The inherent ability of the phages to bind to their target pathogen has been exploited to design biosensor surfaces using physical and chemical functionalization. The physical functionalization is achieved by surface adsorption, which is simple and straightforward, but gives inconsistent and unstable immobilization density. Physical adsorption nonetheless has been used for detection of Staphylococcus aureus using lytic phages (detection limit ∼ 10⁴ cfu·mL⁻¹) by surface plasmon resonance (SPR) [22] and Salmonella using magnetoelastic sensor in suspension [23] and fat free milk (detection limit ∼ 10³ cfu·mL⁻¹) [24]. Similarly, physical adsorption of the phages on sugar and amino acid modified gold surfaces [25] as well as surface modified silica particles [26] for pathogen capture have also been reported in the literature. The ELISA-based binding strength study of phages versus monoclonal antibody against β-galactosidase in E. coli reveals that phages (K_d ∼ 21 ± 2 nM) bind to their target with similar or better affinity compared to monoclonal antibody (K_d ∼ 26 ± 2 nM) [27]. Despite these successful functionalization reports, strong chemical immobilization of phages on a sensor platform is preferred due to several advantages.

The anchoring of phages by chemical bonds on a biosensor detection platform is pertinent to development of a consistent and stable detection system. The advantage of surface modification and chemically anchored immobilization approach was revealed by a methodical study that demonstrates a 7-fold and 37-fold improvement in the phage density respectively, on cysteamine-modified and gluteraldehyde activated gold substrate compared to that by physical adsorption on bare gold susbstrate [25]. This two-step method was further improved by application of dithiobis(succinimidyl propionate) (DTSP) self-assembled monolayer (SAM) where the thiol group binds to the gold surface while the free succinimidyl interacts with the surface amine groups on the phages [28]. Silane chemistry has similarly been applied for silicon based substrates to facilitate P22 phage immobilization for Salmonella capture [29] as well as study of phage receptor-host ligand binding strength using atomic force microscopy [30]. In yet another example, electrochemical oxidation was used to generation of carboxyl group on carbon surface followed by amide coupling of T4 phages for subsequent E. coli capture [31].

Purity of the phage suspension is an important criterion to consider for chemical functionalization. Phages are amplified in their host bacterial culture to achieve high titers; and despite repeated centrifugation, the contamination of bacterial protein, lipids and carbohydrates could severely affect the efficiency of immobilization and binding ability of the phages. Phage lysate have therefore been purified by a host of methods such as ultra-high speed centrifugation [25], ultra-filtration [32], poly(ethylene glycol) precipitation-gradient centrifugation [33], chromatofocusing [34] and size exclusion chromatography [35]. An interesting study demonstrates that purified phage lysate can be essentially used to systematically study the binding kinetics of the phages on to an activated surface for chemical immobilization. Study with T4, P22 and NCTC 12673 phages model systems revealed that the phage binding kinetics does not follow the idealized and homogenous Langmuir adsorption isotherm but is governed by heterogenous adsorption closely related to Brouers-Sotolongo isotherm [35]. Such rigorous surface binding studies are extremely important for understanding of phage immobilization on a surface and cannot be realized in the presence of contaminations in the lysate. Wild-type phages have been extensively explored for functionalization of biosensor platforms and subsequent pathogen detection.

Wild-type intact phages however suffer from certain drawbacks that limit their application on biosensor platform. Intact phages are biologically active and thus result into the lysis of the host bacterium upon infection that would lead to loss of signal on a biosensor platform [36]. Besides, some phages show enzymatic activity towards their host bacterial surface receptor. For example, P22 phage shows endorhaminosidase activity towards the O-antigen on the surface of Salmonella enterica, which results in binding and subsequent detachment of the bacterium. Sf6 phages targeting Shigella flexnari shows similar endorhamnosidase activity [37]. Such enzymatic activity would lead to inconsistent signals on a biosensor platform, leading to changes in detection efficiency. Results also suggest that intact phages bound on sensor platform lose their bacterial binding capability upon drying. This could be explained due to the fact that intact phages collapse on the sensor surface upon drying, and consequently their tail fibers are unavailable to bind to the bacterial host [36]. In addition, intact phages have relatively large sizes, which limits their application as bioreceptors on particular sensor platforms such as in the surface plasmon resonance based sensor where detection signal is distance dependent.

The ability to manipulate the genetic material of microorganisms has paved the way for tremendous possibilities of creating novel recognition systems for biosensor applications. The following section will discuss the prospective application of genetically modified phage based recognition elements as bio-probes for pathogen detection.

Surface Plasmon Resonance is the oscillation phenomenon that exists at the interface between any two materials. SPR sensors measure the refractive index near the sensor surface that changes as a result of interaction of target analyte in solution with bioreceptors on transducer surface. SPR has been widely used for real time monitoring of biochemical interactions of small analyte such as DNA hybridization, cell-ligand, protein-peptide, and protein-lipid. SPR systems have also been modified to enable the direct label-free detection of larger biomarkers such as bacterial pathogens. Bacteriophages have been immobilized on SPR sensor surface as probes to provide specificity of recognition for detection of bacteria. The successful detection of S. aureus [22], E. coli K12 [28], E. coli O157:H7 and methicillin-resistant Staphylococcus aureus (MRSA) [69] have been demonstrated on phage-immobilized SPR methods. The limit of detection was typically in the range of 10²–10³ cfu·mL⁻¹.

Bacteriophage receptor binding proteins have also been used as biorecognition probes on SPR platforms for specific detection of bacteria. For example, Singh et al. immobilized genetically engineered tailspike proteins (TSP) from P22 bacteriophage onto the gold-coated SPR plates, and demonstrated a selective real-time detection of Salmonella with the sensitivity of 10³ cfu·mL⁻¹ of bacteria [36]. They also developed a similar detection platform for detection of Campylobacter jejuni bacteria by immobilizing the receptor binding protein (RBP) of Campylobacter bacteriophage NCTC 12673 on SPR plates [64]. They expressed GP48 RBPs as a glutathione S-transferase-Gp48 (GST-Gp48) fusion protein and used glutathione self-assembled monolayers (GSH SAM) to immobilize onto surface plasmon resonance (SPR) surfaces. They could thus achieve a limit of detection of 10² cfu·mL⁻¹ [64].

Bioluminescence assays are sensitive, rapid, and simple techniques for the quantitative detection of bacteria in samples by measuring the level of light emission from intercellular components. The first step of this assay is bacteria cell lysis to release interacellular components, which are then measured using a bioluminescent reaction with luciferase. The major drawback of this technique is the lack of specificity. The lytic phage is used as a recognition probe to detect and lyse the target bacteria. Blasco et al. developed an ATP bioluminescence assay for detection of E. coli and Salmonella Newport using lytic phage as bioprobe and lysis agent [70]. The sensitivity of bioluminescence assay was improved 10- to 100-fold when adenylate kinase (AK) was used as an alternative cell marker, and fewer than 10⁴ cfu·mL⁻¹ E. coli could be detected in less than 1 h [70]. A similar assay for Salmonella was slower and took up to 2 h [70]. Wu et al. showed that the amount of released AK form bacterial cells depends on the bacterial type, the growth stage, the phage type, and the infection time [71]. The use of lytic phage as biorecognition probe provides sensitivity and eliminates the need for lengthy conventional microbiological methods and selective media.

In this technique, fluorescence-labeled bacteriophages are used as staining agents for bacteria. The fluorescently stained bacteriophages recognize and bind to their host bacteria. The complex of phage-bacteria is then detected using flow cytometry or epifluorescent filter technique. The average sensitivity reported so far is around 10²–10³ cfu·mL⁻¹ for epifluorescent microscopy and is 10⁴ cfu·mL⁻¹ for flow cytometric detection [78–80]. Goodridge et al. combined this technique with immunomagnetic separation method, and could detect between 10 to 10² cfu·mL⁻¹ E. coli O157:H7 in artificially contaminated milk after 10 h enrichment [65] and 10⁴ cfu·mL⁻¹ concentration of E. coli O157:H7 in broth [66].

Edgar et al. [68] and Yim et al. [73] further improved the sensitivity of this approach by using fluorescent quantum dots (QD) to tag bacteriophages. QD improves the intensity and stability of fluorescent signal, and improves the sensitivity of detection platforms such flow cytometry and epifluorescence microscopy. The bacteriophage was engineered with biotin binding peptide on the head. The streptavidin coated QDs were allowed to bind strongly to the biotinylated phages. This method enabled detection of as low as 20 E. coli cells in 1 mL water sample in 1 h [68].

The fluorescent assays have also been used for detection of bacterial toxins. Goldman et al. applied phage display to select a 12-mer peptide that could bind to staphylococcal enterotoxin B (SEB), which causes food poisoning [72]. They could detect as low as 1.4 ng of SEB per sample well in a fluorescence-based immunoassay using a fluorescently labeled SEB-binding phages. Array biosensors were also developed based on a similar principle to simultaneously detect Bacillus globigii, MS2 phage and SEB [81].

A quartz crystal microbalance (QCM) is a very sensitive mass sensor with capability for detection of nanogram changes in mass. A QCM sensor is made of a thin piezoelectric plate coated on both sides with two metallic electrodes. The application of an electrical field across the quartz crystal excites the mechanical resonance. The fundamental wavelength (λ) and resonance wavelength (λ = 2d/n) are determined based on the plate thickness d, and thus the corresponding resonant frequency: (1) f n = n . f 0 = n . v 2 dwhere ν is the sound velocity, and f₀, f_n are the fundamental and the n^th overtone resonant frequency, respectively. The adsorption of mass onto the electrode surface shifts the resonance to lower frequencies. The rate of frequency change is proportional to the adsorbed mass according to the Sauerbrey Equation: (2) Δ f = − 2 f 0 2 Δ m A ( μ ρ ) 1 / 2in which μ is the shear modulus of quartz (2.947 × 10¹¹ g·cm⁻¹s⁻²), A is the piezoelectrically active crystal area, and ρ is the density of the quartz (2.648 g·cm⁻³). This relation is valid for thin rigid films with uniform mass distribution. Therefore, QCM sensors can be used to measure the mass of various target analytes by immobilizing specific probes on a sensor surface. Bacteriophage probes can be combined with these sensors for specific detection of bacteria. Olsen et al. showed that the physical adsorption of ∼3 × 10¹⁰ phages·cm⁻² on piezoelectric transducer surface provides a sensitive platform for rapid detection of Salmonella typhimurium. This phage-immobilized QCM sensor had a low detection limit of 10² cells·mL⁻¹ with a wide linear range of 10–10⁷ cells·mL⁻¹ and a rapid response time of less than 180 s [74].

Magnetoelastic sensors oscillate mechanically when an AC magnetic field is applied. The resonance occurs when the frequency of the applied field equals to the natural frequency of sensors. The fundamental resonant frequency of longitudinal oscillations is given by: (3) f = E ρ ( 1 − σ 2 ) 1 2 Lwhere E is the Young's modulus of elasticity, ρ the density of the sensor material, σ the Poisson's ratio, and L is the long dimension of the sensor. The addition of non-magnetoelastic material to the sensor surface dampens the mechanical oscillation shifting the resonance frequency to lower values: (4) Δ f = − f 2 Δ m Mwhere f is the initial resonance frequency, M the initial mass, Δm (smaller than M) the mass change and Δf is the shift in the resonant frequency of the sensor. The response of the magnetoelastic sensors can be measured in the absence of direct physical wire contacts to the sensor, making the possibility of real time and in vivo bio-detection systems possible.

Magnetoelastic sensors have been immobilized with filamentous bacteriophages for the detection of various bacteria including Salmonella typhimurium and Bacillus anthracis spores in different food matrixes such as fat free milk, and fresh tomato [24,67,75]. The limit of detection was typically in the range of 10³ cfu·mL⁻¹.

Amperometry is the most common electrochemical detection method for pathogen detection, and offers better sensitivity compared to other methods. Amperometric biosensors are composed of a reference electrode and a working electrode. A bias voltage is applied to these electrodes to produce a current in the analyte. The current produced directly depends on the rate of electron transfer, which changes with variation in ionic concentration of analyte. Amperometry detects ions in solution by measuring the changes in electric current. Many researched have reported amperometric detection of foodborne pathogens. Neufeld et al. combined amperometric technique with phage typing for specific detection of E. coli K12, Mycrobacterium smegmatis, and Bacillus cereous bacteria [77]. These sensors work based on the principle that the phage infection results in bacteria lysis leading to release of bacteria cell content, such as enzyme, into the surrounding medium. This enzymatic activity can be measured and quantified using specific substrate. The product of the reaction between substrate and enzyme is oxidized at the carbon anode at the reference electrode, producing a current. They could achieve limit of detection of 1 cfu·mL⁻¹ within 6–8 h using this technique in combination with filtration and pre-incubation before infecting bacteria with phage.

Electrochemicial impedance spectroscopy (EIS) biosensors measure the changes in impedance over a range of frequencies that occur as a result of biomolecular interaction. EIS biosensors have been exploited for bacterial detection by monitoring the changes in the solution-electrode interface due to the capture of microorganisms on the sensor surface. The capture of target analyte such as bacteria on sensor usually increases the impedance due to the insulating properties. Bacteriophages have been used as a crosslinkage between bacteria and electrode surface. For example, Shabani et al. showed successful detection of E. coli bacteria by immobilizing T4 phage onto the functionalized screen-printed carbon electrode with limit of detection of approximately 10⁴ cfu·mL⁻¹ [31]. They observed a decrease in impedance by increasing the bacteria concentration, which is contrary to normal attachment of intact cells on EIS sensor. The reason behind this observation was due to the lytic activity of phages that led to release of ionic intercellular content and increase in conductivity. The specificity of detection was confirmed by using Salmonella as negative control. In a further attempt, Mejri et al. observed two successive opposite trends over time for detection of bacteria on phage-EIS biosensors. The impedance initially increased due to the capture of bacteria followed by an impedance decrease attributed to phage-induced lysis. Such dual signals are inclusive to the specific detection of bacteria, and is easily distinguishable from those caused by non-specific binding. They specifically detected E. coli with the limit of detection of 10⁴ cfu·mL⁻¹ [82]. Although EIS offers label-free detection of pathogens compared to amperometry technique, its application for pathogen detection is limited due to its lower detection limit compared to other techniques.