Langmuir 2007, 23, 8485-8490
8485
Specific Binding of Large Aggregates of Amphiphilic Molecules to the Respective Antibodies Alexei Nabok,*,† Anna Tsargorodskaya,† Alan Holloway,† Nikolay F. Starodub,‡ and Anna Demchenko‡ Sheffield Hallam UniVersity, Materials and Engineering Research Institute, City Campus, Pond Street, Sheffield S1 1WB, U.K., and Palladin Institute of Biochemistry, National Academy of Sciences of the Ukraine, KieV 02030, Ukraine ReceiVed February 12, 2007. In Final Form: April 17, 2007 The Binding of nonylphenol to respective antibodies immobilized on solid substrates was studied with the methods of total internal reflection ellipsometry (TIRE) and QCM (quartz crystal microbalance) impedance spectroscopy. The binding reaction was proved to be highly specific having an association constant of KA ) 1.6 × 106 mol-1 L and resulted in an increase in both the adsorbed layer thickness of 23 nm and the added mass of 18.3 µg/cm2 at saturation. The obtained responses of both TIRE and QCM methods are substantially higher than anticipated for the immune binding of single molecules of nonylphenol. The mechanism of binding of large aggregates of nonylphenol was suggested instead. Modeling of the micelle of amphiphilic nonylphenol molecules in aqueous solutions yielded a micelle size of about 38 nm. The mechanism of binding of large molecular aggregates to respective antibodies can be extended to other hydrophobic low-molecular-weight toxins such as T-2 mycotoxin. The formation of large molecular aggregates of nonylphenol and T-2 mycotoxin molecules on the surface was proved by the AFM study.
Introduction The registration of toxins in the environment produced either naturally (biotoxins) or as a result of industrial and agricultural activities has become a very important subject in recent times. Some of the toxins, for example, T-2 mycotoxins, because of their high toxicity and relatively easy method of synthesis in large amounts, are considered to be potential agents for bioterrorism.1,2 Another analyte of interest in this work is alkylphenols, which appear in water resources as a result of the biodegradation of commercial alkylphenol ethoxylate surfactants.3 These compounds, apart from being known as toxic and carcinogenic, possess a serious threat to living organisms because of their estrogen-mimicking behavior.4,5 The immune assay approach is very suitable for the registration of the above toxins; it was implemented in immune sensors of T-2 mycotoxin2 and nonylphenol6 and showed high sensitivity and selectivity. However, because of the low molecular weight of the above toxins, the registration of their binding to respective antibodies immobilized on the surface using the conventional optical technique of surface plasmon resonance (SPR) is quite difficult. The more sophisticated and expensive competitive immune reaction technique was used as an alternative to achieve sensitivity in parts per billion range.7,8 The method of total internal * To whom correspondence should be addressed. E-mail:
[email protected]. Phone: +44 114 2253512. Fax: +44 114 2253433. † Sheffield Hallam University. ‡ National Academy of Sciences of the Ukraine. (1) Mirocha, C. J.; Pawlosky, R. A.; Chatterjee, K.; Watson, S.; Hayes, W. J. Assoc. Anal. Chem. 1983, 66, 1485-1499. (2) Kononenko, G. P.; Burkin, A. A.; Soboleva, N. A.; Zotova, E. V. Appl. Biochem. Microbiol. 1999, 35, 457-462. (3) Warhurst, A. M. An EnVironmental Assessment of Alkylphenol Ethoxylates and Alkylphenols; Friends of the Earth: London, 1995. (4) Olsen, C. M.; Meussen-Elholm, E. T. M.; Hongslo, J. K.; Stenersen, J.; Tollefsen, K. E. Comp. Biochem. Physiol., Part C 2005, 141, 267-274. (5) Waring, R. H.; Harris, R. M. Mol. Cell. Endocrinol. 2005, 244, 2-9. (6) Rose, A.; Nistor, C.; Emne´us, J.; Pfeiffer, D.; Wollenberger, U. Biosens. Bioelectron. 2002, 17, 1033-1043. (7) Shankaran, D. R.; Gobi, K. V.; Sakai, T.; Matsumoto, K.; Toko, K.; Miura, N. Biosens. Bioelectron. 2005, 20, 1750-1756. (8) Gobi, K. V.; Miura, N. Sens. Actuators, B 2004, 103, 265-271.
reflection ellipsometry (TIRE) offers much better sensitivity as compared to SPR,9-14 and it was recently successfully implemented for T-2 mycotoxin detection.14 This work is a continuation of our previous research in the development of novel methods of toxin registration. As it was shown earlier,14 the presence of molecules of T-2 mycotoxin at very low concentration down to 0.15 ng/mL was detected using the direct immune assay approach in conjunction with the novel and very sensitive TIRE technique. The reaction of binding T-2 molecules to respective antibodies was found to be very specific, having an association constant in the range of 106-107 mol-1‚L. However, the obtained increase in the thickness of the absorbed layer of 4.5 nm (at saturation) was much larger than anticipated for binding relatively small molecules of T-2 mycotoxin (M ) 466.5). A complementary technique of QCM impedance spectroscopy has also shown an anomalously high added mass of 3.8 µg/cm2 for a saturated adsorbed layer of T-2 molecules. These facts led us to suggest the mechanism of specific binding of larger aggregates of hydrophobic T-2 molecules formed in aqueous solutions. In this work, we applied the same methodology to the registration of nonylphenol, which belong to the family of alkylphenols. Nonylphenol is known to be a carcinogenic and estrogen-mimicking compound5 and is thus of great importance to environmental control, particularly in natural water resources. In addition to the TIRE and QCM methods, the atomic force microscopy (AFM) was used in this work to directly observe the changes in the morphology of adsorbed layers due to the binding of nonylphenol and T-2 mycotoxin molecules. (9) Westphal, P.; Bornmann, A. Sens. Actuators, B 2002, 84, 278-282. (10) Poksinski, M.; Arwin, H. Thin Solid Films 2004, 455-456, 716-721. (11) Arwin, H.; Poksinski, M.; Iohansen, K. Appl. Opt. 2004, 43, 3028-3036. (12) Nabok, A. V.; Tsargorodskaya, A.; Hassan, A. K.; Starodub, N. F. Appl. Surf. Sci. 2005, 246, 381-386. (13) Nabok, A. V.; Tsargorodskaya, A.; Holloway, A.; Starodub, N. F.; Demchenko, A.; Gojster, O. Proc. IEEE Sens. 2004, 1-3, 1195-1198. (14) Nabok, A. V.; Tsargorodskaya, A.; Holloway, A.; Starodub, N. F.; Gojster, O. Biosens. Bioelectron. 2007, 22, 885-890.
10.1021/la700414z CCC: $37.00 © 2007 American Chemical Society Published on Web 07/06/2007
8486 Langmuir, Vol. 23, No. 16, 2007
Nabok et al.
Experimental Details Sample Preparation. Substrates for TIRE measurements were prepared by the consecutive thermal evaporation of 3-5 nm of chromium and 25-30 nm of gold on clean glass microscope slides without breaking the vacuum of 10-6 Torr. Substrates for QCM and AFM measurements were, respectively, AT-cut quartz crystals with golden electrodes (EuroQuartz, 18.434 MHz) and pieces of silicon wafers. The following procedure described in ref 14 was used for the adsorption of biochemicals on the above substrates: (i) A layer of (poly)allylamine hydrochloride (PAH) from SigmaAldrich was electrostatically adsorbed on the surface of Au from its 1 mg/mL solution in Millipore water. The samples were kept in the above solution for 15 min and then rinsed several times in Millipore water. (ii) A layer of protein A from Staphylococcus aureus (SigmaAldrich) was electrostatically adsorbed on a PAH layer from its 0.02 mg/mL solution in 10 mM trizma base/HCl buffer (pH 7.5) using the original trizma-base buffer from Sigma-Aldrich. (iii) A layer of antibodies to nonylphenol (nonylphenol antiserum was acquired from the Institute of Biochemistry, Kiev, Ukraine) was deposited on top of the protein A layer. Protein A was used as an intermediate agent for the orientation of antibodies with their biding sites toward the solution, as was described earlier in ref 15. Initial antibody solutions were diluted 100 times in 10 mM trizma base/HCl buffer (pH 7.5) for further immobilization. The typical incubation time was 15 min. (iv) The coatings were then exposed to aqueous solutions containing different concentrations of nonylphenol (acquired from the Institute of Biochemistry, Ukraine). The initial 600 µg/mL nonylphenol solution in acetonitryl was multiply diluted with 10 mM trizma base/HCl buffer (pH 7.5) to obtain different concentrations from 1.2 to 1000 ng/mL. Deionized water produced by the Millipore Super Q system and having a resistance of no less than 18 MΩ was used for the preparation of all aqueous solutions. To wash out nonspecifically bound nonylphenol molecules, samples were washed either with a 30% methanol/water mixture or acetonitryl and then were placed in buffer solution for TIRE measurements (or Millipore water for QCM and AFM measurements). The same routine was applied to the samples for AFM and QCM experiments, with the only difference being that samples were immersed in the respective solutions instead of using solution injection in TIRE experiments. The samples were thoroughly dried in a flow of nitrogen gas prior to QCM and AFM measurements. A number of samples were prepared for the AFM study of T-2 mycotoxin binding following a similar routine described in detail.14 Antibodies to T-2 mycotoxin (Sigma-Aldrich, monoclonal anti-T2 toxin, clone T2-50) were immobilized on the surface of chromium/ gold-coated glass slides via a layer of PAH and protein A. Then T-2 mycotoxin molecules were specifically bound from the 600 ng/mL solution in trizma base/HCl buffer followed by washing out nonspecifically bound T-2 molecules in 30% methanol, rinsing in Millipore water, and drying with nitrogen gas. Experimental Techniques and Measurement Routine. The TIRE experimental setup, described in detail in refs 12-14, represents the combination of a commercial J. A. Woollam M-2000V spectroscopic rotating analyzer instrument, operating in the 3501000 nm wavelength range, with the Kretschman’s type geometry of SPR measurements.16 As shown in Figure 1 the beam was coupled to the sample (i.e., a chromium/gold (Cr/Au)-plated glass slide (2) with the sensitive layer electrostatically adsorbed on top) through a 68° trapezoidal glass prism (1). A specially designed cell (3), having inlet and outlet tubes and thus allowing the injection of different solutions, was attached to the bottom of the sample. The spectra of two ellipsometric parameters Ψ and ∆ were recorded, representing, respectively, the amplitude ratio Ψ ) arctan(Ap/As) and phase shift ∆ ) φp - φs between p and s components of polarized (15) Starodub, N. F.; Nabok, A. V.; Starodub, V. M.; Ray, A. K.; Hassan, A. K. Ukr. Biochem. J. 2001, 73, 55-64. (16) Kretchmann, E. Z. Phys. 1971, 241, 313-324.
Figure 1. Ellipsometric modeling of thin transparent dielectric films (d ) 5 nm, n )1.42, k ) 0) on a Cr/Au layer (n ) 0.36, k ) 2.9) at λ ) 633 nm and an angle of incidence of 68°: (a) changes in Ψ and ∆ caused by the thickness increase; (b) changes in Ψ and ∆ caused by the refractive index increase. The inset shows the scheme of TIRE measurements comprising a 68° glass prism (1), the goldcoated glass slide (2), and the cell (3). light. Optical parameters of the adsorbed layer, such as the thicknesses (d) and refractive index (n), can be obtained by fitting experimental data to Fresnel equations17 using one of the least-square stechniques. Commercial WVASE32 software was provided by J. A. Woollam Co., Inc. for this task.18 In TIRE experiments, all adsorption steps in the sequence PAH, protein A, antibodies, and toxins with intermediate washings in trizma base/HCl buffer (or Millipore water after the adsorption of PAH) were performed by injecting the required solutions into the cell using a syringe. An incubation time of 15 min was used for the majority of the adsorption steps, and the TIRE spectra were recorded after each step in the adsorption in the cell filled with 10 mM trizma base/HCl buffer solution (pH 7.5). Two similar series of samples were prepared and characterized. QCM impedance measurements, described in detail in ref 19, were performed with Impedance Gain/Phase Analyzer (Solartron 1260), followed by in-situ data fitting to the Butterworth-van Dyke (BVD) equivalent circuit model.19,20 The parameters of the frequency shift (∆f) and resistance change (∆R) obtained from the fitting are related, respectively, to the added mass and changes in the viscoelastic properties of the adsorbed layer on the surface of the quartz crystal, which are both caused by molecular adsorption. AFM images of adsorbed layers were taken with the Nanoscope IIIa instrument operating in tapping mode with an oscillation frequency in the range of 300-500 kHz and a scan rate of about 1 Hz. The tip radius was typically 4-7 nm. Both QCM impedance and AFM measurements were performed at room temperature on dry samples in air after the same sequence (17) Azzam, R. M. A.; Bashara, N. M. Ellipsometry and Polarized Light; North-Holland: Amsterdam, 1992. (18) Woollam, J. A. Guide to Using WVASE32; J. A. Woollam Co., Inc.: Lincoln, NE, 2002. (19) Holloway, A. F.; Nabok, A.; Thompson, M.; Ray, A. K.; Wilkop, T. Sens. Actuators, B 2004, 99, 355-360. (20) Ballatine, D. S.; White, R. M.; Martin, S. J.; Ricco, A. J.; Fryre, G. C.; Zellers E. T.; Wohltjen, H. Acoustic WaVe Sensors: Theory, Design, and PhysicoChemical Applications; Academic Press: New York, 1997.
Binding of Amphiphilic Aggregates to Antibodies
Langmuir, Vol. 23, No. 16, 2007 8487 Table 1. Parameters of the Four-Layer Model in Tire Fitting layer
series 1
series 2
3. ambient (BK7) ) 1.515; ) 0 2. Cr/Au n ) 0.263 ( 0.014 k ) 3.202 ( 0.061 d ) 27.69 ( 0.74 nm 1. Cauchy layers Ana ) 1.396, Bna ) 0.01, Cna ) 0 which gives na ) 1.42, ka ) 0 PAH d ) 0.94 ( 0.58 nm protein A d ) 2.97 ( 1.40 nm nonylphenol AS d ) 16.27 ( 5.12 nm nonylphenol results are shown on the graph in Figure 3 0. water na ) 1.3326, ka ) 0 na
ka
n ) 0.328 ( 0.014 k ) 2.999 ( 0.053 d ) 36.38 ( 1.51 nm
d ) 1.57 ( 0.54 nm d ) 5.04 ( 1.73 nm d ) 22.90 ( 6.07 nm
a Parameters were fixed during fitting, and the values of n and k are given at 633 nm.
Figure 2. Typical set of ∆(λ) TIRE spectra corresponding to the following sequence of adsorption steps: initial spectrum of the Cr/ Au layer (1), after adsorption of the PAH layer (2), after deposition of protein A (3), after deposition of antibodies to nonylphenol (4), and after binding of nonylphenol from its solutions of 3.7 (5) and 63 ng/mL (6). The inset shows the typical dependence of the spectral shift δ∆ on the concentration of nonylphenol. of adsorption steps. Before measurements, the samples were vigorously rinsed in Millipore water and dried with a stream of nitrogen gas.
Results and Discussion TIRE Spectral Measurements. In contrast to the traditional SPR method where just the amplitude of the p component of polarized light is measured, the distinguishing feature of the TIRE method lies in the recording of two spectra Ψ(λ) and ∆(λ), representing the amplitude ratio and phase shift between the p and s components of polarized light, respectively. Typical Ψ(λ) spectra obviously resemble SPR curves, whereas ∆(λ) spectra show a nearly vertical drop in ∆ (from 270° down to -90°) near resonance12-14 (Figure 2). It is known that for thin (less than 10-20 nm) transparent dielectric coatings on reflective substrates changes in the both refractive index and thickness mostly affect the ∆ value, whereas Ψ is virtually intact.17 A simple modeling of the variations in ∆ and Ψ in response to changes in the thickness and refractive index of thin transparent coating on a chromium/ gold substrate at a fixed wavelength of 633 nm shown in Figure 1 demonstrates the much higher sensitivity (7 to 15 times) of the ∆ parameter. In the background of these calculations, which are in line with the previous modeling of the TIRE response,10 the ∆(λ) spectra were chosen in our work for the study of the immune reaction of binding nonylphenol molecules. A typical set of TIRE ∆(λ) spectra in the course of consecutive adsorption steps of PAH, protein A, antibodies to nonylphenol and the binding of nonylphenol from its 3.7 and 63 ng/mL solutions is shown in Figure 2. The spectral shift (δλ), caused by nonylphenol binding, depends on the concentration of nonylphenol, as shown by the inset in Figure 2. It is intriguing that the binding of quite small molecules of nonylphenol having a molecular weight of M ) 220.39 yields a substantial spectral shift of about 45 nm at saturation. This effect is similar to (and even more pronounced than) that reported earlier for the binding of T-2 mycotoxin molecules.14 Values of the thickness of the adsorbed layer were obtained by fitting the experimental Ψ(λ) and ∆(λ) spectra to the fourlayer model (BK7 glass-Cr/Au layer-Cauchy layer-aqueous buffer solution), which was carried out using the dedicated WVASE32 software.18 The fitting procedure was identical to that described earlier.14 First, the optical parameters (thickness d, refractive index n, and extinction coefficient k) of the Cr/Au
Figure 3. Results of TIRE data fitting: the dependence of changes in the adsorption layer thickness on the concentration of nonylphenol.
layer were found by fitting the respective Ψ(λ) and ∆(λ) spectra for the bare Cr/Au coating. Then the parameters of the organic layers were determined by fitting the respective Ψ(λ) and ∆(λ) spectra, whereas the parameters for the Cr/Au layer were fixed. The Cauchy dispersion function for the refractive index18 was used to describe adsorbed organic and biolayers:
n ) An +
Bn λ
2
+
Cn λ4
(1)
The parameters of all layers obtained by fitting are given in Table 1. The values of n and k are given at λ ) 633 nm, although the fitting was performed over the whole spectral range. The effective optical parameters of the Cr/Au coating were different in the two series of samples studied. Because of the ellipsometry restrictions in the simultaneous evaluation of the thickness and refractive index of thin transparent films,17 it was assumed that the refractive indices of all organic (bioorganic) layers are the same and equal to 1.42 at 633 nm. Such an assumption is not strictly correct but reasonable so that all changes in the adsorption layer will be associated with the thickness. Following this assumption, parameters An, Bn, and Cn of the Cauchy layer were fixed during fitting. In addition to the optical parameters of the Cr/Au coating and all adsorbed layers given in Table 1, changes in the film thickness (δd) caused by the binding of the T-2 mycotoxin from its solutions of different concentrations are shown in Figure 3. Similar to our earlier observations on T-2 mycotoxin,14 the main changes in the film thickness occur at concentrations of nonylphenol in the range of 20-200 ng/mL. The saturation of
8488 Langmuir, Vol. 23, No. 16, 2007
Nabok et al.
Figure 4. Typical TIRE kinetics of binding nonylphenol from its 250 ng/mL solution to antibodies. The inset shows the dependence of the exponential factor S ) kaC + kd on the concentration of nonylphenol.
the response is observed at concentrations of nonylphenol higher than 200 ng/mL, which is most likely due to the saturation of antibodies’ binding sites. The minimal registered concentration of nonylphenol was 1.2 ng/mL. The value of δd ) 23 nm at saturation is much larger than the ∼1.5 nm length of the nonylphenol molecule. The observed changes in the adsorption layer thickness are even more dramatic than those reported earlier for T-2 mycotoxin,14 which allowed us to assume the mechanism of binding of aggregates of nonylphenol molecules. TIRE Kinetics Measurements. The kinetics of binding nonylphenol to the respective antibodies was studied by measuring TIRE spectra every 20-30 s during exposure to a nonylphenol solution (typically for 15-20 min). Then time dependence of either Ψ or ∆ can be obtained at a selected wavelength. Typically, the time dependence of Ψ was used because Ψ(λ) spectra give a wider linear range than ∆(λ) spectra.14 Such measurements were repeated for different concentrations of nonylphenol, and typical Langmuir-type kinetics is presented in Figure 4. The analysis of binding kinetics has been done similarly to that in our previous publication on T-2 mycotoxin 14 using the following routine described in refs 21 and 22: (i) The time dependence Ψ(t) obtained experimentally was treated as an integral equation of the sensor response21
Ψ(t) ≈ {1 - exp[-(kaC + kd)t]}
(2)
where ka and kd are, respectively, the coefficients of adsorption and desorption and C is the concentration of nonylphenol in solution. (ii) The experimental kinetics curves for every concentration C were then plotted in semilogarithmic coordinates ln Ψ(t), yielding linear dependences with the slope S ) kaC + kd. (iii) Finally, the values of ka and kd were found, respectively, from the slope and intercept of linear graph S(C), which is shown as the inset of Figure 4. The obtained values are ka ) 1.555 × 103 mol-1 L s-1 and kd ) 9.15 × 10-4 s-1. (iv) The association constant (KA), being the ratio of adsorption and desorption coefficients KA ) ka/kd, is therefore found to be 1.6 × 106 mol-1 L. The value of KA obtained for nonylphenol binding is typical for highly specific reactions of antibodyantigen binding.22 (21) Karlsson, R.; Michaelson, A.; Mattson, L. J. Immunol. Methods 1991, 145, 229-240. (22) Liu, X.; Wei, J.; Song, D.; Zhang, Z.; Zhang, H.; Luo, G. Anal. Biochem. 2003, 314, 301-309.
Figure 5. Typical set of QCM admittance spectra of an uncoated crystal (1), crystals after the deposition of PAH (2), protein A (3), nonylphenol antibodies (4), and after the binding of nonylphenol from its solutions of 30 (5), 125 (6), 250 (7), 500 (8), and 1000 ng/mL (9). The inset shows the dependencies of the frequency shift ∆f vs the nonylphenol concentration.
QCM Impedance Spectra Measurements. As in our previous work on T-2 mycotoxin,14 the method of QCM impedance spectroscopy was used in the current study for the registration of nonylphenol. A typical set of admittance peaks near series resonance, measured after every stage of adsorption (i.e., PAH, protein A, and antibodies to nonylphenol as well as after the binding of nonylphenol from its solutions of different concentrations), is shown in Figure 5. Along with the decrease in the resonance frequency of quartz crystals due to the added mass, the damping and broadening of admittance peaks were observed in the course of adsorption and binding reactions. The obtained admittance spectra were fitted in real time to the BVD equivalent circuit model,19 and the dependence of the frequency shift ∆f versus the cumulative concentration of nonylphenol is shown as the inset in Figure 5. The resistance parameter R, which represents the viscoelastic properties of the adsorbed layer, also increases with concentration from its initial value of 28.4 Ω up to 38.3 Ω at a cumulative concentration of nonylphenol of 2000 ng/mL. The value of the added mass ∆m/A can be calculated from the frequency shift ∆f with respect to the nominal resonance frequency f0 using the Sauerbrey equation:19,20
∆f(Hz) ∆m (g/cm2) ) A 2.26 × 10-6(f02)
(3)
It was found to be 18.3 µg/cm2 at saturation conditions. This corresponds to the concentration of adsorbed nonylphenol molecules of 4.9 × 1015 cm-2, which is more than 3 orders of magnitude higher than the expected concentration of nonylphenol molecules of about 2 × 1012 cm-2 bound to respective antibodies. (This estimation was based on the area occupied by antibodies of about 100 nm2, which can bind two molecules of toxin.) The value of the added mass is 10 times larger than that for T-2 mycotoxin binding reported in our previous publication.14 The mechanism of specific binding of large aggregates of T-2 molecules suggested earlier14 is even more obvious in the case of nonylphenol. An increase in the resistance (∆R), which accompanied the frequency shift, can therefore be associated with the softening of the adsorbed layer. Model of the Nonylphenol Micelle. Both TIRE and QCM impedance methods demonstrated anomalously large responses for the binding of nonylphenol molecules. Because nonspecific binding of nonylphenol and T-2 mycotoxin molecules was
Binding of Amphiphilic Aggregates to Antibodies
Langmuir, Vol. 23, No. 16, 2007 8489
Figure 6. Schematic representation of nonylphenol micelles formed in aqueous solutions.
eliminated by washing in 30% methanol and/or acetonitryl, the only logical explanation remaining was the binding of large molecular aggregates of nonylphenol molecules. The adsorption of solvent (i.e., methanol or acetonitryl) may also contribute to the anomalous increase in both the thickness and mass of the adsorbed layers. However, a direct experimental check did not show any substantial mass changes after washing the samples in the above organic solvents. The amphiphilic molecules of nonylphenol can form large aggregates, such as micelles, in aqueous solutions during the dilution of their initial solutions in organic solvents (methanol or acetonitryl). Figure 6 shows a schematic diagram of such a micelle of nonylphenol molecules in an aqueous medium with hydrophilic phenol groups on the exterior and hydrophobic alkyl chains on the interior; a certain number of solvent molecules can be trapped inside the micelle. The mass of the micelle (Mm) can be estimated from the value of added mass ∆m ) 18.3 µg/cm2 obtained from the QCM experiment and the surface concentration of adsorbed micelles (Cm)
Mm )
∆m ) ∆m(Am) C
(4)
where Am ) 1/Cm is the area occupied by a single micelle specifically bound to an IgG molecule (e.g., an antibody) occupying an area of approximately 100 nm2 so that Am ) 100 nm2. Thus, Mm ) (18.3 × 10-5 kg/m2)(100 × 10-18 m2) ) 1.83 × 10-20 kg. However, the mass of the micelle having a radius r consists of the mass of nonylphenol molecules on the micelle surface (A ) 4πr2) and the mass of solvent trapped in the volume (V ) 4/3πr3) inside the micelle
Mm )
4πF 3 4πM 2 r + r 3 A0NA
(5)
where F ) 789 kg/m3 is the density of ethanol, M ) 0.22039 kg/mol is the molecular weight of nonylphenol, A0 ) 0.2 nm2 ) 2 × 10-19 m2 is the area occupied by an alkyl chain of a nonylphenol molecule in a closely packed monolayer on the micelle surface, and NA is Avogadro’s number (mol-1). Cubic eq 5 can be rewritten in the standard form
ar3 + br2 + cr + d ) 0
(6)
where a ) 4πF/3 ) {4 × 3.14 × 789}/{3} ≈ 3.305 × 103 kg/m3,
Figure 7. AFM tapping mode images of the layer of antibodies immobilized on nonylphenol (a) and the same layer after binding nonylphenol molecules (b).
b ) 4πM/A0NA ) {4 × 3.14 × 0.22039}/{(2 × 10-19)(6.022 × 1023)} ≈ 2.3 × 10-5 kg/cm2, c ) 0, and d ) -Mm ) -3.66 × 10-20 kg. Equation 6 has the only physically feasible solution of r ) 1.56 × 10-8 m ) 15.6 nm. Considering the length of the nonylphenol molecule, l ) 1.5 nm, the exterior diameter of the micelle (Figure 6) is equal to Ø ) 2r + 2l ) (2 × 15.6) + (2 × 1.5) ) 34.2 nm. The obtained values of r and Ø for all three solvents used (ethanol, methanol, and acetonitryl) are summarized in Table 2. Table 2. Size of Nonylphenol and T-2 Mycotoxin Micelles Filled with Different Compounds nonylphenol
T-2 mycotoxin
filling medium
F (kg/m2)
r (nm)
Ø (nm)
r (nm)
Ø (nm)
ethanol methanol acetonitryl air
789 791.8 780 1.168
15.6 15.6 15.7 28.1
34.2 34.2 34.4 59.2
8.8 8.8
19.6 19.6
13.9
29.8
As one can see, the micelle size does not vary much with the type of solvent. The amount of solvent trapped inside the micelle is not really known, so the effective value of the density can vary. In the extreme case of an empty micelle (i.e., a micelle filled with air (F ) 1.168 kg/m), the solution of eq 3 yields r ) 28.1 nm and thus the micelle diameter of 56.2 nm. The actual diameter of nonylphenol micelles therefore lies roughly between 34 and 56 nm. Also, the micelles may not be spherical and can be bound not to one IgG receptor but to two or more. In fact, the obtained area occupied by micelles is therefore much larger than the IgG binding site, which means that binding to two or more antibodies is a very likely scenario. Because of all of the uncertainties mentioned above, the proposed micelle model gives only a very
8490 Langmuir, Vol. 23, No. 16, 2007
Nabok et al.
AFM Study of the Formation of Molecular Aggregates. The formation of large micelles of nonylphenol on the surface was directly observed with AFM. A tapping mode AFM study showed an obvious transformation from a more or less homogeneous (mean roughness Ra ) 0.52 ( 0.12 nm) layer of immobilized antibodies in Figure 7a to a much more rough (Ra ) 4.18 ( 0.35 nm) layer in Figure 7b having distinctive grainy features after binding nonylphenol molecules from its 20 ng/mL solution. The observed small grains of 40-60 nm are clustered on the surface to form larger aggregates of 200-400 nm. This observation is in reasonable agreement with micelle modeling considering a limited lateral accuracy of AFM due to the finite radius (from 4 to 7 nm) of the AFM tips. It also shows that the micelles are rather flat, which could be the result of their collapse after drying the sample. Figure 8 shows similar (but less pronounced) transformations after T-2 mycotoxin binding to respective antibodies. The mean roughness was increased from 1.9 nm in Figure 8a to 3 nm in Figure 8b, and a repeatable grainy structure appears in Figure 8b after binding T-2 mycotoxin molecules.
Conclusions
Figure 8. AFM tapping mode images of the layer of antibodies immobilized on the T-2 mycotoxin (a) and the same layer after binding T-2 mycotoxin molecules (b).
rough estimate of its size. However, the 23 nm value of the thickness increase obtained with the TIRE method seems to be in reasonable agreement with the model considering the incomplete coverage and random distribution of adsorbed micelles. The above model can be expanded to other molecules, for example, T-2 mycotoxin. A previous study of T-2 mycotoxin binding14 showed a 4.5 nm increase in film thickness in TIRE experiments and added mass of 3.8 µg/cm2 in QCM measurements. T-2 mycotoxin is a hydrophobic molecule that is soluble in organic solvents (ethanol and methanol were used) and is therefore prone to form aggregates (micelles) in aqueous solution. Although the T-2 mycotoxin molecule has a more complex 3D configuration as compared to a straight molecule of nonylphenol and the structure of the T-2 micelle is not obvious, the above model can still be formally applied. The resulting dimensions of T-2 mycotoxin aggregates are presented in Table 2. Again, the value of the effective thickness increase of 4.5 nm in TIRE experiments seems to be reasonable.
The study of the specific binding of nonylphenol molecules to respective antibodies immobilized on the surface using the TIRE and QCM impedance methods revealed an anomalously high increase in the film thickness and mass, respectively. A comparison of these findings with the results of the previous study of T-2 mycotoxin14 allowed us to suggest a common mechanism of binding large micelles of low-molecular-weight hydrophobic toxins such as nonylphenol and T-2 mycotoxin to the respective antibodies. The remarkable fact is that these binding reactions are still very specific, with association constants in the range of 106-107 mol-1 L. This effect boosts the sensitivity of the biosensor by 3 to 4 orders of magnitude and allows the exploitation of the direct immune assay approach in conjunction with the very simple and crude transducing technique of QCM. The use of the much more sensitive technique of TIRE will allow the registration of very low concentrations of the above toxins in the sub-parts per billion range, which was previously achievable only by using the competitive immune reaction approach.7,8 The formation of large molecular aggregates on the surface after binding toxins to the respective antibodies and the subsequent mean roughness increase were directly confirmed with AFM measurements. Further work will be focused on detailed experimental investigation into the specificity of micelle binding and modeling. Acknowledgment. This work was supported by a collaborative NATO linkage grant (LST. CLG. 980158) within the NATO program Defense Against Terrorism. LA700414Z
