High resolution microscopy technologies such as electron beam microscopy and confocal fluorescence microscopy, including super-resolution methods (structured illumination microscopy (SIM), stimulated emission depletion microscopy (STED), photo-activated localization microscopy (PALM), and stochastic optical reconstruction microscopy (STORM)) that enable observation of fluorophore-tagged features with sizes below the diffraction limit of light offer the ability to observe nanometer-scale objects [1
]. While super-resolution microscopies are powerful and indispensable research tools, they also require large and costly instrumentation, and thus there is a need to develop methods that are capable of simply observing the presence of a biological analyte, counting analytes with single-unit precision, or performing kinetic measurements of a biomolecular interaction without generating a high-resolution image of the proteins, viruses, or nucleic acids being studied. A further limitation of high resolution microscopy methods is their extremely small field of view, which is insufficient for observing arrays of sensors or biomolecular binding events that may be distributed over many square millimeters of surface area. Therefore, there are compelling needs for technologies that can perform detection and visualization of individual nanometer-scale objects without necessarily obtaining high resolution images of them. Such technologies may be more applicable than high resolution microscopy for applications such as point-of-care diagnostics, and as instruments to study fundamental biological processes at the unit level, in which the analytes interact with each other in a liquid environment.
There has been enormous recent progress in the development of nanoparticles that have physical dimensions that are on the same size scale as biomolecules and viruses, opening the way to tagging biomolecules or viruses with nearly 1-to-1 stoichiometry. Nanoparticles (NP) prepared from dielectric [4
], semiconductor [5
], metal [6
], and magnetic [10
] materials have recently become important elements of biosensor technology due to their ability to prepare their surfaces with ligands that enable them to recognize specific target molecules, and their ability to interact with electromagnetic fields in useful ways. Magnetic NPs can be used to facilitate particle manipulation while at the same time providing a mass amplification tag for acoustic biosensors [12
]. Likewise, metallic NPs, comprised of silver or gold, couple with external illumination sources to generate surface plasmons, which are used to enhance local electric fields on the NP surface [13
]. While many biosensing approaches are capable of sensing the adsorption of large numbers of NPs [15
], several approaches are capable of detecting the presence of a single NP, only if the particle is adsorbed to a specific active location [16
]. For these approaches, the majority of NPs that unfortunately land in an inactive region, remain undetected. Due to the difficulty of directing analytes to precise locations on a substrate surface where a biosensor has sensitivity, an effective approach to overcoming this limitation is to utilize a biosensor surface in which the entire surface area is active as a sensor. Through the use of an imaging detection approach, the adsorption of analytes upon any region within the field of view may be measured. Imaging-based biodetection utilizing optical sensors has been demonstrated using surface plasmon resonance [22
], photonic crystal (PC) biosensors [25
], and dielectric thin film interference sensors [30
]. In these approaches, NPs, may be detected with the potential to observe the attachment of individual targets. Dark field microscopy is a useful tool for NP sensing, but with contrast that decreases as 1/r6
of the NP [9
], and is not able to discriminate scattering centers that are not nanoparticles [8
]. Contrast for sensing NPs can be enhanced when their absorption spectra can be coupled to dielectric high quality-factor (Q-factor) resonators [35
] or moderate Q-factor nanostructured surfaces [38
Recently, ultrasensitive assay technologies with “digital” analyte precision have been introduced, and summarized in reviews [39
]. For example, the Simoa™ system by Quanterix [41
], uses an enzymatically amplified fluorescent reporter attached to antibody-functionalized micron-scale magnetic beads isolated in 50-fl reaction chambers to achieve non-multiplexed fM-scale detection limits with a more complex protocol than enzyme-linked immunosorbant ELISA. Likewise, the Erenna immunoassay technology offered by Singulex claims 1 fM detection limits using functionalized magnetic microparticles, fluorescent dye tags, and a custom format flow cytometer, but is not capable of multiplexed assays. Both of these technologies require enzymatic amplification to achieve detection, representing the ability only to observe reaction endpoints, and an assay protocol that requires chemical amplification reagents. The use of NPs (rather than micron-scale beads) for capture and tagging of analytes represents a situation in which a single tag corresponds very closely to a single analyte (because the nanoparticles are nearly the same size as the biomolecules that they tag) for more accurate quantitation. Further, the NP can be simultaneously used as a capture agent and as an imaging contrast agent when combined with a sensing method with single-NP resolution, thus removing the need for enzymatic amplification processes that add time and complexity to assays. In fact, a major advantage shared by all the detection approaches presented in this review is that the sensor generates an output instantaneously with NP capture, which results in immediate output and the capability for dynamically monitoring the accumulation of analytes to yield kinetic information about the processes taking place. This represents a major advantage over methods that require enzymatic amplification, which can only render a single result at the end point of an assay.
In this review, we will restrict ourselves to describing approaches capable of detecting the presence of NPs through their intrinsic dielectric properties, in which the NP’s complex dielectric permittivity, represented by its refractive index and absorption, is responsible for generating some form of contrast. Thus, the modalities we will consider will utilize the interaction of the NP with an optical transducer that enables an externally observable quantity to be detected through, for example, a shift in resonant wavelength, a change in scattered spectrum, or a change in the intensity of reflected/transmitted light due to scattering or absorption. Although nanometer-scale light emitters, such as phosphors, flurophores, and quantum dots [42
] also interact strongly with optical nanostructures that can substantially modify their emission properties, they will not be covered in this review and interested readers are directed to several excellent papers on this topic [42
]. Likewise, this review will not discuss approaches that include surface enhanced Raman scattering (SERS) from molecules attached single nanoparticles, where the reader is referred to [45
The topics covered in this review are important because single NP detection and the ability to sense many NPs with single-NP (digital) resolution form the underlying principle for novel concepts in biodetection in which a nanometer-sized object (such as a virus particle) is the detection target, or in which a NP is used as a tag to signal when a specific biomolecule-biomolecule interaction has taken place. Direct detection of viruses on an optical transducer that has been prepared with a capture molecule (such as an antibody) that specifically recognizes a protein on the outer coat of the virus is an important capability that is sought for applications that include biological warfare agent defense, HIV viral load monitoring, and food safety. For each of these situations, the number of available pathogens in the test sample may be extremely limited, and thus the ability to capture a large percentage of the pathogens on the active “sensing” region of a transducer is of critical importance, especially in situations in which the presence of even a single virus represents a health threat. Such “direct” detection methods are desirable compared to approaches that require extraction and amplification of pathogen-specific nucleic acid sequences due to potentially improved simplicity (especially through greatly reduced requirement for temperature cycling, extra reagents, and development of “primers” that specifically amplify only the target nucleic acid sequence) and time required to obtain a result. Digital resolution sensing of the presence of specific biomolecules (such as proteins, DNA, or RNA) is an enabling capability for the most demanding applications for in vitro diagnostics such as liquid biopsies for cancer via detection of circulating DNA, messenger RNA (mRNA) or microRNA (miRNA) [46
], where the test sample volume may be limited to only several microliters (such as a droplet of serum from a fingerstick), combined with analyte concentrations that extend below 1 pg/mL. For such scenarios, the number of available target molecules may be between 100–10,000, and thus approaches that require aggregation of many captured molecules into a compact region, such as a microarray spot, will no longer reliably generate features that can be recognized and quantified [50
In this review, we will initially categorize sensing approaches by their capability to function as an imaging modality. Non-imaging discrete transducers may be operated either singly or in the form of multiplexed arrays. Generally, the transducers are comprised of a structure that includes a specific region (such as the perimeter of a whispering gallery mode resonator or the cavity of a photonic crystal) in which the electromagnetic field is substantially greater than surrounding regions, representing the location upon which attachment of a nanoparticle will generate the greatest measurable signal. A key challenge for such devices is preparing them so that the target analytes bind only to the sensitive location, rather than becoming bound to a part of the structure where they cannot have an impact on the measured output. Such approaches may have the ability to observe a single nanoparticle that was fortunate to land upon the sensitive part of the transducer, while many thousands of other analytes are present nearby, but not observed. Approaches that are capable of imaging, however, enable counting of each nanoparticle within a field of view, regardless of where they attach to the transducer surface. Thus, the entire transducer surface can be considered as “active” for sensing, and different regions within the field of view can be designated as experimental controls or as regions for multiplexed detection of many analytes.
This review has described the underlying physical principles, sensor configurations, instrumentation, and applications for nanoparticle-tagged biodetection with the capability for digital resolution. Through the ability to observe the attachment of a single dielectric or metallic nanoparticle on an optical transducer, individual biomolecules or virus particles can yield easily observable signals that can be used to signal their presence and to count them. We have reviewed broad classes of surface plasmon resonant surfaces dielectric resonators, thin film interference effects, and scatting microscopies that share the capability for single-nanoparticle detection limits, and we broadly classified detection methods as those that measure aggregated signals, versus those that can generate images of captured nanoparticle tags.
The technologies described in this review are moving towards approaches that can be applied as life science research tools and in vitro diagnostics assays. Importantly, they can enable direct detection of selectively captured viral particles, and nanoparticle-tagged detection of nucleic acid or protein molecules that serve as biomarkers of disease. Importantly, nanoparticle tags do not require enzymatic amplification and unlike fluorescent reporters, are not subject to the effects of photobleaching, which enables real-time kinetic monitoring of the detection process, with accurate quantification of the analyte. Further, nanoparticle-tagged digital-resolution detection occurs instantaneously when the analyte attaches to the sensor, enabling the collection of dynamic information and the ability to report a result rapidly. We expect, in the near future, that several of the technologies reviewed here will move towards clinical translation and broad adoption in the life science research community. The methods described in this review are expected to become important tools for some of the most demanding challenges in biosensing, in which a very small number of available analytes are present in test samples of limited volumes. These types of applications can include detection of low-abundance nucleic acid (miRNA, mRNA, or circulating tumor DNA) or protein biomarkers within a droplet of serum, detection of a small number of viral particles in a sample of bodily fluid, or studying the proteomics within individual cells. In each of these situations, the ability to digitally count the analytes with a simple detection process will lead to simple workflows, lower costs, and more rapid acquisition of results.