Digital Bioassays: Theory, Applications, and Perspectives - Analytical

Nov 15, 2016 - His primary research interests include the development of microtools for directed evolution of proteins, cell-free protein synthesis, a...
1 downloads 10 Views 2MB Size
Review pubs.acs.org/ac

Digital Bioassays: Theory, Applications, and Perspectives Yi Zhang† and Hiroyuki Noji*,†,‡ †

Department of Applied Chemistry, School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan ‡ Japan Science and Technology Agency, Tokyo 102-0076, Japan



CONTENTS

Theoretical Basis of Digital Counting in Bioassays Compartmentalization Methods Digital PCR Digital ELISA Dynamic Range of Digital Bioassay Multiplexed Digital Bioassay New Trends in Compartmentalization of Digital Bioassay Digital Bioassay as a Tool for Single-Molecule Biophysics Perspectives Conclusion Author Information Corresponding Author ORCID Author Contributions Notes Biographies Acknowledgments References

signal relative to concentration and is most closely associated with the slope of the calibration curves, which determines the ability to discriminate among similar concentrations of the analyte. It is always accompanied by a parameter called “limit of detection (LOD)”, and the term “sensitive” in this Review as well as in other research articles generally means a low LOD.1 In contrast to nucleic acids, which can be exponentially amplified, proteins cannot be utilized as a template for their replication, making ultrasensitive detection of proteins highly challenging.2 Researchers achieved a superior limit of detection of zeptomolar (10−21 M) concentration in 1979, by combining the advantages of radioimmunoassays and enzyme-linked immunosorbent assays (ELISA).3 This significant achievement represents an important milestone; however, the developed method was limited in terms of its applications in the laboratory owing to its complexity and potential health risks associated with the use of radiolabels. Nanowire field-effect transistor (NW-FET) is another excellent example to illustrate what the sensitivity is by improving the signal-to-noise ratio (S/N). In contrast to numerous approaches that enhance the absolute value of detection signals, the absolute intensity of the signal output of NW-FETs is intrinsically weak; however, the success of this approach in detecting single protein molecules depends on the superior S/N of nanowires. The utility of this method has been demonstrated for single-virus detection (by targeting surface proteins on viral particles) and multiplexed tumor biomarker detection,4,5 where the specific binding of single target molecules results in ultrasensitive changes in conductance of nanowires. A commonly used technology for single-molecule detection is total internal reflection fluorescence (TIRF) microscopy.6 A recent report claimed that TIRF combined with coimmunoprecipitation (i.e., antigen−antibody interaction) enables direct visualization of the protein of interest with single-molecule resolution, thereby allowing quantification of the number of protein molecules.7 However, diffraction-limited optics only functions well within a narrow dynamic range (generally at picomolar to nanomolar concentrations) while biomolecular interactions often require concentrations higher than μM (dissociation constant, KD, or Michaelis−Menten constant, KM); this drawback severely limits the application of TIRF-based approaches.8 To overcome the concentration barrier, the “zero-mode waveguide (ZMW)” technique has been developed by using subdiffraction-limited volumes,9 which enabled the analysis of individual molecular events at high concentrations (i.e., representative of physio-

93 94 94 95 96 97 97 97 98 99 99 99 99 99 99 99 100 100

T

he development of bioassays with high sensitivity has received significant attention. Enormous efforts based on a variety of signal amplification techniques have been made for improving the sensitivity of bioassays; however, most of the bioassays developed are only suitable for use in a very limited range of applications. In contrast to conventional bioassays that are commonly performed in a single reactor (tubes, microtiter plates, etc.), the reaction solution is partitioned into a large number of microreactors in digital bioassays, allowing most compartments to be loaded with 0 or 1 target molecule. Because of the binary property of the system, it is called the “digital” bioassay in this Review. The concept of the digital bioassay is unique and easy to understand, while the enrichment effect of the compartmentalization of the reaction solution is immediate and general for all types of bioassays. Digital bioassays are compatible with existing classical techniques widely used in normal biological, chemical, or clinical laboratories. More importantly, this novel research tool has been demonstrated to dramatically reduce labor intensity, cost, and the difficulties associated with the use of conventional analytical methods in biophysical and biochemical studies. This Review provides a comprehensive introduction, from the concept to the practice, of the digital bioassay. Toward quantitative and accurate molecular diagnostics, numerous strategies have been proposed to develop assays with high sensitivity. Sensitivity is officially defined as the change in © 2016 American Chemical Society

Special Issue: Fundamental and Applied Reviews in Analytical Chemistry 2017 Published: November 15, 2016 92

DOI: 10.1021/acs.analchem.6b04290 Anal. Chem. 2017, 89, 92−101

Analytical Chemistry

Review

logical conditions).10 However, the powerful applications of ZMW are real-time sequencing of single DNA molecules and real-time monitoring of protein translation,11,12 rather than the quantification of nucleic acids or proteins in a sample. Because of the relatively simple instrumentation of digital bioassays, which only require partitioning solution into a small confined space, and the solid foundation of quantification in digital bioassays, this method is of great potential utility in practical applications such as rapid diagnosis, with higher prospects for successful commercialization compared with other approaches.

flexibility in practice, as long as m is large enough to ensure the premise of Poisson approximation. This is why digital bioassays are performed with a large number of microcompartments. In addition, note that the parameter k > 1 indicates that several target molecules enter the same microcompartment simultaneously. All of the above describes the physical meaning of the Poisson distribution function in the context of the digital bioassay. The ability to quantify target molecules in a digital bioassay system is enabled by further mathematical derivation. Table 1

THEORETICAL BASIS OF DIGITAL COUNTING IN BIOASSAYS The facile detection of biomolecules with high sensitivity represents the main aim of researchers in the field of bioassay development. The operation of digital bioassays involves partitioning reaction solutions into micrometer-sized compartments. Instead of quantifying the absolute intensity of ensemble signals from tubes or microtiter plates, in digital bioassays, only the fraction of microcompartments showing positive signals is counted. The concept is simple, and the theoretical basis for quantification is solid. The selection of an appropriate mathematical model for describing an object requires careful consideration to give each parameter in the model an explicit physical meaning. A digital bioassay is composed of a large number of identical and independent reactions, with each showing either a positive or a negative signal (i.e., exactly two possible outcomes), satisfying the prerequisites of Bernoulli trials (or binomial trials); therefore, the digital bioassay should be modeled using Poisson distribution, a limiting case of binomial distribution. In the digital bioassay, an assay solution can be considered to be divided into “m” microcompartments with equal small volumes, “v”. Individual reactions are carried out simultaneously in such microcompartments. In this context, “event A” of the binomial distribution can be defined as follows: a given molecule is in a given microcompartment. The probability of event A will be p = 1/m. The total number of target molecules involved in the system, “n”, is just like the number of trials in a Bernoulli process. The probability “P” of event A occurring k times among n molecules obeys a binomial distribution, which is denoted as X ∼ B(n, p):

Table 1. Examples of Probability Calculation by Poisson Distribution



P(X = k) =

⎛n⎞ k (n − k) ⎜ ⎟p (1 − p) ⎝k ⎠

expected value λ 1

0.1

0.01

number of occurrences k

probability P(X = k), %

0 1 2 3 4 5 0 1 2 0 1

36.8 36.8 18.4 6.13 1.53 0.307 90.5 9.05 0.905 99.0 0.990

provides an intuitive insight: with the decrease of λ, i.e., with a decrease in concentration of the target molecules, the fraction of microcompartments showing negative signal P(X = 0) becomes predominant, whereas the fraction of microcompartments showing positive signal Ppositive = P(X ≥ 1) approaches P(X = 1), indicating that most microcompartments contain either 0 or 1 target molecules. Ppositive = 1 − P(X = 0) = 1 − e−λ

or

λ = −ln(1 − Ppositive)

Although the concentration, which is represented by λ, may be calculated using the equation above, a linear relationship between the analytical response (Ppositive) and the analyte concentration (λ) is preferred for convenience (Figure 1). Through a simple transformation based on a proven limit limx→0[−ln(1 − x)] = x, the linear relationship between Ppositive and λ can be represented as follows:

(k = 0, 1, 2, ..., n)

Note that P represents the probability to include k molecules in a microcompartment. The binomial distribution converges toward the Poisson distribution, which is denoted as X ∼ P(n, λ) with parameter λ = np (which can be proven mathematically13): ⎛n⎞ k λk (n − k) ⎜ ⎟p (1 − p) = e −λ n →∞ , p → 0⎝ k ⎠ k! lim

(λ = np , k = 0, 1, 2, ..., n)

Because λ = np = n/m, λ represents the average number of target molecules per one microcompartment. λ can alternatively be given by λ = cvNA, where c is the bulk molar concentration of the target molecule, v is the volume of a single microcompartment as described above, and NA is the Avogadro constant. This derivation implies that the result of digital counting is unrelated to the total number of microcompartments, as the auxiliary parameter m can be removed during the derivation. This property provides digital counting a degree of

Figure 1. Plot of function Ppositive = 1 − e−λ. The linearity between Ppositive and λ increases with the decrease in λ. However, there is no distinct boundary between the so-called linear region and the nonlinear region. 93

DOI: 10.1021/acs.analchem.6b04290 Anal. Chem. 2017, 89, 92−101

Analytical Chemistry

Review

Figure 2. Concept of the digital bioassay. A schematic comparison between analog measurement and digital counting is shown. In digital bioassays (bottom part), the bulk reaction solution is partitioned into extremely small compartments to rapidly concentrate the reaction product. In conventional tube-based assays (top part), the reaction product diffuses very quickly, making a highly diluted product difficult to detect. The exemplary concentration is calculated on the basis of a microcompartment volume of 1 fL (corresponding to height, width, and depth of 1 μm).

lim

Ppositive → 0

λ = Ppositive

Emulsification and droplet microfluidics are the mainstream techniques for generating a large number of droplets efficiently.16,17 In general, aqueous droplets are generated in an oil phase and further transferred into other zones for reaction, coalescence, and sorting. An alternative strategy involved the generation of physically isolated microcompartments in a microfluidic chip using pneumatic pressure,18 hydraulic pressure,19 or oil sealing.20,21 Most of these methods require external equipment or manual intervention. In order to simplify the fluid operation for end users, a degassing-driven self-pumping mechanism was adopted to introduce reaction solution into the device automatically, with minimal manual intervention.22 This principle has been previously proposed:23 a gas-permeable elastomer is degassed in a vacuum and the redissolution of air through the material provides the driving force for the liquid movement when the device is brought back to the atmosphere. The handling of nano-, pico-, or even femtoliter droplets using microfluidic devices provides a convenient way to minimize sample consumption.24 In addition to the facile preparation, as the constant probability p of event A implies, the maintenance of equal volume in each compartment is crucial for quantitative measurement. As an initial attempt, water-in-oil emulsion was used to detect single enzyme molecules, enabling the calculation of the activity as well as the kinetics of the enzyme.25 However, the large size distribution resulted in great inconvenience during data analysis; as a result, some of the data could not be interpreted accurately. On the basis of the microelectro-mechanical system (MEMS) and soft-lithography, the uniform elastomer microchamber array was developed to enclose femtoliter-level quantities of solution, allowing us to measure the activity of single enzyme molecules as well as enzyme concentration.26 Subsequently, several other methods were proposed for the preparation of uniform microchambers.27,28 Eventually, these advancements formed the direct basis of digital ELISA, which can be utilized for the detection of both enzyme and other proteins recognized by enzyme-conjugated antibodies.

As described above, the average number of target molecules per one microcompartment is given by λ = cvNA. Therefore, the concentration c can be calculated using the equation: c = Ppositive/(vNA)

Therefore, from the series of strict approximations described above, we arrive at the clear conclusion that the concentration of the target molecule is proportional to the fraction of positive microcompartments in the digital bioassay system. This approximation also reveals that the digital counting approach is particularly effective for measuring analytes with low concentrations, which represents an intrinsic advantage of the digital bioassay. It should be noted that there is no distinct threshold between digitally quantifiable regions and digitally nonquantifiable regions (Figure 1), although several reports have empirically specified a threshold.14 Following appropriate dilution, target molecules with high concentrations can be quantified using the digital bioassay approach. The Poisson distribution provides an elegant model for the digital bioassay. On the basis of the theoretical foundation of digital counting, interesting outcomes may be predicted accurately even without experiments; this is described further in the following sections.



COMPARTMENTALIZATION METHODS The digital bioassay requires reaction solutions to be partitioned into equal small volumes. The reaction product accumulates inside the microcompartment after the solution is partitioned (Figure 2). Although this concept was proposed in the early 1990s for absolute quantification of the number of DNA molecules in a sample,15 it was initially limited by the large amount of reagent consumed by conventional microtiter plates. However, digital bioassays have become more widespread following the emergence of technologies for the preparation of microreactors. To date, several platforms have been established in industry, and various improvements and variations have been proposed for the application of digital bioassays in academic research.



DIGITAL PCR The development of digital PCR reflects the strong focus on, and enormous efforts involved in, transforming the exponential 94

DOI: 10.1021/acs.analchem.6b04290 Anal. Chem. 2017, 89, 92−101

Analytical Chemistry

Review

stool, demonstrating the great potential of this method for laboratory-based applications and clinical diagnosis using a variety of target samples. The integration of individual functional units makes the digital amplification system more and more practical (e.g., sample-to-answer system) and affordable (e.g., electricity-free heating).

nature of PCR into a linear output. Digital PCR has already been commercialized successfully and is an established alternative for the absolute quantification of DNA or RNA.29 Prior to the development of this technique, the only available method for such purposes was real-time PCR. It is accepted that digital PCR was developed in 1992; however, this technique did not acquire its present name until later.15 Although the total consumption of reagent was much higher than in normal PCR, this pioneering work proposed for the first time the concept of an “all-or-none” reaction scheme using Poisson statistics as a model. Considering PCR technology itself (in particular, the discovery of thermostable DNA polymerase) had only been invented in the late 1980s, the development of digital PCR represented a significant advancement.30 Over the subsequent 20 years, the development of digital PCR focused on the use of larger numbers of, and smaller solution volumes in, partitioned compartments with greater reliability and homogeneity. The first commercial digital PCR instrument was launched by Fluidigm, a university-launched venture company that developed microfluidic chips integrated with large-scale microvalves and micropumps.31,32 Subsequently, several other world leading biotechnology companies, such as Life Technologies (a retired brand of Thermo Fisher Scientific), Bio-Rad, and RainDance Technologies, have launched products with novel technical features; the technology roadmap has been well summarized elsewhere.33 The commercial products developed to date utilize partition volumes, ranging from nanoliter to picoliter scales, with total numbers of microcompartments ranging from 104 to 107. Even smaller volumes have already been achieved, to tens of femtoliters, in digital PCR at the laboratory level.19,34 The motivation for shrinking reaction volumes is to concentrate the small amount of reaction product, allowing the output signal to be detected more rapidly than via tubes or microtiter platebased methods (Figure 2). Accordingly, the ability to perform digital PCR using even smaller reaction volumes should be beneficial for slow biochemical reactions, as well as for overcoming the aforementioned issue of the upper limit of concentration in single-molecule experiments.9 Furthermore, the precision of Poisson approximation (i.e., the Poisson noise) should be improved by increasing the number of counted partitions. In general, PCR is carried out through a series of temperature steps consisting of repeated heating and cooling, which is termed thermocycling. Advancements in the mode of heating have dramatically reduced the complexity of the entire system, which usually involves sacrificing minimal space for time to allow PCR to be carried out in a series of individual heating zones, each of which has a constant temperature.35,36 Another useful strategy, distinct from the above, is isothermal amplification technology,37−39 by which the target nucleic acid can be amplified at a constant temperature. Digital amplification can be implemented in isothermal conditions. Digital recombinase polymerase amplification (RPA) was demonstrated using SlipChip.40 Digital loop-mediated isothermal amplification (LAMP) was demonstrated using a selfdigitization chip,41 a self-priming chip,22 and a continuous flow microfluidic device.42 It is critical to prevent evaporation during the heating process, in both PCR and isothermal amplification.21,42 Rare gene mutations, differential gene expression, and copy number variations have been detected and quantified by digital PCR in various specimens, such as blood, serum, and



DIGITAL ELISA

Because proteins cannot be amplified like DNA, achieving single-molecule detection is more challenging. This challenge has been addressed by the development of digital ELISA. The sensitivity enabled by digital ELISA in some cases is in good concordance with theoretical sensitivity derived from parameters such as concentration, time, and affinity,43,44 making it possible to approach the theoretical limit of detection via an appropriate experimental setup. Digital counting, as a quantitation method, was first demonstrated for enzymes that catalyze the hydrolysis of fluorogenic substrates to produce a directly measurable signal.26 ELISA detects target proteins via an enzyme-conjugated antibody specific to the target. Combining ELISA with the digital counting technique makes the quantification of any kind of protein possible. Specifically, here, ELISA refers to a sandwich ELISA that detects a specific target with a capture and detection antibody pair. The enzyme conjugated to the detection antibody catalyzes the fluorogenic reaction, producing a fluorescent product and allowing the signal to be detected (Figure 3). Digital ELISA has since been successfully applied to the detection of prostate-specific antigen (PSA) in serum at subfemtomolar concentration, which is much lower than that used for conventional ELISA.45 This achievement, which

Figure 3. Schematic representation of digital ELISA. (a) Single target protein molecules are captured by an antibody on the bead and detected by another antibody conjugated with an enzyme molecule. (b) Large amounts of beads, with or without the immunocomplex, are loaded into microchambers for fluorescence imaging. The enzymatic reaction results in the accumulation of the fluorescent product inside the microchamber to allow fluorescence imaging. (Reprinted with permission from Rissin, D. M.; Kan, C. W.; Campbell, T. G.; Howes, S. C.; Fournier, D. R.; Song, L.; Piech, T.; Patel, P. P.; Chang, L.; Rivnak, A. J.; Ferrell, E. P.; Randall, J. D.; Provuncher, G. K.; Walt, D. R.; Duffy, D. C., Nat. Biotechnol. 2010, 28, 595−599 (ref 45). Copyright 2010 Macmillan Publishers Ltd.) 95

DOI: 10.1021/acs.analchem.6b04290 Anal. Chem. 2017, 89, 92−101

Analytical Chemistry

Review

cross-reactivity must be suppressed as much as possible in this kind of assay.

indicated that sensitivity could be improved without significant changes in the biochemical properties based on ligand− receptor binding affinity, represented an important milestone in analytical chemistry. Digital ELISA directly utilizes the wellestablished antigen−antibody interaction, making the transition from conventional ELISA to digital ELISA very easy for end users and laboratories. The first digital ELISA system suffered from several limitations, thus requiring further improvement. The first reported digital ELISA utilized 5 × 104 microchambers and yielded a large coefficient of variation (CV) for low concentrations of target molecules.45 This was attributed to the limited number of positive microchambers that could be counted in such situation, which determines the LOD of the assay system. Increasing the number of microchambers in this system was challenging as the dimensions of the cross section of commercially available optical fiber bundles were fixed. In our independent work, 106 microchambers were utilized and a 25-fold improved LOD and a 12-fold decreased CV were demonstrated.46 Mechanical sealing using a sheet of silicon rubber is also difficult in practice. We therefore proposed a facile and robust oil-sealing strategy for a large array of microchambers.46 This system provided high flexibility by making arbitrary numbers and dimensions of microchambers available. The relatively low technical barrier accelerated the development of integrated instruments for digital ELISA. The first demonstration instrument integrated chip-loading unit, actuator valves, moving stage, LED light source, CCD camera, filter cubes, touch panel, microcomputer, power source, and input/ output ports.47 After the saliva sample and necessary reagents were loaded into the designated reservoir, the chip was loaded onto the instrument and fluid operations and imaging were automatically performed. Six targets were detected using differentially encoded beads in one chip. Subsequently, Quanterix launched a fully automated instrument for multiplexed digital ELISA,48 which adopted an array disc instead of optical fiber bundles28 and demonstrated its application for detection of various kinds of clinically relevant targets.48,49 Owing to the rapid growth of instrumentation, digital ELISA has been widely utilized in clinical diagnosis. In addition to PSA, whose detection has been reported by numerous research papers, to date, over tens of biomarkers, such as Tau,50 Clostridium dif ficile toxins,51 HIV p24 antigen,48 dengue virus IgG and IgM,52 and cytokines,49 have been detected using digital ELISA. The unprecedented sensitivity of this method not only enables prompt diagnosis but also provides novel insights into disease progression. The basic setup of digital ELISA has remained almost the same in various applications. For example, the only choice for the enzyme tag is β-galactosidase. However, any enzyme or substrate, including β-galactosidase, only functions optimally in specific buffer components and a certain pH range. Furthermore, the tetramer property of this large enzyme may make it unsuitable for use in some cases. Therefore, the development of new tags for detection is necessary. In addition to β-galactosidase, several enzymes produce a detectable signal by catalyzing their corresponding substrate. We have shown that alkaline phosphatase, which exhibits optimal activity under alkaline conditions, is a potential candidate enzyme tag.53 In addition, the basic framework of sandwich immunoassays intrinsically requires appropriate antibody pairs. This is particularly important for the multiplexed digital ELISA,54 as



DYNAMIC RANGE OF DIGITAL BIOASSAY

In addition to the sensitivity (LOD) and precision (CV), dynamic range (the range between the minimum and maximum number/concentration of molecules in a sample that can be measured quantitatively) is another important performance metric for any kind of analytical method, including the digital bioassay. The larger the number of microcompartments, the greater the number of molecules accommodated. A reasonable theoretical estimation of dynamic range is given by the concentration beyond which the array becomes completely saturated,21 which indicates that the dynamic range depends solely on the total number of counted microcompartments. Accordingly, we and other groups have shown experimentally that the dynamic range of both digital PCR and digital ELISA can be improved by increasing the number of microcompartments.21,46 However, in practice, the number of microcompartments can only be increased to a limited extent, mainly due to the consequently increased difficulties in preparation or subsequent handling. Several alternative strategies have been proposed to improve the dynamic range without increasing the number of microcompartments.14,55−57 The combination of the detection of single molecules (i.e., digital detection) and an ensemble of molecules (referred to as analog detection) has been shown to extend the working range of the same microchamber array device from the original digital regime to the analog regime. The target protein, at high concentrations, was quantified by introducing a newly defined unit, namely, the average number of enzyme labels per bead.14 Because the unit as well as the quantification was based on the relative intensity of the fluorescence signal, this did not strictly represent a truly high dynamic range; i.e., changing the working range of the device does not change the dynamic range in its digital regime. The application of multivolume microchamber devices is another strategy to improve the dynamic range.55 The mathematical model utilized combined multiple binomial distributions (one for each volume). This strategy was validated to be only effective for digital PCR; this may be because minimization of the inevitable variation in signal intensity of microchambers with highly different volumes is necessary, which is only possible via the exponential amplification enabled by PCR.56 Another strategy based on a concept termed “Brownian trapping with drift” was proposed to improve the dynamic range with uniform microchambers.57 Each microchamber was preloaded with a single bead modified with capture antibodies. The subsequently introduced target molecules with high concentrations drifted and became trapped with an exponentially decaying distribution along the direction of flow, enabling a spatially resolved readout for quantification. It was critical that the exponential decay only occurred when the trapping surface was not saturated by the target during the experiment. As a consequence, however, large changes in the input concentration gave rise to logarithmically smaller changes in the output, as a result of which a linear relationship (the most important characteristic of digital bioassays) between input (analyte concentration) and output (fraction of microcompartment showing positive signals) could not be achieved. 96

DOI: 10.1021/acs.analchem.6b04290 Anal. Chem. 2017, 89, 92−101

Analytical Chemistry



Review

chamber array device compatible with phospholipids.75 The lipid bilayer membranes were formed on the top orifice of the microchambers via sequential injection of aqueous solution and organic solvent (Figure 4a). The successful formation of a lipid

MULTIPLEXED DIGITAL BIOASSAY The multiplexed digital bioassay system is another focus of significant attention in the field of analytical chemistry. Existing strategies to achieve a multiplexed assay can be roughly classified into two categories: labeling and encoding. We are relatively familiar with the labeling strategy, which distinguishes different target analytes by differentially labeled molecules (e.g., organic dyes, quantum dots, etc.). The so-called encoding strategy, such as color-encoding,58−60 wavelength- and intensity-encoding,61−64 compositional encoding,65 shape-encoding,66,67 dot-encoding,68 and positional encoding,69 is relatively new and still under development. The encoding strategy enables circumvention of the problem of spectral overlap, hence dramatically increasing the variety of target analytes that may be distinguished via a single assay. The application of multiplexed digital ELISA is accompanied by several challenges, such as determining whether a carrier bead is present in the microcompartment, identifying the type of beads targeting different target proteins, and determining whether a target molecule is captured by the bead. Although some of the aforementioned approaches are possible, to date, only the wavelength- and intensity-encoding strategy has been tested.70,71 In brief, the beads were spiked with different fluorescent dyes or with the same dye in different ratios. The differentially labeled beads were further conjugated with different capture antibodies. The different target proteins can be captured by corresponding beads and then detected by corresponding detection antibodies conjugated either with a fluorophore or with an enzyme reporter that catalyzes hydrolysis of a fluorogenic substrate. This enables detection of any target available with one color (from the detection antibody) and identification of each target available with corresponding colors (from the encoding beads). However, due to the inevitable spectral overlapping of fluorophores, crossreactivity of antibodies, and the nonspecific binding of the enzyme reporter on the beads, the demonstrated multiplexing capability has been only 4 to 6 targets in a single assay to date. In addition, the multiplexed digital PCR demonstrated multiplexing capability similar to that of the multiplexed digital ELISA.72 Probes targeting different target sequences were labeled with blends of two fluorophores in different ratios, making differentiation of specific PCR reactions possible. Each PCR amplicon finally clustered as a spot on a 2D heat map with a horizontal axis of the intensity of one fluorophore and a vertical axis of the intensity of the other fluorophore. Luminex xMAP, a commercially developed platform, has already achieved a multiplexing capability of up to 500 using the described wavelength- and intensity-encoding strategy for bead modification,73 suggesting the continued development of the multiplexed digital bioassays with higher multiplexing capability in the future.

Figure 4. Arrayed lipid bilayer microchamber system. (a) Formation of a lipid bilayer membrane through sequential liquid injection. (b) Passive transport of fluorescent dye by α-hemolysin. (c) Active transport of protons by FoF1-ATP synthase. (Adapted with permission from Watanabe, R.; Soga, N.; Fujita, D.; Tabata, K. V.; Yamauchi, L.; Kim, S. H.; Asanuma, D.; Kamiya, M.; Urano, Y.; Suga, H.; Noji, H., Nat. Commun. 2014, 5, 4519 (ref 75). Copyright 2014 Macmillan Publishers Ltd.)

bilayer is a prerequisite for membrane protein function and, accordingly, the normal functioning of the membrane protein is evidence for the existence of lipid bilayer. As a proof-of-concept demonstration based on the Poisson statistics, the pumping activity of both passive (α-hemolysin) and active transporter (FoF1-ATP synthase) was confirmed at the single-molecule level (Figure 4b,c). Another lipid can be injected following the injection of the first lipid to create a lipid bilayer array with an asymmetric lipid composition.76 Alternatively, the lipid bilayer can be formed by the mechanical contact of two water-in-oil droplets encapsulated by a lipid monolayer.77 The compositional asymmetry of lipid is an intrinsic property of the biomembrane; therefore, these new approaches enable the study of the membrane protein in an environment representative of physiological conditions.





NEW TRENDS IN COMPARTMENTALIZATION OF DIGITAL BIOASSAY Most integral membrane protein complexes can fold and function only in the presence of a lipid bilayer environment. Owing to their central role in cellular signaling and potential value in clinical treatment, membrane proteins represent important drug targets.74 It is necessary and possible to evaluate the activity of membrane proteins at the singlemolecule level; however, previous efforts have suffered from low throughput. Therefore, we developed a femtoliter micro-

DIGITAL BIOASSAY AS A TOOL FOR SINGLE-MOLECULE BIOPHYSICS Molecular detection is sometimes the purpose of a study; however, this may also be required as a means or a tool for other studies. Digital bioassays serve as irreplaceable tools in basic biophysical research. The biggest difference between the digital bioassay and other analytical methods is that biochemical reactions are carried out in a very small confined space. The size of the reactor is larger than the resolution of the optical 97

DOI: 10.1021/acs.analchem.6b04290 Anal. Chem. 2017, 89, 92−101

Analytical Chemistry

Review

The determination of enzyme activity is one of the primary aims of enzymology studies. The measured activity of single enzyme molecules was used to derive the rate of enzyme production from single cells in real-time.79 Interestingly, the production of low-copy number proteins occurs in bursts, with the number of molecules per burst following an exponential distribution. The heterogeneity as well as fluctuation of individual enzyme molecules has also been extensively investigated via single-molecule enzymology studies.80−84 Confinement by the microchamber provides a facile means to assess the behavior of individual enzyme molecules. In addition, enzyme molecules enclosed in microchambers appear to move freely in solution;26 this is considered beneficial for maximizing protein activity, which may be largely affected by steric hindrance.85 It is well accepted that the fluctuation phenomena may result from transient conformational changes in protein structure or post-translational modifications. Using the microchamber array system, the catalytic turnover of β-galactosidase during the hydrolysis of resorufin-β-D-galactopyranoside was shown to be highly distributed and long-lived, suggesting the presence of stable activity states over long time periods.86 By combining a heating stage with the microchamber array, changes in activity of single β-galactosidase molecules subjected to short heating pulses were monitored in real time.87 The changes in activity were random, and the altered activities were stable between heat pulses for every single molecule. This suggested the possibility of heat-induced conformational changes and different stable confirmations that may be partially responsible for the heterogeneity of enzyme activity. In contrast to β-galactosidase, whose bulk activities were in good agreement with the single-molecule activities, the apparent substrate turnover rate of horseradish peroxidase in the femtoliter microchambers was about 10 times lower than that in bulk; this was attributed to the different two-step redox reaction mechanism of this enzyme.88,89 The large surface-tovolume ratio and the microchamber material were suspected to lead to surface-mediated side reactions of intermediate radicals. In addition to the wild-type enzymes, artificially evolved enzymes were also tested in the same device.90 A partially evolved β-glucuronidase showed a much broader activity distribution among individual enzyme molecules than the wild-type β-glucuronidase, which provided insights into the mechanisms driving the evolution of new enzyme activity. A larger variety of enzymes and substrates should be tested in this system in the future; this is expected to provide statistically significant evidence regarding the heterogeneity of single enzyme molecules.

microscope, and the increase in product concentration occurs at a much faster rate, making sensitive detection by standard microscopy within a reasonable time scale possible. The small change in reactor dimension has enabled new applications that were previously inaccessible. Reaction products can be rapidly enriched in extremely small reactors. To take full advantage of this effect, we performed fine magnetic manipulation of single F1-ATPase in femtoliter microchambers.78 In contrast to the physiological function of ATP synthesis of F1 in the FoF1 complex, the isolated F1 spontaneously rotates in an anticlockwise direction as it hydrolyzes ATP. The single F1 molecule labeled with a magnetic bead was immobilized on the bottom surface and encapsulated into the microchamber, and its rotation was forced into the clockwise direction using a rotating magnetic field, resulting in ATP synthesis (Figure 5b). When the

Figure 5. ATP synthesis by F1-ATPase. (a) Schematic view of the membrane-embedded FoF1-ATP synthase in which the proton-driven Fo rotates F1 in a clockwise direction for ATP synthesis; in contrast, the isolated F1 only hydrolyzes ATP in the reverse direction. (b) To force the ATP synthesis of F1 against the chemical potential, a magnetic bead was attached to the γ-subunit of a single immobilized F1 and rotated clockwise using magnetic tweezers. The bead−enzyme complex was enclosed in a femtoliter microchamber in order to enrich synthesized ATP molecules. (c) Repeated magnetic trapping and release of single F1 molecules; after the magnetic field was switched off, F1 resumed its ATP-hydrolyzing anticlockwise rotation at a speed proportional to the concentration of ATP generated during the period of forced rotation in the microchamber. (Reprinted with permission from Rondelez, Y.; Tresset, G.; Nakashima, T.; Kato-Yamada, Y.; Fujita, H.; Takeuchi, S.; Noji, H., Nature 2005, 433, 773−777 (ref 78). Copyright 2005 Macmillan Publishers Ltd.)



PERSPECTIVES It should be noted that the term “single-molecule sensitivity” does not represent a limit of detection of one molecule, particularly in “ligand-receptor binding” assays; “singlemolecule responsivity” may be a more accurate term. The pursuit of ultimate sensitivity is ceaseless. On the basis of existing studies and estimations,45 the significant improvement of the limit of detection may be achieved by eliminating the nonspecific binding of biomacromolecules, for either bulk measurement or digital counting. Although some strategies based on surface passivation have been proposed,91,92 their effectiveness has not been demonstrated, owing to the lack of knowledge of the mechanisms underlying the interaction between biomacromolecules and surfaces. A breakthrough in

magnetic field was turned off, F1 resumed its anticlockwise rotation at a speed proportional to the concentration of ATP generated during the period of forced rotation (Figure 5c). This was the first direct observation and assessment of mechanochemical coupling at single-molecule resolution, revealing a nearly perfect coupling efficiency of this molecular machine. In this study, energy conversion considered ubiquitous in physics was perfectly represented in the biological world. 98

DOI: 10.1021/acs.analchem.6b04290 Anal. Chem. 2017, 89, 92−101

Analytical Chemistry

Review

ensure that the assay is fully compatible with the water−oil hybrid system of large surface area. It should be noted that there are few studies on nonfluorescent digital bioassays. Therefore, a game-changing methodology is highly anticipated, and it is expected that overcoming these challenges will require multidisciplinary collaboration.

understanding the phenomenon of nonspecific binding is crucial. In analogy with the development roadmap of conventional bioassays, applications of digital bioassays, that have been unexplored to date, may be identified. For example, a digital bioassay for the detection of small molecules (termed haptens) has not yet been developed. Haptens are too small for simultaneous binding with two antibodies, which is required by the sandwich immunoassay, to occur. Several strategies, such as open sandwich ELISA for hapten detection,93 have already been proposed for this purpose in bulk measurement,94 but hapten detection has never been demonstrated in digital bioassays. Hapten detection is important for environmental monitoring and food safety, as well as clinical diagnosis; accordingly, we expect that digital bioassays capable of small molecule analysis will be developed in the near future. State-of-the-art digital bioassays are carried out in micrometer-sized reactors with femtoliter volumes such as microchambers and water-in-oil droplets. Because of the limited ability of microfabrication technologies, further reduction of reactor dimensions is difficult and costly. Several alternative approaches have been proposed to prepare even smaller and more homogeneous reactors in a high-throughput format such as electrospun nanofiber junctions with attoliter volumes,95 virus particles with zeptoliter volumes,96 and DNA-nanostructure templated liposomes with subattoliter volumes.97 These concepts are promising; however, their applicability for versatile applications has yet to be demonstrated. Increasing the total number of microcompartments is a general and effective way to improve the LOD, CV, and dynamic range. It is necessary to take into account the signal acquisition as well as imaging for a large number of microcompartments with dimensions approaching the limit of specific optical settings; this is a general requirement for all imaging experiments. There is a trade-off between resolution, sensitivity, area size, uniformity, and cost. It has been demonstrated that one million picoliter droplets of digital PCR were captured in a single snapshot by an optimally assembled wide-field fluorescence imaging system similar to some commercial stereomicroscopes;98 however, the sensitivity and image quality were compromised. We have developed a more compact lens-less complementary metal-oxide-semiconductor (CMOS) image sensor integrated with femtoliter microchamber array for digital counting as well as digital ELISA, providing a miniaturized imaging system suitable for resource-limited settings.99,100 This system enables discrimination between fluorescent and nonfluorescent spots in a large field of view; however, the resolution was compromised to the limit where one chamber is only represented by one pixel. Unless significant innovation in imaging technologies occurs, a balance between the needs of each specific imaging experiment and quality of data obtained is required. Unlike digital PCR in which target sequences are amplified by a general polymerase, digital bioassays for proteins are highly dependent on the appropriate combination of enzymes and fluorogenic substrates for signal readout. However, the availability of reliable pairs of substrates and enzymes is still limited. In general, an ideal fluorogenic substrate should be highly stable against degradation, highly fluorescent upon hydrolysis, and finally stable against photobleaching. For the digital bioassay, in addition to such general requirements, additional factors, such as surface adsorption of large or small molecules and leakage of reactants, must be considered to



CONCLUSION Digital bioassay is a revolutionary technology for both analytical chemistry and single-molecule studies. The most remarkable difference between digital bioassays and other analytical methods is that, in the former, only the number of a binary (positive or negative) signal from a large population of individual reactors is counted for the determination of concentration, whereas in the latter, the absolute signal intensity is recorded from a single reactor. The digital bioassay or digital counting technique provides an effective tool to interrogate molecular events at the single-molecule level, with high throughput. The solid theoretical basis and advanced micro/nanotechnologies ensure the reliability of this strategy for a wide range of applications. Emulsion droplets, microfluidic droplets, and planar microchamber droplets support the development of this emerging field. The original motivation of the digital bioassay for the ultrasensitive detection of nucleic acids and proteins has been extensively demonstrated, and the platform has been further extended to basic biochemical and biophysical studies. The system integration of digital bioassays with other technologies is expected to provide new opportunities for elucidation of biological events at the single-molecule level; such integration to date, as discussed in this Review, remains limited. However, such strategies show great promise, and we believe that a higher degree of integration with innovative optical, thermal, magnetic, electrical, or acoustic technologies should fulfill this potential.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel: +81-3-58417252. Fax: +81-3-5841-1872. ORCID

Yi Zhang: 0000-0002-1886-737X Author Contributions

Y.Z. and H.N. conceived the structure of this Review. Y.Z. wrote the manuscript. Both authors approved the final version of the manuscript. Notes

The authors declare no competing financial interest. Biographies Yi Zhang received his B.Sc. degree in biotechnology from Huazhong University of Science and Technology (China) in 2008. He studied microfluidics at Peking University (China), where he received his Ph.D. degree in 2013. He worked for two years as a JSPS fellow at The University of Tokyo (Japan) from 2013 to 2015. He is currently a project researcher at The University of Tokyo. His primary research interests include the development of microtools for directed evolution of proteins, cell-free protein synthesis, and ultrasensitive bioanalysis. Hiroyuki Noji is a single-molecule biophysicist. He was trained under the supervision of Prof. Masasuke Yoshida and received his Ph.D. from Tokyo Institute of Technology in 1997. After a postdoctoral fellowship at the laboratory of Prof. Kazuhiko Kinosita, Jr., he was appointed as an Associate Professor at the Institute of Industrial Science, The 99

DOI: 10.1021/acs.analchem.6b04290 Anal. Chem. 2017, 89, 92−101

Analytical Chemistry

Review

(19) Men, Y. F.; Fu, Y. S.; Chen, Z. T.; Sims, P. A.; Greenleaf, W. J.; Huang, Y. Y. Anal. Chem. 2012, 84, 4262−4266. (20) Cohen, D. E.; Schneider, T.; Wang, M.; Chiu, D. T. Anal. Chem. 2010, 82, 5707−5717. (21) Heyries, K. A.; Tropini, C.; VanInsberghe, M.; Doolin, C.; Petriv, O. I.; Singhal, A.; Leung, K.; Hughesman, C. B.; Hansen, C. L. Nat. Methods 2011, 8, 649−651. (22) Zhu, Q. Y.; Gao, Y. B.; Yu, B. W.; Ren, H.; Qiu, L.; Han, S. H.; Jin, W.; Jin, Q. H.; Mu, Y. Lab Chip 2012, 12, 4755−4763. (23) Hosokawa, K.; Sato, K.; Ichikawa, N.; Maeda, M. Lab Chip 2004, 4, 181−185. (24) Zhang, Y.; Jiang, X. Microfluidic tools for DNA analysis. In DNA Nanotechnology; Fan, C., Ed.; Springer: New York, 2013; pp 113−153. (25) Rotman, B. Proc. Natl. Acad. Sci. U. S. A. 1961, 47, 1981−1991. (26) Rondelez, Y.; Tresset, G.; Tabata, K. V.; Arata, H.; Fujita, H.; Takeuchi, S.; Noji, H. Nat. Biotechnol. 2005, 23, 361−365. (27) Rissin, D. M.; Walt, D. R. Nano Lett. 2006, 6, 520−523. (28) Kan, C. W.; Rivnak, A. J.; Campbell, T. G.; Piech, T.; Rissin, D. M.; Mosl, M.; Peterca, A.; Niederberger, H. P.; Minnehan, K. A.; Patel, P. P.; Ferrell, E. P.; Meyer, R. E.; Chang, L.; Wilson, D. H.; Fournier, D. R.; Duffy, D. C. Lab Chip 2012, 12, 977−985. (29) Hindson, C. M.; Chevillet, J. R.; Briggs, H. A.; Gallichotte, E. N.; Ruf, I. K.; Hindson, B. J.; Vessella, R. L.; Tewari, M. Nat. Methods 2013, 10, 1003−1005. (30) Saiki, R. K.; Gelfand, D. H.; Stoffel, S.; Scharf, S. J.; Higuchi, R.; Horn, G. T.; Mullis, K. B.; Erlich, H. A. Science 1988, 239, 487−491. (31) Unger, M. A.; Chou, H. P.; Thorsen, T.; Scherer, A.; Quake, S. R. Science 2000, 288, 113−116. (32) Thorsen, T.; Maerkl, S. J.; Quake, S. R. Science 2002, 298, 580− 584. (33) Baker, M. Nat. Methods 2012, 9, 541−544. (34) Leman, M.; Abouakil, F.; Griffiths, A. D.; Tabeling, P. Lab Chip 2015, 15, 753−765. (35) Kopp, M. U.; de Mello, A. J.; Manz, A. Science 1998, 280, 1046− 1048. (36) Wu, W.; Trinh, K. T. L.; Lee, N. Y. Analyst 2012, 137, 2069− 2076. (37) Gill, P.; Ghaemi, A. Nucleosides, Nucleotides Nucleic Acids 2008, 27, 224−243. (38) Craw, P.; Balachandran, W. Lab Chip 2012, 12, 2469−2486. (39) Zhao, Y. X.; Chen, F.; Li, Q.; Wang, L. H.; Fan, C. H. Chem. Rev. 2015, 115, 12491−12545. (40) Shen, F.; Davydova, E. K.; Du, W. B.; Kreutz, J. E.; Piepenburg, O.; Ismagilov, R. F. Anal. Chem. 2011, 83, 3533−3540. (41) Gansen, A.; Herrick, A. M.; Dimov, I. K.; Lee, L. P.; Chiu, D. T. Lab Chip 2012, 12, 2247−2254. (42) Rane, T. D.; Chen, L. B.; Zec, H. C.; Wang, T. H. Lab Chip 2015, 15, 776−782. (43) Chang, L.; Rissin, D. M.; Fournier, D. R.; Piech, T.; Patel, P. P.; Wilson, D. H.; Duffy, D. C. J. Immunol. Methods 2012, 378, 102−115. (44) Dinh, T. L.; Ngan, K. C.; Shoemaker, C. B.; Walt, D. R. Anal. Chem. 2016, DOI: 10.1021/acs.analchem.6b03192. (45) Rissin, D. M.; Kan, C. W.; Campbell, T. G.; Howes, S. C.; Fournier, D. R.; Song, L.; Piech, T.; Patel, P. P.; Chang, L.; Rivnak, A. J.; Ferrell, E. P.; Randall, J. D.; Provuncher, G. K.; Walt, D. R.; Duffy, D. C. Nat. Biotechnol. 2010, 28, 595−599. (46) Kim, S. H.; Iwai, S.; Araki, S.; Sakakihara, S.; Iino, R.; Noji, H. Lab Chip 2012, 12, 4986−4991. (47) Nie, S.; Henley, W. H.; Miller, S. E.; Zhang, H.; Mayer, K. M.; Dennis, P. J.; Oblath, E. A.; Alarie, J. P.; Wu, Y.; Oppenheim, F. G.; Little, F. F.; Uluer, A. Z.; Wang, P. D.; Ramsey, J. M.; Walt, D. R. Lab Chip 2014, 14, 1087−1098. (48) Wilson, D. H.; Rissin, D. M.; Kan, C. W.; Fournier, D. R.; Piech, T.; Campbell, T. G.; Meyer, R. E.; Fishburn, M. W.; Cabrera, C.; Patel, P. P.; Frew, E.; Chen, Y.; Chang, L.; Ferrell, E. P.; von Einem, V.; McGuigan, W.; Reinhardt, M.; Sayer, H.; Vielsack, C.; Duffy, D. C. J. Lab. Autom. 2016, 21, 533−547. (49) Rivnak, A. J.; Rissin, D. M.; Kan, C. W.; Song, L. A.; Fishburn, M. W.; Piech, T.; Campbell, T. G.; DuPont, D. R.; Gardel, M.;

University of Tokyo, in 2001. In 2005, he moved to the Institute of Scientific and Industrial Research, Osaka University, as a full professor. Since 2010, he has been a Professor at the Department of Applied Chemistry, The University of Tokyo. He has been studying the chemomechanical coupling mechanism of FoF1 ATP synthase using single-molecule techniques. He is also known as an inventor of the femtoliter chamber array system for single-molecule enzymatic assays, which is currently applied in single-molecule digital ELISA.



ACKNOWLEDGMENTS Y.Z. gratefully acknowledges financial support from the JSPS Postdoctoral Fellowship for Overseas Researchers. This work was funded by Grant-in-Aid for JSPS Fellows (13F03378) and ImPACT Program of Council for Science, Technology and Innovation (Cabinet Office, Government of Japan).



REFERENCES

(1) Valcarcel, M., Ed. Sensitivity. In Principles of Analytical Chemistry: A Textbook, 1st ed.; Springer: New York, 2000; pp 65−69. (2) Zhang, Y.; Guo, Y. M.; Xianyu, Y. L.; Chen, W. W.; Zhao, Y. Y.; Jiang, X. Y. Adv. Mater. 2013, 25, 3802−3819. (3) Harris, C. C.; Yolken, R. H.; Krokan, H.; Hsu, I. C. Proc. Natl. Acad. Sci. U. S. A. 1979, 76, 5336−5339. (4) Patolsky, F.; Zheng, G. F.; Hayden, O.; Lakadamyali, M.; Zhuang, X. W.; Lieber, C. M. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 14017− 14022. (5) Zheng, G. F.; Patolsky, F.; Cui, Y.; Wang, W. U.; Lieber, C. M. Nat. Biotechnol. 2005, 23, 1294−1301. (6) Funatsu, T.; Harada, Y.; Tokunaga, M.; Saito, K.; Yanagida, T. Nature 1995, 374, 555−559. (7) Jain, A.; Liu, R. J.; Ramani, B.; Arauz, E.; Ishitsuka, Y.; Ragunathan, K.; Park, J.; Chen, J.; Xiang, Y. K.; Ha, T. Nature 2011, 473, 484−488. (8) Holzmeister, P.; Acuna, G. P.; Grohmann, D.; Tinnefeld, P. Chem. Soc. Rev. 2014, 43, 1014−1028. (9) Levene, M. J.; Korlach, J.; Turner, S. W.; Foquet, M.; Craighead, H. G.; Webb, W. W. Science 2003, 299, 682−686. (10) Zhu, P.; Craighead, H. G. Annu. Rev. Biophys. 2012, 41, 269− 293. (11) Eid, J.; Fehr, A.; Gray, J.; Luong, K.; Lyle, J.; Otto, G.; Peluso, P.; Rank, D.; Baybayan, P.; Bettman, B.; Bibillo, A.; Bjornson, K.; Chaudhuri, B.; Christians, F.; Cicero, R.; Clark, S.; Dalal, R.; Dewinter, A.; Dixon, J.; Foquet, M.; Gaertner, A.; Hardenbol, P.; Heiner, C.; Hester, K.; Holden, D.; Kearns, G.; Kong, X. X.; Kuse, R.; Lacroix, Y.; Lin, S.; Lundquist, P.; Ma, C. C.; Marks, P.; Maxham, M.; Murphy, D.; Park, I.; Pham, T.; Phillips, M.; Roy, J.; Sebra, R.; Shen, G.; Sorenson, J.; Tomaney, A.; Travers, K.; Trulson, M.; Vieceli, J.; Wegener, J.; Wu, D.; Yang, A.; Zaccarin, D.; Zhao, P.; Zhong, F.; Korlach, J.; Turner, S. Science 2009, 323, 133−138. (12) Uemura, S.; Aitken, C. E.; Korlach, J.; Flusberg, B. A.; Turner, S. W.; Puglisi, J. D. Nature 2010, 464, 1012−1017. (13) Bertsekas, D. P.; Tsitsiklis, J. N. The Bernoulli process. In Introduction to Probability, 2 ed.; Athena Scientific: Belmont, MA, 2008; p 307. (14) Rissin, D. M.; Fournier, D. R.; Piech, T.; Kan, C. W.; Campbell, T. G.; Song, L. A.; Chang, L.; Rivnak, A. J.; Patel, P. P.; Provuncher, G. K.; Ferrell, E. P.; Howes, S. C.; Pink, B. A.; Minnehan, K. A.; Wilson, D. H.; Duffy, D. C. Anal. Chem. 2011, 83, 2279−2285. (15) Sykes, P. J.; Neoh, S. H.; Brisco, M. J.; Hughes, E.; Condon, J.; Morley, A. A. Biotechniques 1992, 13, 444−449. (16) Teh, S. Y.; Lin, R.; Hung, L. H.; Lee, A. P. Lab Chip 2008, 8, 198−220. (17) Griffiths, A. D.; Tawfik, D. S. Trends Biotechnol. 2006, 24, 395− 402. (18) Ottesen, E. A.; Hong, J. W.; Quake, S. R.; Leadbetter, J. R. Science 2006, 314, 1464−1467. 100

DOI: 10.1021/acs.analchem.6b04290 Anal. Chem. 2017, 89, 92−101

Analytical Chemistry

Review

Sullivan, S.; Pink, B. A.; Cabrera, C. G.; Fournier, D. R.; Duffy, D. C. J. Immunol. Methods 2015, 424, 20−27. (50) Randall, J.; Mortberg, E.; Provuncher, G. K.; Fournier, D. R.; Duffy, D. C.; Rubertsson, S.; Blennow, K.; Zetterberg, H.; Wilson, D. H. Resuscitation 2013, 84, 351−356. (51) Song, L. A.; Zhao, M. W.; Duffy, D. C.; Hansen, J.; Shields, K.; Wungjiranirun, M.; Chen, X. H.; Xu, H.; Leffler, D. A.; Sambol, S. P.; Gerding, D. N.; Kelly, C. P.; Pollock, N. R. J. Clin. Microbiol. 2015, 53, 3204−3212. (52) Gaylord, S. T.; Abdul-Aziz, S.; Walt, D. R. J. Clin. Microbiol. 2015, 53, 1722−1724. (53) Obayashi, Y.; Iino, R.; Noji, H. Analyst 2015, 140, 5065−5073. (54) Wu, D. L.; Milutinovic, M. D.; Walt, D. R. Analyst 2015, 140, 6277−6282. (55) Kreutz, J. E.; Munson, T.; Huynh, T.; Shen, F.; Du, W. B.; Ismagilov, R. F. Anal. Chem. 2011, 83, 8158−8168. (56) Shen, F.; Sun, B.; Kreutz, J. E.; Davydova, E. K.; Du, W. B.; Reddy, P. L.; Joseph, L. J.; Ismagilov, R. F. J. Am. Chem. Soc. 2011, 133, 17705−17712. (57) Ge, S. C.; Liu, W. S.; Schlappi, T.; Ismagilov, R. F. J. Am. Chem. Soc. 2014, 136, 14662−14665. (58) Lee, H.; Kim, J.; Kim, H.; Kim, J.; Kwon, S. Nat. Mater. 2010, 9, 745−749. (59) Geiss, G. K.; Bumgarner, R. E.; Birditt, B.; Dahl, T.; Dowidar, N.; Dunaway, D. L.; Fell, H. P.; Ferree, S.; George, R. D.; Grogan, T.; James, J. J.; Maysuria, M.; Mitton, J. D.; Oliveri, P.; Osborn, J. L.; Peng, T.; Ratcliffe, A. L.; Webster, P. J.; Davidson, E. H.; Hood, L. Nat. Biotechnol. 2008, 26, 317−325. (60) Cunin, F.; Schmedake, T. A.; Link, J. R.; Li, Y. Y.; Koh, J.; Bhatia, S. N.; Sailor, M. J. Nat. Mater. 2002, 1, 39−41. (61) Han, M. Y.; Gao, X. H.; Su, J. Z.; Nie, S. Nat. Biotechnol. 2001, 19, 631−635. (62) Li, Y. G.; Cu, Y. T. H.; Luo, D. Nat. Biotechnol. 2005, 23, 885− 889. (63) Steemers, F. J.; Ferguson, J. A.; Walt, D. R. Nat. Biotechnol. 2000, 18, 91−94. (64) Zhao, Y. J.; Shum, H. C.; Chen, H. S.; Adams, L. L. A.; Gu, Z. Z.; Weitz, D. A. J. Am. Chem. Soc. 2011, 133, 8790−8793. (65) Nicewarner-Pena, S. R.; Freeman, R. G.; Reiss, B. D.; He, L.; Pena, D. J.; Walton, I. D.; Cromer, R.; Keating, C. D.; Natan, M. J. Science 2001, 294, 137−141. (66) Qin, L. D.; Park, S.; Huang, L.; Mirkin, C. A. Science 2005, 309, 113−115. (67) Matthias, S.; Schilling, J.; Nielsch, K.; Muller, F.; Wehrspohn, R. B.; Gosele, U. Adv. Mater. 2002, 14, 1618−1621. (68) Pregibon, D. C.; Toner, M.; Doyle, P. S. Science 2007, 315, 1393−1396. (69) Zhang, Y.; Sun, J. S.; Zou, Y.; Chen, W. W.; Zhang, W.; Xi, J. J.; Jiang, X. Y. Anal. Chem. 2015, 87, 900−906. (70) Rissin, D. M.; Kan, C. W.; Song, L. N.; Rivnak, A. J.; Fishburn, M. W.; Shao, Q. C.; Piech, T.; Ferrell, E. P.; Meyer, R. E.; Campbell, T. G.; Fournier, D. R.; Duffy, D. C. Lab Chip 2013, 13, 2902−2911. (71) Nie, S.; Benito-Pena, E.; Zhang, H. B.; Wu, Y.; Walt, D. R. J. Visualized Exp. 2013, e50726. (72) Zhong, Q.; Bhattacharya, S.; Kotsopoulos, S.; Olson, J.; Taly, V.; Griffiths, A. D.; Link, D. R.; Larson, J. W. Lab Chip 2011, 11, 2167− 2174. (73) Luminex. xMAP Cookbook, 3rd ed.; Luminex: Austin, TX, 2016. (74) Yin, H.; Flynn, A. D. Annu. Rev. Biomed. Eng. 2016, 18, 51−76. (75) Watanabe, R.; Soga, N.; Fujita, D.; Tabata, K. V.; Yamauchi, L.; Kim, S. H.; Asanuma, D.; Kamiya, M.; Urano, Y.; Suga, H.; Noji, H. Nat. Commun. 2014, 5, 4519. (76) Watanabe, R.; Soga, N.; Yamanaka, T.; Noji, H. Sci. Rep. 2014, 4, 7076. (77) Watanabe, R.; Soga, N.; Hara, M.; Noji, H. Lab Chip 2016, 16, 3043−3048. (78) Rondelez, Y.; Tresset, G.; Nakashima, T.; Kato-Yamada, Y.; Fujita, H.; Takeuchi, S.; Noji, H. Nature 2005, 433, 773−777. (79) Cai, L.; Friedman, N.; Xie, X. S. Nature 2006, 440, 358−362.

(80) Lu, H. P.; Xun, L. Y.; Xie, X. S. Science 1998, 282, 1877−1882. (81) Xie, X. S.; Lu, H. P. J. Biol. Chem. 1999, 274, 15967−15970. (82) Min, W.; English, B. P.; Luo, G. B.; Cherayil, B. J.; Kou, S. C.; Xie, X. S. Acc. Chem. Res. 2005, 38, 923−931. (83) Grima, R.; Walter, N. G.; Schnell, S. FEBS J. 2014, 281, 518− 530. (84) English, B. P.; Min, W.; van Oijen, A. M.; Lee, K. T.; Luo, G. B.; Sun, H. Y.; Cherayil, B. J.; Kou, S. C.; Xie, X. S. Nat. Chem. Biol. 2006, 2, 87−94. (85) Wong, L. S.; Khan, F.; Micklefield, J. Chem. Rev. 2009, 109, 4025−4053. (86) Rissin, D. M.; Gorris, H. H.; Walt, D. R. J. Am. Chem. Soc. 2008, 130, 5349−5353. (87) Rojek, M. J.; Walt, D. R. PLoS One 2014, 9, e86224. (88) Gorris, H. H.; Walt, D. R. J. Am. Chem. Soc. 2009, 131, 6277− 6282. (89) Ehrl, B. N.; Liebherr, R. B.; Gorris, H. H. Analyst 2013, 138, 4260−4265. (90) Liebherr, R. B.; Renner, M.; Gorris, H. H. J. Am. Chem. Soc. 2014, 136, 5949−5955. (91) Mrksich, M.; Whitesides, G. M. Annu. Rev. Biophys. Biomol. Struct. 1996, 25, 55−78. (92) Hua, B. Y.; Han, K. Y.; Zhou, R. B.; Kim, H. J.; Shi, X. H.; Abeysirigunawardena, S. C.; Jain, A.; Singh, D.; Aggarwal, V.; Woodson, S. A.; Ha, T. Nat. Methods 2014, 11, 1233−1236. (93) Ueda, H.; Tsumoto, K.; Kubota, K.; Suzuki, E.; Nagamune, T.; Nishimura, H.; Schueler, P. A.; Winter, G.; Kumagai, I.; Mahoney, W. C. Nat. Biotechnol. 1996, 14, 1714−1718. (94) Shen, J. W.; Li, Y. B.; Gu, H. S.; Xia, F.; Zuo, X. L. Chem. Rev. 2014, 114, 7631−7677. (95) Anzenbacher, P.; Palacios, M. A. Nat. Chem. 2009, 1, 80−86. (96) Comellas-Aragones, M.; Engelkamp, H.; Claessen, V. I.; Sommerdijk, N. A. J. M.; Rowan, A. E.; Christianen, P. C. M.; Maan, J. C.; Verduin, B. J. M.; Cornelissen, J. J. L. M.; Nolte, R. J. M. Nat. Nanotechnol. 2007, 2, 635−639. (97) Yang, Y.; Wang, J.; Shigematsu, H.; Xu, W.; Shih, W. M.; Rothman, J. E.; Lin, C. Nat. Chem. 2016, 8, 476−483. (98) Hatch, A. C.; Fisher, J. S.; Tovar, A. R.; Hsieh, A. T.; Lin, R.; Pentoney, S. L.; Yang, D. L.; Lee, A. P. Lab Chip 2011, 11, 3838− 3845. (99) Sasagawa, K.; Ando, K.; Kobayashi, T.; Noda, T.; Tokuda, T.; Kim, S. H.; Iino, R.; Noji, H.; Ohta, J. Jpn. J. Appl. Phys. 2012, 51, 02BL01. (100) Takehara, H.; Miyazawa, K.; Noda, T.; Sasagawa, K.; Tokuda, T.; Kim, S. H.; Iino, R.; Noji, H.; Ohta, J. Jpn. J. Appl. Phys. 2014, 53, 04EL02.

101

DOI: 10.1021/acs.analchem.6b04290 Anal. Chem. 2017, 89, 92−101