Anal. Chem. 2009, 81, 3896–3902
Quantitative Detection of Staphylococcal Enterotoxin B by Resonant Acoustic Profiling Mohan Natesan,*,† Matthew A. Cooper,‡,§ Julie P. Tran,†,| Victor R. Rivera,† and Mark A. Poli† Integrated Toxicology Division, United States Army Medical Research Institute of Infectious Diseases, Fort Detrick, Maryland 21702, Cambridge Medical Innovations Ltd., Cambridge, CB4 0GJ, U.K., and Oak Ridge Institute of Science Education, Oak Ridge, Tennessee 37831 A rapid and sensitive detection of staphylococcal enterotoxin B (SEB) was developed using a novel acoustic sensing technique: Resonant Acoustic Profiling (RAP), which utilizes high-frequency piezoelectric quartz resonators for monitoring biomolecular interactions. An automated four-channel instrument consisting of acoustic sensors covalently conjugated with anti-SEB antibodies was used. As the samples flowed across control and active sensors simultaneously, binding was measured as a change in the resonant frequency. The lower limit of detection (LLOD) for the label free direct format was 25 ng/mL. Detection sensitivity was increased by adding mass sequentially to the captured SEB on the sensor in the form of sandwich antibodies and biotin-avidin-based gold nanoparticles. The LLOD for the mass enhanced formats were 5 and 0.5 ng/mL of SEB, respectively. The lowest sensitivity corresponds to 1.3 fM in a 75 µL sample. The total assay time including the enhancement steps was less than 10 min. SEB was detected in both neat urine and PBS buffer-spiked samples, with linear correlations between resonant frequency signals and SEB concentrations (R2 of 0.999 and 0.998, respectively). No significant cross-reactivity was observed with homologue toxins SEA, SED, and TSST, but some crossreactivity was observed with the closely related toxin SEC1 when we used a polyclonal antibody in the assay. SEC1 cross-reactivity was not observed when a SEBspecific monoclonal antibody was employed in the assay. Thus the specificity of the assay presented here was dependent on the quality of the antibodies used. In addition to detection, we evaluated RAP’s ability to measure the toxin in unknown samples rapidly by measuring the initial binding rate of the interaction, thereby further shortening the assay time to 6 min. Resonant Acoustic Profiling (RAP) is an advanced label-free sensor technology that detects biomolecular interactions in real * To whom correspondence should be addressed. E.mail: mohan.natesan@ us.army.mil. Fax: 301-619-2348. † United States Army Medical Research Institute of Infectious Diseases. ‡ Cambridge Medical Innovations Ltd. § Current address: Institute for Molecular Bioscience, University of Queensland, St. Lucia, Qld 4072, Australia. | Oak Ridge Institute of Science Education.
3896
Analytical Chemistry, Vol. 81, No. 10, May 15, 2009
time.1 RAP exploits the quartz crystal microbalance (QCM) technique to analyze molecular binding, in which quartz crystal resonators are electrically driven to oscillate at high frequency. The oscillation is then modulated when molecules interact with receptors bound to the crystal. The change in signal indicates not only the presence of a molecule but also the specificity and affinity of the molecule with the receptor.2,3 Measurement of mass by using quartz crystal resonators was first examined by Sauerbrey,4 who showed that the frequency change of the crystal resonator is a linear function of the mass per area, ms, or absolute mass ∆m:4
∆fm ) -
f02 f02 ∆ms ms ) FqFq FqFq Ael
f0 is the resonance frequency of the unperturbed quartz resonator, Fq the frequency constant of the crystal (Fq ) f0dq), dq the thickness, Fq the mass density, and Ael the electrode size of the crystal resonator. The above equation is only valid for thin, solid layers deposited on the resonator. Initially, the QCM system was used for dry measurements; later on when suitable oscillator circuits were developed, it was possible to carry out measurements under liquid conditions.5 This method led to the use of QCM systems as biosensors to detect molecular interactions. A new equation was derived by Kanazawa6 to explain the relationship between density (Fl) and viscosity (ηl) of the liquid and the frequency of the quartz crystal resonator,5
∆f ) -fq3⁄2
Flηl πFqµq
where Fq and µq are the quartz density and shear modulus, respectively.6 In a two-layer system, these frequency shifts simplistically (ignoring viscoelastic and complex shear modulus contributions) add up to an overall shift:6 (1) Godber, B.; Thompson, K. S. J.; Rehak, M.; Uludag, Y.; Kelling, S.; Sleptsov, A.; Frogley, M.; Wiehler, K.; Whalen, C.; Cooper, M. A. Clin. Chem. 2005, 51, 1962–1972. (2) Thompson, M.; Hayward, G. L. IEEE. Int. Freq. Control Symp. 1997, 114– 119. (3) Janshoff, A.; Galla, H. J.; Steinem, C. Angew. Chem., Int. Ed. 2000, 39, 4004–4032. (4) Sauerbrey, G. Z. Phys. 1959, 155, 206–212. (5) Nomura, T.; Okuhara, M. Anal. Chim. Acta 1982, 142, 281–284. (6) Kanazawa, K. K. Faraday Discuss. 1997, 107, 77–90. 10.1021/ac900086t CCC: $40.75 2009 American Chemical Society Published on Web 04/17/2009
(
∆f ) ∆fm + ∆fl ) -f02
∆ms + FqFqAel
) ηlFl f0πµqFq
RAP exploits these basic principles of piezoelectric sensing and, in addition, implements a number of technical advances. These include the use of high-frequency (up to 90 MHz) resonators, submicroliter dead volume microfluidics, internal (on sensor) reference controls, and higher-capacity surface chemistries. Of particular note, the parallel in-line reference controls in the system facilitate subtraction of background or bulk (nonbinding) signals that can often mask specific binding signals. RAP has been shown to be useful in a variety of biomolecular analyses including detection, quantification, kinetics profiling, and elucidation of ion and drug-induced receptor conformational changes.7-9 RAP is potentially well-suited for the detection of biothreat agents by virtue of fast (3-10 min) response time and rapid quantification of the agent. In this paper, we explored utility of RAP in biothreat agent diagnosis with application to the detection of staphylococcal enterotoxin B (SEB). SEB is a 23-29 kDa protein and belongs to group of seven other staphylococcal enterotoxins (SEs) produced by Staphylococcus aureus. SEB is resistant to proteolytic digestion and hightemperature degradation. SEB is also a superantigen and can stimulate a large population of T cells. SEB causes common food poisoning and may also be involved with toxic shock syndrome and sudden infant death syndrome.10 SEB is classified as B-list agent by CDC11 due to its pyrogenicity and its ease of dissemination. SEB is also considered to be a significant aerosol weapon threat. The early detection and treatment to SEB exposure is critical to the preventive efforts of both military and civilian health departments. A number of assays using biosensors and other methods have been developed for SEB detection.12-21 These technologies include surface plasmon resonance (SPR), electrochemiluminence (ECL), enzyme-linked immunosorbent assay (ELISA), cantilevers, classical QCM, fiber optic, carbon nanotubes, silicon nanowire, and time-resolved fluorometry (TRF). Although most of these assays are useful, many of them have limitations, including long processing times, multiple labor intensive processing steps, or poor detection sensitivity. (7) Cooper, M. A.; Singleton, V. T. J. Mol. Recognit. 2007, 20, 154–184. (8) Uludag, Y.; Li, X.; Coleman, H.; Efstathiou, S.; Cooper, M. A. Analyst 2008, 133, 52–57. (9) McBride, J. D.; Cooper, M. A. J. Nanobiotechnol. 2008, 6, 5–12. (10) Newbould, M. J.; Malam, J.; McIllmurray, J. M.; Morris, J. A.; Telford, D. R.; Barson, A. J. J. Clin. Pathol. 1989, 42, 935–939. (11) Ferguson, J. R. JAMA, J. Am. Med. Assoc. 1997, 278, 357–360. (12) Tempelman, L. A.; King, K. D.; Anderson, G. P.; Ligler, F. S. Anal. Biochem. 1996, 233, 50–57. (13) Kijeck, T. M.; Rossi, C. A.; Moss, D.; Parker, R. W.; Henchal, E. A. J. Immunol. Methods 2000, 236, 9–17. (14) Lin, H. C.; Tsai, W. C. Biosens. Bioelectron. 2003, 18, 1479–1483. (15) Peruski, A. H.; Johnson, L. H.; Peruski, L. F. J. Immunol. Methods 2002, 263, 35–41. (16) Poli, M. A.; Rivera, V. R.; Neal, D. Toxicon 2002, 40, 1723–1726. (17) Slavik, R.; Homola, J.; Brynda, E. Biosens. Bioelectron. 2002, 17, 591–595. (18) Chatrathi, M. P.; Wang, J.; Collins, G. E. Biosens. Bioelectron. 2007, 15, 2932–2938. (19) Campbell, G. A.; Medina, M. B.; Mutharasan, R. Sens. Actuators, B 2007, 126, 354–360. (20) Mishra, N. N.; Maki, W. C.; Cameron, E.; Nelson, R.; Winterrowd, P.; Rastogi, S. K.; Filanoski, B.; Maki, G. K. Lab Chip 2008, 8, 868–871. (21) Yang, M.; Kostov, Y.; Rasooly, A. Int. J. Food Microbiol. 2008, 127, 78–83.
Appropriate clinical matrixes for SEB sampling are human blood, plasma, serum, and urine. Of these, urine is the optimal matrix because significant amounts of SEs are eliminated through urine.22,23 While SEB can be found in the patient’s blood/plasma/ serum, pre-existing antibodies may rapidly clear SEB from blood and may also compete with antibodies in the detection assay.24-26 Further, urine can be obtained easily and noninvasively for testing and no sample treatment is necessary for subsequent analysis. This renders the matrix suitable for rapid turn-around laboratorybased assays and also for diagnosis in the field and biothreat arenas where complex and invasive samples of blood can prove problematic. A significant advantage of acoustic sensors is that they are insensitive to refractive index changes that adversely affect optically based sensors such as SPR. This property permits RAP to analyze clinical samples that possess complex matrixes such as urine, serum, saliva, or blood with little or no sample processing.9 EXPERIMENTAL SECTION Reagents. Staphylococcal enterotoxin (SE) A, B, C1, D, and TSST were purchased from Toxin Technology, Inc. (Sarasota, FL). Stock (1 mg/mL) solutions were prepared in phosphatebuffered saline (PBS) and stored frozen at -20 °C. Working dilutions in the analysis matrix were made immediately before use. Bovine serum albumin (BSA) was obtained form Sigma Chemicals (St. Louis, MO). Normal human urine was purchased from Sigma Chemicals as a lyophilized powder and reconstituted in distilled water. Neutravidin was obtained from Pierce Technology (Rockford, IL). Rabbit and monoclonal antibodies (MAb344) against SEB were obtained from the Critical Reagents Program (CRP) Repository, Edgewood Chemical Biological Center, (Aberdeen Proving Grounds, MD) and Toxin Technology Inc., (Sarasota, FL). Hyperimmune polyclonal goat antibodies were produced at United States Army Medical Research Institute of Infectious Diseases (USAMRIID) and affinity purified as reported earlier.16 Normal rabbit IgG was obtained from Jackson Immunoresearch Laboratories (West Grove, PA). Biotin-Labeling of Antibodies. Antibodies were labeled with a water-soluble biotin analogue, sulfo-NHS-LC-biotin (Pierce Technology, IL), according to the manufacturer’s directions. Briefly, a 10-fold molar excess of sulfo-NHS-LC-biotin (10 mM) was reacted with antibody solution (1 mM) in PBS for 45 min at room temperature. The free unreacted biotin was removed from the biotin-antibody conjugate by dialysis for 36 h against PBS at 4 °C. Neutravidin-Labeling of Gold Nanoparticles. Neutravidin was physically adsorbed to 15 nm diameter gold nanoparticles (British Biocell International, Cardiff, U.K.). Briefly, 5 mL of gold nanoparticles were mixed with 3 mL of neutravidin (1 mg/mL in (22) Crawley, G. J.; Gray, I.; Leblang, W. A.; Blanchard, J. W. J. Infect. Dis. 1966, 116, 48–56. (23) Nagaki, M.; Hughes, R. D.; Keane, H. M.; Goka, J.; William, R. J. Med. Microbiol. 1993, 38, 354–359. (24) Jozefczyk, Z. J. Infect. Dis. 1974, 130, 1–7. (25) Jozefczyk, Z.; Robbins, R. N.; Spitz, J. M.; Bergdoll, M. S. J. Clin. Microbiol. 1980, 11, 438–439. (26) LeClaire, R. D.; Bavari, S. Antimicrob. Agents Chemother. 2001, 45, 460– 463.
Analytical Chemistry, Vol. 81, No. 10, May 15, 2009
3897
Figure 1. The RAPid-4 is an automated four channel instrument (Figure 1a) that applies the principle of QCM and integrates acoustic detection with a continuous flow of the microfluidic delivery system, a thermal control unit, and automated sample handling (Figure 1b).
50 mM phosphate buffer, pH 7.0) for 1 h at room temperature. Additional stabilization of the gold nanoparticles was achieved by adding 1% polyethylene glycol (PEG, MW 200 000, Fluka-SigmaAldrich, St. Louis, MO). The particles were washed three times in phosphate buffer by centrifugation at 16 000g for 30 min each. BSA was added to the gold nanoparticles-neutravidin to a final concentration of 0.2%. The gold conjugate was used at 2.6 × 1011 particles per mL dilution in RAP assays. Instrumentation. An automated four-channel RAPid-4 commercial instrument system (Figure 1) was used. Two pairs of resonating quartz crystals mounted on a quartz plate acted as the sensors. Current applied to quartz crystals caused oscillation at high frequency (16.5 MHz), and the resonant frequency was monitored in real time. Quartz crystal resonators, fabricated on a single wafer of piezoelectric material as described below, were combined with a rapid switching process between active areas employed to eliminate cross talk and interference. System performance was enhanced using a proprietary FPGA-based network analyzer with internal digital synthesizer, rf switches, and calibration elements. Before the signal was sent to the sensor interface, the impedance of the signal path was transferred to match the sensor interface impedance. A microfluidic delivery system allowed samples to be injected simultaneously to all four channels. One of the sensors was used as the “control” sensor 3898
Analytical Chemistry, Vol. 81, No. 10, May 15, 2009
and acted as a subtractive reference. The volume of each flow cell was 900 nL. Samples flowed through the system at 25 µL/ min. HEPES buffered saline (HBS, 10 mmol/L, pH 7.4) was used as the running buffer. A thermal control system maintained the temperature of the system at 25 ± 0.5 °C at all times. Baseline drift observed during the 10 min assay was 0.25 ± 0.15 Hz (n ) 10) after immobilization. Sensor Chips and Surface Preparation. AKTiv covalent sensor chips were prepared from standard quartz wafers using a proprietary protocol at CMI facilities (catalog number 04-02000008). The sensors are coated with proprietary planar surface chemistry optimized for biological compatibility, shear modulus, and penetration depth to maximize acoustic coupling of a binding signal to the sensor. An optimized elastomeric mounting was developed to minimize the impact of thermal and motional stress on the piezoelectric material, while simultaneously providing a submicroliter microfluidic dead volume above the sensor. All solutions used in the RAPid-4 were filtered through a 0.22 µm filter. The surfaces were activated with a 1:1 mixture of 400 mmol/L of 1-ethyl-3-[3-dimethlyaminopropyl]carbodiimide hydrochloride (EDC) and 100 mmol/L of N-hydroxysuccinimide (NHS). Equal volumes of EDC and NHS solutions were mixed and injected simultaneously across the sensor surfaces for 3 min. Affinity purified rabbit anti-SEB (Toxin Tech) and normal rabbit IgG (Jackson Immunoresearch) were injected at 50 µg/mL in buffer (10 mM sodium acetate, pH 5.5) for 3 min. The average response for immobilized rabbit anti-SEB was 821 ± 11 Hz (n ) 20), indicating high consistency and reproducibility of protein coupling capacity of the sensor chips. Unreacted NHS esters were then capped with 1 M ethanolamine, pH 8.5. Finally, a 1% (w/v in PBS) BSA solution was pumped through the sensors to block any remaining free sites on the surface of the sensor chips. SEB Binding Assay. SEB samples (HBS and urine spiked) were injected over sensors immobilized with rabbit anti-SEB and control rabbit IgG. Signal amplification was achieved by injecting free goat anti-SEB to form a sandwich. In another format, biotinlabeled monoclonal antibody (MAb344, anti-SEB) was used to form a sandwich, followed by another injection of neutravidinlabeled gold nanoparticles. Each injection was for 3 min. Sensor surfaces were regenerated using a 30 s exposure to 100 mM HCl followed by 60 s exposure to a 20 mM NaOH solution. Regeneration conditions were optimized to provide greater than 95% regeneration efficiency, as determined by the response observed over several cycles of binding and regeneration. All samples were tested in triplicate. HBS was used as a blank vehicle control. The data were analyzed by RAPid-4 Workbench Software v1.0.25. The normal rabbit IgG antibody immobilized “control” sensor responses were subtracted from the rabbit anti-SEB immobilized “active” flow cell responses to normalize for any bulk shift changes. All responses were then superimposed and aligned to the start of the test sample injection. Initial binding rates were calculated from a linear fit model applied to the initial 10-15 s of each binding curve. Initial binding rate slopes were plotted against SEB concentration to generate a concentration-response calibration curve. SEB concentrations in unknown samples were determined by interpolation from this calibration curve. The sensitivities of the SEB detection assays were calculated using the average background plus 3 standard deviations. The maximum binding
Figure 2. Signal amplification response of the RAP SEB assay to different amplification formats. Reagents were sequentially injected as shown in the figure. After 1 µg/mL of SEB was captured on the sensor, signals were amplified by injecting biotin-labeled monoclonal antibody (MAb344), followed by neutravidin-conjugated gold nanoparticles (solid line). The control sensor (dashed line) was immobilized with normal rabbit IgG. The insert shows a schematic representation of the different steps used in RAP signal amplification formats.
responses were calculated by averaging data for 10 s after 180 s of sample injection. Statistical analysis was performed using Microsoft Excel. RESULTS AND DISCUSSION Examples of robust high-sensitivity sensing using acoustic wave devices in the literature are rare due to the significant demands on system sensitivity for low-mass detection. The purpose of this study was to demonstrate the suitability of RAP technology for the rapid and sensitive detection of biothreat toxins. We selected SEB, a class B list toxin as our model and evaluated the detection in a clinically relevant matrix (urine). The RAPid-4 instrument with four individually addressable sensors was used in a format, in which one sensor acted as the “control.” This sensor was immobilized with normal rabbit IgG. A number of assay parameters were evaluated including the best capture antibodies, of the best antibody pairs for a sandwich format, and the use of biotin-labeled antibodies for gold nanoparticles-based amplification. The combinations generating best signal-to-noise ratio were used in subsequent RAP assays for SEB. Three different SEB detection formats were used to show the sensitivity of RAP technology. Figure 2 depicts the detection of SEB using (sequentially) direct toxin binding, sandwich antibody, and then gold nanoparticle amplification assays. Because acoustic sensors fundamentally detect mass changes, adding additional mass to the captured SEB increases the signal observed. Gold nanoparticle promoted signal amplification has been employed in both SPR and QCM8,27 The sandwich format, where another SEB antibody was added over the captured SEB resulted in a 3-fold signal increase, compared to the direct format. The addition of a neutravidin-labeled gold nanoparticle conjugate improved the signal 5-fold over the direct format. These results demonstrate (27) He, L.; Musick, M. D.; Nicewarner, S. R.; Salinas, F. G.; Benkovic, S. J.; Natan, M. J.; Keating, C. D. J. Am. Chem. Soc. 2000, 122, 9071–9077.
Figure 3. Direct assay format RAP frequency response to SEB diluted in buffer at concentrations shown. Data are aligned and overlaid responses for 3 min of flow of SEB samples on sensors immobilized with rabbit anti-SEB. Binding curves for five different SEB concentrations in triplicate are shown. The control channel was immobilized with normal rabbit IgG.
that the sensitivity of the assay can be improved substantially by adding mass to the captured analyte. It is possible to add more mass sequentially than what we have shown here and still monitor binding signals, unlike SPR where the surface plasmon evanescent wave-sensing distance is limited to a short distance (
