Homogeneous Entropy-Driven Amplified Detection of Biomolecular

Jul 27, 2016 - While a range of artificial biochemical circuits is likely to play a significant role in biological engineering, one of the challenges ...
1 downloads 8 Views 2MB Size
Download PDF
Homogeneous Entropy-Driven Amplified Detection of Biomolecular Interactions Donghyuk Kim,† Omai B. Garner,‡ Aydogan Ozcan,§,∥ and Dino Di Carlo*,†,∥,⊥ †

Department of Bioengineering, ‡Department of Pathology & Laboratory Medicine, §Department of Electrical Engineering, California NanoSystems Institute, and ⊥Jonsson Comprehensive Cancer Center, University of California, Los Angeles, Los Angeles, California 90095, United States ∥

S Supporting Information *

ABSTRACT: While a range of artificial biochemical circuits is likely to play a significant role in biological engineering, one of the challenges in the field is the design of circuits that can transduce between biomolecule classes (e.g., moving beyond nucleic acid only circuits). Herein, we design a transduction mechanism whereby a protein signal is transduced into an amplified nucleic acid output using DNA nanotechnology. In this system, a protein is recognized by nucleic acid bound recognition elements to form a catalytic complex that drives a hybridization/displacement reaction on a multicomponent nucleic acid substrate, releasing multiple target single-stranded oligonucleotides in an amplified fashion. Amplification power and simple one-pot reaction conditions lead us to apply the scheme in an assay format, achieving homogeneous and rapid (∼10 min) analyte detection that is also robust (operable in whole blood and plasma). In addition, we demonstrate the assay in a microfluidic digital assay format leading to improved quantification and sensitivity approaching single-molecule levels. The present scheme we believe will have a significant impact on a range of applications from fundamental molecular interaction studies to design of artificial circuits in vivo to high-throughput, multiplexed assays for screening or point-of-care diagnostics. KEYWORDS: DNA-based molecular machinery, protein-to-nucleic acid transduction, biomolecular interaction, homogeneous assay, amplified assay, digital assay

N

in disease treatment, the ability to quickly and accurately identify the presence of multiple types of biomolecules indicative of disease has always been crucial. Numerous nucleic-acid-based approaches had been applied for clinical diagnostics and genetic analysis;3,8,10 however, most have been limited to nucleic-acid-only applications. Immuno-polymerase chain reaction (ImmunoPCR) is one of the very initial attempts to utilize nucleic acids as a tool to generate amplified signal from protein analytes. Due to its specificity, achieved by antibodies, and low limit of detection (LOD), afforded by enzymatic amplification, ImmunoPCR has been widely used in research, industrial, and clinical applications to identify disease biomarkers, pathogens, and pollutants.13−19 In conventional ImmunoPCR, or its digital implementations, however, molecular recognition elements are typically immobilized on a solid support, and unfortunately, this immobilization of, for example, capture antibodies limits the number of antibodies available to bind to target analytes and reduces assay sensitivity.20 Also, ImmunoPCR is typically not a homogeneous technique, meaning that it suffers from nonspecific signal

ucleic acids have proven to be ideal building blocks for artificial molecular machinery. The complexity and diversity of nucleic-acid-based molecular circuits, controlled by simple design principles using Watson−Crick base pairing, have led to numerous innovative functional and structural molecular modules that perform a variety of tasks including assembly into three-dimensional structures,1 DNA nanodevices/machines,2 and molecular logic gates.3−5 Such modules result in a wide range of applications from molecular sensing to embedding artificial molecular pathways into cells.6−8 Furthermore, the ability of nucleic acids to recognize sequences through noncovalent bonding has successfully wired molecular inputs into catalytically amplified outputs so that the circuits can be operable with a low abundance of triggers.4 However, most of the circuits are nucleic-acid-only circuits incapable of transducing different classes of biomolecules. As most natural molecular circuits in biology are composed of multiple classes of biomolecules, controlled transduction of different classes of molecules is likely to play a significant role in a wide range of future biological/biomedical engineering solutions.7−9 One of the opportunities for schemes that transduce between different classes of biomolecules is in the development of sensing systems. Considering the importance of early diagnosis © 2016 American Chemical Society

Received: March 24, 2016 Accepted: July 27, 2016 Published: July 27, 2016 7467

DOI: 10.1021/acsnano.6b02060 ACS Nano 2016, 10, 7467−7475

Article

www.acsnano.org

Article

ACS Nano

Figure 1. (a) Schematic illustration of the assay reaction mechanism. (b) Oligonucleotide sequences used in the optimized HEBA reaction.

assay.26 In the entropy-driven nucleic acid displacement assay, displacement of a target sequence by a catalytic sequence initiates the entire reaction. In HEBA, binding of two recognition elements such as biotin or antibodies to a single target analyte brings together two different sequences that only together can displace a target sequence. We first validate HEBA using biotin−streptavidin as the specificity pair and identify optimal catalytic precursors and operating conditions. We then employ HEBA for identification of influenza A nucleoprotein (NP) spiked in universal storage M5 buffer, and last, we employ HEBA on a digital microfluidic assay platform to achieve ultrasensitive detection of influenza A nucleoproteins.

generation without several washing steps reducing its sensitivity, speed, and ease of assay automation. Some recent developments, the proximity ligation- or extension-based techniques, have achieved similar level of performance without utilizing a solid support while still requiring sequential additions of reagents.23−25 Also, these methods require activities of one or more polymerases and/or thermocycling, and long assay times (∼1 h). There have been several homogeneous (one-pot) assays developed to transduce inputs into nucleic acid outputs to address these challenges of speed and automation with a molecular assay. In a homogeneous assay, the key challenge is to trigger signal generation only in the presence of an analyte. The dominant strategy, so-called molecular beacons, to achieve specificity relies on changing the conformation of fluorophoreand quencher-conjugated nucleic acids in the presence of an input nucleic acid.21,22 Unfortunately, these approaches consist of nucleic-acid-only circuits, and detection limits are not satisfactory for many applications due to signal generation without amplification. Recently, researchers have combined aptamers and DNAzymes to transduce protein-based inputs into amplified nucleic acid outputs for sensing purposes.11,12 While they successfully transduce protein inputs into nucleic acid outputs, only nanomolar-level sensitivity is achieved with the performance limited by the DNAzyme activity, which is often slow or sensitive to modifications in its structure. We herein report a simple approach to transduce a protein input into an amplified nucleic acid output. The proposed machinery is demonstrated as a protein assay, a homogeneous entropy-driven biomolecular assay (HEBA) that achieves onepot, catalytically amplified signal generation with no use of enzymes or precise temperature cycling. In HEBA, reagents are solely added at the beginning of the assay, no temperature changes are required to drive the reaction, and the assay yields an amplified signal specific to the presence of analyte within ∼10 min. The juxtaposition of two catalyst precursor oligonucleotides conjugated to recognition elements (e.g., biotin or antibodies) act as a catalyst in HEBA to augment a recently reported entropy-driven nucleic acid displacement

RESULTS AND DISCUSSION Assay Reaction Design and Validation. HEBA is designed to start with a double-stranded multiplex sequence (MS), consisting of three different single-stranded DNA oligonucleotides (ssDNAs) and two other ssDNA pieces serving as catalytic sequence (CS) and dummy sequence (DS). The steps for the overall reaction are (1) CS binding to MS displacing TS1 and (2) DS binding to the intermediate from (1) by which TS2 and CS are displaced. Thus, the net reaction is MS + CS + DS, which yields TS1 + TS2 + CS and waste sequence WS (Figure S1).14 To expand this nucleic-acidbased strategy to a general molecular assay platform, we incorporated molecular recognition elements to generate CS in an analyte-dependent manner (Figure 1a). In HEBA, the CS is split into two pieces (CS1 and CS2), and each piece is functionalized to have a recognition moiety (e.g., biotin or antibody). Exposure of the system to analyte generates CS1− analyte−CS2 complexes, which in turn can act as CS in the net MS + DS + CS ⇄ WS + TS1 + TS2 + CS reaction. Because individual CS precursors are not capable of accelerating the net reaction, catalytic accumulation of free TS1 and TS2 over time reports the presence of analyte. HEBA was first demonstrated using biotin−streptavidin as the specificity pair. The 5′ end of CS1 and 3′ end of CS2 were biotinylated (Figure 1b) so that a CS1−streptavidin−CS2 complex could be generated in the presence of streptavidin. 7468

DOI: 10.1021/acsnano.6b02060 ACS Nano 2016, 10, 7467−7475

Article

ACS Nano

Figure 2. Detection of streptavidin and influenza A nucleoprotein with analogue HEBA. (a) HEBA-specific signal generation in the presence of 5 nM streptavidin using an oligonucleotide-based fluorescence probe targeting TS2 (as shown in Figure 1a). (b) Impact of individual CS piece structures. Reaction condition: 20 °C for 10 min with [streptavidin] = 5 nM. (c) Detection limit and dynamic range of HEBA using a biotin−streptavidin pair. Reaction conditions: 20 °C for 10 min. (d) Detection limit and dynamic range of HEBA for influenza A nucleoprotein spiked into M5 buffer. Reaction conditions: 20 °C for 10 min. Fluorescence ratio is the ratio of fluorescence signal of an analyte-containing well to that of a well with all reaction components but without analyte. Error bars are mean ± SD, and dotted lines in plots indicate the baseline signal (mean + 3SD).

Figure 3. Digital HEBA extends the dynamic range of detection. (a) Digital HEBA (dHEBA) schematic with expected reaction summary for the “ON” wells (with analyte) and “OFF” wells (without analyte). (b) Representative images obtained from dHEBA assays using 4 aM (∼15 molecules, top) and 100 aM (360 molecules, bottom) of influenza A nucleoproteins as analyte. (c) dHEBA analysis of influenza A nucleoproteins from 4 aM (∼15 molecules) to 10 fM (36 000 molecules). (d) Combined representation of HEBA and dHEBA performance for identification of influenza A NPs over 8 orders of magnitude. Error bars are mean ± SD.

accelerate the net reaction (Figure 2a). To note, there is a baseline fluorescence signal potentially due to the reaction equilibrium, and thus, throughout this work, we normalized analyte signal (CS1 + CS2 + analyte) with respect to this

The addition of 5 nM streptavidin to the reaction mix led to a fluorescent signal that was larger than for CS1 or CS2 alone and CS1 and CS2 together without streptavidin, demonstrating that CS1−streptavidin−CS2 acts as a single catalytic element to 7469

DOI: 10.1021/acsnano.6b02060 ACS Nano 2016, 10, 7467−7475

Article

ACS Nano control fluorescence signal (CS1 + CS2) (see Figure S2a). Reagent mixing order played a role in the HEBA performance: addition of analyte (streptavidin) following both CS1 and CS2 was critical for its success; otherwise, CS1−streptavidin−CS1 or CS2−streptavidin−CS2 complexes can be generated at higher rates, leading to assay performance variation. Defining Chemistry Requirements for HEBA and Assay Characterization. Using the streptavidin−biotin system, we also explored how to design the catalytic sequence to maximize signal generation above background. We compared two split points (Figure 2c and Table S1): a CS1−CS2 pair, in which CS1 had both toehold binding and branch migration functionality to displace TS1, and a xCS1−xCS2 pair, in which xCS1 has no toehold region to localize it prior to replacing the complementary portion of TS1. While CS2 (xCS2) localizes CS1 (xCS1) close to the MS backbone, designing CS1 such that it contains a toehold binding region was favored for optimal HEBA performance. In the CS1−CS2 scheme, a 2 bp toehold region in CS1 acts to localize CS1 on the MS backbone and accelerate the displacement of TS1. These results were consistent with previous findings in literature on toehold binding and branch migration,27,28 and building off these results, we find that the CS1−CS2 structure can be further optimized in the future to take advantage of high affinity toehold binding to accelerate branch migration.27,28 Using this model HEBA system, we next characterized the optimal temperature and time scale of operation. HEBA operated over a range of moderate temperatures (15−35 °C, Figure S2c), and we found a signal increase with respect to control (no analyte) conditions beyond the mean + 3SD within 10 min (Figure S2b). To note, although we generate sufficient signal within 10 min, at this time point, the reaction has not gone to completion. Signal increases further over longer times, but signal to background may not increase substantially beyond 10 min (Figure S2b and Figure 3). We next evaluated how stoichiometry of the reactants affects HEBA performance. The assay requires an analyte sandwiched by CS1 and CS2 to form the catalytic complex (CS1−analyte− CS2). Therefore, if the analyte concentration is too high compared to the [CS1] or [CS2], the probability of creating a CS1−analyte−CS2 complex diminishes, and analyte−CS1 or analyte−CS2 complexes become dominant (Figure S4). Obtained data were consistent with such an effect fluorescence signal diminished for high streptavidin concentrations (200 nM in Figure 2c, 2 mM in Figure S5a). Previous work reported that an optimal stoichiometry to achieve a biotin−streptavidin−biotin complex is ∼10:1 biotin/streptavidin, which is consistent with our observations.16 The stoichiometry between [analyte] and [CSs] is also a parameter to be adjusted for future use of HEBA, depending on the expected analyte concentration within an assay. Because of the amplified nature of the readout, HEBA identified the presence of streptavidin down to 1 fM in TrisEDTA (TE) buffer with 6−7 orders of magnitude in dynamic range (Figure 2b and Figure S5a) in a 10 min 250 μL scale reaction. Since HEBA amplification is non-enzymatic, the detection limit was not substantially reduced in body fluids, achieving sub-picomolar-level limit of detection in plasma whole blood with a similar dynamic range (Figure S5b,c). Development of HEBA for Influenza Nucleoprotein Detection. A HEBA was then developed to detect influenza A proteins, toward a point-of-care diagnostic assay. We chose influenza virus nucleoprotein as the target analyte because it is a

main conserved protein stabilizing the viral RNA strand and serves as the target for most lateral flow point of care influenza assays.29 We evaluated HEBA performance with full length NPs in M5 buffer, which is the universal transport media used in clinical microbiology laboratories to store suspected influenza samples. While noticeable windows for further optimization were observed (discussed in Supplementary Text 1 and 2), this influenza HEBA was able to detect sub-picomolar NPs within 10 min per test at room temperature (Figure 2d and Figure S6). We also examined the performance of HEBA identifying the presence of influenza A nucleoproteins in real influenza A patient samples. Influenza A negative patient samples (examined by qPCR) were obtained from a microbiology laboratory at the University of California Los Angeles (Dr. Omai Garner) and spiked in with varying concentrations of influenza A nucleoproteins. HEBA was capable of identifying the presence of nuceloproteins down to 10 fM concentrations (Figure S8). Although we observed a low detection limit, the quantitative performance of the assay is not ideal in an analogue format. That is, signal increase is small with increasing analyte concentration over orders of magnitude, which led us to investigate a digital assay for improved quantification. Microfluidic Digital HEBA. We next adapted the influenza A HEBA to a microfluidic digital assay platform to improve quantification accuracy and further reduce its limit of detection. A reduced limit of detection could allow for fractionating of samples to conduct multiplexed assays, even when starting with small sample volumes. In the digital HEBA (dHEBA), the reaction mixture is segmented in a microfluidic well array such that largely 0 or 1 analyte molecules would be present within a given microwell (Figure 3a).30−32 The microfluidic dHEBA platform was designed in a 10 × 3 array of 3364 microwells (2.6 pL each), and positive wells were identified as having a mean fluorescence intensity that is higher than the mean +3SD of background wells (Figure 3b). dHEBA successfully detected influenza A NPs down to 4 aM in the initial input solution, which corresponds to ∼15 wells or molecules detected. As opposed to the analogue assay, dHEBA response was relatively linear to [NP] between 4 aM and 1 fM (R2 = 0.94 for 4 aM to 1 fM), while the response started to deviate from linearity above [NP] = 1 fM. Deviation is expected for digital assays because, when assuming a Poisson distribution of molecules into the segmented volumes, the probability of a single NP per volume decreases with increasing analyte concentration (Figure 3c and Supplementary Text 3). In fact, the number of wells expected to have more than 2 NP molecules becomes greater than one at [NP] = 1 fM in the current dHEBA platform, which supports our observation of dHEBA performance deviating from linearity within the 1−10 fM range (Supplementary Text 3). One key aspect for HEBA is the design of the analytecontaining catalyst that is formed upon sandwich recognition of analyte. As pictured in Figure 1, the catalyst, CS1−analyte− CS2, in HEBA is unlike the catalyst in the proximity ligation assay, where two nucleotides in close proximity are ligated to form a single signaling molecule. In HEBA, the two separate oligonucleotides (CS1 and CS2) without ligation act as a single piece to bind and displace in a cooperative manner because of their proximity (and most likely multivalency due to the streptavidin backbone used). The design rule herein was to consider the binding process of the catalyst as a three-step process (1-toehold binding, 2-toehold binding, 3-branch migration) not a two-step process (toehold binding and branch migration). The CS pair is designed to achieve toehold binding 7470

DOI: 10.1021/acsnano.6b02060 ACS Nano 2016, 10, 7467−7475

Article

ACS Nano Table 1. Assay Performance Table assay c,43

BiFC molecular beacon44 FRET/BRETd,45 biobarcode46 proximity ligation47 PCR48 ImmunoPCR48 HEBA

typical target analyte

amplification powera (S/Nb)

LOD

procedure (thermal control)

assay time

protein nucleic acid protein nucleic acid protein nucleic acid protein protein

no amplification no amplification no amplification mo amplification ∼106 (∼10) 2nf (10−100) 2nf (10−100) ∼106g (∼2)

nM nM nM nM−sub-nM pM−sub-pM single-molecule single-molecule fM

simple (no) simple (no) simple (no) moderate (no) moderate (minimale) intensive (yes) intensive (yes) simple (no)

min min min min h h h ∼10 min

a

Number of signaling molecules generated compared to the number of analyte molecules. bRatio of signal increase compared to baseline control. Bimolecular fluorescence complementation. dFRET: Forster resonance energy transfer. BRET: Bioluminescence resonance energy transfer. e Depending on application, but typically isothermal. fn = the number of cycles, typically ∼40, leading to an amplification power of ∼1012. gAssuming the control to be all reagents with no TS. c

respectively (Figure 2 and Figure S5). Potential causes would be interfering immunoglobulin or proteins, blood cells, and cell debrisamong various methods to handle matrix effects, dilution could be considered as a more reliable option for HEBA given that many analytes will be present at levels much higher than tens to hundreds of molecules needed with digital HEBA for accurate detection. As seen in Figure 3c,d, the digital implementation of HEBA, despite a narrow dynamic range due to the limited number of wells, provided better LOD and quantitation performance compared to its analogue version. Considering the fact that a common clinical practice is to split samples for multiplexed examination, and that dHEBA consumes a very small volume of sample (few microliters) while achieving sensitivity approaching the single-molecule level, it will open additional potential of dHEBA in splitting samples for dilution to reduce matrix effects and/or multiplexed examination for multiple markers. Of course dilution is not a universal solution for all types of molecules, and thus, future work will involve and HEBA developers must consider examination of other matrix effect handling approaches such as removal of interfering antibodies and addition of blocking agents depending on analyte/sample matrices. The advantage of dHEBA over analogue HEBA was not limited to LOD. In analogue HEBA, false negative results can be possible due to higher variations in the signal generation in analogue HEBA when a sample is present. For example, when considering the mean − 2SD (fluorescence intensity, 556.1 − (2 × 54.46) = 447.18) for [streptavidin] = 10 pM, this fell below the mean + 3SD (fluorescence intensity, 452.7 + (3 × 15.9) = 500.4) for [streptavidin] = 0 signal in plasma (Figure S5b). However, in dHEBA (Figure S10), even the mean − 5SD (well count, 4629 − (5 × 807.3) = 592.5) for [streptavidin] = 1 pM samples generated higher than the mean + 3SD (140.2 + (3 × 24.81) = 214.63) for [streptavidin] = 0 pM in plasma (Figure S10). The signal generation module of HEBA has been previously examined by Zhang et al.26 In the previous work, while not optimized for rapid readout as explored here, the data clearly hinted that the signal in the presence of catalytic sequence is already clear at earlier time points before signal saturation. As such, we pursued a field-test-suitable time point (10 min, Figure 2a) to achieve rapid readout. Considering that the signal increase is catalytic in the presence of analyte and the equilibrium of the MS disassembly in the absence of analyte, however, the timing of the readout could likely be further optimized in the future particularly for improved quantification. Notably, the signal generated in response to catalytic nucleic acid was not strongly increased with increased concentration in

cooperatively; the short CS2 toehold binds to MS, providing CS1−analyte−CS2 initial stability on the MS so that the longer CS1 has time to interact with the MS. Here, the longer CS1 interacts with MS by toehold binding first; this secondary toehold binding provides CS1−analyte−CS2 additional stability on the MS and is then followed by the higher activation energy process of branch migration. This approach to staged sequence recognition and displacement is similar the short protospacer adjacent motif in the CRISPR/Cas9 system providing preliminary binding for the Cas9 to then locate the proper target sequence.52 We also expected the length of the linker, bridging the oligonucleotide, and the specificity module to play a significant role in the operation of the catalyst, though our initial examination with PEG4 and PEG12 linkers did not show a significant impact. The CS pair design, the specificity of the CS1−analyte−CS2 complex, is also paramount in order to achieve analytedependent generation of an amplified signal. Particularly in body fluids such as blood or plasma, there are myriad potential interfering components that can suppress/enhance specificitydetermining binding events leading to either false positive or negative signals.53−55 As in sandwich ELISA, antibodies in HEBA are designed to target different epitopes on the analyte rather than one targeting the Fc or other regions of the primary antibody. This minimizes potential interference from cross reaction due to endogenous/structural similarities for the primary antibody or heterophile antibodies.54 In addition, in order for the DNA-based molecular machinery to quickly drive the hybridization/displacement reaction upon the generation of the CS1−analyte−CS2, we chose to work with “short” DNA oligonucleotides. The use of short oligonucleotides helps designing the circuit elements with a minimal chance of self-complementary folding and secondary structures. This helps the machinery quickly respond to the catalyst complex while maintaining its stability in diverse matrices. In HEBA, all “active” zones for hybridization/ displacement on oligonucleotides were over relatively short lengths (less than 30 bp) with minimal chance of selfcomplementary folding and secondary structures (folded alone and pairwise using the mFold Web server),56 and HEBA performs stably over a wide range of salt concentrations (6 and 12.5 mM Mg2+ with 0−300 mM Na+ in TE buffer; data not shown). Currently, HEBA, however, is still somewhat sensitive to nonspecific interference from sample matrices (Figure S5). Signal in buffer still is higher by a factor of 110−130 and 150− 170% when compared to undiluted plasma and whole blood, 7471

DOI: 10.1021/acsnano.6b02060 ACS Nano 2016, 10, 7467−7475

Article

ACS Nano

analytes for which standard sandwich ELISA or proximity ligation assays have been developed. Antibody pairs that bind to two separate moieties on a target protein, which have been developed for these well-established assays,57−59 should function well for this assay where two-site binding is required. However, the developer should pay attention to the length of the linker between the antibody and catalytic sequence which might need to be tuned based on the protein in a similar manner as for the proximity ligation assay. In the current example of streptavidin, the size of streptavidin is about 5 nm in diameter, and thus, there is expected to be an optimal or at least minimum linker length required to create a CS1−streptavidin− CS2 complex. Herein, the linker bridges CS oligos to biotin molecules that bind to streptavidin.

this previous work, potentially explaining the relatively poor quantitative performance of HEBA in an analogue format (Supplementary Text 4 and 5). Also, the stoichiometry of the reaction components was adjusted in the HEBA to (i) obtain a better contrast between conditions with and without analyte and (ii) achieve [CSs] that minimize hook effects in readout. Concentration of CSs is one of the key aspects of HEBA to maintain a quantitative link to analyte concentration (Figures S3 and S4) with minimized risks for false negatives. Stoichiometry, CS conjugation site and number per antibody, and linker length must be considered with care when developing HEBA for a new target. As discussed above, HEBA needs further optimization to become a more quantitative assay in diverse sample matrices; however, it successfully addresses several issues that conventional molecular assays struggle to address in a point-of-care environment: sensitivity, complex procedures, and robustness (Table 1). PCR or ImmunoPCR would be good examples of conventional assays that provide high sensitivity.48 Due to inherent signal amplification, these assays can theoretically achieve single-molecule-level sensitivity.48 With regard to the ease and robustness of the assays, however, despite progress in simplifying nucleic acid amplification procedures (e.g., loopmediated isothermal amplification and rolling circle amplification), the use of enzymes still limits the assay conditions and complicates assay procedures, especially for protein-based detection. For HEBA, on the other hand, we have eliminated the use of enzymes, which makes the assay completely homogeneousone simple mixing of assay reagents at the beginning is all that is required to perform the assay. Also, as further discussed below, when not relying on enzymes, the assay appears to become less sensitive to sample matrices such as body fluids.49−51 This simple nature makes the assay easily adaptable to a microfluidic digital assay platformunlike other digital assays, there is no need for washing steps, solid substrates like microbeads, or concerns for 1:1 well-to-bead occupancy.19 Of course, there are other molecular recognition approaches based on which a homogeneous, point-of-care diagnostic can be developed: Forster resonance energy transfer (FRET), 45 bioluminescence resonance energy transfer (BRET),45 bimolecular fluorescence complementation,43 and molecular beacons.44 However, the sensitivity of those assays is limited because the molecular mechanism does not provide signal amplification while having to overcome a high background noise due to, for example, the excitation light source. As such, in more realistic samples like body fluids, the sensitivity often remains in the tens to hundreds of nanomolar level, limiting the ability to identify low abundance targets or the potential for sample-splitting-based multiplexing or dilutionbased reduction of sample matrix effects. The molecular mechanism of HEBA involves catalytic amplification of signalgenerating oligonucleotides overcoming the sensitivity issue of other homogeneous approaches. One other assay, the proximity ligation assay (PLA), achieves an amplified reaction with a similar or slightly higher level of amplification power.47 However, unlike HEBA that is completely homogeneous, PLA still requires the sequential addition of reagents and a long assay time due to the use of ligation enzymes followed by polymerases. Further, the requirement for functionality of many enzymes could limit the effective assay conditions and use in diagnostic settings with variable sample matrices. One of the strengths of the molecular machinery described here is that it shares features and should be applicable to most

CONCLUSIONS The performance of HEBA suggests significant potential for use as a point-of-care diagnostic system in the future; HEBA identified the presence of analyte over an extremely wide range (15−4 × 1010 NP molecules, Figure 3d), within 10 min, without any sample purification and/or washing steps, in biofluids such as whole blood and undiluted plasma and at over a range of mild temperatures (15−35 °C) without thermal control during the assay process. Improving the quantitation performance and multiplexing HEBA would be important directions to pursue in the future, especially as a means to include an internal control within each reaction well that could account for sample matrix effects.33 Also, future work can focus on developing a low-cost reader platform as well as minimizing the potential hook effect observed in the current study (Figure 2c,d).34,35 Further, the proximity nature of HEBA opens up applications in the study of multimeric proteins, protein− protein interactions, and colocalization.23,36−42 The general transduction mechanism that converts a protein recognition event into a catalytic nucleic acid complex should also find broad uses in DNA nanotechnology to facilitate industrially important protein sensing capabilities into this field. We believe the simple to implement nature of HEBA should find applications in conjunction with other droplet or digital microfluidic platforms for various drug screening/cell analysis applications in which the amplification provided leads to lowcost complete solutions. METHODS DNA Oligonucleotides. All DNA oligonucleotides (Table S1), including biotinylated oligos, used in this work were purchased through custom oligonucleotide synthesis services from Sigma-Aldrich (St. Louis, MO) and Integrated DNA Technologies, IDT (Coralville, IA). Antibodies. Monoclonal influenza A nucleoprotein antibodies were purchased from SouthernBiotech (Cat. #10770-01 and #1078001, Birmingham, AL). While the vendor provides data concerning the performance of antibodies in binding to NP, the binding to influenza A NPs was confirmed in our lab using full-length NP protein (Novus, Littleton, CO) by sandwich ELISA. HEBA Assay with Biotin−Streptavidin Pair. Necessary DNA oligonucleotide components, multiplex substrate (MS), dummy sequence (DS), catalytic piece 1 and 2 (CS1 and CS2), and the dye sequences (either TS1Dye or TS2Dye conjugated to TET by IDT, Coralville, IA) were prepared in tris(hydroxymethyl)aminomethane ethylenediaminetetraacetic acid (Tris-EDTA) buffer (100× stock from Sigma-Aldrich) supplemented by the concentrations specified below of Mg2+ (Sigma-Aldrich, St. Louis, MO). To note, mixing order of the components is important, and throughout the work in this report, the order varied except that addition of streptavidin or NP samples was 7472

DOI: 10.1021/acsnano.6b02060 ACS Nano 2016, 10, 7467−7475

Article

ACS Nano always following the presence of both CS1 and CS2 in the solution. This is to minimize the formation of CS1−analyte−CS1 and CS2− analyte−CS2 complexes that are incapable of driving the entire reaction forward. Concentration of individual components is as follows: MS, 300 nM; DS, 300 nM; TS1Dye or TS2Dye, 300 nM; CS1 or CS2, 30 nM. This set of concentrations is fixed in this demonstration unless specified. All reactions are at the 250 μL scale in 96-well plates. HEBA Assay for Influenza A Nucleoprotein. Necessary DNA oligonucleotide components, multiplex substrate (MS), dummy sequence (DS), antibody-conjugated catalytic piece 1 and 2 (AbCS1 and Ab-CS2), and the dye sequence (either TS1Dye or TS2Dye) were prepared in Tris-EDTA buffer supplemented with 12.5 mM Mg2+ (Sigma-Aldrich, St. Louis, MO). Again, mixing order of the components varied except for analyte nucleoprotein, which was added after both Ab-CS1 and Ab-CS2. Concentration of individual components is as follows: MS, 300 nM; DS, 300 nM; TS1Dye or TS2Dye, 300 nM; Ab-CS1 or Ab-CS2, 30 nM. All reactions were performed at 250 μL scale in 96-well plates. Analyte samples were universal transport M5 buffer spiked with varying concentrations of nucleoproteins and control samples consisted of the same buffer without NP. Digital HEBA for Influenza A Nucleoprotein. The assay platform, an array of pL wells, was fabricated by standard photolithography techniques in the California NanoSystems Institute at the University of California, Los Angeles. Well dimensions were 15 μm in diameter and 15 μm in height, which results in a ∼2.65 pL volume for each of the 100 920 individual wells. Arrays are designed as hierarchical 10 × 3 arrays of 58 × 58 2.65 pL wells, with each of the 58 × 58 arrays designed to be within a ∼2 mm × 2 mm field of view of our microscope. Experiments for dHEBA were conducted in an identical fashion as analogue HEBA on a well plate except that it was at a 10 μL scale. After being mixed, 5 μL of HEBA solution was put on the polydimethylsiloxane (PDMS) microwells following UV exposure (or oxygen plasma treatment) to facilitate wetting of the wells by HEBA reaction solution as well as bonding to the glass bottom substrate to firmly compartmentalize the reaction solution into the wells. Note that this UV exposure may cause nonspecific adsorption of molecules, such as target proteins, onto the PDMS surface. This nonspecific adsorption is an uncontrolled aspect in this digital assay that affects the expected Poisson distribution if we use the actual assay volume of 260 nL to interpret data. As such, we interpreted data using the 5 μL reaction mixture volume.

ACKNOWLEDGMENTS The authors acknowledge support from the National Science Foundation Emerging Frontiers in Research and Innovation program, Grant No. 1332275. REFERENCES (1) Han, D.; Pal, S.; Nangreave, J.; Deng, Z.; Liu, Y.; Yan, H. DNA Origami with Complex Curvatures in Three-Dimensional Space. Science 2011, 332, 342−346. (2) Andersen, E. S.; Dong, M.; Nielsen, M. M.; Jahn, K.; Subramani, R.; Mamdouh, W.; Golas, M. M.; Sander, B.; Stark, H.; Oliveira, C. L. P.; Pedersen, J. S.; Birkedal, V.; Besenbacher, F.; Gothelf, K. V.; Kjems, J. Self-Assembly of a Nanoscale DNA Box with a Controllable Lid. Nature 2009, 459, 73−76. (3) Modi, S.; Nizak, C.; Surana, S.; Halder, S.; Krishnan, Y. Two DNA Nanomachines Map pH Changes Along Intersecting Endocytic Pathways Inside the Same Cell. Nat. Nanotechnol. 2013, 8, 459−467. (4) Zhang, D. Y.; Winfree, E. Dynamic Allosteric Control of Noncovalent DNA Catalysis Reactions. J. Am. Chem. Soc. 2008, 130, 13921−13926. (5) Shapiro, E.; Ran, T. DNA computing: Molecules Reach Consensus. Nat. Nanotechnol. 2013, 8, 703−705. (6) Ko, S.; Liu, H.; Chen, Y.; Mao, C. DNA Nanotubes as Combinatorial Vehicles for Cellular Delivery. Biomacromolecules 2008, 9, 3039−3043. (7) Goodman, R. P.; Schaap, I. A.; Tardin, C. F.; Erben, C. M.; Berry, R. M.; Schmidt, C. F.; Turberfield, A. J. Rapid Chiral Assembly of Rigid DNA Building Blocks for Molecular Nanofabrication. Science 2005, 310, 1661−1665. (8) Rudchenko, M.; Taylor, S.; Pallavi, P.; Dechkovskaia, A.; Khan, S.; Butler, V. P., Jr.; Rudchenko, S.; Stojanovic, M. N. Autonomous Molecular Cascades for Evaluation of Cell Surfaces. Nat. Nanotechnol. 2013, 8, 580−586. (9) Chen, Y.-J.; Groves, B.; Muscat, R. A.; Seelig, G. DNA Nanotechnology from the Test Tube to the Cell. Nat. Nanotechnol. 2015, 10, 748−760. (10) Hemphill, J.; Deiters, A. DNA Computation in Mammalian Cells: MicroRNA Logic Operations. J. Am. Chem. Soc. 2013, 135, 10512−10518. (11) Zhou, W.-H.; Zhu, C.-L.; Lu, C.-H.; Guo, X.; Chen, F.; Yang, H.-H.; Wang, X. Amplified Detection of Protein Cancer Biomarkers using DNAzyme Functionalized Nanoprobes. Chem. Commun. 2009, 6845−6847. (12) Zhao, X.-H.; Gong, L.; Zhang, X.-B.; Yang, B.; Fu, T.; Hu, R.; Tan, W.; Yu, R. Versatile DNAzyme-Based Amplified Biosensing Platforms for Nucleic Acid, Protein, and Enzyme Activity Detection. Anal. Chem. 2013, 85, 3614−3620. (13) Pierangeli, S. S.; Harris, E. N. A Protocol for Determination of Anticardiolipin Antibodies by ELISA. Nat. Protoc. 2008, 3, 840−848. (14) Cuendet, M.; Mesecar, A. D.; DeWitt, D. L.; Pezzuto, J. M. An ELISA Method to Measure Inhibition of the COX Enzymes. Nat. Protoc. 2006, 1, 1915−1921. (15) Ambrosi, A.; Airò, F.; Merkoçi, A. Enhanced Gold NanoparticleBased ELISA for a Breast Cancer Biomarker. Anal. Chem. 2010, 82, 1151−1156. (16) Brena, B. M.; Arellano, L.; Rufo, C.; Last, M. S.; Montano, J.; Cerni, E. E.; Gonzalez-Sapienza, G.; Last, J. A. ELISA as an Affordable Methodology for Monitoring Groundwater Contamination by Pesticides in Low-Income Countries. Environ. Sci. Technol. 2005, 39, 3896−3903. (17) Kang, X.; Li, Y.; Fan, L.; Wei, J.; Zhu, X.; Hu, Y.; Li, J.; Chang, G.; Zhu, Q.; Liu, H.; Yang, Y. Development of an ELISA-Array for Simultaneous Detection of five Encephalitis Viruses. Virol. J. 2012, 9, 56. (18) Asensio, L.; González, I.; García, T.; Martín, R. Determination of Food Authenticity by Enzyme-Linked Immunosorbent Assay (ELISA). Food Control 2008, 19, 1−8.

ASSOCIATED CONTENT S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.6b02060. Additional details, materials and methods, figures, and tables (PDF)

AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Author Contributions

D.K. and D.D. brainstormed and developed the HEBA concept. D.K., D.D., A.O., and O.G. planned the experiments. D.K. implemented HEBA, optimized performance, and performed the experiments. D.K. and O.G. contributed to reagent selection. All of the authors discussed and analyzed the data and contributed to writing the manuscript. Notes

The authors declare no competing financial interest. 7473

DOI: 10.1021/acsnano.6b02060 ACS Nano 2016, 10, 7467−7475

Article

ACS Nano (19) 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. Single-Molecule Enzyme-Linked Immunosorbent Assay Detects Serum Proteins at Subfemtomolar Concentrations. Nat. Biotechnol. 2010, 28, 595−599. (20) Zhang, S.; Garcia-D’Angeli, A.; Brennan, J. P.; Huo, Q. Predicting Detection Limits of Enzyme-Linked Immunosorbent Assay (ELISA) and Bioanalytical Techniques in General. Analyst 2014, 139, 439−445. (21) Mocanu, M.-M.; Váradi, T.; Szöllosi, J.; Nagy, P. Comparative Analysis of Fluorescence Resonance Energy Transfer (FRET) and Proximity Ligation Assay (PLA). Proteomics 2011, 11, 2063−2070. (22) Ullman, E. F. The Immunoassay Handbook; Elsevier, 2013; pp 246−254. (23) Fredriksson, S.; Gullberg, M.; Jarvius, J.; Olsson, C.; Pietras, K.; Gustafsdottir, S. M.; Ostman, A.; Landegren, U. Protein Detection Using Proximity-Dependent DNA Ligation Assays. Nat. Biotechnol. 2002, 20, 473−477. (24) Gullberg, M.; Gustafsdottir, S. M.; Schallmeiner, E.; Jarvius, J.; Bjarnegard, M.; Betsholtz, C.; Landegren, U.; Fredriksson, S. Cytokine Detection by Antibody-Based Proximity Ligation. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 8420−8424. (25) Lundberg, M.; Thorsen, S. B.; Assarsson, E.; Villablanca, A.; Tran, B.; Gee, N.; Knowles, M.; Nielsen, B. S.; Gonzalez Couto, E.; Martin, R.; Nilsson, O.; Fermer, C.; Schlingemann, J.; Christensen, I. J.; Nielsen, H. J.; Ekstrom, B.; Andersson, C.; Gustafsson, M.l; Brunner, N.; Stenvang, J.; Fredriksson, S. Multiplexed Homogeneous Proximity Ligation Assays for High-Throughput Protein Biomarker Research in Serological Material. Mol. Cell. Proteomics 2011, 10, M110.004978. (26) Zhang, D. Y.; Turberfield, A. J.; Yurke, B.; Winfree, E. Engineering Etropy-driven Ractions and Ntworks Ctalyzed by DNA. Science 2007, 318, 1121−1125. (27) Zhang, D. Y.; Winfree, E. Control of DNA Srand Diplacement Kietics Using Toehold Exchange. J. Am. Chem. Soc. 2009, 131, 17303− 17314. (28) Neish, C. S.; Martin, I. L.; Henderson, R. M.; Edwardson, J. M. Direct Visualization of Ligand-Protein Interactions Using Atomic Force Microscopy. Br. J. Pharmacol. 2002, 135, 1943−1950. (29) Portela, A.; Digard, P. The Influenza Virus Nucleoprotein: a Multifunctional RNA-Binding Protein Pivotal to Virus Replication. J. Gen. Virol. 2002, 83, 723−734. (30) Kelley, S. O.; Mirkin, C. A.; Walt, D. R.; Ismagilov, R. F.; Toner, M.; Sargent, E. H. Advancing the Speed, Sensitivity and Accuracy of Biomolecular Detection Using Multi-Length-Scale Engineering. Nat. Nanotechnol. 2014, 9, 969−980. (31) Zhang, C.; Xing, D. Single-Molecule DNA Amplification and Analysis Using Microfluidics. Chem. Rev. 2010, 110, 4910−4947. (32) Kreutz, J. E.; Munson, T.; Huynh, T.; Shen, F.; Du, W.; Ismagilov, R. F. Theoretical Design and Analysis of Multivolume Digital Assays with Wide Dynamic Range Validated Experimentally with Microfluidic Digital PCR. Anal. Chem. 2011, 83, 8158−8168. (33) Bell, J.; Bonner, A.; Cohen, D. M.; Birkhahn, R.; Yogev, R.; Triner, W.; Cohen, J.; Palavecino, E.; Selvarangan, R. Multicenter Clinical Evaluation of the Novel Alere i Influenza A&B Isothermal Nucleic Acid Amplification Test. J. Clin. Virol. 2014, 61, 81−86. (34) Koydemir, H. C.; Gorocs, Z.; Tseng, D.; Cortazar, B.; Feng, S.; Chan, R. Y. L.; Burbano, J.; McLeod, E.; Ozcan, A. Rapid Imaging, Detection and Quantification of Giardia Lamblia Cysts Using MobilePhone Based Fluorescent Microscopy and Machine Learning. Lab Chip 2015, 15, 1284−1293. (35) Wei, Q.; Luo, W.; Chiang, S.; Kappel, T.; Mejia, C.; Tseng, D.; Chan, R. Y. L.; Yan, E.; Qi, H.; Shabbir, F.; Ozkan, H.; Feng, S.; Ozcan, A. Imaging and Sizing of Single DNA Molecules on a Mobile Phone. ACS Nano 2014, 8, 12725−12733. (36) Gu, L.; Li, C.; Aach, J.; Hill, D. E.; Vidal, M.; Church, G. M. Multiplex Single-Molecule Interaction Profiling of DNA-Barcoded Proteins. Nature 2014, 515, 554−557.

(37) Söderberg, O.; Gullberg, M.; Jarvius, M.; Ridderstrale, K.; Leuchowius, K.-J.; Jarvius, J.; Wester, K.; Hydbring, P.; Bahram, F.; Larsson, L.-G.; Landegren, U. Direct Observation of Individual Endogenous Protein Complexes in situ by Proximity Ligation. Nat. Methods 2006, 3, 995−1000. (38) Buckhout-White, S.; Spillmann, C. M.; Algar, R.; Khachatrian, A.; Melinger, J. S.; Goldman, E. R.; Ancona, M. G.; Medintz, I. L. Assembling Programmable FRET-Based Photonic Networks Using Designer DNA Scaffolds. Nat. Commun. 2014, 5, 5615. (39) Wang, L.; Hall, J. G.; Lu, M.; Liu, Q.; Smith, L. M. A DNA Computing Readout Operation Based on Structure-Specific Cleavage. Nat. Biotechnol. 2001, 19, 1053−1059. (40) Penchovsky, R.; Breaker, R. R. Computational Design and Experimental Validation of Oligonucleotide-Sensing Allosteric Ribozymes. Nat. Biotechnol. 2005, 23, 1424−1433. (41) Seelig, G.; Soloveichik, D.; Zhang, D. Y.; Winfree, E. EnzymeFree Nucleic Acid Logic Circuits. Science 2006, 314, 1585−1588. (42) Seelig, G.; Yurke, B.; Winfree, E. Catalyzed Relaxation of a Metastable DNA Fuel. J. Am. Chem. Soc. 2006, 128, 12211−12220. (43) Kerppola, T. K. Design and Implementation of Biomolecular Fluorescence Complementation (BiFC) Assays for the Visualization of Protein Interactions in Living Cells. Nat. Protoc. 2006, 1, 1278−1286. (44) Tyagi, S.; Kramer, F. R. Molecular Beacons: Probes that Fluoresce Upon Hybridization. Nat. Biotechnol. 1996, 14, 303−308. (45) Xu, Y.; Piston, D. W.; Johnson, C. H. A Bioluminescence Resonance Energy Transfer (BRET) System: Application to Interacting Circadian Clock Proteins. Proc. Natl. Acad. Sci. U. S. A. 1999, 96, 151−156. (46) Lee, H.; Park, J.-E.; Nam, J.-M. Bio-Barcode Gell Assay for MicroRNA. Nat. Commun. 2014, 5, 3367. (47) Fredriksson, S.; Gullberg, M.; Jarvius, J.; Olsson, C.; Pietras, K.; Gustafsdottir, S. M.; Ostman, A.; Landegren, U. Protein Detection Using Proximity-Dependent DNA Ligation Assays. Nat. Biotechnol. 2002, 20, 473−477. (48) Niemeyer, C. M.; Adler, M.; Wacker, R. Detecting Antigens by Quantitative Immuno-PCR. Nat. Protoc. 2007, 2, 1918−1930. (49) Al-Soud, W. A.; Radstrom, P. Purification and Characterization of PCR-Inhibitory Components in Blood Cells. Journal of Clinical Microbiology 2001, 39, 485−493. (50) Bickley, J.; Short, J. K.; McDowell, D. G.; Parkes, H. C. Polymerase Chain Reaction (PCR) Detection of Listeria Monocytogene in Diluted Milk and Reverse PCR Inhibition Caused by Calcium Ions. Lett. Appl. Microbiol. 1996, 22, 153−158. (51) Funes-Huacca, M. E.; Opel, K.; Thompson, R.; McCord, B. R. A Comparison on the Effects of PCR Inhibition in Quantitative PCR and Forensic STR Analysis. Electrophoresis 2011, 32, 1084−1089. (52) Cong, L.; Ran, F. A.; Cox, D.; Lin, S.; Barretto, R.; Habib, N.; Hsu, P. D.; Wu, X.; Jiang, W.; Marraffini, L. A.; Zhang, F. Multiplex Genome Engineering Using CRISPR/Cas Systems. Science 2013, 339, 819−823. (53) Wood, M. W.; Nordone, S. K.; Vaden, S. L.; Breitschwerdt, E. B. Assessment of Urine Solute and Matrix Effects on the Performance of an Exzyme-Linked Immunosorbent Assay for Measurement of Interleukin-6 in Dog Urine. J. Vet. Diagn. Invest. 2011, 23, 316−320. (54) Chiu, M. L.; Lawi, W.; Snyder, S. T.; Wong, P. K.; Liao, J. C.; Gau, V. Matrix Effects - A Challenge Toward Automation of Molecular Analysis. JALA 2010, 15, 233−242. (55) Ganda, V.; Patel, R. Review Article on Matrix Effect in Bioanalytical Method Development. International Journal of Medipharm Research 2015, 1, 159−165. (56) Zuker, M. Mfold Web Server for Nucleic Acid Folding and Hybridization Prediction. Nucleic Acids Res. 2003, 31, 3406−341. (57) Dixit, C. K.; Vashist, S. K.; O’Neill, F. T.; O’Reilly, B.; MacCraith, B. D.; O’Kennedy, R. Development of a High Sensitivity Rapid Sandwich ELISA Procedure and its Comparison with the Conventional Approach. Anal. Chem. 2010, 82, 7049−7052. (58) Fredriksson, S.; Gullberg, M.; Jarvius, J.; Olsson, C.; Pietras, K.; Gustafsdottir, S. M.; Ostman, A.; Landegren, U. Protein Detection 7474

DOI: 10.1021/acsnano.6b02060 ACS Nano 2016, 10, 7467−7475

Article

ACS Nano Using Proximity-Dependent DNA Ligation Assay. Nat. Biotechnol. 2002, 20, 473−477. (59) Gullberg, M.; Gustafsdottir, S. M.; Schallmeiner, E.; Jarvius, J.; Bjarnegard, M.; Betsholtz, C.; Landegren, U.; Fredriksson, S. Cytokine Detection by Antibody-Based Proximity Ligation. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 8420−8424.

7475

DOI: 10.1021/acsnano.6b02060 ACS Nano 2016, 10, 7467−7475