Anal. Chem. 2005, 77, 2960-2965
Stable and Fluid Ethylphosphocholine Membranes in a Poly(dimethylsiloxane) Microsensor for Toxin Detection in Flooded Waters K. Scott Phillips, Yi Dong, David Carter, and Quan Cheng*
Department of Chemistry, University of California, Riverside, California 92521
Highly stable and fluid supported bilayer membranes were fabricated by fusion of positively charged ethylphosphocholine (DOPC+) vesicles into poly(dimethylsiloxane) (PDMS) microchannels for immunosensing of cholera toxin (CT) in flooded waters. Compared to phosphatidylcholine (PC) layers in the microchannels, DOPC+ membranes show exceptionally strong resistance to air-dry damage, as demonstrated by fluorescence recovery after photobleaching (FRAP) measurements and protein adsorption studies. In FRAP experiments, the mobile fraction of PC membranes was found to decrease by 10% upon drying/rehydration and the lateral diffusion coefficient decreased from 2.2 to 1.6 µm2/s, whereas the mobile fraction and diffusion coefficient for DOPC+ membranes remain virtually unchanged during this process. Characterization by confocal microscopy reveals that only 1% of the DOPC+ membrane in the microchannels was removed by the drying/rehydration process, as compared to 11% for PC. Protein adsorption trends indicate that the charge of DOPC+ membranes allows for tuning of solution conditions to enable the desired proteinmembrane interaction to predominate at the interface. A flow-based immunoassay for bacterial toxin was developed with 5% GM1/DOPC+ membranes in PDMS channels, and a detection limit of 250 amol for CT was obtained from the calibration curves. The assay was successfully applied to detection of CT spiked in water samples from the Santa Ana River, with nearly identical response and sensitivity. Recent advances in supported bilayer membranes (SBMs) have exciting implications for bioanalytical research1,2 because of the increasing capacity to mimic cell membranes, in which vital biological processes such as energy transduction, signaling, and transport occur.3 SBMs have been used in the study of membraneassociated processes such as membrane-protein interactions,4 ion transport,5,6 toxin-induced pore formation,7,8 and cellular * To whom correspondence should be addressed. Phone: (951) 827-2702. E-mail:
[email protected]. (1) Trojanowicz, M. Membr. Sci. Technol. Ser. 2003, 7, 807-845. (2) Tien, H. T.; Ottova, A. L. J. Membr. Sci. 2001, 189, 83-117. (3) Oellerich, S.; Lecomte, S.; Paternostre, M.; Heimburg, T.; Hildebrandt, P. J. Phys. Chem. B 2004, 108, 3871-3878.
2960 Analytical Chemistry, Vol. 77, No. 9, May 1, 2005
recognition.9 They are also an effective sensing platform for analysis of glucose,10 neurotransmitters,11 artificial sweeteners,12 pharmaceutical drugs,13 and bacteria.14 In the burgeoning field of proteomics, SBMs have been applied to investigations of peptide and protein functions, where compatibility with popular detection platforms such as quartz crystal microbalance, surface plasmon resonance, electrochemistry, and fluorescence microscopy presents a critical advantage.15-19 Supported membranes are held together by a delicate balance of electrostatic, hydration, and van der Waals forces,20 and it has been suggested that exposure to air and even shear flow may damage or destroy them.15,21 Therefore, a fundamental challenge in SBM design is to reconcile mechanical stability and lateral fluidity in one structure. One approach is the use of hybrid bilayer membranes, which consist of a self-assembled monolayer (SAM) underlayer and a single membrane leaflet on top and exhibit much higher stability than bilayer membranes on solid substrates. Other methods such as polymerization22 and “heat stabilization”15 have been employed in SBM fabrication. Significant reduction of lateral mobility, however, is often observed in these membranes. Biologi(4) Cezanne, L.; Lopez, A.; Loste, F.; Parnaud, G.; Saurel, O.; Demange, P.; Tocanne, J.-F. Biochemistry 1999, 38, 2779-2786. (5) Bartolommei, G.; Buoninsegni, F. T.; Moncelli, M. R. Bioelectrochemistry 2004, 63, 157-160. (6) Buoninsegni, F. T.; Dolfi, A.; Moncelli, M. R.; et al. J. Inorg. Biochem. 2001, 86, 448-448. (7) Wilkop, T.; Xu, D.; Cheng, Q. Mater. Res. Soc. Symp. Proc. 2003, 774, 223228. (8) Xu, D.; Cheng, Q. J. Am. Chem. Soc. 2002, 124, 4208-4209. (9) Qi, S. Y.; Groves, J. T.; Chakraborty A. K.; Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 6548-6553. (10) Snejdarkova, M.; Rehak, M.; Otto, M. Anal. Chem. 1993, 65, 665. (11) Nikolelis, D. P.; Petropoulou, S.-S. E. Biochim. Biophys. Acta 2002, 1558, 238-245. (12) Nikolelis, D. P.; Pantoulias, S. Anal. Chem. 2001, 73, 5945-5952. (13) Matsuno N.; Murawsky M.; Ridgeway J.; Cuppoletti J. Biochim. Biophys. Acta 2004, 1665, 184-190. (14) Ottova, A. L.; Tien, H. T. J. Surf. Sci. Technol. 2000, 16, 115-148. (15) Stine, R.; Pishko, M. V.; Schengrund, C.-L. Langmuir 2004, 20, 65016506. (16) Terrettaz, S.; Stora, T.; Duschl, C.; Vogel, H. Langmuir 1993, 9, 13611369. (17) Liley, M.; Bouvier, J.; Vogel, H. J. Colloid Interface Sci. 1997, 194, 53-58. (18) Stelzle, M.; Weissmuller, G.; Sackmann, E. J. Phys. Chem. 1993, 97, 29742981. (19) Fang, Y.; Frutos, A. G.; Lahiri, J. Langmuir 2003, 19, 1500-1505. (20) Groves, J. T.; Ulman, N.; Cremer, P. S.; Boxer, S. G. Langmuir 1998, 14, 3347-3350. (21) Holden, M. A.; Jung, S.-Y.; Yang, T.; Castellana, E. T.; Cremer, P. S. J. Am. Chem. Soc. 2004, 126, 6512-6513. 10.1021/ac0500481 CCC: $30.25
© 2005 American Chemical Society Published on Web 03/26/2005
Figure 1. Structure of lipids used in formation of SBMs in PDMS microchannels: PC (top), DOPC+ (middle), and DOTAP (bottom).
cal processes and binding events that require free movement of lipid constituents in the membrane plane can be unfavorably affected. The development of micropatterned SAMs has made it possible to construct arrays of hydrophilic “wells” surrounded by hybrid bilayer walls. The bilayer “patches” assembled in these wells are fluid and have been shown to have improved stability.23 There has been considerable interest in use of specific interactions and functionalization of supporting surfaces for the improvement of SBMs.24 A novel approach was recently demonstrated by Cremer et al. using a monolayer of proteins anchored to streptavidin receptors in the SBM to increase the shelf life in dehydration conditions.21 The use of γ-aminopropylsilane coatings on glass slides has allowed formation of stable SBMs that appear to resist air-dry damage.25 A noticeable trend was observed in the use of electrostatic forces to enhance surface function and improve membrane-substrate interactions, thereby strengthening membrane stability for bioanalytical applications.26 Polyelectrolyte multilayers as a “cushion layer” for generating stable bilayer membranes have been the subject of wide investigation.27,28 Hydrogels tailored with charged functional groups were found effective to increase membrane robustness in an ion channel device.29 In addition, the use of charged supported bilayers has been demonstrated for electrophoresis of DNA molecules electrostatically adsorbed on the membrane30 and electrostatic delivery of membrane components to supported membrane patterns.31 In this work, we report the use of electrostatic interactions to form highly stable and fluid SBMs in poly(dimethylsiloxane) (PDMS) microchannels for immuno-analysis of bacterial toxins. Figure 1 shows the structure of the charged lipids used in SBM formation. The influence of lipid headgroup on membrane properties was studied. The primary focus of the research was placed (22) Johnston, D. S.; Sanghera, S.; Pons, M.; Chapman, D. Biochim. Biophys. Acta 1980, 602, 57-69. (23) Jenkins, A. T. A.; Boden, N.; Bushby, R. J.; Evans, S. D.; Knowles, P. F.; Miles, R. E.; Ogier, S. D.; Schonherr, H.; Vancso, G. J. J. Am. Chem. Soc. 1999, 121, 5274-5280. (24) Ekeroth, J.; Konradsson, P.; Hook, F. Langmuir 2002, 18, 7923-7929. (25) Fang, Y.; Frutos, A. G.; Lahiri, J. J. Am. Chem. Soc. 2002, 124, 2394-2395. (26) Matthews, J. R.; Tencel, D.; Jacobs, R. M. J.; Bain, C. D.; Anderson, H. L. J. Am. Chem. Soc. 2003, 125, 6428-6433. (27) Malzak, K. A.; Ellar, D. J.; Gizeli, E. Langmuir 2004, 20, 1386-1392. (28) Zhang, L.; Longo, M. L.; Stroeve, P. Langmuir 2000, 16, 5093-5099. (29) Anrather, D.; Smetazko, M.; Saba, M.; Alguel, Y.; Schalkhammer, T. J. Nanosci. Nanotechnol. 2004, 4, 1-22. (30) Zhou, D.; Wang, X.; Birch, L.; Rayment, T.; Abell, C. Langmuir 2003, 19, 10557-10562. (31) Schouten, S.; Stroeve, P.; Longo, M. L. Langmuir 1999, 15, 8133-8139.
on lipid ethylphosphocholine (DOPC+), which demonstrates potential to form exceptionally stable membranes on PDMS. Significant effort was spent on characterizing the mobile fraction and lateral diffusion of the supported membranes upon dehydration. We aim to understand the underlying principles that affect membrane integrity and develop new protocols that take advantage of the surface charge to promote desired protein-membrane interactions at the interface. A membrane-based microfluidic sensor chip was demonstrated for the immunoassay of cholera toxin (CT) using cell surface receptor GM1, which requires membrane fluidity for multivalent binding. The feasibility of using this sensor for testing CT in floodwaters was further evaluated with water samples from the Santa Ana River in Southern California. EXPERIMENTAL SECTION Materials and Instrumentation. The monosialoganglioside receptor (GM1) was purchased from Matreya (Pleasant Gap, PA). Phosphatidylcholine (PC), 1,2-dioleoyl-sn-glycero-3-ethylphosphocholine (DOPC+), 1,2-dioleoyl-3-trimethylammonium-propane (DOTAP) and 1-oleoyl-2-[6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]hexanoyl]-sn-glycero-3-phosphocholine (NBD-PC) were obtained from Avanti Polar Lipids (Alabaster, AL). CT, avidin Texas Red conjugate, anti-BSA serum, and rabbit anti-cholera serum were purchased from Sigma. Texas Red conjugated bovine serum albumin (BSA) and AlexaFluor 532 conjugated goat antirabbit IgG were from Molecular Probes. All other chemicals were reagent grade and were used as received. For fabrication of PDMS chips, Dow Corning Sylgard 184 silicon elastomer base and curing agent were obtained from KR Anderson. Glass and PDMS surfaces were treated with a Harrick PDC32G plasma cleaner. An in-house setup was used to perform contact angle measurements. Fluorescence imaging was performed using an Amersham Typhoon 9410 scanner using 532nm excitation and collecting emission between 595 and 625 nm for Texas Red and 545-565 nm for Alexa 532. The fluoresence recovery after photobleaching (FRAP) experiments were carried out using a Meridian Insight confocal laser scanning microscope (CLSM) with argon laser excitation, cooled CCD, and 505-nm longpass emission filter with a 40×/0.75na Achroplan dipping objective. Preparation of SBM Microchips. Preparation and treatment of the PDMS chip was carried out as previously reported.32 Microchannels on PDMS were 200 µm × 200 µm × 2 cm with wider cylindrical ends for connections. A syringe pump from KD Scientific and low-pressure sample injectors from Upchurch were used for flow injection experiments. Vesicles were prepared by probe sonication. Stock solutions of lipids in chloroform were mixed with the appropriate molar ratio, and the solvent was removed under a nitrogen stream. The dried lipids were then resolubilized in a Tris buffer (10 mM with 150 mM NaCl, pH 7.4) and probe sonicated for 10 min, followed by 1-h incubation at 4 °C and purification with an ultracentrifuge (38 000 rpm for 30 min and then at 52 000 rpm for 3 h).33 Characterization of SBMs. (I) Contact Angle Measurements. A degassed mixture of PDMS and Sylgard 184 elastomer (32) Phillips, K. S.; Cheng, Q. Anal. Chem. 2005, 77, 327-334. (33) Barenholz, Y.; Gibbes, B. J. G.; Thompson, T. E.; Carlson, F. D. Biochemistry 1977, 16, 2806-2810.
Analytical Chemistry, Vol. 77, No. 9, May 1, 2005
2961
curing agent (10:1) was poured onto clean glass slides in thin layers and cured at 65 °C for 1 h. After a 30-s oxidation of the PDMS surface, test areas defined by a PDMS frame were filled with buffer or vesicle solutions for 1 h. They were gently washed with buffer three times and dried with a clean air stream before the contact angle measurements were carried out. At least three separate substrates were used for each data point. (II) Fluorescence Techniques. To prepare fluorescence membranes, 2% NBD-PC was added to the lipid solution in chloroform. In FRAP and stability measurements, the prepared chip containing fluorescent membranes was rinsed with Tris buffer for 15 min, disassembled, and secured in a small Petri dish, followed by observation with a CLSM using a dipping objective. The detailed FRAP procedures and data analysis method have been described previously.32 Briefly, the CLSM was used to bleach a perpendicular line in the channels and monitor fluorescence recovery with time, and the data were analyzed by the model developed by Koppel.34 For protein adsorption studies, the fluorescently labeled protein was injected for 1 min at a flow rate of 1 mL/h and then rinsed for 5 min with Tris buffer (pH 7.4). The glass slide was then removed, and the channels were imaged on the Typhoon fluorescence scanner. Two background channel intensities were collected for each chip and used for signal correction. In longer trials, a 2-h flow time was used in place of the 5-min vesicle rinse period. CT Immunosensing. To prepare toxin-binding bilayer membranes on PDMS, 5% ganglioside GM1 receptor molecules were added to the DOPC+ lipids. During the immunoassay, the flow was maintained at 1 mL/h. An injection volume of 37 µL was used for CT, primary antibody, and secondary fluorescent conjugated antibody, with 5-min rinse times between each step. At the end of the assay, an additional 15-min rinse was added to remove any weakly adsorbed antibodies. The chips were then imaged and quantified on the Typhoon scanner. RESULTS AND DISCUSSION Wettability and Stability of Charged Membranes on PDMS. Contact angle measurements were used to investigate wettability of PDMS surfaces upon vesicle treatment. Previous studies show that oxidized PDMS exhibits hydrophobic recovery,35 whereas surfaces covered with PC membranes retain long-standing hydrophilic properties.32 In this work, the wetting capacity of PDMS surfaces treated with vesicles from the cationic lipids DOPC+ and DOTAP was evaluated. DOPC+ treated substrates had an initial contact angle of 70°. This is relatively high compared to PC treated PDMS surfaces, probably due to the ethyl ester group of the headgroup. The DOTAP lipid, in which the glycerol bridge is directly terminated with an amine, resulted in an even higher contact angle of 79°. Neither DOPC+ nor DOTAP surfaces showed noticeable change from the initial angles after 2 h, indicating sustained wettability. We then studied the stability of charged membranes in PDMS microchannels. The supported membranes of zwitterionic PC on glass are known to be highly fragile and have been observed to peel off easily when passed through the air-water interface.36 (34) Koppel, D. E. Biophys. J. 1979, 28, 281-292. (35) Duffy, D. C.; McDonald, J. C.; Schueller, O. J. A.; Whitesides, G. M. Anal. Chem. 1998, 70, 4974-4984. (36) Cremer, P. S.; Boxer, S. G. J. Phys. Chem. B 1999, 103, 2554-2559.
2962 Analytical Chemistry, Vol. 77, No. 9, May 1, 2005
Table 1. Comparison of Lipid Membrane Stability after Dehydration/Rehydrationa lipid
damaged
removed
total
n
PC DOPC+
12 6
11 1
23 7
6 6
a The data are expressed as a fraction (%) of channel damaged after vacuum dessication as determined by confocal fluorescence.
However, we have noticed that PC SBMs in PDMS microchannels appear to be less easily removed. For further comparison, PC and DOPC+ were tested for their resistance to damage from dehydration. Lahiri and co-workers have developed a method to quantify membrane damage by comparing fluorescence intensity measurements before and after drawing the membrane through the air/ water interface.25 This procedure is somewhat unsuitable for membranes in microchannels, where the retention of moisture by the channel structure leads to incomplete dryness. We therefore modified the method by using a desiccated vacuum chamber to achieve gentle and complete dehydration of the supported membrane. The SBM coated chips were rinsed with DI water, disassembled, and placed in a desiccated vacuum chamber for 30 min. At this time, the surface appeared to be evenly dried under an optical microscope. The chips were then reassembled and rinsed for 15 min with Tris buffer. This allowed for reproducible dehydration without causing mechanical damage. The fluorescence intensity in the channels was measured with a confocal microscope. Areas that were completely dark were considered to have peeled off completely, whereas areas that deviated from the normally homogeneous fluorescence of pristine membranes were considered to be damaged. Both types of areas were quantified by total length occupied in the channels and summarized as percentage of the total channel length in Table 1. It is worth noting that PC membranes did not completely peel off from the microchannels after the dehydration process. Only 11% of the membrane was removed and another 12% appeared to be damaged, indicating that the PDMS microchannels protect PC SBMs as compared to glass slides. DOPC+ membranes, on the other hand, showed remarkable stability with only 1% peeling and 6% damage. In other words, DOPC+ membranes yield ∼10-fold greater stability against peeling than the PC membranes, which are the most commonly used and studied SBMs. The positively charged DOPC+ membranes may promote more favorable electrostatic attraction to the negatively charged oxidized PDMS functional groups,37 leading to the observed stability on the surface. Recent work with charged lipids on polyelectrolytes shows that local electrostatic interactions have a profound influence on stability and dynamics of SBMs.38 Effect of Dehydration-Rehydration on SBM Fluidity. Fluidity is an important parameter that defines the properties of bilayer membranes on a supported substrate. The thin water pocket between the lipid membrane and the surface results in characteristic lateral fluidity below the gel-phase transition temperature. A diffusion coefficient (D) of ∼1-5 µm2/s is typically (37) Morra, M.; Occhiello, R. M.; Garbassi, F.; Humphrey, P.; Johnson, D. J. Colloid Interface Sci. 1990, 137, 11-24. (38) Wang, M.; Dilek, I.; Armitage, B. A. Langmuir 2003, 19, 6449-6455.
Figure 2. Fluorescence images of the recovery process after photobleaching for various membranes on PDMS. From top to bottom: dehydrated PC, pristine DOPC+, and dehydrated DOPC+ at 5-, 15-, 35-, and 60-s intervals. All images are background corrected. Table 2. Diffusion Coefficient for PDMS Microchannel Supported PC and DOPC+ Membranes lipid PC DOPC+
glass PDMS rehydrated glass PDMS rehydrated
D (µm2/s)
β
n
r2
4.0 ( 1.3 2.2 ( 0.9 1.6 ( 0.1 3.7 ( 0.7 1.8 ( 0.5 1.7 ( 0.5
1.00 1.00 0.91 0.99 0.97 0.97
14 18 7 9 10 5
0.995 0.994 0.994 0.995 0.992 0.994
observed for membranes on glass39 and is a good indicator that vesicle fusion resulted in a bilayer structure. FRAP experiments were performed in situ in the microchannels to examine lateral fluidity before and after vacuum dehydration. Figure 2 shows typical FRAP images obtained at different times after photobleaching, and Table 2 shows the summary of the results obtained for the two lipids, PC and DOPC+ (β is mobile fraction and n sampling size). Both glass and PDMS surfaces were used as substrate in the experiment for comparison. The diffusion coefficient for DOPC+ on glass is 3.7 µm2/s, which is slightly less than that of PC (D ) 4.0 µm2/s). On PDMS, the diffusion coefficient for DOPC+ is 1.8 µm2/s, which is also less than that for PC (2.2 µm2/s). This value, however, is still in a range considered to be highly mobile for solid supported membranes. The reduced lateral mobility of DOPC+ membranes compared to PC SBMs on both surfaces likely reflects a stronger DOPC+/ surface interaction. The origin of the small 3% immobile fraction (1 - β) for DOPC+ on PDMS is not known. DOTAP membranes did not show any significant recovery within the time scale of the experiments, indicating very limited mobility. When dehydration results in structural damage or nonspecific adsorption, a substantial decrease in the lateral mobility should be observed. To compare the air-dry stability of DOPC+ and PC membranes, the lipid-modified PDMS channels were dehydrated as described above and studied by FRAP. For both lipids, photobleaching of membranes in the dry state resulted in no recovery. After rehydration with buffer, even areas of PC membranes that appeared to be unaffected, as judged by fluorescence (39) Lee, G. M.; Jacobson, K. Curr. Top. Membr. 1994, 40, 111-142.
visualization, exhibited a difference in lipid mobility (Table 2). The mobile fraction of PC decreased by ∼10% and diffusion became much slower, a likely result of membrane structural change or damage. Although this damage was too subtle to be visualized with a confocal fluorescence microscope, it was readily observable with the FRAP technique, which provides additional information about molecular-level fluidity changes. Remarkably, DOPC+ showed no significant change in fluid properties after dehydration. Both mobile fraction and diffusion coefficient of DOPC+ membranes remain virtually the same during the process. This clearly indicates the superb quality of DOPC+ membranes to resist dehydration damage and preserve mobility. The observed behavior is similar to the air-stable bilayers fabricated by Holden et al. on glass21 but without the use of a protective protein layer. Effect of pH, Charge, and Dehydration on SBM Nonspecific Protein Adsorption. One of the most fundamental tasks in biosensing is to control nonspecific adsorption. We have previously demonstrated that nonspecific protein adsorption on PC functionalized PDMS could be reduced by several orders of magnitude compared to the bare PDMS substrate.32 Figure 3 shows the fluorescent signal from the labeled proteins adsorbed on PDMS microchannels treated with each of the three lipids. The values obtained for BSA on zwitterionic PC membranes are similar at all pH values, indicating that the attractive force is not significantly affected by solution conditions. However, the adsorption on ionic SBMs is highly pH dependent. Both cationic lipids DOPC+ and DOTAP show a large pH-dependent change in adsorption for BSA. For DOPC+, minimal adsorption of BSA was obtained at pH 4.5, but adsorption increases ∼10-fold at pH 9.0. This is due to strong electrostatic interaction at higher pH since the pI of BSA is ∼4.8. Similar pH-dependent trends were observed for BSA on DOTAP membranes, but the adsorption levels were noticeably higher. The adsorption of avidin is typically weaker than BSA because it is positively charged at all pH studied (note the scale of the x-axis in Figure 3b vs Figure 3a), and it is reduced to near the detection limit on DOPC+ membranes at neutral or higher pH. These results suggest that the use of ionic lipid headgroups has potential to augment SBM functionality by conferring the ability to selectively attract or reject charged analytes or interfering species. We then investigated whether the protein resistance of supported membranes was compromised by dehydration conditions. Protein adsorption experiments were performed with avidin after purging the channels for 20 min with a nitrogen stream and then rehydrating with the Tris buffer. The results are shown in Figure 4. Avidin adsorption increases by 128 RU on PC and 65 RU on DOTAP membranes after dehydration. However, there is only an increase of 2 RU in DOPC+ treated channels, which is well within the standard deviation in the measurement. This further confirms that both the structure and function of supported DOPC+ membrane are highly stable in air-dry conditions, making it an ideal membrane for real-world applications. Immunosensing of CT Using DOPC+/PDMS Microsensors. Characterization of the DOPC+ membranes in PDMS microchannels revealed favorable properties for design of longlasting sensors with SBM functionality. A flow immunoassay in the PDMS microchannels was therefore developed to detect cholera toxin, one of the most significant bacterial toxins in floodwaters. The DOPC+ membranes were functionalized with 5 Analytical Chemistry, Vol. 77, No. 9, May 1, 2005
2963
Figure 3. Comparison of protein adsorption on lipid membranes in PDMS microchannels at different solution pH. (A) BSA. (B) Avidin.
Figure 4. Fluorescence intensity for avidin adsorption on pristine (gray) and dehydrated/rehydrated membranes (black).
mol % ganglioside GM1 receptors, which insert spontaneously into the lipid vesicles formed during sonication. The use of GM1 receptors to develop CT sensors has been widely adopted.40 The pentasaccharide headgroup of GM1 binds strongly to one of the five B subunits of CT, while the ceramide tail remains anchored firmly in the membrane. Figure 5 shows the calibration curve for 12 different concentrations of CT prepared in Tris buffer. The background-subtracted fluorescent response shows a linear relationship for concentrations up to 20 ng/mL CT. The average reproducibility of the response was 9% at the 2 ng/mL concentration. The detection limit (3σ) is calculated to be 0.4 ng/mL, which represents a concentration of 5 pM and an absolute quantity of 250 amol. This is 1000-fold lower than a 0.5 µg/mL LOD obtained with heat-stabilized glycosphingolipid films using QCM,15 100-fold lower than with a GM1-based fluorescence array biosensor,41 and comparable to the detection limit obtained by an ISFET-based CT sensor42 and a microarray device using GM1 SBMs on γ-aminopropylsilane.19 The high sensitivity obtained here is the result of ideal receptor presentation in the membrane, rapid mass transport in the small channels, and very little background response from nonspecific binding of antibodies. Experiments with pure GM1treated PDMS microchannels yield no detectable specific response, indicating that the lipid membrane as the host matrix is essential for high sensitivity. A control immunoassay performed (40) Kuziemko, G. M.; Stroh, M.; Stevens, R. C. Biochemistry 1996, 35, 63756387. (41) Rowe-Taitt, C. A.; Cras, J. J.; Patterson, C. H.; Golden, J. P.; Ligler, F. S. Anal. Biochem. 2000, 281, 123-133. (42) Zayats, M.; Raitman, O. A.; Chegel, V. I.; Kharitonov, A. B.; Willner, I. Anal. Chem. 2002, 74, 4763-4773.
2964 Analytical Chemistry, Vol. 77, No. 9, May 1, 2005
Figure 5. Calibration curves for immunoassay of cholera toxin using DOPC+/GM1 membranes in PDMS microchannels. Sample prepared in 10 mM Tris buffer (open squares) and spiked in Santa Ana River water sample (circles). Inset shows the expanded view of low concentration data and linear regression lines.
in the same manner with DOPC+ membranes without GM1 receptors confirmed that signal from nonspecific binding of CT to the SBMs was not responsible for the high sensitivity. To prove that GM1 is selectively bound by cholera toxin, an immunoassay with 1 µg/mL BSA and anti-BSA primary antibody was performed on DOPC+-GM1 treated channels. The fluorescent response was similar to background, suggesting that the sensor can be used in complex biological matrixes without interference by endogenous proteins. The microfluidic sensor with robust DOPC+ membranes was further used to test for CT in environmental samples. A major advantage of heterogeneous flow immunoassay is that no separation step is needed to detect analytes in complex mixtures. The capability to detect CT in floodwaters would help identify and prevent waterborne outbreaks that occur frequently during heavy rains and flood seasons in Asia.43 In recent years, cases have been on the rise globally in 94 nations, including the emergence of new serotypes that complicate analysis.44 Real sampling conditions often place greater demands on analytical methods because of matrix interferences. The DOPC+ membranes in PDMS micro(43) WHO Cholera Outbreak: Assessing the outbreak response and improving preparedness; World Health Organization: Geneva. 2005. WHO/CDS/CPE/ ZFK/2004.4
channels, because of greater stability and the ability to tune membrane-protein interactions, should be highly suitable for use in these conditions. The environmental water sample was collected from the Santa Ana River in Riverside, CA. The river water is known to be polluted by discharge from several upstream cities and is listed with the EPA as a 303(d) impaired waterway for high levels of pathogens, nutrients, and suspended solids. All samples were filtered, stored in Nalgene containers at 5 °C, and used soon after collection. A nine-point calibration curve for the CT spiked Santa Ana water is shown in Figure 5. The response was very similar to that achieved for CT prepared in Tris buffer, suggesting that sensors developed with the DOPC+ membranes do not suffer significant interference from the river matrix. CONCLUSIONS A significant improvement in SBM performance has been obtained through the use of DOPC+ vesicles in PDMS microchannels. The resulting membranes are highly fluid and nearly undamaged when dehydrated. The realization of an air-stable and laterally mobile system is a large step forward for developing analytical methods with biomimetic membranes. Membrane-based microdevices can be simplified if they do not have the requirement of being stored in solution at all times. In addition, increased stability allows for events such as chip disassembly and air plugs, enabling advanced applications such as array patterning or internal valves/actuators for sophisticated fluid pathways. In working with the DOPC+ membranes, we have observed that they are more forgiving of real-world conditions than the fragile PC SBMs. The immunoassay demonstrated here proves that DOPC+ SBMs are not only desirable for the properties mentioned above (44) Chhotray, G. P.; Pal, B. B.; Khuntia, H. K.; Chowdhury, N. R.; Chakraborty, S.; Yamasaki, S.; Ramamurthy, T.; Takeda, Y.; Bhattachary, S. K.; Nair, G. B. Epidemiol. Infect. 2002, 128, 131-138. (45) Lenz, P., Ajo-Franklin, C. M., Boxer, S. G. Langmuir 2004, 20, 1109211099.
but also improve the performance of the GM1/CT sensing compared to PC in normal operating conditions. The high sensitivity demonstrated here allows them to be adopted for analysis of other protein toxin targets. The results for detection of CT spiked in river water show that the membranes can yield reproducible response with sensitivity comparable to or better than the time- and reagent-consuming conventional ELISA performed in microwell plates. Further work is needed to determine the basis of improved membrane stability and pinpoint the critical factors involved. Although we believe it is likely electrostatic in nature, the molecular-level interaction of headgroup constituents of DOPC+ with PDMS remains elusive. This is particularly true in the case of similarly charged DOTAP SBMs, which had properties noticeably different from DOPC+ SBMs. A recent paper by Boxer et al. on supported membranes on PDMS shed some light on the role of surface chemical properties in dictating membrane structure (monolayer vs bilayer).45 We are currently testing DOPC+ membranes on PDMS with varied composition in hopes of completely eliminating hydrophobic recovery and thereby allowing dehydrated DOPC+ membranes to remain stable for weeks or even months. This would open the door to commercialization of SBMs that can be stored in conventional conditions and used as needed for research in life sciences, laboratory pathology, and environmental field work. ACKNOWLEDGMENT Support from Eli Lilly Young Analytical Chemist Grant and UC Riverside is acknowledged. Received for review January 10, 2005. Accepted March 2, 2005. AC0500481
Analytical Chemistry, Vol. 77, No. 9, May 1, 2005
2965
