Label-free detection and manipulation of single biological ...chenglab/pdf/DeSantisCheng2016WIRES.pdfAdvanced Review Label-free detection and manipulation of single biological nanoparticles

Advanced Review

Label-free detection andmanipulation of single biologicalnanoparticlesMichael C. DeSantis1 and Wei Cheng1,2*

In the past several years, there have been significant advances in the field ofnanoparticle detection for various biological applications. Of considerable inter-est are synthetic nanoparticles being designed as potential drug delivery systemsas well as naturally occurring or biological nanoparticles, including viruses andextracellular vesicles. Many infectious diseases and several human cancers areattributed to individual virions. Because these particles likely display differentdegrees of heterogeneity under normal physiological conditions, characterizationof these natural nanoparticles with single-particle sensitivity is necessary for elu-cidating information on their basic structure and function as well as revealingnovel targets for therapeutic intervention. Additionally, biodefense and point-of-care clinical testing demand ultrasensitive detection of viral pathogens particu-larly with high specificity. Consequently, the ability to perform label-free virussensing has motivated the development of multiple electrical-, mechanical-, andoptical-based detection techniques, some of which may even have the potentialfor nanoparticle sorting and multi-parametric analysis. For each technique, thechallenges associated with label-free detection and measurement sensitivity arediscussed as are their potential contributions for future real-world applications.© 2016 Wiley Periodicals, Inc.

How to cite this article:WIREs Nanomed Nanobiotechnol 2016. doi: 10.1002/wnan.1392

INTRODUCTION

Nanoparticle research is increasingly becomingone of the most studied fields in science as a

result of the sheer diversity of nanoparticles and theirpotential biomedical and technological applications.As early as 1857, the unusual optical properties ofnanometer-scale metals were first reported as beingdifferent compared with that of the bulk materiallargely owing to the high surface area to volumeratios.1 These size-dependent properties includequantum confinement in semiconductor physics,

surface plasmon resonance (SPR), and superpara-magnetism in various nanoparticles. Moreover,metallic nanoparticles can be modified for manydiagnostic and clinical applications in medicineincluding fluorescent labeling, drug and gene deliverysystems, and the detection and purification of biolog-ical molecules. Consequently, polymer- andliposome-based synthetic nanoparticles are now inpractical use throughout the pharmaceutical industryyielding better control over size, particle dispersaland absorption, surface functionalization with pro-teins or other desired chemicals, and lower toxicityprofiles. Nevertheless, characterization of naturallyoccurring (biological) nanoparticles, specificallyviruses, remains of vital interest for elucidatinginformation on basic structure and function, reveal-ing novel targets for therapeutic intervention aswell as ultrasensitive detection for biodefense; forthis reason, they will be the primary focus of thisreview.

*Correspondence to: [email protected] of Pharmaceutical Sciences, College of Pharmacy,University of Michigan, Ann Arbor, MI, USA2Department of Biophysics, University of Michigan, Ann Arbor,MI, USA

Conflict of interest: The authors have declared no conflicts of inter-est for this article.

© 2016 Wiley Per iodica ls , Inc.

In addition to infectious diseases, viruses havebeen linked to several human cancers. The detailedmechanism of infection is not always known but canusually be attributed to a single virion. Consequently,detection and characterization of single biologicalnanoparticles is necessary, particularly under physio-logical conditions, such that solutions may containheterogeneous populations of particles with differentphysical and optical properties. To unequivocallycharacterize natural nanoparticles, the following cri-teria are of importance: (1) label-free sensing, (2) highsensitivity for single nanoparticles and, if possible,multi-parametric analysis [to examine properties suchas size, fluorescence, refractive index (RI), electriccharge, etc.], (3) high throughput/speed, (4) low tech-nical difficulty (which can be achieved through auto-mation), (5) nanoparticle heterogeneity (for a rangeof particle sizes, materials, and protein incorpora-tion), (6) manipulation and/or nanoparticle sorting,and (7) preservation of the specimen’s biologicalactivity. Among these various attributes, the need forsingle-particle sensitivity is critical because these par-ticles likely display significant heterogeneity evenwithin a single population.2

In this article, we will review recent advance-ments for label-free detection of single, natural nano-particles, and the extent to which these characterizedparticles can be further manipulated for scientificstudy. Electron microscopy and other methods thatusually lead to destruction of the sample are not idealand will not be emphasized. Moreover, mostfluorescence-based detection techniques, such as fluo-rescence resonance energy transfer and super-resolution microscopy, by themselves are not suitableas they require fluorescence labeling of the nanoparti-cle. Additionally, fluorescence correlation spectros-copy provides quantitative information typically onan ensemble of diffusing particles instead of individ-ual particles. To focus on label-free sensing and sort-ing of single biological nanoparticles in this review,we have categorized the methods according toelectrical-, mechanical-, and optical-based detection.For each technique, the challenges associated withlabel-free detection and measurement sensitivity arediscussed as are their potential contributions forfuture real-world applications.

ELECTRICAL-BASED DETECTION

Advances in nano-electronics have afforded newclasses of optical biosensors incorporating photovol-taic cells, chemical, electrochemical, and photodiodeelements. Traditionally, electric charge-based

detection uses a local sensor to measure a change inimpedance, ionic conductance, or capacitance as bio-molecules transiently bind to or transit along con-ducting/semiconducting surfaces. Polymer- andmetal-based nanowires were among the first materi-als used in the detection of whole bacteriophages andinfluenza A viruses3,4 while thin film carbon nano-tube (CNT)-based biosensors have been employedmore recently.5,6 Here, nanowires are highly sensitiveto perturbations to the local environment/nanowiresurface with respect to its electrical properties result-ing from either the adsorption of charged moleculesor changes in ion concentration (Figure 1). Measure-ments include increased nanowire resistance owing toa gating effect between two contact electrodes orconductance changes when using field-effect transis-tors. Single-virus sensitivity has been reported.3

However, in the absence of other measurement tech-niques, these devices primarily provide only a highconfidence level for stochastic sensing due to incuba-tion with antibodies for selectivity, which ultimatelylimits what biomolecules can be detected. In

FIGURE 1 | Nanowire-based detection of single influenza Aviruses. (a) Two nanowires incubated with different antibodies topromote specific binding. (b) The corresponding changes inconductance reflect the surface charge of a virus that binds andunbinds to nanowire 2. (Reprinted with permission from Ref 3.Copyright 2004 National Academy of Sciences of the United States ofAmerica)

Advanced Review wires.wiley.com/nanomed

© 2016 Wiley Per iodicals , Inc.

addition, biomolecules in heterogeneous solutions aredetected simultaneously prohibiting individual selec-tion and subsequent manipulation. Furthermore, adecline in detection sensitivity may occur under phys-iological conditions where ionic concentrations arehigher.

Many of these limitations are overcome usingnanochannels or nanopores coupled with electricalsensing technology which permits single particle anal-ysis and sorting. Once again, changes in conductanceare measured as a nanoparticle transits through ananopore(s) without the requirement of specific bind-ing to antibody-coated surfaces (Figure 2). Accord-ingly, individual viruses can be characterized withhigh throughput. In one such study, the size and con-centration of T7 bacteriophages in both salt solutionsand blood plasma were determined as the particlespassed through two voltage-biased electrodesembedded within a microfluidic channel.7 A commontechnique, however, is to utilize (tunable) resistive-pulse sensing in which the observed current changeor blockade magnitude is proportional to the volume

of electrolyte displaced by the particle. Consequently,determination of particle size and sample concentra-tion in addition to electrophoretic mobility and sur-face charge has been performed in the case ofhepatitis B virus (HBV) capsids,8,9 extracellular vesi-cles including microvesicles and exosomes,10 as wellas a host of other biomolecules within the biomedicaland pharmaceutical industries. Moreover, McMullenet al. have used a solid-state nanopore to align andtransiently trap the stiff, filamentous virus fd as wellas measure its diffusion coefficient as a result of anelectric field gradient.11 Nanopore-based detectionoffers relatively high speed sizing and analysis ofindividual nanoparticles that can also be incorpo-rated into microfluidic platforms (i.e. ‘lab-on-a-chip’or LOC devices) greatly increasing particle through-put. The high signal-to-noise ratio of resistive-pulsesensing allows for discrimination and characteriza-tion of virus-sized nanoparticles, such that for smal-ler particles (<100 nm), the salt concentration of theelectrolyte can be increased or the size of the nano-pore can be decreased to induce detectable currentchanges. However, a potential problem of thisapproach is the change in osmotic pressure. For bio-logical particles, such as enveloped viruses, thisunbalanced tonicity may result in morphologicalchanges of these particles. Moreover, this approachis limited by the size of nanopores. Although smallerpores offer greater sensitivity, large particles or par-ticulate matter/components in buffered solutions canlead to clogging of the nanopores, such that a popu-lation of smaller nanoparticles may fall out of therange of detection.

MECHANICAL-BASED DETECTION

Arguably, the most direct way to detect and/or iso-late single nanoparticles is based on physical mea-surements. Unfortunately, as the size of the desiredparticle decreases so does the number of availablemethods which must rely on increasingly smallernanostructures either in the form of nanofluidic chan-nels or microcantilevers. In the first case, fabricationof planar nanochannels using photolithography tocreate LOC devices allows for laminar flow that canbe driven by capillary action. Connecting multiplenanochannels with decreasing heights in series,injected nanoparticles become physically trappedwhen their sizes exceed the channel’s dimensions.Fluorescence-based detection at the trapping interfacehas been demonstrated with polystyrene beads butalso with biological materials including capsids fromherpes simplex virus I (HSV-1) and HBV.12

FIGURE 2 | Nanochannel design for resistive-pulse sensing ofhepatitis B virus capsids. (a) Schematic of the microfluidic chamberwith two channels for delivery of biological nanoparticles.(b) Scanning electron microscope image of the two microchannelswith an enlarged view of a 2.5-μm long pore-to-pore channel.(c) Atomic force microscopy image of this channel incorporating poresthat are 45 nm wide, 45 nm deep, and 430 nm long. (Reprinted withpermission from Ref 9. Copyright 2014 American Chemical Society)

WIREs Nanomedicine and Nanobiotechnology Detection of biological nanoparticles


Alternatively, combining these nanochannels as abranched network and using electrical sensors (seesection Electrical-Based Detection) at several junc-tions permits real-time sorting based on the measuredelectric current and molecule length.13 While the ana-lytes can be retrieved in the latter design after beingsorted into desired outlets, only DNA molecules havebeen examined thus far.

Most mechanical detectors, however, utilizecantilevers that undergo deflection in response toadsorbed nanoparticles. Unlike atomic force micros-copy whose operational modes can work in liquidenvironments, single nanoparticle detectors typicallyrequire a vacuum for accurately measuring the fre-quency shifts of oscillating microcantilevers; a highmechanical quality factor (high-Q), which is a meas-ure of damping, as well as the resonator’s mass pri-marily dictate the system’s resolution. However, Leeet al. have eliminated viscous loss by using suspendednanochannel resonators (SNRs) in which fluid isplaced directly inside the resonator. By fabricatingnanochannels into the cantilever, SNRs haveachieved a mass resolution of 2.7 × 10−20 kg in a1 kHz bandwidth such that gold nanoparticles withdiameters down to 20 nm can be detected.14 As aresult, detection and weighing of individual nanopar-ticles, including viruses and protein aggregates withcomparable masses, should be possible. In general,cantilever-based detectors are expensive, demand thatsamples are adsorbed onto substrates which forbidsubsequent sorting, and are susceptible to artifactsarising from unusual sample/nanoparticle topologyor from coarse probes. Nevertheless, further minia-turization can lead to increased sensitivity andbroader applications in both biophysics and materi-als science.

OPTICAL-BASED DETECTION

For the past 400 years, optical microscopy has beenthe conventional technique used by scientists for visu-alization of larger, micron-sized particles often rely-ing on visible light. Modern developments in charge-coupled device (CCD) and complementary metal–oxide semiconductor (CMOS) cameras have affordedhigh spatiotemporal resolution while allowing thecapture of digital images to aid in data analysis. As aresult, many subfields have emerged including spec-troscopy as well as absorption and fluorescencemicroscopy for studying nanoparticles. While fluores-cence microscopy can achieve single-molecule sensi-tivity using super-resolution methods or singleparticle tracking analysis, additional labeling is

normally required to visualize the biomolecule orprotein of interest. In lieu of this, advancements inlight scattering techniques have allowed imaging ofsingle nanoparticles with physical sizes well belowthe diffraction limit. In contrast to dynamic lightscattering which monitors the time-dependentchanges in the scattering intensities for an ensembleof particles in bulk suspensions, nanoparticle track-ing analysis (NTA)15 and evanescent wave light scat-tering16 directly image scattered light to track theBrownian motion of an individual particle (>10 nm);by assuming spherical particles, the size of individualparticles can be derived from the measured diffusioncoefficients based on the Stokes–Einstein equation.The NTA technique has been commercialized andcan perform label-free measurements of nanoparticlesand protein aggregates in suspension. However, thereremain challenges associated with the reproducibilityand peak identification of the calculated size distribu-tions. Moreover, photothermal heterodyne imagingof virus-like nanoparticles with various absorptionproperties17 and far-field emission of photonicmolecule-conjugated nanoparticles18 have beendemonstrated. Here, we focus on several traditionaloptics-based methods that have recently character-ized viruses, some of which are capable of nanoparti-cle sorting.

Refractive-Index Based DetectionThe RI, defined as the ratio between the speed oflight in vacuum and its phase velocity in the materialof interest, is a fundamental optical property of mate-rials. Consequently, it potentially offers a label-freeparameter for distinguishing materials of differentcompositions. Because the transient binding orimmobilization of biomolecules at an interface altersthe local RI, there are corresponding changes to themeasured optical signals which are subsequently usedto detect these events. Various structures include sur-faces composed of thin metal layers or metallic nano-particles for SPR, photonic crystals (PCs), opticalring resonators, or other nanocavities to achieve lightenergy confinement; several of these structures havealready been incorporated into microfluidic channelsto increase throughput and further minimize solutionvolumes. In the context of biological nanoparticles,we shall discuss two of these techniques in greaterdetail.

Surface Plasmon ResonanceTo measure this change in the local RI, one form ofimplementation among many others, are devices thatutilize SPR. In a typical setup, a thin layer of metal



(the interface, usually gold) is sandwiched betweenglass and aqueous media. A laser beam is projectedto the back of the metal layer under total internalreflection conditions to create a field of evanescentwaves. When the angle of laser incidence equals theso-called ‘resonance angle’, the electron clouds (‘plas-mon’) in the metal will absorb photon energy fromthe evanescent field, and the resulting loss of lightintensity can be detected by measuring the reflectedlaser beam. This resonance angle depends on thelocal RI, which changes in response to moleculesbinding at the interface. Therefore, by measuring theangle of laser reflection, laser intensity changes at afixed wavelength, or wavelength shifts in the lightspectrum, one can measure the binding of moleculesin real time. The sensitivity of RI-based sensors thatinvolve resonant modes is often reported as a ratio ofthe magnitude of the spectral shift to the correspond-ing RI change, the latter being measured in refractiveindex units (RIUs). In comparison with other meth-ods, SPR can achieve higher detection sensitivities of>10,000 nm/RIU, depending on the resonance wave-length used.19 Also important is the system’s resolu-tion which denotes the smallest spectral shift that canbe accurately measured. Together, these quantitiesdetermine the detection limit quantifying the mini-mum amount of adsorbed protein the sensor canmeasure.

While SPR has been used extensively to studymolecular adsorption of polymers, DNA, and pro-teins, detection of single nanoparticles, includingviruses and exosomes, has only recently been demon-strated. In 2010, Wang et al. measured the intensitiesof both human cytomegalovirus and H1N1 influenzaviruses alongside silica nanoparticles (with compara-ble size, d = 98 nm and RI, n = 1.46), on surfacesfunctionalized with and without antibodies yieldingsize and mass distributions with a detection limitof 0.2 fg/mm2 (Figure 3).20 Nanoplasmonic-baseddetection has also been used to analyze intacthuman immunodeficiency viruses (HIV) of varyingsubtypes selectively captured on antibody immobi-lized biosensing surfaces at clinically relevant concen-trations from unprocessed whole blood.21 For thecase of exosomes (50–100 nm in diameter), Imet al. recently built an SPR-based sensor called thenano-plasmonic exosome assay consisting of a seriesof nanohole arrays coated with specific antibodiesfor profiling surface-exposed proteins to analyze exo-somes derived from ovarian cancer cells22; similarassays have also been developed.23,24 SPR-baseddetection techniques can achieve single-particle sensi-tivity and provide a low-cost, high throughput designfor biomolecule analysis, especially for point-of-care

(POC) medical testing, independent of solutionconditions.

Optical Ring ResonatorsSimilar in idea to the above SPR devices, the local RIchange due to binding of molecules at the interfacecan also be measured in a dielectric microparticle, orthe so-called ‘resonant microcavity’.25 At the reso-nant wavelength, light coupled to the cavity via aninput waveguide will increase in intensity due to con-structive interference as it cycles within the structurebefore detection at an output bus waveguide; opticalcoupling is affected by the distance, the couplinglength, and the difference in RIs. As a result, whisper-ing gallery modes (WGMs) are formed within thecavity such that its evanescent field can be locallyperturbed by binding of molecules at the particle sur-face, which causes a measurable shift in the reso-nance frequency. Early studies demonstrated thesensitivity of optical ring resonators as a biosensorby measuring the binding of proteins, such as biotin,on the microparticle surface; in addition to bulk RIdetection, the corresponding noise equivalent detec-tion limit of 0.14 pg/mm2 had been reported.25,26

Subsequent work with influenza A viruses hasdemonstrated sensitivity for detection of a singlevirus in aqueous buffer. The small size of the influ-enza A virus demanded reduction in the microcavitysize to amplify the magnitude shift of the resonancefrequency (Figure 4).27 Similar work has been donefor bean pod mottle viruses.28 While various environ-mental noise factors can distort measurements basedon resonance frequency shifts, the temporal resolu-tion of WGMs is on par with photon lifetimes,τ = Q/ω ≈ 10−10 seconds despite lower Q valuesassociated with smaller resonator cavities havingincreased energy leakage.

To improve optical coupling as the signalamplitude is sensitive to the location of nanoparticlebinding with respect to the evanescent tail of theWGM, it is recommended that the WGM shifts,broadenings, and splittings be monitored as well.Consequently, Zhu et al. utilized split WGMs in anultrahigh-Q resonator to analyze the size distribu-tions (d = 106.4 � 11.0 nm) and polarizabilities ofsingle influenza A virions.29 They estimated that thedetection limit for dielectric nanoparticles (n = 1.5) isapproximately 20 nm such that the smallest polysty-rene particles accurately measured were 60 nm indiameter. Moreover, a reference interferometer oper-ating in conjunction with an ultrahigh-Q microcavityoffers significantly increased sensitivity to resolve fre-quency shifts in the detection of influenza A vir-ions.30 Optical resonators can easily be adapted for



use on microfluidic or LOC devices requiring littlemode volume to accurately characterize individualnanoparticles, although a significant weakness con-cerns the sequential binding of particles to the reso-nator surface which can change WGM characteristicsover time. In summary, both SPR and WGM techni-ques are sensitive for detection down to single parti-cles and confer high selectivity because biomoleculesrequire antibody-coated surfaces for binding. As aresult, further manipulation of these detected parti-cles is prevented.

InterferometryUnlike absorption microscopy which requires thatnanoparticles have intrinsic absorption propertiesmaking them suitable for study, interferometric meth-ods rely on the scattered intensities of particles illumi-nated by coherent light. While direct imaging of non-fluorescent nanoparticles is possible, as discussedabove, the signal intensities are typically weak forsmaller particles owing to an r6 dependence, thusgreatly reducing detection sensitivity. Consequently,interferometry allows resolution enhancement of

FIGURE 3 | Surface plasmon resonance microscopy for detection of influenza A viruses. (a) Schematic of the experimental setup. (b) Images ofH1N1 influenza A viruses and silica nanoparticles of varying size in PBS buffer. (Insets) Corresponding nanoparticle images produced by numericalsimulation. (Reprinted with permission from Ref 20. Copyright 2010 National Academy of Sciences of the United States of America)

(a)

Microspherecavity

Photodetector

Virus

Sample cell

1.0

0.5T

0.00 10 20

Wavelength sweep (pm)

30

0

0

1

2

3

4

5 × 10–8

Ê� I

nfA

/�

1.5 × 10–8

1 2

Time (seconds)

3

Tunable laser

(b)

FIGURE 4 | Detection of influenza A virions by a microsphere resonator. (a) Illustration of a silica microsphere immersed in aqueous solution.A tunable distributed feedback laser excites whispering gallery modes (WGMs) of a microsphere by evanescent coupling to a tapered optical fiber.(Inset) Typical transmission spectrum for a WGM mode detected while the laser wavelength is tuned. (b) The resonance wavelength shiftsassociated with the binding of single influenza A virions to a microsphere cavity (r = 39 mm) in PBS buffer. (Reprinted with permission fromRef 27. Copyright 2008 National Academy of Sciences of the United States of America)



optical microscopy by numerical combining of partialimages with registered amplitude as well as phase.Based on the frequencies of the initial waves, theresulting interference pattern provides measurementsof small displacements, RI changes, and surface irre-gularities within the examined sample. One of theearliest techniques involved wavefront splitting inter-ferometers to observe the phase difference of lightpassing through an aperture yielding constructive ordestructive interference. Accordingly, HSV-1 virionsbound to antibody-coated waveguide surfaces havebeen detected using a Young interferometer.31 As adetector only, direct measurements, including sizeestimation of the nanoparticles, are not performed.While the analyte concentration can be determined,the detection limit may not be sensitive to singleparticles.

A common method to increase sensitivity is bynon-linear mixing of scattered light with a referencefield through homodyne detection in which the twobeams are of the same wavelength. Heterodyne detec-tion uses a stronger reference beam to amplify thescattered field intensity and shift it to a new fre-quency range as either the sum or difference of therespective frequencies; the new signal will have anintensity proportional to the product of the ampli-tudes of the two input signals. The latter has beenused notably by Mitra et al. for real-time detectionand characterization of several viruses includinginfluenza, Sindbis, and HIV, within nanofluidic

channels (Figure 5).32–34 Single-virus sensitivity hasbeen demonstrated.32 To further improve the detec-tion signal via phase imaging, Daaboulet al. developed an interferometric reflectance ima-ging sensor so as to immobilize nanoparticles, includ-ing influenza H1N1 and vesicular stomatitis virus-pseudotyped virions, on a reflective Si/SO2 sub-strate.35,36 Briefly, a top oxide layer minimizes theoptical path difference at specific wavelengths result-ing in closely aligned scattered and reference fields.The enhanced interferometric response of nanoparti-cles on the surface can then be imaged at differentwavelengths for detection of individual particles. Forthese techniques, prior knowledge about the virus,such as its RI and behavior as non-adsorbing, spheri-cal Rayleigh scatterers as well as experimental para-meters used in the latter’s model, is required foraccurate sizing.

Optical TrappingInvented by Ashkin,37 optical tweezers (OTs) haveemerged as a powerful tool in biophysics for trappingand manipulation of nanometer- and micron-sizedparticles. For typical single-beam gradient OTs, theyare constructed by focusing a laser beam to adiffraction-limited spot using an objective lens with ahigh numerical aperture. At the narrowest point of atightly focused laser beam, radiative pressure due tothe transfer of momentum by incident photons will

Nanofluidic channel

Nanoparticle Objective lens

Objective BFP

Iris

Photodetector

Er (! + �!)

Es (!)

Eexc (!)

FIGURE 5 | Schematic of the dark-field heterodyne interferometric detection technique. Nanoparticles inside a glass nanofluidic channel aredetected as they traverse an illumination spot formed using an excitation laser (Eexc). The detector signal is proportional to the product of thenanoparticle’s scattering field (Es) and an additional frequency-shifted reference field (Er). (Inset) Scanning electron microscope image of thenanofluidic channels. Scale bar = 2 μm. (Reprinted with permission from Ref 33. Copyright 2012 Elsevier)



cause a dielectric particle to experience a scatteringand gradient force. As the name implies, the scatter-ing force will tend to push the particle in the direc-tion of light propagation whereas the gradient force,proportional to the dipole moment of the dielectric,will pull the particle toward the focal region. If theaxial component of the gradient force is greater thanthat of the scattering force, a particle is stablytrapped in three dimensions such that its axial equi-librium position is located slightly beyond the focus.Generally, OTs use continuous-wave (cw) lasersoperating in the infrared or near-infrared (NIR)regime. This is advantageous for biological specimensbecause they have less absorption in the NIR windowwhere light has its maximum depth of penetration intissue. Nevertheless, one should pay attention to thewavelengths selected as potential photodamage totrapped particles can occur. For example, molecularoxygen can absorb around 1064 nm and generatereactive oxygen species in the presence of a sensi-tizer.38 To avoid this photodamage, an 830 nm laseris strongly preferred which also reduces heatingeffects due to water absorption.39,40 Although prima-rily used for trapping, NIR cw lasers are also capableof simultaneous two-photon excitation of fluorescentproteins in optical traps or inside eukaryotic cellswith single-molecule sensitivity.41,42

Compared with micron-sized objects and bacte-rial or mammalian cells, the trapping of biologicalnanoparticles is complicated by their small size andpoorly characterized RIs. While it was previouslydemonstrated that larger-sized viruses, such as thetobacco mosaic virus, and bacteria could be easilytrapped, most animal viruses are typically less than300 nm in size.43 Because the gradient force scales asr3, the photon forces will be considerably weaker forthese particles compared with typical micron-sizedobjects; thus, particles may have a higher tendency toescape after being trapped. To confirm the trappingof these individual particles, it is thus necessary toprovide additional measurements on the trapped par-ticles to show that these are indeed single particles asexpected.

Using a single-beam gradient optical trap incombination with confocal microscopy, Bendix andOddershede has successfully demonstrated opticaltrapping of unilamellar lipid vesicles as small as50 nm that encapsulate high concentrations ofsucrose solutions.44 By combining an optical trapand single-molecule two-photon fluorescence (TPF)detection, Pang et al. has demonstrated optical trap-ping of individual HIV-1 virions in culture media(Figure 6).2 Because the TPF is excited by the trap-ping laser right at the focus and measurable at the

single-molecule level owing to the ultra-small excita-tion volume, this technique is well suited to detectand characterize nanoscale, dielectric particles withsingle-molecule resolution. They have thus namedthis technique ‘virometry’, following ‘cytometry’, thepowerful technique that has been widely used tocharacterize heterogeneity among individual cells. Anadditional advantage of this technique is the use ofback-focal-plane interferometry (BFPI)45 which per-mits ultra-sensitive tracking of the damped Brownianmotion of the trapped particle at high bandwidth(62.5 kHz). Fitting of the corresponding power spec-trum yields the diffusion coefficient and corner fre-quency of the trapped particle with high precision.46

It was observed that these parameters are highly sen-sitive to the simultaneous trapping of multiple parti-cles such that individual virions can be easilydistinguished from the total population. By using anindependent displacement reference, BFPI offers pre-cision measurement of the diameters of the trappedparticles and can thus be used to directly quantitatethe heterogeneity among particles based on their size.Combined with simultaneous TPF detection usingeither external or internal fluorescent labels, virome-try truly offers multi-parametric analysis for an indi-vidually trapped particle and is well suited formeasuring heterogeneity of nanoparticles in solutionwith high sensitivity.

The single-molecule resolution offered by TPFis particularly important for measurement of biologi-cal viruses because these particles are likely to behighly heterogeneous with respect to their proteincontent. One example is the density of envelope gly-coproteins on the surface of individual HIV-1 virions.Previous cryoelectron microscopy studies of individ-ual HIV-1 virions derived from chronically infectedT cell lines indicated a broad heterogeneity in thedensity of these proteins per virion, ranging from4 to 35.47 Using optical trapping virometry, Panget al. showed that the density of these proteins canvary by over one order in magnitude for individualHIV-1 virions derived from transfected cell lines.Given the apparent low copy number of this proteinand the large variation across individual virions,single-molecule sensitivity is therefore essential toreveal these heterogeneities. Although single-moleculefluorescence detection has become routine nowadaysin many labs, for example, through the implementa-tion of total internal reflection fluorescence micros-copy to study the dynamics of HIV-1 envelopeglycoproteins,48 virometry is the only techniqueavailable that offers multi-parametric analysis forindividual, nanoscale particles in solution withsingle-molecule sensitivity. As a result, individual



virions can be characterized in real time and manipu-lated. This capability also makes it possible todirectly deliver a single virion to a host cell for study-ing the early events in viral infection. Additionally,the unprecedented level of control in manipulatingvirus-like particles serves as motivation to incorpo-rate OTs in microfluidic platforms, possibly in com-bination with other optical-based techniques, toachieve high-throughput nanoparticle detection.

Fabrication of optical nanostructures has alsopaved the way for near-field optical trapping includ-ing holographic and plasmonic OTs that are capableof 3D manipulation of nanoscale, dielectric objects.49

Using a PC resonator embedded within a microfluidicchamber to create a highly localized optical cavity,Kang et al. has recently demonstrated optical trap-ping of individual influenza H1N1 virions in freesolution.50 In order to estimate the stoichiometry ofbound antibodies to a single virion, they have devel-oped an effective sphere model relating the change

in trap stiffness and RI after incubation withspecific antibodies. In this approach, evanescent wavelight scattering produces an image of theparticle from which 2D localization measurementsare made. The resulting positional variance is used tocalculate the trap stiffness with the equipartition the-orem. The stoichiometry of bound antibodies pervirion can then be estimated based on this quantita-tive model. This technique has the potential to beexquisitely sensitive to biomolecular binding in freesolutions and provides binding parameters in a label-free manner.

Flow CytometryFlow cytometers are ubiquitous throughout the bio-technology and pharmaceutical industries as theyserve as powerful tools in the identification and anal-ysis of heterogeneous samples of cell populations.Historically, most commercial instruments were

FIGURE 6 | Optical trapping virometry of HIV-1 virions. (a) Experimental schematic for optical trapping of single HIV-1 virions in culture fluid.HIV-1 virions were delivered into a microfluidic chamber and trapped by the IR laser focused at the center of the chamber. The xyz dimensions areshown as indicated, with y perpendicular to the figure plane. (b) Representative two-photon fluorescence (TPF) time courses from individuallytrapped HIV-1 virions. All traces were fit with a single exponential function (red), with time constants for each trace as follows: 60.7 (square), 49.5(cross), 51.5 (circle), and 54.1 seconds (triangle). Time zero started with the onset of TPF collection. (c) The laser deflection signal measured in realtime using back-focal-plane interferometry (BFPI), which can be used to distinguish single HIV-1 particles from aggregates in complete media.(d) Representative Alexa-594 TPF time course from a single virion bound with monoclonal antibody b12 labeled with Alexa-594, where individualphotobleaching steps are indicated with arrows. (Inset) A cartoon of a HIV-1 virion with a single envelope glycoprotein trimer. (Reprinted withpermission from Ref 2. Copyright 2014 Nature Publishing Group)



designed for rapid counting/sorting of cells and simi-larly sized particles in real time; consequently, theyare routinely used diagnostically in the medical com-munity, especially for blood cancer screening. Detec-tion is initially based on forward scattered (FSC) andside-scattered light which convey an object’s size as itpasses through a stationary laser focus; however,multi-parametric analysis is usually performed byincluding additional lasers and photomultipier tubesfor excitation and fluorescence detection, respec-tively. To ensure that particles or cells are in single-file passage through the observation region, the sam-ple is typically delivered by hydrodynamic focusingusing sheath fluid, or through a microcapillary, or acombination of hydrodynamic focusing plus acousticfocusing. Here, analysis and sorting of heterogenousmixtures can be achieved through fluorescence-activated cell sorting. Briefly, a vibrating mechanismcauses the stream to break into individual dropletscontaining no more than one cell. At the same time,an electrical charging ring places a charge on thedroplet according to a fluorescence measurementtaken previously with respect to a parameter of inter-est. Finally, an electrostatic deflection system divertsthe droplet into an appropriate container based onits charge.51

Most flow cytometers on the market can detectparticles with diameters larger than 300 nm. For par-ticles much smaller than the wavelength of the light,the so-called Rayleigh scatterers, the scattered lightvaries as r6 which makes it complicated to detectthese particles with sufficient signal-to-noise ratiosabove background. With significant modification,flow cytometers have been used to analyze nanoparti-cles; however, proper discrimination of single parti-cles (<200 nm) from doublets or aggregates, basedon FSC signals alone, is a major challenge. Indeed,studies using fluorescently labeled Junin viruses(JUNV) yielded FSC and fluorescent signals thatmostly lie at the background noise level and thereforeindistinguishable from the bulk suspension.52 Toovercome this hurdle, researchers can rely onbrighter signals from fluorescently labeled particles,including cell-derived vesicles (d ≈ 100 nm),53 orconjugate magnetic nanoparticles to virions as in thecase of HIV-1 to perform so-called ‘flow virome-try’.54 Thus, it is very difficult to achieve label-freesensing of individual nanoparticles while preservingthe biological activity/infectiousness of viruses. Alter-natively, Zhu et al. has developed a highly sensitiveflow cytometry that allows detection of single silicaand gold nanoparticles as small as 24 and 7 nm indiameter, respectively, based on light scattering.55

This work is made possible by confining the sample

to a narrow sheath flow and isolation of the detec-tion region far away from the walls of the flow cellso as to reduce background scattered light. With thisdevelopment, they have also analyzed doxorubicin-carrying liposomes (d = 61 � 8 nm) and siRNA-loaded lipid nanoparticles. However, it remains to bedemonstrated if this technique will work for extracel-lular vesicles or viruses that have comparativelylow RIs.

CONCLUSION

The past several years have seen tremendousadvances in this fast-growing field of nanoparticledetection for various biological applications. This isimportant for sensing of viruses as well as characteri-zation of viral pathogens and drug deliverynanoparticles,56 just to name a few. Many techniqueswith single-nanoparticle sensitivity have been devel-oped and each technique has its own strengths andweaknesses. In the future, it is likely that differenttechniques will be used in different settings for thespecific application to be addressed. For example, RI-based measurements such as SPR and optical ringresonators are especially well suited for POC applica-tions. They require specific antibodies for capture ofthe nanoparticles but are highly sensitive, low-cost,and potentially high throughput. For quality controlof drug delivery nanoparticles, high throughput isalmost essential. In this regard, additional modifica-tions of existing flow cytometry techniques toincrease detection sensitivity for nanoparticles arelikely to be made. This development is much desiredgiven the multi-parametric analysis capability that isalready built-in among many commercial models. Apotential bottleneck along this development in flowcytometry, however, is the sensitivity for fluorescencedetection. To measure the heterogeneity amongnanoparticles, especially for biological particles suchas viruses that may display broad heterogeneity,single-molecule detection sensitivity is highly desired.Although this problem can be alleviated by engineer-ing brighter fluorophores, this bottleneck residespartly in the fluorescence detection scheme. Toachieve single-molecule sensitivity for particles influid, reduction of the background fluorescence isessential; consequently, confocal or multi-photonexcitation2 is very effective. However, due to thereduced excitation volume, in order to quantitatenanoparticle heterogeneity precisely, one has to guar-antee that the nanoparticle will travel through thelaser focus with minimal deviation. Whether this isfeasible with existing sheath fluid techniques remains



to be seen. Alternatively, other ideas may be testedsuch as finer control to trap single particles in solu-tion using hydrodynamic flow57 or through a feed-back control mechanism to apply corrective voltagesin the anti-Brownian electrokinetic trap.58

Many optical-based detection techniques fornanoparticles involve measurement of scattered lightfrom the nanoparticles of interest. One shouldbe cautioned that the scattered light from the nano-particle will be a function of both the particle’ssize and RI. These two parameters are linked forquantitative interpretation of the scattered light.Due to the lack of information on the RI of biologi-cal nanoparticles,59 many studies in the field haveadopted a convention to assume a RI for the particleof interest which has not been validated by

experimental measurements. Because uncertainties inthe particle’s RI directly propagate into the uncer-tainty of the particle’s size, a reliable measurement ofthe RI for these particles will be essential for accuratedata analysis. Thus, this will be an important areafor future study.

Lastly, as have been seen for several of the tech-niques that are capable of single-virus detection, theirsuccess also relies on the rational design of a micro-fluidic or nanofluidic device. These devices mightbe critical for measurement of nanoparticleswith high sensitivity and throughput as well as thepotential for manipulation and sorting of theseparticles. Accordingly, development of these devicesin parallel for nanoparticle detection is alsoenvisioned.

ACKNOWLEDGMENTS

This work was supported by NIH Director’s New Innovator Award 1DP2OD008693-01 to WC and also inpart by NSF CAREER Award CHE1149670 to WC. MCD is supported by a NIH postdoctoral fellowshipawarded under F32-GM109771.

FURTHER READINGFan X, White IM. Optofluidic microsystems for chemical and biological analysis. Nat Photon 2011, 5:591–597.

Ko J, Carpenter E, Issadore D. Detection and isolation of circulating exosomes and microvesicles for cancer monitoring anddiagnostics using micro-/nano-based devices. Analyst 2016, 141:450–460.

Marago OM, Jones PH, Gucciardi PG, Volpe G, Ferrari AC. Optical trapping and manipulation of nanostructures. NatNanotechnol 2013, 8:807–819.

Yurt A, Daaboul GG, Connor JH, Goldberg BB, Unlu MS. Single nanoparticle detectors for biological applications. Nano-scale 2012, 4:715–726.

REFERENCES1. Faraday M. The Bakerian Lecture: experimental rela-

tions of gold (and other metals) to light. Philos TransR Soc Lond 1857, 147:145–181.

2. Pang Y, Song H, Kim JH, Hou X, Cheng W. Opticaltrapping of individual human immunodeficiencyviruses in culture fluid reveals heterogeneity withsingle-molecule resolution. Nat Nanotechnol 2014,9:624–630.

3. Patolsky F, Zheng G, Hayden O, Lakadamyali M,Zhuang X, Lieber CM. Electrical detection of singleviruses. Proc Natl Acad Sci USA 2004,101:14017–14022.

4. Shirale DJ, Bangar MA, Park M, Yates MV, Chen W,Myung NV, Mulchandani A. Label-free chemiresistiveimmunosensors for viruses. Environ Sci Technol 2010,44:9030–9035.

5. Garcia-Aljaro C, Cella LN, Shirale DJ, Park M,Munoz FJ, Yates MV, Mulchandani A. Carbonnanotubes-based chemiresistive biosensors for detec-tion of microorganisms. Biosens Bioelectron 2010,26:1437–1441.

6. Mandal HS, Su Z, Ward A, Tang XS. Carbon nano-tube thin film biosensors for sensitive and reproduciblewhole virus detection. Theranostics 2012, 2:251–257.

7. Fraikin JL, Teesalu T, McKenney CM, Ruoslahti E,Cleland AN. A high-throughput label-free nanoparticleanalyser. Nat Nanotechnol 2011, 6:308–313.

8. Zhou K, Li L, Tan Z, Zlotnick A, Jacobson SC. Char-acterization of hepatitis B virus capsids by resistive-pulse sensing. J Am Chem Soc 2011, 133:1618–1621.

9. Harms ZD, Haywood DG, Kneller AR, SelzerL, Zlotnick A, Jacobson SC. Single-particle



electrophoresis in nanochannels. Anal Chem 2015, 87(92):699–705.

10. Maas SL, De Vrij J, Broekman ML. Quantificationand size-profiling of extracellular vesicles using tunableresistive pulse sensing. J Vis Exp 2014:e51623.

11. McMullen A, de Haan HW, Tang JX, Stein D. Stiff fil-amentous virus translocations through solid-statenanopores. Nat Commun 2014, 5:4171.

12. Hamblin MN, Xuan J, Maynes D, Tolley HD,Belnap DM, Woolley AT, Lee ML, Hawkins AR.Selective trapping and concentration of nanoparticlesand viruses in dual-height nanofluidic channels. LabChip 2010, 10:173–178.

13. Yamamoto T, Fujii T. Nanofluidic single-moleculesorting of DNA: a new concept in separation and anal-ysis of biomolecules towards ultimate level perfor-mance. Nanotechnology 2010, 21:395502.

14. Lee J, Shen W, Payer K, Burg TP, Manalis SR. Towardattogram mass measurements in solution with sus-pended nanochannel resonators. Nano Lett 2010,10:2537–2542.

15. Filipe V, Hawe A, Jiskoot W. Critical evaluation ofNanoparticle Tracking Analysis (NTA) by NanoSightfor the measurement of nanoparticles and proteinaggregates. Pharm Res 2010, 27:796–810.

16. Matsuoka H. Evanescent wave light scattering: afusion of the evanescent wave and light scattering tech-niques to the study of colloids and polymers near theinterface. Macromol Rapid 2001, 22:51–67.

17. Vieweger M, Goicochea N, Koh ES, Dragnea B. Photo-thermal imaging and measurement of protein shell stoi-chiometry of single HIV-1 Gag virus-likenanoparticles. ACS Nano 2011, 5:7324–7333.

18. Zhang N, Liu S, Wang K, Gu Z, Li M, Yi N, Xiao S,Song Q. Single nanoparticle detection using far-fieldemission of photonic molecule around the exceptionalpoint. Sci Rep 2015, 5:11912.

19. Pang L, Hwang GM, Slutsky B, Fainman Y. Spectralsensitivity of two-dimensional nanohole array surfaceplasmon polariton resonance sensor. Appl Phys Lett2007, 91:123112.

20. Wang S, Shan X, Patel U, Huang X, Lu J, Li J, Tao N.Label-free imaging, detection, and mass measurementof single viruses by surface plasmon resonance. ProcNatl Acad Sci USA 2010, 107:16028–16032.

21. Inci F, Tokel O, Wang S, Gurkan UA, Tasoglu S,Kuritzkes DR, Demirci U. Nanoplasmonic quantitativedetection of intact viruses from unprocessed wholeblood. ACS Nano 2013, 7:4733–4745.

22. Im H, Shao H, Park YI, Peterson VM, Castro CM,Weissleder R, Lee H. Label-free detection and molecu-lar profiling of exosomes with a nano-plasmonic sen-sor. Nat Biotechnol 2014, 32:490–495.

23. Zhu L, Wang K, Cui J, Liu H, Bu X, Ma H, Wang W,Gong H, Lausted C, Hood L, et al. Label-free

quantitative detection of tumor-derived exosomesthrough surface plasmon resonance imaging. AnalChem 2014, 86:8857–8864.

24. Grasso L, Wyss R, Weidenauer L, Thampi A,Demurtas D, Prudent M, Lion N, Vogel H. Molecularscreening of cancer-derived exosomes by surface plas-mon resonance spectroscopy. Anal Bioanal Chem2015, 407:5425–5432.

25. Vollmer F, Braun D, Libchaber A, Khoshsima M,Teraoka I, Arnold S. Protein detection by optical shiftof a resonant microcavity. Appl Phys Lett 2002,80:4057.

26. Li H, Fan X. Characterization of sensing capability ofoptofluidic ring resonator biosensors. Appl Phys Lett2010, 97:011105.

27. Vollmer F, Arnold S, Keng D. Single virus detectionfrom the reactive shift of a whispering-gallery mode.Proc Natl Acad Sci USA 2008, 105:20701–20704.

28. McClellan MS, Domier LL, Bailey RC. Label-free virusdetection using silicon photonic microring resonators.Biosens Bioelectron 2012, 31:388–392.

29. Zhu J, Ozdemir SK, He L, Chen DR, Yang L. Singlevirus and nanoparticle size spectrometry bywhispering-gallery-mode microcavities. Opt Express2011, 19:16195–16206.

30. Lu T, Lee H, Chen T, Herchak S, Kim J-H, Fraser SE,Flagan RC, Vahala K. High sensitivity nanoparticledetection using optical microcavities. Proc Natl AcadSci USA 2011, 108:5976–5979.

31. Ymeti A, Greve J, Lambeck PV, Wink T, vanHovell SW, Beumer TA, Wijn RR, Heideman RG,Subramaniam V, Kanger JS. Fast, ultrasensitive virusdetection using a Young interferometer sensor. NanoLett 2007, 7:394–397.

32. Mitra A, Deutsch B, Ignatovich F, Dykes C,Novotny L. Nano-optofluidic detection ofsingle viruses and nanoparticles. ACS Nano 2010,4:1305–1312.

33. Mitra A, Ignatovich F, Novotny L. Real-time opticaldetection of single human and bacterial viruses basedon dark-field interferometry. Biosens Bioelectron2012, 31:499–504.

34. Block O, Mitra A, Novotny L, Dykes C. A rapid label-free method for quantitation of human immunodefi-ciency virus type-1 particles by nanospectroscopy.J Virol Methods 2012, 182:70–75.

35. Daaboul GG, Yurt A, Zhang X, Hwang GM,Goldberg BB, Unlu MS. High-throughput detectionand sizing of individual low-index nanoparticles andviruses for pathogen identification. Nano Lett 2010,10:4727–4731.

36. Daaboul GG, Lopez CA, Chinnala J, Goldberg BB,Connor JH, Unlu MS. Digital sensing and sizing ofvesicular stomatitis virus pseudotypes in complex



media: a model for Ebola and Marburg detection. ACSNano 2014, 8:6047–6055.

37. Ashkin A. Acceleration and trapping of particles byradiation pressure. Phys Rev Lett 1970, 24:156–159.

38. Landry MP, McCall PM, Qi Z, Chemla YR. Charac-terization of photoactivated singlet oxygen damage insingle-molecule optical trap experiments. Biophys J2009, 97:2128–2136.

39. Neuman KC, Block SM. Optical trapping. Rev SciInstrum 2004, 75:2787–2809.

40. Cheng W, Hou X, Ye F. Use of tapered amplifier diodelaser for biological-friendly high-resolution opticaltrapping. Opt Lett 2010, 35:2988–2990.

41. Hou X, Cheng W. Single-molecule detection using con-tinuous wave excitation of two-photon fluorescence.Opt Lett 2011, 36:3185–3187.

42. Hou X, Cheng W. Detection of single fluorescent pro-teins inside eukaryotic cells using two-photon fluores-cence. Biomed Opt Express 2012, 3:1047–1049.

43. Ashkin A, Dziedzic JM. Optical trapping and manipu-lation of viruses and bacteria. Science 1987,235:1517–1520.

44. Bendix PM, Oddershede LB. Expanding the opticaltrapping range of lipid vesicles to the nanoscale. NanoLett 2011, 11:5431–5437.

45. Gittes F, Schmidt CF. Interference model for back-focal-plane displacement detection in optical tweezers.Opt Lett 1998, 23:7–9.

46. Toli�c-Nørrelykke SF, Schäffer E, Howard J,Pavone FS, Jülicher F, Flyvbjerg H. Calibration of opti-cal tweezers with positional detection in the back focalplane. Rev Sci Instrum 2006, 77:103101.

47. Zhu P, Liu J, Bess J Jr, Chertova E, Lifson JD,Grise H, Ofek GA, Taylor KA, Roux KH. Distributionand three-dimensional structure of AIDS virus envel-ope spikes. Nature 2006, 441:847–852.

48. Munro JB, Gorman J, Ma X, Zhou Z, Arthos J,Burton DR, Koff WC, Courter JR, Smith AB 3rd,Kwong PD, et al. Conformational dynamics of single

HIV-1 envelope trimers on the surface of native vir-ions. Science 2014, 346:759–763.

49. Berthelot J, Acimovic SS, Juan ML, Kreuzer MP,Renger J, Quidant R. Three-dimensional manipulationwith scanning near-field optical nanotweezers. NatNanotechnol 2014, 9:295–299.

50. Kang P, Schein P, Serey X, O’Dell D, Erickson D.Nanophotonic detection of freely interacting moleculeson a single influenza virus. Sci Rep 2015, 5:12087.

51. Fulwyler MJ. Electronic separation of biological cellsby volume. Science 1965, 150:910–911.

52. Gaudin R, Barteneva NS. Sorting of small infectiousvirus particles by flow virometry reveals distinct infec-tivity profiles. Nat Commun 2015, 6:6022.

53. van der Vlist EJ, Nolte-’t Hoen EN, Stoorvogel W,Arkesteijn GJ, Wauben MH. Fluorescent labeling ofnano-sized vesicles released by cells and subsequentquantitative and qualitative analysis by high-resolutionflow cytometry. Nat Protoc 2012, 7:1311–1326.

54. Arakelyan A, Fitzgerald W, Margolis L, Grivel JC.Nanoparticle-based flow virometry for the analysis ofindividual virions. J Clin Invest 2013, 123:3716–3727.

55. Zhu S, Ma L, Wang S, Chen C, Zhang W, Yang L,Hang W, Nolan JP, Wu L, Yan X. Light-scatteringdetection below the level of single fluorescent mole-cules for high-resolution characterization of functionalnanoparticles. ACS Nano 2014, 8:10998–11006.

56. Cressey D. Tiny traits cause big headaches. Nature2010, 467:264–265.

57. Tanyeri M, Schroeder CM. Manipulation and confine-ment of single particles using fluid flow. Nano Lett2013, 13:2357–2364.

58. Cohen AE, Moerner WE. Suppressing Brownianmotion of individual biomolecules in solution. ProcNatl Acad Sci USA 2006, 103:4362–4365.

59. van der Pol E, Coumans FA, Sturk A, Nieuwland R,van Leeuwen TG. Refractive index determination ofnanoparticles in suspension using nanoparticle track-ing analysis. Nano Lett 2014, 14:6195–6201.



Documents

Label-free detection and manipulation of single biological ...chenglab/pdf/DeSantisCheng2016WIRES.pdfAdvanced Review Label-free detection and manipulation of single biological nanoparticles