Method for detecting an unknown contaminant concentration in a substance

Abstract

A method for detecting contaminants in a solution by determining a change in resonant frequency (ΔF) and motional resistance (ΔR) of a crystal microbalance (CM) immunosensor is presented. The method includes measuring ΔF and determining ΔR of a CM immunosenor exposed to various samples including known concentrations and a sample including an unknown concentration of the contaminant. The unknown contaminant concentration may be determined according to ΔR of the samples with the known and unknown contaminant concentrations, or ΔF of the same. If ΔR of the CM immunosensor exposed to the samples with the known contaminant concentrations more accurately reflects the known contaminant concentrations than ΔF does, the unknown contaminant concentration may be determined according to ΔR of the samples with the known contaminant concentrations and the unknown contaminant concentration. Otherwise, the unknown contaminant concentration may be determined according to ΔF of the same.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims priority of U.S. provisional patent application No. 60/642,335, filed Jan. 7, 2005. This provisional application is incorporated herein by reference in its entirety.

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH

The invention was made with United States government support under Grant Number USDA/CREES 99-34211-7563 awarded by the United States Department of Agriculture. The United States government has certain rights in this invention.

BACKGROUND

Generally, quartz crystal microbalance (QCM) immunosensors for bacterial detection solely measure the resonant frequency, and the frequency shift is usually correlated to an elastic mass effect. The quartz crystal microbalance (QCM), as a simple yet powerful technique, has been widely employed in chemical and biological sensing. QCM can be designed as an immunosensor for directly detecting microorganisms without the need of labeled antibodies that are required in sandwich-type immunosensors. QCM immunosensors have been reported for rapid and specific detection of different bacteria. Most QCM immunosensors solely measure the resonant frequency (F₀) using the standard oscillator technique, and the frequency change (ΔF) is usually explained by Sauerbrey equation, which states that the decrease in F₀ (−ΔF) is linearly proportional to the increase in surface mass loading of QCM (Sauerbrey, 1959).

However, the Sauerbrey equation is applicable only to a thin (˜1 μm) and elastic film coupled to the crystal surface, where the mass loading can be up to 0.05% of the crystal mass. The Sauerbrey equation does not apply to the case of cells attached to the QCM surface, largely due to the softness and relatively large size of the cells. In addition to the mass effect, the changes of surface viscoelasticity and other factors also contribute to the frequency change. Due to the additive nature of these effects, the mass effect cannot be differentiated from others when only F₀ is tracked.

High-frequency impedance/admittance analysis can provide more detailed information about the surface/interface changes of QCM. A QCM can be represented by a Butterworth-Van Dyke (BVD) model, which is composed of a static capacitance (C₀) in parallel with a motional branch containing a motional inductance (L_m), a motional capacitance (C_m), and a motional resistance (R_m) in series. These parameters are determined by physical properties of the quartz crystal, perturbing mass layer, and contacting liquid, and can be obtained with a high-frequency impedance analyzer by fitting the measured impedance/admittance data to the BVD model. A simpler way to provide insights into the viscoelastic properties of the bound surface mass is to simultaneously monitor F₀ and R_m or F₀ and the dissipation factor D using a quartz crystal analyzer that is less expensive than the impedance analyzer. This method has been applied to study the behavior of adherent cells in response to chemical, biological, or physical changes in the environment.

The impedance analysis has been used to characterize a QCM immunosensor for detecting Salmonella Typhimurium. A magnetic force was utilized to collect the complexes of Salmonella-magnetic beads on the crystal surface, and R_m was found the most effective and sensitive among the four circuit parameters, which offered a detection limit of about 10³ cells/ml. The sensitivity of the QCM immunosensor in the absence of magnetic beads has not been investigated nor has the measurements of R_m and F₀, therefore it is unclear how much the magnetic beads could affect the detection sensitivity or which of the F₀ and R_m measurements is superior in the presence or absence of the beads.

SUMMARY

A method for detecting contaminants in a solution by determining both a change in resonant frequency (ΔF) and a change in motional resistance (ΔR) of a crystal microbalance (CM) immunosensor is presented. The method may be used to measure contaminants in samples whether or not the samples include immuno-beads, such as immuno-magnetic microbeads. The method generally includes measuring the change in frequency (ΔF) and determining change in motional resistance (ΔR) of a CM immunosenor exposed to various samples that include known concentrations of a contaminant of interest. The method further includes measuring the ΔF and determining the ΔR of a CM immunosensor exposed to a sample containing the contaminant of interest at an unknown concentration. The measurements of ΔF and ΔR create data that relate a given ΔF to a known contaminant concentration and a given ΔR to a known contaminant concentration, respectively.

The unknown contaminant concentration may be determined according to the ΔF of the samples with the known and unknown contaminant concentrations or the ΔR of the samples with the known and unknown contaminant concentrations. The unknown contaminant concentration may be determined according to the ΔR of the CM immunosensor exposed to the samples with the known contaminant concentration and the ΔR of the sample with the unknown contaminant concentration. Alternately, the unknown contaminant concentration may be determined according to ΔF of the CM immunosensor exposed to the samples with the known contaminant concentrations and the ΔF of sample with the unknown contaminant concentration. If ΔR of the CM immunosensor exposed to the samples with the known contaminant concentrations more accurately reflects the known contaminant concentrations, the unknown contaminant concentration may be determined according to the ΔR of the CM immunosensor exposed to the samples with the known contaminant concentrations and the ΔR of the sample with the unknown contaminant concentration. Otherwise, the unknown contaminant concentration may be determined according to ΔF of the CM immunosensor exposed to the samples with the known contaminant concentrations and the ΔF of sample with the unknown contaminant concentration.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a schematic diagram of the QCM test system (top) and vertical-sectional view of the flow cell (bottom).

FIG. 2 is a typical conductance (solid line) and susceptance (dashed) spectra of the QCM. 1) blank QCM; 2) after adsorption of Protein A; 3) after antibody immobilization; and 4) after incubation with 10⁸ cells/ml of S. Typhimurium. All the spectra were obtained in PBS with the same quartz crystal.

FIG. 3 is a bar graph of changes of F₀, R_m, C_m, L_m, and C₀ for the bindings of Protein A, antibody and bacteria, respectively.

FIG. 4 is a graph of temporal responses of resonant frequency and motional resistance during antibody immobilization (1) and the prolonged baselines in PBS (2).

FIG. 5 is a ΔF-ΔR diagram for the layers of protein A (circle), antibody (triangle), and S. Typhimurium cells (squares). An elastic mass effect is represented by line 1, and a pure viscosity-density effect is represented by line 2, which was obtained with 0˜20% sucrose solution. Line 3 shows the ΔF-ΔR relationship for 10⁵-10⁸ cells/ml of S. Typhimurium. Error bars represent standard deviations (n=3˜6, the same below).

FIG. 6 is a fluorescence image of a QCM surface with bound S. Typhimurium cells at 10⁸ cells/ml (bar=4 μm).

FIG. 7 is a graph of temporal responses of resonant frequency and motional resistance during bacterial detection without anti-Salmonella magnetic beads. 1) PBS; 2) Chicken meat sample; 3) 10⁸ cells/ml of E. coli K12 in PBS; 4)10⁷ cells/ml of S. Typhimurium in PBS; 5) 10⁷ cells/ml of S. Typhimurium in chicken meat sample.

FIG. 8 (A) is a bar graph of change of resonant frequency as a function of the cell concentration

FIG. 8 (B) is a bar graph of change of motional resistance as a function of the cell concentration. 1) Chicken meat sample; 2) 10⁸ cells/ml of E. coli K12; 3-6) 10⁵-10⁸ cells/ml of S. Typhimurim. Error bars represent standard deviations (n=3˜6).

FIG. 9 is a graph of temporal responses of resonant frequency and motional resistance during bacterial detection with anti-Salmonella magnetic beads. Numbers indicate concentrations (cells/ml) of S. Typhimurim in chicken meat sample.

FIG. 10 is a graph of temporal responses of motional resistance of the QCM immunosensor (1 and 2) and blank QCM (3) for different samples. 1) 10⁶ cells/ml of S. Typhimurim+anti-Salmonella magnetic beads; 2) 10⁶ cells/ml of S. Typhimurim without anti-Salmonella beads; 3) 10⁷ cells/ml of S. Typhimurim+anti-Salmonella beads.

FIG. 11 is a block diagram of a method for detecting an unknown contaminant concentration in a substance.

FIG. 12 is a block diagram of a method for determining a change in resistance of a CM immunosensor exposed to samples with known contaminant concentrations.

FIG. 13 is a block diagram of a method for determining an unknown contaminant concentration in a substance according to change in resistance or change in resonant frequency of a CM immunosensor exposed to samples with known and unknown contaminant concentrations.

DETAILED DESCRIPTION

A viscoelastic biosensor with enhanced sensitivity has been developed for the detection of bacterial pathogens such as Salmonella Typhimurium based on the use of immuno-magnetic microbeads and the measurement of motional resistance. For Salmonella Typhimurium, the biosensor can be fabricated using Protein A for the antibody immobilization. High-frequency impedance analysis indicates that the changes in resonant frequency and motional resistance (ΔF and ΔR) of the biosensor are significant while the changes in static capacitance, motional capacitance, and motional inductance are insignificant. ΔF and ΔR can be monitored simultaneously in real time during the biosensor fabrication and bacterial detection, and the ΔF˜ΔR diagram can be used to obtain insights into the surface characteristics. It is found that the immobilization of Protein A and antibody cause an elastic mass change while the binding of bacterial cells result in a viscoelastic change. In the direct detection of S. Typhimurium in food samples, ΔF and ΔR are proportional to the cell concentration in the range of 10⁵ to 10⁸, and 10⁶ to 10⁸ cells/ml, respectively. Using anti-Salmonella magnetic microbeads as a separator/concentrator for sample pretreatment as well as a marker for signal amplification, the detection limit is lowered to 10² cells/ml based on the ΔR measurement. There is no interference from E. coli K12 and the sample matrix.

This method is developed for food safety, and can be used in food inspection and monitoring during food processing, storage and market. With minor modification, this method can be adopted for detection of other pathogenic bacteria in food samples, including E. coli O157:H7, Listeria monocytogenes and Campylobacter jejuni. In addition to immediate applications in the food area, the method can also be used in clinical or environmental applications.

Current quartz crystal microbalance (QCM) immunosensors for bacterial detection, with few exceptions, solely measure the resonant frequency, and the frequency shift is usually correlated to an elastic mass effect. In this study, a QCM immunosensor was described for the detection of bacterial pathogens such as Salmonella Typhimurium with simultaneous measurements of the resonant frequency and motional resistance. In the case of Salmonella Typhimurium, the immunosensor was fabricated using Protein A for the antibody immobilization. High-frequency impedance analysis indicated that the changes in resonant frequency and motional resistance (ΔF and ΔR) of the QCM were significant while the changes in static capacitance, motional capacitance, and motional inductance were insignificant. ΔF and ΔR were monitored simultaneously in real time during the immunosensor fabrication and bacterial detection, and the ΔF˜ΔR diagram was used to obtain insights into the surface characteristics. It was found that the immobilization of Protein A and antibody caused an elastic mass change while the binding of bacterial cells resulted in a viscoelastic change. In the direct detection of S. Typhimurium in chicken meat sample, ΔF and ΔR were proportional to the cell concentration in the range of 10⁵ to 10⁸, and 10⁶ to 10⁸ cells/ml, respectively. Using anti-Salmonella magnetic beads as a separator/concentrator for sample pretreatment as well as a marker for signal amplification, the detection limit was lowered to 10² cells/ml based on the ΔR measurements; however, ΔF was not related to the cell concentration. No interference was observed from E. coli K12 or the sample matrix.

EXAMPLE

In this example, ΔF and the change in R_m (ΔR) of a QCM immunosensor were approximately simultaneously monitored for the detection of S. Typhimurium. In addition, the immunosensor with food samples were evaluated and the effect of immuno-magnetic beads on the detection sensitivity was investigated.

Anti-Salmonella CSA-1 antibodies (1 mg) were manufactured by Kirkegaard & Perry Laboratories (Gaithersburg, Md.). Dynabeads® anti-Salmonella (diameter 2.8 μm) were obtained from Dynal Biotech Inc. (Lake Success, N.Y.). Protein A-soluble, from S. aureus (cowan strain) cell walls, was supplied by Sigma-Aldrich (St. Louis, Mo.). Propidium iodide (PI) was purchased from Molecular Probes (Eugene, Oreg.). Phosphate buffered saline (PBS, 0.01 M, pH 7.4) containing 0.138 M NaCl and 0.0027 M KCl, and 1% (w/v) bovine serum albumin (BSA)-PBS (pH 7.4) were received from Fisher Chemical (Fair Lawn, N.J.).

Salmonella Typhimurium (ATCC 14028) as a target pathogen, and Escherichia coli K12 (ATCC 29425) as a competing bacterium were obtained from American Type Culture Collection (Rockville, Md.). The pure culture of S. Typhimurium or E. coli K12 was grown in brain heart infusion (BHI) broth (Remel, Lenexa, Kans.) at 37° C. for 20 h before use. The culture was serially diluted with physiological saline solution and the viable cell number was determined by conventional plate counting. S. Typhimurium was enumerated by surface plating on xylose lysine tergitol (XLT₄) agar (Remel, Lenexa, Kans.). E. coli K12 was enumerated using sorbitol-MacConkey (SMAC) agar (Remel, Lenexa, Kans.). The undiluted cultures were heated in a 100° C. water bath for 15 min to kill all bacteria, and then diluted with PBS or sample solution to desired concentrations for further use.

Chicken breast meat purchased from a local grocery store was used as a tested sample. An amount of 25 g chicken meat was put into a Whirl-plastic bag (Nasco, Fort-Atkinson, Wis.) containing 225 ml of 0.1% buffered peptone water (Difco, Detroit, Mich.) and stomached on a Seward 400 stomacher (Seward, UK) for 2 min. The mixture was filtered by cheesecloth and then centrifuged to remove large debris and particles. An aliquot of 9 ml of the resulting meat solution was added with 1 ml of 10⁹ cells/ml of heat-killed S. Typhimurium to make a sample solution of 10⁸ cells/ml, which was further serially diluted to the desired concentration with the meat solution.

The inoculated sample solutions were analyzed using the QCM immunosensor directly without any other treatment or after immuno-magnetic separation (IMS). In IMS, a total of 20 μl of anti-Salmonella beads (ca. 0.1 mg or 6.6×10⁶ beads) and 1.0 ml of sample solution containing 0-10⁷ cells/ml of S. Typhimurium were added into micro-centrifuge tubes and vortexed for several seconds. The mixture was incubated at room temperature for 60 min with a gentle mixing. Then, the micro-centrifuge tubes were loaded into MPC-S magnetic particle concentrators (Dynal Biotech) and allowed 3 min for separating the magnetic beads from the liquid matrices. The liquid part was discarded and the resulting immuno-complexes of beads and target bacteria were resuspended in 250 μl PBS for further test with the QCM immunosensor.

The immunosensor was fabricated by immobilizing anti-Salmonella antibodies on the gold surface of AT-cut quartz crystals (International Crystal Manufacturing, Oklahoma City, Okla.), which had a diameter of 13.7 mm, a polished Au electrode (5.1 mm diameter, 1,000 Å thickness) deposited on each side, and a resonant frequency of 7.995 MHz. The crystals were pretreated with 1 M NaOH for 20 min and 1 M HCl for 5 min in sequence to obtain a clean Au surface. After each pretreatment the crystals were rinsed by spraying ethanol and water successively, and dried in a stream of nitrogen. Each of the resulting crystals was mounted on a 70-μl acrylic flow cell (International Crystal Manufacturing) as shown in FIG. 1 with only one face exposed to the solution. ΔF and ΔR were simultaneously monitored in real time during antibody immobilization and S. Typhimurium detection using a QCA922 quartz crystal analyzer (Princeton Applied Research, Oak Ridge, Tenn.).

Protein A method was used for the antibody immobilization. First the crystal was flushed with 1 ml PBS to obtain a stable baseline. Secondly, the detection chamber was overflowed by 500 μl of 1 mg/ml Protein A. After 1 h incubation, the detection chamber was flushed with 1 ml PBS 5 times to rinse off the excess Protein A and to obtain a stable baseline. Thirdly, the chamber was overflowed by 500 μl of 200 μg/ml anti-Salmonella antibody solution. Also after 1 h incubation, the chamber was flushed with 1 ml PBS 5 times to rinse off the unimmobilized antibodies and to obtain a stable baseline. All the baselines were obtained in PBS at a stop-flow mode, and the differences between every two neighboring baselines were calculated as the net responses caused by the immobilization of Protein A and antibodies, respectively.

The QCM immunosensor was tested in a stop-flow mode for the detection of S. Typhimurium in PBS or chicken meat sample. First, the immunosensor was incubated with 1% BSA-PBS, blank chicken meat sample solution or 10⁸ cells/ml of E. coli K12 solution for 1 h to block nonspecific binding sites. Then the QCM was flushed with 1 ml PBS 5 times to obtain a stable baseline. Following this, the chamber was overflowed by 1 ml (without magnetic beads) or 250 g (with magnetic beads) of sample solution. After incubation for 1 h, the chamber was flushed with 1 ml PBS 5 times to rinse off nonspecific bindings and to obtain a stable baseline. The difference between the two PBS baselines, both obtained in a stop-flow mode, was correlated to the concentration of S. Typhimurium in the sample solution.

All the experiments were conducted at room temperature, and disposable 1-ml syringes were used to push the reagent/sample solution through the detection chamber.

The QCM sensor was connected to an HP 4291A impedance analyzer (Hewlett Packard Japan, Hyogo, Japan) via an HP 16092A test fixture. The conductance and susceptance spectra (G˜f and B˜j) were measured simultaneously under a linear frequency sweep mode with 201 frequency points and a frequency span of 10 kHz covering the resonant frequency. The measured G˜f and B˜f data were fitted to the BVD model using the following equations (Tan et al., 1999), $\begin{array}{cc}G\left(\omega \right)=\frac{R_m}{R_m^2+{L_m^2\left(\omega -{\omega }_0^2/\omega \right)}^2}& \left(1\right)\\ B\left(\omega \right)=\omega C_0-\frac{L_m\left(\omega -{\omega }_0^2/\omega \right)}{R_m^2+{L_m^2\left(\omega -{\omega }_0^2/\omega \right)}^2}& \left(2\right)\end{array}$
where ω=2πf, f is scanning frequency, ω₀=2πF₀, and F₀ is the resonant frequency (F₀=(2π√{square root over (L_mC_m)})⁻¹). The fittings, which involved minimizing the relative sum of the residual square (Xie et al., 1999) with R_m, L_m, C₀, and F₀ as estimated parameters, were executed using the Excel Microsoft Solver tool. C_m was calculated from F₀ and L_m. Typical fitted results of F₀, R_m, L_m, C_m, and C₀ were 7992987.4 Hz, 9.625 Ω, 17.897 mH, 22.154 fF, and 8.243 pF (including parasitic capacitance in the test fixture) for an unperturbed quartz crystal placed in the air, and 7991347.7 Hz, 346.3 Ω, 25.394 mH, 15.620 fF, and 10.738 pF for a bare crystal (mounted on the flow cell) in PBS, respectively. The values of R_m, L_m and C_m obtained in the air are close to those calculated from the equations (25b-d) of Martin et al (1991) with the relative deviations ranging between 1.5˜19%.

The fluorescence images were taken on Nikon Eclipse 600 Fluorescent Microscope (Nikon Instruments, Lewisville, Tex.) using the Nikon G-2A filter set. Before fluorescent microscopy, the QCM immunosensor was incubated with 10⁸ cells/ml of S. Typhimurium for 1 h. After being rinsed off non-specific bindings, the QCM surface was treated with 1% PI for 2 h to stain the specifically bound cells.

Although extensively used in surface/interface studies, the high-frequency impedance/admittance has been rarely applied to characterize the QCM immunosensor for bacterial detection. In this study, the QCM immunosensor was characterized step by step with the admittance analysis. FIG. 2 shows typical conductance (G) and susceptance (B) spectra of the same quartz crystal subjected to different treatments. All the spectra were measured in PBS after physical adsorptions had been rinsed off. It was demonstrated that both the maximum G (G_max) and the frequency at Gmax decreased with the layers of protein A, antibody and bacterial cells being deposited on the QCM surface successively. The decreased values obtained from FIG. 1A are ca. 0.009, 0.029, and 0.218 mS for the G_max and 50, 200, and 200 Hz for the frequency at G_max, respectively.

To acquire insights into the properties of the films deposited on the QCM surface, the measured admittance data were fitted to the BVD model to extract the values of the four equivalent circuit parameters C₀, L_m, R_m, and C_m along with F₀. Each equivalent circuit parameter has its distinct physical meaning (Martin et al., 1991; Buttry and Ward, 1992): C₀ reflects the dielectric properties between the electrodes located on opposite sides of the insulating quartz crystal; C_m represents the energy stored during oscillation, which corresponds to the mechanical elasticity of the vibrating body; L_m is related to the displaced mass; and R_m is the energy dissipation during oscillation, which is closely related to viscoelasticity of the deposited films and viscosity-density of the adjacent liquid (Muramatsu et al., 1988; Lee et al., 2002). The changes of these parameters are illustrated in FIG. 3 in terms of the signal-to-noise (S/N) ratio. The noise levels of F₀, R_m, C_m, L_m, and C₀, defined as the overnight variations in PBS, were 2.6 Hz, 0.62 Ω, 0.0064 fF, 0.0039 mH, and 0.0068 pF, respectively. As shown in FIG. 3, in all cases including Protein A adsorption, antibody immobilization, and bacterial binding, the change of F₀ was the largest and the change of R_m was the second while the change of C_m, L_m, and C₀ were insignificant with the S/N≦3. The same result was obtained in a repeated test using another quartz crystal. According to Martin et al. (1991), both R_m and L_m increase with simultaneous mass and liquid loading while C₀ and C_m are kept unchanged. L_m did not increase as expected, and this was likely due to the high correlation between L_m and C_m and the difficulty in accurate measurement of L_m (Yang and Thompson, 1993; Noel and Topart, 1994). Kim et al. (2003a) also observed a decrease in L_m, and they corrected the L_m as follows (Yang and Thompson, 1993), $\begin{array}{cc}{\left(L_m\right)}_{\mathrm{Corrected}}=\frac{{\left(L_m{C}_m\right)}_{\mathrm{liquid}}}{{\left(C_m\right)}_{\mathrm{unloaded}}}& \left(3\right)\end{array}$
where the corrected L_m is calculated based on the assumption that C_m is a constant. Since F₀=(2π√{square root over (L_mC_m)})⁻¹, the corrected L_m reflects the same information of F₀, and therefore it is not discussed below while F₀ and R_m are discussed. FIG. 4 shows simultaneous response courses of ΔF and ΔR during the surface modification with Protein A and antibodies successively along with the prolonged baselines in PBS. The surface modification was conducted in the stop-flow mode and PBS was used to rinse off physical adsorptions. To eliminate the background interference, a stable baseline was obtained in PBS before and after the injection of Protein A and antibodies, and the difference between every two neighboring baselines was calculated as the net response caused by the immobilization of Protein A and antibodies, respectively. Similar to the results of the above admittance analysis, the net response of F₀ is more obvious than that of R_m by comparing the S/N ratios.

Simultaneous measurements of ΔF and ΔR can differentiate an elastic mass effect from the viscosity-induced effects. ΔR is a good measure of the viscoelastic change. For an elastic mass change, ΔR will be zero and ΔF will be linearly proportional to the mass change in accordance with the Sauerbrey equation. For a QCM with only one side in contact with a Newtonian liquid, both ΔF and ΔR are linearly proportional to the squared root of the product of viscosity and density of the liquid. Hence, a pure viscosity-density change will result in a linear ΔF˜ΔR plot. As illustrated in FIG. 5, the elastic mass effect can be represented by line 1 where ΔR=0, and the pure viscosity-density effect can be represented by line 2, which was obtained with 0˜20% sucrose solution and had a slope of 0.24 Ω/Hz. In the presence of a viscoelastic change, the ΔR˜ΔF plot will lie between the lines 1 and 2 or even above the line 2. For co-existence of mass and viscosity-induced changes, the absolute value of ΔR/ΔF will be <0.24 Ω/Hz. In general, the smaller the absolute value of ΔR/ΔF, the more predominant the elastic mass effect. The ΔR˜ΔF data obtained in each step including those for specific bacterial bindings, which will be described later in more details, are presented in FIG. 5 to identify an elastic or viscoelastic change. The net responses of ΔF and ΔR caused by the coating of Protein A were −38.3±7.9 Hz (mean±S.D., n=3) and 1.7±0.5Ω, respectively, and the ratio of ΔR to −ΔF was 0.04 Ω/Hz, much smaller than 0.24 Ω/Hz. Considering the ΔR measurement in PBS had a noise level of ˜1Ω, the above ΔR change was approximately negligible, and the ΔF change caused by the adsorption of Protein A was attributed to an elastic mass effect. Corresponding to the immobilization of antibodies, the net responses of ΔF and ΔR were −187.6±17.0 Hz (mean±S.D., n=6) and 6.1±1.9Ω, respectively. Although an obvious ΔR increase was observed, the ratio of ΔR to −ΔF was only 0.03 Ω/Hz, also much smaller than 0.24 Ω/Hz. Hence, the ΔF change caused by the adsorption of antibodies was also primarily due to an elastic mass effect. Therefore, the Sauerbrey equation could be applied to estimate the surface coverages of Protein A and antibodies, which were calculated to be 0.26±0.05 and 1.28±0.12 μg/cm², respectively.

The ΔR˜ΔF data for the binding of Salmonella cells is displayed as line 3. At 10⁵ cells/ml, the ΔF change was obvious but the ΔR change was negligible, indicating an elastic mass effect. When the cell concentration was higher than 10⁶ cells/ml, both a negative ΔF shift and a positive ΔR shift were observed, and the ratio of ΔR to −ΔF was as high as 0.16˜0.48, close to or larger than the slope of the pure viscosity-density response line (line 2). Thus, the layer of bound cells was viscoelastic and the ΔF response did not obey the Sauerbrey equation. Such viscoelastic changes were also observed on certain polymeric films and cells (Zhou et al., 2000).

FIG. 6 shows typical fluorescence image of the QCM surface after it was incubated with 10⁸ cells/ml of S. Typhimurium and stained with PI, a nucleic acid stain for deal cells. The distribution of S. Typhimurium cells is roughly uniform, indicating an approximately homogenous immunosensing surface. The surface coverage of the bound cells was measured to be ca. 9×10³ cells/mm².

The QCM immunosensor was tested in a stop-flow mode for direct detection of S. Typhimurium in PBS as well as in the stomaching solution of chicken meat without using magnetic beads. Typical responses of ΔF and ΔR are given in FIG. 7. Also non-specific bindings were rinsed off by PBS and the interference from sample matrixes was excluded by obtaining a stable PBS baseline before and after the injection of sample solutions. The difference between two adjacent baselines was calculated as the net response induced by specifically bound bacteria. The net responses of ΔF and ΔR caused by 10⁷ cells/ml of S. Typhimurium in PBS (curve 4) or in chicken meat sample (curve 5) were both significantly distinguishable from the negative controls (curves 1˜3). For 10⁷ cells/ml of S. Typhimurium, the net response of ΔR in PBS and that in the chicken meat sample were almost the same, but the net response of ΔF in the chicken meat sample was greater than that in PBS, which might be ascribed to the sensor-to-sensor variations and/or the difference of sample backgrounds. Although the QCM surface was pre-saturated with a blank solution of the chicken meat sample to depress the nonspecific bindings of the sample matrix, the Salmonella cells might carry some food debris or particles and therefore caused lager ΔF response.

The calibration data for the detection of S. Typhimurium in an inoculated chicken meat sample based on the measurements of ΔF and ΔR are presented in FIGS. 8A and 8B, respectively. Chicken meat was selected as a representative sample of poultry product, one of the major vehicle foods of S. Typhimurium. As can be seen, the net responses of ΔF and ΔR were proportional to the concentration of S. Typhimurium in the range of 10⁵ to 10⁸, and 10⁶ to 10⁸ cells/ml, respectively. The ΔF measurement was more sensitive than the ΔR measurement in terms of the detection limit (FIG. 8) as well as in the S/N ratio (FIG. 3). This might be due to the fact that both elastic and viscoelastic mass change resulted in a negative ΔF while the former does not cause a change of ΔR. The layer of the bound Salmonella cells is viscoelastic, but a small portion of the cells can still be probed as an elastic mass.

The relative standard deviations of the sensor-to-sensor determinations varied between 12˜29% (n=3˜6) for the ΔF measurement and 1.5˜28% for the ΔR measurement, respectively. The sample matrix did not interfere with the detection of S. Typhimurium in both the ΔF and ΔR measurements, nor did E. coli K12 although at a concentration as high as 10⁸ cells/ml.

The QCM immunosensors for bacterial detection reported previously typically have a detection limit ranging between 10⁵ and 10⁷ cells/ml and a detection time of minutes to several hours. In this study, without using magnetic beads, a detection limit of 10^5˜106 cells/ml was obtained for the direct detection of S. Typhimurium. However, the infectious dosage of a foodborne pathogen such as S. Typhimurium can be as low as 15-20 cells.

In this example, the effect of magnetic beads was investigated in different ways: the magnetic force was only used for separating the Salmonella-bead complexes from sample matrix, the bead complexes were inducted to the QCM surface simply using a syringe, and ΔF and ΔR were simultaneously monitored in real time. This avoided the use of a complicated test chamber with a magnet and an ultrasonic transducer and the inconvenience of discontinuous impedance measurements.

FIG. 9 shows typical temporal responses of ΔF and ΔR for the detection of S. Typhimurium in chicken meat sample using anti-Salmonella magnetic beads.

A firm and tight attachment of bacteria causes a negative ΔF, in contrast, a flexible attachment results in a positive ΔF. In the absence of anti-Salmonella magnetic beads, the former case applied as the specific binding of Salmonella cells always led to a negative ΔF that was proportional to the cell concentration. In the presence of anti-Salmonella beads, however, ΔF was not related to the cell concentration and was either positive or negative even at the same cell concentration. This was probably because the size of the Salmonella-bead complexes was not uniform from sample to sample. S. Typhimurium is a straight rod bacterium. Typically, the width of a Salmonella cell is 0.7-1.5 μm and its length is 2-5 μm. The magnetic beads used had a diameter of 2.8 μm. The size of the Salmonella-bead complexes thus varied from several microns to tens of microns. Small complexes might generate a tight attachment and a negative ΔF, and oppositely, large complexes and aggregates could cause a flexible attachment and positive ΔF.

A significant net increase in ΔR was always observed at a cell concentration higher than 10² cells, and the net response increased with increasing cell concentration. The effect of anti-Salmonella magnetic beads can be seen more clearly in FIG. 10. The background interference was eliminated by achieving a stable baseline in PBS before and after the sample injection. At the same cell concentration of 10⁶ cells/ml, the net change of ΔR in the presence of anti-Salmonella magnetic beads (curve 1) was positive and 2 times that in the absence of anti-Salmonella beads (curve 2). The blank QCM (without immobilized antibodies) did not give a significant increase in ΔR although tested at a higher cell concentration of 10⁷ cells/ml with anti-Salmonella beads (curve 3). The detection limit based on the ΔR measurement was ca. 10² cells/ml, and this was 1,000 and 10,000 times lower than those of the ΔF and ΔR measurements without magnetic beads, respectively. The improvement of detection sensitivity is attributed to the dual roles of magnetic beads. As a separator/concentrator, the beads separated the target bacteria from the sample matrix and concentrated the sample solution. As a marker, the beads amplified signals of ΔR measurement through increasing the surface viscoelasticity.

ΔF measurement is more sensitive than the ΔR measurement in the direct detection of S. Typhimurium. When magnetic beads were used, however, the ΔR measurement was more reliable, and the sensitivity was improved by 1,000˜10,000 times. The detection limit based on the ΔR measurement was approximately 10² cells/ml, lower than those of the most reported QCM immunosensors for bacterial detection. It was also shown that simultaneous measurement of ΔF and ΔR could provide insights into the surface characteristics: the layers of immobilized Protein A and antibodies were dominantly elastic, the layer of specifically bound Salmonella cells was viscoelastic, and the magnetic beads might increase the viscoelasticity. The QCM immunosensor was successfully applied to the analysis of inoculated food samples with negligible interference form E. coli K12 and the sample matrix.

The same principle can be applied to detect other pathogenic bacteria in food, environmental and clinical samples using specific primary antibodies and immuno-magnetic beads. For example, it can be used to detect infectious bacteria in human blood and urine samples, and pathogenic bacteria in water of rivers, wells and reservoirs. It provides a rapid, sensitive, specific, inexpensive and portable biodetection method for applications in food safety and security, environmental protection and clinical diagnoses.

In addition to the microbeads, immuno-magnetic nanobeads and other types of magnetic beads can be used in this procedure for a QCM immunosensor in the detection of various pathogens. The similar detection limit, time, specificity and sensitivity are expected.

As mentioned above, QCM immunosensors, and crystal microbalance (CM) immunosensors in general, may be used in detecting contaminants in a substance. These contaminants include pathogens, bacteria, viruses, insects, arachnids and other undesirable items. An example of such a method is shown in FIG. 11. The method 1100 illustrated in FIG. 1 may be used to measure samples of a substance. The samples may include some type of immuno-bead such as, as magnetic, micro, nano and any combination of such.

As shown in FIG. 11, the method 1100 generally includes measuring the change in frequency (ΔF) 1102 and determining change in motional resistance (ΔR) 1103 of a CM immunosenor exposed to various samples that include known concentrations of a contaminant of interest. The method 1100 further includes measuring the ΔF and determining the ΔR of a CM immunosensor exposed to a sample containing the contaminant of interest at an unknown concentration. The measurements of ΔF create data that relate a given ΔF to a known contaminant concentration. Likewise, the determinations of ΔR create data that relate a given ΔR to a known contaminant concentration. The unknown contaminant concentration may be determined according to the ΔF of the samples with the known and unknown contaminant concentrations or the ΔR of the samples with the known and unknown contaminant concentrations 1108. Steps 1102, 1103, 1104, and 1106 may be performed in any order. For example, steps 1102 and 1103 may be performed simultaneously. In another example, steps 1104 and 1106 may be performed simultaneously.

ΔF of a CM immunosensor exposed to samples with the known and unknown contaminant concentrations 1102, 1104, respectively, may be measured directly. ΔR of a CM immunosensor exposed to a sample containing the known or unknown contaminant concentration 1103, 1106, respectively, may be determined by direct measurement. For example, the ΔR may be measured directly using a QCA 922. ΔR may also be measured indirectly, an example of which is shown in FIG. 12. Although FIG. 12 refers to determining ΔR of a CM immunosensor exposed to a sample containing the known contaminant concentration 1103, the method shown is equally applicable to determining the ΔR of a sample containing the unknown contaminant concentration 1106.

Referring to FIG. 12, ΔR may be determined indirectly by measuring the conductance spectrum (G_f) 1202 and susceptance spectrum (B_f) 1204 of the CM immunosensor exposed to the sample with the known contaminant concentration. An example of a manner by which these measurements may be made is discussed above. After G_f and B_f have been measured 1202, 1204, respectively, ΔR may be determined as a function of G_f and B_f 1206. For example, ΔR may be determined 1206 by fitting G_f and B_f to the BVD model using equations (1) and (2) or a modified BVD model.

As shown in FIG. 11, the unknown contaminant concentration may be determined according to the ΔR of the CM immunosensor exposed to the samples with the known contaminant concentrations and the ΔR of the sample with the unknown contaminant concentration 1108. Alternately, the unknown contaminant concentration may be determined according to ΔF of the CM immunosensor exposed to the samples with the known contaminant concentrations and the ΔF of sample with the unknown contaminant concentration 1108. The determination of the unknown contaminant concentration is shown in more detail in FIG. 13.

As shown in FIG. 13, determining the unknown contaminant concentration 1108 includes determining whether ΔR or ΔF of the CM immunosensor exposed to the samples with the known contaminant concentrations more accurately reflects the known contaminant concentrations 1302. Several criteria may be used to make this determination. For example, the ΔR or ΔF of the CM immunosensor of the samples with the known contaminant concentrations that more accurately reflects the known contaminant concentrations may include that which is more proportional to the known contaminant concentrations. In another example, the ΔR or ΔF that more accurately reflects the known contaminant concentrations may include that which is more sensitive to the known contaminant concentrations. If ΔF of the CM immunosensor exposed to the samples with the known contaminant concentrations more accurately reflects the known contaminant concentrations, the unknown contaminant concentration may be determined according to the ΔF of the samples with the known concentrations 1306. If, however, ΔR of the CM immunosensor exposed to the samples with the known contaminant concentrations more accurately reflects the known contaminant concentrations, the unknown contaminant concentration may be determined according to ΔR of the samples with the known concentrations 1308.

If the unknown contaminant concentration is to be determined by the ΔF of the CM immunosensor exposed to the samples with the known contaminant concentrations, the ΔF of the CM immunosensor exposed to the sample with the unknown contaminant concentration is compared with that of the known contamination concentration. The contaminant concentration corresponding to the ΔF of the known contaminant concentration that is closest to the ΔF of the unknown concentration approximately equals the unknown contaminant concentration. A similar process may be used if the unknown contaminant concentration is to be determined by the ΔR.

Claims

1. A method for determining an unknown contaminant concentration in a first sample, the method comprising:

determining a change in a first motional resistance of a crystal microbalance (CM) immunosensor exposed to the first sample (ΔR1);

measuring a change in a first resonant frequency of the CM immunosensor exposed to the first sample (ΔF1);

measuring a change in a second motional resistance of the CM immunosensor exposed to a plurality of second samples (ΔR2), wherein the second samples include a plurality of known contaminant concentrations;

measuring a change in a second motional frequency of the CM immunosensor exposed to the plurality of second samples (ΔF2); and

determining the unknown contaminant concentration according to ΔR2 and ΔR, or ΔF2 and ΔF1.

2. The method of claim 1, wherein the crystal includes quartz.

3. The method of claim 1 further comprising creating the first sample from a substance.

4. The method of claim 1, wherein the first sample is created from the substance and a plurality of immuno-beads.

5. The method of claim 1, wherein the first sample is created from the substance and a plurality of immuno-magnetic beads.

6. The method of claim 1, wherein the first sample is created from the substance and a plurality of immuno-magnetic microbeads.

7. The method of claim 1, wherein the first sample is created from the substance and a plurality of immuno-magnetic nanobeads.

8. The method of claim 1, wherein the ΔF1 is measured and ΔR1 is determined simultaneously.

9. The method of claim 1, wherein the ΔF2 is measured and ΔR2 is determined simultaneously.

10. The method of claim 1, wherein determining ΔR, and/or ΔR2 includes measuring ΔR, and/or ΔR2.

11. The method of claim 1, wherein determining ΔR, includes measuring a conductance (Gf1) and a susceptance (Bf1) of the CM immunosensor exposed to the first sample.

12. The method of claim 11, wherein ΔF1, Gf1, and Bf1 are measured approximately simultaneously.

13. The method of claim 11, wherein ΔR, is determined as a function of Gf1 and Bf1.

14. The method of claim 13, wherein the function is a Butterworth-Van Dyke model.

15. The method of claim 1, wherein measuring ΔR2 includes measuring a conductance (Gf2) and a susceptance (Bf2) of the CM immunosensor exposed to the plurality of second samples.

16. The method of claim 15, wherein ΔF2, Gf2, and Bf2 are measured approximately simultaneously.

17. The method of claim 15, wherein ΔR2 is determined as a function of Gf2 and Bf2.

18. The method of claim 17, wherein the function is a Butterworth-Van Dyke model.

19. The method of claim 1, wherein determining the unknown contaminant concentration according to ΔR2 and ΔR, or ΔF2 and ΔF, comprises determining whether ΔR2 more accurately reflects the known contaminant concentrations than does ΔF2.

20. The method of claim 19, wherein determining whether ΔR2 more accurately reflects the known contaminant concentrations than does ΔF2 includes determining whether ΔR2 includes a greater proportionality to the known contamination concentrations than does ΔF2.

21. The method of claim 19, wherein determining whether ΔR2 more accurately reflects the known contaminant concentration than does ΔF2 includes determining whether ΔR2 provides a greater sensitivity than does ΔF2.

22. The method of claim 1, wherein determining the unknown contaminant concentration according to ΔR2 and ΔR, or ΔF2 and ΔF, comprises comparing ΔR, to ΔR2 if ΔR2 more accurately reflects the known contamination concentrations.

23. The method of claim 1, wherein determining the unknown contaminant concentration according to ΔR2 and ΔR1 or ΔF2 and ΔF1 comprises comparing ΔF1 to ΔF2 if ΔF2 more accurately reflects the known contamination concentrations.