Time-Gated FRET and DNA-Based Photonic ... - ACS Publications

Aug 8, 2017 - Electronic Science and Technology Division, Code 6876, U.S.. Naval Research Laboratory, 4555 Overlook Avenue SW, Washington, DC 20375, U...
2 downloads 8 Views 3MB Size
Download PDF
Article pubs.acs.org/acssensors

Time-Gated FRET and DNA-Based Photonic Molecular Logic Gates: AND, OR, NAND, and NOR Melissa Massey,† Igor L. Medintz,‡ Mario G. Ancona,§ and W. Russ Algar*,† †

Department of Chemistry, University of British Columbia, 2036 Main Mall, Vancouver, British Columbia V6T 1Z1, Canada Center for Bio/Molecular Science and Engineering, Code 6900 and §Electronic Science and Technology Division, Code 6876, U.S. Naval Research Laboratory, 4555 Overlook Avenue SW, Washington, DC 20375, United States

‡

S Supporting Information *

ABSTRACT: Molecular logic devices (MLDs) constructed from DNA are promising for applications in bioanalysis, computing, and other applications requiring Boolean logic. These MLDs accept oligonucleotide inputs and generate fluorescence output through changes in structure. Although fluorescent dyes are most common in MLD designs, nontraditional luminescent materials with unique optical properties can potentially enhance MLD capabilities. In this context, luminescent lanthanide complexes (LLCs) have been largely overlooked. Here, we demonstrate a set of high-contrast DNA photonic logic gates based on toehold-mediated strand displacement and time-gated FRET. The gates include NAND, NOR, OR, and AND designs that accept two unlabeled target oligonucleotide sequences as inputs. Bright “true” output states utilize time-gated, FRET-sensitized emission from an Alexa Fluor 546 (A546) dye acceptor paired with a luminescent terbium cryptate (Tb) donor. Dark “false” output states are generated through either displacement of the A546, or through competitive and sequential quenching of the Tb or A546 by a dark quencher. Time-gated FRET and the long luminescence lifetime and spectrally narrow emission lines of the Tb donor enable 4−10-fold contrast between Boolean outputs, ≤10% signal variation for a common output, multicolor implementation of two logic gates in parallel, and effective performance in buffer and serum. These metrics exceed those reported for many other logic gate designs with only fluorescent dyes and with other non-LLC materials. Preliminary three-input AND and NAND gates are also demonstrated. The powerful combination of an LLC FRET donor with DNA-based logic gates is anticipated to have many future applications in bioanalysis. KEYWORDS: molecular logic, phosphorescence, fluorescence, time-resolved FRET, DNA, oligonucleotide, lanthanide, multiplexed

D

silicon-based digital electronics and there has been much interest in mimicking these devices at the molecular level.21 Molecular logic devices (MLDs) differ from semiconductorbased logic devices in that the inputs are (bio)chemical in nature, and the outputs tend to be photonic or electrochemical, but nevertheless have logical input/output characteristics that are analogous to their traditional electronic counterparts.22,23 For example, an ideal molecular AND gate produces a photonic output only if it receives two appropriate (bio)chemical inputs. MLDs are thus promising components for nontraditional information processing systems that function on the nanoscale and in biological environments, whether in the context of computation or bioanalysis. Regarding the latter, there is the potential to use MLDs to concurrently analyze multiple target biomarkers, make an autonomous decision, and produce a single “true/false” output. Exploiting the advantages of DNA as a building block, a variety of oligonucleotide-based MLDs have been developed for

NA oligonucleotides have a long and still growing history as building blocks for the design of biomolecular probes for applications in bioanalysis.1−3 Linear probes (e.g., for in situ hybridization or immobilized on surfaces), hairpins (e.g., molecular beacons), triplex probes, G-quadruplexes, and aptamers have been used to detect targets that range from complementary DNA and various forms of RNA, to small molecules, ions, and proteins.1,4−7 More complex DNA nanostructures have also emerged for applications in bioanalysis, drug delivery, logic gates, and molecular circuitry.1,8−12 Oligonucleotides derive advantages in these applications from their ability to selectively and predictably bind their complements with high affinity, as well as their ease of synthesis and modification, including chemical handles for conjugation (e.g., with fluorescent dyes or nanoparticles).13,14 Such labeling has frequently been leveraged in Fö rster resonance energy transfer (FRET) configurations for fluorescent detection of nucleic acid targets, and for photonic output from DNA machines, logic gates, and circuits.9,15−19 Boolean logic gates are traditionally defined as devices that perform logical operations with two or more inputs to yield a single digital output.20 Logic gates are the underpinning of © XXXX American Chemical Society

Received: May 26, 2017 Accepted: July 24, 2017

A

DOI: 10.1021/acssensors.7b00355 ACS Sens. XXXX, XXX, XXX−XXX

Article

ACS Sensors bioanalysis and molecular computing.2,24−30 Two-input logic gates based on DNA origami, triplex DNA, molecular beacons, and aptamer structures have been reported for DNA, protein, and (deoxy)ribozyme inputs.24,31−34 More sophisticated structures have also been developed; for example, a concatenated series of AND gates was created by transforming a linear oligonucleotide into a three-way DNA junction through an ordered succession of strand displacements with hairpin oligonucleotide inputs.35 Despite rapid progress in research toward DNA-based MLDs, several limitations remain. Although there are three types of two-input Boolean logic gates (AND, OR, XOR) and three types of inverted two-input Boolean logic gates (NAND, NOR, XNOR), many reports address only one or two types of MLD. Signal contrast between the “true/false” logic states also tends to be limited, and several interesting photonic materials, such as luminescent lanthanide complexes (LLCs), have been overlooked as components for MLDs. Fluorescent dyes and, to a much lesser extent, quantum dots and upconversion nanoparticles have been the most prominent optical reporters.26,28,36−38 Here, we design a set of high-contrast, modularly designed, two-input oligonucleotide-based logic gates that rely on an LLC as a FRET donor for generating photonic output. LLCs offer significant optical advantages, including multiple narrow emission lines, resistance to photobleaching, large spectral separation between excitation and emission, and excited-state lifetimes on the order of hundreds of microseconds to milliseconds.39−41 A conventional fluorescent dye can adopt these long lifetimes when it is an acceptor for the LLC in a FRET pair, such that time-gated FRET detection of unlabeled DNA targets is possible. We capitalize on the favorable optical properties of a luminescent terbium(III) cryptate42 (abbreviated as Tb) to design and evaluate two-input AND, OR, NAND, and NOR logic gates. The logic gates comprise assemblies of oligonucleotides in combination with Tb, fluorescent dye labels, quencher labels, and toeholds at strategic positions. Unlabeled target oligonucleotides act as inputs that displace labeled oligonucleotide(s) from the initial logic gate structure and alter competitive and/or sequential FRET pathways with the Tb as a donor. The photonic output is time-gated, FRET-sensitized emission from a fluorescent dye, which was selected as an acceptor for the Tb. The observed contrast between “true/false” output states was among the highest reported to date, detection in serum was demonstrated, two-color multiplexed logic operations were possible, and good preliminary results were obtained for three-input gates. This work provides the initial foundation for designing a new subclass of DNA-based photonic MLDs with FRET and Tb or other LLC donors for sensing applications.

■

Absorbance and PL Measurements. An Infinite M1000 Pro multifunction plate reader (Tecan, Morrisville, NC) was used for absorbance and PL measurements and offered monochromator-based selection of excitation/emission wavelengths, as well as lag time and detector integration time settings for time-gated measurements. Samples were measured in 96-well microtiter plates. Time-gated PL measurements were made with a 1000 μs lag time after flash excitation and a 1000 μs integration time. The combination of long lag time and integration time was found to provide the best contrast in logic function. In most experiments, single-wavelength time-gated measurements were made instead of full spectra. These wavelengths were 490 nm for Tb, 520 nm for A488, and 572 nm for A546. Sample Preparation and Assays. To prepare oligonucleotide logic gates, constituent sequences were mixed together at the desired stoichiometry and quantity in Tris-borate saline (TBS) buffer (90 mM Tris-borate, 137 mM NaCl, 2.7 mM KCl, pH 7.6). These stock solutions of logic gates were heated to 95 °C with subsequent slow cooling to room temperature before aliquoting and dilution to prepare individual samples. Assembled logic gates were mixed with input oligonucleotides and left at room temperature for a minimum of 75 min. Typically, truth tables were done with 1 equiv of input(s) and 25 or 50 nM logic gate, kinetics were done with 2 equiv of input(s) and 100 nM logic gate, and calibrations were done with the indicated numbers of equivalents of input(s) and 25 nM logic gate. Details for serum samples can be found in the SI. Data Analysis. Time-gated PL intensity data were processed as either measured intensities in arbitrary units or normalized to the maximum intensity (i.e., a “true” state). The normalization facilitated interpretation of results in terms of the idealized logic gate outputs of “true/false.” The different combinations of two inputs are denoted by (0,0), (1,0), (0,1), and (1,1), corresponding to no added inputs, one of the two inputs added, and both inputs added. Error bars in graphs are the standard deviations from at least three replicate measurements.

■

RESULTS AND DISCUSSION FRET Pairs and Pathways. The logic gates used three FRET pathways to generate Boolean-like fluorescence signaling. In all cases, the FRET pathway that led to an output value of “true” was FRET from a Tb donor to an A546 acceptor dye. The Förster distance for the Tb-A546 FRET pair was R0 = 6.2 nm. Two strategies were used to generate an output value of “false”. The first strategy, which was used with NOR and NAND logic gates, was to displace the A546-labeled oligonucleotide with the loss of the proximity needed for measurable Tb-to-A546 FRET. The second strategy, which was used with AND and OR logic gates, was introduction of a FRET-based quenching channel, either competitively from the Tb donor to an IabFQ dark quencher (R0 = 5.7 nm for the TbIabFQ FRET pair), or sequentially from a Tb-sensitized A546 to an IabFQ dark quencher (R0 = 5.6 nm for the A546-IabFQ pair). Figure 1A summarizes these FRET strategies for implementing “true/false” output states, and Figure 1B shows the absorption and emission spectra for each luminophore. Figure 1C is relevant to the competitive dark quencher strategy for generating a “false” output, showing the predicted outcome of the FRET competition between the Tb-to-A546 “true” output pathway and the Tb-to-IabFQ “false” output pathway as a function of the separation distance, r, for each FRET pair. The “false” pathway dominates for a wide range of ratios of the separation distances, rTb‑A546/rTb‑IabFQ, and the initial states of the AND and OR gates are within this region. AND-I Gate. A generic AND gate MLD starts with a “false” or dark fluorescent signal, accepts two (bio)chemical inputs, and outputs a “true” or bright fluorescence signal only if both inputs are present. Figure 2A shows the design of our AND-I logic gate. It consisted of two 19 nucleotide (nt) template arms

EXPERIMENTAL SECTION

Materials. Oligonucleotides were from Integrated DNA Technologies (Coralville, IA) and HPLC purified when obtained with dye or linker modifications. Sequences and modifications are listed in the Supporting Information (SI, see Table S1). Alexa Fluor 546 (A546) C5 maleimide and Alexa Fluor 488 (A488) C5 maleimide were from Thermo-Fisher Scientific (Carlsbad, CA). Amine-reactive Lumi4-TbNHS was a kind gift from Lumiphore Inc. (Berkeley, CA) and was developed in the Raymond Lab.42 The protective and highly stable environment that the Lumi4 cryptand provides for the Tb(III) lends exceptional brightness to this LLC. Methods for labeling oligonucleotides with A546 or A488 maleimide and Lumi4-Tb-NHS are described in the SI.43 B

DOI: 10.1021/acssensors.7b00355 ACS Sens. XXXX, XXX, XXX−XXX

Article

ACS Sensors

threshold of ∼0.12 for switching between the “false/true” output states (see SI for discussion). For AND-I and all other logic gates, there was virtually no response when challenged with a noncomplementary, non-input sequence. This data and full PL spectra for AND-I and all other logic gates are shown in Figure S1. When tested in 90% v/v serum, contrast was reduced but still analytically useful with a threshold ≤0.5. Additional experiments were done to further characterize the behavior of the AND-I logic gate. Figure 2C shows the change in the time-gated A546 PL intensity as different numbers of input equivalents (i.e., mole fractions) were added as (χ1,χ2), 0 ≤ χ1, χ2 ≤ 1. The trend for increasing χ1 and χ2 (with χ1 = χ2) had upward curvature versus a linear trend with only one of the two inputs, (χ1,0) or (0,χ2). These trends are most easily understood through the statistics of two independent binding events. With less than 1 equiv of each input, some gates had only one of the two inputs bound, whereas some gates had both inputs bound. The measured A546 PL intensity, IA546, was expected to follow eq 1 (see SI for details), where the coefficients D, C, B, and A are from the data in Figure 2B for (0,0) through (1,1). F and K are fitting parameters to account for the arbitrary units of PL intensity. As shown in Figure 2C, the fit of such a parametrization of eq 1 to the data was excellent for both inputs individually and together. The response saturated beyond 1 equiv of both inputs (see Figure S2). IA546 = F[Aχ1 χ2 + B(1 − χ1 )χ2 + Cχ1 (1 − χ2 ) + D(1 − χ1 )(1 − χ2 )] + K

Figure 1. (A) FRET pathways (with initial excitation of the Tb) utilized for “true/false” time-gated outputs with DNA-based molecular logic gates: (i) “true” output for all logic gates; (ii) “false” output for NAND and NOR logic gates from displacement of A546; and (iii) “false” output for AND and OR gates from competitive or sequential quenching. In (iii), the rates of competitive (1a-b) and sequential (2) FRET pathways are determined by the donor−acceptor distances (r). (B) Normalized absorption and emission spectra for Tb, A546, and IabFQ. (C) Percentage of quenching FRET events (1a) versus total FRET events (1a plus 1b) as a function of donor−acceptor distances. The initial states of the logic gates are expected to lie within the dashed boundary.

(1)

Kinetic traces of the changes in Tb and A546 time-gated PL are shown in Figure 2D and demonstrated that AND-I fully responded to its inputs in less than an hour. The kinetic data also showed that the Tb and A546 time-gated PL intensities increased in parallel, confirming that the “true” output signal was generated by alleviation of the competitive quenching of Tb by IabFQ and restoration of Tb-to-A546 FRET. AND-II Gate. Figure 3A shows the design of our AND-II logic gate. The gate consisted of an 18 nt reporter, labeled with Tb at its 3′ terminus and A546 at its 5′ terminus, that hybridized centrally along the length of an unlabeled 54 nt template oligonucleotide. In the initial state, each flanking end of the template was hybridized with a 13 nt blocking sequence labeled with IabFQ, where one IabFQ label was located adjacent to the Tb and the other was located adjacent to the A546. The result was efficient quenching of the Tb emission and, if there was any FRET sensitization of A546, efficient quenching of its emission (as per Figure 1A-iii, C). When one target oligonucleotide was added, as either the (1,0) or (0,1) input state, it hybridized to its complementary template arm through toehold-mediated strand displacement of the corresponding IabFQ-labeled blocking sequence. The remaining blocking sequence and IabFQ label continued to quench either the Tb emission or A546 emission, as applicable, with only a small increase in FRET-sensitized A546 emission. When both target oligonucleotides were added as the (1,1) input state, both blocks and IabFQ labels were displaced. The result was efficient Tb-to-A546 FRET and a large increase in the timegated FRET-sensitized A546 PL. Figure 3B shows the experimental truth table for AND-II in terms of the relative A546 PL intensity. Normalized outputs of 0.01, 0.10, 0.09, and 1.0 were measured for the (0,0), (1,0), (0,1), and (1,1) input states. Similar to AND-I, robust AND logic was achieved with

joined by an internal amine linker that was labeled with Tb. Both the 3′ and 5′ termini of the template were labeled with A546. In the initial state, each arm was hybridized with a 13 nt blocking sequence labeled with an IabFQ dark quencher, where both IabFQ labels were located in close proximity to the Tb. The result was efficient quenching of the Tb emission and negligible A546 emission (pathway 1a in Figure 1A-iii with expected quenching as per Figure 1C). When one target oligonucleotide was added as either the (1,0) or (0,1) input state, it hybridized to its complementary template arm through toehold-mediated strand displacement of an IabFQ-labeled blocking sequence. The remaining blocking sequence and IabFQ label continued to quench the Tb emission with only a small increase in FRET-sensitized A546 emission. When both target oligonucleotides were added as the (1,1) input state, both blocks and IabFQ labels were displaced. The result was Tb-to-A546 FRET and a large increase in the time-gated FRET-sensitized A546 PL. Figure 2B shows the experimental truth table for AND-I in terms of the relative A546 PL intensity. Normalized outputs of 0.01, 0.09, 0.11, and 1.0 were measured for the (0,0), (1,0), (0,1), and (1,1) input states in buffer. AND logic was achieved with 9-fold contrast and a C

DOI: 10.1021/acssensors.7b00355 ACS Sens. XXXX, XXX, XXX−XXX

Article

ACS Sensors

Figure 2. AND-I gate. (A) Truth table and DNA structures. The template (t), blocking (b1, b2), and input (I1, I2) oligonucleotides, Tb, A546, and IabFQ are labeled. Toeholds for strand displacement by the inputs are circled. Efficient FRET is shown as arrows. Suppressed FRET is shown with xarrows. (B) Relative time-gated A546 PL intensity when challenged with an equivalent of one or both inputs in buffer and 90% v/v serum (normalized to mimic truth table-like output). The “false/true” threshold is the dashed line. (C) Calibration curve when challenged with different equivalents of one or both inputs. The fitted curves are from eq 1 and parametrized with the values in Figure 1B. (D) Kinetic traces of time-gated Tb and A546 PL intensity when the gate was challenged with one or both inputs. The data in panels (C) and (D) are with buffer.

Figure 3. AND-II gate. (A) Truth table and DNA structures. The template (t), blocking (b1, b2), reporter (r), and input (I1, I2) oligonucleotides, Tb, A546, and IabFQ are labeled. Toeholds for strand displacement by the inputs are circled. Efficient/suppressed FRET is shown as an arrow/xarrow. (B) Relative time-gated A546 PL intensity when challenged with an equivalent of one or both inputs in buffer and 90% v/v serum (normalized to mimic truth table-like output). The “false/true” threshold is the dashed line. (C) Calibration curve when challenged with different equivalents of one or both inputs. The fitted curves are from eq 1. (D) Kinetic traces of time-gated Tb and A546 PL intensity when the gate was challenged with one or both inputs. The data in panels (C) and (D) are with buffer.

10-fold contrast and a threshold of ∼0.12 for switching between the “false/true” output states. Contrast was again reduced when tested in serum but still useful with a threshold ≤0.6. Figure 3C shows the change in the time-gated A546 PL intensity with different numbers of input equivalents. Again, the trend with increasing χ1 and χ2 (with χ1 = χ2) had upward curvature, confirming that binding of these inputs was also independent. The data was well-fit by eq 1 parametrized to the values in Figure 3B. Figure 3D shows kinetic traces of the changes in both Tb and A546 time-gated PL when AND-II was challenged with its different input sequences. AND-II was slightly slower to respond than AND-I, with full response requiring closer to 2 h. The kinetic data also highlighted the different signaling mechanisms between AND-I and AND-II. When only the IabFQ blocking sequence adjacent to the Tb was displaced by

its target as a (1,0) input state with AND-II, the Tb intensity increased but the A546 intensity did not, demonstrating that quenching of A546 by its adjacent IabFQ was able to maintain a “false” output signal despite the restored Tb-to-A546 FRET. Similarly, when only the IabFQ blocking sequence adjacent to the A546 was displaced by its target as a (0,1) input state, dominant quenching of the Tb by its adjacent IabFQ prevented significant FRET-sensitization of A546 PL and thus maintained a “false” output signal. OR Gate. A generic OR gate MLD starts with a “false” or dark fluorescent signal, accepts two (bio)chemical inputs, and outputs a “true” or bright fluorescence signal if either one or both of the inputs are present. Figure 4A shows the design of our OR logic gate. The gate consisted of an unlabeled 33 nt template, which was initially hybridized with two other oligonucleotides: (i) a 13 nt blocking sequence labeled with D

DOI: 10.1021/acssensors.7b00355 ACS Sens. XXXX, XXX, XXX−XXX

Article

ACS Sensors

Figure 4. OR gate. (A) Truth table and DNA structures. The template (t), blocking (b), reporter (r), and input (I1, I2) oligonucleotides, Tb, A546, and IabFQ are labeled. Toeholds for strand displacement by the inputs are circled. Efficient/suppressed FRET is shown as an arrow/x-arrow. (B) Relative time-gated A546 PL intensity when challenged with an equivalent of one or both inputs in buffer and 90% v/v serum (normalized to mimic truth table-like output). The “false/true” threshold is the dashed line. (C) Calibration curve when challenged with different equivalents of one or both inputs. The fitted curves are from eq 1. (D) Kinetic traces of time-gated Tb and A546 PL intensity when the gate was challenged with one or both inputs. The data in panels (C) and (D) are with buffer.

Figure 5. NAND gate. (A) Truth table and DNA structures. The template (t), blocking (b), reporter (r), and input (I1, I2) oligonucleotides, Tb, and A546 are labeled. Toeholds for strand displacement by the inputs are circled. Efficient FRET is shown as an arrow. (B) Relative time-gated A546 PL intensity when challenged with an equivalent of one or both inputs in buffer and 90% v/v serum (normalized to mimic truth table-like output). The “true/false” threshold is the dashed line. (C) Calibration curve when challenged with different equivalents of one or both inputs. The fitted curves are from eq 1. (D) Kinetic traces of time-gated Tb and A546 PL intensity when the gate was challenged with one or both inputs. The data in panels (C) and (D) are with buffer.

FRET-sensitized A546 PL. When the other target oligonucleotide was added, as the (0,1) input state, it hybridized to the complementary reporter and displaced it from the template strand. As the IabFQ-labeled blocking sequence was still hybridized to the template oligonucleotide, the close proximity between the Tb and the IabFQ was broken with loss of FRET quenching and a large increase in the time-gated, FRETsensitized A546 PL. When both target oligonucleotides were added as the (1,1) input state, both of the foregoing displacement mechanisms led to a large increase in the timegated, FRET-sensitized A546 PL. Figure 4B shows the experimental truth table for the OR gate in terms of the relative A546 PL intensity. Normalized outputs of 0.10, 1.0, 0.90, and 0.98 were measured for the (0,0), (1,0), (0,1), and

IabFQ at its 5′ terminus; and (ii) an 18 nt reporter labeled with Tb at its 3′ terminus and A546 at its 5′ terminus. The IabFQ label on the blocking sequence was located adjacent to the Tb. A one base-pair mismatch was centrally located in the template/reporter duplex portion of the gate to generate instability and a faster kinetic response (see SI for discussion). The result was efficient FRET quenching of the Tb emission and minimal time-gated A546 emission. When one target oligonucleotide was added, as the (1,0) input state, the input strand hybridized to its complementary site on the template arm through toehold-mediated strand displacement of the IabFQ-labeled blocking sequence. With the loss of the IabFQ, efficient Tb-to-A546 FRET across the reporter oligonucleotide was possible and there was a large increase in the time-gated, E

DOI: 10.1021/acssensors.7b00355 ACS Sens. XXXX, XXX, XXX−XXX

Article

ACS Sensors

Figure 6. NOR gate. (A) Truth table and DNA structures. The template (t), antenna (a), reporter (r), input (I1, I2) oligonucleotides, Tb, and A546 are labeled. Toeholds for strand displacement by the inputs are circled. Efficient FRET is shown as an arrow. (B) Relative time-gated A546 PL intensity when challenged with an equivalent of one or both inputs in buffer and 90% v/v serum (normalized to mimic truth table-like output). The “true/false” threshold is the dashed line. (C) Calibration curve when challenged with different equivalents of one or both inputs. The fitted curves are from eq 1. (D) Kinetic traces of time-gated Tb and A546 PL intensity when the gate was challenged with one or both inputs. The data in panels (C) and (D) are with buffer.

the blocking sequence for input (1,0) remained hybridized to the template. The target sequence for the (0,1) input was only able to bind efficiently when both inputs were present. Such a structure is known as a “hidden toehold”.44 Both blocking sequences were displaced only when both targets were added as the (1,1) input state. The result was loss of Tb-to-A546 FRET, and thus a large decrease in the time-gated A546 PL. Figure 5B shows the experimental truth table for the NAND gate in terms of the A546 PL intensity. Normalized outputs of 0.93, 1.0, 0.86, and 0.23 were measured for the (0,0), (1,0), (0,1), and (1,1) input states, respectively. NAND logic was achieved with ∼4fold contrast and a threshold of ∼0.55 for switching between “true/false” output states. The threshold was similar at ≥0.6 when tested in serum. Figure 5C shows the change in the time-gated A546 PL intensity with different numbers of equivalents of one or both of the inputs, where the latter exhibited the expected linear trend. Equation 1 did not apply with the NAND gate because the hidden toehold ensured that binding of the second input was gated by binding of the first input. Kinetic traces of the changes in Tb and A546 time-gated PL are shown in Figure 5D. Similar to the other logic gates without hidden toeholds, the NAND gate had full response within an hour. The kinetic data also showed that the time-gated Tb and A546 PL intensity increased and decreased in parallel for the (1,1) input, confirming that the “false” output signal was generated by the loss of Tb-to-A546 FRET. NOR Gate. The inverse of an OR gate, a NOR gate starts with a “true” or bright fluorescent signal, accepts two (bio)chemical inputs, and outputs a “false” or dark fluorescent signal if either one or both of the inputs are present. Figure 6A shows the design of our NOR logic gate, which was based on the same oligonucleotide sequences as the OR gate; however, the position of the Tb label was moved from the reporter to a new “antenna” sequence, such that the A546 and Tb were on different strands (i.e., the Tb label in the NOR gate occupied the same position as the IabFQ label in the OR gate). When one target was added as the (1,0) input state, it hybridized to its complementary site on the template through toehold-mediated

(1,1) input states, respectively. OR logic was achieved with 9fold contrast and a threshold of ∼0.13 for switching between the “false/true” Boolean output states. There was good retention of contrast when tested in serum with a threshold ≤0.3. Figure 4C shows the change in the time-gated A546 PL intensity with different numbers of equivalents of inputs. In this case, the trend for increasing χ1 and χ2 (with χ1 = χ2) had downward curvature, which was consistent with independent binding of the inputs. The switch in curvature (cf., upward with AND gates) was expected because of the ability of an OR gate to produce a “true” output for only one input. The response for only one input remained linear. The data was once again wellfit with eq 1 parametrized to the data in Figure 4B. Kinetic traces of the changes in Tb and A546 time-gated PL for the OR gate are shown in Figure 4D and demonstrated full response within an hour. The kinetic data also showed that the Tb and A546 time-gated PL intensities increased in parallel for the (0,1), (1,0), and (1,1) input states, confirming that the “true” output signal was generated by alleviation of the quenching of Tb by IabFQ, whether by removal of the Tb/ A546-labeled reporter strand or by removal of the IabFQlabeled blocking strand from the template oligonucleotide. NAND Gate. A NAND gate is the inverse of an AND gate. It starts with a “true” or bright fluorescent signal, accepts two (bio)chemical inputs, and switches to a “false” or dark fluorescent signal only if both of the inputs are present. Figure 5A shows the design of our NAND logic gate. The gate consisted of two 18 nt template arms joined through an internal amine linker that was labeled with Tb. The template was initially hybridized with an unlabeled 17 nt blocking sequence and an A546-labeled 15 nt reporter sequence with efficient time-gated, FRET-sensitized A546 PL. When one target oligonucleotide was added as the (1,0) input state, it hybridized to its complementary template arm through toehold-mediated strand displacement of the unlabeled 17 nt blocking sequence, and time-gated FRET-sensitized A546 PL was maintained. When only the other oligonucleotide was added as the (0,1) input state, there was no toehold for the (0,1) input because F

DOI: 10.1021/acssensors.7b00355 ACS Sens. XXXX, XXX, XXX−XXX

Article

ACS Sensors strand displacement of the Tb-labeled antenna. The result was loss of Tb-to-A546 FRET and thus a large decrease in the timegated A546 PL. When the second target was added as the (0,1) input state, it hybridized to the complementary reporter sequence, causing displacement from the template. The result was loss of Tb-to-A546 FRET and a large decrease in the timegated A546 PL intensity. When both targets were added as the (1,1) input state, both displacement mechanisms contributed to the decrease in time-gated, FRET-sensitized A546 PL intensity. Figure 6B shows the experimental truth table for the NOR gate in terms of the relative A546 PL intensity. Normalized outputs of 1.0, 0.13, 0.10, and 0.09 were measured for the (0,0), (1,0), (0,1), and (1,1) input states, respectively. Robust NOR logic was achieved with 7.4-fold contrast and a threshold of ∼0.97 for switching between “true/false” output states. Good contrast was retained in serum with a threshold ≥0.75. Figure 6C shows the change in the time-gated A546 PL intensity with different numbers of equivalents of one or both inputs. The trend for increasing χ1 and χ2 (with χ1 = χ2) was again curved, and different than the curvature for AND and OR gates. The data was also well-fit by eq 1 parametrized to the data in Figure 6B, consistent with independent binding of the inputs for the NOR gate. Kinetic traces of the changes in both Tb and A546 timegated PL are shown in Figure 6D and demonstrated that the gate fully responded to its inputs within 2 h. The kinetic data also showed that the Tb intensity increased in parallel with a decrease in the A546 time-gated PL intensity for the (1,0), (0,1), and (1,1) states, further confirming that the “false” output signal was generated by the loss of Tb-to-A546 FRET. Multiplexed Logic Gates. Given the ability to design different types of logic gates, it follows that these gates could be deployed in parallel. Such multiplexed logic would require a resolvable photonic output for each gate. Figure 7A illustrates a two-color combination of the OR and AND-I gates that responded to the same two oligonucleotide inputs, where the AND-I gate used A546 as a FRET acceptor and the OR gate used Alexa Fluor 488 (A488) as an acceptor. The design of the OR gate was unchanged from that presented earlier, except for substitution of A546 with A488. Both the A488 and A546 PL were resolved from the Tb PL, and the logic gates were able to function in concert with a “true/false” threshold of ∼0.2 for the relative dye PL intensity. Full spectra can be found in Figure S3. A second multiplexed experiment was done with a new (and unoptimized) OR gate that shared the first input with the AND-I gate, but did not share the second input, instead accepting a third input. As shown in Figure 7B, the expected response to the input sequences was observed, and the “true/ false” threshold for two-color, three-input logic remained ∼0.2 in terms of relative dye PL intensity. Discussion. An ideal molecular logic gate will emulate its truth table with high contrast between “true/false” states and minimum variation in the real output signal for inputs that generate the same Boolean output; however, many real molecular logic gates fall short of this ideal. For example, DNA-based gates with oligonucleotide inputs and signaling based on FRET between fluorescent dyes have been reported with OR function contrast ratios between 2 and 3.5, and with AND function contrast ratios between 2 and 5 but also less than 1.5.15,17,45−47 NAND and NOR gates in this format have been rarely reported, if at all. The foregoing contrast ratios are much less than the 9(±1)-fold contrast achieved with our AND and OR gates. Our gates also had ≤10% difference in the

Figure 7. Two-color multiplexed logic gates. (A) Two-input and (B) three-input combinations of AND-I and OR: (i) input-output scheme; (ii) truth table; (iii) AND-I gate with time-gated A546 PL output and OR gates with time-gated A488 PL output, where the components and toeholds are labeled as in previous figures; (iv) relative time-gated A488 and A546 PL intensity when challenged with an equivalent number of inputs (normalized to mimic truth table-like output). The approximate “false/true” threshold is the dashed line. NC refers to challenging the gates with a noncomplementary sequence.

magnitude of the output signals (relative to the maximum signal) for a common Boolean output, whereas several of the aforementioned examples had differences of 20−50%.15,17,46,47 Although substitution of quantum dots for dyes in DNA-based logic gates has led to higher contrast, these designs still had lower contrast (4−6-fold) than our AND, OR, and NOR gates, exceeding only our NAND gate.38 Many other molecular logic gate formats, both with non-DNA inputs and non-DNA-based, have also had modest 2−3-fold contrast.48−51 Our logic gates are thus novel in their use of time-gated FRET and advantageous in their scope of logic functions, uniformity in signal magnitude for a common Boolean output, and high contrast between “true/false” signals. These features will ultimately provide more robust logic and greater sensitivity in the context of bioanalysis. At first glance, it would appear that the high contrast came from the dark quencher in the AND and OR gates strongly quenching Tb emission and shutting down FRET to the fluorescent dye reporter; however, there may have also been a secondary effect. We previously elucidated a “sweet-spot” for time-gated FRET with Tb and dye acceptors, representing a G

DOI: 10.1021/acssensors.7b00355 ACS Sens. XXXX, XXX, XXX−XXX

Article

ACS Sensors

the FRET strategies and linear DNA motifs that enabled the two-input (N)AND and (N)OR gates to design XOR and XNOR gates. Reversibility was also not considered in the gate designs. Although reversibility is important for computational applications of molecular logic gates, it is not necessary for the envisioned applications in bioanalysis. Other aspects of gate design were also important, as we found it critical to retain a double-stranded structure. Gate designs that had single-stranded regions of significant length were ineffective, likely because of a previously reported tendency of the Tb and dyes to associate with one another with loss of time-gated FRET emission.43 This limitation also restricted the possible length of toeholds, which impacts strand displacement kinetics.55−57 It was therefore useful in the OR gate to introduce mismatches in addition to a toehold to increase displacement kinetics. More discussion about optimization of the OR gate can be found in the SI, and in general we find that the advantages of Tb with respect to photonic output are still dependent on the molecular dynamics of the logic gate. Future work will need to address ways to use longer toeholds for faster assay times while still retaining maximum contrast in logic function. Finally, the high-contrast photonic output and suitability of the logic gates for sample matrices such as serum raises the possibility of future applications in rapid diagnostic screening. An MLD-based computation could make an autonomous decision about the levels of multiple biomarkers and generate a “true/false” output that corresponds to the health status of an individual. This capability of integrating multiple bits of biochemical information and computing a single output could be useful for simple, rapid, and economical disease screening by nonspecialized personnel. The results could be used as a decision point for further and more rigorous biomedical testing.

compromise between a Tb−dye distance short enough for efficient FRET but not so short as to make FRET efficient enough to shorten the Tb lifetime to a degree that emission is mainly within the instrument lag time for time-gated measurements.52 The reporter dye was placed as close as possible to this sweet spot within the constraints of stable hybridization for blocking sequences and sufficiently long toeholds for actuation of the logic gates. In contrast, the quencher was in close enough proximity to the Tb to shorten its emission lifetime to within the lag time. As a consequence, FRET “leaks” to the reporter dye instead of the quencher would be unlikely to be detected in time-gated measurements. Exploiting this non-sweet-spot effect to compensate for imperfect quenching of FRET to a reporter dye may not only optimize signaling in molecular logic gates, but may also enable future designs with complexity and function that would be impractical or inaccessible when using only conventional fluorescent dyes. The lag time and integration time for time-gated measurements can also be tuned to obtain maximum contrast between “true/false” outputs, as has been demonstrated for obtaining maximum contrast in other sensing applications that use Tb as a FRET donor.53 Time-gated FRET measurements enabled by the Tb donor also made it possible to cleanly measure logic gate output in serum. Poor signal-to-background ratios are obtained in these matrices with only conventional FRET between dyes.52 The time-gated measurement mode suppressed the strong autofluorescence and scattering backgrounds of these matrices with clear resolution of the FRET-sensitized reporter dye fluorescence. Boolean contrast obtained with logic gates in serum matrices may yet tend closer to those obtained in buffer matrices with further optimization of time-gating (see Figures S4−S5). Another advantage offered by the Tb was its narrow emission lines across the visible spectrum. The spectral positions and width of these lines allowed the Tb to be an effective donor for both A546 and A488. It provided the requisite spectral overlap for FRET with both dyes and permitted measurement of both dye emission intensities between the Tb emission lines, while still only requiring a single excitation wavelength. Although we only demonstrated two-dye multiplexing here, it has been shown that Tb can be an effective FRET donor for up to five dyes in parallel while retaining the capability for quantitative bioanalysis.54 A color channel could also be used as an internal standard to establish a ratiometric “true/false” threshold that is independent of the dilution of the logic gates in a sample. Such multiplexing would not be supported by a fluorescent dye donor. While Tb offered many advantages, careful design and optimization of the oligonucleotide sequences remained critical. We utilized two general designs of logic gates: designs with open toeholds and a design with a hidden toehold. For linear oligonucleotide assemblies, the open toehold motif was limited to two inputs, whereas the hidden toehold motif was scalable to more than two inputs. Figures S8−S10 show the design and response of a preliminary three-input AND gate that utilized two hidden toeholds and had a “true/false” threshold between 0.1 and 0.2. A three-input NAND gate was designed similarly, albeit that contrast was significantly lower (“true/false” threshold ∼0.6; Figures S11−S12). The hidden toeholds were not directly amenable to designing three-input OR and NOR gates, and strategies for developing a comprehensive set of three-input logic gates with time-gated FRET will be a subject of future research. Similarly, we were unable to adapt

■

CONCLUSIONS We developed a set of DNA oligonucleotide-based AND, OR, NAND, and NOR photonic logic gates that used toeholdmediated strand displacement and time-gated FRET with a Tb donor for signaling. The two-input designs provided a remarkable 4−10-fold signal contrast between “true/false” output states, enabled in part by the long luminescence lifetime of the Tb, which enabled rejection of directly excited dye emission and background from a complex sample matrix such as serum. In addition, the narrow emission lines of the Tb facilitated multicolor implementation of two logic gates in parallel. Kinetic traces of the PL response of the logic gates provided insight into their mechanism of signaling and the effectiveness of competitive and sequential quenching pathways in creating dark “false” states. In addition to standard two-input logic gates, we also demonstrated the potential for three-input AND and NAND logic gates. Overall, the modular designs of DNA-based logic gates and the unique optical properties of Tb are a powerful combination. Future applications in bioanalysis are envisioned for these and other time-gated FRET and DNAbased photonic MLDs.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acssensors.7b00355. H

DOI: 10.1021/acssensors.7b00355 ACS Sens. XXXX, XXX, XXX−XXX

Article

ACS Sensors

■

(12) Zhang, C.; Yang, J.; Jiang, S.; Liu, Y.; Yan, H. DNAzyme-Based Logic Gate-Mediated DNA Self-Assembly. Nano Lett. 2016, 16, 736− 741. (13) Hermanson, G. T. Bioconjugate Techniques, 3rd ed.; Elsevier/ Academic Press: London; Waltham, MA, 2013; pp 1−1146. (14) Bloomfield, V. A.; Crothers, D. M.; Tinoco, I. Nucleic Acids: Structures, Properties, and Functions; University Science Books: Sausalito, CA, 2000; pp 1−794. (15) Zhang, L.; Bluhm, A. M.; Chen, K. J.; Larkey, N. E.; Burrows, S. M. Performance of Nano-Assembly Logic Gates with a DNA MultiHairpin Motif. Nanoscale 2017, 9, 1709−1720. (16) Wang, F.; Willner, B.; Willner, I. DNA-Based Machines. Top. Curr. Chem. 2014, 354, 279−338. (17) Buckhout-White, S.; Brown, C. W., III; Hastman, D. A., Jr.; Ancona, M. G.; Melinger, J. S.; Goldman, E. R.; Medintz, I. L. Expanding Molecular Logic Capabilities in DNA-Scaffolded MultiFRET Triads. RSC Adv. 2016, 6, 97587−97598. (18) Teo, Y. N.; Kool, E. T. DNA-Multichromophore Systems. Chem. Rev. 2012, 112, 4221−4245. (19) Geiβler, D.; Hildebrandt, N. Recent Developments in Förster Resonance Energy Transfer (FRET) Diagnostics Using Quantum Dots. Anal. Bioanal. Chem. 2016, 408, 4475−4483. (20) Miyamoto, T.; Razavi, S.; DeRose, R.; Inoue, T. Synthesizing Biomolecule-Based Boolean Logic Gates. ACS Synth. Biol. 2013, 2, 72−82. (21) George, A. K.; Singh, H. Enzyme-Free Scalable DNA Digital Design Techniques: A Review. IEEE Trans. Nanobioscience 2016, 15, 928−938. (22) Katz, E.; Poghossian, A.; Schoning, M. J. Enzyme-Based Logic Gates and Circuits-Analytical Applications and Interfacing With Electronics. Anal. Bioanal. Chem. 2017, 409, 81−94. (23) Andréasson, J.; Pischel, U. Molecules With a Sense of Logic: A Progress Report. Chem. Soc. Rev. 2015, 44, 1053−1069. (24) Park, K. S.; Seo, M. W.; Jung, C.; Lee, J. Y.; Park, H. G. Simple and Universal Platform for Logic Gate Operations Based on Molecular Beacon Probes. Small 2012, 8, 2203−2212. (25) Fern, J.; Scalise, D.; Cangialosi, A.; Howie, D.; Potters, L.; Schulman, R. DNA Strand-Displacement Timer Circuits. ACS Synth. Biol. 2017, 6, 190−193. (26) Wu, C.; Wan, S.; Hou, W.; Zhang, L.; Xu, J.; Cui, C.; Wang, Y.; Hu, J.; Tan, W. A Survey of Advancements in Nucleic Acid-Based Logic Gates and Computing for Applications in Biotechnology and Biomedicine. Chem. Commun. 2015, 51, 3723−3734. (27) Orbach, R.; Willner, B.; Willner, I. Catalytic Nucleic Acids (DNAzymes) as Functional Units for Logic Gates and Computing Circuits: From Basic Principles to Practical Applications. Chem. Commun. 2015, 51, 4144−4160. (28) Pu, F.; Ren, J.; Qu, X. Nucleic Acids and Smart Materials: Advanced Building Blocks for Logic Systems. Adv. Mater. 2014, 26, 5742−5757. (29) Deng, W.; Xu, H.; Ding, W.; Liang, H. DNA Logic Gate Based on Metallo-Toehold Strand Displacement. PLoS One 2014, 9, e111650. (30) Groves, B.; Chen, Y. J.; Zurla, C.; Pochekailov, S.; Kirschman, J. L.; Santangelo, P. J.; Seelig, G. Computing in Mammalian Cells with Nucleic Acid Strand Exchange. Nat. Nanotechnol. 2015, 11, 287−294. (31) Yoshida, W.; Yokobayashi, Y. Photonic Boolean Logic Gates Based on DNA Aptamers. Chem. Commun. 2007, 2, 195−197. (32) Yang, J.; Song, Z.; Liu, S.; Zhang, Q.; Zhang, C. Dynamically Arranging Gold Nanoparticles on DNA Origami for Molecular Logic Gates. ACS Appl. Mater. Interfaces 2016, 8, 22451−22456. (33) Gao, W.; Zhang, L.; Zhang, Y.-M.; Liang, R.-P.; Qiu, J.-D. DNA Colorimetric Logic Gates Based on Triple-Helix Molecular Switch. J. Phys. Chem. C 2014, 118, 14410−14417. (34) Baron, R.; Lioubashevski, O.; Katz, E.; Niazov, T.; Willner, I. Logic Gates and Elementary Computing by Enzymes. J. Phys. Chem. A 2006, 110, 8548−8553. (35) Chen, J.; Zhou, S.; Wen, J. Concatenated Logic Circuits Based on a Three-Way DNA Junction: A Keypad-Lock Security System with

Experimental details: labeling of oligonucleotides; calculation of Förster distances, competitive FRET, contrast and threshold values; statistical fitting of calibration curves; normalization of kinetic data. Results and discussion: full emission spectra for logic gates and multiplexed logic, extended calibration curves for AND gates, gate performance in serum, longer oligonucleotide inputs with AND-I, optimization of OR gate design, three-input AND and NAND gates (PDF)

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel: 1-604-822-2464. ORCID

Igor L. Medintz: 0000-0002-8902-4687 W. Russ Algar: 0000-0003-3442-7072 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge support from the Office of Naval Research (ONR). W.R.A. and M.M. acknowledge support from the Canada Foundation for Innovation (CFI), BCKDF, and the Natural Sciences and Engineering Research Council of Canada (NSERC). W.R.A also acknowledges a Canada Research Chair (Tier 2) and a Michael Smith Foundation for Health Research Scholar Award. M.G.A. and I.L.M. acknowledge the NRL NSI and LUCI grants in support of the VBFF program through the OSD. The authors thank Lumiphore, Inc. for the Lumi4-TbNHS reagent.

■

REFERENCES

(1) Tompkins, L. S. Nucleic Acid Probes in Infectious Diseases. Curr. Clin. Top. Infect. Dis. 1989, 10, 174−193. (2) Krishnan, Y.; Simmel, F. C. Nucleic Acid Based Molecular Devices. Angew. Chem., Int. Ed. 2011, 50, 3124−3156. (3) Wang, F.; Liu, X.; Willner, I. DNA Switches: From Principles to Applications. Angew. Chem., Int. Ed. 2015, 54, 1098−1129. (4) Zheng, J.; Yang, R.; Shi, M.; Wu, C.; Fang, X.; Li, Y.; Li, J.; Tan, W. Rationally Designed Molecular Beacons for Bioanalytical and Biomedical Applications. Chem. Soc. Rev. 2015, 44, 3036−3055. (5) Minero, G. A.; Fock, J.; McCaskill, J. S.; Hansen, M. F. Optomagnetic Detection of DNA Triplex Nanoswitches. Analyst 2017, 142, 582−585. (6) Ma, D. L.; Zhang, Z.; Wang, M.; Lu, L.; Zhong, H. J.; Leung, C. H. Recent Developments in G-Quadruplex Probes. Chem. Biol. 2015, 22, 812−828. (7) Zhang, H.; Zhou, L.; Zhu, Z.; Yang, C. Recent Progress in Aptamer-Based Functional Probes for Bioanalysis and Biomedicine. Chem. - Eur. J. 2016, 22, 9886−9900. (8) Schaffert, D. H.; Okholm, A. H.; Sorensen, R. S.; Nielsen, J. S.; Torring, T.; Rosen, C. B.; Kodal, A. L.; Mortensen, M. R.; Gothelf, K. V.; Kjems, J. Intracellular Delivery of a Planar DNA Origami Structure by the Transferrin-Receptor Internalization Pathway. Small 2016, 12, 2634−2640. (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) Zhang, F.; Nangreave, J.; Liu, Y.; Yan, H. Structural DNA Nanotechnology: State of the Art and Future Perspective. J. Am. Chem. Soc. 2014, 136, 11198−11211. (11) Bujold, K. E.; Hsu, J. C.; Sleiman, H. F. Optimized DNA ″Nanosuitcases″ for Encapsulation and Conditional Release of siRNA. J. Am. Chem. Soc. 2016, 138, 14030−14038. I

DOI: 10.1021/acssensors.7b00355 ACS Sens. XXXX, XXX, XXX−XXX

Article

ACS Sensors Visible Readout and an Automatic Reset Function. Angew. Chem., Int. Ed. 2014, 54, 446−450. (36) Ran, X.; Pu, F.; Ren, J.; Qu, X. DNA-Regulated Upconverting Nanoparticle Signal Transducers for Multivalued Logic Operation. Small 2014, 10, 1500−1503. (37) Claussen, J. C.; Hildebrandt, N.; Susumu, K.; Ancona, M. G.; Medintz, I. L. Complex Logic Functions Implemented with Quantum Dot Bionanophotonic Circuits. ACS Appl. Mater. Interfaces 2014, 6, 3771−3778. (38) He, X.; Li, Z.; Chen, M.; Ma, N. DNA-Programmed Dynamic Assembly of Quantum Dots for Molecular Computation. Angew. Chem., Int. Ed. 2014, 53, 14447−14450. (39) Hildebrandt, N.; Wegner, K. D.; Algar, W. R. Luminescent Terbium Complexes: Superior Förster Resonance Energy Transfer Donors for Flexible and Sensitive Multiplexed Biosensing. Coord. Chem. Rev. 2014, 273−274, 125−138. (40) Amoroso, A. J.; Pope, S. J. A. Using Lanthanide Ions in Molecular Bioimaging. Chem. Soc. Rev. 2015, 44, 4723−4742. (41) Bünzli, J. C. Lanthanide Luminescence for Biomedical Analyses and Imaging. Chem. Rev. 2010, 110, 2729−2755. (42) Xu, J.; Corneillie, T. M.; Moore, E. G.; Law, G. L.; Butlin, N. G.; Raymond, K. N. Octadentate Cages of Tb(III) 2-hydroxyisophthalamides: A New Standard for Luminescent Lanthanide Labels. J. Am. Chem. Soc. 2011, 133, 19900−19910. (43) Massey, M.; Ancona, M. G.; Medintz, I. L.; Algar, W. R. TimeResolved Nucleic Acid Hybridization Beacons Utilizing Unimolecular and Toehold-Mediated Strand Displacement Designs. Anal. Chem. 2015, 87, 11923−11931. (44) Seelig, G.; Soloveichik, D.; Zhang, D. Y.; Winfree, E. EnzymeFree Nucleic Acid Logic Circuits. Science 2006, 314, 1585−1588. (45) He, K.; Li, Y.; Xiang, B.; Zhao, P.; Hu, Y.; Huang, Y.; Li, W.; Nie, Z.; Yao, S. A Universal Platform for Building Molecular Logic Circuits Based on a Reconfigurable Three-Dimensional DNA Nanostructure. Chem. Sci. 2015, 6, 3556−3564. (46) Li, X. Y.; Huang, J.; Jiang, H. X.; Du, Y. C.; Han, G. M.; Kong, D. M. Molecular Logic Gates Based on DNA Tweezers Responsive to Multiplex Restriction Endonucleases. RSC Adv. 2016, 6, 38315− 38320. (47) Cannon, B. L.; Kellis, D. L.; Davis, P. H.; Lee, J.; Kuang, W.; Hughes, W. L.; Graugnard, E.; Yurke, B.; Knowlton, W. B. Excitonic and Logic Gates on DNA Brick Nanobreadboards. ACS Photonics 2015, 2, 398−404. (48) Claussen, J. C.; Algar, W. R.; Hildebrandt, N.; Susumu, K.; Ancona, M.; Medintz, I. L. Biophotonic Logic Devices Based on Quantum Dots and Temporally-Staggered Förster Energy Transfer Relays. Nanoscale 2013, 5, 12156−12170. (49) Zhang, S.; Wang, K.; Huang, C.; Li, Z.; Sun, T.; Han, D.-M. An Enzyme-Free and Resettable Platform for the Construction of Advanced Molecular Logic Devices Based on Magnetic Beads and DNA. Nanoscale 2016, 8, 15681−15688. (50) Lee, C.-C.; Liao, Y.-C.; Lai, Y.-H.; Chuang, M.-C. Recognition of Dual Targets by a Molecular Beacon-Based Sensor: Subtyping of Influenza A Virus. Anal. Chem. 2015, 87, 5410−5416. (51) Yang, J.; Dong, C.; Dong, Y.; Liu, S.; Pan, L.; Zhang, C. Logic Nanoparticle Beacon Triggered by the Binding-Induced Effect of Multiple Inputs. ACS Appl. Mater. Interfaces 2014, 6, 14486−14492. (52) Massey, M.; Ancona, M. G.; Medintz, I. L.; Algar, W. R. TimeGated DNA Photonic Wires with Förster Resonance Energy Transfer Cascades Initiated by a Luminescent Terbium Donor. ACS Photonics 2015, 2, 639−652. (53) Scholler, P.; Moreno-Delgado, D.; Lecat-Guillet, N.; Doumazane, E.; Monnier, C.; Charrier-Savournin, F.; Fabre, L.; Chouvet, C.; Soldevila, S.; Lamarque, L.; et al. HTS-Compatible FRET-Based Conformational Sensors Clarify Membrane Receptor Activation. Nat. Chem. Biol. 2017, 13, 372−380. (54) Geiβler, D.; Stufler, S.; Löhmannsröben, H. G.; Hildebrandt, N. Six-Color Time-Resolved Förster Resonance Energy Transfer for Ultrasensitive Multiplexed Biosensing. J. Am. Chem. Soc. 2013, 135, 1102−1109.

(55) Yurke, B.; Mills, A. P. Using DNA to Power Nanostructures. Genet. Prog. Evol. Mach. 2003, 4, 111−122. (56) Zhang, D. Y.; Winfree, E. Control of DNA Strand Displacement Kinetics Using Toehold Exchange. J. Am. Chem. Soc. 2009, 131, 17303−17314. (57) Srinivas, N.; Ouldridge, T. E.; Sulc, P.; Schaeffer, J. M.; Yurke, B.; Louis, A. A.; Doye, J. P.; Winfree, E. On the Biophysics and Kinetics of Toehold-Mediated DNA Strand Displacement. Nucleic Acids Res. 2013, 41, 10641−10658.

J

DOI: 10.1021/acssensors.7b00355 ACS Sens. XXXX, XXX, XXX−XXX