Multiplexed paper test strip for quantitative

Anal Bioanal Chem (2012) 403:1567–1576 DOI 10.1007/s00216-012-5975-x

ORIGINAL PAPER

Multiplexed paper test strip for quantitative bacterial detection S. M. Zakir Hossain & Cory Ozimok & Clémence Sicard & Sergio D. Aguirre & M. Monsur Ali & Yingfu Li & John D. Brennan

Received: 15 February 2012 / Revised: 21 March 2012 / Accepted: 26 March 2012 / Published online: 18 April 2012 # Springer-Verlag 2012

Abstract Rapid, sensitive, on-site detection of bacteria without a need for sophisticated equipment or skilled personnel is extremely important in clinical settings and rapid response scenarios, as well as in resource-limited settings. Here, we report a novel approach for selective and ultra-sensitive multiplexed detection of Escherichia coli (non-pathogenic or pathogenic) using a lab-on-paper test strip (bioactive paper) based on intracellular enzyme (βgalactosidase (B-GAL) or β-glucuronidase (GUS)) activity. The test strip is composed of a paper support (0.5×8 cm), onto which either 5-bromo-4-chloro-3-indolyl-β-D-glucuronide sodium salt (XG), chlorophenol red β-galactopyranoside (CPRG) or both and FeCl3 were entrapped using sol– gel-derived silica inks in different zones via an ink-jet printing technique. The sample was lysed and assayed via lateral flow through the FeCl3 zone to the substrate area to initiate rapid enzyme hydrolysis of the substrate, causing a change from colorless-to-blue (XG hydrolyzed by GUS, indication of nonpathogenic E. coli) and/or yellow to red-magenta (CPRG hydrolyzed by B-GAL, Electronic supplementary material The online version of this article (doi:10.1007/s00216-012-5975-x) contains supplementary material, which is available to authorized users. S. M. Z. Hossain : C. Ozimok : C. Sicard : J. D. Brennan (*) Department of Chemistry and Chemical Biology, McMaster University, 1280 Main St. West, Hamilton, ON L8S 4M1, Canada e-mail: [email protected] S. D. Aguirre : M. M. Ali : Y. Li Department of Biochemistry and Biomedical Sciences, McMaster University, 1200 Main St. West, Hamilton, ON L8N 3Z5, Canada e-mail: [email protected]

indication of total coliforms). Using immunomagnetic nanoparticles for selective preconcentration, the limit of detection was ~5 colony-forming units (cfu) per milliliter for E. coli O157:H7 and ~20 cfu/mL for E. coli BL21, within 30 min without cell culturing. Thus, these paper test strips could be suitable for detection of viable total coliforms and pathogens in bathing water samples. Moreover, inclusion of a culturing step allows detection of less than 1 cfu in 100 mL within 8 h, making the paper tests strips relevant for detection of multiple pathogens and total coliform bacteria in beverage and food samples. Keywords Bacteria detection . Bioactive paper sensor . Colorimetric Abbreviations B-GAL β-galactosidase CI Color intensity cfu Colony-forming units CPRG Chlorophenol red β-galactopyranoside E. coli Escherichia coli GUS β-glucuronidase HB Hydrophobic barrier IMS Immunomagnetic separation LOD Limit of detection MB-Ab Antibody-derivatized magnetic beads PVAm Polyvinylamine XG 5-Bromo-4-chloro-3-indolyl-β-D-glucuronide sodium salt

Introduction Major food- and water-borne bacterial pathogens (such as Escherichia coli O157:H7, E. coli O104:H4 and Listeria

1568

monocytogenes) can cause severe infections in humans, accounting for millions of deaths worldwide each year, especially in children from developing nations [1]. A common indicator of bacterial contamination in food, water, and clinical samples is the coliform group of bacteria (e.g., E. coli), pathogenic or not as they occur concurrently. A key area for testing of coliforms is in recreational and bathing water [2]. Environmental protection agencies have set the limit for the number of E. coli present in recreational water, above which the quality of water is questionable at 500 colony-forming units (cfu) per 100 mL (European Directive 2006/7/EC) and 126 cfu per 100 mL (USA EPA’s Ambient Water Quality Criteria for Bacteria). The requirements for food and drinking water are 0 cells per 100 mL. Traditional microbiology, polymerase chain reaction (PCR), or enzyme-linked immunosorbent assay-based methods are too slow and labor-intensive to allow use at the contamination source and are too costly to deploy in resource-limited settings. To overcome these issues, various biosensing technologies have been developed in the past decade, based on a host of different platforms including liposome-immunoassays [3], microarrays [4], direct-charge transfer biosensors [5], surface plasmon resonance [6, 7], fluoroimmunoassays [8], microelectrical noise transducers [9], and conjugation of semiconductor quantum dots [10] and bioconjugated nanoparticle probes [11, 12] to bacteria. Unfortunately, these methods lack sensitivity, require a lengthy culturing step, or need sophisticated instrumentation to measure signals, making them unsuitable for rapid, inexpensive, selective, and ultrasensitive on-site monitoring [13]. Major gains in sensitivity have been realized by coupling PCR [14] or immunomagnetic separation (IMS) [15–17] to detection methods such as plate counting, laser cytometry, absorbance, or fluorescence, with the best methods achieving detection limits of two to three cells per gram of sample using either a 4-h culturing step [16] or a 2-h PCR amplification step [18]. These methods, however, are neither quantitative nor field-portable. A range of on-site E. coli detection strategies have been reported using either portable (battery powered) instrumentation, competitive displacement assays [19], or immunochromatographic test strips [20]. For low-cost detection, the latter method is the most interesting, and such assays are already commercially available: MaxSignal®, RapidChek®, Gen-Probe®, IQuum®, Watersafe®, and others. The reported limit of detection of bacteria using such strips ranges from 104 to 107 cfu/mL without an enrichment step and is ~105 cfu/mL for E. coli O157:H7. However, to reach a detection limit of 1 cfu/mL, culturing steps of at least 8 h are generally required. Recently, bioactive paper or lab-on-paper devices have been developed as a platform for simple, portable, inexpensive, disposable assays that can detect toxins, heavy metals,

S.M.Z. Hossain et al.

disease markers, illicit drugs, and health markers [21–26]. These paper-based sensing devices can be applied not only in point-of-care testing but also in field analysis in remote locations with limited facilities. While there are several proof-of-concept studies showing the potential of paperbased analytical devices, there are still no paper-based devices reported to date that are able to selectively detect disease-causing microorganisms below the infectious dose without a culturing step. Herein, we introduce a novel method that combines selective immunomagnetic preconcentration of bacteria, lysing of cells, and detection of an intracellular enzyme using an ink-jet printed lateral-flow colorimetric bioactive paper sensor for rapid and sensitive detection of specific bacteria. Importantly, our method is capable of selective, simultaneous detection of pathogenic and non-pathogenic bacteria, as demonstrated with two strains of E. coli. The paper sensors utilize 5-bromo-4-chloro-3-indolyl-β-D-glucuronide sodium salt (XG) and chlorophenol red β-galactopyranoside (CPRG)-based colorimetric assays. The XG is broken down by β-glucuronidase (GUS) into D-glucuronic acid and ClBrindoxyl followed by oxidation of the latter into ClBr-indigo dye, a blue product, while CPRG (yellow color) is broken down by β-galactosidase (B-GAL) into chlorophenol red, a red-magenta product. Coliforms are generally B-GALpositive, and the detection of this enzyme has been extensively used for enumerating total coliforms [27]. Therefore, formation of red-magenta color on paper is indicative of presence of coliforms. Most E. coli strains produce GUS [28] but not the pathogenic E. coli O157:H7 [29]. Thus, the formation of blue color on the paper strips is indicative of non-pathogenic E. coli, whereas absence of blue color (in presence of red-magenta) indicates pathogenic E. coli. All reagents are deposited onto the paper surface via ink-jet printing of layered structures composed of a capture layer (polyvinyl amine), a lower sol–gel-derived silica layer, the reagent, and a top silica layer, with reagents deposited into specific zones as shown in Fig. 1a, b. The paper sensor requires only immersion into a cell lysate for proper functioning. An optional IMS step utilizing antibody-derivatized magnetic beads (MB-Ab) can easily be employed to enhance selectivity and sensitivity without altering the portability of the sensor, as shown in Fig. 1c. Experimental section Organisms and screening of single bacteria Four bacterial strains (E. coli BL21, E. coli 0157:H7, Bacillus subtilis and Salmonella enterica) were used in this study. Two standard media (Luria Bertani or trypticase soy broth) were used for cell culturing. The cultures were grown overnight at 37 °C with shaking at 125 rpm. A standard

Multiplexed paper test strip Fig. 1 Schematic diagram of the paper-based bacteria sensor. A Sensing strategy for detection of single bacterial species, in which either XG or CPRG (color reaction or CR zone) and FeCl3 (Fe zone) are entrapped within sol–gel-derived silica materials in the two dashed regions on a Whatman no. 1 paper strip (0.5×8 cm) via inkjet printing. A hydrophobic barrier (HB zone) composed of MSQ is layered at the top of the sensing zone to prevent leaching of color and thus increase signal intensity. The sensor is dipped into a cell lysate until the liquid front reaches the substrate/sensing region. Color appearance in the CR zone is indicative of the presence of bacteria. B Bidirectional multiplexed sensing strategy in which XG, CPRG, and FeCl3 are entrapped within sol–gel-derived silica materials in separate regions on a paper strip. A HB barrier is introduced between two sensing zones. The bottom of the paper strip is first placed into a cell lysate (E. coli BL21 in this case) and allowed to flow. The sensor is then inverted, immersed into the same lysate, and allowed to flow. The color intensity is monitored after 30 min at room temperature, unless otherwise stated. C Optional preconcentration step in which cells are first isolated from samples via magnetic pulldown, resuspended in a minimal volume of a lysing reagent (10–100-fold less volume than initial sample) and then tested

1569

1570

plate counting method was used to count the numbers of colony-forming units per milliliter (≥10 cells per sample), while for isolation of single E. coli O157:H7 cells, AOAC INTERNATIONAL Methods Committee Guidelines and other published protocols were followed [12, 30] as E. coli O157:H7 and S. enterica are harmful bacteria that should be handled with care. Briefly, 10−7, 10−8, 10−9, and 10−10 dilution stocks (1 mL each) were prepared from the original 10−7 stock (~120 cfu/mL). For each dilution, ten culture tubes (Falcon 2058) with 1.9 mL media were inoculated with 0.1 mL of the respective dilution stock. All inoculations were grown at 37 °C for 24 h. Three different sets of parallel experiments for each dilution were conducted. Construction of paper-based lateral-flow E. coli sensors The sensors were constructed based on two enzyme– substrate systems: (1) GUS vs. XG and (2) B-GAL vs. CPRG (all reagents were obtained from Sigma). A section of Whatman no. 1 paper was cut into small pieces (0.5× 8 cm) on which either XG (3 mM), CPRG (3 mM), or both, and an oxidizing agent, FeCl3 (0.2 mM) were printed using a piezoelectric inkjet printer (Fujifilm Dimatix 2800 N), along with a sol–gel precursor solution to entrap the reagent in different zones (e.g., substrate and oxidizing reagent zones). To prevent leaching of the color, a hydrophobic barrier (HB) was also introduced on the top (Fig. 1a) or middle (Fig. 1b) of the sensing zone(s) using methylsilsequioxane (MSQ), via piezoelectric ink-jet printing of a methytrimethoxysilane (MTMS) sol composed of MTMS (1 mL), EtOH (1.63 mL), and HCl (1 M; 0.5 mL). The sensing regions for GUS and B-GAL were prepared by depositing polyvinylamine (PVAm, BASF, 0.5 wt.% in water)/silica (0.5 g/mL sodium silicate solution prepared as described previously [31, 32])/XG or CPRG/silica layers in the order described. All solutions except the PVAm solution contained 30 % glycerol (v/v) and 0.1 wt.% Triton X-100 to adjust viscosity and surface tension, respectively, to the ranges suitable for ink-jet printing [31, 32]. After printing, the sensor was allowed to dry for at least 1 h in air at room temperature. We are currently investigating alternative printing methods to avoid the use of glycerol, which can affect enzyme activity. Prior to constructing an efficient paper-based E. coli sensor, all reagent concentrations were optimized initially in solution and later on paper (using a separate paper sensor for each concentration) based on single-factor experiments (varying one parameter at a time). E. coli assays using test strips To demonstrate the analytical performance of the sensors under optimized conditions, different concentrations of


bacteria ranging from 0 to ~108 cfu/mL were made. The cell samples were lysed using a 9:1 volume ratio of cell suspension to B-PER® direct bacterial protein extraction reagent (Thermo Scientific) according to the included instructions and assayed by placing the test strips into the lysed cell sample and allowing the liquid to move up the strip until it reached the hydrophobic barrier. The concentration of E. coli in samples was detected by measuring the color intensity (CI), as described below, with incubation times of 30 min, unless otherwise stated. Preconcentration of E. coli using MB-Ab Dynabeads® MAX E. coli O157 kit (BioClone Inc.) was used to selectively preconcentrate E. coli O157:H7, while for preconcentration of E. coli BL21, the anti-E. coli BL21 antibody (20-ER13, Fitzgerald) was conjugated to hydrazide modified MBs (BioClone Inc.) following the manufacturer’s coupling protocol. MB-Ab complexes were strictly used in accordance with included instructions to preconcentrate bacteria from a large volume. Detection of E. coli in food and beverage samples Fresh milk (1 %) and orange juice were purchased from a local grocery store and artificially contaminated with different known numbers of E. coli BL21 (0–8×106 cfu/mL) and E. coli O157:H7 (0–2×106 cfu/mL) cells. The pH of orange juice was adjusted from 3.5 to between 6 and 8 prior to testing by adding NaOH (0.1 M), as the test strips had better performance at this pH (see Electronic supplementary material Fig. S2). Milk samples were tested with no additional processing. Prior to analysis, IMS was used to concentrate the cells (from 10 mL into a volume of 0.09 mL), which were then lysed by addition of lysing agent (0.01 mL) and tested. Uninnoculated orange juice and milk samples were used as controls. Four parallel samples were prepared for each concentration. The skin of head lettuce (8×8 cm) was dipped for 30 s in contaminated water solutions with either E. coli BL21 (~6×106 cfu/mL) or E. coli O157:H7 (~1.3 × 105 cfu/mL) and removed for air drying. The samples of E. coli residue were collected after 1 h through washing with ddH2O (10 mL) or phosphate buffer (10 mL) and tested using the procedure as outlined above without preconcentration. Interference study To evaluate potential interference by non-target organisms, two sets of mixed culture were prepared: (1) a fixed number of B. subtilis (~106 cfu/mL)+E. coli BL21 (~104 to 8× 108 cfu/mL) and (2) a constant number of S. enterica (~106 cfu/mL)+E. coli O157:H7 (~2.5×103 to 108 cfu/mL).

Multiplexed paper test strip

The mixtures of both bacterial suspensions were lysed and tested using paper sensors, and the data were compared with those of individual cell lysates. Data processing and statistics The CI of the sensing areas was monitored either by the naked eye or by obtaining a digital image (Canon G11 12 MegaPixel camera operated in automatic mode with no flash and with the macroimaging setting on) or using an office scanner. Note that, due to a camera setting that automatically adjusts the white balance, the background of the CPRG strips could have a yellowish appearance. The CI was analyzed with ImageJTM software as previously described. The numerical single-color coordinate values were obtained using the ImageJTM color split-channel system. To account for variations in color intensity that are due to differences in environmental illumination and auto-white balance of the camera, a background subtraction was done for each data point. Standard deviations and coefficients of variation were estimated using Excel software. We are currently investigating other methods for colorimetric analysis.

Results and discussion Optimization of paper-based lateral-flow test strips assay format Initial studies focused on optimizing assay time, sensitivity, selectivity, and reproducibility, which required examination of capture agents, concentration of substrates, pH of the running buffer, and time for color development (see Electronic supplementary material Fig. S1 and Fig. S2). These revealed that PVAm was the best cationic capture agent to entrap and preserve the anionic color product [31]. These studies also revealed that XG and CPRG were the most suitable substrates for GUS and B-GAL assays, respectively (data not shown). Our observations are in agreement with the results of previous comparative studies of different chromogenic GUS [33] and B-GAL substrates [34, 35]. Note that, when GUS breaks down XG, ClBrindigo dye appears very slowly (>1 h), but the addition of an oxidizing agent, FeCl3, increases the speed of oxidation of ClBr-indoxyl into ClBr-indigo [36], leading to color formation in 1 min. The reaction between CPRG and BGAL is relatively rapid, with B-GAL being more active and stable in the presence of a small amount of metal ion (e.g., Fe3+) [35]. The optimum pH of the running solution for the test strips was found to be 8. Indeed, the enzyme activities are optimum at pH between 6 and 7 and between 6 and 8 for GUS [37, 38] and B-GAL [39], respectively, and the speed

1571

of oxidation of ClBr-indoxyl into ClBr-indigo increased with pH [36]. A hydrophobic line was introduced to stop solution flow and prevent the colored product from streaking over the entire paper surface. The color intensity (determined using an office scanner followed by image analysis with ImageJ) increased with time, and the maximum signal could be obtained within 1 h for GUS and 30 min for BGAL assays, though color could be detected in 1 min in both cases. For consistency, the images were obtained after 30 min for all subsequent experiments, unless otherwise stated. Analytical performance of E. coli test strips The analytical performance of the paper sensors was evaluated using both non-pathogenic (E. coli BL21) and pathogenic (E. coli O157:H7) bacterial cell suspensions. Prior to the assay, B-PER® direct lysing reagent was added to samples (1:9 volume ratio) to lyse the cells and maximize the free enzyme concentration in the sample. The cost of BPER® is $0.25 per mL, keeping the overall cost of paperbased assay relatively low. For the sake of clarity, as an indicator of non-pathogenic bacteria, we focused on the XG side for E. coli BL21 and on the CPRG side as an indicator of pathogenic E. coli O157:H7. Figure 2a shows preliminary data obtained with paper-based sensors using ~108 cfu/mL of E. coli BL21 (left) and O157:H7 (right), respectively, with signals obtained after 1 min. To improve both the selectivity and limit of detection (LOD, S/N >3, n03) while retaining the portability of the assay, a simple IMS step was integrated into the assay. The Dynabeads® MAX E. coli O157 kit (capture efficiency >95 %) was used to concentrate the pathogenic bacteria while, for non-pathogenic bacteria, the anti-E. coli BL21 antibody was conjugated to MBs and could capture >90 % of E. coli BL21 from samples in the concentration range from 101 to 105 cfu/mL. Even if IMS retains the portability of the assay, drawbacks of this method are that it is unsuitable for untrained personnel and that it increases the cost of the assay by ca. $3.00 per sample. Therefore, we are currently investigating other methods such as a filtration step. Figure 2b, c shows semi-log plots of color intensity (obtained using an office scanner and ImageJ software) with varying E. coli (pathogenic and non-pathogenic) concentrations with and without IMS preconcentration. As the concentration of E. coli increased, the color intensity also increased but not following a linear law, as a result of the combination of several complex parameters including the saturation of the capture agent. Without IMS (curve c), the LOD for E. coli BL21 and E. coli O157:H7 was ~104 and ~2.5×103 cfu/mL, respectively. Given that only 20 μL of sample is used per assay, the strip was able to detect ~200 cells of E. coli BL21 and ~50 cells of E. coli O157:H7, with

1572


Fig. 2 Real-time monitoring of E. coli BL21 (~108 cfu/mL) and E. coli O157:H7 (~108 cfu/mL) using paper strips. A A colorless-to-deep blue color change is observed in the CR zone within 1 min when the strip is dipped into the sample of E. coli BL21 due to GUS hydrolysis of XG (left images), while a light yellow-to-red-magenta color change is seen within 10 s when the strip is dipped into sample of E. coli O157: H7 via breakdown of CPRG by B-GAL (right images). Note that the yellowish appearance of the CPRG strips is due to a camera setting that automatically adjusts the white balance. B Different initial concentrations of E. coli BL21 were prepared and preconcentrated with IMS (a:

100×, b: 10×, or c: 1×), lysed, and tested. C Different initial concentrations of E. coli O157:H7 were prepared and preconcentrated with IMS (a: 100×, b: 10×, or c: 1×), lysed, and tested. D Paper sensor responses as a function of culturing time. Different initial numbers E. coli O157:H7 cells were inoculated into growth media and allowed to grow. The samples were preconcentrated (20-fold) using IMS and tested using the paper sensor. Curves a, b, and c are the timedependent responses of initial cell numbers of 100, 10, and 1 cfu in 2 mL, respectively. For all panels, the values are the average of three replicates; error bars denote ±SD

variation coefficients of less than 5 %. For the assay performed with IMS from different initial sample volumes (1 mL, curve b and 10 mL, curve a) to a final volume of 0.1 mL (prior to lysing and detection), the LOD for E. coli BL21 was 2×102 and 20 cfu/mL, respectively, and for E. coli O157:H7, was 50 and 5 cfu/mL, respectively, representing the expected 10–100-fold concentration enhancement. The achieved LOD is on par with the regulatory limits set for bathing water (126 mL / 100 mL in the USA; 500 cfu / 100 mL in Europe), making the test strips a very promising solution for wide use in testing coliform levels in recreational or bathing water. Further improvements in LOD were still possible by including a culturing step. Figure 2d shows paper sensor

responses for different initial numbers of E. coli O157:H7 cells—100, 10, and 1 cell(s) in 2 mL media, respectively, after different culturing times. All samples were allowed to grow for up to 8 h. The samples were preconcentrated 20fold using IMS (from an initial volume of 2 mL to a final volume of 0.1 mL), lysed, and then tested using the paper sensor. Color was observed after 3 h of culturing for samples containing 100 cells (Fig. 2d–a) while, after 5 h of culturing, color was observed for all samples, and after 8 h of culturing, the color was saturated for all samples. The data show that the sensor is able to detect a single cell in 2 mL with 5 h of culturing and better than 1 cfu in 100 mL after 8 h, making the assay amenable for testing drinking water (where 0 cells per 100 mL is required).


1573

E. coli assay in beverage and food samples

Interferences on assay performance

The paper-based sensors are also suitable for the testing of contaminated beverage and food samples. For both orange juice and milk samples, the data (Fig. 3a, b) were almost identical to those obtained from water samples, which indicates that the sample matrices had a negligible effect on the sensitivity. In the case of food samples, head lettuce was artificially contaminated by dipping the lettuce into media that contained relatively high levels of E. coli BL21 (~6× 106 cfu/mL) or E. coli O157:H7 (~1.3×105 cfu/mL). The samples were washed with phosphate buffer (10 mL) to collect the bacteria, mixed with B-PER lysing solution (9:1 (v/v)), and then tested. The blue color signal in Fig. 3c (CI~14) and red-magenta color signal in Fig. 3d (CI~35) indicated that E. coli BL21 and E. coli O157:H7 cell concentrations collected from the lettuce were between 104–105 cfu/mL and 103–104 cfu/mL, respectively, a recovery of ~10 % in each case. Importantly, all the data were highly reproducible, showing the utility of the assay for monitoring of bacteria on food and in beverages.

An important aspect of the paper-based sensor was a lack of interference by non-target organisms. B. subtilis was used as a model for gram-positive bacteria and tested on the XG side, and S. enterica, which has no intracellular B-GAL, was tested on the CPRG side. B. subtilis and S. enterica alone produced no color changes on the test strips while, for mixed cultures (B. subtilis + E. coli BL21; S. enterica + E. coli O157:H7), deep blue and/or red-magenta color was observed, respectively, as shown for different concentrations of mixed cell lysates (Fig. 4a, b). The data show that there was a negligible effect on the sensitivity, with the LODs for E. coli BL21 and E. coli O157:H7 being similar to those obtained in samples containing only E. coli BL21 or O157:H7.

Fig. 3 Detection of both non-pathogenic (E. coli BL21) and pathogenic (E. coli O157: H7) bacteria in beverage and food samples using bioactive paper strips. Milk (1 %) and orange juice (total volume of 10 mL each) were artificially contaminated with different concentrations of A E. coli BL21 (0 to ~8×106 cfu/mL) or B E. coli O157:H7 (0 to ~2×106 cfu/ mL), separately. The cells were concentrated using IMS from an initial volume of 10 mL to a final volume 0.1 mL (including 0.01 mL of B-PER® lysing agent). C, D Head lettuce was dipped in contaminated solution containing E. coli BL21 (~6×106 cfu/mL) or E. coli O157:H7 (~1.3×105 cfu/mL) for 30 s and removed for air drying. After 1 h drying in air, the cell samples from head lettuce were collected through washing with phosphate buffer (10 mL). All the contaminated samples (milk, juice, head lettuce) were lysed prior to testing. Controls experiments in panels C and D used test strips that were immersed into phosphate buffer instead of the cell lysate solution. All points are means±SD of three measurements for each concentration

Specificity study and multiplexed detection Unlike other bacterial sensors including our recently reported fluorogenic DNAzyme probes [30], the paper

1574


Fig. 4 Interference from non-target organisms, specificity, and multiplex detection using the portable paper sensor. A B. subtilis (~106 cfu/mL, non-target organism) was mixed with different concentrations of E. coli BL21 and tested by using XG-based paper strips without IMS. B S. enterica (~106 cfu/mL, non-target organism) was mixed with different concentrations of E. coli O157:H7 and tested using CPRG-based paper sensors without IMS. Data are means±SD of three determinations. C Dual detection using paper sensor before and after IMS isolation and preconcentration (10×) of E. coli O157:H7 from a mixture of E. coli BL21 (~2.8×106 cfu/mL) and E. coli O157:H7 (~2.4×104 cfu/mL). D Similar experiments (dual detection and specificity studies) conducted using a lateral flow device with an integrated bidirectional sensor and sample inlets. One drop of cell lysate solution (~20 μL) was added into each inlet of the device to allow lateral flow of the sample to the sensing zones

sensors developed herein are capable of specific and multiplexed detection of both pathogenic and non-pathogenic E. coli. Figure 4c shows the ability to selectively detect pathogenic bacteria from a mixture by using IMS. In order to format the sensor as a simple kit, permitting the test to be run even by untrained personnel, the same mixture was tested using a prototypical lateral flow device (made with the paper sensor strip and a plastic frame), which allowed single drops of sample to be placed into the sample loading zones. The results are shown in Fig. 4d. A mixture of E. coli BL21 and E. coli O157: H7 was initially tested without using magnetic separation. After using IMS to selectively preconcentrate E. coli O157:H7 cells by tenfold and remove E. coli BL21, only the red-magenta color was observed, demonstrating that the sensor was able to selectively detect E. coli O157:H7 from the mixed culture. Storage stability of the sensor A major benefit of the current sensors is the lack of any biorecognition element as part of the sensing platform. Even

Fig. 5 Long-term stability of A XG and B CPRG within the layered coating (e.g., silica-XG-silica or silica-CPRG-silica) printed on the paper-based sensor. The XG and CPRG-based strips were immersed into GUS (final concentration of 1 U/mL) and B-GAL (final concentration of 1 U/mL) solution, respectively, every week up to 8 weeks (a separate paper sensor was used for each case). Sensors were stored under dry conditions at room temperature in the dark. Data are means±SD of three determinations


so, oxidation or other changes in the substrates or PVAm capture agents could affect the long-term stability of the sensor. Figure 5 shows that the test strips could be stored for at least 2 months at room temperature (20±2 °C) without a noticeable decrease in the sensing of GUS or B-GAL activity. This suggests that the sensors should have a long shelf life, even when at room temperature, making them amenable to transportation, storage, and use even in resource-limited locations.

1575 Acknowledgments The authors thank the Natural Sciences and Engineering Research Council of Canada and the Sentinel-Canadian Network for the Development and Use of Bioactive Paper for funding. The authors also thank the Canada Foundation for Innovation and the Ontario Ministry of Research and Innovation for support of this work. The authors thank Puneet Chavan for his contribution to experimental data. YL holds the Canada Research Chair in Nucleic Acids Chemistry. JDB holds the Canada Research Chair in Bioanalytical Chemistry and Biointerfaces.

References Conclusion In this study, we have developed, for the first time, a selfcontained, portable, bioactive paper sensor using ink-jet printing of sol–gel entrapped reagents to allow for rapid (seconds to several minutes, depending on cell concentration) and sensitive detection of viable E. coli based on GUS and/or B-GAL activity. Detection could be achieved by eye, or using a digital camera and image analysis software, avoiding the need for instrumentation, trained personnel, or electrical power. A combination of bioactive paper strips and magnetic preconcentration showed rapid, selective, and ultra-sensitive detection of E. coli BL21 (LOD ~20 cfu/mL) and E. coli O157: H7 (LOD ~5 cfu/mL) within 30 min without culturing, and a single bacterium in 2 mL with 5 h of culturing. Milk and orange juice sample matrices had a negligible effect on the E. coli detection sensitivity, and it was possible to detect E. coli quantitatively from food samples. The experimental data show that both XG and CPRG entrapped between two biocompatible silica layers on paper is stable for at least 2 months at room temperature. Overall, the results indicated that the bioactive paper strip could be used as a component of a simple and inexpensive field kit for simultaneous detection of pathogens and total coliform bacteria in recreational water. In addition, the method could be extended to detect less than 10 cfu/mL of E. coli without culturing if IMS was used. Bioactive paper strips could therefore also be suitable for detection of coliforms in drinking water, beverages, or food samples. We believe a filtration step from an initial sample volume of 250 mL to a final volume of 0.1 mL could be performed instead of IMS (currently under investigation), thus maintaining use by untrained personnel. Moreover, a filtration step would discard any enzymes present in the media before cell lysing, enabling selective detection of viable bacteria. The test strips could be adapted to other GUS and B-GAL producing bacteria using IMS specific for a desired bacterial target, or to other intracellular enzyme markers by varying the nature of the reagents printed on the paper test strip. Our efforts in these areas will be described in future work.

1. World Health Organization (2007) Food safety and foodborne illness. http://www.who.int/mediacentre/factsheets/fs237/en/ index.html Accessed on March 20, 2012 2. Edberg SC, Rice EW, Karlin RJ, Allen MJ (2000) J Appl Microbiol 88:106S–116S 3. Ho J-AA, Hsu H-W (2003) Anal Chem 75:4330–4334 4. Lee W, Park K-S, Kim Y-W, Lee WH, Choi J-W (2005) Biosen Bioelectron 20:2292–2299 5. Pal S, Alocilja EC, Downes FP (2007) Biosen Bioelectron 22:2329–2336 6. Waswa J, Irudayaraj J, Debroy C (2007) LWT - Food Sci Technol 40:187–192 7. Huang C-J, Dostalek J, Sessitsch A, Knoll W (2011) Anal Chem 83:674–677 8. Wang L-J, Wu C-S, Hu Z-Y, Zhang Y-F, Li R, Wang P (2008) J Zhejiang Univ Sci B 9:121–131 9. Seo S, Kim HC, Cheng M, Ruan X, Ruan W (2006) J Vac Sci Technol B 24:3133 10. Hahn Ma, Tabb JS, Krauss TD (2005) Anal Chem 77:4861–4869 11. Kalele SA, Kundu AA, Gosavi SW, Deobagkar DN, Deobagkar DD, Kulkarni SK (2006) Small 2:335–338 12. Zhao X, Hilliard LR, Mechery SJ, Wang Y, Bagwe RP, Jin S, Tan W (2004) Proc Nat Acad Sci U S A 101:15027–15032 13. Heo J, Hua SZ (2009) Sensors 9:4483–4502 14. Yeung S-W, Lee TM-H, Cai H, Hsing I-M (2006) Nucl Acids Res 34:e118 15. Lee H-J, Kim BC, Kim K-W, Kim YK, Kim J, Oh M-K (2009) Biosen Bioelectron 24:3550–3555 16. Tu S-I, Reed S, Gehring A, He Y, Paoli G (2009) Sensors 9:717–730 17. Chapman PA (2000) World J Microbiol Biotechnol 16:733–740 18. Guan J, Levin RE (2002) Food Biotechnol 16:135–144 19. Miranda OR, Li X, Garcia-Gonzalez L, Zhu Z-J, Yan B, Bunz UHF, Rotello VM (2011) J Am Chem Soc 133:9650–9653 20. Ngom B, Guo Y, Wang X, Bi (2010) Anal Bioanal Chem 397:1113–1135 21. Abe K, Suzuki K, Citterio D (2008) Anal Chem 80:6928–6934 22. Zhao W, Ali MM, Aguirre SD, Brook MA, Li Y (2008) Anal Chem 80:8431–8437 23. Martinez AW, Phillips ST, Whitesides GM, Carrilho E (2010) Anal Chem 82:3–10 24. Apilux A, Dungchai W, Siangproh W, Praphairaksit N, Henry CS, Chailapakul O (2010) Anal Chem 82:1727–1732 25. Luckham RE, Brennan JD (2010) Analyst 135:2028–2035 26. Khan MS, Thouas G, Shen W, Whyte G, Garnier G (2010) Anal Chem 82:4158–4164 27. Wutor VC, Togo CA, Pletschke BI (2009) WaterSA 35:85–88 28. Kilian M, Bülow P (1976) Acta Patholo Microbiol Scandi B 84:245–251 29. Ratnam S, March SB, Ahmed R, Bezanson GS, Kasatiya S (1988) J Clin Microbiol 26:2006–2012

1576 30. Ali MM, Aguirre SD, Lazim H, Li Y (2011) Angew Chem Int Ed 50:3751–3754 31. Hossain SMZ, Luckham RE, Smith AM, Lebert JM, Davies LM, Pelton RH, Filipe CDM, Brennan JD (2009) Anal Chem 81:5474–5483 32. Hossain SMZ, Luckham RE, McFadden MJ, Brennan JD (2009) Anal Chem 81:9055–9064 33. Delisle GJ, Ley A (1989) J Clin Microbiol 27:778–779 34. Wutor VC, Togo CA, Limson JL, Pletschke BI (2007) Enzyme Microb Technol 40:1512–1517

S.M.Z. Hossain et al. 35. Wutor VC, Togo CA, Pletschke BI (2007) Chemosphere 68:622–627 36. Cotson S, Holt SJ (1958) Proc R Soc Lond B 148:506–519 37. Kim D-H, Jin Y-H, Jung E-A, Han MJ, Kobashi K (1995) Biol Pharm Bull 18:1184–1188 38. Buehler HJ, Katzman PA, Doisy EA (1951) Proc Soc Exp Biol Med 76:672–676 39. Tenu J-P, Viratelle OM, Garnier J, Yon J (1971) Eur J Biochem 20:363–370