Programmable DNA Logic Systems for Applications in Biomedicine

ABSTRACT

DNA has emerged as a versatile molecular engineering material owing to its predictable base‐pairing rules, structural programmability, and excellent biocompatibility. Programmable DNA logic systems capable of integrating diverse biological signals and execute predefined logical operations, bridging molecular computing with biomedical applications. In this review, we summarize recent advances in programmable DNA reaction and logic systems for biomedical applications. We first introduce DNA logic gates responsive to diverse biomolecular and physiological inputs, followed by cascaded DNA reaction circuits and molecular classifiers for amplified signal processing and precision diagnosis. We then discuss DNA nanostructures and intelligent nanomachines that integrate sensing, computation and actuation for therapeutic regulation. Finally, we outline current challenges and future opportunities toward the development of DNA‐based intelligent biomedical systems.

DNA has emerged as a versatile molecular engineering material owing to its predictable base‐pairing rules, structural programmability, and excellent biocompatibility. Programmable DNA logic systems capable of integrate diverse biological signals and execute predefined logical operations, bridging molecular computing with biomedical applications. In this review, we summarize recent advances in programmable DNA reaction and logic systems for biomedical applications. We first introduce DNA logic gates responsive to diverse biomolecular and physiological inputs, followed by cascaded DNA reaction circuits and molecular classifiers for amplified signal processing and precision diagnosis. We then discuss DNA nanostructures and intelligent nanomachines that integrate sensing, computation and actuation for therapeutic regulation. Finally, we outline current challenges and future opportunities toward the development of DNA‐based intelligent biomedical systems.

1. Introduction

Beyond its central role as the genetic carrier of life, DNA has emerged as a highly versatile molecular engineering material owing to its predictable base‐pairing rules, structural programmability, and excellent biocompatibility [1]. Over the past two decades, rapid advances in DNA nanotechnology have transformed DNA from a passive genetic polymer into an active building block for molecular computation, dynamic devices, and functional nanostructures [2]. In particular, the development of DNA strand displacement reactions (SDR), enzyme‐free cascade circuits, and 3D DNA nanostructures has enabled the construction of programmable biochemical networks capable of sensing, computing, and actuating at the molecular level [3, 4]. These technologies collectively give rise to programmable DNA logic and reaction systems, which can integrate diverse biological signals and execute predefined logical operations, thereby bridging molecular computing with biomedical applications [5, 6].

In the biomedical context, disease development is typically accompanied by complex and heterogeneous molecular signatures, including aberrant expression of nucleic acids, dysregulated protein profiles, and microenvironmental changes. Conventional diagnostic and therapeutic strategies often rely on single biomarkers, which can be insufficient to distinguish between different disease states or progression stages. DNA logic systems, in contrast, enable Boolean computation and multi‐input decision‐making, allowing output signals or therapeutic responses to be triggered only when specific molecular conditions are satisfied. By incorporating aptamers, enzymes, functional nucleic acids, and nanostructure scaffolds, these systems can sensitively recognize a wide range of biomarkers, thus offering opportunities for precision diagnostics, cell classification, and smart therapeutics [7, 8]. Meanwhile, cascaded DNA reaction networks further extend this capability from single‐step recognition to hierarchical computation, amplification, and pattern analysis, laying the foundation for molecular classifiers capable of performing intelligent disease diagnosis [9].

In parallel, DNA nanotechnology has evolved from static structural assemblies to dynamic and autonomous nanomachines [10, 11, 12]. Rigid DNA origami frameworks provide programmable platforms for spatially organizing biomolecules with nanometer precision, enabling applications in drug delivery, immune regulation, and vaccine engineering [13, 14, 15]. More advanced reconfigurable nanodevices and nanorobots integrate sensing, computation, and actuation into a single framework, achieving logic‐guided drug release, threshold‐dependent regulation, and autonomous therapeutic decision‐making [16, 17]. Such developments highlight that DNA materials are no longer passive carriers, but molecularly intelligent systems capable of interacting adaptively with complex biological environments.

In this review, we summarize recent advances in programmable DNA reaction and logic systems for biomedical applications (Figure 1). We first introduce DNA logic gates that respond to nucleic acid, protein, and physiological microenvironmental inputs, and summarize their biomedical applications in biosensing, cell identification, and therapeutic regulation. We then discuss cascaded DNA reaction circuits and molecular classifiers to enable amplified signal processing and precision diagnosis in complex biomedical environments. Next, we discuss static, dynamic DNA nanostructures and intelligent nanomachines that enable programmable molecular organization and controlled drug delivery. Then, we highlight current challenges and future opportunities toward DNA‐based intelligent biomedical systems. Finally, we discuss outstanding challenges including molecular stability, in vivo delivery, large‐scale programmability, and clinical translation, and outline future opportunities for advancing DNA nanotechnology toward truly intelligent biomedical systems.

FIGURE 1.

Programmable DNA logic systems for applications in biomedicine.

2. DNA Logic Gates for Biomedical Applications

As the fundamental building blocks of DNA‐based molecular computation, DNA logic gates represent the basic functional units for information‐processing. They implement Boolean logic principles using nucleic acids, leveraging programmable base‐pairing interactions and SDR to convert inputs into defined outputs, thereby enabling logic operations such as AND, OR, and NOT. DNA logic gates provide a powerful strategy for multi‐input biosensing. Only when specific combinations of biomarker signals are present does the logic circuit generate a detectable output, including fluorescence, colorimetric change [18, 19, 20], or downstream circuit activation [21]. With the integration of aptamers, enzymes, and DNA nanostructures, DNA logic systems have evolved to detect multiple categories of biomarkers, including nucleic acids and proteins, while also responding to environmental cues such as pH and light. Hence, in this section, we categorize current DNA logic systems into three hierarchical levels based on their input types and operational complexity: nucleic acid‐responsive gates, protein‐involved systems, and multi‐input integrated platforms. This reflects the functional evolution of DNA computing—moving from fundamental molecular recognition to sophisticated signal integration capable of processing the heterogeneous stimuli found in complex biological environments.

2.1. DNA Logic Gates Based on Nucleic Acid Biomarkers as Inputs

Nucleic acid biomarkers, especially microRNAs (miRNAs), play fundamental roles in gene regulation and disease progression. However, conventional detection approaches, including qPCR and RNA sequencing, frequently involves a series of labor‐intensive and time‐consuming procedures. In 2004, Benenson et al. demonstrated a molecular computation that could recognize specific RNA molecules in vitro and subsequently produce a DNA signal [22]. The authors suggested that such a nanoscopic DNA computer could recognize viral RNA and produce miRNA that activates cellular defense pathways, which would make DNA computations promising tools for at‐home or point‐of‐care diagnosis.

The AND DNA logic gate is the most commonly used tool for in vitro diagnostics, which produces a fluorescent signal only when two or more cancer‐related miRNAs are present simultaneously, thereby avoiding false positive results caused by a single biomarker [23]. Representative designs have demonstrated the feasibility of multi‐input AND logic operations for multiplexed detection of cancer‐associated miRNAs [24, 25]. DNA logic circuits hold promise for precise identification of cancer cells, but they face obstacles such as insufficient signal strength and the lack of an effective therapeutic platform. To address these issues, Zhang et al. developed a transmembrane DNA logic operation to enable DNA nanomachines to manipulate cancer cells in the complex microenvironment of solid tumors (Figure 2a) [26]. The DNA nanomachine is composed of multi‐shell upconversion nanoparticles, which carry aptamers targeting protein tyrosine kinase‐7 on the cancer cell membrane and miR‐21 inside the cells as the two inputs of an AND gate operation. The output DNA strand activates the photosensitizer rose Bengal, thereby generating reactive oxygen species. This system is expected to become a precise therapeutic strategy for cancer cells in the complex microenvironment of solid tumors. However, ensuring their stability and effective delivery to the tumor microenvironment remains a challenge.

FIGURE 2.

DNA logic gates based on nucleic acid biomarkers as inputs. (a) Schematic illustrations of the logic gate‐based transmembrane DNA computation with LA‐apt as input 1, and miRNA‐21 as input 2. Reproduced from reference [26], Copyright, 2021, American Chemical Society. (b) The structure and detection mechanism of DNA NHF with three AND logic gates entering cells to detect microRNAs. Reproduced from reference [28], Copyright, 2021, John Wiley and Sons.

Beyond in vitro diagnostics, nucleic acid‐responsive logic gates have been applied in cellular environments to profile endogenous gene expression patterns. By responding to intracellular miRNA combinations, these systems can distinguish between different cellular phenotypes, developmental stages, or disease states. For example, in 2013, Hemphill et al. designed a DNA‐based AND gate capable of recognizing miRNA inputs within mammalian cells, demonstrating that multi‐input DNA logic gates can sense the endogenous miR‐21 and miR‐122 distributions in living cells, exhibiting molecular recognition of miRNA profile changes [27]. This work represented the first demonstration of DNA logic gate operation at the cellular level. In 2021, Yang et al. constructed a DNA logic gate nano‐hexahedron framework (NHF) for multiple cancer cell subtype identification (Figure 2b) [28]. The DNA nanostructure served as a rigid and programmable scaffold to integrate three recognition modules and three reporting modules to form three “AND” logic gates. Benefiting from the high programmability, structural addressability, and biocompatibility of DNA nanostructures, this logic DNA nanodevice achieved accurate discrimination of different cancer cell subtypes and good intracellular performance.

Despite the promising performance of nucleic acid‐responsive logic gates, several limitations remain. In complex biological environments, the presence of homologous sequences and nonspecific hybridization can compromise selectivity, leading to false‐positive outputs. In addition, signal leakage arising from unintended SDR may reduce computational accuracy. Furthermore, achieving sufficient sensitivity while maintaining low background signals remains a major challenge, particularly for low‐abundance biomarkers in heterogeneous biological matrices.

2.2. DNA Logic Gates Based on Protein Biomarkers as Inputs

Extending DNA logic to protein biomarkers introduces new opportunities for direct cellular sensing. DNA logic systems have been engineered to respond to protein biomarkers through aptamer recognition or enzyme‐mediated reactions. The specific membrane proteins of disease‐related cells have also been used to detect cancer cells [29]. Compared with nucleic‐acid inputs, protein biomarkers often provide more direct information about cellular states and extracellular communication, particularly when located on the cell or vesicle membrane. In a representative example, the author constructed a programmable DNA logic platform capable of autonomously analyzing combinations of two or three cell‐surface biomarkers through modular Boolean gates [30]. This design enables higher‐order marker profiling and allows diagnostic reporting or targeted photodynamic therapy to be activated only under predefined molecular conditions. Moreover, the development of a 3D DNA logic gate nanostructure to achieve bispecific recognition of cancer cell membrane proteins using aptamers (Figure 3a) [31]. Compared to conventional linear logic circuits, the integration of all recognition and computing modules within a single structure results in enhanced cellular targeting and signal output. In the field of biological imaging applications, Feng et al. developed a DNA molecular logic gate nanomachine that performs an AND logic operation on two cancer cell membrane proteins, MUC1 and EpCAM, using aptamer recognition to trigger a downstream signal amplification for highly accurate fluorescence imaging [32]. DNA logic computation has also been extended to sensitive and accurate identification of tumor‐derived extracellular vesicles (tEVs), which holds great promise for noninvasive cancer diagnosis. Li et al. constructed a DNA logic gate system with protein biomarkers as input signals, which could perform molecular‐level computations on the membrane surface of tEVs (Figure 3b) [33]. The researchers used aptamers that recognize HER2 and EpCAM as dual inputs to achieve AND logic discrimination, and subsequently accomplished signal amplification. This system could highly specifically distinguish HER2+ and HER2− breast cancer (BC) patients in clinical serum samples and achieved a diagnostic accuracy of 97%, demonstrating the potential application of DNA computing with protein patterns as input in precise liquid biopsy.

FIGURE 3.

DNA logic gates based on protein biomarkers as inputs. (a) Working principles of the 3D DNA logic gate nanostructure to achieve bispecific recognition of cancer cell membrane proteins using aptamers. Reproduced from reference [31], Copyright, 2018, American Chemical Society. (b) Schematic of the DNA logic gate system which perform molecular‐level computations on the membrane surface of tEVs to achieve cancer diagnosis. The researchers used aptamers that recognize HER2 and EpCAM as dual inputs to achieve AND logic discrimination, and subsequently accomplished signal amplification. Reproduced from reference [33], Copyright, 2021, American Chemical Society.

However, protein‐based logic systems also face important challenges. The specificity of aptamer–protein recognition may be affected by structurally similar proteins or complex serum components, leading to potential interference. In addition, variations in protein expression levels and dynamic cellular environments may influence signal reliability and reproducibility. These factors highlight the need for improved anti‐interference capability and robust system design for practical applications.

2.3. DNA Logic Gates Based on Multiple Types of Inputs

Accurately distinguishing tumor cells from normal cells in the complex tumor microenvironment and the presence of multiple substances presents a significant challenge. Beyond typical biomarker inputs, DNA logic systems have also been engineered to respond to physiological microenvironmental signals and exogenous stimuli, thereby enabling programmable biological regulation. Chen et al. reported a heterogeneity‐driven DNA logic gate that uses membrane protein and extracellular acidity as inputs, in which the latter reflects a characteristic tumor‐associated physiological microenvironmental signal, enabling accurate and sensitive cancer cell identification [34]. Combining DNA logic systems with extra stimuli such as external light irradiation provided spatiotemporal control of DNA logic gates activation. For example, Shang et al. reported a logic‐gate‐activated lysosome‐targeting DNA nanodevice in which UV light and intracellular glutathione (GSH) serve as dual “AND‐gate” inputs to trigger lysosomal trafficking and controlled degradation of specific proteins in living cells [35]. DNA logic devices have also been engineered as intelligent imaging probes. For instance, lysosome‐recognizing framework nucleic acid nanodevices employ pH and ATP levels as logic inputs to activate fluorescence output, thereby achieving highly specific, logic‐guided subcellular bioimaging (Figure 4a) [36]. In a further advance, Chen and co‐workers developed aptamer‐guided, DNA‐controlled receptor assembly systems in which DNA strands and light irradiation function as programmable logic inputs (Figure 4b) [37]. They implemented multi‐input logic to program receptor proximity (such as c‐Met) and thereby modulate downstream cellular signaling and cell behaviors in a programmable manner. Such designs highlight the emerging trend of integrating biochemical cues with external stimuli to construct higher‐order logic systems capable of sophisticated therapeutic decision‐making.

FIGURE 4.

DNA logic gates based on multiple type of biomarkers as inputs. (a) Design and working principle of the AND logic gate framework nucleic acid nanodevices as an imaging probe. Lysosome‐targeting framework nucleic acid nanodevices utilize pH and ATP as dual logic inputs to trigger fluorescence activation, enabling highly specific, logic‐guided subcellular imaging. Reproduced from reference [36], Copyright, 2019, American Chemical Society. (b) Schematic representation of logical aptamer‐controlled receptor assembly for modulating cellular signal transduction. Chen and co‐workers developed an aptamer‐guided, DNA‐controlled receptor assembly system in which DNA strands and light irradiation serve as programmable logic inputs. Reproduced from reference [37], Copyright, 2019, John Wiley and Sons.

Nevertheless, increasing the number and diversity of inputs often leads to higher system complexity, which may reduce robustness and increase the risk of cross‐reactivity and signal interference. Balancing system complexity with operational reliability remains a key challenge for multi‐input DNA logic systems.

3. DNA Cascade Circuit for Biomedical Applications

Building on the fundamental principles of DNA logic gates, DNA cascade circuit expands molecular computation from single‐step logical responses to multilevel, interacting reaction networks. By linking multiple reaction steps, DNA cascade circuits enable enhanced sensitivity, increased computational complexity, and versatile functional outputs. In this section, we discuss recent advances in DNA cascade circuits for biomedical applications, focusing on cascade logic gate‐based systems, signal amplification strategy‐based circuits, and DNA molecular classifiers for disease diagnosis. This progression marks a transition from basic molecular logic operations to more sophisticated computational systems capable of functioning in complex biomedical environments.

3.1. Cascade Logic Gates‐Based DNA Circuits

Cascade logic gate‐based DNA circuits refer to multi‐layered systems in which individual DNA logic gates are interconnected in a sequential manner, such that the output of one logic operation serves as the input for subsequent gates. The advantage of cascaded logic gate lies in the ability to facilitate logic analysis of complex associations within cancers through precise design and manipulation of interactions between nucleic acid strands, resulting in accurate diagnostic outcomes. DNA cascade circuits have been increasingly applied to biosensing and molecular diagnosis [38, 39, 40, 41, 42]. Recently, Ma et al. developed a cascaded AND logic gate‐based electrochemical method with tetrahedral DNA framework probes, using logical combinations of five miRNAs to distinguish pancreatic, breast, and lung cancers (LCs), leveraging AuNPs for signal enhancement and offering a novel approach for precision medicine (Figure 5a) [43]. Beyond molecular diagnosis, cascaded and probabilistic DNA circuits have also been interfaced directly with living cells to modulate receptor clustering on natural killer (NK) cells (Figure 5b) [44]. This work demonstrated that DNA logic computation can be translated into tunable biological outputs, significantly enhancing tumor recognition and cytotoxic activity, and highlighting the therapeutic potential of DNA cascade regulation frameworks. Despite these advances, most DNA logic systems are structurally rigid, so even small functional changes require extensive redesign, limiting scalability and higher‐order cascades. Recently, Zhang et al. developed an allosteric strategy enabling flexible reconfiguration with minimal nucleotide editing, switching among multiple logic modes and functioning in living cells for miRNA‐responsive computation and programmable gene regulation [45]. These results point toward reconfigurable DNA circuits capable of adaptive computation in complex biomedical environments.

FIGURE 5.

DNA cascade logic gates and DNA circuits for biomedical applications. (a) Cascaded DNA logic gates based on miRNA combinations and electrochemical analysis method. Cascaded logic gates were designed and constructed based on logical relationship among five abnormally expressed miRNAs in pancreatic cancer (PC), BC and LC. Reproduced from reference [43], Copyright, 2025, Elsevier. (b) DNA probabilistic circuits are engineered to program receptor clustering on NK cell membranes in a controllable manner. By integrating two specific sequences of DNA input strands (IN1 and IN2), the circuits regulate the generation of output strands that activate artificial receptors and drive their assembly into clusters with tunable density, thereby modulating NK cell–tumor cell interactions and enhancing immune recognition and cytotoxicity in a programmable fashion. Reproduced from reference [44], Copyright, 2022, John Wiley and Sons.

Despite these advances, cascade logic gates‐based DNA circuits are inherently more complex than single‐layer logic systems, making them more susceptible to signal leakage and error propagation across multiple reaction steps. In addition, controlling reaction kinetics and maintaining consistent performance in heterogeneous biological environments remain significant challenges.

3.2. Signal Amplification Strategy‐Based DNA Circuits

Nucleic acids have emerged as important prognostic and diagnostic biomarkers for a wide range of diseases, particularly cancer. However, their extremely low abundance in clinical samples, often at femtomolar to attomolar levels, poses significant challenges for detection. Signal amplification strategy‐based DNA circuits are designed to enhance detectable outputs from low‐abundance inputs by incorporating amplification mechanisms. Unlike cascade logic circuits that primarily increase computational depth, these systems focus on improving sensitivity by generating amplified signals from a single or limited number of input molecules, enabling early disease diagnosis. Typical signal amplification strategies include catalytic hairpin assembly (CHA) [46, 47] and hybridization chain reaction (HCR) [48, 49], which are two well‐established isothermal, enzyme‐free DNA circuits.

CHA is a catalytic, entropy‐driven reaction network in which an initiator strand triggers the successive opening of metastable DNA hairpins to form a stable duplex product, while the initiator is released and recycled for multiple reaction cycles. This catalytic turnover mechanism enables a single target molecule to activate numerous hairpin pairs, resulting in efficient signal amplification. Consequently, CHA typically offers high sensitivity and rapid amplification kinetics. As an advanced CHA‐based amplification strategy, He et al. reported an orthogonally activatable catalytic DNA circuit for in vivo miRNA imaging (Figure 6a) [50]. In their design, CHA is first used to reassemble inactive split‐fuel strands into active fuels only when a specific intracellular mRNA initiator is present. These regenerated fuels then drive a downstream entropy‐driven DNA reaction, forming a self‐reinforced cascade that significantly amplifies the detection signal. However, its performance can be compromised by background leakage caused by spontaneous hairpin opening, necessitating careful sequence design and kinetic optimization.

FIGURE 6.

DNA cascade reaction serves as a signal amplification module using CHA and HCR strategies. (a) Schematic illustration of an orthogonally activated DNA circuit that couples a catalytic hairpin assembly (F‐CHA) module with an entropy‐driven reaction (EDR) network for intracellular microRNA imaging. Reproduced from reference [50], Copyright, 2022, John Wiley and Sons. (b) Schematic for the fabrication and HCR mechanism of the T‐probe for the detection of nucleolin on the surface of cancer cells. The TDN scaffold is functionalized with hairpin probes (H1 and H2) to construct a T‐probe system, which undergoes target‐induced hybridization and assembly, leading to fluorescence signal activation. Reproduced from reference [51], Copyright, 2022, John Wiley and Sons. (c) General design of DNA‐drug conjugates for computed delivery. Drugs of interest are conjugated to DNA hairpins capable of HCRs. Specific binders to biomarkers of interest are conjugated to a split initiator such that the HCR takes place only if the two biomarkers are present in proximity. Reproduced from reference [53], Copyright, 2026, Chen et al., under the terms of the Creative Commons CC BY license.

In contrast, HCR is a noncatalytic chain‐growth amplification process in which a trigger strand sequentially opens hairpins to form long double‐stranded DNA polymers, enabling substantial signal amplification from a single initiator. As an example, Zhang et al. developed an HCR probe based on tetrahedral DNA framework (TDF) to improve in situ tumor imaging sensitivity (Figure 6b) [51]. In this system, a TDN‐enhanced HCR detection system (T‐probe system) is introduced for cancer‐related targets and its versatility is demonstrated by detecting intracellular target miRNA‐21 and cellular membrane target nucleolin. HCR also serves as a powerful platform for constructing programmable DNA nanostructures. Recently, Lv et al. reported an HCR‐assembled DNA nanoarchitecture capable of integrating multiple functional nucleic acids (miRNA, siRNA, sgRNA, and mRNA), enabling enhanced gene silencing, genome editing, and mRNA delivery, as well as real‐time organelle monitoring and environment‐responsive intracellular drug release [52]. More recently, HCR has been extended to programmable therapeutic systems. Chen et al. developed a DNA–drug conjugate platform in which HCR enables logic‐gated and amplified drug delivery in response to specific cellular biomarkers (Figure 6c) [53]. The formation of long DNA polymers carrying multiple drug molecules results in over 100‐fold amplification of both signal and therapeutic payload relative to the input biomarkers using fluorescence quantifications, significantly enhancing intracellular drug accumulation via endocytosis.

Overall, CHA and HCR represent two complementary amplification paradigms in enzyme‐free DNA circuits. CHA provides efficient catalytic amplification with fast kinetics but requires stringent control to suppress leakage, HCR provides robust signal amplification with low background and generates spatially confined polymeric structures, which are advantageous for imaging and drug delivery, but its noncatalytic nature leads to slower kinetics and lower amplification efficiency per target molecule. Consequently, increasing efforts have been devoted to constructing cascaded and hybrid amplification circuits by integrating CHA and HCR with functional modules such as DNAzymes and nanomaterials, enabling synergistic signal amplification, reduced background interference, and improved spatiotemporal programmability [54, 55].

3.3. DNA Molecular Classifier for Disease Diagnosis

DNA molecular neural networks, extended from conventional DNA logic circuits, are molecular intelligent systems capable of implementing weighted summation, threshold‐based decision making, pattern recognition, supervised learning and classification [56, 57, 58, 59, 60]. Building on this concept, DNA molecular classifiers have been developed to map complex nucleic‐acid expression profiles onto discrete diagnostic outputs, providing a promising framework for early disease detection.

A landmark contribution came from Seelig and co‐workers, who in 2018 reported the first molecular multi‐gene classifier capable of directly executing a trained linear SVM model at the molecular level using RNA transcripts as inputs [61]. By assigning positive and negative weights to different RNAs and summing the resulting molecular signals, the system produced a binary diagnostic output, and was successfully applied to early cancer detection and differentiation of bacterial versus viral respiratory infections in synthetic samples. This work clearly established the conceptual and methodological foundation for DNA‐based molecular decision‐making. Building on this foundation, Han and colleagues in 2020 translated the approach into a clinically oriented workflow for LC diagnosis based on circulating miRNA signatures [62]. Their platform combined in‐silico classifier training with a simplified winner‐takes‐all molecular implementation scheme and demonstrated 86.4% diagnostic accuracy in real serum samples, thereby highlighting the translational potential of DNA molecular computation. However, such diffusible reaction networks generally required hours to reach completion, which limited clinical practicality. To overcome the limitations of speed and workflow complexity, they subsequently developed a fully automated DNA computing platform that integrates sample processing, amplification, computation, and reporting into a single device (Figure 7a) [62]. The system achieved rapid etiological diagnosis of acute respiratory infections (ARI) directly from clinical samples, without the need for external computers or expert operators. More recent studies have further improved computational efficiency and robustness [63, 64, 65]. For example, a spatially localized DNA linear classifier assembled on DNA origami scaffolds was shown to accelerate arithmetic operations and improve noise tolerance, enabling faster cancer diagnosis via miRNA signatures in both synthetic and clinical samples (Figure 7b) [66]. In parallel, freeze‐thaw cycling accelerated molecular classifiers have been introduced to shorten computation time of DNA molecular classifier for cancer diagnosis while maintain the performance [67]. These studies demonstrated that DNA molecular classifiers offer a promising route toward intelligent and low‐cost disease diagnosis. More recently, efforts have been directed toward integrating classification capability with autonomous actuation, thereby advancing DNA molecular systems from passive diagnostic tools to active therapeutic platforms. A recently reported all‐in‐one DNA nanodevice integrates molecular classification with programmable drug release, enabling simultaneous cell‐type identification and targeted intervention at the molecular level. In this system, multiple intracellular RNA biomarkers are processed through a DNA‐based computational network that performs weighted analysis and threshold decision‐making, ultimately triggering the release of therapeutic payloads only in target cells [68]. Such designs not only improve diagnostic specificity but also minimize off‐target effects, highlighting the potential of DNA molecular classifiers in precision medicine. These advances indicate a clear trend toward more intelligent, autonomous, and clinically translatable DNA molecular systems, where sensing, computation, and actuation are seamlessly integrated into a unified platform.

FIGURE 7.

DNA molecular classifier for disease diagnosis. (a) Workflow for the classification of ARI etiology with the DNA computation platform, and automated operational flow using DNA molecular classifier for the classification of ARI with clinical whole‐blood samples automatically. Reproduced from reference [62], Copyright, 2022, The American Association for the Advancement of Science. (b) Schematic illustration of a high‐speed, spatially localized DNA linear classifier assembled on DNA origami, in contrast to a nonlocalized DNA classifier with lower classification speed. Reproduced from reference [66]. Copyright, 2024, by Linlin Yang et al., licensed under CC BY 4.0.

4. DNA Nanotechnology for Biomedical Applications

Following the development of DNA‐based logic computation and cascade circuits, DNA nanotechnology further advances these concepts by integrating sensing, computation, and actuation into structurally defined nanosystems. Due to its high programmability, structural precision and excellent biocompatibility, DNA nanotechnology has evolved from a basic molecular engineering tool to an important platform driving biomedical applications. Unlike earlier molecular circuits that primarily operate through diffusible reactions, DNA nanostructures provide spatially organized and programmable platforms that enable precise control over molecular interactions. In this chapter, we first summarize static DNA nanostructures, which serve as programmable carriers for molecular assembly, delivery, immune modulation, and structural analysis in living systems. We then discuss dynamic and intelligent DNA nanomachines, whose reconfigurable architectures enable stimulus‐responsive behavior, logic‐guided therapy, and autonomous biomedical operation. Together, these advances demonstrate how DNA nanotechnology transforms molecular computation into functional nano‐biomedical systems with translational potential.

4.1. Static DNA Nanostructures for Biomedical Application

DNA nanotechnology has been widely used to construct static DNA nanostructures with defined sizes, shapes, and functional regions. These structures serve as precise carriers for spatial organization of proteins, nucleic acids, drugs, and fluorophores, thereby enabling biomedical applications such as targeted drug delivery, vaccine design and immune modulation, as well as biosensing and diagnostics. In this section, we will first introduce the development of DNA nanotechnology, and then focus on the multi‐dimensional applications of static DNA nanostructures in biomedicine.

4.1.1. DNA Nanotechnology

DNA nanotechnology originated in 1982, when Seeman proposed that the cross‐shaped DNA Holliday junctions can serve as building blocks for assembling higher‐order DNA structures [69]. This concept abstracted DNA from its traditional genetic context and redefined it as an engineering material with programmable molecular recognition. By rationally designing complementary DNA segments, four‐arm Holliday junctions (referred as Tile) were created and further assembled into nanoribbons and 2D lattices, which served as scaffolds for organizing proteins and nanoparticles (Figure 8a) [70, 71, 72, 73, 74, 75, 76]. Building on this principle, a variety of DNA tile systems were developed, including DNA brick assemblies capable of producing complex 3D shapes, 2D canvas arrays with addressable sites, and tubular lattices generated by wrapping tiles into nanoscale cylinders [77, 78, 79]. However, Tile‐based self‐assembly relies on numerous short DNA strands and requires precise stoichiometric control and purification, which limits the yield and structural complexity achievable through this method.

FIGURE 8.

DNA nanotechnology. (a) Complex DNA tile motif library and its assembly results. Reproduced from reference [76], Copyright, 2020, by Xuehui Yan et al., licensed under CC BY 4.0. (b) DNA origami nanostructures. Reproduced from reference [80], Copyright, 2006, Springer Nature. (c) Wireframe DNA origami nanostructures. Reproduced from reference [83], Copyright, 2015, Springer Nature.

A significant breakthrough occurred in 2006, when Rothemund introduced DNA origami, in which a long scaffold strand is folded into prescribed shapes by hundreds of short staple strands [80]. This strategy enabled high yield fabrication of complex 2D nanostructures including squares, triangles, stars, pentagons, and even “smiley faces” with nanometer precision (Figure 8b). Furthermore, 3D DNA origami was realized by Shih and colleagues, who stacked multiple DNA helices into multilayer honeycomb and square‐lattice bundles to construct rigid nanostructures such as boxes, prisms, and curved shells [81, 82]. Further innovations introduced wireframe DNA origami, where edges are formed from double‐helical bundles to folding arbitrary polygonal digital meshes including icosahedron, bottle, platonic tiling, and so on (Figure 8c) [83, 84, 85]. To simplify the structural design and improve efficiency, computer‐assisted tools have been developed to design scaffolds and staple routing of DNA nanostructures, achieving fully automated sequence generation [86, 87, 88].

4.1.2. Static DNA Nanostructures for Biomedical Applications

Static DNA nanostructures are not only served as precise carriers for drug and nucleic acid delivery, vaccine development and molecular diagnosis, but also provide unique tools for analyzing the spatial organization of biomolecules and cellular behavior. One of the earliest and most intensively explored biomedical applications of DNA nanostructures is targeted therapeutic delivery [89, 90, 91, 92, 93, 94, 95, 96]. Zhang et al. loaded the chemotherapeutic drug doxorubicin onto the surface of DNA origami nanostructures to achieve targeted enrichment and delivery at tumor sites [97]. Their results demonstrated that the potential of DNA origami nanostructures as innovative platforms for the efficient and safe drug delivery of cancer therapeutics in vivo. Wu et al. recently proposed a “genetically encoded DNA origami” strategy that enables DNA nanostructures to be synthesized directly in vivo through gene expression, thereby overcoming the limitations associated with conventional in vitro fabrication and delivery (Figure 9a) [98]. This work opens a new direction for the long‐term development of DNA origami–based in vivo gene therapy platforms [99].

FIGURE 9.

Static DNA nanostructures for biomedical applications. (a) Schematic illustration of genetically encoded DNA origami for gene therapy, including its construction from gene‐derived strands, lipid coating via DNA‐templated assembly, and subsequent cellular delivery and gene expression. Reproduced from reference [98], Copyright, 2023, American Chemical Society. (b) Schematic showing the pattern recognition process of CD95 signalling activation mediated by the DNA origami nanostructure with a hexagonal CD95L array pattern, and AFM images of DNA origami with proteins of different spacing arrangements. Scale bars, 50 nm. Reproduced from reference [104], Copyright, 2024, Springer Nature. (c) Design and characterization of DNA origami with tunable CpG spatial arrangements: programmable modification sites and CpG spacing (2.5–7.0 nm) on a square DNA origami platform (left), and corresponding morphology and dimensions confirmed by TEM imaging (right) Reproduced from reference [105], Copyright, 2024, Springer Nature. (d) Schematic illustration of nano‐vaccine design, showing programmable control of antigen spacing, antigen copy number within clusters, and cluster number on icosahedral DNA origami to optimize B cell receptor activation. Reproduced from reference [107], Copyright, 2024, by Qingqing Feng et al., licensed under CC BY 4.0.

A unique advantage of DNA origami in biomedical engineering lies not only in its ability to present multiple molecules, but more importantly in its capacity to program nanoscale spatial pattern‐recognition. With its nanometer‐level spatial control, DNA origami enables the rational programming of ligand geometry to exploit pattern‐recognition principles in biomedicine, such as antiviral binding, immune stimulation, and vaccine design [100, 101, 102]. Kwon et al. demonstrated this principle by constructing a star‐shaped DNA nanostructure whose vertex spacing matched the epitope organization on the dengue virus surface, leading to markedly enhanced viral binding inhibition and ultrasensitive detection [103]. More recently, Li et al. reported a reconfigurable DNA origami device that spatially organizes CD95 ligands with defined nanoscale spacing to control receptor clustering and signaling in vivo. By matching the ligand geometry to the organization of CD95 receptors, the origami selectively induced immune tolerance and alleviated disease symptoms in a rheumatoid arthritis model, demonstrating that therapeutic outcomes can be programmed through nanoscale spatial patterning of ligands (Figure 9b) [104]. This concept was further extended to innate immunity, where Zeng et al. used DNA origami to precisely tune the spacing of CpG ligands and showed that Toll‐like receptor 9 activation exhibits a strong preference for a ∼3–4 nm CpG separation, indicating that receptor signaling depends on recognition of a specific nanoscale ligand pattern rather than total CpG dose (Figure 9c) [105]. Complementary studies employing wireframe DNA origami further confirmed that both CpG copy number and spatial geometry synergistically define the magnitude and bias of immune activation [106]. More recently, Zhao and Nie's groups translated this pattern‐recognition principle to adaptive immunity by arranging viral antigens on icosahedral DNA scaffolds, revealing that the topology and clustering of antigen presentation critically regulate B‐cell receptor crosslinking and vaccine immunogenicity (Figure 9d) [107]. Together, these works establish DNA origami as a powerful molecular platform for engineering ligand geometry with nanometer precision.

4.2. Dynamic DNA Nanomachines for Biomedical Applications

DNA nanotechnology has further advanced toward dynamic systems capable of motion, structural transformation, and autonomous operation. These dynamic DNA nanomachines integrate molecular recognition, computation, and actuation within a single nanoscale platform, enabling controllable drug release, signal amplification, and intelligent biomedical intervention. In this section, we highlight recent developments in DNA walkers, reconfigurable nanodevices, and intelligent nanorobots, illustrating how dynamic behavior allows DNA materials to interact adaptively with complex biological environments and move toward truly responsive therapeutic and diagnostic systems.

4.2.1. DNA Walker System for Signal Amplification

DNA walkers are dynamic nucleic acid devices that autonomously move along defined tracks via SDR or DNAzyme reactions [108, 109], and they have been increasingly exploited in biomedical research for signal amplification and intracellular imaging [110]. Early work used DNA walkers as solution‐based fluorescence amplifiers, where repetitive walking steps generate multiple reporter signals from a single target recognition event, thereby enhancing sensitivity for trace biomolecule detection [111]. Based on the same principle, Xiao et al. developed stochastic DNA walkers confined within microdroplets to enable multiplexed bacterial phenotype analysis, demonstrating sensitive detection in complex samples by translating walking‐mediated amplification into a barcode readout (Figure 10a) [112].

FIGURE 10.

DNA walker system for signal amplification. (a) Working principle of the stochastic DNA walker sensor and Real‐time monitoring of single DNA walker sensor by TIRF. Reproduced from reference [112], Copyright, 2019, John Wiley and Sons. (b) Intracellular operation of a DNAzyme motor initiated by a specific miRNA. The DNAzyme motor system is constructed on the AuNP. Reproduced from reference [113], Copyright, 2017, by Hanyong Peng et al., licensed under CC BY 4.0. (c) Schematic overview of UV light‐photoactivated DNA walker based on DNA nanoflares in HeLa cells for dictated signal amplification miRNA imaging. Reproduced from reference [114], Copyright, 2021, American Chemical Society.

For intracellular applications, researchers have adapted walkers to operate inside living cells to detect disease‐related miRNAs. For example, microRNA‐triggered DNAzyme walkers were constructed so that endogenous microRNA activates the walking process, leading to catalytic cleavage of substrate strands and fluorescence amplification in situ (Figure 10b) [113]. This design enables highly sensitive monitoring of disease‐related RNA expression at the single‐cell level. Similarly, Chen et al. designed a photoactivated DNA walker based on DNA nanoflares, in which a photocleavable activation step triggers a DNA walker for signal‐amplified miRNA imaging under controlled light irradiation in single living cells (Figure 10c) [114]. Moreover, Ye et al. reported a near‐infrared (NIR) light–controllable DNA walker that is powered by endogenous ATP and UCNP‐converted NIR excitation, enabling spatiotemporally precise miRNA imaging with low background in living cells [115]. Overall, DNA walkers provide a programmable molecular tool for autonomous and localized signal amplification in biomedical analysis.

4.2.2. Reconfigurable DNA Nanomachines for Drug Release

Allosteric and reconfigurable DNA nanomachines provide a powerful strategy for controlled drug delivery, because their structures can switch between “closed” and “open” states in response to physiological signals. In this section, we summarize representative systems according to the nature of the trigger signal.

A first class relies on pH‐responsive locks, such as triplex or i‐motif domains, that stabilize a closed origami capsule at physiological pH but open in acidic tumor or lysosomal environments. Using this strategy, Ijäs et al. demonstrated a reconfigurable DNA origami nanocapsule that reversibly encapsulates and releases proteins through pH‐dependent structural switching (Figure 11a) [116]. Similar acid‐triggered mechanisms have been applied to immunotherapy, Liu et al. reported a tubular “locked” DNA vaccine platform in which antigens and adjuvants are concealed inside the DNA tube and only exposed upon acid‐triggered unlocking in endo‐lysosomes, thereby enhancing antigen presentation and antitumor immune responses in vivo [117].

FIGURE 11.

Reconfigurable DNA nanomachines for drug release. (a) DNA origami nanocapsule that reversibly encapsulates and releases proteins through pH‐dependent structural switching. Reproduced from reference [116], Copyright, 2019, by Heini Ijäs et al., licensed under CC BY 4.0. (b) The barrel‐shaped DNA nanorobot which conformational switching induced by tumor vascular markers. Reproduced from reference [118], Copyright, 2018, Springer Nature.

A second class is activated by specific molecular recognition events. A typical example is the barrel‐shaped DNA nanorobot, the unlocking is achieved through aptamer conformational switching induced by tumor vascular markers, and local exposure of encapsulated thrombin is triggered, thereby completing target‐dependent antitumor therapy (Figure 11b) [118]. Related strategies include dual‐chain‐locked origami nanocages, which confine CRISPR/Cas9 ribonucleoproteins until tumor‐associated biomarkers trigger unlocking, achieving gene‐editing control linked directly to cellular molecular states [119]. Tubular DNA nanodevices have been designed to co‐load siRNA and chemotherapeutic drugs, allowing programmed intracellular release and enhanced combination therapy [120, 121]. Such dynamic capabilities lay the groundwork for the development of molecularly intelligent devices capable of interacting with biological environments in programmable and adaptive ways.

4.2.3. Intelligent DNA Nanomachines for Biomedical Applications

Unlike conventional nanocarriers that passively release drugs, intelligent DNA nanomachines are designed to sense biological cues, process this information, and respond with controlled therapeutic or diagnostic actions. Early research demonstrated that DNA origami devices can integrate logic gates so that payload exposure only occurs when a defined combination of biomarkers is present. For example, Douglas et al. reported a logic‐gated nanorobot that opens only after recognizing multiple cell‐surface inputs, enabling conditional delivery of antibody fragments to target cells [122]. Amir et al. further showed that interacting DNA origami robots can perform Boolean logic operations in vivo, where “positive” and “negative” robots regulate the activation state of effector robots in living animals, illustrating that DNA robots can perform distributed computation inside biological systems [123].

More recent studies have focused on quantitative decision‐making and threshold recognition rather than simple “on/off” switching. Yang et al. developed an intelligent anticoagulation robot that continuously monitors thrombin concentration and autonomously releases inhibitors only when thrombin exceeds a preset threshold, thus avoiding both under‐ and over‐treatment [124]. In a related direction, Yin et al. reported a thrombin‐responsive DNA origami nanodevice for precision thrombolysis, where interlocking DNA triplex fasteners act simultaneously as thrombin recognizers, threshold controllers and opening switches, ensuring that tissue plasminogen activator is exposed only at pathological thrombus sites (Figure 12a) [125]. These systems highlight a shift from passive delivery carriers to self‐regulated therapeutic platforms capable of dosage control based on real‐time biochemical cues. Intelligent mechanisms have also been used for information processing and spatiotemporally controlled diagnostics. For instance, DNA framework state machines (DFSM) have demonstrated that DNA devices can record, update and report their internal states inside living cells over time (Figure 12b) [126]. Recent efforts have advanced intelligent DNA nanomachines toward integrated spatiotemporal actuation, where environmental cues are directly linked to structural reconfiguration and therapeutic output. In this context, Liu et al. developed an allosteric DNA nanorobot for targeted glioma therapy that exemplifies a higher level of functional integration between sensing, actuation, and drug delivery [127]. Cytosine‐rich motifs function as pH‐responsive elements that sense acidic tumor microenvironments, triggering protonation‐induced C–G·C⁺ triplex formation and driving a conformational transition from a planar DNA origami sheet to a compact tetrahedral DNA nanostructure (TDN). This structural transformation is coupled with functional output: the TDN enhances stability and blood–brain barrier penetration during circulation, while its destabilization under acidic conditions promotes the release of intercalated doxorubicin, enabling precise spatiotemporal drug delivery.

FIGURE 12.

Intelligent DNA nanomachines for biomedical applications. (a) Logical operation for thrombin‐responsive opening in the intelligent core of the DNA nanodevice. The planar DNA scaffold is first functionalized with cargo molecules and subsequently rolled into a tubular architecture, where thrombin aptamers act as molecular recognition elements and DNA locker strands serve as structural switches to regulate device stability and cargo encapsulation. Scale bar, 50 nm. Reproduced from reference [125], Copyright, 2024, Springer Nature. (b) A DNA framework‐based state machine enables sequential, key‐triggered unlocking and programmable transitions between multiple states (S1–S6). This temporally controlled structural reconfiguration allows ordered information processing in living cells, as validated by AFM characterization. Scale bar, 10 nm. Reproduced from reference [126], Copyright, 2023, Springer Nature.

These progression highlights the evolution of DNA nanorobotics toward autonomous, context‐responsive systems for complex in vivo applications. However, the clinical translation of DNA nanostructures remains challenging. Issues such as nuclease degradation, limited circulation stability, inefficient cellular uptake, and potential immune responses may significantly affect their in vivo performance. In addition, large‐scale and cost‐effective production remains a barrier to practical applications.

5. Summary and Outlook

In summary, programmable DNA logic systems provide a molecular platform for sensing, processing, and acting on biological information. These DNA circuits support Boolean computation, cascade amplification, molecular classification, and conditional actuation in a biocompatible format. When combined with DNA nanostructures, DNA chemical networks can couple molecular recognition events to controlled therapeutic release, enabling programmable control over biological functions.

Several unifying general principles and design rules can be distilled across DNA logic gates, cascade reaction networks, and DNA nanotechnology‐based systems. At the most fundamental level, all DNA‐based computational systems rely on programmable molecular recognition to convert biological and physiological inputs into defined outputs. This is achieved through sequence‐specific hybridization or aptamer–target interactions, forming the basis for information encoding and signal transduction. Building on this, SDR serve as a universal computational mechanism, enabling modular logic operations, signal propagation, and cascading across different system scales—from simple logic gates to multi‐layer reaction networks. A key general principle underlying these systems is the tight coupling between thermodynamics and kinetics. Reaction pathways must be energetically favorable while maintaining controlled reaction rates, which are typically regulated through toehold design, strand stability, and local concentration effects. This balance becomes increasingly critical in cascade circuits, where error accumulation and signal attenuation can significantly affect computational fidelity. From a system design perspective, several common rules emerge. First, minimizing leakage and suppressing background reactions are essential for maintaining high computational accuracy, particularly in multi‐step circuits and amplification systems. Second, modularity and orthogonality must be carefully engineered to enable scalable circuit construction without cross‐reactivity between components. Third, signal amplification strategies are often required to enhance sensitivity toward low‐abundance biomarkers in complex biological environments. Importantly, DNA nanotechnology introduces an additional spatial design dimension that distinguishes it from purely diffusible systems. By organizing functional modules within defined nanostructures, local concentration, reaction kinetics, and pathway specificity can be precisely regulated. This spatial confinement not only improves reaction efficiency and reduces leakage, but also enables the integration of sensing, computation, and actuation within a single nanoscale platform. Collectively, these principles indicate that the performance of DNA‐based biomedical systems is governed not only by sequence‐level design, but also by system‐level considerations, including reaction kinetics, spatial organization, noise suppression, and environmental adaptability. The convergence of these factors defines the transition from simple molecular logic to robust and functional DNA‐based intelligent systems.

Several challenges must be addressed before DNA‐based molecular computing can be applied reliably in biomedical settings. At the reaction‐network level, leakage reactions, kinetic variability, and signal attenuation across cascades reduce computational accuracy. Quantitative modeling under physiological conditions remains difficult, particularly in crowded or protein‐rich environments. From an engineering standpoint, scalable synthesis, structural stability, nuclease resistance, and predictable pharmacokinetics remain essential barriers to in vivo use. Additional complications arise at the biological interface, including immune recognition, nonspecific interactions, and variability in cellular uptake. Finally, translation toward clinical use will require standardized manufacturing, reproducible quality control, long‐term safety evaluation, and regulatory approval pathways.

Several emerging trends are shaping the future development of DNA logic systems in biomedicine. First, there is a clear transition from simple Boolean logic gates toward higher‐order and multi‐layered computational systems, including cascaded circuits and molecular classifiers, enabling more sophisticated decision‐making. Second, DNA logic is increasingly being integrated with structural DNA nanotechnology, giving rise to intelligent nanomachines that combine sensing, computation, and actuation within a single platform. Third, the field is moving from in vitro proof‐of‐concept demonstrations toward in vivo and clinically relevant applications, where challenges such as stability, delivery, and immune compatibility become critical. The integration of DNA computing with other technologies, such as CRISPR systems and machine learning‐guided design, is expected to further enhance programmability and functional complexity. With continued progress in molecular engineering and translational research, intelligent DNA nanomachines may develop from experimental demonstrations into practical tools for precision diagnostics and controlled therapy.

Conflicts of Interest

The authors declare that they have no Conflicts of Interest.

Biographies

Junke Wang received her Ph.D. in 2024 under the supervision of Prof. Lianhui Wang and Prof. Jie Chao from Nanjing University of Posts and Telecommunications. She then worked as a postdoc in Prof. Jie Chao's group at Nanjing University of Posts and Telecommunications from 2024 to now. Her current research interests focus on construction of algorithm‐empowered DNA nanomachines and its biomedical applications.

Xiaorui Tian obtained his Bachelor's degree from Nanjing University of Posts and Telecommunications. In 2023, he joined Prof. Chao Jie's group as a master's student at Nanjing University of Posts and Telecommunications. His current research interest is DNA computing for applications in biomedicine.

Baodong He obtained his Bachelor's degree from Nanjing University of Posts and Telecommunications. In 2024, he joined Prof. Jie Chao's group as a master's student at Nanjing University of Posts and Telecommunications. His current research interest is design of intelligent DNA nanomachines.

Lianhui Wang received his Bachelor's degree in 1988 from the Chengdu University of Science and Technology, and Ph.D. in 1998 from the Zhejiang University. After completing all his degrees, he had two years of postdoctoral experience at University of Singapore. In 2000, he was a researcher at the National University of Singapore. In 2005, he was hired by Fudan University and became a professor in the laboratory. In 2011, he joined Nanjing University of Posts and Telecommunications. His current research is focused on nanoparticle photonic materials, bioelectronics, and biosensors.

Jie Chao received her Ph.D. from Nanjing University in 2008. In 2006–2007, she was a visiting student in Prof. Seeman's laboratory at New York University. After her postdoctoral research at Nanjing University, she joined the Shanghai Institute of Applied Physics (SINAP) in 2011 as an associate professor. In 2014, she moved to Nanjing University of Post and Telecommunications as a professor. Her research interests focus on DNA assembly and bioapplications.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.