Logic devices based on nucleic acid self‐assembly

Abstract

Nucleic acids are natural macromolecules with the ability to store and transmit information based on the strict base‐pairing principle. Beyond the natural nucleic acid double helixes, various DNA/RNA nanostructures with customized geometries and functionalities have been fabricated. Featured with programmability and sequence‐dependent responsiveness, DNA/RNA nanostructures have been employed for the rational design and construction of logic devices. When stimulated by internal molecular triggers and/or external stimuli, these logic gate devices can operate at nanoscale level in complex biological environments, performing logic operations and producing corresponding outputs. In this minireview, we summarize the recent advances of nucleic acid logic devices, which are responsive to various stimuli, including DNA/RNA strands, metal ions, small molecules, peptides, proteins, photo‐irradiation, pH changes, and so forth. The applications of these devices in biosensing and biofunction regulation are also included. In the last part of the present study, we discuss the remaining challenges and perspectives of nucleic acid logic devices.

INTRODUCTION

Although silicon-based computing systems have been applied to every aspect of our life because of their outstanding computing power and incredible adaptability,^1–4 conventional silicon chips will be soon reaching their physical limits according to Moore's Law.^5–7 Alternatively, researchers have been exploring the alternative ways for information storage and computing, allowing for the continuous rise in computational power in different application scenarios.

Nanotechnology has offered exciting opportunities in the construction of multiscale logic gate networks from single molecules to customized nanostructures that can process information and realize predetermined functions.^8–10 With the rapid development of DNA and RNA nanotechnology, nucleic acid strands and assemblies have provided intriguing strategies for the design and construction of logic devices. For nucleic acid molecules, only approximately 50 atoms (one base pair) are needed to store one 1-bit information, which exhibits outstanding capability potential for miniaturizing information storage devices.^11,12 DNA and RNA are commonly recognized as natural biomolecules that are essential for the information storage and transfer of almost all living creatures. The discovery of the central dogma revealed that DNA molecules can store vast genetic information by the permutations of only four simple bases (adenine [A], thymine [T], cytosine [C], and guanine [G]). RNA molecules usually play important roles in the transcription and translation processes based on the strict base-pairing principle between nucleic acid molecules. Alternatively, in nucleic acid nanotechnology, both DNA and RNA molecules are taken as basic building blocks to construct various two-dimensional (2D) and three-dimensional (3D) nanostructures. Based on the base-pairing principle, a variety of nucleic acid nanostructure constructing methods have already been reported, for example, tile-based assembly^13–15 and DNA origami.^16–18 Due to the excellent structural programmability and addressability, nucleic acid nanoarchitectures have shown great application potential in the fields of nanophotonics,^19,20 nanoelectronics,^21,22 and biomedicine,^23–27 as well as logic computing and information process.

Programmable DNA/RNA nanotechnology provides a powerful tool to construct complex nucleic acid-based computing systems. Through the design of nucleic acid sequences and site-specific chemical modifications, these logic devices can respond to various inputs such as metal ions, biological molecules, protons, photo-irradiation, and so forth. Once recognizing the inputs, these DNA/RNA devices can be triggered for sequence-specific conformation changes, exhibiting tunable responses that eventually enable logic gating. After the performance of logic operations, these DNA/RNA devices can produce corresponding outputs. The outputs also present in different forms, which are usually chemical bond cleavage or formation, structural reconfiguration, and subsequent changes in spectra.^28,29 In addition, the nature of biomacromolecules enables the remarkable biocompatibility of DNA/RNA logic devices,^30–32 which allows their application in biological environments. In this minireview, we summarize the recent developments in nucleic acid-based logic devices with different initial inputs, and demonstrate their application potential in biosensing and biofunction regulation. Finally, the remaining challenges and perspectives of nucleic acid logic devices are discussed.

NUCLEIC ACID LOGIC SYSTEMS WITH DIFFERENT INPUT SIGNALS

The first proof-of-concept of nucleic acid-based logic computing system was reported by Adleman in 1994.³³ By encoding the graphs of the Hamiltonian path in DNA molecules, Adleman performed the computation with DNA ligation and polymerase chain reaction and solved the directed Hamiltonian path problem in a relatively fast and energy-efficient way. From then on, various DNA/RNA-based logic systems have been developed for logic algorithm and signal process.^34–36 Some of them have been further applied in disease diagnosis and neural network mimics.^37–39 Here, we classify these nucleic acid-based logic gates into following prototypes according to different signal inputs: (1) operated by nucleic acid strands, (2) activated by cations and biomolecules, (3) triggered by photo-irradiation, and (4) manipulated by multiple stimuli.

Nucleic acid strands triggered logic gates

The rigid base complementary between nucleic acid strands has been widely used to trigger the algorithm in nucleic acid-based logic systems. Either DNA or RNA strands can serve as inputs to initiate the downstream logic computing via strand displacement reactions. The strand displacement reaction is an entropy-driven process where the hybridization of the “invader” strand and the dehybridization of the “incumbent” strand with the “substrate” strand take place simultaneously.^40,41 The application of strand displacement reaction triggered nanodevice was pioneered by Yurke and colleagues. They fabricated a pair of nanotweezers with three DNA strands. The nanotweezers were closed and opened reversibly by the addition of “fuel” DNA strand, which could undergo multiple strand displacement cycles through the “toehold” domains.⁴² Since then, strand displacement-induced logic systems with much more complexity have been successively reported.⁴³

In 2007, Frezza et al. developed a complete set of solid-supported two-input DNA logic gates (AND, OR, AND-NOT), which could be integrated into a three-level circuit (Figure 1A).⁴⁴ These DNA logic gates were immobilized on streptavidin-sepharose beads via a 5′-biotin modification, while the solution phase served as a communicating medium between gates. The appropriate Boolean operation was induced by the addition of single-stranded DNA (ssDNA) molecules, and the release of ssDNA outputs was monitored by detecting the Cy3 fluorescence labeled on output strands. These isolated logic gates were further connected in series to build a three-level circuit that possessed XOR Boolean behavior, which takes a step forward to the goal of reproducing the tenets of digital logic. In 2014, Wang et al. designed DNA origami-based logic gates for microRNA diagnostics.⁴⁵ The rectangular DNA origami architecture is entirely addressable, allowing for the localization of functional groups with predesigned numbers and patterns. In their work, each rationally designed DNA origami had a computation module and an output module. MicroRNA-21 and microRNA-195, two indicators of heart failure, were selected as the logic inputs. After the correct inputs were received, the biotinylated signal DNA molecules were released from the computation module and hybridized with complementary DNA (cDNA) sequences in the output module. Streptavidin was then added to visualize the output results by atomic force microscope. The DNA origami logic device presents great potential in the logic analysis of the biological microenvironment and the activation of drugs when disease indicators are spotted.

FIGURE 1. Nucleic acid-based logic devices with nucleic acid strands as input signals. (A) Immobilized DNA-based logic gates with two DNA single strands as inputs for constructing modular multilevel circuits. Reproduced with permission: Copyright 2007, American Chemical Society.44 (B) A cargo-sorting DNA robot, which could sort different cargoes into distinct piles, constructed by a simple algorithm on a two-dimensional (2D) DNA origami. Reproduced with permission: Copyright 2017, American Association for the Advancement of Science.47 (C) A multiple-aptamer-based DNA logic device for accurate cancer cell identification via associative toehold activation on live cell membranes. Reproduced with permission: Copyright 2019, American Chemical Society.52 (D) DNAzyme-based logic gates for directing the self-assembly of DNA tiles onto prescribed DNA origami frames. Reproduced with permission: Copyright 2016, American Chemical Society58

In 2017, Chatterjee et al. constructed circuit architectures with spatially arranged DNA hairpins on DNA origami for fast and modular DNA computing.⁴⁶ In the process of signal propagation, the upstream hairpins were unwound to release the toeholds locked in them, initiating the unwinding of subsequent hairpins. The quencher-labeled strands in the double-stranded reporters were then displaced, and the fluorescence signal was restored. Signal propagation in different lengths and orientations was demonstrated, and multi-input logic circuits were obtained through the combination of elementary logic gates (two-input OR and two-input AND gates). Compared with previous systems with diffusible components, this approach apparently reduced the computation time from hours to minutes without reducing precision, providing a path for enhancing the specificity of theranostic DNA nanodevices. In 2017, Thubagere et al. developed a DNA origami-based robot for performing cargo sorting at the molecular level (Figure 1B).⁴⁷ An ssDNA device was constructed on a 2D DNA origami surface using a simple robot design: one leg-two foot domains for walking and one arm-one hand domains for picking up or dropping off cargoes. The picking-up of cargoes was a result of the random wandering of the robot on the origami surface while the recognition of goals instructed the robot to drop matched cargoes off, by which means the nanorobot could realize the sorting of different cargoes into distinct piles. Different from other robotic systems that display single function within limited steps, this DNA robot realized more sophisticated tasks and performed more steps in logic calculation, making it possible to build molecular robots that can be easily programmed like macroscopic robots but work in microscopic environments.

To avoid the undesired release in the absence of input, Wang et al. developed a series of leakless designs and successfully reduced “toeless” strand displacement, even with a concentration of 100 times larger than previous work.⁴⁸ These leakless designs were compatible with the clamping technique, which introduced an energy barrier for toeless displacement. The initial species in leakless designs need to bind together to generate leaked signals, which would incur a thermodynamic penalty, making the leaking process unfavorable. In addition, a four-layer linear translator cascade with nine strand displacement steps and a three-layer OR circuit consisting of 21 multistranded complexes were demonstrated, showing the applicability of designed leak reduction in more complex chemical systems. Later in 2019, Woods et al. designed a set of ssDNA tiles that could be reprogrammed to implement a wide variety of six-bit algorithms.⁴⁹ This set of tiles was then used to construct 21 circuits that execute a series of algorithms (e.g., copying, sorting, recognizing palindromes and multiples of 3, random walking, obtaining an unbiased choice from a biased random source, electing a leader, simulating cellular automata, generating deterministic and randomized patterns, and counting to 63). The logic device networks suggested that molecular self-assembly could be a reliable algorithmic component within programmable chemical systems and molecular programming. He et al. presented an artificial DNA signaling network that was capable of translating nucleic strand input into easy-to-read temperature output.⁵⁰ The signaling network consisted of two major parts: the signal amplifier and the signal transducer. After the addition of the target strand, the signal amplifier was initiated and released a bunch of strand R by the strand displacement reaction. Meanwhile, the released strand R further liberated the pre-blocked DNA G4 aptamer sequences in the signal transducer through strand displacing. DNA G4 aptamer–hemin complexes were then formed to catalyze the oxidation of 3,3′,5,5′-tetrazmethylbenzidine sulfate (TMB) and subsequent temperature output upon irradiation. Benefiting from the programmability of DNA, other functional nucleic acids, such as aptamers and DNAzymes, could be applied as signal amplifiers, showing the great potential of DNA signaling networks in biomedical research and point-of-care clinical diagnostics.

Strand displacement reaction can be used to build RNA logic gates as well. Oesinghaus et al. constructed guide RNA (gRNA) logic nanodevices using extension strategies for switching the activity of Cas12a through strand displacement reaction.⁵¹ Strand displacement gRNAs (SD gRNAs) were induced by 5′ extensions. An SD gRNA consists of four building blocks: gRNA domain, separator domain, switch domain, and toehold domain. The gRNA handle was first disrupted so that Cas12a binding was suppressed. When RNA triggers were given as input, the secondary structure of the Cas12a handle was restored via toehold-mediated strand invasion into the switch domain, and the activity of Cas12a was regained. It is noteworthy that the SD gRNAs also demonstrated the ability to function inside bacterial cells.

In addition, nucleic acid-induced logic gates can be built by activating toehold-based reactions for the initiation of hybridization chain reaction (HCR). The addition of one connector strand can link two or more target strands together and form an associated toehold domain. This toehold domain is then connected with a readout strand and triggers the HCR reaction to amplify the output signal. Taking advantage of HCR, Chang et al. designed a logic device that can perform AND Boolean logic analysis to develop a single-step cell identification, realizing the accurate identification and isolation of cells (Figure 1C).⁵² This device can be utilized to analyze the multiple biomarkers on the cell surface and acutely label the target cell subtypes from large populations of similar cells, which show great potential in biomedical engineering and personalized medicine.

DNAzymes are catalytically active ssDNA with the ability to cleave specific nucleic acid strands.^53,54 Based on this property, many nucleic acid-triggered logic gates were developed. Elbaz et al. constructed a set of computing circuits and further fabricated a universal set of logic gates and a half-adder/half-subtractor system.⁵⁵ The basic elements of these computing circuits were DNAzyme subunits and fluorophore–quencher pairs labeled substrates. In the presence of input strands, active DNAzyme structures are formed by DNAzyme subunits and their substrates, which result in the cleavage of the substrates and the recovery of the fluorescence. Multilayered gate cascades, fan-out gates, and parallel logic gate operations were demonstrated in their work. They also proved that the input markers-responsive system could regulate the expression of antisense molecules and aptamers, which act as inhibitors for enzymes. Based on the same triggering principle, Orbach et al. developed a series of logically reversible Toffoli and Fredkin gates based on Mg²⁺-dependent DNAzymes.⁵⁶ The design of these reversible DNA logic gates effectively reduced energy dissipation associated with logic operations in dense computing circuits. In comparison to the polymerase-stimulated reversible Fredkin gate, which blocks the effective cascade of logic gates, DNAzyme subunit-based approach showed better performance in constructing reversible logic gates.⁵⁷ Although the Toffoli and Fredkin logic reversible gates still exhibit thermodynamic irreversibility, these logic gates have broad application prospects in the nanomedicine field. Zhang et al. constructed a series of DNAzyme-based logic gates to direct the self-assembling behavior of DNA tiles onto a prescribed DNA origami frame (Figure 1D).⁵⁸ An “YES” gate consisted of double-crossover (DX) tiles and an origami K1 consisted of two holes. Single-stranded sticky ends located at the inner edges of the holes bound to the sticky ends on the two ends of the tiles, forming a complete single-stranded DNAzyme. The substrate strand modified with a pair of carboxyfluorescein–black hole quencher (FAM-BHQ) was then cleaved and the fluorescent signal was recovered. The “OR”, “logic switch”, and “AND” gates were obtained from the improvement of the “YES” gate by introducing DX tiles with different sticky ends and ssDNA protectors, which functioned as locks for the tiles. The study proved that DNAzyme-mediated DNA self-assemblies can be used for the construction of hierarchical logic gates and controllable information propagation.

Cations and biomolecules triggered logic gates

DNA can not only form a duplex structure by complementary base pairing but also interact with cations, small biological molecules, or proteins.^53,59–61 The interactions between cations and nucleic acid strands undergo different mechanisms such as mismatch base pairing by metal ions,^61–63 ion-mediated catalytic deoxyribozyme,^29,56,64 and tetrad structure formation (G-quadruplexes and i-motifs).^65–67 Various ion-mediated logic gates can be designed with these principles. For example, Hg²⁺ ions can bridge T bases, and Ag⁺ ions can specifically bridge C bases.^60,68 In 2009, Freeman et al. designed a set of T-rich-ssDNA-modified quantum dots (QDs) or C-rich-ssDNA-modified QDs for the multiplexed optical analysis of Hg²⁺ and Ag⁺ ions (Figure 2A).⁶⁹ The T-rich or C-rich DNA strands could catch Hg²⁺ or Ag⁺ ions and form rigid hairpin structures, respectively. Excited QDs then transferred their energy to the Hg²⁺–T or Ag⁺–C complexes, resulting in the quenching of QDs. By modifying different nucleic acid sequences onto the surface of two kinds of QDs, AND and OR logic gate operations were achieved, which have great prospects for the sensitive analysis of metal ion pollutants in water and food resources.

FIGURE 2. Nucleic acid-based logic devices driven by cations, biological small molecules, and proteins. (A) Multiplexed analysis of Hg2+ or Ag+ ions by nucleic acid functionalized quantum dot-based logic gates via the ion-mediated base bridging. Reproduced with permission: Copyright 2009, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.69 (B) Colorimetric logic gates (AND and OR) based on the incorporation of aptamer-cross-linked hydrogels using the color change of BSA-modified gold nanoparticle solution as output signals to help visualize the solution/gel transition. Reproduced with permission: Copyright 1996, Royal Society of Chemistry.74 (C) A three-dimensional (3D) DNA-logic gate triangular prism nanomachine for bispecific recognition and computing on target cell surfaces. Reproduced with permission: Copyright 2018, American Chemical Society.89 BSA, bovine serum albumin

Elbaz et al. designed a DNA tweezer that consisted of C-rich arms,⁷⁰ which could be manipulated by reversible pH changes to achieve switching cycles. At acidic pH (~5.2), the DNA arms were coiled into i-motif structures, followed by the release of cross-linking DNA strands and the open of the tweezer. At neutral pH (~7.2), the coiled arms stretched out and captured cross-linking strand, resulting in the closure of the tweezer. In addition, a two-tweezer device was further constructed to realize the concurrent activation of two DNA strands by pH stimuli, which functioned as a logic “SET-RESET” system. Using DNAzymes to test input ions is another example. In 2013, Zhang et al. constructed a series of colorimetric logic gates (OR, AND, INHIBIT) based on Pb²⁺- and Cu²⁺-dependent DNAzymes.⁶³ The horseradish peroxidase (HRP)-mimicking substrates were initially hybridized with ion-dependent DNAzymes. In the presence of Pb²⁺ and Cu²⁺, the hemin/G-quadruplex–HRP-mimicking DNAzymes were formed by the cleavage products of substrates. The colored products were then generated by the catalyzed oxidation of TMB as output signals. A three-input AND logic gate was subsequently fabricated by utilizing the unique interactions between Hg²⁺ ions and the T–T pairing. Compared with logic gates based on gel electrophoresis or fluorescence detection, this approach holds advantages in low cost and high efficiency.

Nucleic acid-based logic gates can be activated by small biological molecules as well. Aptamers are single-stranded oligonucleotides that can be folded into unique tertiary structures with specific molecular recognition ability. Owing to the specific binding of biological molecules, aptamers have been incorporated into logic gate circuitry for applications such as biosensing, chemiluminescence detection, and bioimaging.^59,71–73 Yin et al., for example, developed a system of colorimetric logic gates (AND and OR) based on aptamer-crosslinked hydrogels (Figure 2B).⁷⁴ DNA-modified linear polyacrylamide polymers were used to build functional hydrogels by cross-linking between complementary strands. DNA strands contained the aptamer sequences of ATP, and cocaine were employed to recognize the targets and result in the dissociation of the hydrogel. Bovine serum albumin (BSA)-modified gold nanoparticles (BSA-GNPs) were employed as output signals to help visualize the solution/gel transition. This logic system can be utilized for the visualization of the triggered release of trapped cargoes like the BSA-GNPs, which demonstrated the potential of the aptamer switchable hydrogel in the field of biosensing and construction of nanomechanical devices and drug delivery vehicles. Conventional electrochemical sensors based on the specific ion carrier in a polymeric membrane cannot be applied for potentiometric sensing of multiple analytes.^75–77 To solve this problem, Liu et al. constructed a logical potentiometric aptasensing platform based on a G-quadruplex/hemin DNAzyme for detection of dual analytes.⁷⁸ The probe DNA, a two-aptamer-unit-incorporated oligonucleotide, was initially assembled with a signal reporter (a DNAzyme-sequence-containing oligonucleotide) onto magnetic beads. The “OR” and “INHIBIT” logic functions were manipulated by adding kanamycin and oxytetracycline as inputs. The chronopotentiometric response was mediated by signal-reporter-formed DNAzyme as output. With high sensitivity, promising potentiometric output, and expected multilevel logic operation, this versatile potentiometric aptasensing platform can be applied to detect a variety of analytes for bioanalysis and environmental monitoring.

Protein detection is one of the most important procedures for monitoring biological systems.^79–81 Based on the fact that anti-thrombin aptamer TA-15 can bind to the fibrinogen exosite with a strong inhibitory effect on blood coagulation,^82,83 Han et al. constructed a DNA circuit that can realize autonomous and programmable manipulation of thrombin function.⁸⁴ Each circuit contained an input convertor, a threshold, and an inhibitor generator. In input convertor, input thrombin reacted with aptamer TA-29 in duplex aptamer–input (A–I) and released the complementary ssDNA (named DNA-input) for downstream cascade reactions. DNA-input then continued to the threshold controller and hybridized with an exposed toehold on duplex threshold. When the amount of DNA-input was excessive, the excess DNA-input would enter the inhibitor generator, resulting in the strand displacement-mediated release of inhibitor TA-15, which would bind to thrombin and finally inhibit coagulation. This molecular logic circuit is the first example to manipulate protein activities based on direct nucleic acid–protein interactions. Disease-associated biomarkers on the cell surfaces have already been widely used for accurate diagnosis and therapy.^85–87

In 2014, Tan's group designed a DNA-based device called “Nano-Claw” that could perform autonomous logic-based analysis of multiple cancer cell surface markers and induce a therapeutic effect.⁸⁸ This claw had several structure-switchable aptamers as “Capture Toes” and one logic-gated DNA duplex as the “Effector Toe”. When Capture Toes recognized cancer cell surface markers, cDNAs were released from aptamer/cDNA duplexes and bound to the Effector Toe by toehold-mediated strand displacement reactions, resulting in a diagnostic signal and/or targeted photodynamic therapy. In their subsequent work, an advanced 3D DNA-logic gate triangular prism nanomachine has been constructed (Figure 2C),⁸⁹ which has better molecular targeting ability and internalization.^90–92 In 2015, the same group reported another DNA-based device to realize multicellular marker-based cancer analysis.⁹³ This logic device consisted of two operationally connected components, a short-oligonucleotide-tagged aptamer probe and a reporter probe (a dye or drug-labeled ssDNA or dsDNA). The tags on short-oligonucleotide-marked aptamer probes are first specifically bound to target cells. Then, the reporter probes recognized the tags and anchored on the cell surface through toehold-mediated displacement reaction. Compared with conventional bireceptor-targeting methods, this approach successfully executed programmable and multilayer analysis of various cell surface markers, providing a nucleic acid logic platform for precise disease diagnosis and effective therapy.

Amir et al. produced a DNA origami-based nanorobot that had the capability to dynamically interact with each other in a living animal.⁹⁴ Various logic gates (AND, OR, XOR, NAND, NOT, CNOT, and a half-adder) were created. These logic gates could generate logical outputs by recognizing protein cues in living cells and interacting with each other, and switch molecular payloads on or off. The in vivo experiments were also successfully carried out by using DNA origami robots in living cockroaches (Blaberus discoidalis) to control a molecule that targets their cells, demonstrating its potential as a diagnostic tool in living systems.

Photo-irradiation triggered logic gates

Photo-responsive molecules incorporated nucleic acid nanodevices that have gained increasing attention.^95–97 With light as an external stimulus, such systems can effectively avoid the introduction of chemical inputs, which could eliminate pollution and retain the efficiency and robustness of the systems after several switching cycles.

Prokup et al. reported a photochemically controlled AND gate with different light wavelengths (I₁ = 365 nm and I₂ = 532 nm) as signal inputs (Figure 3A).⁹⁸ In their study, 6-nitropiperonyloxy methylene-caged thymidine nucleotides were released by I₁ irradiation, exposing uncaged nucleic acid pairs with complementary regions. The released thymidine nucleotides then started two steps of toehold-mediated strand displacement, leading to the separation of quencher and fluorophore strands. Upon I₂ irradiation, fluorophores were excited and emission was observed as the output signal. The photochemically controlled logic gate not only realized both spatial control and temporal control but also paved the way to reduce the gap between DNA computation and silicon-based electrical circuitry. In 2018, Haydell et al. fabricated a photo-switchable logic catalytic system for complex DNA computing operations with two distinct photo-switches as execution modules.⁹⁹ Two distinct photo-responsive molecules (2′,6′-dimethylazobenzene [DM-Azo] and N-methyl-arylazopyrazole [AAP]) were incorporated into a split HRP-mimicking DNAzyme, respectively. Upon irradiation at 365 or 350 nm, the system was switched off as the formation of functional HRP-DNAzymes was prevented by cis-conformations of DM-Azo and AAP. Upon irradiation at 590 or 450 nm, DM-Azo and AAP transformed to trans-conformations, which recovered the activity of HRP-DNAzymes and switched the system on. Potentially, this strategy showed more possibilities for developing functional DNA architectures, which could be used in the field of biosensors and DNA nanomachines. In 2020, Xing et al. designed a DNA tile self-assembly system mediated by light-controlled, toehold-mediated DNA circuits (Figure 3B).¹⁰⁰ In the upstream light-controlled circuit, the initial structures were photocleavable nitrobenzyl linker modified DNA hairpins (Pc-Hairpin). After irradiation by UV light, the Pc-Hairpin groups were cleaved and “hidden toehold” sticky ends were exposed. The invading strands then interacted with the “hidden toeholds”, starting a series of strand displacement reactions to release the activator strands. The activator strands are further bound to the downstream incomplete DNA tile precursors to form the activated DNA tiles and assembled into nanotubes. This light-controlled DNA logic circuit shows the possibility of using a wide range of molecules as input to control the simultaneous assembly of various supramolecular structures in a highly programmable manner.

FIGURE 3. Nucleic acid-based logic devices triggered by photo-irradiation. (A) A photochemically controlled AND gate with different light (I1 = 365 nm and I2 = 532 nm) as input signals. Reproduced with permission: Copyright 2012, American Chemical Society.98 (B) A light-controlled DNA tile self-assembly system. Reproduced with permission: Copyright 2020, American Chemical Society100

Multiple stimuli triggered logic gates

Many multistimuli-triggered nucleic acid logic gates have been developed in recent decades. It is worth mentioning that many of these logic gates can be applied for multiparameter biosensing, regulating, and even manipulating biological systems due to the good biocompatibility and responsive ability to multiple biological signals.^101,102

In 2012, Willner and coworkers reported pH-programmable DNA logic arrays for mimicking sophisticated biomachineries. The DNA nanoarrays were constructed by Mg²⁺- and ${\mathrm{UO}}_2^{2+}$-dependent DNAzymes/substrates libraries. Both nucleic acids and pH changes were utilized to manipulate molecular circuits.¹⁰³ According to the distinct activities of the Mg²⁺- and the ${\mathrm{UO}}_2^{2+}$-dependent DNAzymes under acidic, neutral, and slightly acidic conditions, three logic circuits with different complexities were designed. Such systems could be carried out to realize the pH-controlled regulation of cellular functions or biotransformations stimulated by bacteria. Fan's group designed a set of reconfigurable DNA tetrahedra as intracellular logic sensors.¹⁰⁴ Each tetrahedron had one or two nick sites, each of which was positioned in the middle of one edge strand. A Forster resonance energy transfer pair (a rhodamine green fluorophore and a DABCYL quencher) was attached on either side of the nick site. With the introduction of dynamic sequences (i-motif, anti-ATP aptamer, T-rich mercury-specific oligonucleotide, and hairpin structures) into the complementary strand opposite the nick sites, the configuration of the tetrahedron was regulated by specific inputs (protons, ATP, and mercury ions) and the fluorescence signals were observed as the output signal. AND, OR, XOR, INH logic gates, and a half-adder operation were obtained by the rational combination of different dynamic sequences in one tetrahedron. The application prospect of DNA logic tetrahedra for intracellular detection was further demonstrated by detecting ATP in living cells. Later in 2018, the same group constructed a smart plasmonic nanobiosensor for the detection of miR-21 at the single-molecule level.²⁸ The nanobiosensor was composed of tetrahedron-structured DNA (tsDNA) modified Au@Ag core–shell nanocubes immobilized on the surface of an indium-tin oxide) glass slide. Each tsDNA had a miR-21 recognition sequence and the cleavage sites of KpnI and StuI on its three vertical edges. The localized surface plasmon resonance scattering spectra shift and the color changes between green and yellow in dark-field microscopy images were selected as output signals. The biosensor was demonstrated to not only detect a single miR-21 hybridization event but also perform OR and XOR logic operations by the addition of endonucleases KpnI and StuI. This study offered a unique method for single-molecule detection of biomolecules. Wang et al. reported DNA-logic-circuit-controlled framework nucleic acid (FNA) nanocarriers excited by endogenous cellular components (Figure 4A).¹⁰⁵ The logic-controlled FNA device consisted of a truncated square pyramid cage carrying an ATP aptamer (ABA27) and i-motif in its two opposite edges and a built-in duplex cargo, which contained an antisense oligonucleotide for TK1 mRNA sensing. The FNA logic devices first responded to endogenous ATP or/and H⁺ to release at least one side of the cargo. The free side of the cargo then hybridized with TK1 mRNA and triggered the release of Cy5-labeled strands for fluorescent mRNA imaging. As a number of disease-related microRNAs are located in the cytoplasm, these environment-responsive DNA circuits could be employed in cellular target detection and therapy.

FIGURE 4. Nucleic acid-based logic devices induced by multiple stimuli. (A) Environment-recognizing logic framework nucleic acid (FNA) nanocarriers for intracellular transport and mRNA imaging. Reproduced with permission: Copyright 2020, WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.105 (B) DNAzyme-based logic gates triggered by Pb2+ and histidine to control the cleavage of dimer and trimer origami tiles. Reproduced with permission: Copyright 2016, American Chemical Society.106 (C) Tweezer-like logic plasmonic nanodevices for multistimuli sensing. Reproduced with permission: Copyright 2020, Chinese Chemical Society107

Multistimuli-mediated nucleic acid logic gates can be used to study “origami chemistry” where origami tiles acted as atoms to assemble into origami-based “molecules” that could undergo programmed reactions. Wu et al. constructed DNAzyme-based logic gates to control the cleavage of the origami dimer and trimer (Figure 4B).¹⁰⁶ Pb²⁺-dependent or histidine-dependent DNAzyme sequences/substrates were used to connect origami tiles and yield dimers T1–T2 and T3–T4, respectively. In the presence of Pb²⁺ or histidine, origami dimers were cleaved to monomer tiles. Similarly, the logic tiles could also be bridged by the Pb²⁺ ion-dependent DNAzyme sequences and the histidine-dependent DNAzyme sequences and their substrates to form T1–T5–T4 trimers and to yield an “AND” logic gate. The “origami chemistry” provided a novel strategy to control the geometry and functions of biomolecule assemblies and held great potential in the regulation of biophysical processes in living systems. In 2020, Ding and his coworkers reported a series of novel dynamic plasmonic nanodevices to realize logic-gated computing (Figure 4C).¹⁰⁷ Two gold nanorods and computing elements were co-assembled on a tweezer-like DNA origami to create chiral plasmonic nanoarchitectures. The input molecules (DNA strands, glutathione, and adenosine) were used to manipulate the morphology and circular dichroism signal of nanodevices. Through the redesign and multifunctionalization of the device, a series of Boolean logic-gated computing (YES, NOT, AND, OR) and three-input complex logic computing were realized. The intelligent DNA nanodevices are promising for bioanalysis and molecular information processing.

SUMMARY AND OUTLOOK

In this minireview, the recent advances of nucleic acid-based logic systems have been summarized. Featured with sequence-specific recognition and ease of chemical modification, nucleic acid molecules provide a promising candidate for the construction of logic devices. After recognizing the appropriate nucleic acid sequences or other chemical/physical signals as inputs, the DNA/RNA logic devices can operate and perform computing at nanoscale, subsequently generating corresponded outputs, which are usually from the structural reconfiguration and the changes of spectra. Furthermore, DNA/RNA molecules have remarkable biocompatibility as genomic materials and their sequence-dependent responsiveness, providing an intriguing tool for applications in complicated biological environment. DNA/RNA logic devices with multiple ligands could be constructed to accurately and precisely recognize or capture the target malignant cells expressing specific markers. Taking advantage of the logic-gated strategy, the cell discrimination properties of DNA/RNA drug carriers can be enhanced and the off-target effects could be reduced.^24,108 In the future, biological circuitries constructed by nucleic acids could even be designed to process multiplexed physiological/pathological information in vivo, releasing loaded drugs on-site and regulating the expression of specific genes.

Although various DNA/RNA-based logic devices have been successfully fabricated, several crucial issues must be addressed to advance intelligent nanomaterials for future applications. For example, the fact that nucleic acid-based logic devices depend on the diffusion of substances results in the relatively slow reaction kinetics and some unexpected spurious interactions.^35,109 The reaction rates and the effectiveness of DNA/RNA logic devices may be affected by the leak reaction and spurious binding of nucleic acid sequences.^35,110 One of the solutions to these problems is to construct logic devices on constrained platforms, such as nanoparticles,¹¹¹ solid support,⁴⁴ or DNA assemblies.^112,113 Each layer of logic gates can be isolated on spatially separated supports to prevent them from mismatch.⁴⁴ Another widely adopted technique to reduce leak reaction is to replace the “toehold” sequence with a smaller domain called “clamps” to introduce an energy barrier for toeless strand displacement.^35,110,114

The resolution and selectivity of DNA/RNA logic devices still need to be improved. Decreasing the background noise of logic devices is also significant when they work in a complicated biological environment. In addition, the relatively high cost of DNA/RNA production is another factor that hinders the application of nucleic acid-based devices. Recently, the cost of DNA/RNA devices has been greatly reduced with the development of chemical synthesis techniques and biotechnological methods for the mass production of nucleic acid strands.^115–117 We expect nucleic acid logic devices with gradually growing complexity and various biological functions will play an increasingly important role in biomedical application in the future.

ACKNOWLEDGMENTS

This work is supported by the National Natural Science Foundation of China (22025201, 32071389, 21773044, and 51761145044), the National Basic Research Program of China (2016YFA0201601, 2018YFA0208900), Beijing Municipal Science & Technology Commission (Z191100004819008), Key Research Program of Frontier Sciences, CAS, the Youth Innovation Promotion Association, CAS, Grant QYZDBSSW-SLH029, CAS Interdisciplinary Innovation Team, and K. C. Wong Education Foundation (GJTD-2018-03).

CONFLICT OF INTEREST

The authors declare no conflict of interest.