1.

Introduction

Bacillus anthracis (B. anthracis), a pathogen of a zoonotic infectious disease—anthrax, has posed a threat of becoming biological weapons of mass destruction (BWMD).¹ Such terrorization originates from the bacterial virulence factors. These factors include toxins and antiphagocytic capsular polypeptides.^{2, 3} They are encoded by genes harbored in two virulence plasmids, pXO1 and pXO2.⁴ pXO1 carries the genes for three toxin proteins: cya for edema factor, lef for lethal factor, and pagA for protective antigen. These proteins, acting in binary combinations, cause edema and cell death in the host, but nontoxic alone.^{5, 6} pXO2 harbors three genes essential for capsule synthesis.⁷ The capsule prevents phagocytosis of the vegetative bacterial cells and inhibits the host defense mechanisms.⁸ The vegetative bacilli armed with these virulent factors are culprits of the bacteremia and the lethal toxemia as well as the host response, responsible for the overt symptoms of the host; if untreated, the infected individuals die in a few days. These attributes rank B. anthracis high on the list of BWMD and demand rapid detection for early treatment to minimize casualties.

With the advent of bioterrorism, a variety of nucleic acid-based biosensors have been developed. This approach generally constitutes immobilization of DNA probes onto a transducer surface of the biosensor, capturing of DNA analytes through complementary base-pairing, and monitoring of the reaction signals. The fluorescently labeled DNA probes have long been used to generate the signals. The drawbacks of the fluorescence-based biosensing include high cost, background noise, and prolonged operation time required for labeling. The effort to overcome the limits has stimulated the development of label-free biosensing technologies. ^{9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20} Surface plasma resonance (SPR) techniques have been developed for anthrax detection with antibodies,²¹ but DNA-based SPR detection of anthrax has not been reported.

This work aims to develop a DNA-based biosensor for fast detection of anthrax DNA that is label-free, sensitive, specific, and easy to operate. It is based on a prototype of a photonic crystal structure in a total-internal-reflection (PC-TIR) we developed recently.^{22, 23, 24} The PC-TIR configuration creates a unique open microcavity allowing biomolecules to easily access the sensing surface. In this study, the lef-derived single-stranded (ss) DNA was immobilized on the sensor surface to recognize its complementary strand versus the unrelated DNA.

2.

Experimental Procedures

2.1.

Design and Fabrication of PC-TIR Sensors

We used the following optical sensor fabricated for detection of the lef DNA. Specifically, we conceptually split a Fabry–Pérot etalon with a cavity layer sandwiched between two PC structures from the middle and use only half of the structure in a TIR geometry [Fig. 1a]. Owing to the TIR, there is a mirror image of the PC structure [Fig. 1b]. Thus, even with only half a conventional etalon structure, a cavity is still formed between the PC structure and its imaginary part in the TIR configuration. It is important to note that this is an open cavity [Fig. 1c] in contrast to the closed cavity in a conventional etalon. The existence of resonant modes similar to that of a Fabry–Pérot etalon cavity has been verified in our previous experiments utilizing the open cavity for real-time bioassays.^{22, 23, 24}

Fig. 1

Working principle of a PC-TIR sensor. (a) A cavity layer sandwiched by two pieces of PC structures. We conceptually split this structure from the middle layer into two pieces. (b) One piece of the PC structure is used in a TIR configuration. Owing to the TIR, an imaginary PC structure exists and a microcavity is still formed as if there were two pieces of the PC structures. (c) This PC-TIR sensor offers a unique sensing interface open for biomolecular assays.

We designed and fabricated a PC-TIR sensor based on the following theoretical calculations. The sensor is composed of a transparent substrate, a periodic structure of two alternating layers of different dielectric materials (silica and titania), and a polymer thin film as the cavity layer. Assume the incident angle at the substrate layer is θ _i, and the refraction angles in the substrate (S), silica (A), titania (B), and cavity layer (X) are θ _s, θ _A, θ _B, and θ _X, respectively. Let n _s, n _A, n _B, and n _X be the refractive index of the substrate, the silica, the titania, and the cavity layer, respectively. In order to form a photonic bandgap, the thickness of the dielectric multilayer has to satisfy,

Eq. 1

[TeX:] \documentclass[12pt]{minimal}\begin{document}\begin{equation} d_A = \frac{{\lambda _d }}{{4\left({n_A^2 - n_S^2 \sin ^2 \theta _S } \right)^{1/2}}},\quad d_B = \frac{{\lambda _d }}{{4\left({n_B^2 - n_S^2 \sin ^2 \theta _S } \right)^{1/2}}}, \end{equation}\end{document} $$d_A=\frac{{\lambda }_d}{4{\left(n_A^2-n_S^2{\sin }^2{\theta }_S\right)}^{1/2}},d_B=\frac{{\lambda }_d}{4{\left(n_B^2-n_S^2{\sin }^2{\theta }_S\right)}^{1/2}},$$

where d _A and d _B are the thickness for silica and titania, respectively. λ _d is the selected resonant wavelength, which was 632.8 nm in this experiment. An incident angle θ _s was chosen to be 64 deg. At this incident angle, titania and silica layers therefore have a designed thickness of 89.8 and 307.2 nm, respectively. They were used to form five alternating layers of the PC structure and fabricated with electron-beam physical vapor deposition on a BK7 glass substrate.

For obtaining a resonance condition in the cavity, the thickness of the cavity layer d _x should satisfy the following relation:

Eq. 2

[TeX:] \documentclass[12pt]{minimal}\begin{document}\begin{equation} 2 \cdot \frac{{2\pi }}{{\lambda _d }}n_x d_x \cos \theta _x + \alpha = (2m + 1)\pi \quad (m = 0,1,2,\ldots), \end{equation}\end{document} $$2·\frac{2\pi }{{\lambda }_d}{n}_x{d}_x\cos {\theta }_x+\alpha =(2m+1)\pi (m=0,1,2,...),$$

where α represents the Goos–Hänchen phase shift. The factor of 2 in the first term on the left-hand side is due to the fact that the light double passes this layer owing to the TIR. The integer m means multiple resonant conditions can be satisfied. For s-polarization of incident light, substituting the Goos–Hänchen phase shift expression into Eq. 2, one obtains the thickness of the cavity layer:

Eq. 3

[TeX:] \documentclass[12pt]{minimal}\begin{document}\begin{eqnarray} d_X &=& \frac{{\lambda _d }}{{4\pi \left({n_x^2 - n_S^2 \sin ^2 \theta _S } \right)^{1/2}}}\Bigg\{ 2m\pi + \pi - 2\tan ^{ - 1}\nonumber\\ &&\times \left[ {\left({\frac{{n_S^2 \sin ^2 \theta _S - n_H^2 }}{{n_X^2 - n_S^2 \sin ^2 \theta _S }}} \right)^{1/2}} \right] \Bigg\}, \end{eqnarray}\end{document} $$\begin{array}{ccc}\hfill d_X& =& \frac{{\lambda }_d}{4\pi {\left(n_x^2-n_S^2{\sin }^2{\theta }_S\right)}^{1/2}}\{2m\pi +\pi -2{\tan }^{-1}\hfill \\ & & \times \left[{\left(\frac{n_S^2{\sin }^2{\theta }_S-n_H^2}{n_X^2-n_S^2{\sin }^2{\theta }_S}\right)}^{1/2}\right]\},\hfill \end{array}$$

where n _H is the refractive index of the medium above the cavity layer. The cavity layer of the sensor was formed with a thin film (∼330 nm) of poly(methyl methacrylate) (PMMA) (A6, MicroChem), which was spin-coated on the PC structure at 500 revolutions per minute (rpm) for 10 s, followed by 4200 rpm for 45 s. This sensor structure was baked at 60°C for 30 min.

2.4.

Functionalization of the PC-TIR Sensor

The sensor chip was functionalized through the following steps for detection of the lef gene (Fig. 2): 1. Aldehyde groups were introduced onto the PMMA surface of the sensor to enhance immobilization of streptavidin on the sensor surface.^{25, 26, 27} Briefly, the PMMA surface was partly hydrolyzed with NaOH solution at 1N for 30 min at 55°C, and then the surface was fully rinsed with deionized (DI) water until the pH value reached 7. The substrate was dried and immersed in poly(ethyleneimine) (PEI) (MW = 75,000, Sigma-Aldrich) solution (2.5%, pH = 7) at room temperature for 1 h. After washing with DI water, the aminated substrate surface was further immersed in a 2% (w/v) glutaraldehyde solution for 30 min at 25°C. The surface was rinsed with copious DI water and then dried. 2. The PDMS wells were placed on top of the chip by gently pressing them together. Each well was filled with a 200-μl volume of phosphate buffered saline (PBS, pH 7). The resonant wavelengths of the two wells were measured and set as the detection baseline. 3. The PBS in the sample well was then replaced with 200-μl PBS solution containing streptavidin (40 μg/ml, from Invitrogen). After immobilization of streptavidin on the sensor surface had been completed, the well was washed twice with PBS. The amount of streptavindin bound to the sensor surface in a 200-μl volume of PBS at 25°C after washes was measured by the resonant wavelength shift of the sensor. Different from the case that such washes largely removed streptavidin from the unmodified sensor surface, the strong association of streptavidin with the chemically-modified sensor suggests covalent bonding of streptavidin to the surface through interaction of amino groups of streptavidin with the aldehyde groups on the sensor surface as reported.^{25, 26, 27} 4. Biotinylated single-stranded lef-probe DNA (100 nM in 200 μl PBS) was used to replace the PBS in the well for immobilization of the lef-probe on the surface via strong binding between streptavidin and biotin. After washed with PBS twice, the well was filled with 200-μl PBS. The amount of the biotinylated lef-probe bound to the surface was quantified by measuring the resonant wavelength shift. The sensor chip was thus functionalized with the lef DNA and ready for hybridization.

Fig. 2

Schematic illustration of the stepwise process involved in the surface modification and immobilization strategies for functionalization of the PC-TIR sensor surface.

3.

Results

We initially encountered technical challenges when immobilizing DNA probes onto the PMMA surface of the biosensor. Since the net charge of the PMMA surface is negative while the avidin is highly positively charged at pH 7.4, it is possible to directly immobilize avidin molecules on the PMMA surface via electrostatic absorption. We observed near-full coverage of avidin monolayer on the sensor surface. However, the signal for hybridization between biotinylated lef-probe and lef-com was unexpectedly low (see Sec. 4 for details). One likely cause was that the negatively-charged lef-probe tends to lie flat on the positively-charged avidin-coated surface as a result of electrostatic attraction. This effect may have hindered the lef-com and lef-probe hybridization. To solve this problem, we used streptavidin. In contrast to avidin, streptavidin is negatively charged at pH 7.4, thus preventing negatively charged DNA molecules from lying flat on the sensor surface. However, because PMMA has the same sign of charges as streptavidin, immobilization of streptavidin on the PMMA surface cannot be achieved via electrostatic adsorption. We therefore intended to immobilize streptavidin on the PEI-treated PMMA surface through covalent binding. Figure 3a shows several representative reflectance spectra of the sensor at different times during the immobilization of streptavidin. The spectra shift to longer wavelengths, indicating the increased amount of streptavidin immobilized on the sensor surface. Figure 3b plots the shift of resonant wavelength at different stages of the whole functionalization process of the sensor. The curve at stage I shows the detection baseline when PBS was injected on the sensor surface. The resonant wavelength started to shift when the buffer was replaced with 200-μl streptavidin (200 μg/ml in PBS) [Fig. 3b stage II]. The shift reached a plateau, indicative of the status of binding saturation. After washes, the curve at stage III shows the remaining streptavidin molecules bound to the surface.

Fig. 3

Detection of lef by functionalized PC-TIR sensor. (a) Representative resonant reflectance spectra of the sensor. Curve A, PBS injected on the sensor surface. Curves B, C, and D, corresponding to the spectra taken at time 0, 20 min, and 40 min, respectively, after injection of streptavidin onto the sensor surface. There was an immediate change in the resonant wavelength at time 0 because of the difference of refractive indices between PBS and the streptavidin solution. (b) The binding curve of streptavidin on the PEI-treated sensor surface. At stage I: a PBS solution was injected into the sensor well; Stage II: the resonant wavelength change due to the binding of streptavidin molecules to the aldehyde groups on the PEI-modified sensor surface; Stage III: the remaining streptavidin molecules covalently bound to the sensor surface after rinsing; Stage IV: the binding kinetics of biotinlated lef-probe with streptavidin on the sensor surface; Stage V: the bound lef-probe remaining on the sensor surface after rinsing. (c) Hybridization of lef-probe with the complementary lef-com (solid circle) and an unrelated rDNA control (open circle).

The surface is then ready for attaching the biotinylated lef-probe. For that, a 200-μl volume of the biotinylated lef-probe solution at a concentration of 100 nM in PBS was added to the sensor surface immobilized with streptavidin. The binding curve of the biotinylated lef-probe is shown at stage IV in Fig. 3b. After the sensor surface had been washed with PBS to remove molecules with nonspecific binding, the resonant wavelength shift changed from 4.6 to 3.5 nm. The curve at stage V indicates the amount of the lef-probe molecules immobilized on the sensor surface. Hence, functionalization of the sensor chip was completed and the sensor was ready for anthrax DNA detection.

We tested the functionalized sensor for detection of DNA hybridization between the immobilized lef-probe and the targeted lef-com. The temperature of the 200-μl lef-com solution (100 nM in PBS) was raised to 65°C before injection of the target ssDNA onto the lef-probe functionalized sensor surface in order to minimize formation of a secondary structure from ssDNA. After the temperatures had cooled down to allow lef-com to hybridize with the lef-probe on the sensor surface, the sensor was rinsed with PBS, and the reflectance spectra of the sensor were measured. Figure 3c demonstrates the resonant wavelength change of the sensor resulting from the binding of lef-com to lef-probe. Based on our transfer matrix simulation of multilayer interference,²³ the 1.2-nm shift in the resonant wavelength corresponds to an averaged binding thickness of 0.8 nm of lef-com on the sensor surface.

In addition to hybridization of the biotinylated lef-probe with its complementary DNA, the unrelated DNA was used to test specificity of the sensor for lef detection. When a 200-μl volume of the rDNA control (100 nM in PBS) derived from A. baumannii was heated and injected as above, a shift of the resonant wavelength was only 0.2 nm, a reduction of 84% compared with lef-com [Fig. 3c]. The decrease most likely resulted from the reduced base-pairing between the lef probe and the rDNA control, given the fact of the low complementarities between the two ssDNAs. This result indicates that the functionalized PC-TIR sensor is highly specific in detection of anthrax DNA.

In order to determine the sensitivity of the sensor, different concentrations of lef-com were tested. The resonant wavelength shift of the sensor increased with increasing lef-com concentrations (Fig. 4). At 100 nM, the sensor response did not appear to increase linearly with the concentration of lef-com, suggesting binding saturation at this high concentration level. When the concentration of lef-com was as low as 0.1 nM, a clear response of the sensor was still observed.

Fig. 4

Sensor responses to lef-com binding at concentrations of 0.1, 1.0, and 100 nM.

4.

Discussion

In this study, a sensitive photonic-crystal–based biosensor was fabricated to allow a unique open microcavity for rapid label-free biochemical assays. For detection of the B. anthracis lef gene, the sensor was functionalized with the synthetic ssDNA, lef-probe, from the plasmid-borne lef gene of B. anthracis. Significant changes in the resonant wavelength of the sensor were observed when the complementary strand, lef-com, was recognized by the sensor. The detection sensitivity was obtained to be better than 0.1 nM of lef-com. Specificity of the functionalized sensor has also been confirmed by using a negative control of an unrelated ssDNA fragment. These results demonstrate that detection of the anthrax lef gene with the PC-TIR sensor is highly specific and sensitive.

One of the technical challenges for the nuclei acid-based biosensing technology is immobilization of DNA probes onto a biosensor. We initially attempted to take advantage of the avidin-biotin approach to immobilization of DNA. The reason was that the negatively-charged PMMA surface and the positively-charged avidin at pH 7.4 allow direct immobilization of avidin onto the PMMA surface through electrostatic adsorption. After a 200-μl volume of avidin (200 μg/ml in PBS) was injected into the sample well on the PMMA surface, we observed that the resonant reflectance spectra of the sensor shifted as large as 5.5 nm, which indicates almost a full coverage of a monolayer of avidin on the sensor surface.²⁸ Biotinylated lef-probe molecules were then injected on the avidin immobilized sensor surface. A further shift of the resonant wavelength by 0.6 nm was observed due to the binding of lef-probe to the sensor surface. However, the resonant wavelength only shifted by 0.15 nm when a 200-μl volume of lef-com (100 nM in PBS) was added onto the sensor surface. The data suggest that little hybridization of lef-com with lef-probe occurred on the sensor surface. One possible reason for the low hybridization rate may be attributed to the following facts. The immobilized avidin is positively charged at pH 7.4 while lef-probe is negatively charged so that electrostatic attraction may cause lef-probe to lie down on the avidin coated surface, hindering hybridization. Nevertheless, this problem was successfully solved in this study by immobilizing streptavidin via covalent binding.

The capability of directly detecting anthrax DNA is of high impact because it may allow further reconfiguration of the sensor for recognizing DNA not only in the vegetative cells but also in the endospores of B. anthracis. B. anthracis generates endospores, a dormant life form, while the vegetative form is aerobic, Gram-positive, and rod-shaped. The endospores are not only transmittable easily to humans but also resistant to harsh environments,²⁹ all of which have made the anthrax endospores a potential BWMD.³⁰ Once inhaled, the endospores are engulfed either by host defense cells where the spores germinate and assume vegetative forms. The vegetative bacterial cells are disseminated, causing septicemia and producing virulence factors. Thus, rapid detection of the endospores in the environments and the vegetative form of B. anthracis in human bodies is demanded so that early specific treatment can be administered to reduce fatalities. With the PC-TIR sensor functionalized for detecting anthrax DNA, we believe this work has demonstrated a significant progress in the label-free photonic-crystal biosensing technologies, a prerequisite for rapid detection of BWMD.