Fluorescence is one of the most widely used detection modalities for gene expression analysis, DNA sequencing, disease biomarker diagnostic tests, and cell imaging due to the ability for high quantum efficiency tag molecules or nanoparticles to provide excellent signal-to-noise ratio. Many fluorescence-based assays are performed upon solid surfaces, such as glass substrates, upon which immobilized molecules selectively capture analyte molecules from a test sample. Such assays can be readily multiplexed in a microarray format, in which capture molecules are applied to a substrate in distinct spots, enabling detection of many genes (for a DNA microarray) [1
] or disease biomarkers [6
] simultaneously. In conventional fluorescent microarrays performed upon glass substrates, the analyte is tagged with a fluorescent dye, and a focused laser beam is scanned across the array surface to generate an image of fluorescent output as a function of position. Despite the sensitivity afforded by this approach, there is a need to further reduce the limits of detection for microarray analysis in order to observe the presence of genes that are expressed at low levels and to detect disease biomarkers present at lower concentrations. As a result, there is a great deal of interest in techniques that can increase the sensitivity of fluorescent assays through the use of nanostructured optical surfaces that can locally enhance the electromagnetic fields that excite fluorescent emission and to more efficiently couple fluorescent photons to a detection system. Such approaches are broadly referred to as demonstrating enhanced fluorescence (EF).
Several approaches have been investigated for EF. Metal nanostructures can efficiently couple an external laser light source to substrate-immobilized fluorophores through surface plasmons that generate localized regions with enhanced electric field intensity [3
]. When fluorescent molecules are placed close to the regions of these intensified electric fields, a subsequent enhancement in fluorescence emission by 10-100 × has been reported. Despite the ability of metal-based nanostructures to generate evanescent fields, the application of plasmon-based EF has been limited by the absorption of light at optical wavelengths (which limits resonant quality factor and thus the potential electric field enhancement factor) [15
], and by quenching of fluorescent molecules that are in close proximity to metal [17
An alternative approach to EF utilizes dielectric-based optical resonator surfaces, such as photonic crystals (PC). PC surfaces have demonstrated the ability to provide higher quality factor (Q~1000) optical resonances than surface plasmons [18
], a lack of quenching [21
], and the ability to obtain enhancement factors as high as 7500× [18
]. PC enhanced fluorescence (PCEF) has been performed upon plastic-based [19
] and quartz-based [18
] surfaces that can be inexpensively fabricated over large areas (i.e. entire 1×3 in2
microscope slides or 3×5 in2
microplates) by nanoreplica molding [22
] or nanoimprint lithography [25
]. A PC may be designed with multiple resonances that couple with both the wavelength of fluorescent excitation and the wavelength of fluorescent emission, so that the mechanisms of enhanced excitation (increased local electric field at the PC surface) and enhance extraction [20
] (increased collection efficiency of emitted photons) can operate simultaneously with multiplicative effects [18
]. Microarray assays performed upon PC surfaces may be measured by conventional commercially available confocal laser scanners [30
], but greater enhancement factors have been demonstrated using custom designed detection systems that utilize collimated laser illumination rather than focused illumination [28
]. PCEF has been applied to large DNA microarrays used for gene expression analysis [30
], and to protein microarrays for detection of breast cancer biomarkers in serum [32
PC optical resonances have also been exploited as a platform for label-free (LF) detection. The incorporation of biomolecules into the evanescent field region of the PC results in a positive shift of the resonant wavelength (for a fixed incident angle) [33
] due to the increased dielectric permittivity of biomolecules with respect to water. An alternative method for LF detection on a PC surface that allows high resolution imaging of surface-based biomolecular interactions is to measure a shift in the resonant coupling angle, when illuminating the PC surface with a fixed wavelength from a laser [28
]. A microscope-based detection instrument for label-free imaging the density of immobilized microarray capture spots [32
] and cells [29
] has been demonstrated by detecting shifts in the Angle of Minimum Transmission (AMT) as a function of position upon a PC surface with ~3 × 3 μm2
The fact that the resonant coupling conditions of a PC shift due to the presence of immobilized biomolecules poses a unique challenge and opportunity for fully exploiting PCEF. Obtaining the greatest possible enhancement factor for enhanced excitation requires that the PC be illuminated at the precise wavelength/angle combination for optimal resonant coupling. While high quality-factor PC resonances yield the greatest enhancement factor [37
], they also have the most stringent coupling condition. For example, previous work has shown that illumination of a PC surface at the optimum coupling angle/wavelength results in a ~100 × enhancement factor, while a θ=0.4° incident angle deviation reduces the enhancement to only 10× [37
], and a deviation of θ=1.2° completely eliminates the enhancement. The regions of a PC surface with a high density of immobilized capture molecules, such as the capture spots of a microarray, have very different coupling conditions than the regions of a PC surface between the capture spots. Therefore, it is possible to selectively obtain a large enhancement factor from the capture spot regions of a microarray by illuminating them in an optimal “on-resonance” condition, while at the same time illuminating the regions of a microarray between the capture spots in an “off-resonance” condition. In this way, the fluorescent signal emitted from active assay regions of the surface can be maximized, while the background fluorescence between spots can be minimized, thereby substantially improving image contrast. The problem of obtaining optimal and equal excitation conditions for an entire microarray is further exacerbated by the fact that the biomolecular density of immobilized capture spots is not completely uniform due to a variety of factors that include spot buffer concentration variability, capture spot density nonuniformity, surface chemistry nonuniformity, capture molecule binding affinity variation, analyte molecular weight, and analyte concentration in the test sample. As a result, no two spots in a microarray are guaranteed to have the same optimal resonant coupling condition, and optimally resonant conditions can even vary within a single spot.
In this work, we demonstrate a detection instrument and image processing approach that takes full advantage of the optimally available enhancement factor for a PCEF surface for every location of a PC surface. The method utilizes a label-free image of the microarray capture spots to identify the locations of an array surface that are within a capture spot, and to differentiate “on-spot” regions from those that are “between spots.” To obtain optimum on-resonance coupling for the entire microarray in the presence of differences in immobilized spot density, a series of fluorescent images are rapidly gathered using a range of incident angles separated by small angle increments. An image processing algorithm is applied that generates a composite fluorescent image in which the maximum fluorescent intensity is selected for on-spot regions on a pixel-by-pixel basis, while between-spot regions are displayed using data collected in an off-resonance condition. The method is demonstrated to substantially improve the contrast of microarray images while maximizing the uniformity of the fluorescent image through the application of a uniform enhancement factor. The method provides substantial gains in signal to noise for the cases where assays are specific and background fluorescence is low. However, in most biological assays, issues of high substrate autofluorescence, background fluorescence from blocking reagents and non-specific binding can minimize the gains in signal-to-noise ratio (SNR) provided by the PC’s enhancement capabilities. In such situations, selective enhancement is desirable to maximize image contrast and nullify the effects of non-specific binding.
The outline of this paper is as follows: Section 2 describes the basic functions of label-free imaging detection and fluorescence enhancement of the photonic crystal enhanced microscope (PCEM). Section 3 describes the scheme to generate a mask that selects the area for applying enhanced fluorescence detection. Section 4 implements this technique for characterizing the fluorescent antibody microarray on the PC surface. Using tumor necrosis factor-alpha (TNFα) as an example protein biomarker, the advantage of selectively enhanced fluorescence is demonstrated with reduced limit of detection for a small microarray. Section 5 presents this study’s conclusions.