Biosensors and Bioelectronics 55 (2014) 289–293
Contents lists available at ScienceDirect
Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios
Short communication
Quaternized magnetic nanoparticles–fluorescent polymer system for detection and identification of bacteria Yi Wan a, Yan Sun a, Peng Qi a,b, Peng Wang a, Dun Zhang a,n a Key Laboratory of Marine Environmental Corrosion and Bio-fouling, Institute of Oceanology, Chinese Academy of Sciences, 7 Nanhai Road, Qingdao 266071, China b University of Chinese Academy of Sciences, 19 (Jia) Yuquan Road, Beijing 100039, China
art ic l e i nf o
a b s t r a c t
Article history: Received 9 October 2013 Received in revised form 18 November 2013 Accepted 25 November 2013 Available online 12 December 2013
Nanomaterial-based ‘chemical nose’ sensor with sufficient sensing specificity is a useful analytical tool for the detection of toxicologically important substances in complicated biological systems. A sensor array containing three quaternized magnetic nanoparticles (q-MNPs)–fluorescent polymer systems has been designed to identify and quantify bacteria. The bacterial cell membranes disrupt the q-MNP– fluorescent polymer, generating unique fluorescence response array. The response intensity of the array is dependent on the level of displacement determined by the relative q-MNP–fluorescent polymer binding strength and bacteria cells–MNP interaction. These characteristic responses show a highly repeatable bacteria cells and can be differentiated by linear discriminant analysis (LDA). Based on the array response matrix from LDA, our approach has been used to measure bacteria with an accuracy of 87.5% for 107 cfu mL 1 within 20 min. Combined with UV–vis measurement, the method can be successfully performed to identify and detect eight different pathogen samples with an accuracy of 96.8%. The measurement system has a potential for further applications and provides a facile and simple method for the rapid analysis of protein, DNA, and pathogens. & 2013 Elsevier B.V. All rights reserved.
Keywords: Quaternized magnetic nanoparticles Bacteria detection Fluorescent polymer Linear discriminant analysis
1. Introduction The determination, identification, and quantification of pathogen for clinical diagnosis, environmental monitoring, and food safety are crucial for public health protection. Traditional methods for bacteria detection, such as plating and culturing (Gracias and McKillip, 2004), enzyme-linked immunosorbent assay (Chitarra and van den Bulk, 2003), and polymerase chain reaction (Yamamoto, 2002), involve a pre-enrichment step or a selective enrichment step followed by a biochemical test, and the sophisticated series of assays required can take up from several hours to several days. Although these approaches for bacterial detection are sufficiently sensitive and selective, most have several drawbacks such as being time-consuming, cost-intensive, or technically complex. Increasing worldwide demand and concern for their safe development and use require a simple, stable, and sensitive detection assay for pathogen evaluation and environmental monitoring. Over the past decade, many efforts have been made to use efficient and cost-effective techniques based on magnetic nanoparticles (MNPs) as a rapid and reliable tool for pathogen detection (Arruda et al., 2009; Gao et al., 2009; Lu et al., 2012; Sanvicens
n
Corresponding author. Tel./fax: þ 86 532 82898960. E-mail addresses: [email protected], [email protected] (D. Zhang).
0956-5663/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bios.2013.11.080
et al., 2009; Shao et al., 2009, 2010). Several immunomagnetic approaches have been developed for label-free detection of pathogen. For example, the immunomagnetic separation with MNP–antibody conjugates was studied and estimated for pathogenic E. coli in a complex environment. MNP–antibody conjugates were prepared by immobilizing biotin-labeled polyclonal antibodies onto streptavidin-coated MNP or direct chemical immobilization (Ravindranath et al., 2009; Varshney et al., 2005). Our group reported a fast, sensitive and reliable biosensor based on the amplification of the response of vancomycin-coated MNP for the detection of marine pathogen under an external magnetic field (Wan et al., 2010). The feature characteristics of these MNP-based analysis techniques result from the simplification of separation and enrichment steps, or from their high specificity, which allow the detection of specific bacteria in complicated matrices. Recently, some MNPs have also been used as signal labels in analytical techniques. One example is proposed by Kaittanis and his co-workers, who functionalized superparamagnetic iron oxide nanoparticles for quantification of bacteria in clinical and environmental samples through magnetic relaxation (Kaittanis et al., 2006). Another is the work by Weissleder's group, who designed an MNP hybridization assay, involving ubiquitous and specific bacterial probes that target bacterial DNA segments, to detect amplified target bacterial DNAs using a miniaturized nuclear magnetic resonance device (Chung et al., 2013). However, whilst these techniques are relevant for laboratory use,
290
Y. Wan et al. / Biosensors and Bioelectronics 55 (2014) 289–293
they cannot adequately serve the requirements of health practitioners or enable large-scale monitoring of environmental issues. Given these disadvantages, an effort has been made to develop a facile and simple method for target analysis that is labor- and time-saving. The ‘chemical nose’ technique offers a good alternative for biomolecules sensing. In this strategy, a chemical nose needs to be trained with qualified biological samples so as to build a database of reference. Then the sensing system can recognize new samples by comparing biomolecule database. This method has been used to sense a lot of targets, including amino acid (Folmer-Andersen et al., 2006), carbohydrates (Lee et al., 2006), protein (You et al., 2007), bacteria (Phillips et al., 2008) and cancer cells (Bajaj et al., 2009). For example, Phillips et al. who address the issue of rapid identification of bacteria, developed for bacterial sensing using an array of gold–nanoparticle-conjugated polymer constructs (Phillips et al., 2008). However, the presence of the constructs in a well on plate may disrupt the response pattern of the ‘chemical nose.’ In our work, we designed a q-MNP–fluorescent polymer system for detection and identification of bacteria. Three q-MNP–fluorescent polymer constructs are combined to differentiate bacterial cells according to their characteristic response pattern. To avoid the interference of q-MNP–fluorescent polymer constructs in the measurement system, the constructs can be separated quickly by a magnet.
2. Material and methods 2.1. Bacterial cultivation All bacteria were offered from the Key Laboratory of Experimental Marine Biology. Bacteria were seeded and cultured in suspension using the following media: S. oneidensis, V. fischeri, M. luteus, E. tarda and E. coli in Luria–Bertani media; V. alginolyticus and P. aeruginosa in trypticase soy broth; P. pastoris in yeast extract peptone dextrose medium. For these bacteria, a single colony was inoculated in bacterial medium at 37 1C overnight, shaking at 200 rpm. After centrifugation (6000 rpm) for 10 min and PBS washing, the bacterial cells were diluted to the desired concentration or optical density in phosphate buffer. 2.2. Synthesis of fluorescent polymer A polymer functionalized with side-chain carboxylic acid group (PFBT) was synthesized using a method reported with modification as shown in Fig. S1 (Zhang et al., 2008). A mixture of 2,7dibromo-9,9-bis(3-(tert-butyl propanoate))fluorine (monomer A, 0.232 g), 4,7-dibromobenzo[c][1,2,5]thiadiazole (monomer B, 0.176 g), and 9,9-dioctylfluorene-2,7-diboronic acid bis(1,3-propanediol) ester (monomer C, 0.56 g) was dissolved in 20 mL toluene, and then Bu4NBr (12.5 mg) and 20 mL Na2CO3 (2 mol L 1) were added. The reaction system was degassed and refilled with argon after addition of Pd(PPh3)4 (50 mg). The mixture was stirred vigorously at 85 1C for 36 h and phenylboronic acid (0.1 g) predissolved in 1 mL tetrahydrofuran was added in 2 h. Then 1 mL bromobenzene was injected for further 3 h. Methanol (100 mL) was poured into the mixture to form a yellow precipitate. The precipitate was filtered and washed with methanol, water and acetone. The resulting solid was dissolved in CH2Cl2 (10 mL) and filtered using a 0.2 μm membrane. The filtrate liquor was concentrated and reprecipitated in methanol (30 mL). The power (364 mg, 62%) was collected by filtration, washed with methanol and dried in vacuo. The protecting group, tert-butyl esters, was removed by trifluoroacetic acid (TFA). TFA (2 mL) was added into a solution of
polymer (100 mg) in CH2Cl2 (20 mL) at room temperature for 12 h. The organic layer was washed with Mili-Q water and then stirred with NaOH (10%, 10 mL). The aqueous phase containing NaOH and PFBT was subjected to dialysis for 5 days and lyophilization to yield a water-soluble product of PFBT (297 mg, 83%). The PFBT is water-soluble and exhibits absorption and emission maxima in water at λabs ¼ 468 nm and λem ¼ 536 nm, respectively (Fig. S2). The water-soluble product of PFBT was also evaluated using FT-IR spectra shown in Fig. S3.
2.3. Preparation of functionalized magnetic nanoparticles To synthesize iron oxide nanoparticles, we adapted the solvothermal reduction method (Deng et al., 2005) for the fabrication of monodisperse, hydrophilic, and single-crystalline magnetic nanoparticles with modification. In brief, 2.7 g iron (III) chloride hexahydrate was dissolved in 80 mL ethylene glycol to form a yellow solution, followed by the addition of 7.2 g sodium acetate and 2.0 g polyethylene glycol. The system was stirred for 1 h and then sealed in an autoclave at 200 1C for 8 h. The products, magnetic nanoparticles (MNPs), were washed three times with water and ethanol. 2 g of the MNP dispersed in Mili-Q water was taken in a solution of 40 mL CH3CH2OH and 10 mL H2O. To this reaction mixture 0.05 mL of (3-aminopropyl)-trimethoxysilane was added and kept stirring for 24 h. The product, amino functionalized MNP, so formed was dried at 80 1C. To obtain quaternized magnetic nanoparticles (q-MNPs), NaOH solution (2 mol L 1, 20 mL) was added to amino functionalized MNP (0.1 g in 50 mL water), followed by dropping addition of CH3CH2CH2Br (0.50 g) predissolved in isopropanol (1.5 mL). The mixture was stirred at 50 1C for 24 h and q-MNP 1 was obtained through magnetic separation. q-MNP 2 (HOCH2CH2Br quaternized for MNP) and q-MNP 3 (C6H5CH2Br quaternized for MNP) were prepared from the above-mentioned procedures (shown in Fig. S4a).
2.4. q-MNP–fluorescent polymer constructs In fluorescent titration assay, the fluorescence intensity at 538 nm was measured with an excitation wavelength of 486 nm on a Hitachi F-4500 fluorescence spectrophotometer. During the assay, 0.1 mL of PFBT (100 nmol L 1) was placed in the centrifuge tube and the initial fluorescence intensity was recorded. Aliquots of a solution of q-MNP (0.1 mL) were subsequently added into the solution and incubated for 20 min. The q-MNP–fluorescent polymer complex was evaluated using TEM images and zeta potential (shown in Fig. S5). Then the q-MNP–fluorescent polymer complex was removed using magnetic separation and the fluorescence intensity of supernatant was measured again.
2.5. Linear discriminant analysis (LDA) for bacteria identification In the bacteria sensing assay, 0.1 mL of q-MNP–fluorescent polymer complex (0.1 mg mL 1) was placed in six separate centrifuge tubes. Subsequently, 0.1 mL of a bacterial solutions (107 cfu mL 1 or OD600 ¼0.2) was added into each tube and incubated for 20 min. Then the q-MNP was removed using magnetic separation and the fluorescence intensity of supernatant was measured. PBS was used as a negative control and a standard protein, bovine serum albumin (BSA) was used as a positive control. This process was repeated for eight bacteria to generate six replicates of each. The data matrix was processed using LDA in SPSS statistics 17.0. Similar operations were also used to analyze randomly selected bacterial sample.
Y. Wan et al. / Biosensors and Bioelectronics 55 (2014) 289–293
291
Scheme 1. Design of the PFBT conjugated q-MNP system. The response intensity is dependent on the level of displacement determined by the relative q-MNP–PFBT binding strength and bacteria cells–MNP interaction. A–F in the scheme represent bacteria of different species.
3. Results and discussion 3.1. Fabrication of q-MNP–fluorescent polymer complex To solve the issue of rapid detection and identification of pathogen, a strategy has been devised for bacterial sensing using an array of the q-MNP conjugated fluorescent polymer system. Scheme 1 shows a representation of this bacteria detection method based on the q-MNP conjugated PFBT system. In our experiments, an anionic PFBT is first associated with quaternized MNP in aqueous solution to offer a sensing system. When bacteria samples are added into the system, the anionic PFBT will be replaced on the surface of the bacterial membrane due to which both electrostatic and hydrophobic interaction play key roles in the complex system of MNP with bacterial cells. To avoid the interference of q-MNP and q-MNP–fluorescent polymer complex, these particles and complex will be removed using magnetic separation and the fluorescence intensity of the supernatant was measured for three response patterns. For the response replacement strategy, we selected PFBT and three nanoparticles (q-MNP1–q-MNP3) as sensor elements (shown in Fig. S4). This method avoids the need to develop complicated antibody- or specific recognition elementfunctionalized nanoparticles, thus allowing straightforward and complete surface response labels for bacteria detection in the absence of instability recognition elements within a short period of time. Fluorescence titration was performed to assess the binding affinity between PFBT and q-MNP. The fluorescence of PFBT was distinctly decreased and the normalized response intensities of PFBT at 538 nm were plotted versus the ratio of q-MNP and PFBT polymer (shown in Fig. S6). The complex stability constants (Ks) and association stoichiometries (n) were estimated through a mathematical model using nonlinear least-squares curve-fitting analysis (You et al., 2005). As depicted in Fig. S6, the Ks and n are dependent on the side chains of the nanoparticles. Complex stabilities vary within one order of magnitude from 1.7 105 M 1 for q-MNP 3, 4.2 105 M 1 for q-MNP 1 to 2.8 106 M 1 for q-MNP 2, and the association stoichiometries range from 213 for q-MNP 3, 420 for q-MNP 1 to 1204 for q-MNP 2. These results demonstrated that the chemical structural changes of the nanoparticles0 surface affect their interaction for PBFT polymer. Since the different binding abilities of PFBT with q-MNP 1, qMNP 2 and q-MNP 3 have been established, the q-MNP–conjugated PFBT system will be used to detect and sense bacteria. 3.2. ‘Chemical nose’ for bacteria identification As a sensitive and reliable tool, the chemical nose has been applied in sensors to measure the change of response intensity based on biological and chemical molecule recognition. As illustrated in Fig. 1a
and b, addition of bacterial samples (107 cfu mL 1) resulted in a variety of fluorescence responses due to which the PFBT was released in the solution from the competitive binding ability of different bacterial cells. For each bacteria sample, we measured its response change against the three q-MNP polymers. The fluorescence response matrix was analyzed by LDA, a method used in statistics and pattern recognition to find a linear combination of features which characterizes two or more classes of events (Martinez and Kak, 2001). The raw data of matrix were clustered into eight groups for different bacteria using canonical scores of LDA transformation (shown in Fig. 1c). Furthermore, another 32 bacterial samples were tested in an unknown condition, where the method offered a relatively low accuracy of 87.5% (Table S1) due to which the canonical scores area of E. tarda, M. luteus, and S. oneidensis show a partial overlap (Fig. 1c). The response pattern (Fig. 2a) and the three-dimensional representation (Fig. 2b) of response changes where the bacterial concentration is OD600¼0.2 are notably different from those observed from 107 cfu mL 1. As shown in Fig. 2c, LDA can accurately distinguish each bacteria response pattern without overlap between the 95% confidence ellipses. With the canonical score plot in Fig. 2c and in the supporting materials (shown in Table S2), 32 unknown bacteria samples chosen from the eight different bacteria with different concentrations can be identified and detected. Combined with UV– vis measurement, a high accuracy of 96.8% demonstrated reliability (shown in Table S2). This result certainly demonstrates that our ‘chemical nose’ holds a significant promise for identification and detection of pathogen. A series of unknown bacteria samples were successfully estimated using the ‘chemical nose’ with UV–vis measurement. This result unambiguously demonstrates that our sensor holds potential promise for both the detection and identification of bacterial targets. Recently, the efficient and cost-effective techniques based on MNP as a rapid and reliable tool for pathogen detection have already been widely used. For example, Bakthavathsalam et al. reported a rapid method for immunomagnetic separation of Salmonella along with their real time detection via PCR. It enables differentiation of Salmonella typhi and Salmonella paratyphi using a set of four specific primers with costintensiveness and technical complexity (Bakthavathsalam et al., 2013). Kwon et al. use antibody conjugated gold-coated magnetic nanoparticle clusters and magnetophoretic chromatography with a precision pipette for detection of Salmonella (Kwon et al., 2013). This method shows a relatively unstable condition due to which the antibody is easily inactivated. Compared to other analytical techniques based on MNP, it adequately serves the requirements of large-scale monitoring of environmental issues. The ‘chemical nose’ technique offers a good alternative for the biomolecules sensing.
292
Y. Wan et al. / Biosensors and Bioelectronics 55 (2014) 289–293
Fig. 1. Fluorescence response patterns (a) of q-MNP polymer complex in the presence of bacteria (107 cfu m L 1), three-dimensional representation of response changes against the three q-MNP polymer system (b) and canonical score plot (c) for response as determined with LDA. The first two factors collate 93.2% of the variance. Each value is an average of six times measurements, and the error bars are shown.
4. Conclusions A sensitive assay that used q-MNP–fluorescent polymer systems to allow the analysis and capture of bacteria has been demonstrated. In this paper, we have illustrated the assemblies' kinetics of q-MNP with a PFBT polymer that can provide an efficient sensor of pathogen. Through application of LDA, the fluorescence response pattern can be employed to identify and quantify eight different bacteria by canonical scores in a rapid, simple, and low-cost way. The sensitivity of this sensor pattern, which has yet to be fully optimized, presents a potential
Fig. 2. Fluorescence response patterns (a) of q-MNP polymer complex in the presence of bacteria (OD600¼ 0.2), three-dimensional representation of response changes against the three q-MNP polymer system (b) and canonical score plot (c) for response as determined with LDA. The first two factors collate 95.8% of the variance. Each value is an average of six times measurements, and the error bars are shown.
approach for detecting pathogenic bacteria, protein, or DNA analysis.
Acknowledgments This work was supported by the National Natural Science Foundation of China (Grant nos. 41306072 and 41076047), Science & Technology Basic Research Program of Qingdao (Grant no. 13-1-4– 181-jch) and the Shandong Provincial Natural Science Foundation, China (Grant no. BS2013HZ005).
Y. Wan et al. / Biosensors and Bioelectronics 55 (2014) 289–293
Appendix A. Supplementary material Supplementary materials associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.bios.2013. 11.080. References Arruda, D.L., Wilson, W.C., Nguyen, C., Yao, Q.W., Caiazzo, R.J., Talpasanu, I., Dow, D.E., Liu, B.C.S., 2009. Expert Rev. Mol. Diagn. 9 (7), 749–755. Bajaj, A., Miranda, O.R., Kim, I.-B., Phillips, R.L., Jerry, D.J., Bunz, U.H.F., Rotello, V.M., 2009. Proc. Natl. Acad. Sci. USA 106 (27), 10912–10916. Bakthavathsalam, P., Rajendran, V.K., Saran, U., Chatterjee, S., Ali, B.M.J., 2013. Microchim. Acta 180 (13–14), 1241–1248. Chitarra, L.G., van den Bulk, R.W., 2003. Eur. J. Plant Pathol. 109 (5), 407–417. Chung, H.J., Castro, C.M., Im, H., Lee, H., Weissleder, R., 2013. Nat. Nanotechnol. 8 (5), 369–375. Deng, H., Li, X., Peng, Q., Wang, X., Chen, J., Li, Y., 2005. Angew. Chem. Int. Ed. 44 (18), 2782–2785. Folmer-Andersen, J.F., Kitamura, M., Anslyn, E.V., 2006. J. Am. Chem. Soc. 128 (17), 5652–5653. Gao, J.H., Gu, H.W., Xu, B., 2009. Acc. Chem. Res. 42 (8), 1097–1107.
293
Gracias, K.S., McKillip, J.L., 2004. Can. J. Microbiol. 50 (11), 883–890. Kaittanis, C., Naser, S.A., Perez, J.M., 2006. Nano Lett. 7 (2), 380–383. Kwon, D., Joo, J., Lee, J., Park, K.H., Jeon, S., 2013. Anal. Chem. 85 (15), 7594–7598. Lee, J.W., Lee, J.-S., Chang, Y.-T., 2006. Angew. Chem. Int. Ed. 45 (39), 6485–6487. Lu, Y., Dong, L., Zhang, L.C., Su, Y.D., Yu, S.H., 2012. Nano Today 7 (4), 297–315. Martinez, A.M., Kak, A.C., 2001. IEEE T. Pattern Anal. 23 (2), 228–233. Phillips, R.L., Miranda, O.R., You, C.-C., Rotello, V.M., Bunz, U.H.F., 2008. Angew. Chem. Int. Ed. 47 (14), 2590–2594. Ravindranath, S.P., Mauer, L.J., Deb-Roy, C., Irudayaraj, J., 2009. Anal. Chem. 81 (8), 2840–2846. Sanvicens, N., Pastells, C., Pascual, N., Marco, M.P., 2009. TRAC-Trend Anal. Chem. 28 (11), 1243–1252. Shao, H.L., Min, C., Issadore, D., Liong, M., Yoon, T.J., Weissleder, R., Lee, H., 2010. Theranostics 2 (1), 55–65. Shao, H.L., Yoon, T.J., Liong, M., Weissleder, R., Lee, H., 2009. Beilstein J. Nanotechnol. 1, 142–154. Varshney, M., Yang, L.J., Su, X.L., Li, Y.B., 2005. J. Food Prot. 68 (9), 1804–1811. Wan, Y., Zhang, D., Hou, B.R., 2010. Biosens. Bioelectron. 25 (7), 1847–1850. Yamamoto, Y., 2002. Clin. Diagn. Lab. Immunol. 9 (3), 508–514. You, C.-C., De, M., Han, G., Rotello, V.M., 2005. J. Am. Chem. Soc. 127 (37), 12873–12881. You, C.-C., Miranda, O.R., Gider, B., Ghosh, P.S., Kim, I.-B., Erdogan, B., Krovi, S.A., Bunz, U.H.F., Rotello, V.M., 2007. Nat. Nanotechnol. 2 (5), 318–323. Zhang, Y., Liu, B., Cao, Y., 2008. Chem. Asian J. 3 (4), 739–745.
Contents lists available at ScienceDirect
Biosensors and Bioelectronics journal homepage: www.elsevier.com/locate/bios
Short communication
Quaternized magnetic nanoparticles–fluorescent polymer system for detection and identification of bacteria Yi Wan a, Yan Sun a, Peng Qi a,b, Peng Wang a, Dun Zhang a,n a Key Laboratory of Marine Environmental Corrosion and Bio-fouling, Institute of Oceanology, Chinese Academy of Sciences, 7 Nanhai Road, Qingdao 266071, China b University of Chinese Academy of Sciences, 19 (Jia) Yuquan Road, Beijing 100039, China
art ic l e i nf o
a b s t r a c t
Article history: Received 9 October 2013 Received in revised form 18 November 2013 Accepted 25 November 2013 Available online 12 December 2013
Nanomaterial-based ‘chemical nose’ sensor with sufficient sensing specificity is a useful analytical tool for the detection of toxicologically important substances in complicated biological systems. A sensor array containing three quaternized magnetic nanoparticles (q-MNPs)–fluorescent polymer systems has been designed to identify and quantify bacteria. The bacterial cell membranes disrupt the q-MNP– fluorescent polymer, generating unique fluorescence response array. The response intensity of the array is dependent on the level of displacement determined by the relative q-MNP–fluorescent polymer binding strength and bacteria cells–MNP interaction. These characteristic responses show a highly repeatable bacteria cells and can be differentiated by linear discriminant analysis (LDA). Based on the array response matrix from LDA, our approach has been used to measure bacteria with an accuracy of 87.5% for 107 cfu mL 1 within 20 min. Combined with UV–vis measurement, the method can be successfully performed to identify and detect eight different pathogen samples with an accuracy of 96.8%. The measurement system has a potential for further applications and provides a facile and simple method for the rapid analysis of protein, DNA, and pathogens. & 2013 Elsevier B.V. All rights reserved.
Keywords: Quaternized magnetic nanoparticles Bacteria detection Fluorescent polymer Linear discriminant analysis
1. Introduction The determination, identification, and quantification of pathogen for clinical diagnosis, environmental monitoring, and food safety are crucial for public health protection. Traditional methods for bacteria detection, such as plating and culturing (Gracias and McKillip, 2004), enzyme-linked immunosorbent assay (Chitarra and van den Bulk, 2003), and polymerase chain reaction (Yamamoto, 2002), involve a pre-enrichment step or a selective enrichment step followed by a biochemical test, and the sophisticated series of assays required can take up from several hours to several days. Although these approaches for bacterial detection are sufficiently sensitive and selective, most have several drawbacks such as being time-consuming, cost-intensive, or technically complex. Increasing worldwide demand and concern for their safe development and use require a simple, stable, and sensitive detection assay for pathogen evaluation and environmental monitoring. Over the past decade, many efforts have been made to use efficient and cost-effective techniques based on magnetic nanoparticles (MNPs) as a rapid and reliable tool for pathogen detection (Arruda et al., 2009; Gao et al., 2009; Lu et al., 2012; Sanvicens
n
Corresponding author. Tel./fax: þ 86 532 82898960. E-mail addresses: [email protected], [email protected] (D. Zhang).
0956-5663/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bios.2013.11.080
et al., 2009; Shao et al., 2009, 2010). Several immunomagnetic approaches have been developed for label-free detection of pathogen. For example, the immunomagnetic separation with MNP–antibody conjugates was studied and estimated for pathogenic E. coli in a complex environment. MNP–antibody conjugates were prepared by immobilizing biotin-labeled polyclonal antibodies onto streptavidin-coated MNP or direct chemical immobilization (Ravindranath et al., 2009; Varshney et al., 2005). Our group reported a fast, sensitive and reliable biosensor based on the amplification of the response of vancomycin-coated MNP for the detection of marine pathogen under an external magnetic field (Wan et al., 2010). The feature characteristics of these MNP-based analysis techniques result from the simplification of separation and enrichment steps, or from their high specificity, which allow the detection of specific bacteria in complicated matrices. Recently, some MNPs have also been used as signal labels in analytical techniques. One example is proposed by Kaittanis and his co-workers, who functionalized superparamagnetic iron oxide nanoparticles for quantification of bacteria in clinical and environmental samples through magnetic relaxation (Kaittanis et al., 2006). Another is the work by Weissleder's group, who designed an MNP hybridization assay, involving ubiquitous and specific bacterial probes that target bacterial DNA segments, to detect amplified target bacterial DNAs using a miniaturized nuclear magnetic resonance device (Chung et al., 2013). However, whilst these techniques are relevant for laboratory use,
290
Y. Wan et al. / Biosensors and Bioelectronics 55 (2014) 289–293
they cannot adequately serve the requirements of health practitioners or enable large-scale monitoring of environmental issues. Given these disadvantages, an effort has been made to develop a facile and simple method for target analysis that is labor- and time-saving. The ‘chemical nose’ technique offers a good alternative for biomolecules sensing. In this strategy, a chemical nose needs to be trained with qualified biological samples so as to build a database of reference. Then the sensing system can recognize new samples by comparing biomolecule database. This method has been used to sense a lot of targets, including amino acid (Folmer-Andersen et al., 2006), carbohydrates (Lee et al., 2006), protein (You et al., 2007), bacteria (Phillips et al., 2008) and cancer cells (Bajaj et al., 2009). For example, Phillips et al. who address the issue of rapid identification of bacteria, developed for bacterial sensing using an array of gold–nanoparticle-conjugated polymer constructs (Phillips et al., 2008). However, the presence of the constructs in a well on plate may disrupt the response pattern of the ‘chemical nose.’ In our work, we designed a q-MNP–fluorescent polymer system for detection and identification of bacteria. Three q-MNP–fluorescent polymer constructs are combined to differentiate bacterial cells according to their characteristic response pattern. To avoid the interference of q-MNP–fluorescent polymer constructs in the measurement system, the constructs can be separated quickly by a magnet.
2. Material and methods 2.1. Bacterial cultivation All bacteria were offered from the Key Laboratory of Experimental Marine Biology. Bacteria were seeded and cultured in suspension using the following media: S. oneidensis, V. fischeri, M. luteus, E. tarda and E. coli in Luria–Bertani media; V. alginolyticus and P. aeruginosa in trypticase soy broth; P. pastoris in yeast extract peptone dextrose medium. For these bacteria, a single colony was inoculated in bacterial medium at 37 1C overnight, shaking at 200 rpm. After centrifugation (6000 rpm) for 10 min and PBS washing, the bacterial cells were diluted to the desired concentration or optical density in phosphate buffer. 2.2. Synthesis of fluorescent polymer A polymer functionalized with side-chain carboxylic acid group (PFBT) was synthesized using a method reported with modification as shown in Fig. S1 (Zhang et al., 2008). A mixture of 2,7dibromo-9,9-bis(3-(tert-butyl propanoate))fluorine (monomer A, 0.232 g), 4,7-dibromobenzo[c][1,2,5]thiadiazole (monomer B, 0.176 g), and 9,9-dioctylfluorene-2,7-diboronic acid bis(1,3-propanediol) ester (monomer C, 0.56 g) was dissolved in 20 mL toluene, and then Bu4NBr (12.5 mg) and 20 mL Na2CO3 (2 mol L 1) were added. The reaction system was degassed and refilled with argon after addition of Pd(PPh3)4 (50 mg). The mixture was stirred vigorously at 85 1C for 36 h and phenylboronic acid (0.1 g) predissolved in 1 mL tetrahydrofuran was added in 2 h. Then 1 mL bromobenzene was injected for further 3 h. Methanol (100 mL) was poured into the mixture to form a yellow precipitate. The precipitate was filtered and washed with methanol, water and acetone. The resulting solid was dissolved in CH2Cl2 (10 mL) and filtered using a 0.2 μm membrane. The filtrate liquor was concentrated and reprecipitated in methanol (30 mL). The power (364 mg, 62%) was collected by filtration, washed with methanol and dried in vacuo. The protecting group, tert-butyl esters, was removed by trifluoroacetic acid (TFA). TFA (2 mL) was added into a solution of
polymer (100 mg) in CH2Cl2 (20 mL) at room temperature for 12 h. The organic layer was washed with Mili-Q water and then stirred with NaOH (10%, 10 mL). The aqueous phase containing NaOH and PFBT was subjected to dialysis for 5 days and lyophilization to yield a water-soluble product of PFBT (297 mg, 83%). The PFBT is water-soluble and exhibits absorption and emission maxima in water at λabs ¼ 468 nm and λem ¼ 536 nm, respectively (Fig. S2). The water-soluble product of PFBT was also evaluated using FT-IR spectra shown in Fig. S3.
2.3. Preparation of functionalized magnetic nanoparticles To synthesize iron oxide nanoparticles, we adapted the solvothermal reduction method (Deng et al., 2005) for the fabrication of monodisperse, hydrophilic, and single-crystalline magnetic nanoparticles with modification. In brief, 2.7 g iron (III) chloride hexahydrate was dissolved in 80 mL ethylene glycol to form a yellow solution, followed by the addition of 7.2 g sodium acetate and 2.0 g polyethylene glycol. The system was stirred for 1 h and then sealed in an autoclave at 200 1C for 8 h. The products, magnetic nanoparticles (MNPs), were washed three times with water and ethanol. 2 g of the MNP dispersed in Mili-Q water was taken in a solution of 40 mL CH3CH2OH and 10 mL H2O. To this reaction mixture 0.05 mL of (3-aminopropyl)-trimethoxysilane was added and kept stirring for 24 h. The product, amino functionalized MNP, so formed was dried at 80 1C. To obtain quaternized magnetic nanoparticles (q-MNPs), NaOH solution (2 mol L 1, 20 mL) was added to amino functionalized MNP (0.1 g in 50 mL water), followed by dropping addition of CH3CH2CH2Br (0.50 g) predissolved in isopropanol (1.5 mL). The mixture was stirred at 50 1C for 24 h and q-MNP 1 was obtained through magnetic separation. q-MNP 2 (HOCH2CH2Br quaternized for MNP) and q-MNP 3 (C6H5CH2Br quaternized for MNP) were prepared from the above-mentioned procedures (shown in Fig. S4a).
2.4. q-MNP–fluorescent polymer constructs In fluorescent titration assay, the fluorescence intensity at 538 nm was measured with an excitation wavelength of 486 nm on a Hitachi F-4500 fluorescence spectrophotometer. During the assay, 0.1 mL of PFBT (100 nmol L 1) was placed in the centrifuge tube and the initial fluorescence intensity was recorded. Aliquots of a solution of q-MNP (0.1 mL) were subsequently added into the solution and incubated for 20 min. The q-MNP–fluorescent polymer complex was evaluated using TEM images and zeta potential (shown in Fig. S5). Then the q-MNP–fluorescent polymer complex was removed using magnetic separation and the fluorescence intensity of supernatant was measured again.
2.5. Linear discriminant analysis (LDA) for bacteria identification In the bacteria sensing assay, 0.1 mL of q-MNP–fluorescent polymer complex (0.1 mg mL 1) was placed in six separate centrifuge tubes. Subsequently, 0.1 mL of a bacterial solutions (107 cfu mL 1 or OD600 ¼0.2) was added into each tube and incubated for 20 min. Then the q-MNP was removed using magnetic separation and the fluorescence intensity of supernatant was measured. PBS was used as a negative control and a standard protein, bovine serum albumin (BSA) was used as a positive control. This process was repeated for eight bacteria to generate six replicates of each. The data matrix was processed using LDA in SPSS statistics 17.0. Similar operations were also used to analyze randomly selected bacterial sample.
Y. Wan et al. / Biosensors and Bioelectronics 55 (2014) 289–293
291
Scheme 1. Design of the PFBT conjugated q-MNP system. The response intensity is dependent on the level of displacement determined by the relative q-MNP–PFBT binding strength and bacteria cells–MNP interaction. A–F in the scheme represent bacteria of different species.
3. Results and discussion 3.1. Fabrication of q-MNP–fluorescent polymer complex To solve the issue of rapid detection and identification of pathogen, a strategy has been devised for bacterial sensing using an array of the q-MNP conjugated fluorescent polymer system. Scheme 1 shows a representation of this bacteria detection method based on the q-MNP conjugated PFBT system. In our experiments, an anionic PFBT is first associated with quaternized MNP in aqueous solution to offer a sensing system. When bacteria samples are added into the system, the anionic PFBT will be replaced on the surface of the bacterial membrane due to which both electrostatic and hydrophobic interaction play key roles in the complex system of MNP with bacterial cells. To avoid the interference of q-MNP and q-MNP–fluorescent polymer complex, these particles and complex will be removed using magnetic separation and the fluorescence intensity of the supernatant was measured for three response patterns. For the response replacement strategy, we selected PFBT and three nanoparticles (q-MNP1–q-MNP3) as sensor elements (shown in Fig. S4). This method avoids the need to develop complicated antibody- or specific recognition elementfunctionalized nanoparticles, thus allowing straightforward and complete surface response labels for bacteria detection in the absence of instability recognition elements within a short period of time. Fluorescence titration was performed to assess the binding affinity between PFBT and q-MNP. The fluorescence of PFBT was distinctly decreased and the normalized response intensities of PFBT at 538 nm were plotted versus the ratio of q-MNP and PFBT polymer (shown in Fig. S6). The complex stability constants (Ks) and association stoichiometries (n) were estimated through a mathematical model using nonlinear least-squares curve-fitting analysis (You et al., 2005). As depicted in Fig. S6, the Ks and n are dependent on the side chains of the nanoparticles. Complex stabilities vary within one order of magnitude from 1.7 105 M 1 for q-MNP 3, 4.2 105 M 1 for q-MNP 1 to 2.8 106 M 1 for q-MNP 2, and the association stoichiometries range from 213 for q-MNP 3, 420 for q-MNP 1 to 1204 for q-MNP 2. These results demonstrated that the chemical structural changes of the nanoparticles0 surface affect their interaction for PBFT polymer. Since the different binding abilities of PFBT with q-MNP 1, qMNP 2 and q-MNP 3 have been established, the q-MNP–conjugated PFBT system will be used to detect and sense bacteria. 3.2. ‘Chemical nose’ for bacteria identification As a sensitive and reliable tool, the chemical nose has been applied in sensors to measure the change of response intensity based on biological and chemical molecule recognition. As illustrated in Fig. 1a
and b, addition of bacterial samples (107 cfu mL 1) resulted in a variety of fluorescence responses due to which the PFBT was released in the solution from the competitive binding ability of different bacterial cells. For each bacteria sample, we measured its response change against the three q-MNP polymers. The fluorescence response matrix was analyzed by LDA, a method used in statistics and pattern recognition to find a linear combination of features which characterizes two or more classes of events (Martinez and Kak, 2001). The raw data of matrix were clustered into eight groups for different bacteria using canonical scores of LDA transformation (shown in Fig. 1c). Furthermore, another 32 bacterial samples were tested in an unknown condition, where the method offered a relatively low accuracy of 87.5% (Table S1) due to which the canonical scores area of E. tarda, M. luteus, and S. oneidensis show a partial overlap (Fig. 1c). The response pattern (Fig. 2a) and the three-dimensional representation (Fig. 2b) of response changes where the bacterial concentration is OD600¼0.2 are notably different from those observed from 107 cfu mL 1. As shown in Fig. 2c, LDA can accurately distinguish each bacteria response pattern without overlap between the 95% confidence ellipses. With the canonical score plot in Fig. 2c and in the supporting materials (shown in Table S2), 32 unknown bacteria samples chosen from the eight different bacteria with different concentrations can be identified and detected. Combined with UV– vis measurement, a high accuracy of 96.8% demonstrated reliability (shown in Table S2). This result certainly demonstrates that our ‘chemical nose’ holds a significant promise for identification and detection of pathogen. A series of unknown bacteria samples were successfully estimated using the ‘chemical nose’ with UV–vis measurement. This result unambiguously demonstrates that our sensor holds potential promise for both the detection and identification of bacterial targets. Recently, the efficient and cost-effective techniques based on MNP as a rapid and reliable tool for pathogen detection have already been widely used. For example, Bakthavathsalam et al. reported a rapid method for immunomagnetic separation of Salmonella along with their real time detection via PCR. It enables differentiation of Salmonella typhi and Salmonella paratyphi using a set of four specific primers with costintensiveness and technical complexity (Bakthavathsalam et al., 2013). Kwon et al. use antibody conjugated gold-coated magnetic nanoparticle clusters and magnetophoretic chromatography with a precision pipette for detection of Salmonella (Kwon et al., 2013). This method shows a relatively unstable condition due to which the antibody is easily inactivated. Compared to other analytical techniques based on MNP, it adequately serves the requirements of large-scale monitoring of environmental issues. The ‘chemical nose’ technique offers a good alternative for the biomolecules sensing.
292
Y. Wan et al. / Biosensors and Bioelectronics 55 (2014) 289–293
Fig. 1. Fluorescence response patterns (a) of q-MNP polymer complex in the presence of bacteria (107 cfu m L 1), three-dimensional representation of response changes against the three q-MNP polymer system (b) and canonical score plot (c) for response as determined with LDA. The first two factors collate 93.2% of the variance. Each value is an average of six times measurements, and the error bars are shown.
4. Conclusions A sensitive assay that used q-MNP–fluorescent polymer systems to allow the analysis and capture of bacteria has been demonstrated. In this paper, we have illustrated the assemblies' kinetics of q-MNP with a PFBT polymer that can provide an efficient sensor of pathogen. Through application of LDA, the fluorescence response pattern can be employed to identify and quantify eight different bacteria by canonical scores in a rapid, simple, and low-cost way. The sensitivity of this sensor pattern, which has yet to be fully optimized, presents a potential
Fig. 2. Fluorescence response patterns (a) of q-MNP polymer complex in the presence of bacteria (OD600¼ 0.2), three-dimensional representation of response changes against the three q-MNP polymer system (b) and canonical score plot (c) for response as determined with LDA. The first two factors collate 95.8% of the variance. Each value is an average of six times measurements, and the error bars are shown.
approach for detecting pathogenic bacteria, protein, or DNA analysis.
Acknowledgments This work was supported by the National Natural Science Foundation of China (Grant nos. 41306072 and 41076047), Science & Technology Basic Research Program of Qingdao (Grant no. 13-1-4– 181-jch) and the Shandong Provincial Natural Science Foundation, China (Grant no. BS2013HZ005).
Y. Wan et al. / Biosensors and Bioelectronics 55 (2014) 289–293
Appendix A. Supplementary material Supplementary materials associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.bios.2013. 11.080. References Arruda, D.L., Wilson, W.C., Nguyen, C., Yao, Q.W., Caiazzo, R.J., Talpasanu, I., Dow, D.E., Liu, B.C.S., 2009. Expert Rev. Mol. Diagn. 9 (7), 749–755. Bajaj, A., Miranda, O.R., Kim, I.-B., Phillips, R.L., Jerry, D.J., Bunz, U.H.F., Rotello, V.M., 2009. Proc. Natl. Acad. Sci. USA 106 (27), 10912–10916. Bakthavathsalam, P., Rajendran, V.K., Saran, U., Chatterjee, S., Ali, B.M.J., 2013. Microchim. Acta 180 (13–14), 1241–1248. Chitarra, L.G., van den Bulk, R.W., 2003. Eur. J. Plant Pathol. 109 (5), 407–417. Chung, H.J., Castro, C.M., Im, H., Lee, H., Weissleder, R., 2013. Nat. Nanotechnol. 8 (5), 369–375. Deng, H., Li, X., Peng, Q., Wang, X., Chen, J., Li, Y., 2005. Angew. Chem. Int. Ed. 44 (18), 2782–2785. Folmer-Andersen, J.F., Kitamura, M., Anslyn, E.V., 2006. J. Am. Chem. Soc. 128 (17), 5652–5653. Gao, J.H., Gu, H.W., Xu, B., 2009. Acc. Chem. Res. 42 (8), 1097–1107.
293
Gracias, K.S., McKillip, J.L., 2004. Can. J. Microbiol. 50 (11), 883–890. Kaittanis, C., Naser, S.A., Perez, J.M., 2006. Nano Lett. 7 (2), 380–383. Kwon, D., Joo, J., Lee, J., Park, K.H., Jeon, S., 2013. Anal. Chem. 85 (15), 7594–7598. Lee, J.W., Lee, J.-S., Chang, Y.-T., 2006. Angew. Chem. Int. Ed. 45 (39), 6485–6487. Lu, Y., Dong, L., Zhang, L.C., Su, Y.D., Yu, S.H., 2012. Nano Today 7 (4), 297–315. Martinez, A.M., Kak, A.C., 2001. IEEE T. Pattern Anal. 23 (2), 228–233. Phillips, R.L., Miranda, O.R., You, C.-C., Rotello, V.M., Bunz, U.H.F., 2008. Angew. Chem. Int. Ed. 47 (14), 2590–2594. Ravindranath, S.P., Mauer, L.J., Deb-Roy, C., Irudayaraj, J., 2009. Anal. Chem. 81 (8), 2840–2846. Sanvicens, N., Pastells, C., Pascual, N., Marco, M.P., 2009. TRAC-Trend Anal. Chem. 28 (11), 1243–1252. Shao, H.L., Min, C., Issadore, D., Liong, M., Yoon, T.J., Weissleder, R., Lee, H., 2010. Theranostics 2 (1), 55–65. Shao, H.L., Yoon, T.J., Liong, M., Weissleder, R., Lee, H., 2009. Beilstein J. Nanotechnol. 1, 142–154. Varshney, M., Yang, L.J., Su, X.L., Li, Y.B., 2005. J. Food Prot. 68 (9), 1804–1811. Wan, Y., Zhang, D., Hou, B.R., 2010. Biosens. Bioelectron. 25 (7), 1847–1850. Yamamoto, Y., 2002. Clin. Diagn. Lab. Immunol. 9 (3), 508–514. You, C.-C., De, M., Han, G., Rotello, V.M., 2005. J. Am. Chem. Soc. 127 (37), 12873–12881. You, C.-C., Miranda, O.R., Gider, B., Ghosh, P.S., Kim, I.-B., Erdogan, B., Krovi, S.A., Bunz, U.H.F., Rotello, V.M., 2007. Nat. Nanotechnol. 2 (5), 318–323. Zhang, Y., Liu, B., Cao, Y., 2008. Chem. Asian J. 3 (4), 739–745.
