Microenvironment-confined kinetic elucidation and implementation of a DNA nano-phage with a shielded internal computing layer

Tang, Decui; He, Shuoyao; Yang, Yani; Zeng, Yuqi; Xiong, Mengyi; Ding, Ding; Wei, Weijun; Lyu, Yifan; Zhang, Xiao-Bing; Tan, Weihong

doi:10.1038/s41467-025-56219-9

Download PDF

Article
Open access
Published: 22 January 2025

Microenvironment-confined kinetic elucidation and implementation of a DNA nano-phage with a shielded internal computing layer

Nature Communications volume 16, Article number: 923 (2025) Cite this article

5554 Accesses
4 Citations
1 Altmetric
Metrics details

Subjects

Abstract

Multiple receptor analysis-based DNA molecular computation has been developed to mitigate the off-target effect caused by nonspecific expression of cell membrane receptors. However, it is quite difficult to involve nanobodies into molecular computation with programmed recognition order because of the “always-on” response mode and the inconvenient molecular programming. Here we propose a spatial segregation-based molecular computing strategy with a shielded internal computing layer termed DNA nano-phage (DNP) to program nanobody into DNA molecular computation and build a series of kinetic models to elucidate the mechanism of microenvironment-confinement. We explain the contradiction between fast molecular diffusion and effective DNA computation using a “diffusion trap” theory and comprehensively overcome the kinetic bottleneck of DNP by determining the rate-limiting step. We predict and verify that identifying trace amount of target cells in complex cell mixtures is an intrinsic merit of microenvironment-confined DNA computation. Finally, we show that DNP can efficiently work in complex human blood samples by shielding the interference of erythrocytes and enhance phagocytosis of macrophages toward target cells by blocking CD47-SIRPα pathway.

Structure sampling for computational estimation of localized DNA interaction rates

Article Open access 16 June 2021

Highly shape- and size-tunable membrane nanopores made with DNA

Article 28 April 2022

Nanopore sequencing of DNA-barcoded probes for highly multiplexed detection of microRNA, proteins and small biomarkers

Article Open access 25 September 2023

Introduction

Cell membrane receptors perform important biological functions by participating in various biological processes like cell cycle^1,2, proliferation^3,4,5, communication^{6,7,8,9,10,11}, and migration^12,13,14, and thus are considered as valuable targets for regulating physiological and pathological states. Numerous targeting strategies have been developed by adopting specific binding ligands as the recognition moieties to bind membrane receptors and have found wide applications in realizing real-time molecular imaging^15,16,17,18, targeted drug delivery^19,20,21,22, specific protein degradation^{23,24,25,26,27}, dynamic biomolecule manipulation^16,28,29,30 and enhanced immunoregulation^11,31,32,33. However, in living organisms, cell membrane receptors are commonly shared by multiple cell types or are overexpressed in both diseased cells and normal cells so that single-receptor-targeted strategies may lead to off-target effects and nonspecific cytotoxicity^34,35,36.

DNA molecular computation combining multiple receptor analysis and logic-gated operation offered an effective solution to mitigate the off-target effect of single targets and improve the specificity of cellular recognition caused by nonspecific expression of receptors^37,38,39,40. By using aptamers as recognition modules, the expression of dual or multiple receptors can be converted into DNA secondary signals through a binding-induced strand displacement reaction and passed on to subsequent Boolean operation^{28,37,39,41,42}. However, since a DNA computing process is a multi-step kinetic process including molecular diffusion, DNA dynamic reaction and target binding in cellular microenvironment, so far there is still no comprehensively quantitative model to reveal the molecular mechanism of efficient system operation. Additionally, the involvement of other ligands apart from aptamers, for example, nanobodies (Nbs), into molecular computation will greatly expand the cellular recognition toolbox and promote the performance in sensing, manipulating, and responding to intricate cellular milieus. Nbs, also known as single-domain antibodies, are antibody fragments (variable region of heavy chain of heavy-chain-only antibody, VHH) with many good features like small size (2–4 nm in diameter, about 15 kDa), compact structure, high stability, superior water solubility, easy cloning and the capability of binding to difficult-to-access antigens^43,44. Compared with aptamers that can be easily integrated into DNA computation as signal convertors through structural switchable strategies, it is quite difficult to involve nanobodies as the recognition modules because of the “always-on” response mode and the inconvenient molecular programmability. Moreover, for some functional membrane receptors, binding to ligands would immediately activate downstream signaling pathways, thus disturbing the target system and invalidating the Boolean operation process. Therefore, strict recognition order and ligands biofunction shielding are usually required in multi-layer DNA molecular computation for precise targeting and functional regulation of specific membrane receptors.

To solve these problems, here we propose a spatial segregation-based molecular computing platform with a shielded internal computing layer termed DNA nano-phage (DNP), as an analogy of the structure of T4 bacteriophage in nature, and build series of kinetic models to comprehensively elucidate the molecular mechamism of DNA computation in cellular microenvironment (Fig. 1). With a head-like DNA cage and tail fiber-like external nano-claws, DNP can recognize receptors on target cell surface, convert multi-input bio-signal into DNA signal, execute a multi-layer molecular computation and release the encapsulated Nb-DNA strand complex which is prepared based on a stoichiometric and site-specific Nb labeling strategy. We explain the contradiction between fast molecular diffusion and effective DNA computation using a “diffusion trap” theory, therefore elucidating the mechanism of microenvironment-confinement effect. Through kinetic analysis, we improve the computing speed and signal intensity of DNP by optimizing the rate-limiting step. By considering by-standing cells in our model, we predict and verify that DNP could accurately identify trace amount of target cells in complex cell mixtures, which is also proved to be an intrinsic merit of microenvironment-confined DNA computation. Taking the immune checkpoint protein CD47 that expressed on both tumor cells and normal cells as an expample, we prove that DNP can efficiently work in complex human blood samples by shielding the interference of erythrocytes and enhance the immune-phagocytosis of macrophages by specifically blocking the CD47-SIRPα pathway on target cells without causing off-target effects on “bystander” cells, thus improving the specificity and biosafety of CD47-SIRPα pathway as an immunotherapy target.

Results

The design principle and construction of DNP

As illustrated in Fig. 1a, a DNP consists of two functional moieties, a head-like DNA cage and a tail fiber-like external DNA nano-claw. The computation process of DNP contains three computing layers: 1) input computing layer, 2) shielded internal computing layer, 3) output computing layer (Fig. 1b). During the input computing layer, the DNA nano-claw works as a signal convertor of DNP by equipping with one or more aptamer-based specific recognition toes. The recognition toes are essentially “Yes” gates that can recognize the receptors on cell membrane as the bio-input signals and convert the bio-input signals into downstream DNA signals by releasing complementary strands (c-aptamer) through aptamer binding-mediated strand displacement reaction. The number of recognition toes can be designed flexibly to achieve multiple input or multi-valence input. During shielded internal computing layer, the converted DNA strand c-aptamer can then diffuse into the cage and execute DNA strand displacement-based molecular computation. If the computation result is “true”, then the encapsulated CD47 Nb-DNA strand complex (Nb-S) will be released from the cage. The output computing layer is essentially an “AND” gate, during which the released Nb-S searches for and binds with CD47 on target cell membrane. Since the binding between CD47 nanobody and CD47 on target cell membrane is much stronger than that between SIRPα on macrophage membrane and CD47 on target cell membrane, the CD47-SIRPα pathway way can be effectively blocked to enhance the macrophage phagocytosis of target cells.

We first constructed the DNA cage via hybridization-driven assembly⁴⁵. The length of the edges of DNA cage is about 10 nm (30–32 bp). To encapsulate the CD47 Nb-DNA complex Nb-S, a 71 nt carrier strand was anchored inside the cage by hybridizing with two opposing edges (Supplementary Fig. 1). To prepare Nb-DNA complex Nb-S, we here developed an efficient Nb labeling strategy with a good yield to achieve accurate 1:1 and site-specific DNA modification⁴⁶. As shown in Fig. 1c and Supplementary Fig. 2, NH₂-PEG-N₃ was specifically labeled at N terminal of Nb as a linker by glutamate transaminase, then DBCO-modified DNA strand was reacted with N₃ through click chemistry. In contrast, using glutamate transaminase to directly conjugate Nb with NH₂-modified DNA strand with different spacer length (Supplementary Fig. 3) gave no obvious yield, probably because the negatively charged DNA impeded the approaching of glutamate transaminase. Flow cytometry analysis showed that the DNA modification only slightly affected the binding affinity between Nb and target cell CEM (Supplementary Fig. 4) with a dissociation constant (K_D) of 4.28 nM. We compared the binding affinity between Nb and commercialized monoclonal antibodies B6H12 through binding competing assay and obtained a binding affinity order of Nb > Nb-S > B6H12 (Fig. 1d and Supplementary Fig. 5). These results also suggested that Nb and B6H12 had the same binding site on CD47. The Nb-DNA complex Nb-S was then encapsulated into the DNA cage through hybridization to form cage-Nb. The atomic force microscopy (AFM) images revealed that cage-Nb were monodisperse cage structures (Fig. 1e), further demonstrating the successful preparation of DNPs. Confocal imaging and flow cytometry analysis demonstrated the DNA cage effectively protected the encapsulated Nb from binding with target CEM cells (Fig. 1f and Supplementary Fig. 6) while free Nbs showed obvious binding with target CEM cells. Besides, the concentration-dependent non-specific cell binding was negligible, as demonstrated by flow cytometry (Supplementary Fig. 7). Then the DNPs can be constructed by integrating the cage-Nb and different recognition toe-equipped nano-claws, as will be demonstrated in the following sections.

Implementation of three-layer two-input DNP

Next, we investigated the specific recognition and Boolean operation-based molecular computation of DNP. We chose aptamer sgc8c that specifically bind protein tyrosine kinase 7 (PTK7) to construct the recognition toe in the nano-claw and built a three-layer two-input (PTK7 and CD47) DNP (DNP1) (Fig. 2a). PAGE and DLS were used to characterize the assembly of DNP1 (Supplementary Fig. 8). To monitor the binding of aptamer and Nb, we labeled a TAMRA on sgc8c and a FAM on Nb-S, and labeled corresponding quenchers on the complementary strands. The logic flowchart and truth table of three-layer two-input DNP is shown in Fig. 2b. The slightly slowed release of Nb from DNP was tested by real-time fluorescent kinetics measurement, which is attributed to the inhibition of DNA cage on displacement kinetics (Fig. 2c). Four cell lines, target CEM cell, and nontarget Ramos, K562 and RPMI-8226 cells were utilized to study whether DNP1 could perform accurate target cell identification. The binding pattern between aptamer/Nb and different cell lines is shown in Supplementary Fig. 9. Flow cytometry results (Fig. 2d and Supplementary Fig. 10) shows that target CEM cells exhibited obvious fluorescent intensity in both TAMRA (red) and FAM (green) channels after incubating with DNP1 and showed negligible fluorescence in both channels after incubating with cage-Nb. In contrast, nontarget Ramos, K562 and RPMI-8226 cells showed no obvious fluorescence signal in both channels no matter incubated with cage-Nb or DNP1 (Fig. 2e). Similar results were also confirmed by confocal imaging as shown in Supplementary Figs. 11–14. These results demonstrate that the DNP1 can accurately identify target cells through a three-layer two-input Boolean operation, in which both aptamer and Nb-based molecular recognitions are involved into the molecular computation.

**Fig. 2: Implementation and molecular mechanism of three-layer two-input DNP.**

Molecular mechanism of microenvironment-confined DNA computation

Althouth the successful implementations of DNP1 as well as previous works^37,39,41, the molecular mechanism of microenvironment-confined DNA computations remain obscure due to the complexity of the kinetic process including molecular diffsusion, DNA dynamic reaction and target binding. A schematic diagram of reaction pathways of DNP1-based computation in target cellular microenvironment is shown in Fig. 2f. We first built a kinetic model to describe the molecular diffusion by solving the second-order partial differential equation of Fick’s second law (Supplemenatry Note 1). The time and distance-dependent simulation result shows that the diffusion between cellular microenvironment and surrounding buffer solution is extremely rapid and will finish within 1 s (Fig. 2g). By comparing the diffusion kinetics and the DNA reacton or target binding kinetics, it is clear that the diffusion rate is several orders of magnitude faster than reaction rates (Fig. 2h). If that’s the case, then effective DNA computation should not have happened because the DNA strands should have quickly diffused away from microenvironment into surrounding buffer solution once uppon realeased from DNP, which obviously contradicts the experimental results. Here we propose a theory called “diffusion trap” to explain this contradiction. As shown in Fig. 2i, molecular diffusion is essentially based on Brownian motion, whose trajectory is like a three-dimensional “random walk”. For a DNA strand generated in cellular microenvironment, the random walk will cause continual collisions between the DNA strand and nearby membranes so that hybridization or target binding will definitely happen before the DNA strand diffusing away. Therefore, in a DNP-based computing process, once a DNA strand like c-sgc8c or Nb-S was released near the target cell membrane, there will be no chance for them to diffuse away into surround buffer solution in spite of the fast diffusion rate. To validate our therory, we simulated the diffusion trajectories with or without a difusion boundary using MATLAB codes (Supplementary Note 2). As shown in Supplementary Fig. 15, when a molecule is near the boundary, its diffusion trajectory will expand laterally along the boundary accompanied by a large number of collisions. The “diffusion trap” therory allowed us to elucidate the molecular mechanism of microenvironment-confined DNA computation by computationally establishing a kinetic model based on diffusion kinetics, strand displacement reaction kinetics, and binding kinetics. A set of differential equations following the kinetic profile was formulated and provided in Supplementary Note 3. The set of rate constants acquired by fitting the time-dependent concentration changes of species is given in Supplementary Table 1. By using this set of rate constants, we simulated reaction rate of different layers with or without considering the “diffusion trap” theory. As shown in Fig. 2j, when considering “diffusion trap” theory, the computation can happen with a rational reaction rate, according to the simulation result using the kinetic model. In contrast, extremely low reaction rates in layer 2 and layer 3 were simulated when not considering “diffusion trap” theory. The time-dependent concentration changes of species during DNP1-based molecular computation were computationally estimated as shown in Fig. 2k which corresponded to the experimental data. It should be noted that the our kinetic model is univeral and can be adjusted to describe the kinetic behavior of almost all previously reported microenvironment-confined DNA computing process as implemented in our codes.

Implementation and kinetic simulation of three-layer three-input DNP

The constructed kinetic model can help us to predict the feasibility of more complex DNA computations in cellular microenvironment. Therefore, we next challenge to build a three-layer three-input DNP by using two different aptamers in the input computing layer. A unique advantage of cell-specific aptamers is that aptamers can be screened to specifically bind with target cells without previous knowledge of the target protein. To take advantage of such merits, here we chose sgc4f, which can target unknown overexpressed receptors (marked as Tsgc4f) on several leukemia cells, to construct the second recognition toe on the nano-claw of the three-layer three-input DNP (DNP2) (Fig. 3a). PAGE and DLS were used to characterize the assembly of DNP2 (Supplementary Fig. 16). The logic flowchart and truth table of DNP2 is shown in Fig. 3b and the reaction pathways are shown in Fig. 3c. The kinetic model to formulate the DNP2-based computing process was computationally established in Supplementary Note 4 using a set of differential equations following the kinetic profile and a set of fitted rate constants in Supplementary Table S2. The time-dependent concentration changes of species were computationally simulated as shown in Fig. 3d, indicating the feasibility of DNP2-based three-layer three-input computation. To experimentally investigate the computing performance of DNP2, we labeled a Cy5 on sgc4f, a TAMRA on sgc8c and a FAM on Nb-S and labeled corresponding quenchers on complementary strands. The release of Nb-S from DNP2 was tested by real-time fluorescent kinetics measurement (Fig. 3e). Five different cell lines, including target CEM cells and nontarget Ramos, K562, RPMI-8226 and Jurkat cells were utilized to study the specific recognition ability of DNP2. The binding pattern between aptamer/Nb and different cell lines was shown in Supplementary Fig. 17. Then different cell lines were incubated with cage-Nb, cage-Nb-sgc4f, cage-Nb-sgc8c and DNP2, respectively and analyzed by flow cytometry. For target CEM cells incubated with DNP2, two “Yes” gates in the input layer were activated, and the released c-sgc8c and c-sgc4f acted as the input signal to activate the “AND” gate in the internal computing layer. FAM-labeled Nb-S was then released to bind CD47 on cell membrane. As a result, fluorescence was observed in Cy5 (sgc4f binding), TAMRA (sgc8c binding) and FAM channel (Nb binding) (Fig. 3f). For K562 cells, no fluorescence was observed in any channel; for Ramos and RPMI-8226 cells, fluorescence was only observed in Cy5 channel; for Jurkat cells, fluorescence was only observed in TAMRA channel (Fig. 3g and Supplementary Fig. 18). The fluorescent intensity after computation was further confirmed by confocal imaging as shown in Fig. 3h and Supplementary Figs. 19–22. These results demonstrated that the three-layer three-input DNP was successfully constructed for muti-parameter cell identification by involving multiple aptamers and Nb into molecular computation, just as predicted by the kinetic model.

Kinetic analysis-directed optimization of computing efficiency of DNP

A widely existed bottleneck of microenvironment-confined DNA computing strategies is the slow computing speed and weak output signal. By simulating the accelerated reaction rate in different computing layers, we find that the signal conversion in the input computing layer (layer 1) is the rate determining step that limit the computing rate of DNPs (Fig. 4a). Therefore, we predict that only by increasing the reaction rate of the first layer can enhance computing efficiency (Fig. 4b). To verify our prediction, we next constructed DNP3 with two identical recognition toes on its nano-claw as the simplest multivalent design to increase the concentration of reactants and used DNP1 with only one recognition toe as the control (Fig. 4c). PAGE and DLS were used to characterize the assembly of DNP3 (Supplementary Fig. 23). Kinetic model-based simulation shows that DNP3 has a higher reaction rate than DNP1 (Fig. 4d). Flow cytometry results show that the fluorescent intensity in FAM (green) channel of target CEM cells incubated with DNP3 is about twice that incubated with DNP1 (Fig. 4e). In contrast, control cell groups showed no obvious fluorescence signal no matter incubated with DNP3 or DNP1 (Fig. 4f, and Supplementary Fig. 24). Confocal imaging was also performed as shown in Fig. 4g and Supplementary Fig. 25 and fluorescent intensity quantification of individual cells was shown in Fig. 4h. These results demonstrate that the kinetic model can determine the rate-limiting step and help to optimize computing efficiency.

Kinetic analysis of signal leakage of DNA computation in complex cell mixtures

The signal leakage in complex cell mixtures is another problem of microenvironment-confined DNA compution. To discuss the performance of DNP in complex cell mixtures containing both target cells and by-standing cells, we updated the the kinetic model by considering the substance exchange between the microenvironments of target cells and by-stranding cells and the surrounding environment. Since DNP3 can be activated by two coexisting membrane proteins on cell membrane, here two kinds of by-standing cells each with only one membrane protein (by-standing cell 1 and 2) are considered in the kinetic model (Fig. 5a). A set of differential equations following the kinetic profile was formulated and provided in Supplementary Note 5 and a set of fitted rate constants was provided in Supplementary Table S3. For microenvironment-confined DNA compution, there are mainly two kinds of leakages: 1) the leakage caused by cross talk between target cells and by-standing cells (type I leakage); 2) the leakage caused by cross talk between by-standing cells (type II leakage). The type I leakage mainly occurs when the expression level of input protein 2 (e.g. CD47) on target cell is lower than input protein 1 (e.g. PTK7). As a result, the excessive accumulation of upstream signals in the cellular microenvironment will lead to the failure of “diffusion trap”, which can be clearly described using our kinetic model. As shown in Fig. 5b, if the expression level of CD47 is higher than PTK7, the Nb-S strand difussed into the surrounding solution is limited to an extreamly low concenration by the “diffusion trap”. However, if the expression level of CD47 is lower than PTK7, the concentration of the Nb-S strand difussed into the surrounding solution will increase sharply, leading to the cross between target cells and by-standing cells. The type II leakage mainly occurs when input proteins are separately expressed in different by-standing cells. Both type I and type II leakage of DNP3 in complex cell mixtures can be simulated computationally. The simulation result shows that type I leakage increases with increased concentration of target cells while type II leakage increases with increased concentration of by-standing cells (Fig. 5c). Interestingly, both type I and type II leakages are not significant compared with the strong output signal on target cells, because the diffused DNA strand will be quickly diluted in the surround solution so that the kinetics of leakage reaction in the microenvironment of by-standing cell is quite slow. The simulation result also indicates that identifying trace amount of target cells in complex cell mixtures is an intrinsic merit of microenvironment-confined DNA computation because the reaction kinetics in the microenvironment of target cell will not be affected by extremely low concentration of target cell (Supplementary Fig. 26).

Then we tried to verify these predictions using experiments. The target cells were pre-stained with cell-nuclear dye Hoechst (blue) and mixed with nontarget cells Ramos, K562 and PMI-8226. After incubating with DNP1 or DNP3, only target CEM cells displayed obvious fluorescence in FAM channel (Fig. 5d). The fluorescent intensity distributions of the target and nontarget cells treated by DNP1 and DNP3 were calculated by ImageJ, respectively, further confirming the strong fluorescent intensity of target CEM cells and very weak fluorescent intensity of nontarget cell mixtures (Supplementary Fig. 27). We further investigated whether DNPs could selectively identify target cells mixed with human blood leukocytes. Healthy human blood samples were centrifuged to remove erythrocytes, and the resuspended PBMC were mixed with tumor cells (CEM, Ramos, K562 and RPMI-8226) pre-stained with mitochondrial red in the same ratio (30,000 tumor cells and 100,000 blood leukocytes). The mixture was subsequently incubated with DNP3 (Supplementary Fig. 28). As shown in Fig. 5e, only target CEM cells showed obvious fluorescence in FITC channel, indicating that CEM was accurately identified among the mixed cells. These results verified the above prediction that microenvironment-confined DNA compution can effectively identify target cells in complex cell mixtures without nonspecific signal output on nontarget cells.

Shielding the interference of erythrocytes and enhancing the phagocytosis of macrophages

A current problem in CD47 blockade is that indiscriminate CD47-SIRPα blockade has a limited effect on attenuating tumor immune escape due to the competitive binding of by-standing erythrocytes and can cause severe hemolytic effects^47,48,49,50. Considering the huge proportion of erythrocytes in blood cells, precisely targeting trace amount of cancer cells in human blood samples remains a great challenge. To test the performance of DNP in shielding the interference of erythrocyte and identifying trace amount of cancer cells in human blood samples, we obtained human erythrocytes by centrifugation from healthy human blood. Serum stability experiments (Supplementary Fig. 29) shows that the DNPs remain stable after 6 h incubation in culture medium with 10% FBS, which is consistent with the previous reports⁴⁵. Classical CCK-8 assay shows negligible cytotoxicity of DNPs (Supplementary Fig. 30), indicating the high biocompatibility of DNPs. The effective protection of DNA-cage on encapsulated Nb from binding with erythrocytes was demonstrated using flow cytometry (Supplementary Fig. 31). We then mixed the obtained erythrocytes with different proportion (0.25%, 0.5%, 1%) of CEM cells pre-stained with Mitochondrial Red to prepare the mimic blood samples with trace amount of cancer cells (Supplementary Fig. 32). Then mimic blood samples were incubated with Nb-S, DNP1 and DNP3, respectively. As shown in Fig. 6a and Supplementary Fig. 33, the mixed cells incubated with Nb-S showed obvious FITC fluorescence, indicating that the indiscriminate binding of CEM and erythrocytes. For samples incubated with DNP1 and DNP3, CEM cells showed obvious FITC fluorescence while erythrocytes showed no fluorescence, indicating the improved CD47 binding specificity of Nb and accurate target CEM identification with reduced off-target effect in clinic human blood. These results emphasized the intrinsic merits of DNP in accurate and sensitive identification of trace amount of cancer cells in human blood samples as predicated by kinetic analysis. We also tested the hemolysis effect of DNPs. Using Triton X-100 as the positive control, we found that blood samples treated with DNP3 showed no obvious hemolysis. In contrast, commercialized monoclonal antibodies B6H12 caused enhanced hemolysis in lower concentration range (from 0 nM to 85 nM), but caused weakened hemolysis in higher concentration range (from 85 nM to 680 nM) (Fig. 6b and Supplementary Fig. 34). This phenomenon is consistent with previous reports and can be explained by the high dose hook effect.

**Fig. 6: Shielding the interference of erythrocyte and enhancing the phagocytosis of macrophages.**

Finally, we tested whether DNP3 could specifically enhance phagocytosis of tumor cells (Supplementary Fig. 35). Human derived M1 macrophages were induced by THP-1 cells⁵¹ (Supplementary Fig. 36). The expression of CD80 was tested using flow cytometry to confirm the successful differentiation of M1 macrophages (Supplementary Fig. 37). We also characterized the expression of SIRPα on M1 macrophages as shown in Supplementary Fig. 38. The DNP3 mediated phagocytosis enhancement was verified by treating cancer cell/M1 macrophages mixture with Nb-S, cage-Nb, or DNP3. CEM, K562, Ramos and PMI-8826 cells were used as the model cancer cells. Cancer cells were stained with CellTrace™ Far Red dye and M1 macrophages were stained with CellTrace™ CFSE for optical differentiation. Therefore, macrophages that phagocytosed cancer cells will be stained by Far Red and exhibit enhanced fluorescence in Far Red channel. Cells were then collected for flow cytometry analysis to quantify phagocytosis by calculating the proportion of macrophages with enhanced fluorescence in Far Red channel in the total macrophages (Fig. 6c and Supplementary Fig. 39). As shown in Fig. 6d, CEM cells showed enhanced phagocytosis after DNP3 treatment while control cells showed no significant difference in phagocytosis with control groups. Confocal imaging further confirmed these results. As shown in Fig. 6e and Supplementary Fig. 40, DNP3-treated CEM cells resulted in enhanced phagocytosis by macrophages, with numerous Far red-stained cancer cell fragments localized inside of the CFSE-stained M1 macrophages. The significantly enhanced phagocytosis was verified by statistical analysis as shown in Fig. 6f. The time-depended macrophage-cancer cell interaction and phagocytosis were also recorded, indicating obvious phagocytosis caused by DNP3 treatment (Fig. 6g). These results indicated that the specificity and biosafety of CD47-SIRPα pathway as an immunotherapy target were significantly improved by DNP-based molecular computation.

Discussion

In this work, we reported a spatial segregation-based molecular computing platform termed DNA nano-phage to program nanobodies into DNA molecular computation. By developing a stoichiometric and site-specific DNA-Nb labeling strategy, we successfully encapsulated Nb into the shielded internal computing layer of DNP through hybridization. Benefit from the spatial segregation strategy, DNP can respond to multiple bio-inputs, execute multi-layer molecular computation, and release the encapsulated Nb-DNA complex with programmed order. The released Nb can block CD47/SIRP-α signaling between target cells and macrophages and eliminate “don’t eat me” signal to promote phagocytosis, therefore improving the specificity and biosafety of CD47 blockade and providing approaches for enhanced immune therapy. We expect that the rational design of DNPs will lead the development of other smart DNA nanodevices and expanded applications in multi-target recognition, controlled signal regulation, precision biomedicine and immunotherapy. Theoretically, other recognition molecules such as small molecular ligands, peptides, antibodies, aptamers, as well as reactive molecules with covalent groups (e.g., maleimide and succinimide) or hydrophobic groups (e.g., cholesterol) can also be involved into molecular computation using such a strategy to eliminate off-target effect.

Based on DNP, we further established a series of kinetic models to reveal the molecular mechanism and predict the computing performance of microenvironment-confined DNA molecular computation. We explored the rapid diffusion behavior of DNA molecules in the cellular microenvironment and proposed a “diffusion trap” theory to explain the kinetics of microenvironment-confined DNA computation. Through kinetic analysis, we improved the computing efficiency by optimizing the rate-limiting step and demonstrated that microenvironment-confined DNA computation has intrinsic merit in discriminating trace amount of target cells in complexed cell mixtures. These kinetic models quantitatively describe the complex and cascading dynamic behavior of DNA molecules in non-uniform environments, therefore successfully elucidate the molecular mechanism of microenvironment-confined DNA computation. We believe that the principle we used in model construction can be applied to various types of microenvironment-confined systems and can provide insights for optimizing the performance of DNA computing.

Methods

The study was approved by the Ethics Committee at Renji Hospital, School of Medicine, Shanghai Jiao Tong University (protocol number: KY2024-070-C). All methods were performed in accordance with these approved guidelines.

Materials and apparatus

All DNA Sequences shown in Supplementary Table S4 were synthesized, modified and purified by Sangon Biotech Co. Inc. (Shanghai, China). Unless otherwise specified, reagents were purchased from Sinopharm Chemical Reagent Company, Ltd. (Shanghai, China). Ultrapure water was obtained from a Milli-Q system (Billerica, PA, USA). The fluorescence spectra were determined by FS5 Spectrofluorometer Fluorescence Spectrophotometer (Edinburgh Instrument, UK). The AFM characterizations were carried out on a Bruker Multimode V8 Scanning Probe Microscope (Bruker, Bremen, Germany). Flow cytometry analysis was performed by a CytoFLEX flow cytometer (Beckman Counter, Inc., USA).

Synthesis of Nb-S complex

The amino acid sequence of CD47 Nb is (patent number: CN115737845A):

MHHHHHHGVKLVESGGGLVQPGGSLTLSCVASGFDFNSAHMRWYRQGPGKEREMVASISTTGGVTIYEDSVKGRFTISRDNADNTAYLRMNSLKPEDTAVYYCRAYGFGIDYWGQGTQVTVSSGGGGSGGGGSLLQS.

To prepare Nb-DNA complex Nb-S, we here developed an efficient and indirect Nb labeling strategy to achieve accurate 1:1 and site-specific DNA modification by using the bifunctional molecule NH₂-PEG-N₃ as the linker. First, the CD47 Nb was mixed with NH₂-PEG-N₃ at a ratio of 1:50 in PBS buffer, and then glutamine transaminase was added to react at 37 °C and shake at 1000 rpm for 16 h. Excessive NH₂-PEG-N₃ was removed using His-Ni column, and the recovered intermediate product Nb-PEG-N₃ was concentrated for subsequent crosslinking reaction. The concentrated Nb-PEG-N₃ was mixed with DBCO-S strand in the ratio of 1 : 3 and the mixture were slight shaken at 1000 rpm at 37 °C for 5 h. The cross-linked product Nb-S was purified by Fast Protein Separation Chromatography (FPLC) and verified with SDS-PAGE and 10% native polyacrylamide gel electrophoresis (native-PAGE).

The binding competition between CD47 Nb and monoclonal antibody

When using monoclonal antibody B6H12-PE (cat.no.12283-MM07-P, Sino Biological, China) to complete the binding between CD47 Nb and target cell, cells were first washed with DPBS three times and incubated with Nb or Nb-S at 4 °C for 30 min. Then cells were washed with DPBS three times and incubated with B6H12-PE at 4 °C for 30 min. Finally, cells were washed with DPBS three times and analyzed by flow cytometry. When using CD47 Nb to complete the binding between monoclonal antibody B6H12-PE and target cell, cells were first washed with DPBS three times and incubated with B6H12-PE at 4 °C for 30 min. Then cells were washed with DPBS three times and incubated with Nb or Nb-S at 4 °C for 30 min. Finally, cells were washed with DPBS three times and analyzed by flow cytometry. Cells incubated with monoclonal antibody B6H12-PE were used as a positive control whose binding efficiency was set as 100%.

Assembly of DNA phages

DNA phages were constructed by DNA hybridization-based self-assembly. Firstly, the DNA strands in Supplementary Table S4 with equal molar ratio (100 nM for PAGE and fluorescence analysis, 500 nM for flow cytometry and cell experiments) were mixed in 1 × TAE/Mg²⁺ buffer (40 mM Tris-HCl, 20 mM CH₃COOH, 2 mM EDTA, 12.5 mM MgCl₂, pH = 7.4) and annealed from 95 °C to 25 °C over 6.5 h to construct DNA cage and Y1/Y2. Then Nb-S and DNA cage were mixed in 1×TAE/Mg²⁺ buffer and incubated at 37 °C for 2 h to encapsulate Nb-S into DNA cage and construct cage-Nb via hybridization. At last, sgc8c-17+c-sgc8c (1 : 1), Y1+sgc8c-Y+c-sgc8c (1 : 2 : 2) and Y2+sgc8c-Y+c-sgc8c+sgc4f +c-sgc4f (1 : 1 : 1 : 1 : 1) were mixed in 1×TAE/Mg²⁺ buffer respectively, and incubated for 1 h at 37 °C to construct different DNA nano-claws (claw 1,claw 2 and claw 3), Then different DNA nano-claws were equipped on constructed cage-Nb via hybridization to construct DNP1, DNP2 and DNP3.

Native polyacrylamide gel electrophoresis (native PAGE) analysis

We used 8% native-PAGE to characterize the step-by-step assembly of DNA cage and used 5% native-PAGE to analyze the loading of functional module into DNA cage. The samples were running in 1 × TAE/Mg²⁺ buffer under 80 V for 80 min in ice-water bath. The gels were stained with Gel-Green for 10 min and imaged by Bio-Rad ChemiDoc XRS System (Bio-Rad, USA) with imaging software under UV light.

Fluorescence analysis

All fluorescence spectra were measured using a FS5 Spectrofluorometer fluorescence spectrophotometer (Edinburgh Instruments, UK). The fluorescence excitation wavelength was set to 488 nm and the emission range was set to 500–550 nm. Unless otherwise stated, the concentration of DNA strands, Cage-Nb, or DNP used for was fluorescence analysis 100 nM and the bandwidth was set to 2 nm. For real-time fluorescent kinetics analysis, fluorescent intensity was measured every 30 s.

Serum stability

To demonstrate the serum stability of nano-claws, DNA cages and DNPs, samples were incubated with 10% FBS-supplemented RPMI 1640 medium at 37 °C for different times (0, 1, 2, 3, 4, 5 and 6 h). Then 2 µl of 6 × loading buffer was added into 10 µl sample for 1% agarose gel electrophoresis (AGE) analysis. AGE was run under 60 V for 40 min with ice-water bath in 1 × TAE/Mg²⁺ buffer. DNA bands were stained by Gel-Green and imaged using the Bio-Rad ChemiDoc XRS System with imaging software under UV light.

AFM imaging

5 µl of Ni²⁺ (30 mM) was dropped to the freshly torn mica’s surface to enhance the adhesion of the DNA samples. Then 10 µl of 100 nM samples were deposited and left for 10 min to adsorb. Finally, the mica was rinsed three times with DPBS. The samples were imaged using a Multimode 8 Scanning Probe Microscope (Bioscope system, Bruker, Germany) in “ScanAsyst in fluid” mode and the images were analyzed by NanoScopeAnalysis software.

DLS characterization

100 µl of 500 nM purified samples (DNA claws, DNA cages, DNPs) were prepared to measure the hydrodynamic diameter with Zetasizer Nano-ZS (Malvern Instruments). These experiments were repeated three times.

Cell culture

RPMI-8226, Jurkat and THP-1 was purchased from Procell Life Science & Technology Co., Ltd. (Wuhan, China), other cancer cell lines were purchased from American Type Culture Collection (Manassas, VA, USA). RPMI-CEM (Human acute lymphoblastic leukemia T lymphocytes), Ramos (Human B lymphocytoma cells), K562 (Human chronic myelogenous leukemia cells), RPMI-8226 (Human multiple myeloma peripheral blood B lymphocytes) and Jurkat (Human T-lymphocyte leukemia cells) were cultured in RPMI 1640 media (Thermo Fisher Scientific, USA) supplemented with 10% fetal bovine serum, 100 U/ml penicillin, and 100 mg/ml streptomycin at 37 °C in 5% CO₂. THP-1 (Human monocytic leukemia cell line) was cultured in 1640 media (Procell Life Science & Technology Co., Ltd., China) supplemented with 10% fetal bovine serum, 100 U/ml penicillin, 100 mg/ml streptomycin and 50 µg/mL 2-Mercaptoethanol at 37 °C in 5% CO₂.

Binding assay using flow cytometry

All flow cytometry experiments are carried out on a CytoFLEX flow cytometer (Beckman Counter, Inc., USA). To characterize the binding ability of Nb-S and aptamers (Sgc8c and Sgc4f) to different cell lines, 1 × 10⁵ RPMI-CEM, Ramos, K562, RPMI-8226 and Jurkat were incubated at 37 °C for 1 h with 100 nM FAM-labeled Nb-S, TAMRA-labeled Sgc8c, and Cy5-labeled Sgc4f, respectively. To test the computing performance of DNPs, 100 nM DNPs (DNP1, DNP2 and DNP3) were incubated with 3 × 10⁵ cells in 100 µl of binding buffer for 1 h at 37 °C. After washing with washing buffer (5 mM MgCl₂ and 4.5 g/L glucose in DPBS) three times to remove unbound probes, fluorescence signals of cells were analyzed using flow cytometer (Beckman Counter, Inc., USA) by counting 10,000 cells.

Confocal laser scanning microscopy

All confocal images were recorded using Nikon TIE + A1 SI confocal laser-scanning microscope (Nikon, Japan). RPMI-CEM, Ramos, K562, RPMI-8226 and Jurkat cells were incubated with 100 nM DNPs (DNP1, DNP2 and DNP3) for 1 h at 37 °C. Then cells were washed with washing buffer three times and imaged using confocal microscope.

Target cell identification from the cell mixture

1 × 10⁵ or 5 × 10⁴ CEM cells were pre-stained with Mito-tracker Red CMXRos (Beyotime Biotechnology, Shanghai, China) for flow cytometry. 2 × 10⁵ CEM cells were pre-stained with Hoechst (Beyotime Biotechnology, Shanghai, China) for fluorescence confocal microscopy imaging. The pre-stained CEM cells were washed three times with washing buffer, and mixed with K562, Ramos and RPMI-8226 (1 × 10⁵, each cell line) cell mixtures in binding buffer. Then cell mixtures were incubated with 100 nM DNP2 at 37 °C for 1 h. The samples were washed three times with washing buffer and analyzed using flow cytometry and imaged using confocal microscope.

Target cell identification from the blood sampled

All human blood analyzed in this study were collected with informed consent from a healthy volunteer (female, 28-year-old) and approved protocols that complied with all relevant ethical regulations at Renji Hospital, School of Medicine, Shanghai Jiao Tong University. Erythrocytes in the blood samples were pre-removed by gradient centrifugation (800 × g) and leukocytes were washed the sample three times with DPBS. Then the resuspended leukocytes were mixed with Mito-tracker Red-labeled tumor cells (CEM, K562, Ramos and RPMI-8226) in a ratio of 1:0.3 (leukocyte : tumor cells). The cell mixtures were incubated with 100 nM DNP2 at 37 °C for 1 h and washed three times with washing buffer. The samples were further analyzed by flow cytometry.

Hemolysis assay

Leukocytes and plasma in the blood samples were removed by gradient centrifugation (800 × g) and erythrocytes were washed three times with DPBS. The erythrocytes were resuspended with DPBS to prepare a 20% erythrocyte suspension. 100 µl of samples and 100 µl of 20% erythrocyte suspension were added to each well of 96-well plates. Triton X-100 and B6H12 (cat.no.14-0479-82, Thermo Fisher Scientific, USA) were used as positive control. The samples were incubated at 37 °C for 3 h. After incubation, the samples were transferred into 1.5 ml tubes and centrifuged for 10 min at 5000 rpm. 100 µl of supernatant from the control group and experimental group were transferred into a 96-well plate for absorption analysis using a microplate reader at 450 nm. The hemolysis was calculated using the following equation:

$${Hemolysis}\left(\%\right)=\frac{\left({A}_{{sample}}-\left({A}_{{negative}}-{A}_{{control}}\right)\right)}{\left(\left({A}_{{positive}}-{A}_{{control}}\right)-\left({A}_{{negative}}-{A}_{{control}}\right)\right)}\,*\, 100\%$$

Induction of macrophage differentiation

To induce the differentiation of macrophage, 1 × 10⁵ THP-1 cells were seeded in 24-well cell culture plates or 4-well confocal dishes, and incubated with 100 ng/ml PMA (Phorbol 12-myristate 13-acetate, MACKLIN, China) for 24 h to induce the THP-1 cells differentiate into M0 macrophages. Then M0 macrophages were stimulated with 100 ng/ml LPS and 20 ng/ml INF-γ (Novoprotein, Shanghai, China) for 24 h to generate M1 macrophages. DPBS was used to wash samples in flow cytometry and confocal imaging experiments.

In Vitro Phagocytosis assay

For the flow cytometry assay, 5 × 10⁵ cancer cells labeled with Far-red (Thermo Fisher Scientific, USA) were mixed with CFSE-labeled (Thermo Fisher Scientific, USA) M1 macrophages and incubated with 100 nM Nb-S, cage-Nb and DNP2 for 3 h at 37 °C in serum-containing medium. After washed twice with DPBS, the cell mixture was treated with enzyme-free digestion solution for 8 min. The detached cells were collected and resuspended in cell culture medium containing serum. Phagocytosis was analyzed using a CytoFLEX flow cytometer (Beckman Coulter). For confocal imaging, Far-red-labeled cancer cells were mixed with CFSE-labeled M1 macrophages and treated with DNPs in 4-well dishes at 37 °C for 3 h. Images were recorded with the Nikon TIE + A1 SI confocal laser-scanning microscope with a × 40 water immersion objective.

Inclusion & ethics

This study does not involve multi-region collaborations. Researcher contributions and authorship criteria have been carefully considered throughout the research process.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The data supporting the findings of this study are available within the article and its Supplementary Information. Source data are provided with this paper and available via FigShare (https://doi.org/10.6084/m9.figshare.27312108.v2). Source data is available for Figs. 1–6, and Supplementary Figs. S1-S27, S29-S31, S33-S34, S36-S40. Source data are provided with this paper.

Code availability

The code (Supplementary code 1-6) that supports the findings of this study are available from FigShare (https://doi.org/10.6084/m9.figshare.26132110.v1).

References

Lemmon, M. A. & Schlessinger, J. Cell Signaling by Receptor Tyrosine Kinases. Cell 141, 1117–1134 (2010).
Article CAS PubMed PubMed Central Google Scholar
Rozenfeld, R. & Devi, L. A. Receptor heteromerization and drug discovery. Trends Pharmacol. Sci. 31, 124–130 (2010).
Article CAS PubMed PubMed Central Google Scholar
Hwang, J. R., Byeon, Y., Kim, D. & Park, S. G. Recent insights of T cell receptor-mediated signaling pathways for T cell activation and development. Exp. Mol. Med. 52, 750–761 (2020).
Article CAS PubMed PubMed Central MATH Google Scholar
Zhu, J. & Thompson, C. B. Metabolic regulation of cell growth and proliferation. Nat. Rev. Mol. Cell Biol. 20, 436–450 (2019).
Article CAS PubMed PubMed Central MATH Google Scholar
Yang, X. et al. The circular RNA CDR1as regulate cell proliferation via TMED2 and TMED10. BMC Cancer 20, 312 (2020).
Article CAS PubMed PubMed Central MATH Google Scholar
Jahn, R., Cafiso, D. C. & Tamm, L. K. Mechanisms of SNARE proteins in membrane fusion. Nat. Rev. Mol. Cell Biol. 25, 101–118 (2023).
Article PubMed MATH Google Scholar
Li, D. & Wu, M. Pattern recognition receptors in health and diseases. Signal Transduct. Target. Ther. 6, 291 (2021).
Article CAS PubMed PubMed Central MATH Google Scholar
Li, J. et al. Nongenetic engineering strategies for regulating receptor oligomerization in living cells. Chem. Soc. Rev. 49, 1545–1568 (2020).
Article CAS PubMed MATH Google Scholar
Zosel, F., Soranno, A., Buholzer, K. J., Nettels, D. & Schuler, B. Depletion interactions modulate the binding between disordered proteins in crowded environments. Proc. Natl. Acad. Sci. USA. 117, 13480–13489 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Morsut, L. et al. Engineering customized cell sensing and response behaviors using synthetic notch receptors. Cell 164, 780–791 (2016).
Article CAS PubMed PubMed Central MATH Google Scholar
Williams, J. Z. et al. Precise T cell recognition programs designed by transcriptionally linking multiple receptors. Science 370, 1099–1104 (2020).
Article ADS CAS PubMed PubMed Central MATH Google Scholar
SenGupta, S., Parent, C. A. & Bear, J. E. The principles of directed cell migration. Nat. Rev. Mol. Cell Biol. 22, 529–547 (2021).
Article CAS PubMed PubMed Central MATH Google Scholar
Huttenlocher, A. Cell polarization mechanisms during directed cell migration. Nat. Cell. Biol. 7, 336–337 (2005).
Article CAS PubMed MATH Google Scholar
van Helvert, S., Storm, C. & Friedl, P. Mechanoreciprocity in cell migration. Nat. Cell. Biol. 20, 8–20 (2018).
Article PubMed Google Scholar
Li, J. et al. Amplified Visualization of Protein-Specific Glycosylation in Zebrafish via Proximity-Induced Hybridization Chain Reaction. J. Am. Chem. Soc. 140, 16589–16595 (2018).
Article CAS PubMed Google Scholar
Chen, S. et al. Logic-Gate-Actuated DNA-Controlled Receptor Assembly for the Programmable Modulation of Cellular Signal Transduction. Angew. Chem. Int. Ed. 58, 18186–18190 (2019).
Article ADS CAS Google Scholar
Suzuki, K. G. et al. Transient GPI-anchored protein homodimers are units for raft organization and function. Nat. Chem. Biol. 8, 774–783 (2012).
Article CAS PubMed MATH Google Scholar
Bhatia, D. et al. Quantum dot-loaded monofunctionalized DNA icosahedra for single-particle tracking of endocytic pathways. Nat. Nanotechnol. 11, 1112–1119 (2016).
Article ADS CAS PubMed PubMed Central MATH Google Scholar
Kuai, H. et al. Circular Bivalent Aptamers Enable in Vivo Stability and Recognition. J. Am. Chem. Soc. 139, 9128–9131 (2017).
Article CAS PubMed PubMed Central MATH Google Scholar
Zhou, F. et al. Molecular engineering-based aptamer-drug conjugates with accurate tunability of drug ratios for drug combination targeted cancer therapy. Angew. Chem. Int. Ed. 58, 11661–11665 (2019).
Article CAS MATH Google Scholar
Yang, Y. et al. Aptamer-Based Logic Computing Reaction on Living Cells to Enable Non-Antibody Immune Checkpoint Blockade Therapy. J. Am. Chem. Soc. 143, 8391–8401 (2021).
Article CAS PubMed MATH Google Scholar
Huang, Z. et al. Regulating the anticancer efficacy of sgc8-combretastatin A4 conjugates: A case of recognizing the significance of linker chemistry for the design of aptamer-based targeted drug delivery strategies. J. Am. Chem. Soc. 143, 8559–8564 (2021).
Article CAS PubMed MATH Google Scholar
Ahn, G. et al. LYTACs that engage the asialoglycoprotein receptor for targeted protein degradation. Nat. Chem. Biol. 17, 937–946 (2021).
Article CAS PubMed PubMed Central MATH Google Scholar
Banik, S. M. et al. Lysosome-targeting chimaeras for degradation of extracellular proteins. Nature 584, 291–297 (2020).
Article ADS CAS PubMed PubMed Central MATH Google Scholar
Miao, Y. et al. Bispecific Aptamer Chimeras Enable Targeted Protein Degradation on Cell Membranes. Angew. Chem. Int. Ed. 60, 11267–11271 (2021).
Article CAS MATH Google Scholar
Zhang, H. et al. Covalently Engineered Nanobody Chimeras for Targeted Membrane Protein Degradation. J. Am. Chem. Soc. 143, 16377–16382 (2021).
Article CAS PubMed Google Scholar
Liu, Y. et al. Aptamer-Based Targeted Protein Degradation. ACS Nano 17, 6150–6164 (2023).
Article CAS PubMed MATH Google Scholar
Zhang, J. et al. Scan and unlock: a programmable DNA molecular automaton for cell-selective activation of ligand-based signaling. Angew. Chem. Int. Ed. 60, 6733–6743 (2021).
Article ADS CAS MATH Google Scholar
He, F., Wang, M., Wang, J., Wang, H. H. & Nie, Z. An extracellular miRNA-responsive artificial receptor via dynamic DNA nano-assembly for biomarker-driven therapy. Angew. Chem. Int. Ed. 62, e202305227 (2023).
Article MATH Google Scholar
Li, H. et al. A DNA Molecular Robot that Autonomously Walks on the Cell Membrane to Drive Cell Motility. Angew. Chem. Int. Ed. 60, 26087–26095 (2021).
Article CAS Google Scholar
Gray, M. A. et al. Targeted glycan degradation potentiates the anticancer immune response in vivo. Nat. Chem. Biol. 16, 1376–1384 (2020).
Article CAS PubMed PubMed Central MATH Google Scholar
Zhang, D. et al. Equipping natural killer cells with specific targeting and checkpoint blocking aptamers for enhanced adoptive immunotherapy in solid tumors. Angew. Chem. Int. Ed. 59, 12022–12028 (2020).
Article CAS MATH Google Scholar
Labanieh, L., Majzner, R. G. & Mackall, C. L. Programming CAR-T cells to kill cancer. Nat. Biomed. Eng. 2, 377–391 (2018).
Article CAS PubMed MATH Google Scholar
Zhou, H., Jiao, P., Yang, L., Li, X. & Yan, B. Enhancing cell recognition by scrutinizing cell surfaces with a nanoparticle array. J. Am. Chem. Soc. 133, 680–682 (2011).
Article CAS PubMed MATH Google Scholar
Ren, K. et al. A DNA dual lock-and-key strategy for cell-subtype-specific siRNA delivery. Nat. Commun. 7, 13580 (2016).
Article ADS CAS PubMed PubMed Central MATH Google Scholar
Oostindie, S. C. et al. Logic-gated antibody pairs that selectively act on cells co-expressing two antigens. Nat. Biotechnol. 40, 1509–1519 (2022).
Article CAS PubMed PubMed Central MATH Google Scholar
Peng, R. et al. Engineering a 3D DNA-Logic Gate Nanomachine for Bispecific Recognition and Computing on Target Cell Surfaces. J. Am. Chem. Soc. 140, 9793–9796 (2018).
Article CAS PubMed PubMed Central MATH Google Scholar
Yin, Y. et al. FINDER: a fluidly confined CRISPR-based DNA reporter on living cell membranes for rapid and sensitive cancer cell identification. Angew. Chem. Int. Ed. 62, e202309837 (2023).
Article CAS Google Scholar
Wang, D., Li, S., Zhao, Z., Zhang, X. & Tan, W. Engineering a Second-Order DNA Logic-Gated Nanorobot to Sense and Release on Live Cell Membranes for Multiplexed Diagnosis and Synergistic Therapy. Angew. Chem. Int. Ed. 60, 15816–15820 (2021).
Article CAS Google Scholar
Gao, Q. et al. Highly Specific, Single‐step Cancer Cell Isolation with Multi‐Aptamer‐Mediated Proximity Ligation on Live Cell Membranes. Angew. Chem. Int. Ed. 59, 23564–23568 (2020).
Article CAS Google Scholar
Yuan, K. et al. Extracellular Milieu and Membrane Receptor Dual-Driven DNA Nanorobot for Accurate in Vivo Tumor Imaging. CCS Chem. 4, 1597–1609 (2022).
Article CAS MATH Google Scholar
Engelen, W., Sigl, C., Kadletz, K., Willner, E. M. & Dietz, H. Antigen-Triggered Logic-Gating of DNA Nanodevices. J. Am. Chem. Soc. 143, 21630–21636 (2021).
Article CAS PubMed PubMed Central Google Scholar
Schumacher, D., Helma, J., Schneider, A. F. L., Leonhardt, H. & Hackenberger, C. P. R. Nanobodies: Chemical Functionalization Strategies and Intracellular Applications. Angew. Chem. Int. Ed. 57, 2314–2333 (2018).
Article CAS Google Scholar
Rahman, M. M. et al. Chimeric nanobody-decorated liposomes by self-assembly. Nat. Nanotechnol. 19, 818–824 (2024).
Article CAS PubMed MATH Google Scholar
Fu, X. et al. Size-selective molecular recognition based on a confined DNA molecular sieve using cavity-tunable framework nucleic acids. Nat. Commun. 11, 1518 (2020).
Article ADS CAS PubMed PubMed Central MATH Google Scholar
Jeger, S. et al. Site‐Specific and Stoichiometric Modification of Antibodies by Bacterial Transglutaminase. Angew. Chem. Int. Ed. 49, 9995–9997 (2010).
Article CAS MATH Google Scholar
Lu, Y. et al. Immunological conversion of solid tumours using a bispecific nanobioconjugate for cancer immunotherapy. Nat. Nanotechnol. 17, 1332–1341 (2022).
Article ADS CAS PubMed PubMed Central MATH Google Scholar
Chen, Y. C., Shi, W., Shi, J. J. & Lu, J. J. Progress of CD47 immune checkpoint blockade agents in anticancer therapy: a hematotoxic perspective. J. Cancer Res. Clin. Oncol. 148, 1–14 (2022).
Article CAS PubMed MATH Google Scholar
Logtenberg, M. E. W., Scheeren, F. A. & Schumacher, T. N. The CD47-SIRPalpha Immune Checkpoint. Immunity 52, 742–752 (2020).
Article CAS PubMed PubMed Central Google Scholar
Feng, M. et al. Phagocytosis checkpoints as new targets for cancer immunotherapy. Nat. Rev. Cancer 19, 568–586 (2019).
Article CAS PubMed PubMed Central MATH Google Scholar
Hu, Y. et al. Molecular Force Imaging Reveals That Integrin-Dependent Mechanical Checkpoint Regulates Fcgamma-Receptor-Mediated Phagocytosis in Macrophages. Nano Lett 23, 5562–5572 (2023).
Article ADS CAS PubMed MATH Google Scholar

Download references

Acknowledgements

This work is supported by grants from the National Key Research and Development Project (2020YFA0909000), the National Natural Science Foundation of China (NSFC 22107027), the Natural Science Foundation of Hunan Province (2023JJ20003), the Science and Technology Innovation Program of Hunan Province (2024RC3099), and the Scientific Research Program of FuRong Laboratory (2023SK2088) to Y.L.; by grants from the National Natural Science Foundation of China (22104032) to M.X.

Author information

Authors and Affiliations

Molecular Science and Biomedicine Laboratory (MBL), State Key Laboratory of Chemo/Biosensing and Chemometrics, College of Chemistry and Chemical Engineering, Aptamer Engineering Center of Hunan Province, Hunan University, Changsha, Hunan, China
Decui Tang, Shuoyao He, Yani Yang, Yuqi Zeng, Mengyi Xiong, Yifan Lyu & Xiao-Bing Zhang
Institute of Molecular Medicine (IMM), Renji Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China
Ding Ding & Weihong Tan
Department of Nuclear Medicine, Institute of Clinical Nuclear Medicine, Renji Hospital, School of Medicine, Shanghai Jiao Tong University, Shanghai, China
Weijun Wei
Furong Laboratory, Changsha, Hunan, China
Yifan Lyu
Zhejiang Cancer Hospital, Hangzhou Institute of Medicine (HIM), Chinese Academy of Sciences, Hangzhou, Zhejiang, China
Weihong Tan

Authors

Decui Tang
View author publications
Search author on:PubMed Google Scholar
Shuoyao He
View author publications
Search author on:PubMed Google Scholar
Yani Yang
View author publications
Search author on:PubMed Google Scholar
Yuqi Zeng
View author publications
Search author on:PubMed Google Scholar
Mengyi Xiong
View author publications
Search author on:PubMed Google Scholar
Ding Ding
View author publications
Search author on:PubMed Google Scholar
Weijun Wei
View author publications
Search author on:PubMed Google Scholar
Yifan Lyu
View author publications
Search author on:PubMed Google Scholar
Xiao-Bing Zhang
View author publications
Search author on:PubMed Google Scholar
Weihong Tan
View author publications
Search author on:PubMed Google Scholar

Contributions

Y.L., X.Z. and W.T. conceived the project. D.T. and Y.L. design the experiments. D.T. performed the most experiments and data analyses. S.H. and Y.Z. performed the cells culture and DNA structure assembly. Y.Y. performed the AFM imaging. M.X., D.D. and W.W. participated in data analyses and provided experimental advice. Y.L., D.T., X.Z. and W.T. wrote the paper. All the authors discussed the results and commented on the paper.

Corresponding authors

Correspondence to Yifan Lyu, Xiao-Bing Zhang or Weihong Tan.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Chang Feng and the other, anonymous, reviewer for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Reporting Summary

Transparent Peer Review file

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Tang, D., He, S., Yang, Y. et al. Microenvironment-confined kinetic elucidation and implementation of a DNA nano-phage with a shielded internal computing layer. Nat Commun 16, 923 (2025). https://doi.org/10.1038/s41467-025-56219-9

Download citation

Received: 30 June 2024
Accepted: 13 January 2025
Published: 22 January 2025
Version of record: 22 January 2025
DOI: https://doi.org/10.1038/s41467-025-56219-9

This article is cited by

Interaction between ELMO1 DNA methylation and Med31 promotes H. pylori-induced gastric cancer EMT and intestinal metaplasia via M2 polarization
- Tianyu Lu
- Tingting Yu
- Jian Song
Scientific Reports (2026)