Surface Plasmon Fluorescence Measurements of Human Chorionic

Mar 1, 2005 - A. Toby A. Jenkins*,†. Department of Chemistry, University of Bath, Bath, BA2 7AY, United Kingdom, Max-Planck-Institut für Polymerfor...
0 downloads 0 Views 252KB Size
Anal. Chem. 2005, 77, 2426-2431

Surface Plasmon Fluorescence Measurements of Human Chorionic Gonadotrophin: Role of Antibody Orientation in Obtaining Enhanced Sensitivity and Limit of Detection Margarida M. L. M. Vareiro,† Jing Liu,‡ Wolfgang Knoll,‡ Kris Zak,§ David Williams,§ and A. Toby A. Jenkins*,†

Department of Chemistry, University of Bath, Bath, BA2 7AY, United Kingdom, Max-Planck-Institut fu¨r Polymerforschung, Ackermannweg 10, Mainz, Germany D55128, and Unipath Ltd., Priory Business Park, Bedford, MK44 3UP, United Kingdom

This paper describes the determination of limits of detection (LODs) of interactions between an antigen, human chorionic gonadotrophin (hCG), and antibodies, anti-rhCG and anti-β-hCG, using a sandwich assay by surface plasmon field-enhanced fluorescence spectroscopy (SPFS). Randomly biotinylated antibodies were adsorbed onto a structured self-assembled monolayer (SAM)-streptavidin matrix, tethered to gold via a SAM consisting of biotinylated thiol molecules interspersed with hydroxyalkanethiol molecules. The influence of the concentration of biotinylated thiol on the binding of biotinylated antibody and its functionality, in terms of its ability to bind to the hCG antigen, was studied. This allowed determination of the optimum biotin-thiol mole fraction in the mixed thiol solution and consequently in the SAM, to maximize binding of hCG of the streptavidin-bound antibody. SPFS studies of the binding of a secondary fluorescently labeled antibody to hCG immobilized on the optimized SAMstreptavidin-antibody layer showed that a LOD of hCG of 2 mIU mL-1 (4 × 10-12 mol L-1) could be realized. The system was further optimized by using a more oriented and organized surface by adsorbing monobiotinylated Fab-hCG in place of the whole antibody. A LOD of 0.3 mIU mL-1 (6 × 10-13 mol L-1) was achieved for this system. This work illustrates the importance of antibody orientation, both on the planar surface and in terms of position of binding site, in maximizing sensor sensitivity. In recent years, many affinity-based optical biosensors have been developed. Examples include surface plasmon resonance (SPR)-based immunosensors that have been successfully commercialized and are able to provide both qualitative information, e.g., whether two molecules interact, and quantitative information such as kinetics and equilibrium constants.1 SPR uses an evanes* To whom correspondence should be addressed. [email protected]. Fax: + 44 1225386231. † University of Bath. ‡ Max-Planck-Institut fu ¨ r Polymerforschung. § Unipath Ltd. (1) Myszka, D. G. J. Mol. Recognit. 1999, 12, 390-408.

2426 Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

E-mail:

cent wave phenomenon to detect changes in the refractive index of the surface medium and has been used in a variety of applications, especially in biosensing, and has been shown to be applicable to a wide range of molecules.2,3 However, the labelfree nature of SPR, considered to be an advantage, does limit its ultimate sensitivity. This is particularly the case when detecting binding of small molecules, e.g., DNA, or looking at very low binding density of larger molecules, since SPR effectively measures changes in mass density at the surface. This problem is illustrated by attempts to measure the direct binding interactions of low molecular weight molecules, where alternative methods such as equilibrium analysis and competitive assays have had to be adopted.4-6 Indirect optical immunosensors using the field enhancement phenomenon provides method to overcome many of the limitations of the direct sensors and offers attractive routes to improving sensitivity.7-10 Briefly, this technique, surface plasmon field-enhanced fluorescence spectroscopy (SPFS), uses the strong evanescent optical field, produced by optical resonance of surface plasmons, to excite fluorophores located at or near to the metal-dielectric liquid interface, resulting in a strong fluorescence signal. The evanescence field increases in intensity as the maximum resonance condition of the surface plasmons is reached at θres. The field intensity is enhanced by as much as 16× on gold and 50× on silver when excited by laser light (λ ) 633 nm). The evanescent field decays exponentially in the dielectric medium with a penetrating depth of approximately Lz ) 150 nm, depending on the wavelength of the excitation light. The binding of the analyte induces a change in the refractive index that can be sensed by an angular shift of the SPR resonance minimum, (2) Mullett, W. M.; Lai, E. P. C.; Yeung, J. M. Methods 2000, 22, 77-91. (3) Rich, R. L.; Myszka, D. G. J. Mol. Recognit. 2001, 14, 273-294. (4) Karlsson R. Analytical Biochemistry 1994, 33, 142-151. (5) Gestwicki, J. E.; Hsieh, H. V.; Pitner, J. B. Anal. Chem. 2001, 73, 57325737. (6) Adamczyk, M.; Moore, J. A.; Yu, Z. Methods 2000, 20, 319-328. (7) Liebermann, T.; Knoll, W.; Sluka, P.; Herrmann, R. Colloid Surf. A 2000, 169, 337-350. (8) Lakowicz, J. R. Principles of fluorescence spectroscopy, 2nd ed.; Kluwer Academic/Plenum Publishers: New York, 1999. (9) Malicka, J.; Gryczynski, I.; Gryczynski, Z.; Lakowicz, J. R. J. Phys. Chem. B 2004, 108, 19114-19118. (10) Matveeva, E.; Gryczynski, Z.; Gryczynski, I.; Malicka, J.; Lakowicz, J. R.; Anal. Chem. 2004, 76, 6287-6292. 10.1021/ac0482460 CCC: $30.25

© 2005 American Chemical Society Published on Web 03/01/2005

which is related to the change in the surface mass density. If a fluorescent dye is present at the interface, it can be excited by the electromagnetic field and an intense fluorescence signal can be measured. Hence, both fluorescence and the reflected light intensity signal could be measured simultaneously. Recently, Liebermann et al.7 described a setup for SPFS capable of measuring changes in both the refractive index and fluorescence emission simultaneously. It was shown to be able to detect DNA hybridization and antibody-antibody interactions.7,11,12 We present a reproducible and methodical application of SPFS to measure binding interactions between protein and antibodies using a well-studied supramolecular structured surface to minimize steric hindrance effects and optimize limit of detection (LOD). For the purpose of this study, the well-characterized human chorionic gonadotrophin (hCG) sandwich assay was chosen as a model system. An attempt to sense hCG in serum using the phenomenon of fluorescence emission due to the presence of an evanescent wave formed on a silver substrate was previously described by Attridge.13 A sensitivity lower than 80 mIU mL-1 hCG was determined (protein with different activity of ∼8400 IU/ mg). However, neither surface structure studies of capture antibody surface orientation nor distance control of fluorophore to metal surface was performed. One difficulty with performing fluorescence measurements close to a gold surface is that gold is an efficient fluorescence quencher, via fluorescence resonance energy transfer (FRET). Since FRET is a distance-related phenomenon, it is critical to design a sensor surface that provides sufficient spatial separation between the quencher, the gold film, and the fluorophore. Experimentally, it has been determined that this distance is required to be at least 20 nm.14 Such a distance can be achieved by a number of surface modification strategies. In this work, the well-characterized biotin-self-assembled monolayer (SAM)-streptavidin system is employed.15 This strategy allows the control not only of fluorophore distance from the surface in the Z direction but also of capture antibody density on the surface, in the X-Y plane. The utilization of the biotin-SAMstreptavidin matrix, and further separation by bound hCG and capture antibody, represents a total distance of the fluorescently tagged antibody from the gold surface of ∼20 nm (thiols and streptavidin thickness ∼1 nm and ∼4 nm, respectively;15 antibody vertical height is ∼12 nm;16 X-ray crystallographic measurements of hCG molecular diameter were estimated at 4 nm,17 (Protein Data Bank code 1HCN)). Approximate calculations and assumptions about protein orientation on the surface were made. In this paper, it is shown that by applying different dilutions of biotinylated thiol molecules interspersed with hydroxyalkanethiol molecules (presented in Figure 1), tethered to gold, one can control steric hindrance effects on the adsorbed protein layers, (11) Yu, F.; Yao, D. F.; Knoll, W. Anal. Chem. 2003, 75, 2610-2617. (12) Liebermann, T.; Knoll, W. Langmuir 2003, 19, 1567-1572. (13) Attridge, J. W.; Daniels, P. B.; Deacon, J. K.; Robinson, G. A.; Davison, G. P.; Biosens. Bioelectron. 1991, 6, 201-214. (14) Neumann, T.; Johansson, M.; Kambhampati, D.; Knoll, W. Adv. Funct. Mater. 2002, 12, 575-586. (15) Knoll, W.; Zizlsperger, M.; Liebermann, T.; Arnold, S.; Badia, A.; Liley, M.; Piscevic, D.; Schmitt, F. J.; Spinke, J. Colloid Surf. A 2000, 161, 115-137. (16) Green, R. J.; Davies, J.; Davies, M. C.; Roberts, C. J.; Tendler, S. J. B. Biomaterials 1997, 18, 405-413. (17) Wu, H.; Lustbader, J. W.; Liu, Y.; Canfield, R. E.; Hendrickson, W. A. Structure 1994, 2, 545-558.

Figure 1. Structures of the OH-terminated thiol 1 and biotin-thiol 2 used to prepare the mixed SAM on the gold film capable of binding streptavidin monolayer and preventing nonspecific binding due to hydrophilic surface.

thus controlling the lateral packing of proteins and hence the binding of the antigen hCG. Indirect detection of bound hCG was performed by injecting Alexa Fluor 647 fluorescently labeled secondary antibodies. A LOD of 2 mIU mL-1 (4 × 10-12 mol L-1) was determined. In a previous study by Spinke et al.,18 another supramolecular structure, Fab-hCG (hCG antigen binding fragment), was used to detect hCG. This was measured only by SPR with the LOD determined on this surface being in the order of 10 × 10-9 mol L-1. These measurements revealed that this system had a LOD 2 orders of magnitude lower than was necessary for a practical application, ∼50 × 10-12 mol L-1. In comparison, in the last part of this work, a LOD of 0.3 mIU mL-1 (6 × 10-13 mol L-1) was achieved by careful design of the sensor surface and SPFS detection with a fluorescently labeled detection antibody. MATERIALS AND METHODS Materials. Gold 99.99% was obtained from Advent, 11-mercapto-1-unadecanol (1) was obtain from Sigma-Aldrich and biotinthiol (systemic name: 1H-thieno[3,4-d]imidazole-4-pentanamide, hexahydro-N-[2-[2-[2-[(11-mercapto-1-oxoundecyl)amino]ethoxy]ethoxy]ethyl]-2-oxo-, [3aS-(3aa,4b,6aa)]) (2) shown in Figure 1 was custom synthesized at the Max-Planck-Institut fu¨r Polymerforschung.19 hCG with a molecular mass of 39.5 kDa and an activity of 11 243 IU/mg, anti-R-hCG, and anti-β-hCG were supplied by Unipath, Ltd.. The Fab-hCG fragment, monobiotinylated in the hinge region was kindly supplied by Roche Diagnostics GmbH. Anti-R-hCG and anti-β-hCG were labeled with biotin-NHS (Sigma) and Alexa Fluor 647 (Molecular Probes) following the standard protocols provided by Sigma and Molecular Probes, respectively. The degree of antibody labeling (fluorochemical labels/antibody) was measured spectrophotometrically with a dyeto-antibody ratio of ∼5:1. Randomly biotinylated anti-R-hCG and biotinylated anti-β-hCG are abbreviated as b-anti-R-hCG and b-antiβ-hCG, respectively. Alexa Fluor 647-labeled anti-R-hCG and labeled anti-β-hCG are abbreviated as AF-anti-R-hCG and AF-antiβ-hCG, respectively. All protein solutions were prepared with PBSTA buffer (PBS 10 mM, Tween 0.005% (v/v), NaN3 0.01% (w/v) pH 7.4). The 10 mM glycine-HCl pH 1.7 was used as regeneration buffer. Instrumentation. Measurements were performed on a homebuilt surface plasmon-enhanced fluorescence spectrometer based on a design by Knoll.11 Briefly, the system operates by coupling p-polarized HeNe laser light (633 nm) via a prism into a 50-nmthick gold film thermally evaporated onto high refractive index (18) Spinke, J.; Liley, M.; Schmitt, F. J.; Guder, H. J.; Angermaier, L.; Knoll, W. J. Chem. Phys. 1993, 99, 7012-7019. (19) Ha¨ussling, L.; Michel, B.; Ringsdorf, H.; Ro¨hrer, H. Angew. Chem.-Int. Ed. Engl. 1991, 30, 569-572.

Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

2427

glass LaSFN9 (Berliner glass). As in many conventional SPR instruments, the angle of incidence of the light to the prism is controlled by a goniometer, with reflected light intensity being detected with a photodiode. In this way, the system measures surface plasmon resonance by detecting the reflected light minimum, which is close to the point of maximum coupling of the incident light wave vector with the surface plasmon wave vector on the gold film. Angle-dependent measurements and kinetic information by measuring changes in the light reflected intensity at fixed angle, can thus be made. The important change to the conventional SPR system described above is the addition of a photomultiplier tube (PMT) (Hamamatsu), placed so that it detects light emitted perpendicular to the plane of the gold film and at 90° to the photodiode. In front of the PMT, an interference filter was placed (λ ) 670 nm, ∆λ ) (10 nm, 70% transmittance, LOT-Oriel), which prevents transmission of laser light at 633 nm and is specially designed for Alexa Fluor 647 dyes. Furthermore, a photomultiplier protection unit is employed to prevent PMT damage on light overload, and a universal counter quantifies fluorescence intensity from the PMT. Finally, a computer-controlled laser shutter, positioned before the prism, is employed to prevent photobleaching of fluorophores by the laser light. The SPR and fluorescence systems are all controlled by custom-written programs.20 Flow Cell. A home-built quartz flow cell with a volume of 85 µL was connected to a peristaltic pump (Rego Analog, Ismatec). The flow cell is placed onto a low fluorescent quartz slide (Herasil, Schott) and sealed with two Viton O-rings. The gold-LaSFN9 glass slide is placed onto the flow cell. Introduction of analyte and buffer solutions was performed via the pump/flow system. The circulating volume is ∼400 µL with sample volume, with a circulation rate of 4 mL/min for optimized analyte delivery minimizing mass transport effects. Substrate Preparation and Modification. Gold was thermally evaporated onto LaSFN9 glass in a thermal evaporator (Emitech K975X) at a pressure of 5 × 10-6 mbar. The substrates were thermally annealed for 1 min at 450 °C. The slides were left to cool and then placed in ethanolic solution containing two thiol moieties, 11-mercapto-1-unadecanol spacer molecule 1 and a biotin-thiol 2, shown in Figure 1. The total thiol concentration was 0.5 × 10-3 mol L-1, with the mole fraction of the two thiols being varied systematically in the first part of the experiment. The thiol moieties were allowed to self-assemble on the gold film for >16 h, before rinsing in ethanol and drying under nitrogen. The mole fractions present in the SAM are roughly the same as in the prepared solutions.21 Following self-assembly, the slide was attached to the flow cell and placed in the SPFS instrument. Variation of the mole fraction of two thiol moieties in solution allowed control of the mass density of the subsequent adlayers. Optimization of Antibody Sensor Surface. After incubation of the different gold slides with the thiol moieties, streptavidin was injected on the SAM layer at a concentration of 500 × 10-9 mol L-1. Following streptavidin layer preparation, b-anti-β-hCG at a concentration of 100 × 10-9 mol L-1 in running buffer was injected at a fixed flow rate of 4 mL min-1. The hCG antigen was then injected at the same flow rate with a concentration of 286 IU (20) Scheller, A. Wasplas, Max-Planck-Institut fur Polymerforshung, 2003. (21) Lahiri, J.; Isaacs, L.; Tien, J.; Whitesides, G. M. Anal. Chem. 1999, 71, 777790.

2428

Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

mL-1 in running buffer. After every injection, the signal was measured for 30 min and then the system was rinsed with running buffer for 15 min in order to minimize any bulk effects and eliminate nonspecific binding of analyte. Binding of the antigen hCG to the capture antibodies was followed by regeneration of the surface using 10 mM glycine-HCl buffer, letting it run for 3 min, and rinsing thoroughly with PBS-TA buffer. SPR angular shift for binding of streptavidin, biotinylated antibodies, and hCG was measured as a function of mole fraction of biotin-thiol 2. The surface mass density can be quantified by measuring the change in angle of the SPR minimum relative to the critical angle. Previous calibration measurements have shown that a change in angular minimum of 0.1° equates to a mass density change at the surface of ∼0.53 × 10-9 g mm-2.22 Determination of LOD. For the determination of the LOD, a slide was prepared with the optimized thiol moiety, streptavidin, and b-anti-β-hCG layers, different concentrations of hCG ranging from 0 to 100 IU mL-1 were injected, and cyclic measurements were performed similar to that described in ref 23. Briefly, the surface was prepared for antigen binding by performing a primary regeneration step. The regeneration buffer was injected for 3 min followed by rinsing with running buffer. The sensor surface was then tested for nonspecific binding by injecting fluorescently labeled antibody, the detection antibody, at a concentration of 50 × 10-9 mol L-1 onto the sensing surface with no hCG present: this was considered to be the response for 0 IU/mL hCG. Both SPR and SPFS signals were measured simultaneously. Figure 2 illustrates the capture surface with the immobilized hCG “sandwich” assay. For a final study of the system, biotinylated Fab fragments, with a concentration of 50 × 10-9 mol L-1, were immobilized on the biotin-SAM-streptavidin layer, in place of whole capture antibodies. The same procedure was followed for determination of the LOD on this surface. Binding events for different concentrations of hCG were repeated three times per substrate in order to obtain a dose-response curve. RESULTS AND DISCUSSION Optimization of Antibody Surface Sensor. To maximize the sensitivity of the sensor surface, the first experiment performed was to study how changes in the composition of the binary SAM might affect the measured binding of hCG to a layer of capture antibodies immobilized on the streptavidin-SAM matrix. This experiment was carried out using SPR only, with relatively high concentration, 286 IU mL-1 hCG, to observe binding. It was assumed that one biotinylated antibody will bind to one streptavidin molecule.15 Therefore, variations in the biotin-thiol concentration on the surface allow for control of surface streptavidin density and consequently control of antibody density at the surface. Antibodies are large proteins and are affected by interprotein steric hindrance, and therefore, the optimum sensor surface in terms of antigen mass density may not necessarily correspond to the maximum surface density of the immobilized capture antibody. Moreover, since biotin labeling can take place on many lysine residues, not all antibodies will be optimized in terms of antigen binding site orientation away from the surface. (22) Yu, F.; Persson, B.; Lofas, S.; Knoll, W. J. Am. Chem. Soc. 2004, 126, 89028903. (23) Knoll, W.; Yu, F.; Neumann, T.; Schiller, S.; Naumann, R. Phys. Chem. Chem. Phys. 2003, 5, 5169-5175.

Figure 2. Schematic representation of immunoassays used in this work. Formation of different layers was followed by SPR and binding of fluorescently labeled antibody was followed by SPFS. (i) Sensor surface using randomly biotinylated antibody. An attempt to illustrate steric hindrance effects on the sensor surface; (ii) sensor surface using Fab-hCG monobiotinylated fragment. A more oriented and organized surface is obtained. (a) LaFSN9 glass; (b) 50-nm gold; (c) binary mixed thiol SAM; (d) streptavidin; (e1) randomly biotinylated b-anti-β-hCG; (e2) FabhCG monobiotinylated in the hinge region to allow orientation of binding site on the surface; (f) hCG; (g) fluorescently labeled antibody.

Figure 3. Correlation between mole fraction of biotin-thiol in SAMforming solution to SPR angular shift for streptavidin and b-anti-βhCG (left-hand axis) and hCG (right-hand axis). For χ ) 0.01, 63% of the available antibodies bound to hCG.

Figure 2i illustrates the role that steric hindrance might play on the efficiency of antigen binding. A systematic study was carried out, varying the concentration of the biotin-thiol moiety 2 and the lateral spacer 1 in a binary mixed SAM. The following mole fractions of biotin-thiol 2 were used: χ ) 0.002, 0.005, 0.01, 0.05, and 0.1. The mole fraction in the thiol solution will correlate with surface mole fraction, as both molecules have a C11 alky chain tether.24 In this work, b-anti-βhCG was chosen to form the capture antibody layer because it has higher affinity toward hCG than anti-R-hCG. Figure 3, left-hand axis, shows that as the concentration of biotin-thiol on the mixed SAM increases from χ ) 0.002 to χ ) 0.1 there is an increase of both streptavidin and b-anti-β-hCG although b-anti-β-hCG attachment levels off at 0.1 mole fraction biotin-thiol. Figure 3, right-hand axis, shows the angular shift for hCG binding to the surface-bound b-anti-β-hCG for the various mole fractions of biotin-thiol. The binding of hCG to the surfacebound b-anti-β-hCG does not show the same pattern. It can be seen that maximum angular shift, which corresponds to maximum hCG binding, is observed for a mole fraction of biotin-thiol at χ (24) Ulmann, A. An Introduction to Ultrathin Organic Films: From LangmuirBlodgett to Self-Assmbly, ed.; Academic Press: New York, 1991.

) 0.01 (1 mol %). At mole fractions of 0.05 or greater, even though there is more surface-bound antibody, hCG binding is lower than at χ ) 0.01. A possible explanation for this effect is that surfacebound antibodies need sufficient space to bind to their antigen. These results can be considered in the context of the work by Schneider,25 which studied the binding of hCG to biotinylated randomly orientated 3H-Mab-A94 antibodies bound to avidincoated SiO2. It was found that a maximum level of antigen binding was obtained for a surface concentration of antibodies calculated to be 3.52 ng mm-2, which equated to 28% of the available antibodies binding to a hCG antigen. However, in the experiments carried out in this work, the surface antibody density obtained was lower. For χ ) 0.01, a mass density of b-anti-β-hCG of ∼1.58 × 10-9 g mm-2 was calculated from the SPR angular minimum shift, a higher degree of hCG/antibody binding was achieved, with a calculated 63% of antibodies binding to an hCG antigen. However, when capture antibody concentration was increased, with a biotin-thiol mole fraction of χ ) 0.05, a higher mass density (2.63 × 10-9 g mm-2) of antibody was obtained but a lower total hCG surface density was achieved. Thus, a significantly lower hCG/capture antibody ratio binding was determined, with only ∼34% of the antibodies bound to hCG. This illustrates the importance of steric effects on reducing the effectiveness of antigen binding on to the antibody capture layer due to overcrowding of sensor surface. Measurement of Dose Response of hCG on a Randomly Organized Capture Antibody Layer. With the sensor surface now optimized, the dose response for hCG in the b-anti-β-hCG/ hCG/AF-anti-R-hCG sandwich assay was studied using SPFS. To determine the dose response of the optimized sensor surface, the χ ) 0.01 biotin-thiol SAM was prepared as described previously. Streptavidin was added and b-anti-β-hCG immobilized. The effects of mass transport on the rate and degree of hCG binding on this system are minimized by using this relatively low surface capacity of capture antibody and working on a planar surface. On dextran chips, such as the system described by Yu et al.,22 where Alexa Fluor rabbit anti-mouse IgG) ( AF-RaM) binding to mouse IgG was studied, a very high capture antibody density was achieved on the three-dimensional surface; however, antibody orientation (25) Schneider, B. H.; Dickinson, E. L.; Vach, M. D.; Hoijer, J. V.; Howard, L. V. Biosens. Bioelectron. 2000, 15, 13-22.

Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

2429

Figure 4. Determination of hCG dose response by cyclical injection of hCG followed by addition of 50 × 10-9 mol L-1 AF-anti-R-hCG. Note that the SPR response (left-hand y-axis) shows small changes in signal in all concentration range s due to small changes in temperature, flow, or refractive index of the solution. On the right-hand y-axis, the fluorescence signal change due to change in concentration of hCG is shown.

could not be controlled and diffusion of antigen to antibodies buried in the dextran capture matrix affected the response since the capture surface becomes effectively nonuniform. In our work, nonspecific binding of AF-anti-R-hCG on the sensor surface was initially tested by measuring the fluorescence signal after injection of fluorescently labeled antibody prior to addition of hCG. The signal obtained of ∼8 × 104 counts/ was considered to be the background signal at 0 IU mL-1 hCG. No nonspecific binding was found since fluorescence intensity dropped back to its original value after rinsing with running buffer. Hence, the fluorescence background was used as baseline relative to which all other changes in the fluorescence signal were determined; this was termed the absolute intensity fluorescence IFL. Various known standard concentrations of hCG in a range of 0-10 IU mL-1 were injected cyclically and allowed to bind for 30 min, followed by rinsing and injection of 50 × 10-9 mol L-1AFanti-R-hCG. The increase in fluorescence signal as the labeled antibody binds to hCG was measured for 10 min. Subsequently, the surface was rinsed in running buffer and then regenerated with 10 mM glycine pH 1.7. Regeneration of the surface removed both hCG and AF-anti-R-hCG but retained the b-anti-β-hCG capture surface, which was used for further measurements. All measurements for a specific hCG concentration were repeated twice and found to be reproducible. A typical fluorescence-hCG concentration response is shown in Figure 4. The change in SPR response is due to small variations in the circulating solution temperature and refractive index. Figure 5 shows the calibration curve in which the absolute fluorescence intensity changes as a function of hCG concentration. A signal from a concentration of hCG as low as 1 × 10-3 IU mL-1 (2.2 × 10-12 mol L-1) was resolved, thus illustrating the measurement sensitivity of the SPFS technique on this sensor surface.26 The sensitivity obtained is comparable with other work presented,13,25,27-32 where different measurement methodologies, such (26) Ekins, R.; Edwards, P. Clin. Chem. 1998, 44, 1773-1776.

2430 Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

Figure 5. Dose-response curve for determination of hCG using the system with b-anti-β-hCG and detection with AF-anti-R-hCG. Line in main graph is a Langmuir isotherm fit. Line in inset graph is linear relationship at low hCG concentrations for LOD determination.

as the Hartman interferometer, were used. This sensitivity is 1 order of magnitude higher than that found by the pioneering work of Attridge in 1991.13 To determine the LOD, the IUPAC methodology was employed. The average fluorescence value of the baseline measurement added to three times the standard deviation was calculated and the corresponding concentration of hCG, assuming linearity between 0 and 10 mIU L-1 hCG (see Figure 5 inset), was (27) Boozer, C.; Yu, Q.; Chen, S.; Lee, C.-Y.; Homola, J.; Yee, S. S.; Jiang, S. Sens. Actuators, B 2003, 90, 22-30. (28) Dostalek, J.; Ctyroky, J.; Homola, J.; Brynda, E.; Skalsky, M.; Nekvindova, P.; Spirkova, J.; Skvor, J.; Schrofel, J. Sens. Actuators, B 2001, 76, 8-12. (29) Chetcuti, A. F.; Wong, D. K. Y.; Stuart, M. C. Anal. Chem. 1999, 71, 40884094. (30) Schneider, B. H.; Dickinson, E. L.; Vach, M. D.; Hoijer, J. V.; Howard, L. V. Biosens. Bioelectron. 2000, 15, 597-604. (31) Schult, K.; Katerkamp, A.; Trau, D.; Grawe, F.; Cammann, K.; Meusel, M. Anal. Chem. 1999, 71, 5430-5435. (32) Zhang, B.; Mao, Q.; Zhang, X.; Jiang, T.; Chen, M.; Yu, F.; Fu, W. Biosens. Bioelectron. 2004, 19, 711-720.

determined and found to be 2 mIU mL-1 (0.2 × 10-9 g mL-1). This value is comparable in magnitude to the work of Schneider et al., where a LOD of 0.1 × 10-9 g mL-1 was obtained.25,30 Further Optimization of Sensor Surface: Utilization of Biotinylated Fab Fragments for hCG Sensing. Studies by Spinke et al.18 have shown that steric hindrance, among other factors, limits the sensitivity of antibody sensor surfaces. From our initial study described above, it can also be seen that by reducing the number of binding sites on a planar surface it is possible to minimize access effects. As a consequence, in the sandwich assay configuration, a bound protein-detection antibody conjugate will not influence the binding of detection antibodies to neighboring binding sites. To improve ultimate sensor sensitivity, Fab fragments monobiotinylated in the hinge region were immobilized on a χ ) 0.01 SAM-streptavidin matrix (vida supra). The biotinylated Fab-hCG fragment ensured that the hCG antigen binding site was optimally oriented away from the surface minimizing problems of structuring of the hCG layer and reducing steric hindrance effects on binding of the labeled antibody. Figure 2ii illustrates the new sensor system where the Fab fragments replace the randomly organized antibody layer. The Fab-hCG fragment binds specifically to the C terminal region of hCG, which is positioned close to the R region of hCG.33 Hence, when the Fab-hCG is immobilized on the surface, the β region of hCG will be more accessible to the detection antibody. Therefore, to optimize the level of detection, AF-anti-β-hCG detecting antibody was employed, instead of the AF-anti-R-hCG used earlier. The same hCG titration procedure employing the cyclic binding and surface regeneration was used as before. No nonspecific binding was detected, and a background fluorescence signal of 8 × 104 counts/s was measured for 0 mIU mL-1 hCG seven times and taken as the baseline. No increase in the fluorescence background was found after consecutive surface regenerations. A concentration of 50 × 10-9 mol L-1 AF-anti-βhCG detection antibody was used. The dose-response curve is shown in Figure 6, where the absolute fluorescence intensity versus hCG concentration is plotted for this system. The surface was tested for 0.1mIU mL-1 (2.24 × 10-13 mol L-1) and an increase in the fluorescence signal was found; however, no fluorescence increase was resolved for a 0.01 mIU mL-1 (2.24 × 10-14 mol L-1) hCG concentration. The system sensitivity was determined and found to be 0.1 mIU mL-1 hCG. For all hCG concentrations, the measurements were repeated three times. The LOD, estimated following the procedure described above, was determined as 0.3 mIU mL-1 hCG (6 × 10-13 mol L-1) for this system. This is 1 order of magnitude lower than in the work described by Schneider et al.25 CONCLUSION This study has investigated a number of factors that affect the sensitivity of an immunosensor: the influence of steric effects on analyte binding to the capture surface; the orientation of antigen binding sites on the capture surface; the utilization of SPFS for the detection of hCG. Initially, an SPR study showed that (33) Berger, P.; Sturgeon, C.; Bidart, J. m.; Paus, E.; Gerth, R.; Niang, M.; Bristow, A.; Birken, S.; Stenman, U. H. Tumor Biol. 2002, 23, 1-38.

Figure 6. Dose-response curve for determination of hCG using an Fab-hCG fragment to optimize orientation of antigen binding site on the sensor surface. Linearity is obtained for hCG concentrations between 0.1 and 1 mIU mL-1. Line in main graph is Langmuir isotherm fit. Line in inset graph is linear relationship at low hCG concentrations for LOD determination.

maximum binding of hCG to surface-immobilized b-anti-R-hCG occurred for surfaces with χ ) 0.01 density of a biotin-thiol 2 molecule in a binary SAM with a 11-mercapto-1-unadecanol 1 lateral spacer. From this result we concluded that a lower density of capture antibodies on the surface gives a higher sensitivity. This is probably due to reduction in steric interaction between hCGs on the capture surface (Figure 3). The low mass density of bound hCG required relatively high concentrations of analyte for SPR detection (286 IU mL-1). To increase the detection sensitivity of the system, SPFS was employed. SPFS allowed measurement of fluorescently labeled detection antibodies on the bound hCG layer, thus allowing a higher analyte detection sensitivity. Two studies to determine the LOD using SPFS were carried out. First, a randomly organized biotinylated capture antibody layer was immobilized on the optimized SAMstreptavidin layer used previously. On this system, a LOD of 2 mIU mL-1 was achieved. Second, to orientate the antigen binding site a monobiotinylated Fab-hCG was utilized. The same optimized SAM-streptavidin layer used previously was employed. A LOD of 0.3 mIU mL-1 was calculated. From this work it can be concluded that spatial control of the capture antibody surface, both on the x-y plane and in terms of the orientation of binding sites, is critical for optimizing biosensor sensitivity. ACKNOWLEDGMENT The authors thank Unipath Ltd., England, for providing us with antibodies and human chorionic gonadotrophin proteins and Roche Diagnostics, Pensberg, for providing us with the biotinylated Fab fragment. The authors also thank Unipath Ltd. and the University of Bath for funding.

Received for review November 26, 2004. Accepted January 20, 2005. AC0482460 Analytical Chemistry, Vol. 77, No. 8, April 15, 2005

2431