Article pubs.acs.org/ac
Natural and Biomimetic Materials for the Detection of Insulin Romana Schirhagl,† Usman Latif,†,‡ Dagmar Podlipna,† Hans Blumenstock,§ and Franz L. Dickert*,† †
Department of Analytical Chemistry, University of Vienna, Waehringer Strasse 38, A-1090 Vienna, Austria Department of Chemistry, COMSATS Institute of Information Technology, Tobe Camp, University Road, 22060 Abbottabad, Pakistan § Sanofi Germany, Process Development Biotechnology, D-65926 Frankfurt, Germany ‡
ABSTRACT: Microgravimetric sensors have been developed for detection of insulin by using quartz crystal microbalances as transducers, in combination with sensitive layers. Natural antibodies as coatings were compared with biomimetic materials to fabricate mass-sensitive sensors. For this purpose polyurethane was surface imprinted by insulin, which acts as a synthetic receptor for reversible analyte inclusion. The sensor responses for insulin give a pronounced concentration dependence, with a detection limit down to 1 μg/mL and below. Selectivity studies reveal that these structured polymers lead to differentiation between insulin and glargine. Moreover, antibody replicae were generated by a double imprinting process. Thus, biological recognition capabilities of immunoglobulins are transferred to synthetic polymers. In the first step, natural-immunoglobulinimprinted nanoparticles were synthesized. Subsequently, these templated particles were utilized for creating positive images of natural antibodies on polymer layers. These synthetic coatings, which are more robust than natural analogues, can be produced in large amount. These biomimetic sensors are useful in the biotechnology of insulin monitoring.
I
during polymerization and after removal the template leaves behind a highly patterned sensitive material for reversible and selective inclusion of analytes. In this paper, different sensing materials are synthesized and utilized in combination with mass sensitive devices13 to design a fast and label free detection system. At first natural antibodies are used to compare these results with measurements obtained by biomimetic receptors. Natural antibodies are immobilized on a gold electrode surface of dual channel QCM for the detection of insulin. The difference in responses between both electrodes is used to observe the immobilization and, subsequently, interaction of antibodies with respective analyte. Natural antibodies being highly selective toward their respective antigens are widely used as sensitive coatings for the detection of different analytes.14−16 However; these antibodies are expensive, have limited lifetime, and are practicable in biocompatible measuring conditions only. To overcome these limitations, we moved to new biomimetic receptors for
nsulin hormone regulates carbohydrate and fat metabolism in the body and its disorder causing severe illness and dysfunctions in humans named as diabetes mellitus. In 2000, about 2.8% of the adult population was suffering from diabetes; however, an increase to 4.4% is anticipated until 2030.1 Furthermore, diabetes patients are reliant on intake of adequate and accurate amount of insulin. Additionally, insulin is also used as a doping drug in sports as a way of cheating. Papers can be found for the detection of insulin by using different analytical approaches such as immunoaffinity chromatography with subsequent LC/MS analysis,2 immunoassays,3 electrochemical,4 amperometric,5 surface plasmon resonance,6 and mass-sensitive biosensors7 to serve for this purpose. To design a biomimetic sensor device for the detection of insulin at low levels is no easy task. However, molecular imprinting turns out to be a promising tool for synthesizing a sensitive layer that in combination with a suitable transducer leads to fabrication of a sensor device for the detection of molecules,8,9 viruses,10 or entire cells.11 The imprinted materials12 can be synthesized by combining model compound (template) with functional monomers. The template may be the target analyte or any appropriate compound to structure the sensitive layer. The monomers engulf the template by a self-organization process © 2012 American Chemical Society
Received: June 30, 2011 Accepted: April 2, 2012 Published: April 2, 2012 3908
dx.doi.org/10.1021/ac201687b | Anal. Chem. 2012, 84, 3908−3913
Analytical Chemistry
Article
bioanalytes, such as directly imprinted polymers and antibody replicae17 for the detection of insulin. Directly imprinted polymers are synthesized by surface imprinting the polyurethane polymer with analyte itself and concentration dependent frequency shifts are obtained with exposure to analyte solutions. Moreover, selectivity studies reveal the successful imprinting effects as well as advantages of microgravimetric sensors over previous mass-sensitive biosensors that pronounced sensor responses are observed toward templated analytes. Antibody replicae18 are generated by utilizing a double imprinting approach. The implementation of such sensitive layers on gravimetric sensors enables us to develop robust, fast, and labelfree detection systems. These artificial immunoglobulins can analyze different concentrations of insulin, which ensures the successful transfer of biological information to plastics. These synthetic receptors make possible monitoring insulin in biotechnology due to their robustness. Furthermore, cross sensitivities can be analyzed to modify insulin and protein contaminations.
■
EXPERIMENTAL SECTION All chemicals were purchased from Sigma Aldrich, Merck, Fluka, or VWR in the highest available purity. Natural Antibodies. Biosensors based on natural antibodies are fabricated by immobilizing the antibodies on the gold electrode surface of a quartz crystal microbalance. For this purpose, guinea pig polyclonal anti-insulin antibodies (Peninsula Laboratories INC.) were injected into the measuring cell and adsorbed onto the gold surface via sulfur moieties, which was confirmed by a decrease in frequency of the mass-sensitive sensor. These antibodies can be used directly for the detection of insulin; the schematic design of these interactions is shown in Figure 1a. Directly Imprinted Polymers. An artificial receptor was generated by surface imprinting the polyurethane layer with insulin crystals. For this purpose, 42.2 mg of bisphenol A, 15.6 mg of phloroglucinol, and 42.2 mg of 1,1-diisocyanatodiphenylmethane are dissolved in 100 μL of tetrahydrofuran (THF). Afterward, 15 μL of pyridine was added to the solution to start the polymerization, which takes almost 15 min at 70 °C. The end point can be confirmed by taking IR spectra of the isocyanate, giving the most prominent band at 2263 cm−1, which will diminish quickly near the gel point. The resulting polymer was diluted with THF in 1:8 ratios and used for spin coating on the transducer surface. Then, stamps for surface imprinting were prepared by dropping 10 μL of an insulin suspension on 8 × 8 mm2 glass plates. Subsequently, 5 μL of diluted polymer was spin-coated (2000 rpm) on one of the electrodes of QCM and the dried stamps were pressed onto the polymer with a clamp, whereas another electrode was coated with nonimprinted polymer, serving as a reference to compensate for physical effects. The polymer is allowed to stay for 48 h to ensure complete polycondensation; then, the stamp was removed from the imprinted layer by washing with 0.5 M HCl/0.1% H2O2 for 1 h. The successful pattern results from insulin crystals, which was confirmed by atomic force microscopy (AFM) with Veeco/digital instruments Nanoscope IVa in contact mode. The resulting sensor is integrated into the measuring cell followed by equilibration in solvent and injection of the analytes. The principle of synthesizing a molecularly imprinted polymer is shown in Figure 1b. Antibody Replicae. Another strategy was developed to create artificial antibodies for the detection of insulin (shown in
Figure 1. Different detection methods: (a) Natural antibodies are immobilized on the gold surface of a QCM via sulfur groups. (b) Holes are created by molecular imprinting which selectively incorporates their template. (c) Positive structures of natural antibodies are synthesized by following the double imprinting process.
Figure 1c). For this purpose, 60 mg of dihydroxyethylenebisacrylamide (DHEBA), 20 mg of vinylpyrrolidone, and 50 mg of methacrylic acid were dissolved in 800 μL of distilled water. After dissolving the cross-linker (DHEBA) at 70 °C, the pH of the prepolymer solution was adjusted to 7 with KOH to prevent the antibody from denaturating. Finally, 1.5 mg of sodium peroxydisulfate was added to initiate the radical polymerization. Subsequently, 0.2 mg of natural immunoglobulin (guinea pig polyclonal anti-insulin) was added to 0.4 mL of the monomer mixture and the solution was prepolymerized 3909
dx.doi.org/10.1021/ac201687b | Anal. Chem. 2012, 84, 3908−3913
Analytical Chemistry
Article
under UV light for 15 min at room temperature. Afterward, antibody-imprinted nanoparticles were precipitated by adding prepolymerized solution into acetonitrile (20 mL/1 L) under vigorous stirring condition. The nonimprinted reference nanoparticles were generated in the same way but without adding natural antibodies. The monomer mixture is miscible in acetonitrile, whereas, with the generation of distinct size oligomers during polymerization, solubility decreases and they precipitated out together with natural antibodies, in the form of nanoparticles. Large sized particles can be produced by increasing either the prepolymerization time or amount of prepolymerized solution per milliliter of acetonitrile. The suspension is stirred overnight followed by centrifugation and the supernatant liquid is discarded. The templated particles were sonicated with water followed by centrifugation. This procedure was repeated twice to ensure complete removal of antibodies. After washing, the nanoparticles were suspended in 2 mL of water. The presence or absence of natural antibodies in templated nanoparticles can be verified by UV spectroscopy or so-called xanthoprotein reaction, which reveals that aromatic moieties of proteins form a yellow dye product after their reaction with concentrated nitric acid. A 20 μL volume of the particle suspension is dropped onto 5 × 5 mm2 glass plates which are used as stamps for surface imprinting, after drying. The polymer consists of the same monomers as the particles (solely with 30 mg of DHEBA instead of 60 mg and no pH adjustment). The monomer solution was polymerized at 70 °C for 30 min and diluted 1:2 with THF before coating. For the imprinting process, 5 μL of the prepolymerized solution was spin-coated (3000 rpm) on a QCM electrode and a stamp with printed particles was pressed into the polymer. Afterward, it is allowed to dry overnight at room temperature, and removal of the stamp from the polymer layer leaves behind a positive structure of natural antibodies, named as antibody replicae. We mounted the resulting sensor on the measuring cell and used it for the detection of insulin. Measuring Setup. The mass-sensitive measurements were performed by dual-electrode quartz crystal microbalances of fundamental frequency 10 MHz, with a diameter 15.5 mm and a thickness of 0.168 mm.19 The dual-electrode pattern was screen printed on a QCM with gold paste (GGP 2093126 W. C. Heraeus). The QCM was mounted on a measurement cell having a maximum capacity of 140 μL analyte volume. Sensor responses are obtained by coating one electrode with a sensitive layer (natural antibodies, directly imprinted polymers, antibody replicae), while the second electrode served as a reference channel for differential measurements. The serial resonance of the 10 MHz QCMs was monitored via a homemade dual twostage amplifier. Data acquisition and transfer was performed by LabVIEW via HP 53131A frequency counter.
Figure 2. Natural antibodies are immobilized on the gold surface of a QCM for the detection of their antigens (insulin). Immobilization of antibodies corresponds to frequency decrease, indicating that antibodies are now anchored to the surface via sulfur groups. When the frequency is constant, the sensor is used to detect insulin.
sulfur containing functional groups within proteins and the gold surface of the electrode. These antibodies become immobilized on the transducer surface and can be used for sensing purposes without the risk of washing them off (Figure 1a). A frequency shift of 500 Hz corresponds to an effective thickness of 20 nm of immunoglobulin antibodies (1 kHz ≈ 40 nm),20 when these are adsorbed on the gold surface of the electrode. Subsequently, an analyte solution of 0.25 mg/mL insulin was injected into the measuring cell, resulting in a frequency shift, which indicates binding of insulin with natural antibodies, anchored on transducer surface. The sensor was regenerated by flushing with 2 M guanidine hydrochloride for 45 s. This substance is known to be chaotropic, being able to break the hydrogen bond between antibody−antigen complexes.21,22 The immunoglobulin can be completely removed from the transducer surface by washing with 2 M guanidine hydrochloride for 15 min. Measurements with Directly Imprinted Polymers. By utilizing the molecular imprinting approach, we have created artificial receptors by structuring a polymer surface with the respective analyte, for the detection of insulin. Instead of strong covalent bonds, various noncovalent interactions such as hydrogen bonds, hydrophobic interactions, dipole interactions, or π/π interactions are favoring the reversibility of the sensor response. Consequently, an appropriate polymer should contain the binding counterparts present in the analyte being able to form an exact structure of the template, which is more or less fixed by the cross-linker.23,24 In our case, a polyurethane layer consisting of bisphenol A, phloroglucinol, and 1,1diisocyanato-diphenylmethane turned out to be a feasible recognition material for proteins. Figure 3 shows AFM images of the insulin-imprinted polyurethane surfaces. The pattern produced on the layer is actually obtained by surface imprinting with insulin crystals. In this way, receptor sites are generated on these sensitive layers for reversible inclusion of insulin molecules. The formation of molecular cavities cannot directly be detected by AFM according to a limited resolution of this method. Thus, we can say that, the pattern obtained reflects the shape of insulin crystals. The proof of a successful imprinting process is a sensor response via selective analyte recognition, especially comparing imprinted and nonimprinted coating. In addition, pH dependency of insulin inclusion has to be analyzed for exact measuring conditions. Figure 4 shows this optimization concerning pH, which directly relates to the solubility of insulin. Insulin has its iso-electrical point at a pH of 5.4 and is more soluble at lower pH values. It can be concluded from the graph that the optimal sensitivity is between pH 4.5 and 7, with much higher or lower pH values causing denaturing.
■
RESULTS AND DISCUSSION Measurements with Natural Antibodies. Each immunoglobulin-G molecule consists of four peptide chains2 light and 2 heavy onesand possesses two binding sites. The recognition sites are actually the paratopes that is specific for particular epitopes on target antigens. These antibodies are mounted on the gold surface of a QCM via their sulfur moieties for the detection of respective antigens. Figure 2 shows the immobilization of immunoglobulins to the surface, which can be verified by observing a decrease in the frequency of the mass-sensitive sensor. The natural antibodies anchor to bare electrode due to formation of covalent linkage between the 3910
dx.doi.org/10.1021/ac201687b | Anal. Chem. 2012, 84, 3908−3913
Analytical Chemistry
Article
Figure 5. Measurement of insulin with directly imprinted polyurethane layer. PBS is injected alternately with different concentrated insulin solutions (1−1000 μg/mL).
trations, 200−1000 μg/mL of insulin, a curvature is observed, which indicates some saturation tendency. The noise level of some hertz in Figure 5 can drastically be reduced by analogue mixing between measuring and reference channels. In this way an instantaneous fluctuation of frequency due to temperature can be compensated for, which leads perhaps to an enhanced sensitivity by a factor of 10 (0.1 μm/mL).7 A 100 μg/mL insulin sample leads to a sensor response of 159 Hz, which commensurates to a layer height of about 5−6 nm (1 kHz corresponds to 40 nm). A monolayer of insulin heaving a molar mass of 5800 g/mol would result in a layer height of approximately 2 nm. Thus, a complete coverage of the sensor should occur at this concentration of 100 μg/mL. No further sensor response should be observed at higher insulin concentrations due to saturation effects. An increasing sensor response at higher concentrations according to Figure 5 can be explained, however, by the formation of insulin hexamers.25 A monolayer of insulin−hexamers leads to a layer height that approximately corresponds to the observed sensor effect. At high concentrations the sensor effects will increase due to an extended insulin aggregation, which is favored by a reduced insulin solubility. The quality of the sensor can be accessed by selectivity studies. For this purpose, the insulin-imprinted sensor was exposed to equal concentrations (2 mg/mL) of insulin, lysozyme, trypsin, and pepsin as shown in Figure 6, since enzyme contaminations must be considered during the biotechnological process. The measurements ensure that
Figure 3. AFM image of insulin (crystal)-imprinted surface. After removing the analyte, recognition sites will be generated which are geometrically and chemically fit to the insulin.
Figure 4. pH dependency of the sensor response determined by exposing 0.5 mg/mL of insulin.
The pH range from 4.5−6.5 (Figure 4) leads to a reduced solubility in comparison to more acid condition. In this way crystallization is favored according to insulin aggregation. This phenomenon causes a higher mass-sensitive effect in comparison to a complete insulin monolayer. Different insulin concentrations were exposed to printed polyurethane, as shown in Figure 5, illustrating sensor characteristics in the form of concentration-dependent frequency shifts. In the concentration range from 1 to 100 μg/mL of insulin a nearly linear characteristic is obtained, whereas at higher concen-
Figure 6. Insulin-imprinted polyurethane layer was exposed to equal concentrations (2 mg/mL) of insulin, lysozyme, trypsin, and pepsin. Pronounced sensor response was achieved for the templated insulin in comparison to other analytes. 3911
dx.doi.org/10.1021/ac201687b | Anal. Chem. 2012, 84, 3908−3913
Analytical Chemistry
Article
replicae as the sensitive layer, as shown in Figure 8a. During mass-sensitive measurements, distilled water of pH 7 was used
successful patterning was achieved by surface imprinting with insulin crystals, which in turn produce chemically and sterically fit cavities for the reinclusion of insulin. Pronounced sensor response was achieved for templated analyte, whereas a minor cross sensitivity was observed for lysozyme and the sensor remains almost insensitive to trypsin and pepsin. Considering the differential response between printed and nonimprinted QCM electrodes no cross sensitivity at all was observed in the case of trypsin and pepsin. Cross-sensitivity studies to modified insulin were carried out by surface imprinting the polyurethane layer with insulin and glargine, respectively (structural differences between insulin and glargine: glycine is on position A21 and two arginines are on the C terminus). Subsequently, these imprinted layers were exposed to templated analytes, as shown in Figure 7. These investigations reveal that each sensor shows
Figure 8. Detection of insulin with artificial immunoglobulin: The sensor was regenerated by washing with 0.5 M HCl with 0.1% H2O2 (measurement paused) and later on with water: (a) time behavior and (b) characteristic.
to equilibrate the sensor, followed by the injection of insulin. Insulin sensor was regenerated after flushing with 0.5 M HCl containing 0.1% H2O2. In this case, HCl makes insulin more soluble and H2O2 denaturizes the protein by oxidation. In comparison to the directly printed layer, the measuring effect observed with antibody replica is 3.25 times higher and turns out to be 1.5 times higher compared with the natural antibody. This effect is due to the larger surface area and thus the higher imprinting density caused by the use of nanoparticles. In Figure 8b the concentration dependence of the sensor effect is shown. Again, in the beginning of the characteristic a linear behavior is found followed by some saturation effects.
Figure 7. Cross-selectivity studies were carried out while exposing the insulin- and glargine-imprinted polyurethane polymers to their templated analytes.
the highest immune response to its target analyte and a lower frequency shift was observed for the cross-reactive substances. Surface-imprinted acrylate polymers were also designed for the detection of insulin but these sensor materials give a response of only 30% toward insulin. Obviously, polyurethanes with polar and apolar moieties lead to an optimized interaction between layer and analyte. The sensitivity of the layers can be enhanced by applying the prepolymer to the electrodes by drop coating. In this way the roughness is enhanced, which will give more interaction sites which lead to an improvement of sensitivity by a factor of 2. This procedure, however, needs some cleverness to avoid noise due to the formation of an inhomogeneous layer. Measurements with Antibody Replicae. Another strategy was followed for generating recognition sites in the sensitive layer by casting the geometrical and chemical structure of natural antibodies into polymer. For this purpose, nanoparticles were precipitated out along with immunoglobulin-G and subsequently, removal of antibodies leaves behind imprinted particles. Afterward, these templated particles were utilized for structuring the polymer layer by surface imprinting to produce the plastic antibodies, whose interacting capabilities toward respective analytes are similar to natural ones. The small sizes of the nanoparticles make them accessible for the solvent and enable us to separate them from the printed polymer easily. Additionally, the large surface area of particles leads to a relatively increased imprinting density and, thus, higher measuring effects. Concentration-dependent frequency shifts were obtained by injecting 0.5 mg and 0.25 mg/mL of insulin into the measuring cell, respectively; containing antibody
■
CONCLUSIONS The molecular imprinting technique enables us to synthesize cost-effective and robust biomimetic materials. The combination of these nanostructured sensitive surfaces with QCMs leads to fast, label free, and most notably regenerable detection systems for insulin. Further sensitivity enhancement can be obtained (Figure 5) by two or three orders of magnitude in going to high-frequency SAWs.26,27 Selectivity studies ensure the successful imprinting effects and produce pronounced sensor effects toward templated analytes. The interaction behavior of antibody replicae is more pronounced than that of other sensitive layers due to enhanced imprinting density. The recognition strategy of antibody replicae is totally different from that of directly imprinted polymers in such a way that replicae bind to epitopes of antigens, just like their natural counter parts, whereas imprinted polymers interact with the whole surface of analytes. Additionally, the direct use of the analyte can be circumvented, while detecting either dangerous or unstable analytes (susceptible to be destroyed or denatured in the direct imprinting process). Moreover, it is an advantage that a template is required for the first time during the synthesis of the particle stamp and afterward, a large number of sensitive layers can be generated. In other words, the use of a doubleimprinting process allows us to create robust polymers: 3912
dx.doi.org/10.1021/ac201687b | Anal. Chem. 2012, 84, 3908−3913
Analytical Chemistry
Article
(25) Lougheed, W.; Woulfe-Flanagan, H.; Clement, J.; Albisser, A. Diabetologia 1980, 19, 1−9. (26) Dickert, F. L.; Forth, P.; Bulst, W. A.; Fischerauer, G.; Knauer, U. Sens. Actuators, B 1998, 46, 120−125. (27) Mujahid, A; Afzala, A.; Glanzing, A.; Leidl, A.; Lieberzeit, P. A.; Dickert, F. L. Anal. Chim. Acta 2010, 675, 53−57.
promising materials to work in harsh conditions. The strategy can simply be modified to detect other analytes by changing the starting antibody to generate biomimetic receptors for target antigens.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We would like to acknowledge Higher Education Commission (HEC) of Pakistan for providing funding for PhD studies to Usman Latif.
■
REFERENCES
(1) Mytas, D. Z.; Stougiannos, P. N.; Zairis, M. N.; Foussas, S. G.; Pyrgakis, V. N.; Kyriazis, I. A. J. Diabetes Complications 2009, 23, 273− 282. (2) Thomas, A.; Thevis, M.; Delahaut, P.; Bosseloir, A.; Schänzer, W. Anal. Chem. 2007, 79, 2518−2524. (3) Neal, J.; Han, W. Endocr. Pract. 2008, 14, 1006−1010. (4) Wang, Y.; Li, J. Anal. Chim. Acta 2009, 650, 49−53. (5) Arvinte, A.; Westermann, A. C.; Sesay, A. M.; Virtanen, V. Sens. Actuators, B 2010, 150, 756−763. (6) Frasconi, M.; Tortolini, C.; Botrè, F.; Mazzei, F. Anal. Chem. 2010, 82, 7335−7342. (7) Lieberzeit, P. A.; Afzal, A.; Podlipna, D.; Krassnig, S.; Blumenstock, H.; Dickert, F. L. Sens. Actuators, B 2007, 126, 153−158. (8) Nopper, D.; Lammershop, O.; Wulff, G.; Gauglitz, G. Anal. Bioanal. Chem. 2003, 377, 608−613. (9) Latif, U.; Rohrer, A.; Lieberzeit, P.; Dickert, F. Anal. Bioanal. Chem. 2011, 400, 2457−2462. (10) Jenik, M.; Schirhagl, R.; Schirk, C.; Hayden, O.; Lieberzeit, P.; Blaas, D.; Paul, G.; Dickert, F. L. Anal. Chem. 2009, 81, 5320−5326. (11) Hayden, O.; Mann, K.-J.; Krassnig, S.; Dickert, F. L. Angew. Chem., Int. Ed. 2006, 45, 2626−2629. (12) Qu, P.; Lei, J.; Zhang, L.; Ouyang, R.; Ju, H. J. Chromatogr., A 2010, 1217, 6115−6121. (13) Jung, Lu, C.; Jin, C.; Zellers, E. T. J. Environ. Monit. 2006, 8, 270−278. (14) Matschulat, D.; Deng, A.; Niessner, R.; Knopp, D. Analyst 2005, 130, 1078−1086. (15) Brenn-Struckhofova, Z.; Füreder, C.; Cichna-Markl, M.; RazzaziFazeli, E. J. Chromatogr., A 2009, 1216, 5828−5837. (16) Fodey, T.; Leonard, P.; O’Mahony, J.; O’Kennedy, R.; Danaher, M. TrAC, Trends Anal. Chem. 2011, 30, 254−269. (17) Hoshino, Y.; Kodama, T.; Okahata, Y.; Shea, K. J. J. Am. Chem. Soc. 2008, 130, 15242−15243. (18) Schirhagl, R.; Podlipna, D.; Lieberzeit, P. A.; Dickert, F. L. Chem. Commun. 2010, 46, 3128−3130. (19) (a) Hayden, O.; Bindeus, R.; Haderspöck, C.; Mann, K.-J.; Wirl, B.; Dickert, F. L. Sens. Actuators, B 2003, 91, 316−319. (b) Schirhagl, R.; Seifner, A.; Husain, F. T.; Cichna-Markl, M.; Lieberzeit, P. A.; Dickert, F. L. Sensor Lett. 2010, 8, 399−404. (20) Dickert, F. L.; Lieberzeit, P. A.; Achatz, P.; Palfinger, C.; Fassnauer, M.; Schmid, E.; Werther, W.; Horner, G. Analyst 2004, 129, 432−437. (21) Dandliker, W. B.; Alonso, R.; de Saussure, V. A.; Kierszenbaum, F.; Levison, S. A.; Schapiro, H. C. Biochemistry 1967, 6, 1460−1467. (22) Banerjee, D.; Verma, P. K.; Pal, S. K. Photochem. Photobiol. Sci. 2009, 8, 1441−1447. (23) Vidyasankar, S.; Arnold, F. H. Curr. Opin. Biotechnol. 1995, 6, 218−224. (24) Ye, L.; Mosbach, K. Chem. Mater. 2008, 20, 859−868. 3913
dx.doi.org/10.1021/ac201687b | Anal. Chem. 2012, 84, 3908−3913
