|Home | About | Journals | Submit | Contact Us | Français|
The use of plasmonic nanostructures for the detection of interactions between biomolecules is currently a very active research field. In this report we demonstrate that localized surface plasmons of metallic particles provide the opportunity for sensitive detection of protein-protein interactions in one-step fluoroimmunoassays. Glass coated with silver island films was used for the construction of a bioactive sensing surface where binding results in fluorescence intensity amplification and highly decreased lifetime of the bound probes. The one-step fluoroimmunoassay is demonstrated for detection of TNF-α using secondary antibody labeled with DY488 dye. Kinetic and end-point measurements were performed using intensity and lifetime (phase shift and modulation) measurements. The ability of variation in the ratio of bound to unbound probes provides the mean for tunability of the assay sensitivity using lifetime measurement modalities. The estimated enhancement in sensitivity is about 100-fold compared to the standard glass surface assay because of intensity amplification of 8.4-fold and similar lifetime decrease. The discussion of experimental results is supported with theoretical calculations with specific parameters for TNF-α assay and DY488 fluorophore. The combined effects of plasmonic structures on intensity and lifetime of fluorophores and the use of phase-modulation fluorometry open the opportunity for bioassay design with high sensitivities and simplified biochemical procedures.
The major goal in the field of bioassays is to develop fluorescence biosensors capable of performing rapid and reliable selective immunoassays on labeled samples using relatively simple instrumentation. Innovative approaches to obtain sensitivity enhancements with the potential for single molecule detection have begun to emerge from the nanotechnology field. Recently, several research groups have started to explore the application of metallic nanostructures for the detection of biomolecule interactions. Because metallic nanostructures display unique optical properties due to a strong interaction between free electrons and an incident electromagnetic field, they have the ability to transduce bimolecular binding events into measurable optical signals. There are several approaches to using changes in scattering and extinction spectra of localized surface plasmon resonance which are strongly affected by particle aggregation or changes in refractive index due to biomolecule interactions on their surfaces.1-9
The properties of metallic nanoparticles and nanofilms have also been explored in combination with fluorescent probes for detection of biomolecules interactions. Gold and silver nanostructures provide a means to greatly enhance fluorescence intensities and reduce lifetimes of practically every fluorophore in the visible and infrared wavelength range due to the increased excitation field and increased radiative decay rate.10-15 The near-field effect of fluorophore-plasmon interactions occurs when fluorophores are positioned within a distance of about 3-50 nm from the surface of metallic nanostructures. These effects are frequently known as metal-enhanced fluorescence (MEF) or surface-enhanced fluorescence (SEF). The decrease in fluorescence lifetime in the presence of metallic nanostructures results in better photostability of fluorophores because of the decreased time it takes for photochemistry to occur during the excited state of molecules. Also, the fluorophores are less prone to optical saturation.16 The large practical distance range where fluorophore-plasmon interactions occur is adequate to design surface-based assays with a biorecognitive capture layer and a suitable labeled biomolecule for detection. Several assay designs that exploited the fluorescence intensity enhancement have been described using a variety of fluorescent probes and metallic nanostructures.17-20
In this report, we describe a biosensor that employs the intensity enhancement and reduction in fluorescence lifetime of the probe when in close proximity of metallic nanostructures. It has been reported that the lifetime decreases many folds due to the interaction of an excited fluorophore with particle plasmons and can be as short as 10-20 ps.13-14, 21-23 The highly reduced lifetime of MEF has not been regarded as a practical signal transducer for bioassay design because of requirements for ultrafast time-resolved instrumentation. We demonstrate that the use of phase-modulation (PM) fluorometry allows for the design of biosensors that exploit MEF effects, enhanced intensity and reduced lifetime simultaneosly, which results in more than two orders of magnitude improvement in sensitivity relative to standard intensity measurements. The large intensity amplification and large decrease in lifetime of MEF permits the design of biosensors with high sensitivity, rapid readout, minimal sample preparation, and flexibility of choice of fluorescence labels. The designs we describe use the silver nanostructures in the form of silver islands on glass substrate onto which a bioactive layer is adsorbed. The sensing mechanism of the MEF-PM biosensor is based on changes in the ratio of bound-to-unbound labeled detection biomolecules. Because the bound labeled biomolecules display MEF, the detection is performed in the presence of unbound probes by direct measurement of changes in intensity, phase shift and/or modulation. As an example of the application of MEF-PM for ultrasensitive immunoassays, a tumor necrosis factor alpha (TNF-α) assay is used.
Reagents for human tumor necrosis factor alpha (TNF-α) assays were obtained from Pierce Biotechnology (Rockwood, IL). The TNF-α assay consisted of components of the ELISA kit including recombinant human TNF-α (RTNFAI), capture monoclonal antibody (MTNFAI, binding affinity of 0.14 nM) and secondary detection monoclonal antibody (M3TNFABI, binding affinity of 1.3 nM) labeled with dye DY488 with a dye-to-antibody ratio of about 2:1.
Silanized glass substrates, purchased from Sigma, were used for the deposition of silver islands. The chemical deposition method used has been described elsewhere. 14, 24 On the surface of the MEF substrate, a microwell array was attached with dimensions of 2.5 mm diameter and 2 mm depth that accommodated about 10 uL solutions in each well for preparing bioactive surfaces and subsequent sample incubation and measurement.
TNF-α immunoassays were performed in a sandwich format. The sensing MEF substrate surface was prepared by immobilization of capture antibody by direct physical adsorption. The phosphate buffer solution (pH 7.4) of 50 μg/ml of antibody was incubated for 2 hours, followed with blocking solution (StartingBlock (PBS) Blocking Buffer from Pierce Biotechnology) at room temperature for 1 hour. Next, samples of various TNF-α concentrations were mixed with detection antibody labeled with DY488. The resulting mixtures contained a fixed concentration of Ab2-DY488 of 27 nM and various concentrations of the TNF-α ranging from 7 pM to 24 nM. The Ab2-DY488 concentration of 27 nM (4 ug/ml) was chosen to produce nearly complete saturation of TNF-α with Ab2-DY488 (about 95% saturation calculated based on Kd = 1.3 nM for TNF-α binding to the Ab2 (M3TNFAI)). In this case the competition for binding sites on the surface from free TNF-α is minimized. The sample mixtures were used for the end-point and binding kinetics measurements. For the end-point measurements the samples were incubated with the sensing surfaces for 2 hours using 10 uL solutions within wells (2.5 mm diameter and 2 mm depth). To monitor the binding between complexed TNF-α, the measurements were performed immediately after contacting samples with sensing surfaces. The schematic configuration of sandwich one-step TNF-α assay is shown in Figure 1.
Intensity, phase and modulation measurements were performed using epi-illumination with excitation from a blue LED at 470 nm through a band pass filter of 460/40 nm. The excitation intensity was modulated by applying a RF driving current to the LED [25, 26]. The emission was observed through a band pass filter 525/50 nm. The phase-modulation intensity decays were measured using frequency-domain fluorometer (K2 from ISS, Champagne, IL). For detailed lifetime characterization of reporter probes when free in solution and when bound to MEF substrates, we used sweep mode in the frequency range of 10 - 250 MHz. For assay readouts, a single modulation frequency of 100 MHz was used. Our discussion on phase-modulation fluorometry (see Supplemental Material) has included only fundamental equations to illustrate the principles of phase-modulation for two component system as they relate to the binding interactions. A more detailed description of the phase-modulation technique can be found elsewhere .
We used a four parameter logistic function (4-PL) which is a frequently used analytical function for analysis of interaction between antibody and antigen 
where A1 and A2 are the asymptotic values, CM is the concentration at midpoint and p is the parameter related to the curve slope. The midpoint concentration is interpreted as the effective concentration which produces 50% maximal response (EC50) in the observed measured parameter (intensity, phase shift and modulation) for the TNF-α assay.
The concept of the MEF-PM based immunoassay relies on the difference in spectral properties of the free and bound probes. These differences (intensity and lifetime) are the effect of several factors such as quality of silver nanostructures, selected fluorescent probe and the geometry of the surface assay design. The schematic configuration of sandwich one-step TNF-α assay is shown in Figure 1. The excitation light is incident from the front side thus exciting the free probes in solution and those bound to the surface. The observed emission consists of contributions from both forms of probe. The relative increase in intensity and decrease in lifetime is directly correlated with the antigen concentration. The performance of the MEF-PM-based immunoassay depends strongly on the difference between lifetimes of the reporter protein labeled with dye when in solution and when bound to the sensing surface as well as on amplification of the intensity of bound relative to free probe.
In order to determine the characteristics of free and bound probes, phase-modulation intensity decays were performed. First, the intensity decays of free probe Ab2-DY488 was measured in the absence of the TNF-α. We also measured intensity decay of complexes of Ab2-DY488 with TNF-α in solution and found them similar to that of uncomplexed Ab2-DY488. Secondly, the intensity decay of bound probe was measured after washing out of free probes. The intensity decay results are shown in Figure 2 and the detailed analysis is summarized in Table I. The average lifetime of the free probe (3.71 ns) is reduced 8.8-fold when the probe binds to the TNF-α in the assay sandwich format (0.42 ns). It is important to note that the substantial reduction in lifetime of DY488 in the sandwich assay results in large differences in the phase shifts and modulations between the free and bound probes which allows for the design of a sensitive MEF-PM based bioassay for TNF-α. For example, at a frequency of 100 MHz (dashed line in Figure 2), the difference in phase shift between free probe in solution and that bound in the sandwich assay is about 37 degrees and the difference in modulation is about 0.385. We also determined the intensity enhancement of 10.4-fold by comparison of intensity from a monolayer of bound probes (in sandwich assay) on silver island surface to intensity of identical sample on bare glass.
The geometrical configuration of assay on the sensing surface is very important for MEF-PM assay format because of the distance of the fluorophore to the metal surface. The approximate distance of the DY488 in the TNF-α sandwich assay is about 15-20 nm. One can obtain even larger differences in phase shift and modulation between free and bound Ab2-DY488 if the capture antibody would be replaced with a reduced in-size fragment, Fb, which will cause that the DY488 to be placed closer to the MEF surface upon binding. For example, the lifetime of directly immobilized Ab2-DY488 on the SIFs surface resulted in about 31-fold decreased lifetime (0.12 ns) and intensity enhancement in about 21-fold compared to the free probe in solution and probe bound to glass, respectively. This is because the distance between DY488 and metal surface has been reduced to approximately (average) 5 nm. This observation indicates that direct binding of reporter probe to the metallic sensing surface need to be minimized in assay design to not compromise the MEF-PM assay sensitivity. To this point, the direct binding has been minimized in the subsequent assay design for TNF-α, by using blocking solution after capture antibody immobilization and performing the assay incubation in the presence of proteins (Array Diluent Solution, Pierce Biotechnology).
It is important to discuss the potential effect of probes in solution that diffusively can be placed in proximity to the metal surface and act as bound probes. In the TNF-α assay design presented here only bound probes are enhanced, and those in solution are highly discriminated from MEF. This is because of short near-field effect of particle plasmons (~30 nm) and low concentration of detection antibody. For example, 27 nM of detection probe in 30 nm layer of solution above the surface will be equal to about 8×10-17 mol/cm2 which is about five orders of magnitude less than the typical surface density of capture antibodies of few pmol/cm2. Thus the contribution of MEF from unbound probes will be marginal compared to that bound specifically. While the effect of MEF from free probes is expected to be marginal, one should design the sensing surface to prevent the non-specific binding by using effective blocking agents.
The binding efficiency can be estimated from the equilibrium conditions defined by the concentration of analyte, density of binding sites and dissociation constants between interacting biomolecules. In Appendix 1 (Supplemental Material) we provided such calculations. Due to the binding of antigen-(Ab2-DY488) to the surface, the observed intensity increases and average lifetime decreases. The relative intensity increase and average lifetime decrease depends on the molar ratio of the bound to free probes; therefore the observed emission is providing information about the analyte behavior and its concentration in the sample.
The acquisition of phase-modulation data over the range of modulation frequencies and the analysis of intensity decay are in fact not necessary to perform the MEF-PM-based sensing. However, we provide them in order to better understand the behavior of the fluorescent label in the bioassay when free in solution and bound to the MEF substrate. In the Appendix 2 (Supplemental Material) we provide theoretical calculations for two specie system where phase shift and modulation values at the selected single modulation frequency are calculated. The calculations illustrate that the heterogeneous intensity decays of free and bound probes are not important but the effective values of phase shift and modulation at selected modulation frequency have an effect on the performance of the MEF-PM-based bioassay.
The TNF-α samples with four different concentrations (0.29, 1.14, 5.1 and 24 nM) were initially premixed with a fixed concentration of Ab2-DY488 (27 nM). Next, the samples were contacted with a sensing surface in wells and covered with a glass coverslip. Temporal measurements of intensity, phase shift, and modulation were subsequently performed. Time-dependent binding of TNF/Ab2-DY488 complexes to the capture antibody on the surface are shown in Figure 3. The time-dependent intensity and phase shift were monitored over the course of three hours. The samples were illuminated only during the measurements (~10 second per point) to minimize the effect of photobleaching. The observed temporal interactions indicate that the static incubation conditions are dependent on diffusion of complexes from solution to the surface with gradually decreased concentration of complexes (depletion) during the binding process. The binding capacity of the surface is estimated of about 5 nM (See Appendix 1), thus comparable or higher then the concentration of the complexes in the solution. However, the estimated depletion of unbound probes (free Ab2-DY488 and complexes Ab2-DY488/TNF-α) is small, the largest being 16% at 24 nM of TNF-α. Thus, the relative changes in intensity and phase shift values are due to increased ratio of bound to free probes without significant changes in concentration of the free unbound probes during the binding process.
The control measurements without TNF-α show small change in intensity (~1.1-fold increase) and in phase shift (decrease of about 3 degrees) which are due to non-specific interactions. In the control sample on bare glass (24 nM of TNF-α), we were unable to observe changes in intensity and phase shift beyond the experimental error. However, after washing the unbound probes, we found that the signals on the glass were on average 10.4-fold less than those on the MEF substrate. This implies that the intensity and lifetime of bound probes to the glass substrate were not different (or very little) than of free probes in solution. This observation is in agreement with the fact that binding interactions between biomolecules usually do not change the spectral properties of conjugated fluorescent probes. This is the reason that all surface-based fluorometric bioassays require washing out the unbound probes before readout. The two well established fluorometric approaches, fluorescence resonance energy transfer (FRET) and fluorescence polarization (FP), allow detection of binding without washing steps but they have significant limitations in applications to immunoassays. FRET requires labeling both interacting biomolecules with matching fluorophores (donor and acceptor). The typical distance between donor and acceptor should not exceed 6-8 nm for efficient FRET, thus limit the FRET application to relatively small size of interacting biomolecules. 29 Fluorescence polarization assay relies on the changes in rotational motion of free and bound probes and is limited to the use of small biomolecules labeled with fluorophores of relatively long lifetime (mostly fluorescein with about 4 ns lifetime).30
The measurements of binding kinetics using MEF (intensity) and MEF-PM (intensity and lifetime) demonstrate several desired capabilities for immunoassays; (1) no requirement for washing unbound probes, (2) no restrictions on the size of interacting biomolecules (limited by near-field effects of about 30 nm), (3) flexibility of the choice of fluorescent labels. Such measurements can be designed to determine the binding constants, perform bioassay optimization, and for rapid detection of biomarkers. Thus MEF can be an alternative or supplementary method to the widely used label-free SPR technique. In fact, both techniques exploit interaction of light with surface plasmons; change in refractive index in SPR and change in fluorescence properties in MEF during the binding of biomolecules to the metallic substrates.
The end-point measurements were performed after incubation of TNF-α samples (TNF-α mixed with Ab2-DY488) with the sensing surface for 2 hours. The wells with 10 uL samples (2.5 mm diameter and 2 mm depth) were covered with glass coverslip and measurements of intensities, phase shifts, and modulations were performed. All measurements for a specific TNF-α concentration were performed at least three times using various surface areas within a single well. In order to obtain the saturation of the surface immobilized capture antibodies, one of the wells was filled with a high concentration of TNF-α (171 nM) and was incubated for 2 hours. Before the spectroscopic measurements, the well was washed and filled with Ab2-DY488 at a concentration of 27 nM. This sample allowed us to obtain data point with the highest signal from the bound probes. The dose response curves using intensity, phase shift, and modulation measurements are shown in Figure 4 and and5.5. The data points were fit to the four parameter logistic function as shown with lines. Two series of measurements were performed as marked with A (squares) and B (circles). Series A represents a truly one-step TNF-α assay without a washing step. The obtained dose responses display different assay sensitivities as can be justified by the EC50 values; 3.51 nM (intensity), 1.87 nM (modulation) and 0.92 nM (phase shift). The better sensitivities from PM measurements indicate the additional effects of decreased lifetime. The different sensitivities between the phase shift and the modulation-based assay result because of different weighting of short and long lifetime components in these measurable parameters. Despite the large free probe concentration, the magnitude of changes in phase shift and modulation are substantial which provide ability for accurate detection of TNF- α.
The measurements for series B were performed at reduced signal from the free probes. The measurements were repeated in the presence of lower effective concentration of Ab2-DY488 to facilitate higher ratios of bound to free probes. The silicone spacer of 2 mm thickness was replaced with 0.3 mm. Decreasing the cell thickness causes the ratios of bound to free probes to increase resulting in larger relative changes in intensities (Figure 4). While the relative intensity (Figure 4, curve B) increased several fold, the EC50 value changed moderately from 3.21 to 2.58 nM. The effect of decreased cell thickness, however, resulted in significant shifts in the phase shift- and modulation-based dose responses towards lower TNF-α concentration (Figure 5). The EC50 value of phase shift changed from 0.92 to 0.22 nM and for modulation from 1.87 to 0.62 nM, which indicate an improvement in sensitivities of 4.3-fold and 3.0-fold, respectively. Moreover, the dynamic ranges of phase shift and modulation have been changed moderately using lower detection antibody concentrations. The comparison of midpoint values indicates that measurements of the phase shift results in the most sensitive assay which is more than one order of magnitude (11.7-fold) better than using respective intensity measurements (0.22 nM vs 2.58 nM). The estimated limit of detection was determined by a 3X standard deviation of the phase shift of the sample without TNF-α below the upper baseline of the phase shift dose response and was found to be approximately 3 pM (Figure 5, line B). Taking into consideration that the intensity dose response includes the 8.4-fold intensity enhancement over the glass surface assay, the total estimated sensitivity improvement is 98-fold over a similar but standard assay on bare glass.
The observed intensity, phase shift, and modulation dose responses for the TNF- α assays agree well with those calculated based on binding affinity of antibodies (Appendix 1) and properties of detection probe (Ab2-DY488) free in solution and when bound in the sandwich assay (Appendix 2). The calculated phase responses are shown in Figure 5 for phase shift (dotted lines). The calculations illustrate that the assay performance can be predicted to a relatively high degree. As predicted from the simulated data (Appendix 2), decreasing the contribution from the unbound probes leads to increased sensitivities using phase shift and modulation measurements. The ability for theoretical modeling of assay performance is desired in terms of assay optimization and knowing theoretical limits of selected components; binding affinities of antibodies, surface density of immobilized antibodies, selected fluorophore and the metallic structures enhancement factor. For example, the calculated phase shift dose responses (Appendix 2) do not account for possible effects of non-specific binding which are very difficult to avoid when working with the secondary labeled antibodies. In MEF-PM, the non-specific bindings resulted in a small decrease in phase shift baseline from 65.1 degrees (series A) to 59.8 degrees (series B). The presence of non-specific binding resulted that the experimental calibration curves are slightly different that those calculated. This small shift in the baseline minimally compromises the performance of the phase shift-based assay where the dynamic range for phase shift is still very large (about 39.7 degrees compared with theoretical 44.5 degrees). However, the systematic studies of using the most efficient blocking solution as well as optimizing the immobilization of capture antibody will certainly result in decreased non-specific bindings and superior performance of MEF-PM-based assays. In this report, we focus on the principles of MEF-PM detection technology and less on the optimization of the sensing surface.
To determine the limit of detection (LOD), two times the standard deviation was subtracted from the average values of the phase shift baseline (Figure 5, Curve B) and the corresponding concentration of TNF-α was determined. The obtained value of about 3 pM (about 50 pg/ml) is comparable in magnitude with other methods where the optical amplifications were employed. For example, detection of 100 pg/ml of hCG was achieved with the refractometric method using gold nanoparticles to amplify the changes in refractive index 31 and 200 pg/ml using surface-plasmon field-enhanced fluorescence spectroscopy (SPFS) with optimized sensing surface.32
One can expect that the sensitivity of MEF-PM can be improved with an optimized biosensor (higher intensity enhancement and optimized sensing surface).
A new approach for the design of surface-based fluoroimmunoassays is demonstrated that uses the optical amplification of signal from bound probes. Phase-modulation fluorometry is very suitable for the detection of interacting biomolecules because combines two effects of surface plasmons on fluorescent probes, the intensity enhancement and lifetime reduction. The obtained sensitivity of TNF-α (50 pg/ml) is comparable with other optically enhanced methods. It is expected that sensitivity of single pg/ml can be obtained using an optimized biosensing surface that includes metallic nanostructures, an immobilized capture antibody, and effective blocking reagents. The large number of available fluorophores that can be conjugated to antigens and antibodies as well as many commercially available immunoassay kits can be utilized in the proposed method. These makes that MEF-PM method can be regarded as new detection platform for design of immunoassays.
This research was supported by NIH grants 1R21CA134386, EB006521 and HG002655.