A digital microarray using interferometric detection of plasmonic nanorod labels
Derin Sevenler, George Daaboul, Fulya Ekiz-Kanik, M. Selim Unlu
A digital microarray using interferometric detection of plasmonic nanorod labels
Derin Sevenler , George Daaboul , Fulya Ekiz-Kanik , and M. Selim Ünlü Department of Electrical and Computer Engineering, Boston University, Boston, MA 2.
NanoView Diagnostics, Boston, MA
PRELIMINARY DRAFT 23 January 2017
DNA and protein microarrays are a high-throughput technology that allow the simultaneous quantification of tens of thousands of different biomolecular species. The mediocre sensitivity and dynamic range of traditional fluorescence microarrays compared to other techniques have been the technology’s Achilles’ Heel, and prevented their adoption for many biomedical and clinical diagnostic applications. Previous work to enhance the sensitivity of microarray readout to the single-molecule (‘digital’) regime have either required signal amplifying chemistry or sacrificed throughput, nixing the platform’s primary advantages. Here, we report the development of a digital microarray which extends both the sensitivity and dynamic range of microarrays by about three orders of magnitude. This technique uses functionalized gold nanorods as single-molecule labels and an interferometric scanner which can rapidly enumerate individual nanorods by imaging them with a 10x objective lens. This approach does not require any chemical enhancement such as silver deposition, and scans arrays with a throughput similar to commercial fluorescence devices. By combining single-nanoparticle enumeration and ensemble measurements of spots when the particles are very dense, this system achieves a dynamic range of about one million directly from a single scan.
Protein and DNA microarray technologies continue to be useful in a myriad of biomedical and clinical applications, such high-throughput genetic or transcriptional analysis and multiplexed protein detection. Insufficient sensitivity and dynamic range are the two most commonly cited weaknesses of the technology, and have motivated the widespread adoption of newer methods based on DNA sequencing or sample compartmentalization to perform sensitive and multiplexed nucleic acid or protein analysis . This practical limit is not imposed by the microarray format itself, but rather the sensitivity of conventional fluorescence readers. A theoretically ideal transducer that could quantify the absolute number of immobilized targets with no background would be limited only by Poisson process variability . Since a typical 100 µ m-wide microarray spot contains about one billion probe oligonucleotides , the ideal microarray transducer would also have a dynamic range as high as 100 million. In practice, virtually all fluorescent scanners have a dynamic range of only 100-1,000, despite the fact that single fluorophores are routinely detected in scientific microscopy . The reason for this discrepancy is that single fluorophore detection requires a high numerical aperture (NA) objective lens with a tight focus tolerance and a small field of view, while microarrays are often larger than 1 cm . It is a significant technical challenge to maintain the required focus tolerance (usually less than 300 nm) while scanning across a large array. Most single-fluorophore scanners simply cannot scan large arrays, or otherwise require a sophisticated focus-tracking system . Even then, the scanning throughput remains fundamentally limited by the quantum “speed limit” of the fluorophore’s emission lifetime, which sets the fluorophore’s minimum obtainable average time between photon emission events. In contrast, measurements of light scattering by nanoparticles have no saturated emission rate or photobleaching. The speed and throughput of light scattering measurements therefore tend to be limited only by the available light power, or maximum allowable local heating of the particle. Gold nanoparticles are routinely used place of fluorescent probes for microarray labeling, and either detected individually directly based on their light scattering or indirectly via silver deposition . All of these techniques have successfully enhanced the sensitivity of microarrays by several orders of magnitude, and typically have a limit of detection of roughly 1 femtomolar. However, the former of these approaches have all required a high-NA lens—reducing throughput to less than 20 spots—while the latter methods suffer from reduced dynamic range unless multiple rounds of silver enhancement and re-scanning is performed. To our knowledge, no method exists to enumerate individual nanoparticle labels across a very large surface with a throughput comparable to commercial fluorescence scanners. The most obvious way to increase throughput would be to use a lower magnification lens to increase the instrument field of view. However, lower magnification lenses are less efficient at collecting light, and reduced light collection is unacceptable for many of these designs. For dark-field detection, the signal scales (approximately, for NA<0.6) with the fourth power of the NA. Interferometric detection is more resilient than dark-field to very weak signals since the scattered light amplitude is measured rather than intensity. Interferometric reflectance imaging sensing (IRIS) is one of a family of similar optical techniques for interferometric detection of nanoparticles immobilized on a substrate . IRIS uses a substrate of polished silicon with a thin film of thermally-grown silicon dioxide. The substrate is imaged with a reflectance microscope with Köhler illumination from an LED source. In the absence of any nanoparticle, the microscope camera observes a featureless reflection of the illumination light on the substrate surface. If a nanoparticle is present, light scattered by the particle is also imaged onto the camera where it forms a faint diffraction-limited interference pattern with the reflected light. The ‘normalized intensity’ of this interference pattern is obtained by dividing by the intensity of the reflected field alone. If the substrate and illumination were both ideally smooth and uniform, arbitrarily weak signals could be detected by collecting enough photons until the shot noise was reduced below the normalized intensity of the signal. In practice, IRIS substrates are slightly heterogeneous, resulting in about 0.5% fluctuations in reflectivity across the chip surface. Therefore, the normalized intensity of nanoparticles must be at least 2-3%, to be robustly detected. Figure 1. (a) Schematic of gold nanorod (GNR) detection with interferometric reflectance imaging sensing (IRIS). Circularly polarized plane wave illumination 𝐄 " is reflected as a circularly polarized plane wave by the IRIS substrate 𝐄 %&’ but scattered by the GNR as a spherical wave 𝐄 ($) that is linearly polarized along the rod’s longitudinal axis 𝜃 . (b) Schematic of the reflectance microscope used to image the IRIS chip. Both the reflected (dotted lines) and scattered light (red shadow) are imaged onto the camera with a 10x objective ( LP - linear polarizer, QWP - quarter wave plate). (c) At the camera, the phase between scattered and reflected light depends on both particle orientation 𝜃 and focus position 𝑧 . All GNRs are robustly detected regardless of their orientation by acquiring a z-stack and measuring the difference between the maximum and minimum at each (x, y) pixel. This is the main challenge to performing interferometric detection with a low-NA objective—reducing the NA reduces the collection of scattered light but not that of reflected light, which reduces the normalized intensity of the particle image. If the normalized intensity drops below the 3% visibility threshold, the particle will no longer be detectable with IRIS. The reduced scattering could be compensated for, if the reflected light could be attenuated. However, this cannot be done with a simple neutral density filter since it takes the same optical path from the chip surface to the camera. We therefore developed a method that utilizes the unique depolarization properties of plasmonic gold nanorods to selectively attenuate the reflected light.
RESULTS AND DISCUSSION Interferometric detection of single plasmonic nanorods with a 10x objective lens.