Anal Bioanal Chem DOI 10.1007/s00216-014-7947-9
PAPER IN FOREFRONT
Microfluidic biosensor for cholera toxin detection in fecal samples Natinan Bunyakul & Chamras Promptmas & Antje J. Baeumner
Received: 30 March 2014 / Revised: 28 May 2014 / Accepted: 4 June 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract Sample preparation and processing steps are the most critical assay aspects that require our attention in the development of diagnostic devices for analytes present in complex matrices. In the best scenarios, diagnostic devices should use only simple sample processing. We have therefore investigated minimal preparation of stool samples and their effect on our sensitive microfluidic immunosensor for the detection of cholera toxin. This biosensor was previously developed and tested in buffer solutions only, using either fluorescence or electrochemical detection strategies. The microfluidic devices were made from polydimethylsiloxane using soft lithography and silicon templates. Cholera toxin subunit B (CTB)-specific antibodies immobilized onto superparamagnetic beads and ganglioside GM1-containing liposomes were used for CTB recognition in the detection system. Quantification of CTB was tested by spiking it in human stool samples. Here, optimal minimal sample processing steps, including filtration and centrifugation, were optimized using a microtiter plate assay owing to its highthroughput capabilities. Subsequently, it was transferred to the microfluidic systems, enhancing the diagnostic Published in the topical collection celebrating ABCs 13th Anniversary. N. Bunyakul : C. Promptmas Department of Clinical Chemistry, Faculty of Medical Technology, Mahidol University, Nakhon Pathom 73170, Thailand A. J. Baeumner (*) Department of Biological and Environmental Engineering, Cornell University, Ithaca, NY 14853, USA e-mail: [email protected] A. J. Baeumner e-mail: [email protected] Present Address: A. J. Baeumner Institute of Analytical Chemistry, Chemo- and Biosensors, University of Regensburg, 93053 Regensburg, Germany
characteristic of the biosensor. It was found that the debris removal obtained through simple centrifugation resulted in an acceptable removal of matrix effects for the fluorescence format, reaching a limit of detection of only 9.0 ng/mL. However, the electron transfer in the electrochemical format was slightly negatively affected (limit of detection of 31.7 ng/ mL). Subsequently, cross-reactivity using the heat-labile Escherichia coli toxin was investigated using the electrochemical microfluidic immunosensors and was determined to be negligible. With minimal sample preparation required, these microfluidic liposome-based systems have demonstrated excellent analytical performance in a complex matrix and will thus be applicable to other sample matrices. Keywords Biosensor . Microfluidic . Liposomes . Cholera toxin . Fecal sample
Introduction From the introduction of the first microfluidic device developed in the 1970s [1], microfluidic device technology has entered various fields, including chemical, biological, and diagnostic analyses, owing to its advantages in terms of the small amounts of reagent and sample required, high sensitivity, rapid response, portability, and its simple adaptation to multifunctionality and high-throughput analyses [2]. Diagnostic assay development is moving toward the use of micro total analysis systems in which detection and sample preparation are integrated. In the case of microfluidic immunosensors, antibodies are used as biorecognition elements, similarly to traditional immune biosensors, which find ample application in point-of-care diagnostics [3]. Both optical and electrochemical detection approaches have been realized with microfluidic immunosensors. Their application has been successfully demonstrated for a variety of analytes. For
N. Bunyakul et al.
example, optical microfluidic immunosensors have been demonstrated to be successful in the detection of cardiac biomarkers [4], viral particles [5, 6], pathogen-specific immunoglobulins [7], and cancer biomarkers [8]. Electrochemical microfluidic immunosensors also have been developed for many analytical applications, such as for cancer biomarkers [9–11], mycotoxins [12], bacterial toxins [13], and cardiac biomarkers [14]. Cholera is an acute diarrheal infection caused by ingestion of food or water contaminated with the bacterium Vibrio cholerae. The patient can die within an hour if left untreated. Every year, there are an estimated three million to five million cholera cases of cholera and 100,000–120 000 deaths from cholera. The short incubation period of 2 h to 5 days compounds the potentially explosive pattern of outbreaks. Cholera is now endemic in many countries [15]. Cholera toxin produced from V. cholerae is the actual causative agent of the disease. Cholera toxin is composed of two subunits, cholera toxin subunit A and cholera toxin subunit B (CTB). The B subunit’s function is to bind specifically to the ganglioside GM1 receptor on intestinal cell walls, whereas the A subunit is the active function protein that activates the production of cyclic AMP, causing a dramatic efflux of ions and water from infected enterocytes, leading to watery diarrhea [16]. In a recent study [17], we developed a microfluidic immunosensor with GM1-containing liposome for CTB detection in purified samples. Two assay formats were developed, i.e., fluorescence and electrochemical. Good sensitivity was found for both assay formats, and electrochemical signal recording was significantly simpler than the optical approach. In the present study, simple sample preparation procedures, real sample matrices, and their effect on both formats were investigated. The microfluidic assay with sample preparation procedures and the biological recognition principles are shown schematically in Fig. 1.
Tygon® tubing was supplied by Small Parts (Miami Lakes, FL, USA). PEEK™ tubing was obtained from Upchurch Scientific. (Oak Harbor, WA, USA). Syringes and syringe filters were purchased from Becton, Dickinson (Franklin Lakes, NJ, USA). Rare earth neodymium–iron–boron disc magnets were obtained from National Imports (Vienna, VA, USA). The silicone elastomer kit, Sylgard®184 containing polydimethylsiloxane (PDMS) prepolymer and catalyst, was obtained from Dow Corning (Midland, MI, USA). The Cornell Nanofabrication Facility provided clean-room facilities, chemicals, and equipment for silicon template fabrication. The Plexiglas® housing was constructed in the machine shop located in the School of Chemical and Biomolecular Engineering, Cornell University. Stool sample collection and simulated (spiked) stool sample preparation Watery stools from patients with diarrhea admitted to Nakhon Pathom Hospital were collected, pooled (total approximately 300 mL), and sent to the laboratory to serve as stool sample matrices for the CTB spiking study. Prior to shipment to the laboratory, the pooled samples were stripped of all patient information. The pooled stool sample was stored at -20 °C in order to minimize the growth of bacteria. Prior to the spiking with CTB, a standard bacterial culture was performed, verifying that no V. cholerae bacteria were present in the pooled sample received. The simulated (spiked) stool samples were then prepared by adding known concentrations of purified CTB to the original stool samples. The samples were mixed well in a microtube. The same procedure was followed for each CTB concentration studied. The spiked stool samples were kept at 4 ° C prior to analysis and were used within 9 h after spiking. Interdigitated ultramicroelectrode array fabrication
Experimental Materials and reagents General laboratory chemicals, including n-octyl-β- D glucopyranoside (OG) and buffer reagents, were purchased from VWR Scientific Products, (New York, NY, USA). Sulforhodamine B (SRB) was purchased from Molecular Probes (Eugene, OR, USA). GM1 ganglioside and the nontoxic form of cholera toxin, i.e., its subunit B (CTB), were supplied by Calbiochem (San Diego, CA, USA). Heat-labile toxins from Escherichia coli were obtained from Quadratech Diagnostics. The biotinylated anti-CTB was purchased from United States Biological (Swampscott, MA, USA). Superparamagnetic beads (Dynabeads MyOne streptavidin) were obtained from Dynal Biotech (Lake Success, NY, USA).
The interdigitated ultramicroelectrode arrays (IDUAs) were fabricated by standard clean-room photolithographic and liftoff techniques on glass wafers, as described previously [18]. Briefly, with use of GCA Autostep 200 DSW i-line wafer stepper, the pattern was transferred to Pyrex glass wafers (Corning 7740; Corning, Corning, NY, USA) coated with positive photoresist (SPR955-CM-0.9, Shipley, Marlborough, MA, USA). Then, image reversal was performed using NH3 gas in a YES-58SM image reversal oven. The wafers were flushed using an HTG system III-HR contact aligner. Subsequently, 70 Å of titanium and 500 Å of gold were evaporated using an e-gun source (CVC 4500 evaporator). The wafers were then soaked in Microposit remover 1165 (Shipley, Marlborough, MA, USA) to lift off the excess metal and photoresist, leaving the patterned electrode arrays. The
Microfluidic biosensor for cholera toxin detection in fecal
Fig. 1 The microfluidic assay for cholera toxin subunit B (CTB) with sample preparation procedures (from a CTB-spiked sample) and the biological recognition principles
resulting IDUAs had 420 pairs of fingers with widths of 2.5 μm, lengths of 1,000 μm, and gap sizes of 4.5 μm. Microchannel fabrication and construction of the device housing Figure 2 shows the microfluidic devices for both fluorescence and electrochemical formats. The microchannels with 50-μm depths were produced by standard photolithography and dry etching techniques as described previously [17, 19, 20]. With use of soft lithography, the microchannels were realized in PDMS covered with a glass slide and a glass slide bearing the IDUA for the fluorescence and electrochemical formats, respectively. The PDMS microchannels and the glass slide were placed between two Plexiglas® plates with the glass slide below and the microchannels above. The top Plexiglas® plate had PEEK™ tubing glued into locations lined up with the inlet and outlet holes of the microchannel. The top plate also had another hole to accommodate the permanent magnet required to capture magnetic beads in the detection zone. The two Plexiglas® plates were held in place by four or eight screws to provide sufficient pressure for leakproof sealing between the PDMS and glass slide layers. Preparation of the GM1-containing liposomes for the fluorescence and electrochemical assay formats Reverse-phase evaporation was used during liposome preparation for both the fluorescence and the electrochemical
format, as described previously [17]. SRB and ferricyanide/ ferrocyanide were the encapsulated markers for the fluorescence and electrochemical assay formats, respectively. The average sizes of both types of liposomes were determined by a DynaPro™ dynamic light scattering machine (Wyatt Technology, Santa Barbara, CA, USA). The phospholipid concentration of SRB-encapsulating GM1-containing liposomes was determined using Bartlett’s assay [21], wherein one mole of phosphorus represents one mole of phospholipid. The phospholipid concentration of ferricyanide/ferrocyanideencapsulating GM1-containing liposomes cannot be determined from Bartlett’s assay because phosphorus from a potassium phosphate buffer, which is used for preparing the liposomes, will interfere with the phosphorus determination.
Immunomagnetic separation microtiter plate assay for CTB in the stool samples Ten micrograms of the anti-CTB antibody conjugated beads, prepared as described previously [17], was added to each well of the 0.01 % (w/v) bovine serum albumin blocked microtiter plates. Subsequently, CTB-spiked stool samples were centrifuged at 8,000 rcf for 20 min, and 100 μL of the supernatant solution was added to each well using pipette mixing. The plate was then incubated in the dark at room temperature with 30 min of shaking. After magnetic separation on a magnetic separator plate (3 min), the supernatant was removed and tris(hydroxymethyl)aminomethane-buffered saline (200 μL per well) was added as a washing step. This washing was repeated twice [1× HEPES saline sucrose buffer (HSS) instead
N. Bunyakul et al.
Fig. 2 The microfluidic devices for cholera toxin subunit B detection: a fluorescence detection format; b electrochemical detection format
of tris(hydroxymethyl)aminomethane-buffered saline was used din the final step]. Then, 100 μL of SRB-encapsulating GM1-containing liposomes with a phospholipid concentration of 0.01 mM was added to each well and the resulting mixtures were incubated in the dark with 30 min of shaking. The plate was then washed three times as described before using 1× HSS buffer. The supernatant was aspirated and 100 μL of 30 mM OG was added to each well for 10 min at room temperature with shaking to lyse the bounded liposomes in the plate. Finally, 50 μL of the supernatant solution was transferred into an empty unblocked well of the black microtiter plate for measurement of the fluorescence signal (545 nm for λex and 590 nm for λem) using a microplate fluorescence reader (Wallac Victor 1420, PerkinElmer).
Fluorescence microfluidic assay for CTB in the stool samples CTB-spiked stool samples were centrifuged at 8,000 rcf for 20 min, and 1 μL of the supernatant solution was mixed with 1 μL of anti-CTB antibody conjugated magnetic bead solution at 1 μg μL-1, 1 μL of SRB-encapsulating GM1-containing liposome solution at a phospholipid concentration of 0.075 mM, and 1 μL of 1× HSS buffer (pH 7.0) in a microcentrifuge tube. The sample mixture was then incubated at room temperature with shaking for 30 min. Subsequently, the sample mixture was applied into the sample injection inlet of the optical format microfluidic chip by using a syringe pump (New Era Pump Systems, Farmingdale, NY, USA) at a flow rate of 5 μL min-1, followed immediately by 26 μL of 1× HSS buffer at the same flow rate. During these two steps, the liposome–CTB– bead complexes formed were immobilized at the magnet area and nonbound liposomes were washed away. Then, 30 mM OG solution was applied through the detergent injection inlet at 0.8 μL min-1 [19] in order to lyse the liposomes and release the SRB molecules into the microchannel. Fluorescence of lysed liposomes in the detection zone was visualized using an IX70 inverted microscope (Olympus, USA). The fluorescence images of the detection zone were obtained with a CCD camera
(Olympus, USA) coupled with Image-Pro Plus (Media Cybernetics, Rockville, MD, USA) for color intensity measurement. The average pixel value of the detection zone was considered to be the fluorescence intensity of a given sample. Electrochemical microfluidic assay for CTB in the stool samples The electrochemical microfluidic assay was conducted following the fluorescence microfluidic assay protocol. Slight differences included the use of 1 μL of 1:50 dilution of the ferricyanide/ferrocyanide-encapsulating GM1-containing liposome stock solution and 1 μL of potassium phosphate buffer (pH 7.0) with sucrose to control the osmotic pressure and a detergent flow rate of 0.1 μL min-1 in order to lyse the liposomes and release the electroactive ferricyanide/ ferrocyanide compounds into the microchannel. Released ferricyanide/ferrocyanide was detected on the IDUA positioned downstream of the capture zone, and signals were recorded by using a potentiostat (Metrohm Autolab, Utrecht, The Netherlands), which was connected to the contact pads of the IDUA in the microfluidic device via copper wires. Data analysis All of the signals in terms of the fluorescence signal (relative fluorescence units), color intensity (pixels), and current peak area (μA s) derived from the immunomagnetic separation (IMS), the optical microfluidic, and the electrochemical microfluidic assay, respectively, were correlated to CTB concentrations using a four-parameter logistic (Eq. 1) using XLfit (IDBS, Guildford, UK): y¼aþ
b−a ; 1 þ 10ðc − xÞd
ð1Þ
where a is the response at the maximum concentration, b is the response at zero concentration, x is the target concentration, c
Microfluidic biosensor for cholera toxin detection in fecal
is the concentration yielding 50 % response, and d is a slope factor [22]. The limits of detection were defined as the concentration equivalent to the background signal plus three times the standard deviation [23].
Results and discussion The analytical performance of the microfluidic devices using fluorescence and electrochemical detection via liposome amplification for CTB detection in real (spiked) stool samples was studied. Diagnostic devices will require minimal yet effective sample preparation steps, and effort here was put toward the use of truly minimal steps that focus only on the removal of particulates. First, a simple sample preparation procedure consisting of a quick centrifugation step was optimized in a microtiter plate format. Since small particles present in stool samples can cause microchannel clogging, their removal was required. Second, the effect of the sample matrix on the assay’s sensitivity was evaluated in the microtiter plate format prior to transferring it into the microfluidic format. Finally, the cross-reactivity by another toxin was studied also in the microfluidic format.
Stool sample preparation study To investigate the possibility of applying the microfluidic immunosensors for real sample application, the detection of the toxin in stool matrices was performed by using the simulated (spiked) stool as the sample. The stool sample preparation by centrifugation was optimized using the IMS microtiter plate assay. The parameters studied were the time of spiking and the centrifugal force and time. From the study, no significant difference in the signal detected by the IMS microtiter plate assay was obtained when compared with spiked samples stored for up to 9 h at 4 °C after spiking. Also, the optimal centrifugation condition was determined to be 8,000 rcf for 20 min. This resulted in the removal of all visible debris from the sample. A dose–response curve was generated for CTB in this assay (Fig. 3). The limit of detection was 45.6 ng mL-1 with a dynamic range up to 2,000 ng mL-1. At the maximum signal, a signal-to-noise ratio of 21 was obtained. The assay, therefore, performed very well, with minimal sample preparation needed; however, the limit of detection was somewhat higher than that determined previously for CTB in a buffer solution (8 ng mL-1) [17]. This is to be expected as it has been shown previously by other researchers that matrix components in stool inhibit the antibody–analyte binding reaction [24, 25]. Finally, to ensure that no particulates would clog the
Fig. 3 Dose–response curve for CTB detection by the microtiter plate assay. Stool samples were spiked with CTB. The samples were centrifuged prior to analysis to remove debris present. Triplicate analyses were performed and standard deviations were plotted. RFU relative fluorescence units
N. Bunyakul et al.
microfluidic channels, an additional filtration step was included. Here, the samples were injected into the device used in the microtiter plate assay by passing them through a syringe filter (Whatman® GD/X, 0.45-μm pore-size). The additional filtration step did not affect the microtiter plate assay (data not shown). We therefore postulate that CTB does not attach to small particulate matter present in the sample after centrifugation, which makes the additional filtration step ideal for incorporation in the microfluidic assay systems. Dose–response curves of the fluorescence and electrochemical microfluidic GM1-containing liposome immunosensors for CTB detection in spiked stool samples The optimum conditions for the sample preparation procedure and the fluorescence microfluidic GM1-containing liposome immunosensor developed in our previous work [17] were used to generate a dose–response curve for CTB in stool samples, as shown in Fig. 4. The limit of detection of 9.0 ng mL-1was calculated from a linear graph plotted from the low range of CTB spiking concentrations (0–100 ng mL1 ), and was defined as the concentration equivalent to the background signal plus three times the standard deviation. The optimum procedure of the electrochemical microfluidic assay for CTB as described in our previous report
[17] and the optimum conditions of the sample preparation procedure were used to generate a CTB dose–response curve, as shown in Fig. 5. The limit of detection of 31.7 ng mL-1 was calculated from a linear graph plotted from the low range of CTB spiking concentrations (0–100 ng mL-1), and was defined as the concentration equivalent to the background signal plus three times the standard deviation. As demonstrated by the results shown in Figs. 4 and 5, both formats of the microfluidic assay can detect the toxin within a total assay time of 1.5 h, including a simple sample preparation step. The limits of detection of the microfluidic immunosensors in stool samples were 9.0 and 31.7 ng mL-1 for the optical and electrochemical formats, respectively. The sensitivity of the electrochemical format is about two times higher than that of the optical format when assaying the stool matrices. This is in contrast to earlier findings where the electrochemical format showed a lower limit of detection for CTB in pure buffer solutions [17], i.e., 1.0 ng mL-1 versus 6.6 ng mL-1 for the fluorescence approach. It is therefore assumed that hydrophobic contents of the stool sample, such as lipids, may negatively influence the IDUA electrodes and therefore result in lower reactivity [26]. Therefore, more refined sample preparation procedures will be required or coating of the electrode materials will be necessary. Alternatively, assay designs can be devised where only the lysed liposome
Fig. 4 Dose–response curve of the microfluidic sulforhodamine B-encapsulating GM1-containing liposome immunosensor for CTB detection in spiked stool samples. Triplicate analyses were performed and standard deviations are plotted
Microfluidic biosensor for cholera toxin detection in fecal
Fig. 5 Dose–response curve of the electrochemical microfluidic immunosensor for CTB detection in spiked stool samples. Triplicate analyses were performed and standard deviations are plotted
solution will pass over the IDUA surface and all earlier solutions will be transported through an alternative channel (Fig. 6). Cross-reactivity study with heat-labile E. coli toxin using the electrochemical microfluidic sensor Extensive studies have been done investigating the specificity of the GM1 receptor and the anti-CTB antibodies for crossreactivity to other toxins. The receptor itself can bind to a wide Fig. 6 Microchannel layout (top view) with an additional outlet channel for the sample matrix solution. IDUA interdigitated ultramicroelectrode array
range of toxins, including cholera toxin [27] and heat-labile E. coli toxin [28]. The anti-CTB antibody, however, provides excellent specificity toward CTB, as shown by Lian et al. [29]. Ahn-Yoon et al. [30] have shown that the combination of the GM1 receptor and anti-CTB antibody biorecognition elements provides a high-specificity assay for CTB [30]. To demonstrate that this inherent specificity can also be translated into the microfluidic assay, the heat-labile E. coli toxin was chosen for detection with the electrochemical format. Heat-labile E. coli toxin is similar in structure and function to cholera
N. Bunyakul et al. Fig. 7 Dose–response signals from purified samples of CTB and heat-labile Escherichia coli toxin (LT) detected by the electrochemical microfluidic immunosensor. Triplicate analyses were performed and standard deviations are plotted
toxin. Both toxins have two subunits, A and B, which form an AB5 complex. Both toxins also can bind to GM1 ganglioside receptors and cause watery diarrhea via activation of adenylate cyclase by ADP-ribosylation [31]. Cross-reactivity signals were investigated by using the electrochemical microfluidic immunosensor detecting various concentrations (0–1,000 ng mL-1) of heat-labile E. coli toxin compared with CTB in purified samples (Fig. 7). The signals obtained for heat-labile E. coli toxin were negligible, providing signals only at a level of less than 0.4 μA s, which is at around the limit of detection for signals obtained for CTB. Thus, the highly specific combination of anti-CTB and GM1-receptor antibodies also provides specific detection in the microfluidic setup.
Conclusion The study described here focused on the application of microfluidic immunosensors for CTB detection in stool samples. Since rapid and on-field screening diagnosis of cholera would make possible improved health care in outbreak situations, it is an active area of research that deserves attention, especially to the sample preparation details. Several research groups have demonstrated interesting approaches for purified cholera toxin detection in microfluidic devices, such as using an immunoassay [32], electrochemical impedance spectroscopy [13], and a flow cytometer [33]. With respect to stool matrices, microfabricated PCR assays for rotavirus [34] and Salmonella [35] have demonstrated general feasibility. In these cases, sample preparation consisted of centrifugation followed by filtration [34] and
magnetic capture techniques [35]. However, studies in buffer samples are not sufficient anymore for microfluidic analytical systems, and matrix effects of real-world samples are necessary. At the same time, simple analytical steps are important, so that the overall assay remains field-portable. In our studies, we determined that simple debris removal from stool samples was sufficient for fluorescence detection approaches; however, it was not appropriate for an electrochemical system. The electron transfer was affected by remaining matrix effects in comparison with pure buffer solutions. Thus, in the future, detailed studies of electrode surfaces may assist in the design of required sample preparation steps in order to remove those matrix components that affect the electron transfer. Using orthogonal detection approaches, such as the fluorescent and electrochemical systems used here, will assist in the identification of matrix effects of the transduction system itself. Similarly, the specificity studies and the orthogonal approaches demonstrated that the matrix effects here did not influence the biological recognition. Similar studies will be conducted in the future to investigate optimal sample preparation steps in the detection of CTB in stool samples.
Acknowledgments The authors thank Wijit Wonglumsom of Mahidol University for help with the bacterial biochemical tests. We thank Sutas Boonyong of Nakhon Pathom Hospital, Thailand, for help in obtaining the stool samples. We also thank John Connelly for IDUA electrode fabrication. This work was performed in part at the Cornell N a n o f a b r i c a t i o n F a c i l i t y, a m e m b e r o f t h e N a t i o n a l Nanofabrication Users Network, which is supported by the National Science Foundation (grant ECS-0335765). This research was also supported in part by the Thailand Research Fund through the RGJ-PhD program in Thailand and the Commission on Higher Education, Ministry of Education, Thailand.
Microfluidic biosensor for cholera toxin detection in fecal
References 1. Terry SC, Jerman JH, Angell JB (1979) A gas chromatographic air analyzer fabricated on a silicon wafer. IEEE Transl Electron Devices 26:1880–1886 2. Mir M, Homs A, Samitier J (2009) Integrated electrochemical DNA biosensors for lab-on-a-chip devices. Electrophoresis 30:3386–3397 3. Choi S, Goryll M, Sin LYM, Wong PK, Chae J (2011) Microfluidicbased biosensors toward point-of-care detection of nucleic acids and proteins. Microfluid Nanofluid 10:231–247 4. Darain F, Yager P, Gan KL, Tjin SC (2009) On-chip detection of myoglobin based on fluorescence. Biosens Bioelectron 24:1744– 1750 5. Heinze BC, Song JY, Lee CH, Najam A, Yoon JY (2009) Microfluidic immunosensor for rapid and sensitive detection of bovine viral diarrhea virus. Sensors Actuators B 138:491–496 6. Heinze BC, Gamboa JR, Kim K, Song JY, Yoon JY (2010) Microfluidic immunosensor with integrated liquid core waveguides for sensitive Mie scattering detection of avian influenza antigens in a real biological matrix. Anal Bioanal Chem 398:2693–2700 7. Pereira SV, Bertolino FA, Fernández-Baldo MA, Messina GA, Salinas E, Sanz MI, Raba J (2011) A microfluidic device based on a screen-printed carbon electrode with electrodeposited gold nanoparticles for the detection of IgG anti-Trypanosoma cruzi antibodies. Analyst 136:4745–4751 8. Zhang H, Fu X, Liu L, Zhu Z, Yang K (2012) Microfluidic beadbased enzymatic primer extension for single-nucleotide discrimination using quantum dots as labels. Anal Biochem 426:30–39 9. Liu Y, Wang H, Huang J, Yang J, Liu B, Yang P (2009) Microchipbased ELISA strategy for the detection of low-level disease biomarker in serum. Anal Chim Acta 650:77–82 10. Kellner C, Botero ML, Latta D, Drese K, Fragoso A, O'Sullivan CK (2011) Automated microsystem for electrochemical detection of cancer markers. Electrophoresis 32:926–930 11. Chikkaveeraiah BV, Mani V, Patel V, Gutkind JS, Rusling JF (2011) Microfluidic electrochemical immunoarray for ultrasensitive detection of two cancer biomarker proteins in serum. Biosens Bioelectron 26:4477–4483 12. Arévalo FJ, Granero AM, Fernández H, Raba J, Zón MA (2011) Citrinin (CIT) determination in rice samples using a micro fluidic electrochemical immunosensor. Talanta 83:966–973 13. Chiriacò MS, Primiceri E, D’Amone E, Ionescu RE, Rinaldi R, Maruccio G (2011) EIS microfluidic chips for flow immunoassay and ultrasensitive cholera toxin detection. Lab Chip 11:658–663 14. Liang W, Li Y, Zhang B, Zhang Z, Chen A, Qi D, Yi W, Hu C (2012) A novel microfluidic immunoassay system based on electrochemical immunosensors: an application for the detection of NT-proBNP in whole blood. Biosens Bioelectron 31:480–485 15. WHO (2013) Cholera. http://www.who.int/topics/cholera/en/. Accessed 25 Mar 2013 16. McDowall J (2005) Cholera toxin. http://www.ebi.ac.uk/interpro/ potm/2005_9/Page1.htm. Accessed 7 Jan 2008 17. Bunyakul N, Edwards KA, Promptmas C, Baeumner AJ (2009) Cholera toxin subunit B detection in microfluidic devices. Anal Bioanal Chem 393:177–186 18. Goral VN, Zaytseva NV, Baeumner AJ (2006) Electrochemical microfluidic biosensor for the detection of nucleic acid sequences. Lab Chip 6:414–421 19. Zaytseva NV, Goral VN, Montagna RA, Baeumner AJ (2005) Development of a microfluidic biosensor module for pathogen detection. Lab Chip 5:805–811 20. Kwakye S, Baeumner AJ (2003) A microfluidic biosensor based on nucleic acid sequence recognition. Anal Bioanal Chem 376:1062– 1068
21. Bartlett GR (1959) Phosphorus assay in column chromatography. J Biol Chem 234:466–468 22. Gottschalk PG, Dunn JR (2005) The five-parameter logistic: a characterization and comparison with the four-parameter logistic. Anal Biochem 343:54–65 23. Thompson V, Schatzlein D, Mercuro D (2003) Limits of detection in spectroscopy. Spectroscopy 18(12):112–114 24. Rivero-Marcotegui A, Olivera-Olmedo JE, Valverde-Visus FS, Palacios-Sarrasqueta M, Grijalba-Uche A, García-Merlo S (1998) Water, fat, nitrogen, and sugar content in feces: reference intervals in children. Clin Chem 44(7):1540–1544 25. Selby C (1999) Interference in immunoassay. Ann Clin Biochem 36: 704–721 26. Zhaoa H, Bian X, Galligan JJ, Swain GM (2010) Electrochemical measurements of serotonin (5-HT) release from the guinea pig mucosa using continuous amperometry with a boron-doped diamond microelectrode. Diam Relat Mater 19:182–185 27. Holmgren J, Lonnroth I, Mansson J, Svennerholm L (1975) Interaction of cholera toxin and membrane GM1 ganglioside of small intestine. Proc Natl Acad Sci U S A 72:2520–2524 28. Chen J, Chang Y, Wu S, Chao D, Chang C, Li C, Ho T, Hsiang C (2007) Inhibition of Escherichia coli heat-labile enterotoxin-induced diarrhea by Chaenomeles speciosa. J Ethnopharmacol 113:233–239 29. Lian W, Wu D, Lim DV, Jin S (2010) Sensitive detection of multiplex toxins using antibody microarray. Anal Biochem 401:271–279 30. Ahn-Yoon S, DeCory TR, Baeumner AJ, Durst RA (2003) Ganglioside-liposome immunoassay for the ultrasensitive detection of cholera toxin. Anal Chem 75:2256–2261 31. Spangler BD (1992) Structure and function of cholera toxin and the related Escherichia coli heat-labile enterotoxin. Microbiol Rev 56: 622–647 32. Phillips KS, Cheng Q (2005) Microfluidic immunoassay for bacterial toxins with supported phospholipid bilayer membranes on poly(dimethylsiloxane). Anal Chem 77:327–334 33. Kim JS, Anderson GP, Erickson JS, Golden JP, Nasir M, Ligler FS (2009) Multiplexed detection of bacteria and toxins using a microflow cytometer. Anal Chem 81:5426–5432 34. Li Y, Zhang C, Xing D (2011) Integrated microfluidic reverse transcription-polymerase chain reaction for rapid detection of food- or waterborne pathogenic rotavirus. Anal Biochem 415: 87–96 35. Verdoy D, Barrenetxea Z, Berganzo J, Agirregabiria M, Ruano-López JM, Marimón JM, Olabarría G (2012) A novel real time micro PCR based point-of-care device for Salmonella detection in human clinical samples. Biosens Bioelectron 32:259–265
Natinan Bunyakul received his Ph.D. degree in medical technology from Mahidol University (Thailand) in 2010. He is now a lecturer and researcher in the Faculty of Medical Technology, Mahidol University. His research focuses on the development of biosensors, especially lab-on-a-chip devices for environmental and clinical applications.
N. Bunyakul et al. Chamras Promptmas is Assistant Professor at Mahidol University. He received his diploma in biotechnology from the University of Kent at Canterbury, UK, in 1988 and his Ph.D. degree in biochemistry from Mahidol University in 1994. He has been working at Mahidol University for several years on the development of biosensors for laboratory diagnosis.
Antje Baeumner is Professor and Director of the Institute of Analytical Chemistry, Chemo and Biosensors at the University of Regensburg, Germany. In 2013 she moved there from her previous scientific home, the Department of Biological and Environmental Engineering at Cornell University, where the research described also was performed. Her research focuses on the development of miniaturized biosensors and micro total analysis systems for the detection of pathogens and toxins for food, environmental, and clinical applications.
PAPER IN FOREFRONT
Microfluidic biosensor for cholera toxin detection in fecal samples Natinan Bunyakul & Chamras Promptmas & Antje J. Baeumner
Received: 30 March 2014 / Revised: 28 May 2014 / Accepted: 4 June 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract Sample preparation and processing steps are the most critical assay aspects that require our attention in the development of diagnostic devices for analytes present in complex matrices. In the best scenarios, diagnostic devices should use only simple sample processing. We have therefore investigated minimal preparation of stool samples and their effect on our sensitive microfluidic immunosensor for the detection of cholera toxin. This biosensor was previously developed and tested in buffer solutions only, using either fluorescence or electrochemical detection strategies. The microfluidic devices were made from polydimethylsiloxane using soft lithography and silicon templates. Cholera toxin subunit B (CTB)-specific antibodies immobilized onto superparamagnetic beads and ganglioside GM1-containing liposomes were used for CTB recognition in the detection system. Quantification of CTB was tested by spiking it in human stool samples. Here, optimal minimal sample processing steps, including filtration and centrifugation, were optimized using a microtiter plate assay owing to its highthroughput capabilities. Subsequently, it was transferred to the microfluidic systems, enhancing the diagnostic Published in the topical collection celebrating ABCs 13th Anniversary. N. Bunyakul : C. Promptmas Department of Clinical Chemistry, Faculty of Medical Technology, Mahidol University, Nakhon Pathom 73170, Thailand A. J. Baeumner (*) Department of Biological and Environmental Engineering, Cornell University, Ithaca, NY 14853, USA e-mail: [email protected] A. J. Baeumner e-mail: [email protected] Present Address: A. J. Baeumner Institute of Analytical Chemistry, Chemo- and Biosensors, University of Regensburg, 93053 Regensburg, Germany
characteristic of the biosensor. It was found that the debris removal obtained through simple centrifugation resulted in an acceptable removal of matrix effects for the fluorescence format, reaching a limit of detection of only 9.0 ng/mL. However, the electron transfer in the electrochemical format was slightly negatively affected (limit of detection of 31.7 ng/ mL). Subsequently, cross-reactivity using the heat-labile Escherichia coli toxin was investigated using the electrochemical microfluidic immunosensors and was determined to be negligible. With minimal sample preparation required, these microfluidic liposome-based systems have demonstrated excellent analytical performance in a complex matrix and will thus be applicable to other sample matrices. Keywords Biosensor . Microfluidic . Liposomes . Cholera toxin . Fecal sample
Introduction From the introduction of the first microfluidic device developed in the 1970s [1], microfluidic device technology has entered various fields, including chemical, biological, and diagnostic analyses, owing to its advantages in terms of the small amounts of reagent and sample required, high sensitivity, rapid response, portability, and its simple adaptation to multifunctionality and high-throughput analyses [2]. Diagnostic assay development is moving toward the use of micro total analysis systems in which detection and sample preparation are integrated. In the case of microfluidic immunosensors, antibodies are used as biorecognition elements, similarly to traditional immune biosensors, which find ample application in point-of-care diagnostics [3]. Both optical and electrochemical detection approaches have been realized with microfluidic immunosensors. Their application has been successfully demonstrated for a variety of analytes. For
N. Bunyakul et al.
example, optical microfluidic immunosensors have been demonstrated to be successful in the detection of cardiac biomarkers [4], viral particles [5, 6], pathogen-specific immunoglobulins [7], and cancer biomarkers [8]. Electrochemical microfluidic immunosensors also have been developed for many analytical applications, such as for cancer biomarkers [9–11], mycotoxins [12], bacterial toxins [13], and cardiac biomarkers [14]. Cholera is an acute diarrheal infection caused by ingestion of food or water contaminated with the bacterium Vibrio cholerae. The patient can die within an hour if left untreated. Every year, there are an estimated three million to five million cholera cases of cholera and 100,000–120 000 deaths from cholera. The short incubation period of 2 h to 5 days compounds the potentially explosive pattern of outbreaks. Cholera is now endemic in many countries [15]. Cholera toxin produced from V. cholerae is the actual causative agent of the disease. Cholera toxin is composed of two subunits, cholera toxin subunit A and cholera toxin subunit B (CTB). The B subunit’s function is to bind specifically to the ganglioside GM1 receptor on intestinal cell walls, whereas the A subunit is the active function protein that activates the production of cyclic AMP, causing a dramatic efflux of ions and water from infected enterocytes, leading to watery diarrhea [16]. In a recent study [17], we developed a microfluidic immunosensor with GM1-containing liposome for CTB detection in purified samples. Two assay formats were developed, i.e., fluorescence and electrochemical. Good sensitivity was found for both assay formats, and electrochemical signal recording was significantly simpler than the optical approach. In the present study, simple sample preparation procedures, real sample matrices, and their effect on both formats were investigated. The microfluidic assay with sample preparation procedures and the biological recognition principles are shown schematically in Fig. 1.
Tygon® tubing was supplied by Small Parts (Miami Lakes, FL, USA). PEEK™ tubing was obtained from Upchurch Scientific. (Oak Harbor, WA, USA). Syringes and syringe filters were purchased from Becton, Dickinson (Franklin Lakes, NJ, USA). Rare earth neodymium–iron–boron disc magnets were obtained from National Imports (Vienna, VA, USA). The silicone elastomer kit, Sylgard®184 containing polydimethylsiloxane (PDMS) prepolymer and catalyst, was obtained from Dow Corning (Midland, MI, USA). The Cornell Nanofabrication Facility provided clean-room facilities, chemicals, and equipment for silicon template fabrication. The Plexiglas® housing was constructed in the machine shop located in the School of Chemical and Biomolecular Engineering, Cornell University. Stool sample collection and simulated (spiked) stool sample preparation Watery stools from patients with diarrhea admitted to Nakhon Pathom Hospital were collected, pooled (total approximately 300 mL), and sent to the laboratory to serve as stool sample matrices for the CTB spiking study. Prior to shipment to the laboratory, the pooled samples were stripped of all patient information. The pooled stool sample was stored at -20 °C in order to minimize the growth of bacteria. Prior to the spiking with CTB, a standard bacterial culture was performed, verifying that no V. cholerae bacteria were present in the pooled sample received. The simulated (spiked) stool samples were then prepared by adding known concentrations of purified CTB to the original stool samples. The samples were mixed well in a microtube. The same procedure was followed for each CTB concentration studied. The spiked stool samples were kept at 4 ° C prior to analysis and were used within 9 h after spiking. Interdigitated ultramicroelectrode array fabrication
Experimental Materials and reagents General laboratory chemicals, including n-octyl-β- D glucopyranoside (OG) and buffer reagents, were purchased from VWR Scientific Products, (New York, NY, USA). Sulforhodamine B (SRB) was purchased from Molecular Probes (Eugene, OR, USA). GM1 ganglioside and the nontoxic form of cholera toxin, i.e., its subunit B (CTB), were supplied by Calbiochem (San Diego, CA, USA). Heat-labile toxins from Escherichia coli were obtained from Quadratech Diagnostics. The biotinylated anti-CTB was purchased from United States Biological (Swampscott, MA, USA). Superparamagnetic beads (Dynabeads MyOne streptavidin) were obtained from Dynal Biotech (Lake Success, NY, USA).
The interdigitated ultramicroelectrode arrays (IDUAs) were fabricated by standard clean-room photolithographic and liftoff techniques on glass wafers, as described previously [18]. Briefly, with use of GCA Autostep 200 DSW i-line wafer stepper, the pattern was transferred to Pyrex glass wafers (Corning 7740; Corning, Corning, NY, USA) coated with positive photoresist (SPR955-CM-0.9, Shipley, Marlborough, MA, USA). Then, image reversal was performed using NH3 gas in a YES-58SM image reversal oven. The wafers were flushed using an HTG system III-HR contact aligner. Subsequently, 70 Å of titanium and 500 Å of gold were evaporated using an e-gun source (CVC 4500 evaporator). The wafers were then soaked in Microposit remover 1165 (Shipley, Marlborough, MA, USA) to lift off the excess metal and photoresist, leaving the patterned electrode arrays. The
Microfluidic biosensor for cholera toxin detection in fecal
Fig. 1 The microfluidic assay for cholera toxin subunit B (CTB) with sample preparation procedures (from a CTB-spiked sample) and the biological recognition principles
resulting IDUAs had 420 pairs of fingers with widths of 2.5 μm, lengths of 1,000 μm, and gap sizes of 4.5 μm. Microchannel fabrication and construction of the device housing Figure 2 shows the microfluidic devices for both fluorescence and electrochemical formats. The microchannels with 50-μm depths were produced by standard photolithography and dry etching techniques as described previously [17, 19, 20]. With use of soft lithography, the microchannels were realized in PDMS covered with a glass slide and a glass slide bearing the IDUA for the fluorescence and electrochemical formats, respectively. The PDMS microchannels and the glass slide were placed between two Plexiglas® plates with the glass slide below and the microchannels above. The top Plexiglas® plate had PEEK™ tubing glued into locations lined up with the inlet and outlet holes of the microchannel. The top plate also had another hole to accommodate the permanent magnet required to capture magnetic beads in the detection zone. The two Plexiglas® plates were held in place by four or eight screws to provide sufficient pressure for leakproof sealing between the PDMS and glass slide layers. Preparation of the GM1-containing liposomes for the fluorescence and electrochemical assay formats Reverse-phase evaporation was used during liposome preparation for both the fluorescence and the electrochemical
format, as described previously [17]. SRB and ferricyanide/ ferrocyanide were the encapsulated markers for the fluorescence and electrochemical assay formats, respectively. The average sizes of both types of liposomes were determined by a DynaPro™ dynamic light scattering machine (Wyatt Technology, Santa Barbara, CA, USA). The phospholipid concentration of SRB-encapsulating GM1-containing liposomes was determined using Bartlett’s assay [21], wherein one mole of phosphorus represents one mole of phospholipid. The phospholipid concentration of ferricyanide/ferrocyanideencapsulating GM1-containing liposomes cannot be determined from Bartlett’s assay because phosphorus from a potassium phosphate buffer, which is used for preparing the liposomes, will interfere with the phosphorus determination.
Immunomagnetic separation microtiter plate assay for CTB in the stool samples Ten micrograms of the anti-CTB antibody conjugated beads, prepared as described previously [17], was added to each well of the 0.01 % (w/v) bovine serum albumin blocked microtiter plates. Subsequently, CTB-spiked stool samples were centrifuged at 8,000 rcf for 20 min, and 100 μL of the supernatant solution was added to each well using pipette mixing. The plate was then incubated in the dark at room temperature with 30 min of shaking. After magnetic separation on a magnetic separator plate (3 min), the supernatant was removed and tris(hydroxymethyl)aminomethane-buffered saline (200 μL per well) was added as a washing step. This washing was repeated twice [1× HEPES saline sucrose buffer (HSS) instead
N. Bunyakul et al.
Fig. 2 The microfluidic devices for cholera toxin subunit B detection: a fluorescence detection format; b electrochemical detection format
of tris(hydroxymethyl)aminomethane-buffered saline was used din the final step]. Then, 100 μL of SRB-encapsulating GM1-containing liposomes with a phospholipid concentration of 0.01 mM was added to each well and the resulting mixtures were incubated in the dark with 30 min of shaking. The plate was then washed three times as described before using 1× HSS buffer. The supernatant was aspirated and 100 μL of 30 mM OG was added to each well for 10 min at room temperature with shaking to lyse the bounded liposomes in the plate. Finally, 50 μL of the supernatant solution was transferred into an empty unblocked well of the black microtiter plate for measurement of the fluorescence signal (545 nm for λex and 590 nm for λem) using a microplate fluorescence reader (Wallac Victor 1420, PerkinElmer).
Fluorescence microfluidic assay for CTB in the stool samples CTB-spiked stool samples were centrifuged at 8,000 rcf for 20 min, and 1 μL of the supernatant solution was mixed with 1 μL of anti-CTB antibody conjugated magnetic bead solution at 1 μg μL-1, 1 μL of SRB-encapsulating GM1-containing liposome solution at a phospholipid concentration of 0.075 mM, and 1 μL of 1× HSS buffer (pH 7.0) in a microcentrifuge tube. The sample mixture was then incubated at room temperature with shaking for 30 min. Subsequently, the sample mixture was applied into the sample injection inlet of the optical format microfluidic chip by using a syringe pump (New Era Pump Systems, Farmingdale, NY, USA) at a flow rate of 5 μL min-1, followed immediately by 26 μL of 1× HSS buffer at the same flow rate. During these two steps, the liposome–CTB– bead complexes formed were immobilized at the magnet area and nonbound liposomes were washed away. Then, 30 mM OG solution was applied through the detergent injection inlet at 0.8 μL min-1 [19] in order to lyse the liposomes and release the SRB molecules into the microchannel. Fluorescence of lysed liposomes in the detection zone was visualized using an IX70 inverted microscope (Olympus, USA). The fluorescence images of the detection zone were obtained with a CCD camera
(Olympus, USA) coupled with Image-Pro Plus (Media Cybernetics, Rockville, MD, USA) for color intensity measurement. The average pixel value of the detection zone was considered to be the fluorescence intensity of a given sample. Electrochemical microfluidic assay for CTB in the stool samples The electrochemical microfluidic assay was conducted following the fluorescence microfluidic assay protocol. Slight differences included the use of 1 μL of 1:50 dilution of the ferricyanide/ferrocyanide-encapsulating GM1-containing liposome stock solution and 1 μL of potassium phosphate buffer (pH 7.0) with sucrose to control the osmotic pressure and a detergent flow rate of 0.1 μL min-1 in order to lyse the liposomes and release the electroactive ferricyanide/ ferrocyanide compounds into the microchannel. Released ferricyanide/ferrocyanide was detected on the IDUA positioned downstream of the capture zone, and signals were recorded by using a potentiostat (Metrohm Autolab, Utrecht, The Netherlands), which was connected to the contact pads of the IDUA in the microfluidic device via copper wires. Data analysis All of the signals in terms of the fluorescence signal (relative fluorescence units), color intensity (pixels), and current peak area (μA s) derived from the immunomagnetic separation (IMS), the optical microfluidic, and the electrochemical microfluidic assay, respectively, were correlated to CTB concentrations using a four-parameter logistic (Eq. 1) using XLfit (IDBS, Guildford, UK): y¼aþ
b−a ; 1 þ 10ðc − xÞd
ð1Þ
where a is the response at the maximum concentration, b is the response at zero concentration, x is the target concentration, c
Microfluidic biosensor for cholera toxin detection in fecal
is the concentration yielding 50 % response, and d is a slope factor [22]. The limits of detection were defined as the concentration equivalent to the background signal plus three times the standard deviation [23].
Results and discussion The analytical performance of the microfluidic devices using fluorescence and electrochemical detection via liposome amplification for CTB detection in real (spiked) stool samples was studied. Diagnostic devices will require minimal yet effective sample preparation steps, and effort here was put toward the use of truly minimal steps that focus only on the removal of particulates. First, a simple sample preparation procedure consisting of a quick centrifugation step was optimized in a microtiter plate format. Since small particles present in stool samples can cause microchannel clogging, their removal was required. Second, the effect of the sample matrix on the assay’s sensitivity was evaluated in the microtiter plate format prior to transferring it into the microfluidic format. Finally, the cross-reactivity by another toxin was studied also in the microfluidic format.
Stool sample preparation study To investigate the possibility of applying the microfluidic immunosensors for real sample application, the detection of the toxin in stool matrices was performed by using the simulated (spiked) stool as the sample. The stool sample preparation by centrifugation was optimized using the IMS microtiter plate assay. The parameters studied were the time of spiking and the centrifugal force and time. From the study, no significant difference in the signal detected by the IMS microtiter plate assay was obtained when compared with spiked samples stored for up to 9 h at 4 °C after spiking. Also, the optimal centrifugation condition was determined to be 8,000 rcf for 20 min. This resulted in the removal of all visible debris from the sample. A dose–response curve was generated for CTB in this assay (Fig. 3). The limit of detection was 45.6 ng mL-1 with a dynamic range up to 2,000 ng mL-1. At the maximum signal, a signal-to-noise ratio of 21 was obtained. The assay, therefore, performed very well, with minimal sample preparation needed; however, the limit of detection was somewhat higher than that determined previously for CTB in a buffer solution (8 ng mL-1) [17]. This is to be expected as it has been shown previously by other researchers that matrix components in stool inhibit the antibody–analyte binding reaction [24, 25]. Finally, to ensure that no particulates would clog the
Fig. 3 Dose–response curve for CTB detection by the microtiter plate assay. Stool samples were spiked with CTB. The samples were centrifuged prior to analysis to remove debris present. Triplicate analyses were performed and standard deviations were plotted. RFU relative fluorescence units
N. Bunyakul et al.
microfluidic channels, an additional filtration step was included. Here, the samples were injected into the device used in the microtiter plate assay by passing them through a syringe filter (Whatman® GD/X, 0.45-μm pore-size). The additional filtration step did not affect the microtiter plate assay (data not shown). We therefore postulate that CTB does not attach to small particulate matter present in the sample after centrifugation, which makes the additional filtration step ideal for incorporation in the microfluidic assay systems. Dose–response curves of the fluorescence and electrochemical microfluidic GM1-containing liposome immunosensors for CTB detection in spiked stool samples The optimum conditions for the sample preparation procedure and the fluorescence microfluidic GM1-containing liposome immunosensor developed in our previous work [17] were used to generate a dose–response curve for CTB in stool samples, as shown in Fig. 4. The limit of detection of 9.0 ng mL-1was calculated from a linear graph plotted from the low range of CTB spiking concentrations (0–100 ng mL1 ), and was defined as the concentration equivalent to the background signal plus three times the standard deviation. The optimum procedure of the electrochemical microfluidic assay for CTB as described in our previous report
[17] and the optimum conditions of the sample preparation procedure were used to generate a CTB dose–response curve, as shown in Fig. 5. The limit of detection of 31.7 ng mL-1 was calculated from a linear graph plotted from the low range of CTB spiking concentrations (0–100 ng mL-1), and was defined as the concentration equivalent to the background signal plus three times the standard deviation. As demonstrated by the results shown in Figs. 4 and 5, both formats of the microfluidic assay can detect the toxin within a total assay time of 1.5 h, including a simple sample preparation step. The limits of detection of the microfluidic immunosensors in stool samples were 9.0 and 31.7 ng mL-1 for the optical and electrochemical formats, respectively. The sensitivity of the electrochemical format is about two times higher than that of the optical format when assaying the stool matrices. This is in contrast to earlier findings where the electrochemical format showed a lower limit of detection for CTB in pure buffer solutions [17], i.e., 1.0 ng mL-1 versus 6.6 ng mL-1 for the fluorescence approach. It is therefore assumed that hydrophobic contents of the stool sample, such as lipids, may negatively influence the IDUA electrodes and therefore result in lower reactivity [26]. Therefore, more refined sample preparation procedures will be required or coating of the electrode materials will be necessary. Alternatively, assay designs can be devised where only the lysed liposome
Fig. 4 Dose–response curve of the microfluidic sulforhodamine B-encapsulating GM1-containing liposome immunosensor for CTB detection in spiked stool samples. Triplicate analyses were performed and standard deviations are plotted
Microfluidic biosensor for cholera toxin detection in fecal
Fig. 5 Dose–response curve of the electrochemical microfluidic immunosensor for CTB detection in spiked stool samples. Triplicate analyses were performed and standard deviations are plotted
solution will pass over the IDUA surface and all earlier solutions will be transported through an alternative channel (Fig. 6). Cross-reactivity study with heat-labile E. coli toxin using the electrochemical microfluidic sensor Extensive studies have been done investigating the specificity of the GM1 receptor and the anti-CTB antibodies for crossreactivity to other toxins. The receptor itself can bind to a wide Fig. 6 Microchannel layout (top view) with an additional outlet channel for the sample matrix solution. IDUA interdigitated ultramicroelectrode array
range of toxins, including cholera toxin [27] and heat-labile E. coli toxin [28]. The anti-CTB antibody, however, provides excellent specificity toward CTB, as shown by Lian et al. [29]. Ahn-Yoon et al. [30] have shown that the combination of the GM1 receptor and anti-CTB antibody biorecognition elements provides a high-specificity assay for CTB [30]. To demonstrate that this inherent specificity can also be translated into the microfluidic assay, the heat-labile E. coli toxin was chosen for detection with the electrochemical format. Heat-labile E. coli toxin is similar in structure and function to cholera
N. Bunyakul et al. Fig. 7 Dose–response signals from purified samples of CTB and heat-labile Escherichia coli toxin (LT) detected by the electrochemical microfluidic immunosensor. Triplicate analyses were performed and standard deviations are plotted
toxin. Both toxins have two subunits, A and B, which form an AB5 complex. Both toxins also can bind to GM1 ganglioside receptors and cause watery diarrhea via activation of adenylate cyclase by ADP-ribosylation [31]. Cross-reactivity signals were investigated by using the electrochemical microfluidic immunosensor detecting various concentrations (0–1,000 ng mL-1) of heat-labile E. coli toxin compared with CTB in purified samples (Fig. 7). The signals obtained for heat-labile E. coli toxin were negligible, providing signals only at a level of less than 0.4 μA s, which is at around the limit of detection for signals obtained for CTB. Thus, the highly specific combination of anti-CTB and GM1-receptor antibodies also provides specific detection in the microfluidic setup.
Conclusion The study described here focused on the application of microfluidic immunosensors for CTB detection in stool samples. Since rapid and on-field screening diagnosis of cholera would make possible improved health care in outbreak situations, it is an active area of research that deserves attention, especially to the sample preparation details. Several research groups have demonstrated interesting approaches for purified cholera toxin detection in microfluidic devices, such as using an immunoassay [32], electrochemical impedance spectroscopy [13], and a flow cytometer [33]. With respect to stool matrices, microfabricated PCR assays for rotavirus [34] and Salmonella [35] have demonstrated general feasibility. In these cases, sample preparation consisted of centrifugation followed by filtration [34] and
magnetic capture techniques [35]. However, studies in buffer samples are not sufficient anymore for microfluidic analytical systems, and matrix effects of real-world samples are necessary. At the same time, simple analytical steps are important, so that the overall assay remains field-portable. In our studies, we determined that simple debris removal from stool samples was sufficient for fluorescence detection approaches; however, it was not appropriate for an electrochemical system. The electron transfer was affected by remaining matrix effects in comparison with pure buffer solutions. Thus, in the future, detailed studies of electrode surfaces may assist in the design of required sample preparation steps in order to remove those matrix components that affect the electron transfer. Using orthogonal detection approaches, such as the fluorescent and electrochemical systems used here, will assist in the identification of matrix effects of the transduction system itself. Similarly, the specificity studies and the orthogonal approaches demonstrated that the matrix effects here did not influence the biological recognition. Similar studies will be conducted in the future to investigate optimal sample preparation steps in the detection of CTB in stool samples.
Acknowledgments The authors thank Wijit Wonglumsom of Mahidol University for help with the bacterial biochemical tests. We thank Sutas Boonyong of Nakhon Pathom Hospital, Thailand, for help in obtaining the stool samples. We also thank John Connelly for IDUA electrode fabrication. This work was performed in part at the Cornell N a n o f a b r i c a t i o n F a c i l i t y, a m e m b e r o f t h e N a t i o n a l Nanofabrication Users Network, which is supported by the National Science Foundation (grant ECS-0335765). This research was also supported in part by the Thailand Research Fund through the RGJ-PhD program in Thailand and the Commission on Higher Education, Ministry of Education, Thailand.
Microfluidic biosensor for cholera toxin detection in fecal
References 1. Terry SC, Jerman JH, Angell JB (1979) A gas chromatographic air analyzer fabricated on a silicon wafer. IEEE Transl Electron Devices 26:1880–1886 2. Mir M, Homs A, Samitier J (2009) Integrated electrochemical DNA biosensors for lab-on-a-chip devices. Electrophoresis 30:3386–3397 3. Choi S, Goryll M, Sin LYM, Wong PK, Chae J (2011) Microfluidicbased biosensors toward point-of-care detection of nucleic acids and proteins. Microfluid Nanofluid 10:231–247 4. Darain F, Yager P, Gan KL, Tjin SC (2009) On-chip detection of myoglobin based on fluorescence. Biosens Bioelectron 24:1744– 1750 5. Heinze BC, Song JY, Lee CH, Najam A, Yoon JY (2009) Microfluidic immunosensor for rapid and sensitive detection of bovine viral diarrhea virus. Sensors Actuators B 138:491–496 6. Heinze BC, Gamboa JR, Kim K, Song JY, Yoon JY (2010) Microfluidic immunosensor with integrated liquid core waveguides for sensitive Mie scattering detection of avian influenza antigens in a real biological matrix. Anal Bioanal Chem 398:2693–2700 7. Pereira SV, Bertolino FA, Fernández-Baldo MA, Messina GA, Salinas E, Sanz MI, Raba J (2011) A microfluidic device based on a screen-printed carbon electrode with electrodeposited gold nanoparticles for the detection of IgG anti-Trypanosoma cruzi antibodies. Analyst 136:4745–4751 8. Zhang H, Fu X, Liu L, Zhu Z, Yang K (2012) Microfluidic beadbased enzymatic primer extension for single-nucleotide discrimination using quantum dots as labels. Anal Biochem 426:30–39 9. Liu Y, Wang H, Huang J, Yang J, Liu B, Yang P (2009) Microchipbased ELISA strategy for the detection of low-level disease biomarker in serum. Anal Chim Acta 650:77–82 10. Kellner C, Botero ML, Latta D, Drese K, Fragoso A, O'Sullivan CK (2011) Automated microsystem for electrochemical detection of cancer markers. Electrophoresis 32:926–930 11. Chikkaveeraiah BV, Mani V, Patel V, Gutkind JS, Rusling JF (2011) Microfluidic electrochemical immunoarray for ultrasensitive detection of two cancer biomarker proteins in serum. Biosens Bioelectron 26:4477–4483 12. Arévalo FJ, Granero AM, Fernández H, Raba J, Zón MA (2011) Citrinin (CIT) determination in rice samples using a micro fluidic electrochemical immunosensor. Talanta 83:966–973 13. Chiriacò MS, Primiceri E, D’Amone E, Ionescu RE, Rinaldi R, Maruccio G (2011) EIS microfluidic chips for flow immunoassay and ultrasensitive cholera toxin detection. Lab Chip 11:658–663 14. Liang W, Li Y, Zhang B, Zhang Z, Chen A, Qi D, Yi W, Hu C (2012) A novel microfluidic immunoassay system based on electrochemical immunosensors: an application for the detection of NT-proBNP in whole blood. Biosens Bioelectron 31:480–485 15. WHO (2013) Cholera. http://www.who.int/topics/cholera/en/. Accessed 25 Mar 2013 16. McDowall J (2005) Cholera toxin. http://www.ebi.ac.uk/interpro/ potm/2005_9/Page1.htm. Accessed 7 Jan 2008 17. Bunyakul N, Edwards KA, Promptmas C, Baeumner AJ (2009) Cholera toxin subunit B detection in microfluidic devices. Anal Bioanal Chem 393:177–186 18. Goral VN, Zaytseva NV, Baeumner AJ (2006) Electrochemical microfluidic biosensor for the detection of nucleic acid sequences. Lab Chip 6:414–421 19. Zaytseva NV, Goral VN, Montagna RA, Baeumner AJ (2005) Development of a microfluidic biosensor module for pathogen detection. Lab Chip 5:805–811 20. Kwakye S, Baeumner AJ (2003) A microfluidic biosensor based on nucleic acid sequence recognition. Anal Bioanal Chem 376:1062– 1068
21. Bartlett GR (1959) Phosphorus assay in column chromatography. J Biol Chem 234:466–468 22. Gottschalk PG, Dunn JR (2005) The five-parameter logistic: a characterization and comparison with the four-parameter logistic. Anal Biochem 343:54–65 23. Thompson V, Schatzlein D, Mercuro D (2003) Limits of detection in spectroscopy. Spectroscopy 18(12):112–114 24. Rivero-Marcotegui A, Olivera-Olmedo JE, Valverde-Visus FS, Palacios-Sarrasqueta M, Grijalba-Uche A, García-Merlo S (1998) Water, fat, nitrogen, and sugar content in feces: reference intervals in children. Clin Chem 44(7):1540–1544 25. Selby C (1999) Interference in immunoassay. Ann Clin Biochem 36: 704–721 26. Zhaoa H, Bian X, Galligan JJ, Swain GM (2010) Electrochemical measurements of serotonin (5-HT) release from the guinea pig mucosa using continuous amperometry with a boron-doped diamond microelectrode. Diam Relat Mater 19:182–185 27. Holmgren J, Lonnroth I, Mansson J, Svennerholm L (1975) Interaction of cholera toxin and membrane GM1 ganglioside of small intestine. Proc Natl Acad Sci U S A 72:2520–2524 28. Chen J, Chang Y, Wu S, Chao D, Chang C, Li C, Ho T, Hsiang C (2007) Inhibition of Escherichia coli heat-labile enterotoxin-induced diarrhea by Chaenomeles speciosa. J Ethnopharmacol 113:233–239 29. Lian W, Wu D, Lim DV, Jin S (2010) Sensitive detection of multiplex toxins using antibody microarray. Anal Biochem 401:271–279 30. Ahn-Yoon S, DeCory TR, Baeumner AJ, Durst RA (2003) Ganglioside-liposome immunoassay for the ultrasensitive detection of cholera toxin. Anal Chem 75:2256–2261 31. Spangler BD (1992) Structure and function of cholera toxin and the related Escherichia coli heat-labile enterotoxin. Microbiol Rev 56: 622–647 32. Phillips KS, Cheng Q (2005) Microfluidic immunoassay for bacterial toxins with supported phospholipid bilayer membranes on poly(dimethylsiloxane). Anal Chem 77:327–334 33. Kim JS, Anderson GP, Erickson JS, Golden JP, Nasir M, Ligler FS (2009) Multiplexed detection of bacteria and toxins using a microflow cytometer. Anal Chem 81:5426–5432 34. Li Y, Zhang C, Xing D (2011) Integrated microfluidic reverse transcription-polymerase chain reaction for rapid detection of food- or waterborne pathogenic rotavirus. Anal Biochem 415: 87–96 35. Verdoy D, Barrenetxea Z, Berganzo J, Agirregabiria M, Ruano-López JM, Marimón JM, Olabarría G (2012) A novel real time micro PCR based point-of-care device for Salmonella detection in human clinical samples. Biosens Bioelectron 32:259–265
Natinan Bunyakul received his Ph.D. degree in medical technology from Mahidol University (Thailand) in 2010. He is now a lecturer and researcher in the Faculty of Medical Technology, Mahidol University. His research focuses on the development of biosensors, especially lab-on-a-chip devices for environmental and clinical applications.
N. Bunyakul et al. Chamras Promptmas is Assistant Professor at Mahidol University. He received his diploma in biotechnology from the University of Kent at Canterbury, UK, in 1988 and his Ph.D. degree in biochemistry from Mahidol University in 1994. He has been working at Mahidol University for several years on the development of biosensors for laboratory diagnosis.
Antje Baeumner is Professor and Director of the Institute of Analytical Chemistry, Chemo and Biosensors at the University of Regensburg, Germany. In 2013 she moved there from her previous scientific home, the Department of Biological and Environmental Engineering at Cornell University, where the research described also was performed. Her research focuses on the development of miniaturized biosensors and micro total analysis systems for the detection of pathogens and toxins for food, environmental, and clinical applications.
