Heat-Stabilized Glycosphingolipid Films for Biosensing Applications

Langmuir 2004, 20, 6501-6506

6501

Heat-Stabilized Glycosphingolipid Films for Biosensing Applications Rory Stine and Michael V. Pishko* Departments of Chemical Engineering, Chemistry, and Materials Science and Engineering, The Pennsylvania State University, University Park, Pennsylvania 16802

Cara-Lynne Schengrund Department of Biochemistry and Molecular Biology, Milton Hershey Medical Center, The Pennsylvania State University, 500 University Drive, Hershey, Pennsylvania 17033 Received February 20, 2004. In Final Form: May 17, 2004 We have investigated a means of producing thin, oriented lipid monolayers which are stable under repeated washing and which may be useful in biosensing or surface-coating applications. Phosphatidylcholine and the glycosphingolipid GM1 were used as representative lipids for this work. Initially, a mixed selfassembled monolayer of octanethiol and hexadecanethiol was produced on a gold surface. This hydrophobic monolayer was then brought into contact with a thin lipid film that had been assembled at the liquid/air interface of a solution, allowing the lipid to deposit on the gold surface through hydrophobic interactions. The lipid layer was then heated to cause intermingling of the fatty acid and alkanethiol chains and cooled to form a highly stable film which withstood repeated rinsing and solution exposure. Presence and stability of the film were confirmed via ellipsometry, Fourier transform infrared spectroscopy, and quartz crystal microbalance (QCM), with an average overall film thickness of ∼3.5 nm. This method was then utilized to produce GM1 layers on gold-coated QCM crystals for affinity sensing trials with cholera toxin. For these sensing elements, the lower detection limit of cholera toxin was found to be approximately 0.5 µg/mL, with a logarithmic relationship between toxin concentration and frequency response spanning over several orders of magnitude. Potential sites for nonspecific adsorption were blocked using serum albumin without sacrificing toxin specificity.

Introduction The utilization of biological molecules for sensing applications has received a great deal of attention in recent years. The highly specific binding nature of enzymes, antibodies, and other biomolecules offers a potentially selective way of detecting proteins and other biologically important compounds. In particular, the recent emergence of new pathogens1 and the increasing threat of bioterrorism2 have placed the need for effective biotoxin sensors in the spotlight. Traditionally, biosensing has relied on the use of protein-based recognition molecules, such as antibodies, as its primary detection method. More recently, however, the use of glycolipids has gained acceptance as an effective recognition molecule for biotoxin sensing. Recent studies have successfully used glycolipids in fluorescent,3 surface plasmon resonance (SPR),4 and quartz crystal microbalance (QCM)5 based biotoxin sensors. Glycosphingolipids (GSLs), consisting of a carbohydrate headgroup attached to a sphingosine-derived tail, make up a small percentage of the lipid component of cell membranes. For many biological toxins, including cholera, ricin, botulinum, and tetanus, binding to the carbohydrate portion of a GSL is the first step in their interaction with * To whom correspondence should be addressed. Phone: (814) 863-4810. Fax: (814) 865-7846. E-mail: [email protected]. (1) Shi, Y.; Yi, Y.; Li, P.; Kuang, T.; Li, L.; Dong, M.; Ma, Q.; Cao, C. J. Clin. Microbiol. 2003, 41, 5781-5782. (2) Cloud, D. In Wall Street Journal; 2002. (3) Song, X.; Swanson, B. I. Anal. Chem. 1999, 71, 2097-2107. (4) Gustafson, I. Colloids Surf., B 2003, 30, 13-24. (5) Spangler, B. D.; Tyler, B. J. Anal. Chim. Acta 1999, 399, 51-62.

cells.6 The specificity of this binding has been utilized in a number of methods developed to detect such toxins as cholera,3-5,7-10 the heat-labile enterotoxin of E. coli,5,11 ricin,4 botulinum,7 and tetanus.7,8 Several fluorescencebased detection methods which utilize liposomes containing the GSL of interest as well as a fluorescent tag have been investigated.7,9 However, these techniques require a secondary antibody and significant preparation prior to testing. For continuous flow analysis, the GSLs must be immobilized on the detection surface in a manner that allows for the carbohydrate headgroup to present itself at the surface/fluid interface. Using GSLs in this manner presents several advantages over the traditional use of antibodies. These include the ability to easily orient the carbohydrate portion of the GSL toward the surface/fluid interface through hydrophobic/hydrophilic interactions, more active sites for protein binding per unit area, and the greater stability of GSLs when compared to antibodies under a variety of conditions. Structure,12-14 electrochemical behavior,15,16 and biosensing17,18 properties of supported lipid layers have been (6) Berg, J.; Tymoczko, J.; Stryer, L. Biochemistry, 5th ed.; W. H. Freeman and Company: New York, 2002. (7) Singh, A. K.; Harrison, S. H.; Schoeniger, J. S. Anal. Chem. 2000, 72, 6019-6024. (8) Fang, Y.; Frutos, A. G.; Lahiri, J. Langmuir 2003, 19, 15001505. (9) Ahn-Yoon, S.; DeCory, T. R.; Baeumner, A. J.; Durst, R. A. Anal. Chem. 2003, 75, 2256-2261. (10) Fisher, M. I.; Tjarnhage, T. Biosens. Bioelectron. 2000, 15, 463471. (11) Spangler, B. D.; Wilkinson, E. A.; Murphy, J. T.; Tyler, B. J. Anal. Chim. Acta 2001, 444, 149-161. (12) Tang, Z.; Jing, W.; Wang, E. Langmuir 2000, 16, 1696-1702. (13) Brechling, A.; Sundermann, M.; Kleinberg, U.; Heinzmann, U. Thin Solid Films 2003, 433, 281-286.

10.1021/la049554w CCC: $27.50 © 2004 American Chemical Society Published on Web 06/25/2004

6502

Langmuir, Vol. 20, No. 15, 2004

studied. Supported lipid layers have also been incorporated into several biotoxin detection systems that could potentially be developed for continuous flow analysis using fluorescence,3,8 SPR,4 and/or QCM5,11,19 based sensing techniques. In each of the reported studies, however, the lipid layers were merely adsorbed to the supporting surface. Cremer and colleagues20 found that similar lipid bilayers are prone to detachment under shear stress. In a continuous flow analysis system this could pose problems such as loss of activity over time and exposure of the underlying supportive layer to nonspecific protein adhesion and fouling. Spangler and colleagues reported using GM1 on a QCM crystal to detect cholera toxin5 and heatlabile enterotoxin.5,11 However, the GM1 layers used in those studies were simply adsorbed to the QCM crystal, and the issues of lipid layer detachment and nonspecific response due to nonspecific protein attachment were not addressed. As we will show, certain lipid layers are prone to shear-induced detachment even when anchored to the supporting surface via a hydrophobic self-assembled monolayer. Stabilization of supported lipid bilayers has been accomplished via photoinitiated cross-linking of the lipid film,21 but this method is only suitable for lipid molecules that have the functional groups that are necessary for the cross-linking reaction. Here, we report on a heat-stabilized, supported lipid monolayer that could be rinsed repeatedly without significant loss when compared to a control layer that had not been subjected to heat stabilization. This study investigates in greater detail the effects of heat curing on lipid films that was described by Rowe-Taitt and colleagues22 as being essential to immobilizing lipid monolayers on a sensor surface. Measurements of lipid monolayer depletion were made using ellipsometry and quartz crystal microbalance. We have used this method to create a stabilized layer of GM1 ganglioside for the detection of a GM1 binding toxin, such as cholera, using QCM. Experimental Section Materials and Equipment. Gold-coated, AT-cut QCM crystals with a base resonant frequency of 10 MHz and an electrode area of 0.2 cm2 were purchased from ICM Co, Inc., Oklahoma City, OK. Silicon wafers were purchased from Wafer World, West Palm Beach, FL. Gold coating of the silicon wafers was done by Lance Goddard Associates, Foster City, CA. The silicon wafers were sputter coated on one side, first with 200 Å of chromium to aid the adhesion of the gold layer to the silicon and then with 1000 Å of gold. Reagent grade ethanol, concentrated sulfuric acid, hexane, and 30% hydrogen peroxide were purchased from VWR International. Hexadecanethiol (HDT), octanethiol (OT), cholera toxin B-chain (CTB), and phosphatidylcholine (PC) were purchased from Sigma-Aldrich. Bovine serum albumin (BSA) was purchased from Cal-biochem. GM1 was purified from the gray matter of bovine brain.23 Ellipsometry measurements were done using a Gaertner LSE ellipsometer at a 70° angle of incidence, utilizing a single (14) Meuse, C. W.; Krueger, S.; Majkrzak, C. F.; Dura, J. A.; Fu, J.; Connor, J. T.; Plant, A. L. Biophys. J. 1998, 74, 1388-1398. (15) Hepel, M. J. Electroanal. Chem. 2001, 509, 90-106. (16) Buoninsegni, F. T.; Becucci, L.; Moncelli, M. R.; Guidelli, R. J. Electroanal. Chem. 2001, 500, 395-407. (17) Vikholm, I.; Albers, W. M. Langmuir 1998, 14, 3865-3872. (18) Zhang, Y.; Telyatnikov, V.; Sathe, M.; Zeng, X.; Wang, P. G. J. Am. Chem. Soc. 2003, 125, 9292-9293. (19) Chang, H.; Yang, C.; Yeh, T. Anal. Chim. Acta 1997, 340, 4954. (20) Cremer, P. S.; Boxer, S. G. J. Phys. Chem. B 1999, 103, 25542559. (21) Ross, E. E.; Spratt, T.; Sanchao, L.; Rozanski, L. J.; O’Brien, D. F.; Saavedra, S. S. Langmuir 2003, 19, 1766-1774. (22) Rowe-Taitt, C. A.; Cras, J. J.; Patterson, C. H.; Golden, J. P.; Ligler, F. S. Anal. Biochem. 2000, 281, 123-133. (23) Ledeen, R. W.; Yu, R. K. Methods Enzymol. 1982, 83, 139-191.

Stine et al. wavelength light source from a helium-neon laser. Infrared spectroscopy measurements were done using a Thermo-Nicolet Nexus 670 FTIR in external reflectance mode with an MCT/A detector. QCM measurements were taken using an ICM 10 MHz lever oscillator and an Agilent digital multimeter and frequency counter. The flow cell used for QCM trials was also purchased from ICM. Film Deposition. PC and GM1 films were first deposited onto gold-coated silicon wafers for ease of surface characterization using ellipsometry and FTIR. Films were then deposited on goldcoated QCM crystals for film stability (PC films) and affinity sensing trials (GM1 films). Gold-coated silicon wafers were cleaned by submersion in 3:1 sulfuric acid/hydrogen peroxide solution for 20 min. They were then rinsed sequentially with deionized water and ethanol, sonicated in ethanol for 5 min, rinsed with ethanol again, and dried with a stream of nitrogen. After this, wafers were placed into a solution of 2 mM OT and 2 mM HDT in ethanol and allowed to react for 3 h at room temperature. To protect the electrode connections on a QCM crystal, the sulfuric acid/hydrogen peroxide solution was just swabbed onto the surface before sonicating. For self-assembled monolayer deposition, the crystals were placed into a specially designed Teflon holder that allowed the gold-coated surface to be exposed while isolating the electrode connections. For the addition of the lipid layer, a procedure described by Meuse and colleagues14 was used, with a few minor modifications. In short, the desired lipid was dissolved in hexane, and this solution was carefully added to a beaker containing water to create a two-phase liquid system. Upon evaporation of the hexane, a lipid film was left at the air/water interface with the hydrophobic fatty-acid tail of the lipid oriented toward the air. Finally, the HDT/OT-coated surface was brought into contact with the lipid film, creating a single lipid layer on top of the hydrophobic selfassembled monolayer. While no special measures were taken to ensure that a tightly packed lipid monolayer was formed, repeated deposition and subsequent ellipsometry analysis showed the addition of a reproducible lipid film with a thickness corresponding to values reported for lipid monolayer films in the literature.14,16 Films that are referred to as heat stabilized were then placed in a desiccated oven at 105 °C overnight. This temperature was chosen to be above the range of values that have been reported for achieving phase transition in selfassembled monolayers.24 PC was used in experiments to determine film stability, both to avoid using expensive GM1 during procedure development and to ensure that stabilization was not a result of heat-induced cross-linking between carbon-carbon double bonds in GM1. Once a method was found that would give sufficient film stability, GM1 was deposited for affinity sensing trials. Film Stability Trials. For ellipsometry tests, PC films were first deposited onto gold-coated silicon wafers, and their average film thickness was determined. The wafers were then extruded through an air/water interface and dried with nitrogen, and the average film thickness was again measured via ellipsometry. Film stability was also tested by depositing PC films on goldcoated QCM crystals and monitoring baseline frequency change as the surface was repeatedly rinsed with PBS buffer. Affinity Sensing Trials. For ellipsometry and FTIR characterization, a layer of GM1 was deposited onto a gold-coated silicon wafer using the above procedure and heat-stabilized overnight, and the surface was characterized by ellipsometry and FTIR. Next, the wafer was placed in a 10 µg/mL solution of CTB in PBS for 2 h at room temperature. After this time, the wafer was removed, rinsed with deionized water to remove nonbound surface protein, and again characterized by ellipsometry and FTIR. For QCM trials, a GM1 layer was deposited onto a gold-coated QCM crystal and heat stabilized overnight. After this time, the crystal was removed from the oven and placed into a 2000 µg/mL solution of BSA in PBS for 3 h at room temperature in order to block any exposed surface sites and prevent nonspecific protein adhesion. The crystal was then placed into the flow cell and allowed to reach a steady baseline frequency. For specificity tests, 100 µL of a 1000 µg/mL BSA solution was first added and (24) Self-assembled monolayers of thiols; Ulman, A., Ed.; Academic Press: New York, 1998.

Glycosphingolipid Films

Figure 1. Graph showing the changes in film thickness as measured via ellipsometry for lipid film stability trials. The prerinse data were taken after heat stabilization of the appropriate samples and before the initial rinse step. The heatstabilized lipid films remain at a relatively constant thickness through repeated rinsing, while the non-heat-stabilized lipid films were easily rinsed away. By the second rinse, the nonstabilized film thickness had returned to the value of the underlying self-assembled monolayer. The error bars represent standard deviations over five trials. allowed to sit at steady state for 5 min. Then, 20 cell volumes of PBS were flushed through the system to remove the BSA solution and any nonbound surface proteins, and the crystal was again allowed to reach a steady frequency. After that, 100 µL of 1000 µg/mL CTB solution was added to the cell and allowed to sit at steady state for 5 min. The time frame was chosen upon examination of the data, which showed that the frequency response of the crystal began to plateau in this range. After this time, another 20 cell volumes of PBS were flushed through the system, and the crystal was allowed to reach a steady frequency. Frequency response changes were measured as the difference in steady-state frequencies from one rinse step to the next, thus minimizing any viscosity-induced frequency changes caused by the presence of the more viscous protein solutions. For frequency response vs concentration data, 100 µL of the appropriate CTB solution was added to the flow cell and allowed to sit at steady state for 5 min. Then, 20 cell volumes of PBS were flushed through the system, and the crystal was allowed to reach a steady frequency. Frequency response changes were measured as the difference in steady baseline frequencies from one rinse step to the next.

Results and Discussion PC Film Stability. A mixed self-assembled monolayer of HDT and OT was chosen as the initial base for anchoring the lipid monolayer to the gold substrate surface. Brechling and colleagues13 found that a mixed monolayer leads to more complete surface coverage with fewer discontinuities than its homogeneous monolayer counterpart when forming supported lipid films. Figure 1 shows film layer thickness for the underlying self-assembled monolayer and subsequent PC layers both before and after rinsing. The initial lipid monolayer thickness of 2.14 nm is in reasonably good agreement with values that have been reported elsewhere.14,16 As Figure 1 shows, the thickness of the heat-stabilized PC monolayers remained relatively constant over several rinse steps, whereas the PC monolayers that had not been heated showed a significant decrease in thickness after a single rinse step and were almost completely removed in a second. Work done by Nakamura and colleagues25,26 has shown that alkanethiol self-assembled monolayers undergo a sort of phase transition above a critical temperature and that the film’s packing structure loosens as a result. With this in mind, we can hypothesize that the added stability of our lipid (25) Nakamura, T.; Aoki, K.; Chen, J. Electrochim. Acta 2002, 47, 2407-2411. (26) Nakamura, T.; Aoki, K.; Chen, J. Electrochem. Commun. 2002, 4, 521-526.

Langmuir, Vol. 20, No. 15, 2004 6503

films after heating is due to the intermingling of the fatty acid portions of the lipid molecules with the loosened alkane self-assembled monolayer. Upon cooling of the film back to room temperature, the monolayer recrystallizes, thereby helping to stabilize the lipid film on the substrate surface. We are not sure of the extent of the fatty-acid chain integration into the self-assembled monolayer. Initial ellipsometry tests showed no variation outside of the standard deviation of the results, which leads us to believe that the amount of fatty acid chain integration is only on the order of a few angstroms. To check stability over time, one of the heat-stabilized samples was allowed to soak in deionized water for 4 days at room temperature (data not included), at which point the film was again checked via ellipsometry. No loss of PC from the surface was detected. A PC film was also deposited onto a gold-coated QCM crystal and its detachment under shear stress monitored. A large initial drop in frequency occurred when PBS was initially introduced to the previously dry system. This was followed by a short equilibration period where the signal slowly rose to give a steady baseline. Admittedly, any PC loss that occurred during this time period could have been obscured by the equilibration of the system in the aqueous environment. However, we feel that the previously mentioned ellipsometry data as well as the affinity sensing data presented give reasonably good assurance that there is no significant loss of lipid from the film during the first introduction of aqueous media. The lipid-coated crystal was then exposed to four rinses, in which approximately 20 cell volumes of PBS were flushed through the system. We found that there was no discernible increase in the resonant frequency of the crystal, as would be expected if film detachment had occurred. A quick calculation of the difference between the initial steady baseline frequency and the baseline frequency after the final rinse showed a change of only 0.1 Hz, a value that is within the range of the random background fluctuations of the instrument. Hepel and colleagues15 have previously argued that supported lipid layers behave as rigid, solid films and may therefore be analyzed using the well-known Sauerbrey equation to relate frequency response to change in surface mass load. Using the Sauerbrey equation in this case, it was found that the 0.1 Hz change corresponds to a 450 pg/cm2 decrease in surface mass. On the basis of calculations by Petrache and colleagues,27 the total mass for a PC monolayer should be approximately 47 µg/cm2, meaning that the total loss of PC from the film during rinsing is a negligibly small percentage of the total. While it is not the intention of this work to show the validity of a direct correlation between frequency change and mass load in systems similar to ours (a topic which is still under much debate), we felt that this quick calculation could offer a good sense of scale for the amount of variation in our film under repeated stress. GM1 Film Characterization. Figure 2 shows ellipsometry data that was collected for a GM1 sample on a gold-coated silicon wafer and depicts the film thickness increase for each successive layer. For the addition of GM1 to the HDT/OT monolayer, an increase of 2.36 nm is observed, a value that agrees well with both our previous values for a lipid monolayer as well as values that have been published in the literature.14,16 After this layer of GM1 was exposed to a solution containing cholera toxin, another distinct increase of 2.63 nm was observed, a value (27) Petrache, H. I.; Dodd, S. W.; Brown, M. F. Biophys. J. 2000, 79, 3172-3192.

6504

Langmuir, Vol. 20, No. 15, 2004

Figure 2. Ellipsometry measurements showing the increase in film thickness at each deposition step. A distinct thickness increase is observed after the deposition of the GM1 lipid layer, followed by a second increase after the film is exposed to CTB. These increases offer evidence supporting the presence and proper orientation of the stabilized lipid film. The error bars represent standard deviations over seven trials.

that corresponds well with published values28,29 for the thickness of a protein monolayer. With these results in mind, we proceeded to test the samples using FTIR. Figure 3 shows the results for scans on both the GM1 and the GM1 plus CTB layers of the sample. The GM1 scan shows several peaks that support the presence of the expected layer, including an amide peak at 1638 cm-1, a carbonoxygen peak at 1265 cm-1, and a large peak at 1078 cm-1 that indicates the presence of oxygen containing rings such as those that would be found in the carbohydrate headgroup of GM1. For the CTB layer scan, several changes are observed that support the presence of a protein layer. There is a sharp increase in the intensity of the peak at 1265 cm-1 that can be attributed to the addition of carbon-nitrogen bonds, which are prevalent in proteins. The most interesting difference, though, is the significant increase of the amide peaks. There is the appearance of a substantial amide II peak at 1545 cm-1, which was only vaguely seen in the amide region for GM1. Also, the amide I peak has gained in intensity, as well as shifting from 1638 cm-1 in the GM1 scan to 1660 cm-1 in the CTB scan. The gain in amide intensity would be expected for the presence of a protein layer, since amide bonds are the

Stine et al.

predominant structural feature of proteins. Also, the shift in the amide I peak is indicative of a change from an alkyl amide, such as the one found in GM1, to a polypeptide amide, such as those that make up the primary structure of proteins. These FTIR data, along with the previously mentioned ellipsometry data, indicate not only that the GM1 layer is present on the sample surface but also that it has maintained the proper orientation to be effectively bound by cholera toxin at the surface/fluid interface. Affinity Sensing Trials. First, the GM1 coated QCM sensors were tested for specificity in their response to CTB over a control protein, BSA. Nonspecific protein adhesion can result in false-positive readings for many different types of biosensors, but it is a particular problem with QCM. This is due to the fact that QCM measures changes in the resonating frequency of the crystal. Therefore, any type of surface attachment, be it from a specific, desired interaction such as the one between GM1 and CTB or from the nonspecific adhesion of a different protein, will be counted and seen as a positive response.30,31 Hence, it is essential that the sensor surface is coated with a ligand that is only recognized by the compound or compounds of interest. BSA was used initially as a blocking agent on the GM1 coated sensor to protect any areas of the crystal surface that were not thoroughly covered by the GM1. BSA has been used for this purpose in many different techniques.32-34 We then further used BSA as the control protein for these experiments, both to test that the surface of the crystal was thoroughly protected against nonspecific adhesion and to determine whether additional BSA would aggregate and attach at sites where BSA was already present as a blocking agent. Figure 4 shows the results of these tests. A simple average of the frequency changes obtained from five trials shows that the response in the presence of CTB is approximately 2 orders of magnitude larger than the response seen for BSA. The error bars in Figure 4, which represent the standard deviation for the measurements, show that the response of the sensors to the addition of BSA did not differ

Figure 3. FTIR scans of the GM1 layer before (A) and after (B) exposure to CTB. The presence of the GM1 carbohydrate rings can be seen in A. The position shift and increase in the intensity of the amide peaks shown in B are indicative of the addition of a protein layer. These data offer further evidence of both the presence and proper orientation of the heat-stabilized lipid films.

Glycosphingolipid Films

Figure 4. Bar graph showing the frequency response upon exposure of GM1 modified QCM crystals to CTB and a BSA control. The response of the sensor elements to CTB is approximately 2 orders of magnitude greater than the response to the control. The error bars represent standard deviations over six trials.

Langmuir, Vol. 20, No. 15, 2004 6505

to the flow cell, while Figure 5B presents a plot of frequency response for various CTB concentrations. It can be seen that the lower detection limit of the sensor, defined here as the point where no signal greater than twice the magnitude of the random background fluctuation could be detected, was found to be approximately 0.5 µg/mL. This value compares well with detection limits found in previous studies in which GSLs were used in conjunction with SPR to detect biotoxins4 and is an order of magnitude lower than previous limits obtained using GSLs with QCM to detect protein toxins.11 Other researchers have achieved lowerdetectionlimitsusingfluorescence-basedmethods,7-9,35 but these methods are more complex and may not be suitable for fast, continuous monitoring in a constant flow system. Song and colleagues3 have achieved lower detection limits (50 pM or 3 ng/mL) using a fluorescence-based technique that does not require secondary labeling; however this technique is dependent on the GM1 film’s ability to move fluidly and aggregate at sites where CT is present, leading to fluorescence quenching. This dependence on a nonimmobilized lipid film could lead to GM1 loss from the surface and degradation over time for a constant monitoring system. While the lower detection limit of 0.5 µg/mL that we have shown may be inadequate for some toxin-sensing applications, this may be improved by using a higher frequency resonator. The lipid film immobilization method investigated here could be adapted for use with other signal transduction methods to achieve lower detection limits. The upper detection limit, or the point where viable surface binding sites become saturated, was not reached in our experiments, which still showed a good logarithmic agreement between toxin concentration and frequency response at the highest concentration tested (100 µg/mL). The exact location of the saturation plateau was not determined, since the lower detection limit is typically the area of interest in biotoxin sensing. Conclusions

Figure 5. (A) Graph showing the change in QCM frequency upon the addition of 100 µg/mL of CTB. (B) Graph showing the frequency response of GM1 modified QCM crystals as a function of CTB concentration. A logarithmic relationship between response and concentration is observed, with a lower detection limit at 0.5 µg/mL.

significantly from baseline resonant frequency. Negative standard deviation bars for the BSA adsorption experiments were due to small negative frequency changes that were a part of the intrinsic instrument background noise. These results confirmed that the sensing elements were recognized by CTB. Unlike previous studies in which GM1 was used to detect biotoxins via QCM,5,11 the results obtained in this work have addressed the critical issue of specificity, as well as the potential problem of stability of the GSL layer under shear stress. It can be seen that the response was constant after repetitive washing, there was no film detachment, and there was no significant nonspecific adhesion of the BSA. Results of experiments to determine the applicable sensing range of the QCM based sensors are shown in Figure 5. Figure 5A shows the decrease in absolute sensor frequency as a 100 µg/mL solution of CTB is introduced

We have produced a heat-stabilized lipid film that was found to be very stable under shear stress. These films may be of significant importance in a variety of biomaterial and biosensing applications and should be suitable for use with any native lipid molecule, without the need for specially engineered functional groups for cross-linked stabilization. Further, the data provided indicates that these films can be used to immobilize GSLs at a surface interface in a manner suitable for sensing applications. The carbohydrate headgroup was found to be oriented on the surface of a QCM crystal in a manner that made it available for binding by the B subunit of cholera toxin. The lower limit of detection was approximately 0.5 µg/mL of toxin, but this value could most likely be improved by using a detection platform that is more sensitive than QCM. Potential uses for these films include their use in improving biocompatibility of nonthrombogenic surfaces (28) Facci, P.; D. A.; Andolfi, L.; Schnyder, B.; Kotz, R. Surf. Sci. 2002, 504, 282-292. (29) Berzina, T. S.; V. I. T.; Petrigliano, A.; Alliata, D.; Tronin, A. Yu.; Nicolini, C. Thin Solid Films 1996, 284-285. (30) O’Sullivan, C. K.; Guilbault, G. G. Biosens. Bioelectron. 1999, 14, 663-670. (31) Marx, K. A. Biomacromolecules 2003, 4, 1099-1120. (32) Ben-Dov, I.; Willner, I.; Zisman, E. Anal. Chem. 1997, 69, 35063512. (33) Fung, Y. S.; Wong, Y. Y. Anal. Chem. 2001, 73, 5302-5309. (34) Disley, D. D.; Cullen, D. C.; You, H.; Lowe, C. R. Biosens. Bioelectron. 1998, 13, 1213-1225. (35) Alfonta, L.; Willner, I.; Throckmorton, D. J.; Singh, A. K. Anal. Chem. 2001, 73, 5287-5295.

6506

Langmuir, Vol. 20, No. 15, 2004

by preventing nonspecific protein adhesion, as well as for the detection of other biotoxins that bind to GSLs. Acknowledgment. This work was supported by the NIH (EB000684-01) and the Commonwealth of Pennsylvania. We also thank NASA for additional fellowship support. This project is also funded, in part, under a grant

Stine et al.

with the Pennsylvania Department of Health’s Health Research Formula Funding Program (State of PA, Act 2001-77-PA Tobacco Settlement Legislation). The Department specifically disclaims responsibility for any analyses, interpretations, or conclusions. LA049554W