Author Correction: DNA-based platform for efficient and precisely targeted bioorthogonal catalysis in living systems

As one of the typical bioorthogonal reactions, copper(I)-catalyzed azide-alkyne cycloaddition (CuAAC) reaction holds great potential in organic synthesis, bioconjugation, and surface functionalization. However, the toxicity of Cu(I), inefficient catalytic activity, and the lack of cell specific targeting of the existing catalysts hampered their practical applications in living systems. Herein, we design and construct a DNA-based platform as a biocompatible, highly efficient, and precisely targeted bioorthogonal nanocatalyst. The nanocatalyst presents excellent catalytic efficiency in vitro, which is one order of magnitude higher than the commonly used catalyst CuSO4/sodium ascorbate. The theoretical calculation further supports the contribution of DNA structure and its interaction with substrates to the superior catalytic activity. More importantly, the system can achieve efficient prodrug activation in cancer cells through cell type-specific recognition and produce a 40-fold enhancement of transformation compared to the non-targeting nanocatalyst, resulting in enhanced antitumor efficacy and reduced adverse effects. In vivo tumor therapy demonstrates the safety and efficacy of the system in mammals. Copper-click reaction has been used for a wide range of bio-conjugations but does suffer from toxicity issues. Here, the authors report on the growth of copper nanoparticles on DNA with linked aptamer targeting and demonstrate high catalytic effect and improved application due to targeting and biocompatibility.

B ioorthogonal reactions are emerging as promising approaches for intracellular labeling bio-targets and modulating biological processes without disturbing other biomolecules and function [1][2][3][4][5][6][7][8][9][10][11] . As one of the typical bioorthogonal reactions, CuAAC reaction has been applied as a powerful synthetic tool for bioconjugations or coupling functional groups and molecules with fast kinetics and high yields under mild conditions [12][13][14][15] . Despite great advance in organic synthesis, material science, and medicinal chemistry [16][17][18] , the implementation of CuAAC in living systems remains in its infancy. The inherent toxicity of the active catalytic species Cu(I) can damage normal cells and healthy tissues and cause numerous side effects, limiting the application of CuAAC within biological systems 19 . To eliminate the toxic effect of Cu(I), several approaches were explored to develop biocompatible catalysts, including utilization of ligand-chelated or support-stabilized Cu(I), in-situ reduction of Cu(II) salts with the addition of sodium ascorbate, metallic copper nanoparticles and nanostructured copper oxide [20][21][22][23][24][25] . However, these approaches still suffer from intrinsic drawbacks, which hamper their biological applications. First, the reaction time is rather long in general, and the yields are much insufficient. In some cases, high temperature, microwave or non-eco-friendly reagents was employed to enhance the reaction rate 26 , which cannot fulfill the requirements of CuAAC in living cells. Second, these catalysts themselves cannot recognize target cells 27 , limiting the applicable dosage to specific regions, compromising the catalytic transformation efficacy, even damaging normal cell populations, and causing numerous side effects due to nonspecific distribution. Recently, metal-organic frameworks supported heterogeneous copper nanocatalyst has been reported to catalyze CuAAC in mitochondria by decorating a triphenylphosphonium (TPP) group 28 . Nevertheless, the nanocatalyst showed low tumor selectivity and unsatisfactory treatment specificity since overcoming the barriers of extracellular matrix and cell membrane to selectively target specific cells is the premise of the subcellular organelle-targeted delivery 29 . Besides, complicated and tedious preparation processes, and difficulty in control of size, morphology, dispersion and stability which determine catalysis performance, are matters of concern.
DNA as a powerful and intelligent building block for construction of functional nanomaterials has attracted great attention [30][31][32] . Aside from excellent properties, including biocompatibility, programmability, and easy synthesis and modification, DNA can be used as a template to synthesize metal nanoclusters or nanoparticles, such as Au, Ag, and Cu 33,34 , whose size can be exquisitely controlled by the length and sequence of DNA 35 . These DNA-templated nanoparticles have been used in many areas, such as nanoelectronics, sensing, and bioimaging 36 . Meanwhile, aptamers, single-stranded DNA or RNA sequences selected through SELEX technology to recognize various targets with high affinity and specificity, have been widely utilized for analytical chemistry and nanomedicines [37][38][39] . Aptamer conjugated on the surface of nanoparticles can specifically bind to overexpressed receptors on tumor cells, and then improve cellular uptake through the receptor-mediated endocytosis 40 . In the meantime, DNA strands were recently exploited to regulate catalytic activity of nanozymes 41,42 . For example, DNA could enhance activity of Fe 3 O 4 for the peroxidation of 3,3′,5,5′-tetramethylbenzidine by 10-fold 43 . The enhanced reactivity could be attributed to highly effective molarity of reaction components in or near the DNA facilitated by electrostatic attraction, hydrogen bonding and aromatic stacking 43 . These properties open the opportunities for designing DNA-based platform as biocompatible, highly active, and targeted bioorthogonal catalyst.
Inspired by these unique properties, we construct a DNA-based platform for efficient and precisely targeted bioorthogonal catalysis in living systems. The introduction of DNA offers an attractive catalyst synthesis method. The platform can catalyze azide-alkyne cycloaddition reaction in vitro with short reaction time and high yield. The theoretical calculation further supports the contribution of DNA to the enhanced catalytic activity. More importantly, the work addresses the issues of the lack of precise targeting and inefficient catalytic activity of the bioorthogonal catalyst in living system. The highly efficient, targeted, and biocompatible nanocatalyst can be applied in tumor therapy via in situ activation of prodrugs. The precise and efficient catalytic synthesis of drugs in tumor tissues can ensure satisfactory and specific antitumor efficacy and reduce the serious adverse effects. Furthermore, the programmability of nucleic acids enables the generality of DNA-based nanocatalyst by simply using different sequences. The work provides exciting opportunities to design versatile bioorthogonal catalysts for in vivo applications in complex multicellular organisms.

Results and discussion
The design and preparation of DNA-based copper nanoparticles. The design principle of the DNA-based nanocatalyst is illustrated in Fig. 1a. The single-stranded DNA is composed of three domains. One portion is a thymine-rich sequence which can serve as a template for the formation of copper nanoparticles (CuNPs) with small sizes, good dispersion, stability, and low toxicity to cells. Since the size of the nanocatalyst is of utmost importance in catalysis for providing a large surface-to-volume ratio and increased number of defect sites 44,45 , CuNPs with different sizes and fluorescence intensities can be obtained by changing the number of thymine. Herein, DNA strands containing different lengths of thymine-rich regions (T20, T30, and T40) are chosen. The second portion is aptamer which can target cancer cells. Mucin 1 (MUC1), a cell-surface receptor protein, is known to be uniquely and abundantly expressed on the surface of adenocarcinoma cells rather than on normal cells 46 . MUC1 aptamer is selected for targeting MUC1 protein on the surface of cancer cells (Fig. 1b). Besides, AS1411 aptamer possessing high affinity to nucleolin overexpressed in tumor cells is also selected 47 . The use of different aptamers targeting different types of cancer cells allows for the versatile applicability. The third portion is a 15-base strand which links the different lengths of thymine-rich regions and different types of aptamers. The resulting nanocomposites are termed as Apt-Cu 20 , Apt-Cu 30 , and Apt-Cu 40 and can be used for bioorthogonal catalysis.
First, DNA-templated CuNPs were synthesized according to the previously reported method. The copper content of the DNAtemplated CuNPs after synthesis was shown in Supplementary  Table 2. The obtained DNA-templated CuNPs were monodispersed ( Fig. 1c-e). The average sizes of CuNPs were 1.78, 3.12, and 4.79 nm for DNA strands containing T20, T30, and T40, respectively (Supplementary Figs. 1-3, and Fig. 1f). Atomic force microscope (AFM) images showed one nanoparticle formed on one DNA template ( Supplementary Fig. 4). The zeta potential measurement showed that the nanocomposites were negatively charged (Fig. 1g). X-ray photoelectron spectroscopy (XPS) spectrum of Cu2p showed two distinct peaks at 932.6 eV (Cu 2p 3/2 ), and 952.2 eV (Cu 2p 1/2 ), corresponding to Cu (0) nanoparticles (Fig. 1h). Then the fluorescence spectra of CuNPs were investigated. As shown in Fig. 1i, CuNPs presented fluorescence emissions at 600 nm upon excitation at 340 nm. The intensities increased significantly with the increase of thymine numbers. Under the irradiation of UV lamp, the nanocomposites exhibited red fluorescence. The results indicated the successful synthesis of DNA-templated CuNPs.
Catalytic performance of the DNA-based nanocatalysts in vial. The catalytic capability of the DNA-based nanocatalysts was tested by catalyzing the CuAAC reaction in vitro. Transformation of precursors 1 (3-azido-7-hydroxycoumarin) and 2 (phenylacetylene) into triazole 3 in the presence of catalyst was chosen as a model reaction (Fig. 2a). 1 and 2 are non-fluorescent (Supplementary Fig. 5), while the product 3 emits strong fluorescence (Fig. 2b, Supplementary Fig. 6) (λ ex = 340 nm; λ em = 460 nm). In the experiment, Apt-Cu 20 , Apt-Cu 30 and Apt-Cu 40 were incubated with the mixture of 1 and 2 in vials, respectively. The samples instantaneously showed cyan-blue fluorescence, suggesting the superior catalytic activity of these CuNPs. The products were measured using fluorescence spectra. As shown in Fig. 2b, all three Apt-Cu nanocatalysts could induce increase of fluorescence intensity at 460 nm upon excitation at 340 nm, implying the formation of triazole 3. Moreover, the fluorescence intensities increased in the order of Apt-Cu 30 > Apt-Cu 40 > Apt-Cu 20 . Subsequently, to investigate the catalytic efficiency and kinetic process, the changes of fluorescence intensity at 460 nm with time were recorded. As shown in Fig. 2c, in the absence of nanocatalyst, the fluorescence remained constant over the period. Upon addition of nanocatalyst, the fluorescence turned up instantaneously. Within the first 10 min, a rapid increase in fluorescence was clearly observed (Fig. 2d, e), suggesting the excellent catalytic activity. The reaction rates increased in the order of Apt-Cu 30 > Apt-Cu 40 > Apt-Cu 20 ( Supplementary  Fig. 7). The conversion rates of the three Apt-CuNPs at 10 min were analyzed by high-performance liquid chromatography (HPLC), respectively ( Supplementary Fig. 8). The results were consistent with the fluorescent results. The fluorescence intensity reached the plateau after 60 min, indicating the catalytic reaction was nearly completed. Furthermore, the conversion rates of Apt-Cu 30 at different time points were analyzed by HPLC. As shown in Supplementary Figs. 9-11, in the presence of Apt-Cu 30 , the conversion rate increased with time. 90% of the substrate was converted into the product within 60 min, supporting the result of the fluorescence measurement. Meanwhile, the commonly used catalyst CuSO 4 /sodium ascorbate was tested for comparison. As shown in Supplementary Table 3, the catalytic transformation using Apt-Cu 30 was 26 times higher than that of using CuSO 4 / sodium ascorbate in 0.5 h. Besides, CuNPs synthesized using DNA strand without aptamer region (WA-Cu 30 ) were used for comparison. Both Apt-Cu 30 and WA-Cu 30 could catalyze CuAAC reaction in H 2 O, and there was no obvious difference of fluorescence intensity, indicating aptamer did not influence the catalytic efficiency of CuNPs in vitro ( Supplementary Fig. 12). Moreover, DNA strand without thymine-rich region was tested as control. Due to the lack of thymine-rich sequence as a template, CuNPs could not be synthesized. The fluorescence spectra of click reaction showed weak signal (Supplementary Fig. 13), indicating the catalytic capability of this DNA strand was low even in a long reaction time (Supplementary Table 4). Sizes, supporting materials and surface sites of nanocatalysts have major impacts on their performance. Generally, the smaller size possesses the larger specific surface area, leading to the higher catalytic activity and the higher atom utilization rate. Herein, the size of CuNPs obtained increased with the length of DNA templates. However, their catalytic activities did not completely follow the size-activity relationship, and Apt-Cu 20 presented the poorest activity. Therefore, it was deemed important to explore other factors. It is reported that more poly T bases could provide more stability for CuNPs 48 . Apt-Cu 20 possessed the worst stability due to the shortest DNA template, which might result in the lower catalytic activity. Cu is easily oxidized due to its low reduction potential. XPS analysis was performed to determine the oxidation state of Apt-Cu 20 . As shown in Supplementary Fig. 14, two peaks at 932.4 and 952.2 eV assigned to Cu (0) were observed. Meanwhile, there were two peaks at 933.6 eV (Cu 2p 3/2 ) and 953.6 eV (Cu 2p 1/2 ) identical to Cu (II), indicating the presence of Cu (II). The content of Cu (0) accounts for only 38.8%. The reduced content of Cu (0) and the existing oxidation state on the surface of Apt-Cu 20 would compromise its catalytic activity. To the end, the synergic effects of size, supporting materials and surface sites determined the catalytic efficiency of these CuNPs.
In vitro, aptamer did not present specific recognition because there are no cells. However, the nanocatalysts still showed higher catalytic activity. To further clarify the role of DNA in the highly efficient catalysis, molecular dynamics simulation was used to explore the structure of DNA and their interaction with substrates 49,50 . The DNA structure (T30-15 base linker) was first relaxed by molecular dynamics simulation. The root-meansquare deviation (RMSD) values during the simulation indicated that the DNA sequence can reach a stable state in 20 ns ( Supplementary Fig. 15). During the 30 ns simulation, the 15base motif (linker sequence) can quickly fold to a hairpin-like secondary structure. In contrast, the T30 motif did not form any stable secondary structure. The last frame of simulation was employed for docking with 1 and 2. As illustrated in Fig. 2f, the hydroxyl group of 1 inserted in the hairpin-like structure and generated two hydrogen bonds with A33 and C40. What's more, the N 3 group exposed to the bulk water, which would not affect the combination with 2. Meanwhile, the hydrophobic interactions between 1 and nucleic acids A32, C34, T39 and A41 stabilized the complex. For 2, the complex was stabilized only by hydrophobic interactions between the ligand and the nucleic acids C31, A32 and A41 (Fig. 2g). The interactions between DNA and substrates indicated high molarity of reaction components in or near the DNA and proximity to the catalytic sites, resulting in more efficient catalytic activity. We also tested the catalytic capability of these nanocatalysts under different physiological conditions (Fig. 2h, Supplementary  Fig. 16). Similar results were obtained in these buffers, suggesting the nanocatalyst could be used in biological systems without loss of catalytic activity.
Cell targeting ability and intracellular catalysis of the nanocatalyst. After demonstrating the catalytic activity of Apt-Cu in vial, we next tested their targeting capability. Due to the optimal catalytic activity of T30-templated CuNPs among the three CuNPs, MUCI aptamer-T30-templated-CuNPs (MApt-Cu 30 ) were used for further research. CuNPs synthesized using DNA strand without aptamer region (WA-Cu 30 ) were used as a control. First, the cytotoxicity of MApt-Cu 30 (Fig. 3b, Supplementary  Fig. 23a) and high fluorescence intensity ratio (Fig. 3c) in comparison with MCF-7 and A549 cells treated with WA-Cu 30 . Slight or no shift and fluorescence change was observed for normal cells or MUC1 − cancer cells under identical conditions, confirming the target recognition of MApt-Cu 30 for type-specific cells. In addition, we tested the targeting capability of a single DNA sequence MApt and WA without growth of CuNPs. The flow cytometry analysis results proved that the target recognition capability was not influenced after the synthesis of CuNPs ( Supplementary Figs. 23b, 24, 25).
Subsequently, the ability of MApt-Cu 30 to target cancer cells to perform catalysis was determined by monitoring the signals of cell membrane dye-DiO (green), CuNPs (red) and the product triazole 3 (blue) using confocal microscopy. As shown in Fig. 4a, MCF-7 cells treated with precursors+WA-Cu 30 presented weak red fluorescence, while cells treated with precursors+MApt-Cu 30 exhibited strong red fluorescence. It indicated that aptamer could increase CuNPs uptake via receptor-mediated endocytosis. MCF-7 cells treated with precursors or precursors+WA-Cu 30 displayed weak blue fluorescence, while cells treated with precursors +MApt-Cu 30 showed intense blue fluorescence. It implied that higher transformation efficiency of precursors 1 and 2 into triazole 3 in the presence of MApt-Cu 30 was achieved, which could be attributed to the higher accumulation of CuNPs as catalyst in the living cell. The merged fluorescence image of CuNPs and triazole 3 revealed that the CuAAC reaction was efficiently achieved within the MUC1 + cancer cells. The LC-MS analysis result also demonstrated the synthesis of triazole 3 in MCF-7 cells (Supplementary Fig. 26). Similar results were obtained for A549 cells (Supplementary Fig. 27). For MDA-MB-231 cells treated with precursors+WA-Cu 30 and precursors +MApt-Cu 30 , they presented weak red fluorescence and there was no obvious difference in red fluorescence (Fig. 4b). It indicated that MUC1 aptamer could not recognize MUC1 − cells without related receptors to enhance CuNPs uptake, leading to the lower amount of CuNPs in these cells. Consequently, these cells presented low blue fluorescence due to the low yield of triazole 3 catalyzed by a limited amount of CuNPs. Similar phenomena were observed for HepG2, RAW 264.7, NIH-3T3 and HEK293 (Supplementary Figs. [28][29][30][31]. Furthermore, the flow cytometric analysis showed that MApt-Cu 30 caused above 180fold increase of blue fluorescence in MUC1 + cells incubated with precursors 1 and 2 compared with the cells without MApt-Cu 30 (Fig. 4c, d). In sharp contrast, WA-Cu 30 presented rather low catalytic activity since the DNA strand could not specifically recognize these cells. Above 40-fold fluorescence enhancement was obtained for MApt-Cu 30 catalyzed CuAAC reaction compared that with WA-Cu 30 . The above results confirmed that the higher catalytic yield of triazole 3 could be attributed to aptamer-mediated cell selectivity and enhanced uptake of CuNP catalyst.
To demonstrate the DNA-based CuNP catalyst as a universal platform for aptamer-mediated cell selectivity, AS1411 aptamer which can specifically bind to nucleolin overexpressed on human cervical cancer (HeLa) cells was selected as another model. First, the cytotoxicity of AS1411 aptamer-T30-templated-CuNPs (AApt-Cu 30 ) was tested through MTT experiment. As shown in Supplementary Fig. 32, AApt-Cu 30 was biocompatible to HeLa cells even at the concentration of 50 μM. Next, the cellular internalization and intracellular catalysis performance of AApt-Cu 30 was monitored by confocal microscopy images (Supplementary Figs. 33,34). HeLa cells treated with AApt-Cu 30 presented stronger green, red and blue fluorescence compared with cells treated with WA-Cu 30 . The overlap of green fluorescence from FAM fluorescein labeled on AApt-Cu 30 , red fluorescence of AApt-Cu 30 , and the blue fluorescence of the product triazole 3 indicated that AApt-Cu 30 was specifically internalized by HeLa cells and exhibited good intracellular catalytic activity. The results confirmed the universality of the DNA-based platform by simply using different aptamers.
Targeted prodrug activation in living cells. Prodrug activations by bioorthogonal reactions have been considered as a powerful strategy for tumor inhibition with satisfactory effectiveness [51][52][53][54] .
Having demonstrated the superior cell-specific recognition of MApt-Cu 30 and the resulting excellent catalytic effect in CuAAC reaction, we next explored the application of the nanocatalyst in cancer therapy using Resveratrol (3,5,4′-trihydroxystilbene; Rsv) analog 6 as model drug 55 . Rsv and its analogs have elicited a vast interest due to its multiple beneficial properties, such as chemopreventive and antitumor activities 56 . However, the compounds can target many biomolecules and have different effect in different environments. The in-situ generation of Rsv analogues 6 could improve its anti-tumor effect and minimize the side effects. We synthesized the prodrug 4, 5 and then utilized a MTT assay to investigate the cytotoxicity of 4, 5, and the resulting Rsv analogues 6.
As shown in Fig. 5a and Supplementary Figs. 35, 36, in the absence of MApt-Cu 30 , low cytotoxicity was found for the prodrugs 4 and 5 even at the concentration of 50 μM in various cells. Upon addition of MApt-Cu 30 , MUC1 + cells showed a notable decrease in cell viability while MUC1cells and normal cells showed a slight decrease (Fig. 5b, Supplementary Fig. 36). It implied that Rsv analogues 6 could be synthesized under the catalysis of CuNPs and resulted in a cytotoxicity to cells. The difference in cytotoxicity to MUC1 + cells and MUC1 -/normal cells resulted from specific recognition of MUC1 aptamer. To further prove the effect of specific binding of MApt-Cu 30 to MUC1 on prodrug activation, the control experiment was performed. First, a DNA strand complementary to MUC1 aptamer was mixed with MApt-Cu 30 and denoted as MApt-Cu 30 -block (Supplementary Table 1). Then, CuNPs were synthesized using MUCI aptamer-T30 with two mutants and denoted as mutMApt-Cu 30 . MTT assay was carried out to investigate the prodrug activation by MApt-Cu 30 -block and mutMApt-Cu 30 . The results showed that the cell viabilities of MCF-7 treated with MApt-Cu 30 -block and mutMApt-Cu 30 were higher than that treated with MApt-Cu 30 (Supplementary Fig. 37). For MApt-Cu 30 -block, the complementary sequence could inhibit aptamer-protein binding. For mutMApt-Cu 30 , the mutant aptamer had much lower affinity to MUC1 compared to MUC1 aptamer. Therefore, the accumulation of the two CuNPs was much lower, resulting in lower catalytic efficiency. Flow cytometry analysis by double staining showed that the death of MUC1 + cell was primarily associated with apoptosis (Fig. 5c). As negative controls, the percentage of apoptosis was much lower in MUC1 − cells. Combined with HPLC ( Supplementary Fig. 38) and LC-MS ( Supplementary Fig. 39) analysis, it can be proved that the CuAAC reaction proceeded smoothly in the living cells and Rsv analogues 6 with the molecular weight of 270.0 was generated. These experiments demonstrated that MApt-Cu 30 could significantly promote the CuAAC reaction for prodrug activation, thereby exhibiting stronger antitumor efficacy in specific type cancer cells and minimizing the side effects to normal cells.
Targeted prodrug activation in vivo. With the outstanding targeted catalytic performance of MApt-Cu 30 in living cells, we next investigated its behaviors in vivo. We initially evaluated the in vivo toxicology of MApt-Cu 30 using Caenorhabditis elegans (C. elegans) model ( Supplementary Fig. 40a). The N 2 wild-type strain worms were chosen to monitor the toxicity and phenotype alteration. Similar to the control group, the worms incubated with  Fig. 40c). The results showed that MApt-Cu 30 could effectively achieve CuAAC reaction in vivo with biocompatibility. Furthermore, we systematically studied the toxicology of MApt-Cu 30 to further clarify their safety in vivo using athymic nude mouse model. Hemolysis test of MApt-Cu 30 was first carried out. No or negligible hemolysis of red-blood cells occurred at concentrations ranging from 3.125 to 50 μM (Supplementary Fig. 41). After intravenous injection of the nanocatalyst into MCF-7 tumor bearing nude mice, the biodistribution of MApt-Cu 30 was studied at 4 h, 12 h, and 24 h (Supplementary Fig. 42) via ICP-MS analysis. Owing to the high specificity and affinity of MUC1 aptamer to MUC1 + cells, the accumulation of MApt-Cu 30 in MCF-7 tumor region was largely completed within 4 h. In addition, the accumulation of MApt-Cu 30 was only decreased 16%, elucidating that MApt-Cu 30 could be remained in tumors at a high concentration for at least 24 h. The results indicated MApt-Cu 30 with favorable accumulation and retention could be used as effective modality for selective targeted cancer therapy. Hematology analysis and blood biochemical assay were performed to evaluate the long-term toxicity of MApt-Cu 30 . The parameters in the MApt-Cu 30 group were compared with those in the control group ( Supplementary  Fig. 43). The injection of MApt-Cu 30 had negligible impact on the mice growth ( Supplementary Fig. 44). Hematoxylin and eosin (H&E) stained images showed that no organ damage or adverse effects occurred after administration of MApt-Cu 30 (Supplementary Fig. 45). These biosafety results indicated MApt-Cu 30 was nontoxic as well as favored their further bioapplications.
Encouraged by the satisfactory treatment result at the cellular level, we further explored the performance of MApt-Cu 30 in CuAAC reaction in vivo for cancer therapy. MCF-7 tumorbearing athymic nude mouse model, characterized by MUC1 + cell lines, was established (Fig. 6a). The mice were randomly divided into six groups (five mice per group): (1) (Supplementary Fig. 46). After 4 h, prodrugs 4 and 5 were administered intraperitoneally. PBS was used as the control. The administrations of nanocatalysts and prodrugs were repeated every 5 days. The body weights and tumor sizes of each group were monitored to conduct separate efficacy study during the course of treatments (Fig. 6b, c). By comparing the sizes of excised tumors mass from each group, it could be proved that 4 + 5 + MApt-Cu 30 greatly reduced tumor growth rate and exhibited admirable antitumor efficacy in MCF-7 tumors model (Fig. 6d). As a result of significantly enhanced accumulation of MApt-Cu 30 nanocatalyst in MCF-7 tumor, prodrugs could be effectively activated via CuAAC reaction. The staining assay showed apoptotic cells significantly increased, implying the tumors cells were badly damaged and adequately reduced (Fig. 6e). Above all, 4 + 5 + MApt-Cu 30 possessed desirable therapeutic effect on in vivo tumor and negligible side effects to normal tissues, resulting from specific cancer cell-targeting ability of nanocatalyst for CuAAC reaction.
To further elucidate the behaviors in vivo, pharmacokinetic analysis of prodrug 4, prodrug 5 and product 6 via HPLC was performed. As shown in Supplementary Table 5 and Supplementary Fig. 47, both prodrugs could reach maximum in the blood circulation system within 30 min, while the content of product 6 was relatively low. The result suggested very few 6 was synthesized through bioorthogonal reaction in the blood circulation system, demonstrating the specific targeting of the nanocatalyst.
In summary, a DNA-based nanocatalyst for bioorthogonal catalysis was successfully designed and constructed. The DNA strand not only served as the building block of nanocarrier and template of CuNPs, but also performed a specific recognition to target cells. The resulting nanocatalyst presented good biocompatibility, superior catalytic transformations, and type-specific cell recognition in vitro. The catalytic transformation efficiency was improved by an order of magnitude compared to the commonly used catalyst CuSO 4 /sodium ascorbate, and 40 fold compared to the nontargeting nanocatalyst. Molecular dynamics simulation was performed and showed that DNA structure and its interaction with substrates contributed to the powerful catalytic activity of the nanocatalyst. The system revealed a promising prospect in the practical use of prodrug activation in vivo. Animal studies using tumor-bearing mouse model validated our nanocatalyst and prodrugs with excellent tumor inhibition and negligible side effect to major organs. Furthermore, the programmability of nucleic acids enables the generality of aptamer-based nanocatalyst by simply using different aptamers. This proof-of-concept meant flexible targeting can be achieved, representing an easy and promising approach for the personalized therapy of cancers. It is expected this precision-targeted bioorthogonal nanocatalyst based on nucleic acid with superior performance to be a significant advance in the application of nanomaterials in biomedicine and the work will pave a new way for the treatment of important human diseases.

Methods
Preparation of DNA-based Cu nanoparticles. DNA strands (1 μM) containing thymine-rich regions were dissolved in 3-(N-Morpholino) propanesulfonic acid (MOPS) buffer (10 mM, pH 7.6) and sodium ascorbate solution (2 mM). After blending completely, CuSO 4 solution (100 μM) was added and incubated for 15 min at room temperature. The mixtures were centrifuged via ultrafiltration centrifuge tube (retained molecular weight: 10 K) to remove unreacted substance. Fluorescent copper nanoparticles (CuNPs) were formed within 5 min (λex = 340 nm). The fluorescence spectra were generally obtained after five minutes reaction in our experiments. The optimized synthetic condition of Apt-Cu 30 was obtained by testing different Cu (II) concentration and T30 concentration.
The oligonucleotide sequences used in the experiments are listed in a separate Supplementary Table. Catalytic performance in vial. Apt-Cu (with an equivalent amount of copper 5 µM), 1 (10 µM), and 2 (10 µM) were mixed in water at 25°C. The fluorescence intensities of the mixtures were measured by fluorescence spectrometer. The catalytic efficiency and kinetic process were measured by HPLC and fluorescence spectra. Molecular dynamics simulation of DNA structure. The timestep was set as 2 fs and the system was run for 30 ns. The last frame of simulation was used for docking. The RMSD values during the simulation indicated that the DNA sequence can reach a stable state in 20 ns. During the 30 ns simulation, the 15-base motif can quickly fold to a hairpin-like secondary structure. In contrast, the T30 motif did not form any stable secondary structures.
Cellular internalization of MApt-Cu 30 and WA-Cu 30 quantified by ICP-MS. MCF-7 cells were preincubated with MApt-Cu 30 (5 µM) or WA-Cu 30 (5 μM) for different time, then washed with PBS for 3 times. After the cells were digested by trypsin, lysis buffer (300 µL) was added. The obtained cell lysate was treated with mixture of HNO 3 and H 2 O 2 (3:1 mL) overnight. Then aqua regia (3 mL) was added to the samples for another 2-3 h. The obtained samples were dried and redispersed into the water, followed by analysis using ICP-MS.

MApt-Cu 30 -mediated click reactions in living cells
Confocal microscopy imaging. MCF-7, A549, NIH-3T3, RAW, MDA-MB-231 and HepG2 cells were seeded on sterilized cover slips for 6 h in 24-well plates. MApt-Cu 30 or WA-Cu 30 (5 µM) was added and incubated with the cells for 4 h. Then the cells were washed with PBS for 3 times. Next, azide coumarin 1 (10 µM) and alkyne 2 (10 µM) were added and incubated with the cells for 12 h, followed by washing with PBS for 2 times. After staining cytomembrane by DiOC18 (3), the fluorescence images were taken by OLYMPUS-BX51 microscopes with U-25ND25 filter.
Flow cytometric analysis. MCF-7, A549, NIH-3T3, RAW, MDA-MB-231, and HepG2 cells were plated in 6-wellplates. Apt-Cu 30 (5 μM) was incubated with the cells for 4 h, and then washed with PBS for 3 times. Next, azide coumarin 1 (10 µM) and alkyne 2 (10 µM) were added and incubated with the cells for 12 h. After being washed with PBS for 2 times, the cells were harvested using trypsin and resuspended in PBS. The intracellular fluorescence of product 3 was analyzed using flow cytometry (450/50).
For MApt-Cu 30 -mediated synthesis of resveratrol analogue inside living cells, the experimental conditions were the same as above, except for replacing azide coumarin 1 and alkyne 2 with 3 and 4. The apoptosis of these cells induced by resveratrol analogue was analyzed by double staining with Annexin VFITC and propidium iodide using commercial apoptosis detection kit. The flow cytometry data were obtained by BD LSRFortessa™ Cell Analyzer and analyzed using FlowJo_V10. software.
LC-MS analysis. The MCF-7 cells were plated in 6-well plates for CuAAC reaction. Apt-Cu 30 (5 μM) was incubated with MCF-7 cells for 4 h, then washed with PBS for 3 times. Then, azide coumarin 1 (10 µM) and alkyne 2 (10 µM) were added and incubated with the cells for 12 h. After the reaction was completed, the cells were treated with trypsin, collected in ultrapure water, and then lysed by sonication. The resulting lysate was centrifuged at 12,470 g for 5 min. The supernatant was extracted to mix with cold acetone and kept at −20°C overnight. Then the mixture was centrifuged at 12,470 g for 15 min to collect the precipitate. The precipitate was re-dispersed in methanol and analyzed by LC-MS.
Caenorhabditis elegans (C. elegans) experiments. Wild-type (N2) strain were cultured in nematode growth medium (NGM) containing E. coli (OP50) at 20°C. MApt-Cu 30 (5 µM) was incubated with the N2 worms (about 50 worms per plate) for worm paralysis assay. After 4 h, worms were washed with PBS for 3 times to remove excess MApt-Cu 30 . Next, azide coumarin 1 (10 µM) and alkyne 2 (10 µM) were added, and incubated for 12 h. Worms were washed with PBS twice and fixed with 4% paraformaldehyde. The images of worms were taken using a fluorescence microscope after being washed by PBS.

ARTICLE
Animal experiments. Healthy six-week-old BalB/c nude mice (14~16 g) were purchased from the Laboratory Animal Center of Jilin University (Changchun, China). They were housed in a specific pathogen-free environment at 26 ± 1°C and 50 ± 5% humidity, with a 12 h light-dark cycle. The handling procedures of animal were in accordance with the guidelines of the Animal Ethics Committee of Jilin University for Animal Experiments.
The biocompatibility of MApt-Cu 30 in athymic nude mouse. Healthy BalB/c-nu mice were randomly divided into two groups. The mice in control group and experimental group were intravenously injected with PBS and MApt-Cu 30 (3.5 mg kg −1 ), respectively. Blood samples of the mice at different periods were collected for whole blood panel analysis and serum biochemical analysis. The blood biochemistry markers (WBC, RBC, MCV, HCT, MCH, MCHC, PLT and HGB), liver function markers (ALT, ALP, ALB, AST, and TP) and kidney function markers (BUN and CREA) were tests. Within 28 days after injection, the weight of mice was measured every 3 days. On the 28th day, the heart, liver, spleen, lung, and kidney of mice were collected, and fixed with 4% paraformaldehyde for subsequent H&E staining.
Biodistribution and pharmacokinetics of the nanocatalyst in tumor-bearing mice. The mice were intravenously injected with MApt-Cu 30 (3.5 mg kg −1 ). At 4, 12 and 24 h after injection, the mice were sacrificed. The organs and tumors were collected and digested in aqua regia at 80°C. The accurate distribution was evaluated by ICP-MS. H&E staining. The major organs (heart, liver, spleen, lung, and kidney) of the mice in the above seven groups were collected, and fixed with neutral buffered formalin (10%). Then the organs were embedded into paraffin and sectioned into 4 μm thickness for H&E staining. Furthermore, to evaluate the pathological damages to tumors, the tumor tissues sections were collected for H&E staining. The images of these sections were taken using a microscope under bright field.
Statistical analysis. All of the data were presented as the mean ± standard deviation (SD). All figures were obtained from a minimum of three independent experiments. Data analysis was performed using Origin 2020 and Excel 2018 software. Statistical evaluation was performed using two-tailed Student's t test analysis. Asterisks indicated significant differences (*P < 0.01, **P < 0.005, ***P < 0.001). Notably, no samples and animals were excluded from the analysis.