Spectroscopic methods exploit different optical characteristics of molecules in different range of electromagnetic spectrum. The presence of Cyanobacteria toxic metabolites in natural water was determined by exploiting the color change of bromine thymol blue indicator [6]. Ho et al. used fluorescence for the detection of cholera toxin [7].

Mycotoxins are secondary metabolites produced by filamentous fungi that can have toxic effects on vertebrates. In order to improve food safety the presence of mycotoxins must be determined during the process of food and feed preparation. The fast screening is most conventionally done with ELISA. Quantitative methods of analysis for most mycotoxins use immunoaffinity clean-up with high performance liquid chromatography (HPLC) separation in combination with UV and/or fluorescence detection. There is a strong trend towards the use of multi-mycotoxin methods for the simultaneous analysis of several of the important mycotoxins, which is best achieved by LC–MS/MS (liquid chromatography with tandem mass spectrometry) [8].

Another physical method that can be used in toxin detection is amperometry. The principle of amperometry is based on the measurement of the current between the two electrodes, which is induced by a redox reaction at one of the electrode. The conditions are chosen in such a way that the current is directly proportional to the concentration of a redox active species in the analyte solution. An amperometric sensor system for measuring okadaic acid, a diarrheic shellfish poisoning toxin, has been developed recently [9].

Quartz crystal microbalance sensors for the detection of ricin were presented by Stine et al. [10]. Antibodies have been used traditionally as the detection molecules for these types of biosensors. However, recent studies have shown that glycosphingolipids may be used for recognition of certain protein toxins, including ricin. Glycosphingolipids had lower detection limits (5 μg/mL), approximately five times lower than were found for antibodies (25 μg/mL) [10]. Quartz crystal microbalance and amperometry detection were used in many cases for the detection of low molecular weight toxins in environment and food [11–13].

One of the methods that offer quick toxin detection is surface plasmon resonance (SPR). This is a relatively new technique, which became more popular with the commercialization of biosensors by the company Biacore in the 90’s. SPR is used in biochemistry to study molecular interactions and it provides significant advantages over other biophysical approaches [14–16]. SPR allows the kinetic parameters of the molecular interaction to be determined, but it is also used for quantitative purposes, i.e. for the determination of unknown concentrations of analytes in complex samples. The significant advantage over other biophysical approaches is that the interaction is studied in real time and without the need of labelling. The method is very sensitive, as concentrations in the picomolar range may be determined. The limit for the detection is around 200 Da, depending from the instrument used. Smaller molecules may also be detected with competition type assay or by including one or more amplification steps, which lowers the sensitivity of detection to femtomolar levels [17]. Finally, the technology can be included in portable instruments [18–20]. Field detection is desirable in monitoring and identification of biological agents as well as environmental pollutants. The portable SPR platforms were among other included in detection of ricin [21], enterotoxin B [22], 2,4-dichlorophenoxyacetic acid [23] and atrazine [24].

Biosensors based on SPR utilize a thin metal film between two transparent media of different refractive index, e.g. a glass prism and sample solution. Gold is preferably used in many SPR refractometers. A plane-polarized light beam entering the higher refractive index medium (glass prism) can undergo total internal refraction above a critical angle of incidence. Under these conditions, an electromagnetic field component of the light, the evanescent wave, will penetrate into the gold film. At a specific angle of incidence, interaction of this wave with free oscillating electrons at the gold film surface will cause the excitation of surface plasmons, resulting subsequently in a decrease in the reflected light intensity. This phenomenon is called surface plasmon resonance and occurs only at a specific angle of incident light. SPR system thus detects changes in the refractive index of the surface layer of a solution in contact with the sensor chip (Figure 1A).

SPR is observed as a sharp dip in reflected intensity at a certain angle, which is dependent on the refractive index of the medium on the non-illuminated side of the surface. This angle shifts when biomolecules bind to the surface and change the refractive index of the surface layer (Figure 1B). The sensorgram is a plot of the SPR angle against time, and displays the real-time progress of the interaction at the sensor surface (Figure 1C). Most of the SPR biosensors utilize response units (RU). The signal is proportional to the amount of the bound molecule (Figure 2). For proteins it was estimated that approximately 1,000 RU corresponds to surface coverage of 1 ng/mm² [25].

The SPR signal is sometimes not large enough to detect analytes at the desired level with an SPR sensor, and this may be a problem when detecting small molecules. The localized surface plasmon resonance effect may be employed, by using Au nanoparticles as a further signal amplification approach [26]. For example, substrate-selective polymer gels with gold nanoparticles as the molecular recognition elements were designed. The sensing is based on swelling of the polymer gel that is triggered by an analyte binding. This increases the distance between the gold nanoparticles and substrate, which causes the shift of an SPR dip [27]. This approach have been recently used in atrazine detection. Detection of 5 pM atrazine in acetonitrile was achieved [28].

The majority of publications using commercial SPR biosensor platforms employed Biacore technology [15,29], although other products also exist on the market that employ different optical and liquid-handling formats [30]. The essential components of a Biacore analytical system are the sensor chip, optical detector and an integrated microfluidic cartridge for the delivery of the sample to the surface of the sensor chip. Sensor chips are glass slides with a thin layer of functionalized gold. Biacore offers large variety of different sensor chips, that allow capture of almost any molecule. Hence, Biacore biosensors are readily used in basic science to study protein-protein, protein-DNA, protein-membrane, protein-small molecule, small molecule-membrane, etc., interactions [15,29,31]. The SPR imaging technology on array-based systems takes SPR analysis one step further. Biacore’s FLEXChip is an array-based system which can monitor up to 400 reactions simultaneously. In SPR imaging instruments, the SPR angles of many regions of interests may be monitored simultaneously by camera, which optically image the sensor surface.

The successful detection of toxins by SPR requires a binding partner. Preferentially, this should be a molecule which recognizes the toxin with great specificity and high affinity. Non-specific binding and cross-reactivity of other toxins or contaminants in the sample may largely be prevented by the careful choice of a binding partner. Examples presented in the literature mostly use antibodies specific for a certain toxin (e.g. reference [32]). Other molecules from the cell that are bound by toxins may also be used. Nakamura et al. have used a photosynthetic reaction centre of Rhodobacter sphaeroides to specifically detect atrazine [33]. Fonfria et al. have employed phosphodiesterases to detect yessotoxin, since it was shown that yessotoxin activates them in cells [34]. And finally, glycolipids were used as ligands for specific detection of ricin, due to the fact that ricin binds sugars present on cell-surface [35].

It is possible that toxins may non-specifically bind to the components of the sensor chip, i.e. to the dextran matrix and molecules associated with it. However, this should be accounted for by optimization experiments, where different types of sensor chips, buffers, additives, etc. are tested. It is also well-known that ligand immobilization on the sensor chip may affect unspecific binding. This may be accounted for by employing a different immobilization strategy.

In SPR the detector response is proportional to the mass of the analyte that binds to the ligand and, therefore, the responses resulting from binding of small molecules are low. SPR offers high sensitivity, the limit of detection for examples of toxin detection presented in this paper are in μM - nM concentration range. With the amplification it is possible to extend the limit of detection to pM - fM levels (see below). Direct measurement of the analyte may be performed when the analyte is large enough to produce significant responses even at low molar concentrations. Naimushin et al. showed direct detection of Staphylococcus aureus enterotoxin B (SEB) (Figure 2). The toxin with molecular weight 28.400 Da was detected at concentrations above 0.5 nM [17]. Competitive approaches may be more suitable for the analysis of small molecules (Figure 3). Two types of competitive assay formats are commonly used.

Surface competition methods employ a larger analogue of an analyte, usually a protein conjugate, which is used as a competitor for binding to immobilized ligand on the sensor surface. The second approach is a solution competition method, where a known amount of a detecting molecule, e. g. an antibody, is mixed with the analyte, and the amount of free detecting molecule remaining in the mixture is measured. In both competition approaches, the response obtained is inversely related to the concentration of analyte in the sample. The maximum signal is obtained when no free analyte is present. The calibration curve is constructed from the responses of the known concentrations of the analyte (Figure 4).

Both approaches were used in isoproturon determination [32]. The direct binding of various herbicide concentrations to the surface, immobilized with anti-isoproturon monoclonal antibodies, is shown in Figure 4A. However, the SPR response due to the binding of isoproturon was measured only for concentrations above 10 μg/mL. The surface competition method was found to enhance the detection. Isoproturon-BSA conjugate was used as a competitor in the indirect detection approach. After optimizing the conjugate concentration, the calibration curves were obtained using three different ligand densities on sensor surface (Figure 4B). The lower ligand density was shown to be more sensitive to isoproturon [32].

Most of the toxins that are detected with SPR have low molecular weight and indirect tests offer better detection limits for their detection in samples. Several assays employing SPR for measuring mycotoxin concentration were described (Table 1). The indirect test was used for the determination of aflatoxin B1 [36]. Aflatoxins are highly toxic fungal secondary metabolites from Aspergillus species, which may contaminate foodstuffs and feeds. A conjugate consisting of aflatoxin B1 and bovine serum albumin was immobilized on the dextran gel surface. The samples mixed with known concentration of polyclonal antibody against aflatoxin B1 were injected over the surface. When the concentration of aflatoxin B1 in the samples was high, the response signals were low. Such inhibition assay enabled detection of four different mycotoxins simultaneously within a time frame of 25 minutes, including the time needed for sensor regeneration [36].

Daly et al. [37] produced a polyclonal anti-aflatoxin B1 antibody, which could be regenerated after the binding of the analyte by using an organic solution consisting of ethanolamine with acetonitrile, and was, therefore, usable for detection of several samples. An SPR immunosensor was developed to determine concentrations of the fumonisin B1 in spiked samples. After optimization of the antibody overlayer surface, a detection limit of 50 ng/mL was obtained for the direct assay with an analysis time under 10 min [38]. A rapid deoxynivalenol specific biosensor assay was developed based on a simple procedure that includes extraction with acetonitrile and 10-fold dilution with the running buffer. The majority of the assay results were in agreement with LC-MS/MS results [39].

A broad range of synthetically produced chemicals can represent danger to health of humans or other organisms (Table 2). It includes herbicides, pesticides, antibiotic residues, etc. Atrazine is a widely used herbicide for growth control of grasses and crops. Farre et al. used a portable biosensor platform to detect atrazine in water with limit of detection at 20 ng/mL [24]. In another study, a direct detection system for atrazine inhibiting photosynthetic electron transfer was developed using the photosynthetic reaction centre from the purple bacterium. It was shown before that atrazine interacts with the D1 protein of the photosynthetic reaction centre [40]. Due to structural and functional homologies between plant and purple bacterial reaction centres the later was used in atrazine detection. Atrazine was detectable in the range from 1 to 100 mg/mL within 1 min. The obtained detection limit in this system, 1 mg/mL atrazine, was not enough to detect the guideline concentration in drinking water (2 ng/mL, World Health Organization). The authors suggest to improve the sensitivity with the competitive approach [33].

A simple SPR immunosensor, enabling parallel analysis of multiple analytes or multiple samples of an analyte, has been investigated for detection of a low-molecular-weight toxic substance, 2,4-dichlorophenoxyacetic acid. Low detection limit of 0.1 ppb (ng/mL) was established [23].

Gouzy et al. described the sensor for monitoring isoproturon, a selective and systemic herbicide. The detecting rat monoclonal anti-isoproturon antibody was reversibly immobilized through the use of a capture mouse anti-rat monoclonal antibody, which was covalently immobilized on the sensor chip surface. The capture of the anti-isoproturon antibody could be carried out up to 120 times (immobilization/regeneration cycles) without any evidence of activity loss. Using a strategy based on a surface competition assay easily enhanced the limit of detection. The sensitivity and working range of the indirect format were found to be dependent on the surface density of the anti-isoproturon monoclonal antibody. As expected for competitive formats, the lowest surface coverage allowed a lower detection of the herbicide isoproturon with a calculated limit of detection of 0.1 μg/L [32] (Figure 4).

A rapid and sensitive biosensor immunoassay was developed for residues of the antiparasitic agent ivermectin in bovine liver. A detection limit of 19.1 ng/g was achieved [41].

The antibiotics streptomycin and dihydrostreptomycin are used in modern agriculture for the treatment of bacterial infections in cattle, sheep, pigs and poultry. Their due detection in food is important because antibiotic residues could be harmful to consumers. The assay allowed the direct analysis of bovine whole milk, honey samples required dilution with buffer, while kidney and muscle samples from pigs were homogenized in an aqueous extraction buffer and clarified by centrifugation [42].

In recent years there has been an increase in the use of tylosin in apiculture due to resistance to oxytetracycline. Mass spectrometry based methods have been developed for the detection, but up until now there has been no complementary screening method available, capable of sub 10 μg/kg detection limits. Caldow et al. reported the assay with the detection limit 2.5 μg/kg [43]. Honey samples containing trace residue levels of tylosin were analyzed by both the biosensor screening method and an LC-MS/MS confirmatory procedure. The results were in good agreement [43].

The wide range of kits for screening for some other drug residues is commercially available on the market. They offer rapid, precise and highly sensitive determination. Studies have demonstrated that the results obtained with these kits correlates well with other confirmatory methods (http://www.biacore.com/food/products/biacore_q/qflex_kits/index.html).

An interesting approach for the detection of phenol, employing living cells, was presented by Choi et al. [44]. They fabricated the sensor surface with E. coli O157:H7 strain and exposed such surface to phenol. The damage following the injection of phenol had a significant impact on the total amount of intracellular biomaterials in attached cells, which induced the change of SPR signal. The detection limit was determined to be 5 ppm of phenol [44].

The mouse bioassay is most widely used to detect biotoxins produced by marine dinoflagellates in shellfish samples. Alternative methods of toxin detection were developed, due to animal welfare concerns. Fonfria et al. developed a new detection and quantification method for yessotoxin. The assay is based on the interaction of toxin with phosphodiesterase enzyme, which was immobilised on the sensor surface [34]. Detection protocols for okadaic acid [45] and saxitoxin [46] were also established. When the concentrations of okadaic acid determined with the biosensor method were compared with the values obtained by LC–MS in contaminated shellfish samples, the correlation between the two analytical methods was found to be highly satisfactory (r²=0.991) [45].

Bacterial and plant toxins have large molecular weight and can be detected directly when using SPR. Enterotoxin B is an exotoxin produced by Staphylococcus aureus. It is one of the toxins responsible for staphylococcal food poisoning in humans and has also been produced as a biological weapon. The toxin was determined in different samples, including milk, seawater and mushrooms (Table 4). Naimushin et al. designed the SPR platform that is ideal for the field use. The enterotoxin was detectable at sub-nanomolar levels and with amplification at femtomolar levels [17].

The SPR immunoassay was compared with an ELISA and the traditional bioassay for Clostridium perfringens β-toxin. The SPR detection proved to be less sensitive than ELISA, however, the sample treatment lasted only 20 min, while with ELISA it lasted six hours [47]. Tetanus toxin is large protein with molecular weight 150 kDa. Due to its large size its direct detection is easy even in samples with low concentrations [48].

Ricin represents one of the most potent plant toxins. Its detection has been performed by many methods, but not with SPR until recently. Feltis et al. developed a hand-held biosensor. They were able to detect ricin at 200 ng/mL, which is approximately 2,500 times less than the minimum lethal dose, in 10 min [21].

Two additional SPR-based assays for ricin detection were published in 2008. Tran et al. identified two suitable antibodies with strong affinity to all six ricin variants. The assay was linear over a wide range of ricin concentrations (up to at least 750 ng/mL) with a limit of detection of 0.5 ng/mL [49]. The fact that ricin binds cell-surface oligosaccharides was employed in another assay. Analysis using carbohydrate probes allowed researchers to detect the toxin in a highly sensitive and facile manner (10 pg/mL, 5 min) [35].

Chinowsky et al. have developed multi-analyte SPR instrument and showed detection of ricin A among other five distinct analytes ranging from small molecules to whole microbes during the course of a single experiment [22].

Abrin is a highly potent and lethal type II ribosome inactivating toxin from Abrus precatorius. It is very similar to ricin in structure and biochemical mechanism of action, but has been stated to be approximately 75 times more toxic in mice. The authors described the selection and characterization of two human monoclonal antibodies, which are capable of binding native abrin with high affinity and specificity. The limit of detection was 35 and 75 ng/mL [50].