) 5492. [24] M.S. Ibrahim, R.S. Lofts, P.B. Jahrling, E.A. Henchal, V.W. Weedn, M.A. Northrup, P. Belgrader, Anal. Chem. 70 (1998) 2013. [25] P. Belgrader, C.J. Elkin, S.B. Brown, S.N. Nasarabadi, R.G. Langlois, F.P. Milanovich, B.W. Colston, G.D. Marshall, Anal. Chem. 75 (2003) 3446. [26] M. Panning, M. Asper, S. Kramme, H. Schmitz, C. Drosten, Clin. Chem. 50 (2004) 702. [27] M. Laassri, V. Chizhikov, M. Mikheev, S. Shchelkunov, K. Chumakov, J. Virol. Meth. 112 (2003) 67. [28] J.A. Higgins, S. Nasarabadi, J.S. Karns, D.R. Shelton, M. Cooper, A. Gbakima, R.P. Koopman, Biosens. Bioelectron. 18 (2003) 1115. [29] H.J. Lin, P.T. Charles, J.D. Andreadis, A.M. Churilla, D.A. Stenger, J.J. Pancrazio, Anal. Chim. Acta 457 (2002) 97. [30] A.J. Baeumner, B. Leonard, J. McElwee, R.A. Montagna, Anal. Bioanal. Chem. 380 (2004) 15. [31] J.J. Gooding, Electroanalysis 14 (2002) 1149. [32] T.G. Drummond, M.G. Hill, J.K. Barton, Nat. Biotechnol. 21 (2003) 1192. [33] J.H. Graber, M.J. ODonnell, C.L. Smith, C.R. Cantor, Curr. Opin. Biotech. 9 (1998) 14. [34] C.B.V. Christensen, Talanta 56 (2002) 289. [35] M. Boncheva, L. Scheibler, P. Lincoln, H. Vogel, B. Akerman, Langmuir 15 (1999) 4317. [36] R. Levicky, T.M. Herne, M.J. Tarlov, S.K. Satija, J. Am. Chem. Soc. 120 (1998) 9787. [37] R. Georgiadis, K.P. Peterlinz, A.W. Peterson, J. Am. Chem. Soc. 122 (2000) 3166. [38] E.L.S. Wong, E. Chow, J.J. Gooding, Langmuir 21 (2005) 6957. [39] G.P. Mitchell, C.A. Mirkin, R.L. Letsinger, J. Am. Chem. Soc. 121 (1999) 8122. [40] S. Takenaka, K. Yamashita, M. Takagi, Y. Uto, H. Kondo, Anal. Chem. 72 (2000) 1334. [41] D.J. Maxwell, J.R. Taylor, S.M. Nie, J. Am. Chem. Soc. 124 (2002) 9606. [42] S.J. Park, T.A. Taton, C.A. Mirkin, Science 295 (2002) 1503. [43] K.M. Hansen, H.F. Ji, G.H. Wu, R. Datar, R. Cote, A. Majumdar, T. Thundat, Anal. Chem. 73 (2001) 1567. [44] R.M. Umek, S.W. Lin, J. Vielmetter, R.H. Terbrueggen, B. Irvine, C.J. Yu, J.F. Kayyem, H. Yowanto, G.F. Blackburn, D.H. Farkas, Y.P. Chen, J. Mol. Diagn. 3 (2001) 74. [45] J.J. Gooding, A. Chou, F.J. Mearns, E. Wong, K.L. Jericho, Chem. Commun. (2003) 1938. [46] S.A. Kushon, K.D. Ley, K. Bradford, R.M. Jones, D. McBranch, D. Whitten, Langmuir 18 (2002) 7245. [47] S.A. Kushon, K. Bradford, V. Marin, C. Suhrada, B.A. Armitage, D. McBranch, D. Whitten, Langmuir 19 (2003) 6456. [48] Y. Kuang, I. Biran, D.R. Walt, Anal. Chem. 76 (2004) 2902. [49] R.L. Edelstein, C.R. Tamanaha, P.E. Sheehan, M.M. Miller, D.R. Baselt, L.J. Whitman, R.J. Colton, Biosens. Bioelectron. 14 (2000) 805. [50] M. Thompson, L.M. Furtado, Analyst 124 (1999) 1133. [51] O. Lioubashevski, F. Patolsky, I. Willner, Langmuir 17 (2001) 5134. [52] A. Bardea, F. Patolsky, A. Dagan, I. Willner, Chem. Commun. (1999) 21. [53] E.M. Boon, D.M. Ceres, T.G. Drummond, M.G. Hill, J.K. Barton, Nat. Biotechnol. 18 (2000) 1096. [54] E.L.S. Wong, J.J. Gooding, Anal. Chem. 75 (2003) 3845. [55] M.U. Kopp, A.J. de Mello, A. Manz, Science 280 (1998) 1046. [56] Z. Zhao, Z. Cui, D.F. Cui, S.H. Xia, Sens. Actuators B 108 (2003) 162. [57] J. Liu, C. Hansen, S.R. Quake, Anal. Chem. 75 (2003) 4718. [58] G.H.W. Sanders, A. Manz, Trac-Trends Anal. Chem. 19 (2000) 364. [59] C.F. Edman, D.E. Raymond, D.J. Wu, E.G. Tu, R.G. Sosnowski, W.F. Butler, M. Nerenberg, M.J. Heller, Nucl. Acids Res. 25 (1997) 4907.

150

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151 [96] B.J. Hindson, A.J. Makarewicz, U.S. Setlur, B.D. Henderer, M.T. McBride, J.M. Dzenitis, Biosens. Bioelectron. 20 (2005) 1925. [97] B.J. Hindson, M.T. McBride, A.J. Makarewicz, B.D. Henderer, U.S. Setlur, S.M. Smith, D.M. Gutierrez, T.R. Metz, S.L. Nasarabadi, K.S. Venkateswaran, S.W. Farrow, B.W. Colston, J.M. Dzenitis, Anal. Chem. 77 (2005) 284. [98] B.J. Hindson, S.B. Brown, G.D. Marshall, M.T. McBride, A.J. Makarewicz, D.M. Gutierrez, D.K. Wolcott, T.R. Metz, R.S. Madabhushi, J.M. Dzenitis, B.W. Colston, Anal. Chem. 76 (2004) 3492. [99] M.T. McBride, S. Gammon, M. Pitesky, T.W. OBrien, T. Smith, J. Aldrich, R.G. Langlois, B. Colston, K.S. Venkateswaran, Anal. Chem. 75 (2003) 1924. [100] M.T. McBride, D. Masquelier, B.J. Hindson, A.J. Makarewicz, S. Brown, K. Burris, T. Metz, R.G. Langlois, K.W. Tsang, R. Bryan, D.A. Anderson, K.S. Venkateswaran, F.P. Milanovich, B.W. Colston, Anal. Chem. 75 (2003) 5293. [101] W. Lu, A.G. Fadeev, B.H. Qi, E. Smela, B.R. Mattes, J. Ding, G.M. Spinks, J. Mazurkiewicz, D.Z. Zhou, G.G. Wallace, D.R. MacFarlane, S.A. Forsyth, M. Forsyth, Science 297 (2002) 983. [102] R. ODonnell, Proc. IEEE 91 (2003) 1993. [103] D. Marculescu, R. Marculescu, N.H. Zamora, P. Stanley-Marbell, P.K. Khosla, S. Park, S. Jayaraman, S. Jung, C. Lauterbach, W. Weber, T. Kirstein, D. Cottet, J. Grzyb, G. Troster, M. Jones, T. Martin, Z. Nakad, Proc. IEEE 91 (2003) 1995. [104] R.F. Service, Science 301 (2003) 909. [105] J.G. Bruno, J.L. Kiel, Biosens. Bioelectron. 14 (1999) 457. [106] S.S. Iqbal, M.W. Mayo, J.G. Bruno, B.V. Bronk, C.A. Batt, J.P. Chambers, Biosens. Bioelectron. 15 (2000) 549. [107] C.K. OSullivan, Anal. Bioanal. Chem. 372 (2002) 44. [108] E. Luzi, M. Minunni, S. Tombelli, M. Mascini, Trends Anal. Chem. 22 (2003) 810. [109] H.Y. Mason, C. Lloyd, M. Dice, R. Sinclair, W. Ellis, L. Powers, Biosens. Bioelectron. 18 (2003) 521. [110] C. Estes, A. Duncan, B. Wade, C. Lloyd, W. Ellis, L. Powers, Biosens. Bioelectron. 18 (2003) 511. [111] A.D. Sheehan, J. Quinn, S. Daly, P. Dillon, R. OKennedy, Anal. Lett. 36 (2003) 511. [112] D. Di Guisto, W.A. Wlassoff, J.J. Gooding, B.A. Messerle, G.C. King, Nucl. Acids Res. 33 (2005) e64. [113] A.M. Moulin, S.J. OShea, M.E. Welland, Ultramicroscopy 82 (2000) 23. [114] F. Quist, V. Tabard-Cossa, A. Badia, J. Phys. Chem. B. 107 (2003) 10691. [115] M. Lahav, C. Durkan, R. Gabai, E. Katz, I. Willner, M.E. Welland, Angew. Chem. -Int. Ed. 40 (2001) 4095. [116] A. Gupta, D. Akin, R. Bashir, Appl. Phys. Lett. 84 (2004) 1976. [117] F. Patolsky, G.F. Zheng, O. Hayden, M. Lakaamyali, X.W. Zhuang, C.M. Lieber, Proc. Natl. Acad. Sci. U.S.A. 101 (2004) 14017. [118] Y. Cui, Q.Q. Wei, H.K. Park, C.M. Lieber, Science 293 (2001) 1289. [119] S. Boussaad, N.J. Tao, R. Zhang, T. Hopson, L.A. Nagahara, Chem. Commun. (2003) 1502. [120] K.A. Williams, P.T.M. Veenhuizen, B.G. de la Torre, R. Eritja, C. Dekker, Nature 420 (2002) 761. [121] P.G. Collins, K. Bradley, M. Ishigami, A. Zettl, Science 287 (2000) 1801. [122] G.F. Zheng, F. Patolsky, Y. Cui, W.U. Wang, C.M. Lieber, Nat. Biotechnol. 23 (2005) 1294. [123] PamGene, URL: http://www.pamgene.com/. [124] S. Chan, P.M. Fauchet, Y. Li, L.J. Rothberg, B.L. Miller, Phys. Status Solidi A, Appl. Res. 182 (2000) 541. [125] P. Reece, G. Lerondel, J. Mulders, W.H. Zheng, M. Gal, Proceedings of the Third International Conference on Porous Semiconductors Science and Technology (PSST 2002), Tenerife, Spain, 2002. [126] F. Patolsky, G.F. Zheng, O. Hayden, M. Lakaamyali, X.W. Zhuang, C.M. Lieber, Proc. Natl. Acad. Sci. U.S.A. 101 (2004) 14017. [127] X.D. Song, B.I. Swanson, Anal. Chem. 71 (1999) 2097.

[60] Y. Huang, K.L. Ewalt, M. Tirado, T.R. Haigis, A. Forster, D. Ackley, M.J. Heller, J.P. OConnell, M. Krihak, Anal. Chem. 73 (2001) 1549. [61] M.J. Heller, A.H. Forster, E. Tu, Electrophoresis 21 (2000) 157. [62] M.T. Taylor, P. Belgrader, B.J. Furman, F. Pourahmadi, G.T.A. Kovacs, M.A. Northrup, Anal. Chem. 73 (2001) 492. [63] BioVeris Corporation, URL: http://www.bioveris.com/. [64] Response Biomedical Corp. URL: http://www.responsebio.com/. [65] QTL Biosystems, URL: http://www.qtbio.com. [66] E.R. Goldman, A.R. Clapp, G.P. Anderson, H.T. Uyeda, J.M. Mauro, I.L. Medintz, H. Mattoussi, Anal. Chem. 76 (2004) 684. [67] R.A. Ogert, J.E. Brown, B.R. Singh, L.C. Shriverlake, F.S. Ligler, Anal. Biochem. 205 (1992) 306. [68] G.P. Anderson, K.D. King, L.K. Cao, M. Jacoby, F.S. Ligler, J. Ezzell, Clin. Diagn. Lab. Immunol. 5 (1998) 609. [69] G.P. Anderson, M.A. Jacoby, F.S. Ligler, K.D. King, Biosens. Bioelectron. 12 (1997) 329. [70] L.K. Cao, G.P. Anderson, F.S. Ligler, J. Ezzell, J. Clin. Microbiol. 33 (1995) 336. [71] L.C. Shriverlake, R.A. Ogert, F.S. Ligler, Sens. Actuators B 11 (1993) 239. [72] C.A. Rowe-Taitt, J.J. Cras, C.H. Patterson, J.P. Golden, F.S. Ligler, Anal. Biochem. 281 (2000) 123. [73] C.A. Rowe-Taitt, J.P. Golden, M.J. Feldstein, J.J. Cras, K.E. Hoffman, F.S. Ligler, Biosens. Bioelectron. 14 (2000) 785. [74] J.B. Delehanty, F.S. Ligler, Anal. Chem. 74 (2002) 5681. [75] C.R. Taitt, G.P. Anderson, B.M. Lingerfelt, M.J. Feldstein, F.S. Ligler, Anal. Chem. 74 (2002) 6114. [76] F.S. Ligler, C.R. Taitt, L.C. Shriver-Lake, K.E. Sapsford, Y. Shubin, J.P. Golden, Anal. Bioanal. Chem. 377 (2003) 469. [77] C.R. Taitt, J.P. Golden, Y.S. Shubin, L.C. Shriver-Lake, K.E. Sapsford, A. Rasooly, F.S. Ligler, Microbial. Ecol. 47 (2004) 175. [78] R. International in Research International, RAPTOR, Vol. Research International. [79] Unipath PLC, URL: http://www.unipath.com/clinicaldiagnostics.cfm. [80] Biowarfare Agent Detection Device (BADD), URL: http://osbornscientic.com/. [81] S. Ahn-Yoon, T.R. DeCory, R.A. Durst, Anal. Bioanal. Chem. 378 (2004) 68. [82] T. OBrien, L.H. Johnson, J.L. Aldrich, S.G. Allen, L.T. Liang, A.L. Plummer, S.J. Krak, A.A. Boiarski, Biosens. Bioelectron. 14 (2000) 815. [83] S. Chan, S.R. Horner, P.M. Fauchet, B.L. Miller, J. Am. Chem. Soc. 123 (2001) 11797. [84] D.W. Branch, S.M. Brozik, Biosens. Bioelectron. 19 (2004) 849. [85] L. Alfonta, I. Willner, D.J. Throckmorton, A.K. Singh, Anal. Chem. 73 (2001) 5287. [86] M. Zayats, O.A. Raitman, V.I. Chegel, A.B. Kharitonov, I. Willner, Anal. Chem. 74 (2002) 4763. [87] A.N. Naimushin, S.D. Soelberg, D.K. Nguyen, L. Dunlap, D. Bartholomew, J. Elkind, J. Melendez, C.E. Furlong, Biosens. Bioelectron. 17 (2002) 573. [88] A.N. Naimushin, S.D. Soelberg, D.U. Bartholomew, J.L. Elkind, C.E. Furlong, Sens. Actuators B 96 (2003) 253. [89] A.N. Naimushin, C.B. Spinelli, S.D. Soelberg, T. Mann, R.C. Stevens, T. Chinowsky, P. Kauffman, S. Yee, C.E. Furlong, Sens. Actuators B 104 (2005) 237. [90] J.J. Pan, D. Charych, Langmuir 13 (1997) 1365. [91] Q. Cheng, T.Z. Peng, R.C. Stevens, J. Am. Chem. Soc. 121 (1999) 6767. [92] T.Z. Peng, Q. Cheng, R.C. Stevens, Anal. Chem. 72 (2000) 1611. [93] Q. Cheng, S.M. Zhu, J. Song, N. Zhang, Analyst 129 (2004) 309. [94] T.H. Rider, M.S. Petrovick, F.E. Nargi, J.D. Harper, E.D. Schwoebel, R.H. Mathews, D.J. Blanchard, L.T. Bortolin, A.M. Young, J.Z. Chen, M.A. Hollis, Science 301 (2003) 213. [95] R. Casagrande, Scientic American, 2002, p. 82.

J.J. Gooding / Analytica Chimica Acta 559 (2006) 137151 [128] D. Kelly, K.M. Grace, X. Song, B.I. Swanson, D. Frayer, S.B. Mendes, N. Peyghambarian, Opt. Lett. 24 (1999) 1723. [129] S.P. Pogorelova, M. Zayats, A.B. Kharitonov, E. Katz, I. Willner, Sens. Actuators B 89 (2003) 40. [130] R.S. Marks, A. Novoa, D. Thomassey, S. Cosnier, Anal. Bioanal. Chem. 374 (2002) 1056.

151

[131] B. Leshem, G. Sarfati, A. Novoa, I. Breslav, R.S. Marks, Luminescence 19 (2004) 69. [132] L.H. Chen, D.W. McBranch, H.L. Wang, R. Helgeson, F. Wudl, D.G. Whitten, Proc. Natl. Acad. Sci. U.S.A. 96 (1999) 12287. [133] J.H. Thomas, S.K. Kim, P.J. Hesketh, H.B. Halsall, W.R. Heineman, Anal. Chem. 76 (2004) 2700.