Hierarchical core–shell DNA condensates enable programmable information storage and encryption Authors: Likang Chu, Liqi Wan, Haixia Wang, Peng Wang, Linyi Deng, Yuan Gao, Pei Guo, Lei He, Da Han Published: 2025 ABSTRACT AbstractUtilizing DNA’s molecular programmability, massive parallelism, and minimal energy requirements, it emerges as a transformative medium for secure data storage and encryption. However, practical implementation is limited by technical challenges, particularly the development of robust, programmable systems for flexible data encoding, precise molecular operations, and reliable encryption. Here, we present a molecular information storage and encryption platform that integrates hierarchical core–shell DNA condensates with biomolecular computing networks. It allows programmable and rapid execution of core operations such as information encoding, erasure, rewriting, replication, and repair within readily accessible and readable DNA-based condensates. Moreover, programmable biomolecular computing circuits endowed the system with unprecedented encryption capabilities, including multi-level logic encryption, time-dependent dynamic encryption, living system-driven encryption and fine-grained access control of information. Together, this work represents a meaningful advancement in molecular information storage and encryption, with notable advantages in dynamic editability and system scalability.Subject terms: DNA computing, Bioanalytical chemistry Introduction In the digital era, the transformation of information processing has exposed the inherent limitations of traditional storage and encryption technologies in managing vast and dynamic datasets. DNA-based information systems have garnered significant attention due to their remarkable molecular programmability, exceptional theoretical storage density and long-term stability1–3. Currently, the most common approach involves directly encoding information into DNA sequences, offering significant advantages in storage capacity. Nonetheless, the static nature of DNA sequence-based information encoding fundamentally restricts real-time manipulation and dynamic reconfiguration, thereby limiting its suitability in scenarios requiring adaptive data processing and programmable encryption4–6. In contrast, dynamic information storage and encryption systems leveraging DNA self-assembly are gaining increasing attention by offering a compromise between static architectures and controllable conformational transitions7–10. These systems typically integrate static structures with dynamic functional units, the latter enabling programmable conformational changes through DNA strand displacement reactions or external stimuli such as pH changes, light, or enzymatic reactions11–14. Such dynamic features facilitate information storage and encryption via structural rearrangements and reconfigurations. However, current technologies struggle to achieve diverse conformational states—with encoded states typically limited to only a few—thus severely restricting potential information capacity. Furthermore, the inability to facilitate arbitrary and direct transitions between different conformational states limits both the programmability and scalability of these systems. These challenges underscore the urgent need for a novel molecular platform that offers high storage capacity while enabling dynamic operability and robust encryption capabilities. Recently, hierarchical core-shell DNA condensates, formed via liquid-liquid phase separation, have emerged as a promising material platform15–18. In contrast to traditional methods that construct nanoscale DNA structures through sequence-specific supramolecular interactions, hierarchical core-shell DNA condensates use temperature-gradient-driven regulation to combine macroscopic phase separation with precise duplex strands recognition, thus enabling the assembly of complete structures at the microscale. This heating–cooling cycling strategy not only broadens the structural size range but also promotes the spontaneous formation of stable hierarchical organizations during the cooling process. Our recent studies have demonstrated that hierarchical core-shell DNA condensates offer a versatile and efficient platform for constructing intelligent molecular systems with multiple functionalities19,20. Nonetheless, the application of hierarchical core-shell DNA condensates for high-capacity information storage and programmable encryption is still in its early stages, presenting immense, untapped potential for innovation. In this study, we introduce and develop a molecular information processing platform based on hierarchical core-shell DNA condensates. The hierarchical architecture of this platform brings significant advances in both molecular design complexity and operational flexibility. Firstly, owing to their microscale spatial partitioning, hierarchical core-shell DNA condensates facilitate independent functional encoding in distinct spatial layers (e.g., core and shell), and the synergistic interactions between these regions significantly enhance the multidimensionality and organization of information storage. Secondly, the microscale organization enables precise decoding of fluorescent barcodes across different spatial partitions using conventional microscopy techniques (e.g., confocal microscopy), thereby improving information retrieval efficiency and reducing decoding complexity. Last but not least, the semi-open nature of these condensates permits molecular diffusion between different spatial regions and the external environment, which not only enhances the efficiency of information exchange and cooperative interactions but also dramatically improves the editability and flexibility of information storage. For instance, operations such as erasure, rewriting, and replication of information can be efficiently executed within the system. Our experiments further confirm the robustness of this platform, showing that hierarchical core-shell condensates retain their structural integrity and information readability even under stress conditions such as dehydration or repeated freeze-thaw cycles. In instances where high-temperature-induced structural disruptions cause information loss, we successfully restore the original information using an Erase–and–Diffuse strategy. Notably, by conducting parallel molecular computations across various spatial hierarchical structures within the DNA condensates, we establish a computational framework capable of facilitating arbitrary transitions between encoded states. This framework lays the foundation for designing dynamically complex, multi-layered encryption schemes. As a proof of concept, we integrated DNA-based molecular circuits to develop diverse programmable encryption strategies, encompassing multi-layer logic, bio-system interactions, time-dependent dynamics, and fine-grained access control. These advances provide a novel experimental paradigm for building molecular information platforms that combine high stability with flexible operability. Overall, this study not only significantly broadens the application scope of DNA materials but also establishes a practical foundation for the multifunctional integration of future intelligent molecular systems. Results Information encoding on DNA condensates Firstly, owing to their microscale spatial partitioning capability, DNA condensates facilitate independent fluorescence encoding in distinct structural layers (e.g., core and shell). These distinct fluorescence encodings can be further integrated through synergistic interactions between regions, thereby enhancing the multidimensionality and organization of the encoded information. Moreover, to accurately differentiate fluorescence encodings within the core-shell structure, the fluorophores employed in different hierarchical regions are required to maintain strict orthogonality. As shown in Fig. 1a, labeling DNA strands complementary to the barcode sequences located in distinct layers of the condensate—within the core (m*) and on the shell (n*)—with orthogonal fluorophores achieves spatially segregated distributions of the fluorophores. Specifically, each fluorophore adopts one of three possible distributions: exclusively on the shell, exclusively in the core, or completely absent from the condensate. As illustrated in Fig. 1b, for a three-color fluorescence encoding system, three orthogonally distributed fluorophores can produce 27 (3³) unique encoding modes. Each encoding mode corresponds to a discrete piece of information, thereby representing a range of values. Consequently, these 27 encoding modes are sufficient for comprehensively encoding the entire English alphabet. Figure 1c illustrates the correlation between fluorescence distributions and their corresponding alphabet representations. Importantly, the mapping between fluorescence encoding patterns and characters is arbitrary and fully reconfigurable; users can redefine it at will, allowing reassignment for alternative symbol sets. Notably, by expanding to four fluorescent channels, the encoding capacity increases exponentially to 81 (3⁴) unique modes (Supplementary Fig. 1), thereby establishing a robust framework for developing more comprehensive international encoding systems. In general, standard confocal fluorescence microscopes can distinguish up to five channels, constrained by spectral overlap. Beyond this limit, extending the spectral range into the near-infrared windows or, orthogonally, exploiting fluorescence-lifetime encoding offers significant potential for further expansion of encoding capacity. In this conceptual study, unless otherwise specified, we focus on the three-color encoding system described above as the representative example. Thus, once the entire English alphabet is constructed using the three-color encoding system, the encoded letters can be arranged and combined in specific sequences to generate more complex textual information. As shown in Fig. 1d, DNA condensates encoding different letters were organized in a predetermined order within a 384-well plate, and information was subsequently retrieved using confocal microscopy. As a proof-of-concept demonstration, we successfully encoded and decoded Shakespeare’s iconic quote, “TO BE OR NOT TO BE, THAT IS THE QUESTION”. These results indicate that DNA condensates offer a versatile platform for multi-layer information encoding, suggesting a reliable approach to advancing molecular data storage and processing. Information encoding in the DNA condensate system. a Schematic illustration of the synthesis of DNA condensates and their spatial-fluorescence encoding mechanism. Specifically, the DNA condensates are formed by a heat-induced self-compartmentalization of two single-stranded DNA (ssDNA) multiblock copolymers, namely, p(A20-m) and p(T20-n). Under 95 °C, p(A20-m) undergoes phase separation, forming the core of the DNA condensate. During the subsequent cooling process, the T20 segments from p(T20-n) hybridize with the A20 segments on the condensate surface, resulting in the formation of the outer shell. As both p(A20-m) and p(T20-n) contain specific encoding sequences (m or n, respectively), this setup facilitates a partitioned yet synergistic information encoding strategy within the DNA condensates. By integrating fluorescence labels into distinct spatial regions, up to 3n unique encoding modes can be achieved, where n denotes the number of available fluorescent channels. b The encoding table for the 26 letters. Encoded information is categorized into monocolor, bicolor, and tricolor modes, based on the number of fluorescent colors incorporated within the DNA condensates. Specifically, when only one fluorescent color is present, the DNA condensates generate six encoding modes, corresponding to letters A through F. Similarly, when two fluorescent colors are present, twelve encoding modes are generated, representing letters G through R. Finally, when three fluorescent colors are present, eight encoding modes emerge, corresponding to letters S through Z. In the absence of fluorescence, a Null encoding mode is produced, signifying a blank information element. c Confocal laser scanning microscopy (CLSM) images depicting the decoding of the 26 encoded letters. The large panel on the right highlights the merged fluorescence signals of multiple encoded letters, while the smaller panels on the left offer detailed visualizations of the spatial fluorescence distribution corresponding to individual letters, arranged sequentially as Merge, Cy5, Cy3, and FAM channels. Scale bar: 5 μm. d Information storage based on the DNA condensate system. Schematic illustration created in BioRender. He, L. (https://BioRender.com/tbnewg9). Red, green, and blue in all CLSM images represent Cy5, Cy3, and FAM fluorescence, respectively. Scale bar: 1 μm. All CLSM images shown are representative of three independent experiments. Information erasure and rewriting in DNA condensate systems Achieving molecular-level information erasure and rewriting not only enhances the reusability of storage media but also lays the foundation for advanced dynamic information storage and encryption systems21–23. Next, we demonstrate precise control over information erasure and rewriting within a DNA condensate system (Fig. 2a). Specifically, light-responsive regulatory units were introduced into the DNA condensates to construct a photo-responsive molecular switch system, achieved by selectively modifying DNA strands complementary to the barcode sequences with photosensitive PC-linker groups. Upon ultraviolet (UV) light irradiation, the PC-linker undergoes controlled cleavage, leading to the dissociation of the DNA strands complementary to the barcode sequences from the DNA condensate structure24,25. Taking the DNA condensates encoding the message “SAVE” as an example, this process enables the spatially directed removal of encoded information from the condensates (Supplementary Fig. 2). Notably, this light-controlled cleavage not only achieves information erasure but also exposes reprogrammable single-stranded DNA barcode regions (m and n) in both the core and shell of the condensate, thereby providing the molecular basis for subsequent information rewriting. Building on this, we demonstrated that introducing new fluorescently labeled complementary DNA strands (m* and n*) into the system enables precise rewriting of information. This erase–and–rewrite cycle effectively establishes a dynamic molecular information storage system that is capable of supporting multiple iterative operations in theory (Fig. 2b). As a proof of concept, multicycle writing, erasure, and rewriting of month-related information was successfully demonstrated using DNA condensates. As illustrated in Fig. 2c, we initially encoded empty DNA condensates with the month “NOV”. Following UV light treatment, the “NOV” encoding was completely erased, reverting the condensates to the Null state. Subsequently, upon addition of a new encoding strand bearing a distinct fluorescent label, the condensates were reencoded with the new month “AUG”. Leveraging this approach, the system enabled an iterative cycle of encoding, erasure, and reencoding (e.g., Null–NOV–Null–AUG–Null–FEB–Null). Importantly, the decoding accuracy remains consistently high (~100%) even after three full cycles of writing and erasing (Supplementary Fig. 3). This experiment demonstrates the potential of the system for dynamic information processing applications. Information erasure and rewriting in DNA condensate systems. a Diagram illustrating the principles of information erasure and rewriting, enabled by the modulation of DNA condensates. b Detailed mechanism describing the molecular processes involved in information erasure and rewriting. c CLSM images demonstrating the results of three consecutive cycles of writing, erasure, and rewriting, showcasing reproducibility and stability in the system. Scale bar: 5 μm. Red, green, and blue in all CLSM images represent Cy5, Cy3, and FAM fluorescence, respectively. All CLSM images shown are representative of three independent experiments. Information replication and repair in DNA condensate systems Creating multiple backups is crucial for preserving stored information against data loss and enables efficient sharing with minimal impact on the original system. Traditional DNA sequence-based storage systems rely on polymerase chain reaction (PCR) for replication; however, this process is prone to errors—such as mutations, deletions, and insertions—that compromise data fidelity26,27. Moreover, PCR suffers from sequence biases, resulting in uneven amplification that undermines data integrity and retrieval accuracy28. In contrast, DNA structure-based storage systems encounter even greater challenges owing to the absence of a mature, universal method (comparable to PCR) for efficiently replicating their dynamic architectures29,30. This limitation arises from the lack of established strategies for reliable replication—a consequence of their fundamentally distinct storage mechanisms. To address this critical gap, we propose an innovative Erase–and–Diffuse strategy that enables high-fidelity replication of information within DNA condensate systems (Fig. 3a). Importantly, although the replication process induces some redistribution of fluorescence signals in the parent DNA condensate, our system leverages high-density encoding sites generated via dual-round DNA rolling circle amplification (RCA), ensuring minimal information loss. For example, by encoding the information “RICE”, we performed consecutive rounds of replication (Fig. 3b). The replicated DNA condensates retained excellent readability, and the encoded information was precisely analyzed using confocal microscopy, thereby demonstrating the low-loss characteristics of this replication method. Naturally, for extended multi-round replication, gradual signal attenuation may eventually affect decoding fidelity. In such cases, signal levels can be sustained by adjusting the parent-to-progeny ratio or supplementing the system with fresh labeled strands, thereby ensuring robust information readout throughout multiple cycles. Information replication and repair in the DNA condensate system. a Principle of information replication. Specifically, the core of this strategy is the use of two carefully designed universal DNA strands (X and Y) in conjunction with a Null DNA condensate serving as the acceptor. First, the DNA displacement strands (X and Y) selectively hybridize with the information-encoding strands (m* and n*) located on both the interior and surface of the DNA condensates via strand displacement reactions, thereby temporarily erasing the encoded information from DNA condensates. Moreover, the DNA strands X and Y are engineered to incorporate photosensitive PC-linker groups that undergo controlled cleavage upon brief ultraviolet (UV) light exposure. This cleavage releases the original information-encoding strands (m* and n*) back into their active, single-stranded form. Thereafter, the reactivated strands rehybridize with their corresponding barcode regions (m and n) within the parent DNA condensate, thereby restoring the original information. Simultaneously, the released information-encoding strands (m* and n*) diffuse to the “Null” DNA condensate and hybridize with the barcode regions on the new carrier, thereby completing the replication process. b Demonstration of information replication. During the first round, the template “RICE” is replicated into three identical copies. In the second round, one of these copies serves as a new template, further producing three additional replicated copies. c Mechanistic principle of information damage and repair. Structural damage to the DNA condensate encoding information can be restored through a defined information repair pathway. d Demonstration of information repair. CLSM images display the encoded information “HELLO” at three stages: prior to thermal damage, following thermal damage, and after successful repair of the information. Scale bar: 5 μm. Red, green, and blue in all CLSM images represent Cy5, Cy3, and FAM fluorescence, respectively. All CLSM images shown are representative of three independent experiments. Another critical challenge for DNA structure-based storage systems is structural stability. Maintaining structural integrity is essential for reliable information storage and retrieval. As shown in Supplementary Fig. 4, DNA condensates encoding the target information retained their spatial hierarchical structure and enabled accurate retrieval of encoded data after storage at 4 °C for several months, with decoding fidelity remaining above 94%. In parallel, DNA condensates preserved in aqueous solution and subjected to slow evaporation at room temperature also maintained retrievable encoded information, with the decoding accuracy retained at ~98% (Supplementary Fig. 5). Extreme temperature fluctuations may disrupt the intricate architecture of DNA condensates and negatively impact the readability of encoded information. As shown in Supplementary Fig. 6, DNA condensates subjected to repeated freeze–thaw cycles at low temperatures remained physically stable, with intact encoded information that was correctly read, maintaining an accuracy of ~98%. This result demonstrates the excellent freeze–resistance of the DNA condensates. However, exposure to high temperatures disrupted the hierarchical spatial organization of the DNA condensates, resulting in encoded information becoming difficult to interpret (Supplementary Fig. 7). For example, using the encoded message “HELLO”, we observed that at room temperature the information could be accurately retrieved. However, following high-temperature treatment, the fluorescence-encoded spatial information within the DNA condensates was disrupted, rendering the stored message unreadable. Remarkably, despite the structural destabilization, the spatial and color information encoded by the m* and n* DNA strands remained essentially intact. By applying our “Erase–and–Diffuse” replication strategy, the damaged DNA condensates successfully transferred their preserved information to “Null” DNA condensates (Fig. 3c, d). This transfer allowed the originally unreadable information to be fully reconstructed and restored the decoding accuracy to 99% (Supplementary Fig. 8). Overall, this process highlights the exceptional robustness of our system in preserving and recovering encoded information under stressful conditions. Dynamic interconversion and encryption of information states in DNA condensate systems Programmable interconversion between distinct information states is crucial for encryption, yet current technologies are limited by the diversity of encoded states and control over their transitions. Next, we develop a system that enables unprecedented flexibility in information state interconversion by performing parallel addition and subtraction molecular computation on the distinct spatial hierarchical structures of DNA condensates. As a proof of concept, we specifically designed two distinct molecular regulatory mechanisms: a photo-controlled switching mechanism utilizing photosensitive PC-linkers and a molecular programming mechanism based on DNA strand displacement reactions. As shown in Supplementary Fig. 9, both mechanisms enable precise molecular addition and subtraction computation of fluorescent signals at various spatial hierarchies of the DNA condensates. Using the encoding of the letter “H” as an example, we demonstrate that UV-triggered cleavage of PC-linkers or DNA strand displacement can selectively release or incorporate fluorescent/quencher-labeled DNA strands. This precise molecular computation enables programmable, direct transitions between information states via spatially controlled color addition and subtraction. Importantly, the design is modular and general: each mechanism requires only 12 fundamental DNA circuit elements (3 fluorophores × 2 localization of shell or core × 2 operations of addition or subtraction), and their combinations cover all pairwise interconversions among the 27 encoded states, thereby eliminating the need for bespoke circuits. Moreover, hierarchical spatial and color regulation supports both simple arithmetic and one-step direct color transitions (Supplementary Fig. 10), enhancing operational efficiency and information processing. Building on this, we develop a programmable encryption strategy designed to mislead unauthorized users (Supplementary Fig. 11a). In this example, the encoded true message “NO” is concealed within the DNA condensate system, while only a decoy message “YES” is externally visible. Even if adversaries gain access to the system, they can only recover the misleading information “YES”. The correct “key” (demonstrated here as either optical stimuli or nucleic acid molecules) activates the hierarchical structure-specific molecular computation within the DNA condensates. Coordinated addition and subtraction computation across different hierarchical structures of the condensates precisely converts the false information “YES” into the true message “NO” (Supplementary Fig. 11b, c). By combining programmable molecular computation with hierarchical coding strategies, the DNA condensate system offers a transformative approach to achieving secure, highly dynamic, and versatile information encryption. Logic gate-driven DNA condensate information decryption system Molecular logic gates, which regulate chemical signals, optical feedback, and dynamic interactions, condense logic operations to the molecular scale, paving the way for advancements in information processing and encryption31–34. Further, we leveraged the high compatibility and programmability of DNA condensates with DNA-based molecular logic circuits to develop a programmable encryption system driven by molecular logic gates. As shown in Fig. 4a, we used an AND logic gate as a proof of concept to demonstrate the coupling between logic computation and DNA condensate-based information encryption. In the initial state, the true message “NO” is concealed within three different DNA condensates, each individually encrypting the characters “N”, “O” or “Null” as “Y”, “E” or “S”, so that only the decoy message “YES” is visibly accessible. Upon the simultaneous input of two keys—an optical stimulus and a specific nucleic acid molecule—the cooperative interaction between these inputs activates multiple dynamic DNA arithmetic reactions within different regions of the DNA condensate, including hierarchical addition and subtraction computation performed in parallel. These coordinated molecular operations trigger the conversion of the decoy message “YES” into the true message “NO”. Conversely, when only one key or no key is supplied, the system retains its default decoy state, displaying the preset false message “YES” (Fig. 4b, Supplementary Fig. 12 and Supplementary Fig. 13). We also implemented an AND-gate encryption in which decryption is triggered only when two distinct ssDNA keys are present (Supplementary Fig. 14). Collectively, these features ensure precise logical encryption in response to specific input conditions. Logic gate and life system-driven DNA condensate information decryption system. a Principle of the programmable encryption system based on an AND logic gate. Information decryption requires the simultaneous presence of two keys. Only upon UV-triggered cleavage of the PC-Linkers can ssDNA undergo strand displacement reaction to selectively release fluorescent or quencher-labeled DNA strands. This reaction enables color subtraction or addition at distinct spatial levels within the DNA condensates. Consequently, under these precise conditions, the system achieves transitions among arbitrarily encoded information states. b CLSM images demonstrating the implementation of the multi-level encryption system driven by the AND logic gate. c Principle of the DNA condensate decryption system driven by dynamic biological interactions. In this system, natural cells serve as inputs, releasing decryption keys to seamlessly unlock the information concealed within DNA condensates. d CLSM images demonstrating the implementation of DNA condensate decryption system driven by dynamic biological interactions. In the presence of PTK7-negative cells, which lack the capability to produce the required decryption key, the genuine encoded information remains concealed. In contrast, PTK7-positive cells generate the decryption key, enabling the disclosure of the encoded information “CEM”. Scale bar: 5 μm. Red, green, and blue in all CLSM images represent Cy5, Cy3, and FAM fluorescence, respectively. All CLSM images shown are representative of three independent experiments. Living system-driven DNA condensate information decryption system The integration of information systems with living systems is pivotal for the convergence of information technology and life sciences. Traditional encryption methods generally lack the capacity to interact with dynamic biological signals, limiting their applicability in life-driven processes35. Thus, we present a DNA condensate-based encryption platform that combines molecular computation, dynamic biosensing, and information encryption, enabling interactive decryption triggered by cellular signals. As illustrated in Fig. 4c, this platform combines aptamer-driven DNA molecular circuits with the DNA condensate system to establish a highly specific, living cell-triggered dynamic decryption mechanism. The aptamer Sgc8c, renowned for its exceptional specificity and affinity for membrane biomarkers on target cells such as CEM cells, functions as a molecular receptor36–38. Upon recognizing and binding to its target biomarker on the cell surface, Sgc8c undergoes a conformational transformation, releasing a complementary nucleic acid key—an essential element of the encryption–decryption process. This key seamlessly integrates with the hierarchical encryption network within the DNA condensates, triggering a cascaded decryption process and selectively revealing the embedded encoded information, such as the word “CEM”. In non-target cell environments where the requisite surface biomarkers are absent, the aptamer remains in its original conformation and no nucleic acid key is generated (Fig. 4d). This ensures that the encrypted information remains securely encapsulated within the DNA condensate, preventing unauthorized access and maintaining robust data security. Time-driven DNA condensate systems for dynamic information encryption Time-driven dynamic information encryption is an emerging high-security strategy characterized by the ability to make encrypted information accessible only within a specific time window, thereby significantly reducing the risk of information leakage39,40. Next, we proposed and demonstrated two types of time-driven DNA condensate systems for dynamic information encryption: transient multi-read encryption and self-destructive encryption. For transient multi-read encryption, this strategy facilitates time-resolved multi-read decryption for transient encryption (Fig. 5a, b). Experimental validation (Supplementary Fig. 15) demonstrates the successful execution of three consecutive rounds of information retrieval from DNA condensates under in vitro conditions. The addition of the decryption key transiently activated the fluorescence signal in a time-dependent manner, followed by its automatic erasure within ~1 h. Crucially, fluorescence intensity remained stable across three successive cycles, highlighting the robustness and reliability of this system in achieving transient, time-controlled multi-read decryption. Inspired by these results, we further encoded the information “CHINA” as a proof of concept. The experiments demonstrated that upon the addition of a key, the types of fluorescent signals and their spatial distribution within the DNA condensates representing different information states gradually emerged, allowing the encoded information “CHINA” to be clearly and accurately decoded. However, over time, the key was gradually depleted, leading to the erasure of the fluorescent signal types and their spatial information within the DNA condensates. Consequently, the information was automatically erased and rehidden. When a new key was introduced, the encoded information “CHINA” could be decoded again and subsequently erased (Fig. 5c). Through this repetitive cycle, we successfully demonstrated the system’s capability for time-driven, transient, and multiple-read dynamic encryption of information within DNA condensates. Dissipative reaction-driven time-dependent encryption. a Schematic illustration of transient multi-read encryption. After decryption, the information gradually re-encrypts itself, making it accessible for only a limited period. This decryption–auto-encryption cycle can be repeated multiple times. b Principle of transient multi-read encryption. Specifically, the fluorescence of DNA condensates is initially quenched, keeping the information in an encrypted state. Upon addition of the key, a strand displacement reaction releases the quencher strand from the condensates, restoring fluorescence and enabling decryption. However, this decryption process is reversible and time-dependent: RNase H specifically degrades the key strand within the key–quencher duplex, allowing the quencher strand to rehybridize with the condensates. As a result, the fluorescence signal gradually fades and the information is automatically re-encrypted. The information reading process can be reactivated by adding the key again, enabling transient multi-read decryption. c CLSM images showing the fluorescence state changes of DNA condensates during three rounds of transient multi-read encryption. d Schematic illustration of self-destructive encryption. After decryption, the information is gradually and irreversibly self-destructed, rendering it readable for only a limited time. Once erased, the information cannot be decrypted again. e Principle of self-destructive encryption. Specifically, the addition of the key triggers a toehold-mediated strand displacement reaction, releasing the quencher strand from the DNA condensate, restoring the fluorescence signal, and decrypting the information. However, unlike transient multi-read encryption, RNase H specifically cleaves the toehold region of the duplex formed between the key strand and the quencher strand. On one hand, this cleavage disrupts the duplex, allowing the quencher strand to rehybridize with the DNA condensates, which causes the fluorescence signal to disappear and renders the encoded information unreadable. On the other hand, the permanent damage to the quencher strand’s toehold region prevents subsequent strand displacement reactions, even if additional key strands are introduced. f CLSM images showing the fluorescence changes of DNA condensates during the self-destructive encryption process. Scale bar: 5 μm. Red, green, and blue in all CLSM images represent Cy5, Cy3, and FAM fluorescence, respectively. All CLSM images shown are representative of three independent experiments. Compared to transient multi-read encryption, self-destructive encryption further enhances information security by enabling single-read decryption, ensuring that the information is automatically destroyed after decryption (Fig. 5d, e). First, as shown in Supplementary Fig. 16, we successfully demonstrated self-destructive encryption of DNA condensate-encoded information under in vitro conditions. The first addition of the key transiently activated the fluorescence signal in a time-dependent manner, followed by rapid signal destruction within ~30 min. Importantly, even with subsequent additions of the key, the information could not be re-read. Similarly, we encoded the information “CHINA” as a proof of concept. Experimental results demonstrated that upon the addition of a key, the types of fluorescent signals and their spatial distribution within the DNA condensates representing different information states gradually emerged, enabling the encoded information “CHINA” to be clearly and accurately decoded. However, over time, the key was gradually degraded, leading to the erasure of the fluorescent signal within the DNA condensates, and consequently, the automatic clearance of the encoded information. Moreover, this process resulted in permanent damage to the toehold region of the quencher strand. As a result, when a new key was introduced, the original key-triggered information decoding reaction could no longer occur, rendering the encoded information “CHINA” unreadable (Fig. 5f). Through this experiment, we demonstrated the time-dependent and irreversible nature of this system in single-read decryption, further highlighting its potential application in the field of dynamic encryption. Fine-grained DNA condensate information encryption systems Fine-grained encryption enables precise and dynamic control over data access by tailoring decryption privileges to specific users, devices, or contextual conditions. Its multi-dimensional, hierarchical data stratification fortifies security, mitigates unauthorized access and data breaches, and establishes a resilient framework for managing and safeguarding information in complex environments41. Therefore, we further develop a molecular-level fine-grained encryption system using DNA condensates as the central carrier. This system allows for the encryption of information at distinct hierarchical levels, with decryption privileges defined by a layered design that associates specific “keys” with corresponding access permissions. Only keys matching the designated privilege level are capable of decrypting and accessing certain levels of information, while higher-tier encrypted data remains inaccessible without the appropriate authorization. As shown in Fig. 6b, the data is divided into three hierarchical levels and encoded using DNA condensates. For instance, the first level encodes the information “CELL”, the second level encodes “B” and the third level encodes “MEMORY”. Users are assigned decryption keys corresponding to their access permissions. Higher-level keys not only allow users to decrypt information corresponding to their own access level but also include all elements of lower-level keys, thereby enabling access to all lower-level information. Specifically, as illustrated in Fig. 6b, a user with a first-level key can only decrypt “CELL” and cannot access higher-level information such as “B” or “MEMORY”. A user with a second-level key can decrypt both “B” and the lower-level information “CELL”, but cannot access the top-level information “MEMORY”. Conversely, a user with the highest-level key can seamlessly decrypt all information, achieving full access. This innovative approach significantly enhances the flexibility and security of encrypted data and opens new avenues for hierarchical information storage and management. Fine-grained encryption for precise information access control. a Principle of fine-grained information encryption. The decryption keys are ssDNA that can initiate a strand displacement reaction to release the quencher-labeled DNA from the DNA condensates encoding the corresponding information level, thereby enabling information decryption. High-level keys contain the sequences of all lower-level keys, allowing access to information at all levels. In contrast, lower-level keys lack the sequences of higher-level keys and cannot decrypt information that requires higher permissions. b CLSM images demonstrating the decryption results at different permission levels. Scale bar: 5 μm. Red, green, and blue in all CLSM images represent Cy5, Cy3, and FAM fluorescence, respectively. All CLSM images shown are representative of three independent experiments. Discussion In conclusion, this work establishes a transformative framework for engineering hierarchical molecular assemblies capable of performing complex information processing tasks. By harnessing the power of dynamic DNA nanotechnology, we realize key functionalities such as information encoding, erasure, rewriting, replication, and repair, thereby endowing DNA condensates with unprecedented programmability and encryption capabilities. This platform integrates sophisticated molecular computation mechanisms, enabling precise control over seamless interconversion between multiple encoded states. By overcoming critical limitations of current DNA-based systems, such as restricted editability and scalability, our approach sets a new standard for secure, programmable molecular information systems. Beyond addressing these fundamental challenges, the platform introduces advanced encryption modalities, including multi-layered logic encryption, time-dependent dynamic encryption, cell signal-responsive encryption, and fine-grained access control. These innovations significantly enhance the flexibility, security, and operational versatility of molecular information systems (see Supplementary Note 1 for detailed comparison of different DNA information systems). Regarding information density and scalability, in the proof-of-concept stage each character was stored in an individual well of a 384-well plate, yielding an information density of ~0.051 bits/mm³—a constraint dictated by experimental convenience rather than intrinsic system capability. Recent advances in microfabrication and microfluidics now enable high-density chips and droplet-based microarrays with thousands to millions of individually addressable compartments, far smaller than conventional well plates42,43. Coupling DNA condensate platform with such architectures could elevate storage density to ~7.93 × 109 bits/mm³, while preserving spatial ordering and enabling automated, high-throughput molecular information processing. Although this density remains below that of DNA sequence-based storage (8 × 1013 bits/mm³), it is already comparable to that of hard drive44 (see Supplementary Note 2 for detailed information density calculations). In addition, information density could likewise be increased by storing multiple condensates within a shared solution. In small-scale mixtures comprising only a few condensate types, no appreciable circuit cross-talk was observed (Supplementary Fig. 17). However, to ensure precise addressability in more complex mixtures, orthogonal sequence barcodes could be assigned to different condensates, thereby allowing selective manipulation of specific characters. Collectively, these two strategies would substantially enhance the system’s scalability and practical utility. In terms of encryption strength, the platform combines fluorescence-spatial multiplexing with a fully reconfigurable code-to-letter mapping, exponentially enlarging the encoding space. In addition, modular molecular operations—combinations of color addition and subtraction—enable programmable conversion between any characters. Together, these features effectively preclude brute-force decoding attempts based on enumerating fluorescence patterns. The system also supports multiple decryption pathways and logic gate integration, further expanding the encryption space. Information can only be decoded when all keys are input, thereby significantly enhancing security. Nonetheless, the platform’s security is relative rather than absolute, as advanced molecular analyses could still reveal the composition of the information carriers. Overall, the mechanism is systematic, versatile, and highly customizable, enabling a wide range of advanced and adaptable encryption protocols. Looking forward, the potential applications of this platform are vast and transformative. In data security, its dynamic encryption capabilities, including transient multi-read and self-destructive encryption, could enhance secure communication systems by mitigating residual data risks and preventing unauthorized access. Furthermore, the platform’s capacity to interact dynamically with biological signals positions it as a powerful tool for synthetic biology and biocomputing, where real-time, precise control over biological information processing is crucial. By bridging the gap between programmable molecular systems and real-world applications, this work lays the foundation for a new paradigm in molecular information science, with far-reaching implications across data security, biomedicine and synthetic biology. Methods Materials T4 DNA Ligase (400,000 U/mL), Exonuclease I (20,000 U/mL), Exonuclease III (100,000 U/mL), φ29 DNA Polymerase (10,000 U/mL), Inorganic Pyrophosphatase (100 U/mL), RNase H (5,000 U/mL) and dNTP (100 mM) solution were purchased from New England Biolabs (NEB). The 1× TE buffer (10 mmol/L Tris-HCl; 1 mmol/L EDTA, pH 7.8–8.2), magnesium acetate tetrahydrate (MgAc₂), and 5 M NaCl were purchased from Sangon Biotech. Dulbecco’s phosphate-buffered saline (1× DPBS) was purchased from Biosharp, and RPMI Medium 1640 was obtained from Gibco. Ultrapure water (resistivity of 18.2 MΩ·cm) was produced using a Genie Water System (Rephile). All DNA were synthesized and purified by high-performance liquid chromatography (HPLC) by Sangon (China). RNA and DNA–RNA chimeric oligonucleotides were synthesized and HPLC-purified by Accurate (China). The complete list of nucleic acid sequences is provided in the Supplementary Table. All oligonucleotides were suspended at 20 μM in 1× TE buffer and stored at 4 °C or −20 °C until use. Cell Culture The CCRF-CEM (CL-0058) and Ramos (CL-0483) cell lines were obtained from Pricella (China) and maintained at 37 °C in a humidified incubator with 5% CO₂ to ensure optimal growth conditions. All cell lines were cultured in RPMI-1640 basal medium (Gibco) supplemented with 10% fetal bovine serum (FBS, Gibco) and 1% penicillin-streptomycin (Gibco) to provide essential nutrients and prevent microbial contamination. Prior to aptamer incubation, cells were rinsed with 1× DPBS to remove residual media and ensure consistency in subsequent binding assays. For cell density determination, each sample was measured twice and the results were averaged. For each measurement, 10 μL of the cell suspension was loaded onto Cell Counting Chamber Slides (Thermo Fisher) and counted using a cell counter (Invitrogen Countess3). Confocal Imaging Prior to Confocal Laser Scanning Microscopy (CLSM) imaging, the DNA condensate solution was transferred to a confocal culture dish (Cellvis) or a 384-well plate (Cellvis) and allowed to settle naturally for a defined period. The DNA condensate concentration was adjusted to 150–200 nM (based on barcode n) to achieve optimal density for confocal imaging. Imaging was conducted on a NIKON A1 HD25 single-photon laser confocal microscope with a 100× objective lens. Image acquisition was carried out with NIS-Elements AR software (version 5.30.02) controlling the confocal microscope. Laser excitation and emission settings were as follows: DAPI (405 nm excitation, 425–475 nm emission), FAM (488 nm excitation, 500–550 nm emission), Cy3 (561 nm excitation, 570–620 nm emission), and Cy5 (640 nm excitation, 663–738 nm emission). Pseudo-color assignments were as follows: for three-color imaging, FAM (blue), Cy3 (green), and Cy5 (red); for four-color imaging, DAPI (blue), FAM (green), Cy3 (yellow), and Cy5 (red). During image acquisition, sequential channel scanning was used to minimize crosstalk between different channels. A low laser power was employed to reduce photobleaching, and the imaging quality was further optimized by adjusting the detector sensitivity (HV). For each experiment, several regions were imaged, and representative results are presented. Image processing and analysis were performed using NIS-Elements Viewer (version 5.21.00) or ImageJ (version 1.54 f). Fluorescence kinetics monitoring The fluorescence intensity was monitored over time using a Spark Multimode Microplate Reader (Tecan) to evaluate the reaction kinetics. Fluorescence kinetics monitoring was conducted in a white 96-well microplate (biosharp), which minimizes background fluorescence and ensures consistent optical properties across wells. For Cy5 detection, the excitation wavelength was set to 626 nm, and the emission wavelength was collected at 669 nm, which corresponds to the optimal fluorescence signal of Cy5. The kinetic measurements were performed at room temperature (~22–25 °C), with fluorescence data collected every 15 s to capture dynamic changes in real time. Prior to initiating the reaction, the system was pre-equilibrated for 5–10 min to establish a stable fluorescence baseline and minimize signal fluctuations. The resulting fluorescence kinetic curves were plotted and analyzed using Origin 2021. General procedure for DNA condensate preparation The nucleic acid sequences used in this section are listed in Supplementary Table 3. The DNA condensate preparation method was based on previously reported protocols, with modifications to optimize reaction conditions and improve yield15,19. (1) Synthesis of circular ssDNA template. The circular ssDNA template was synthesized by annealing a 5’-phosphorylated ssDNA template with its corresponding linker strand at an equimolar ratio, with a final concentration of 1 μM in 100 μL TE buffer. The solution was heated to 85 °C for 5 min and gradually cooled to 20 °C at 0.01 °C/s in a thermal cycler (Bio-Rad) for precise DNA strand annealing. After annealing, 20 μL of 10× T4 DNA ligase reaction buffer (NEB), 79 μL of ultrapure water, and 1 μL of T4 DNA ligase (NEB, 400 U/μL) were added. The mixture was stirred for 10 min and incubated at room temperature for 4 h. The ligase was inactivated by heating at 70 °C for 20 min to terminate the ligation reaction. To remove unligated linear DNA, 20 μL of exonuclease I (NEB, 20 U/μL) and 20 μL of exonuclease III (NEB, 100 U/μL) were added to the reaction mixture and incubated overnight at 37 °C with stirring (300 rpm). The exonucleases were deactivated by heating at 80 °C for 40 min to prevent further degradation of DNA. The circular ssDNA template was purified using an Amicon Ultra centrifugal filter (Merck Millipore, 10 kDa molecular weight cutoff) and washed three times with TE buffer. The final solution was diluted to 1 μM in TE buffer. (2) Rolling circle amplification for multiblock ssDNA polymer synthesis. A reaction mixture containing 10 μL of 1 μM circular template, 134 μL of water, 20 μL of 10× φ29 DNA polymerase reaction buffer (NEB; 500 mM Tris-HCl, 100 mM (NH₄)₂SO₄, 40 mM DTT, 100 mM MgCl₂, pH 7.5), 2 μL of exonuclease-resistant primer (10 μM, TE buffer), 4 μL of φ29 DNA polymerase (NEB, 10 U/μL), 20 μL of pyrophosphatase (NEB, 0.1 U/μL), and 10 μL of dNTP mixture (NEB; total dNTP concentration of 100 mM, with each nucleotide ratio matching the sequence composition of the target ssDNA polymer) was prepared. The reaction mixture was incubated at 30 °C with constant stirring (300 rpm) for 60 h to ensure efficient polymerization. The resulting polymer was purified using an Amicon Ultra centrifugal filter (Merck Millipore, 30 kDa molecular weight cutoff) and washed three times with 400 μL of TE buffer. The ssDNA concentration was quantified using UV-Vis spectroscopy (Nano-400A Microvolume Spectrophotometer, China). (3) Synthesis of DNA condensates. To prepare DNA condensates, multiblock ssDNA polymers were mixed in TE buffer at target concentrations (typically, p(A20-m) = 0.2 g/L, corresponding to m = 10 μM; p(T20-n) = 0.04 g/L, corresponding to n = 2.5 μM; with a molar ratio of p(A20-m) to p(T20-n) of 4:1). The mixture was heated to 95 °C for 15–20 min to fragment long ssDNA polymers into smaller units for subsequent assembly. Then, MgAc₂ solution (0.5 M) was added to a final concentration of 50 mM, followed by heating to 95 °C for 5 min and cooling to room temperature at a ramp rate of 3 °C/s. All DNA condensates were stored in TE buffer containing 100 mM NaCl. Spatial-fluorescence dual encoding of DNA condensates Following the character encoding scheme, an equimolar amount (1:1 molar ratio) of the corresponding fluorophore-labeled m* or n* was added to the DNA condensate suspension (the sequences are shown in Supplementary Table 3). After brief vortex mixing, the mixture was incubated at room temperature with gentle agitation for 1 h to prevent precipitation of DNA condensates and ensure efficient hybridization with the fluorophore-labeled sequences for accurate information encoding. To construct specific information sequences, DNA condensates encoding distinct characters were spatially allocated to designated wells of a 384-well confocal plate (Cellvis) according to the desired target sequence. Finally, CLSM imaging was conducted to visualize the spatial arrangement and fluorescence signals of the DNA condensates, allowing accurate information reading and decoding. Information erasure and rewriting in DNA condensate system Information encoding was accomplished by hybridizing DNA condensates with the DNA sequences listed in Supplementary Table 4. Equimolar amounts of fluorophore-labeled barcode sequences were added to the DNA condensate suspension. Information rewriting began with UV irradiation at 300 nm for 1 h to completely remove residual fluorescence signals from the DNA condensates, effectively erasing the previously encoded information. Following fluorescence erasure, an equimolar amount of new fluorophore-labeled barcode sequences were hybridized into the DNA condensates according to the defined encoding scheme. The mixture was subsequently agitated gently at room temperature for 1 h to ensure efficient hybridization and achieve accurate re-encoding. The efficiency of encoding, erasure, and rewriting was evaluated through CLSM, with uniform imaging parameters maintained across all experiments. Information rewriting experiments were conducted in confocal culture dishes (Cellvis) with a reaction volume of 200 μL and a DNA condensate concentration of 177 nM, calculated based on the molarity of barcode n. Information replication in DNA condensate systems The replication process was initiated by hybridizing DNA condensates (150 nM, calculated based on barcode n) with the corresponding DNA sequences listed in Supplementary Table 5 at an equimolar ratio (m*:m = 1, n*:n = 1), encoding the parental information. To enable replication, Copy instruction strands X and Y were added to the parent DNA condensates at a total final concentration of 750 nM. The mixture underwent brief vortex mixing and was incubated at room temperature with gentle shaking for 2 h, resulting in the dissociation of fluorescent barcodes from the parent DNA condensates. Subsequently, the mixture was centrifuged at 14,000 rpm for 5 min to separate the erased parent DNA condensates as the pellet from the dissociated fluorescent barcodes in the supernatant. The collected supernatant was evenly divided into three portions: two portions were combined with unencoded DNA condensates, while the remaining portion was retained with the erased parent DNA condensates. All three mixtures were subjected to UV irradiation (300 nm) for 15 min, followed by incubation at room temperature for an additional hour. This treatment cleaved the replication initiator ssDNA, releasing the fluorescent barcodes, which subsequently hybridized with the DNA condensates to complete the first round of replication. This process generated two identical DNA condensate copies while simultaneously restoring the parental information. To perform a second round of replication, the same procedure was repeated using one of the newly generated copies as the template. Evaluation of the stability of encoded DNA condensates under extreme conditions To assess the stability of DNA condensate-encoded information under extreme environmental conditions, DNA condensates were hybridized with equimolar amounts of the corresponding sequences listed in Supplementary Table 5. The encoded DNA condensates were subsequently stored in TE buffer supplemented with 100 mM NaCl. To simulate low-temperature stress, the encoded DNA condensates were subjected to freeze-thaw cycles by freezing at −80 °C for 30 min, followed by thawing at room temperature for 15 min. This process was repeated three times to mimic repeated exposure to freezing conditions. To evaluate stability under dehydration, the encoded DNA condensates were allowed to air dry at room temperature until the solvent completely evaporated. To simulate high-temperature stress, the encoded DNA condensates were heated to 95 °C for 10 min, followed by gradual cooling to room temperature. These experiments collectively assessed the resilience of encoded DNA condensates to extreme thermal and environmental stressors. Information repair in DNA condensate systems To repair heat-damaged DNA condensates, 25 μL of heat-treated DNA condensates (containing 4.45 pmol of barcode n) was combined with a total of 22.25 pmol of Copy instruction strands X and Y, corresponding to an equimolar amount of barcodes n and m. The reaction mixture was subjected to brief vortex mixing and incubated at room temperature for 1 h, allowing the repair initiator strands to bind to and displace the fluorescent barcodes on the damaged DNA condensates. Subsequently, 10 μL of unencoded DNA condensate solution (containing 4.45 pmol of barcode n) was added to the reaction mixture. The sample was then irradiated with 300 nm UV light for 15 min to cleave the repair initiator strands and release the fluorescent barcodes. Thereafter, the reaction was incubated at room temperature for an additional hour, enabling the released fluorescent barcodes to hybridize with the intact DNA condensates, thereby restoring the encoded information. Dynamic interconversion of DNA condensate information induced by light and DNA To encode information responsive to light, DNA condensates were hybridized with the sequences listed in Supplementary Table 6. The encoded DNA condensates were then irradiated with UV light (300 nm) for 1 h, resulting in the cleavage of the PC linker within the sequences and subsequently altering the encoded information. For encoding information responsive to DNA, DNA condensates were first hybridized with the sequences provided in Supplementary Table 7 or Supplementary Table 8. Following this, 1 equivalent of Key (DNA strand) was introduced into the encoded DNA condensates. The mixture was vortexed briefly, followed by incubation at room temperature for 2 h, inducing changes in the encoded information. To encode information that responds to a combination of light and molecular stimuli, DNA condensates were hybridized with the sequences listed in Supplementary Table 8. The encoded DNA condensates were initially subjected to UV irradiation for 1 h, after which 1 equivalent of Key2 (DNA strand ab) was added. The mixture was subsequently incubated at room temperature for 2 h, ultimately producing changes in the encoded information. Living system-driven DNA condensate information decryption system To implement biological encryption, DNA condensates were assembled with the corresponding sequences listed in Supplementary Table 9. For information decryption, Sgc8c and Key were annealed at a 1:1 molar ratio in TE buffer (containing 100 mM NaCl) to yield a 5 μM Sgc8c/Key complex. The resulting complex was subsequently mixed with encrypted DNA condensates (containing 4.8 pmol of barcode n) while maintaining a Key-to-n molar ratio of 1:1. Washed CEM or Ramos cells were then added to the mixture to a final count of 4 × 104 cells. Glucose and MgAc₂ were added, and the reaction volume was adjusted to 25 μL with TE buffer (100 mM NaCl), yielding final concentrations of 192 nM Sgc8c/Key, 5 mM MgAc₂, and 4.5 g/L glucose. The samples were incubated for 2 h at room temperature to allow Key release and strand-displacement–mediated decryption. The decrypted state was subsequently examined using CLSM. Time-driven dynamic DNA condensate information encryption systems (1) To construct a system capable of transient multi-read encryption, the corresponding sequences from Supplementary Table 10 were first annealed and incubated with DNA condensates, encoding specific information. The encoded DNA condensates (final n concentration of 150 nM) were subsequently mixed with 10× RNase H Reaction Buffer (adjusted to a final concentration of 1×) and RNase H (final concentration of either 75 U/mL or 120 U/mL) in TE buffer supplemented with 100 mM NaCl, thereby assembling the transient multi-read encryption system. To decrypt the encoded information, 2 equivalents of the key sequence were introduced into the system, and CLSM was employed to capture the decrypted state. After a sufficient period of time (~6 h) to allow for complete information erasure, the system was re-examined using CLSM to confirm the transient nature of the encryption. The decrypt–auto-erase cycle was repeated three times. (2) For the construction of a self-destructive encryption system, the corresponding sequences from Supplementary Table 10 were similarly annealed and incubated with DNA condensates to encode specific information. The encoded DNA condensates (final n concentration of 160 nM) were then mixed with 10× RNase H Reaction Buffer (final concentration of 1×) and RNase H (final concentration of either 75 U/mL or 200 U/mL) in TE buffer containing 100 mM NaCl, completing the assembly of the self-destructive encryption system. For each decryption process, 2 equivalents of the key sequence were added to the system to decrypt the information, and CLSM was used to record the decoded state. After allowing sufficient time (~6 h) for the automatic degradation of the encoded information, CLSM was employed again to capture the final self-destructed state of the system. Fine-grained DNA condensate information encryption The corresponding sequences from Supplementary Table 11 were annealed and subsequently hybridized with DNA condensates to achieve fine-grained encryption of distinct information. To decrypt the encrypted information, an equimolar amount of the key (final concentration of 900 nM) was added to the DNA condensate solution, where the concentrations of barcode n and barcode m were 180 nM and 720 nM, respectively. The reaction mixture was vortexed briefly to ensure thorough mixing and incubated at room temperature for 4 h, allowing the strand displacement reaction to reach completion and thereby successfully decrypting the encoded information. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Reporting Summary Transparent Peer Review file Source data Source Data Acknowledgements This work was supported by the National Key R&D Program of China (2021YFA0909400 to D.H. and L.H.), the Yangtze River Delta Science and Technology Innovation Community Joint Basic Research Project (2024CSJZN1200 to L.H.), the National Natural Science Foundation of China (22225402 to D.H., 22474127 to L.H., and 32341017 to D.H.), and the Zhejiang Provincial Jianbing Lingyan Research and Development Project (2023SDYXS0002 to D.H., L.H., and P.G.). The authors also acknowledge support from the Scientific Experiment Center, Hangzhou Institute of Medicine (HIM), Chinese Academy of Sciences. Author contributions D.H., L.H., and L.C. conceived the study. L.C. planned and performed the most wet-lab experiments with guidance from D.H. and L.H. L.W., H.W., L.D., and Y.G. performed cell-related measurements. P.W., Y.G., L.D. and P.G. fabricated and optimized the synthesis of DNA condensates. D.H., L.H. and L.C. wrote the paper with input from all authors. Peer review Peer review information Nature Communications thanks Sung Ha Park, Albert Markvoort and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available. Data availability All data supporting the findings in this study are provided in the main article, Supplementary Information or Source Data file. The full image dataset is available from the corresponding author upon request. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Lei He, Email: helei@hnu.edu.cn. Da Han, Email: dahan@sjtu.edu.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-025-67093-w. References Associated Data Supplementary Materials Supplementary Information Reporting Summary Transparent Peer Review file Source Data Data Availability Statement All data supporting the findings in this study are provided in the main article, Supplementary Information or Source Data file. The full image dataset is available from the corresponding author upon request. Source data are provided with this paper.