A DNA-based hydrogel for exosome separation and biomedical applications

The DNA hydrogel–based separation technology included three key steps: enzymatic amplification to synthesize two ultralong DNA chains, the pretreatment of medium by low-speed centrifugation, and the separation of EXOs by DNA hydrogel (Fig. 1). The two ultralong DNA chains (DNA-chain-1 and DNA-chain-2) were, respectively, produced by RCA using two circular DNA (circ-DNA) as templates (Fig. 1A and SI Appendix, Fig. S1 and Table S1). The complementary sequence of AptCD63 was integrated to circ-DNA-1, and thus DNA-chain-1 contained repeated sequences of AptCD63 (polyvalent AptCD63). The EXO-containing medium, such as cell culture medium and patient serum, was pretreated by low-speed centrifugation to remove cells and cell debris, during which the components with small size and low density remained in the supernatant, including EXOs, nontarget vesicles such as exomeres and high-density lipoprotein (HDL) (25). Upon adding DNA-chain-1 and gently shaking for 0.5 h at 37 °C, the EXOs in supernatant were anchored on DNA-chain-1 through the high affinity between AptCD63 and CD63 (Fig. 1B). Next, DNA-chain-2 was added into the system, and incubated for another 0.5 h at 37 °C. Through the 22-nt complementary sequences in DNA chains, DNA-chain-2 hybridized with DNA-chain-1 driven by hydrogen bond, forming DNA hydrogel. During the formation of DNA hydrogel, the captured EXOs were encapsulated in the hydrogel and separated from other components.

The formation of DNA hydrogel by the two DNA chains was first confirmed. The solutions of two DNA chains were mixed in a 1.5-mL tube, and shaken for 0.5 h at 37 °C; as a result, a bulk of DNA hydrogel formed (Fig. 2A). The rheology results showed that the storage modulus (G′, ~4 Pa) was higher than the loss modulus (G″, ~0.5 Pa), indicating the solid state and super-softness of DNA hydrogel (Fig. 2B). The porous microstructure of DNA hydrogel was important for the diffusion of EXOs in separation process. The images of scanning electron microscopy (SEM) and fluorescence microscopy showed the microscale three-dimensional (3D) porous structure of DNA hydrogel (Fig. 2 C and D). The results of mercury intrusion assay showed that the porosity of DNA hydrogel was 78.66%, and the size of pores in the hydrogel was larger than 100 nm, which facilitated the diffusion of EXOs in hydrogel (SI Appendix, Fig. S2).

For the construction and optimization of DNA hydrogel, BMSC-derived EXOs that were obtained by ultracentrifugation were employed as a demonstration. To verify the specific recognition of AptCD63 aptamer toward EXOs, EXOs were stained by red fluorescent dye CM-DiI, and AptCD63 was modified with green fluorophore FAM. After incubating EXOs with AptCD63 for 0.5 h, the colocalization of green fluorescence with red fluorescence was observed by fluorescence microscopy (Fig. 2E and SI Appendix, Fig. S3). As a control, noncorrelative ssDNA (NC) modified with FAM was incubated with EXOs. Pearson’s correlation coefficients were quantified as 0.81 (AptCD63) and 0.35 (NC), respectively, indicating the coincidence of AptCD63 and EXOs (Fig. 2E). These results demonstrated that AptCD63 was able to specifically recognize EXOs, which was an important prerequisite for the following EXO separation.

Next, the specific encapsulation of EXOs using DNA hydrogel was tested. The integration of AptCD63 sequence into DNA-chain-1 was verified using circular dichroism spectra (SI Appendix, Fig. S4), ensuring the specific recognition of DNA hydrogel toward EXOs. In fluorescence microscopy images, the fluorescence of EXOs colocalized well with that of DNA hydrogel (Fig. 2F). Fluorescence 3D stacking was further performed to observe the 3D spatial distribution of EXOs in DNA hydrogel. The images exhibited that EXOs were three-dimensionally distributed in the hydrogel as fluorescence dots, and the magnified region of interest showed that a large fluorescence dot contained a cluster of EXOs (Fig. 2G). These results indicated that EXOs were successfully encapsulated in the DNA hydrogel.

One of the important parameters of the separation technology was the capacity of DNA hydrogel to encapsulate EXOs. In the separation process, the volume ratio of DNA-chain-1 to DNA-chain-2 solution was 1:1. The numbers of encapsulated EXOs based on 50 μL DNA hydrogel were calculated. According to the linear relationship between EXO number and the fluorescence intensity of CM-DiI (SI Appendix, Fig. S5), the numbers of encapsulated EXOs were calculated by measuring the fluorescence intensity of the supernatant before and after separation. As shown in Fig. 2H, with initial EXO numbers increasing from 1 × 10⁷ to 1 × 10⁸, the numbers of encapsulated EXOs gradually increased, finally reaching a plateau with an encapsulation number of ~7.6 × 10⁷. These results indicated that the maximum encapsulation number to EXOs of 50 μL DNA hydrogel was 7.6 × 10⁷.

The controlled release of captured EXOs from DNA hydrogel was explored. Taking the advantage of enzymatic degradation property of DNA, deoxyribonuclease I (DNase I) was used to degrade DNA hydrogel (26) (Fig. 3A). To study the process of enzymatic degradation, DNA hydrogel was incubated with different units of DNase I (0, 0.01, 0.1 and 1 U) at 37 °C, respectively. By measuring the DNA concentration in supernatant, we obtained the degradation rate of DNA hydrogel (SI Appendix, Fig. S6). The degradation rate significantly increased with the increase in DNase I concentration. Under the catalysis of 1 U DNase I, the degradation rate reached 100% at 60 min, much higher than that in 0.1 U (68.36%) and 0.01 U (43.22%) DNase I. During degradation, the cumulative release rate of EXOs was also monitored by measuring the fluorescence intensity of supernatant. When the concentration of DNase I was 0.1 U/mL, the cumulative release rate of EXOs from the DNA hydrogels with different concentrations (volume ratios of DNA hydrogel to buffer were 1:5, 1:10 and 1:15) at 60 min was all higher than 90% (Fig. 3B and SI Appendix, Fig. S7A); while the cumulative release rate of EXOs in the system without DNase I was only 6.54% (SI Appendix, Fig. S7B). The quick release of EXOs by enzymatic degradation allowed to efficiently obtain the dispersed EXOs for the experiments such as cell viability and cell proliferation tests. 3D stacking image showed the distribution of EXOs changing from 3D to single-layered, along with the degradation of hydrogel (Figs. 2G and 3C). In addition, considering that this DNA hydrogel would be utilized in biological media that contained nucleases, the release rate of EXOs was further monitored in 10% fetal bovine serum at 37 °C. The release rate of EXOs rapidly reached 54.99% in the first 4 h, and became slow after 8 h (SI Appendix, Fig. S8). These results demonstrated that enzymatic degradation achieved the efficiently controlled release of EXOs from DNA hydrogel.

To verify the structural integrity of the EXOs released via enzymatic degradation, the size, morphology, and biomarkers of the EXOs in degradation product were characterized (27). The hydrodynamic diameter of EXOs was ~110 nm measured by nanoparticle tracking analysis (Fig. 3D and SI Appendix, Fig. S9), and the saucer-shaped morphology of EXOs was observed in transmission electron microscopy (TEM) image (Fig. 3E and SI Appendix, Fig. S10), matching the typical structural characteristics of EXOs. The biomarkers of MCF-7-derived EXOs, CD63, and TSG101 (28), were detected by western blot assay. The nonexosomal markers, calnexin that was located in endoplasmic reticulum and Rab5 that was located in early endosome and plasma membrane, were detected in cell lysates, with much higher expression than EXOs (Fig. 3F and SI Appendix, Fig. S11). These results demonstrated that our DNA hydrogel–based technology was able to specifically separate EXOs, and the EXOs released via enzymatic degradation retained structural integrity.

Next, we designed another route, strand-displacement, to release EXOs, without disintegrating the DNA hydrogel. As mentioned above, the separation by our DNA hydrogel was based on the high affinity between AptCD63 and EXOs, which was closely related to the spatial configuration of AptCD63. Consequently, the nondestructive release of EXOs by changing the spatial configuration of AptCD63 to reduce the affinity between AptCD63 and EXOs was achieved (Fig. 3G). Three displacement strands (DSs) with different base numbers were designed, including DS₁₅, DS₂₀, and DS₃₂ (SI Appendix, Table S1). In our design, the DSs were completely complementary to AptCD63 and used to displace the captured EXOs from DNA hydrogel through hybridizing with the AptCD63 sequences on DNA-chain-1. To verify the competition effect of strand-displacement, 12% polyacrylamide gel electrophoresis (PAGE) was executed. To visualize the location of AptCD63 and DSs, AptCD63 was labeled with FAM (green fluorescent dye), and DSs were labeled with TRMRA (red fluorescent dye). As shown in Fig. 3H, two types of fluorescence overlapped well after mixing AptCD63 with DSs (lines 5 to 7, yellow fluorescence), and the bands moved much slowly compared with AptCD63 (line 1) as well as single DS (lines 2 to 4), indicating that all the three DSs effectively hybridized with AptCD63. To simulate the competition process, AptCD63 was mixed with EXOs, and then incubated with different DSs. As the base numbers in DS increasing, the bands in lines 10 to 12 showed that red fluorescent smears (DS) faded, and the fluorescent intensity of yellow bands (AptCD63+DS) was enhanced, indicating that more AptCD63 strands were displaced by DS from EXOs, and DS₃₂ showed the most significant competition effect. Consequently, DS₃₂ was chosen as the displacement strand in the following experiments.

The release process of EXOs from DNA hydrogel by strand-displacement was monitored by fluorescence microscopy. Significant green fluorescence was observed on the DNA hydrogel that was incubated with FAM-labeled DS₃₂ (Fig. 3 I, Left), demonstrating the efficient hybridization of DS₃₂ with AptCD63 on DNA hydrogel. Then, the EXO-encapsulated DNA hydrogel was incubated with FAM-labeled DS₃₂. Compared with the group without DS₃₂ (Fig. 3 I, Middle), the red fluorescence of EXOs in hydrogel was significantly weakened (Fig. 3 I, Right), demonstrating that DS₃₂ successfully displaced EXOs from DNA hydrogel. To verify the structural integrity of the EXOs released through strand-displacement, the size, morphology, and biomarkers of the EXOs were characterized. The hydrodynamic diameter of EXOs was approximately 110 nm (Fig. 3J), and the morphology was saucer-shaped in TEM image (Fig. 3K), matching the typical structural characteristics of EXOs. Biomarkers CD63 and TSG101 were detected in the supernatant of release system by western blot; nonexosomal markers calnexin and Rab5 were detected in cell lysates (Fig. 3L and SI Appendix, Fig. S11). These results demonstrated that the EXOs released by strand-displacement remained structural integrity.

Exosomal miRNA is a category of reliable and noninvasive biomarker for the early diagnosis of cancer (29–31). However, the low abundance in EXOs and the tedious process of extraction often limited the sensitivity of exosomal miRNA detection. In our separation technology, after adding DNA-chain-2, the EXOs captured by DNA-chain-1 can be efficiently enriched as the formation of DNA hydrogel, which achieved a high local concentration of EXOs and was beneficial to improve the sensitivity of miRNA detection; moreover, the DNA hydrogel can work as a framework for the integration of detection units, e.g., oligonucleotide molecular beacon (MB) and single probe (SP) that have been utilized in in situ detection of EXOs (32, 33), achieving high specificity and accuracy. The sequences of MB and SP were designed and integrated to DNA hydrogel to develop a detection method toward exosomal microRNA-21 (miR-21) detection, a biomarker of breast cancer (31). MB was a hairpin-structure DNA that was modified with a red fluorophore Cy3 at 5′ and a quencher (BHQ2) at 3′; SP was an ssDNA modified with FAM at 5′ (Fig. 4A and SI Appendix, Table S1). When MB was at hairpin state, the fluorescence of Cy3 was quenched by BHQ2. Both MB and SP were designed to hybridize with DNA-chain-1: MB hybridized with AptCD63 sequence, and SP hybridized with nonaptamer sequence on DNA-chain-1 (Fig. 4A). When both MB and SP hybridized with DNA-chain-1, the fluorescence of FAM on SP was quenched by BHQ2 on MB due to their closer distance.

After hybridizing with MB and SP, DNA-chain-1 was added into biological medium to capture EXOs. When AptCD63 recognized CD63 on EXOs, CD63 displaced MB from DNA-chain-1 due to the higher affinity toward AptCD63, and the green fluorescence of FAM was recovered due to the weakened quenching effect from BHQ2. When the displaced MB penetrated into EXOs, the sequence on the loop of MB recognized miR-21 and hybridize to form double-stranded structure, which resulted in the weakened quenching effect of BHQ2 to Cy3 and the recovery of red fluorescence of Cy3. Finally, DNA-chain-2 was added to form DNA hydrogel. The enrichment of EXOs by DNA hydrogel enhanced the fluorescence signals of FAM and Cy3, so that the signals were able to be detected by fluorescence microscopy. In particular, the green fluorescence was used as reference to minimize the influence of EXO concentration on detection accuracy by built-in correction (34). Consequently, the fluorescence intensity ratio of red to green was calculated to reflect the level of exosomal miR-21.

In our design, the specificity of MB was based on the complementary base pairing between MB and target miRNA, which was proved by 12% PAGE (SI Appendix, Fig. S12A). To verify the feasibility of fluorescence recovery, the fluorescence intensity of MB was measured with and without target miR-21, respectively. The fluorescence intensity increased 20.8 times in the presence of miR-21 (SI Appendix, Fig. S12B), demonstrating the efficient configuration transition of MB from hairpin to double helix, thus weakening the quenching effect of BHQ2 to Cy3. Then, three miRNAs (miR-21, miR-27a, miR-375) and the corresponding MBs (MB-21, MB-27a, MB-375) were used to study the specificity of MB in an orthogonal assay, and MB with noncorrelated sequence (MB-NC) was used as a control. As shown in Fig. 4B, the fluorescence intensity of MBs was not affected by noncomplementary miRNAs, demonstrating the specificity of MB toward miRNA.

We next studied the feasibility of MB for the detection of miRNAs in the human breast cancer cell (MCF-7)–derived EXOs from cell culture medium. The penetration of MB into EXOs has been reported by the previous literatures (32, 35). We performed fluorescent colocalization experiment and proved the high penetration efficiency of MB-21 into MCF-7 EXOs by the well-overlapped fluorescence (SI Appendix, Fig. S13). To investigate the penetration process of MB-21 into EXOs, the fluorescence intensity of Cy3 was monitored after mixing MB-21 with MCF-7 EXOs (SI Appendix, Fig. S14). In the first 45 min, the intensity was gradually increased, indicating that the MB-21 was penetrating into the EXOs and recognizing miR-21; after 45 min, the intensity reached a plateau. In addition, MB-NC was used as a control to investigate the detection specificity of MB-21 toward exosomal miR-21, and only the group of MB-21 with MCF-7 EXOs showed the significant fluorescence of Cy3 (SI Appendix, Fig. S15). These results demonstrated the potential of MB-21 for specific detection of exosomal miR-21.

Two circ-DNAs of RCA were designed to produce new DNA chains (DNA-chain-1 and DNA-chain-2) for hybridization with MB-21 and SP, and the synthesis was proved by 12% PAGE (SI Appendix, Fig. S16). The time of RCA was optimized to obtain DNA chains that hybridized with more MB-21. When the time of RCA was 4 h, the fluorescence intensity of the MB-21 (without BHQ2) bound on hydrogel was the highest, and the fluorescence intensity of free MB-21 in supernatant was the lowest, indicating the highest hybridization efficiency of MB-21 with hydrogel (SI Appendix, Fig. S17). Next, the DNA chains produced by 4-h-RCA were utilized to hybridize with MB-21 and SP, and further formed DNA hydrogel. To ensure a high signal–noise (S/N) ratio of the detection, the ratio of MB-21 (with BHQ2) and SP added in the hydrogel was adjusted, and the corresponding S/N ratio by monitoring the two fluorescence of different hydrogels was calculated (SI Appendix, Fig. S18). After analysis, the optimal mole ratio of MB-21 to SP was determined as 1:0.6 (SI Appendix, Fig. S19). Finally, a fluorescent colocalization experiment was performed, demonstrating that the hydrogel formed by DNA-chain-1 and DNA-chain-2 was able to separate EXOs from cell culture medium (SI Appendix, Fig. S20).

Based on the above results, we applied DNA hydrogel–based separation technology to the detection of exosomal miR-21 in clinic samples of breast cancer patients (BCs). It has been proved that miR-21 was differentially expressed in normal tissues and cancer regions (36, 37). The feasibility of the semiquantitative detection of exosomal miR-21 from different cell lines was first investigated, including MCF-7, another human breast cancer cell (MDA-MB-231) and human bronchial epithelial cells (BEAS-2B). After treating the culture media from three cell lines by this technology, three DNA hydrogels encapsulating different EXOs were obtained and then observed by fluorescence microscopy (Fig. 4C). According to the microscopy images, the fluorescence intensity ratio of red to green was calculated to reflect the level of exosomal miR-21. As shown in Fig. 4D, the ratios of MCF-7 EXOs and MDA-MB-231 EXOs were approximately 3.42-fold and 6.47-fold higher than those of BEAS-2B EXOs, respectively. To evaluate the credibility of detection results, we also used RT-PCR as a standard to measure the expression levels of miR-21 in the three types of EXOs (Fig. 4D). The detection results from these two detection technologies showed a similar tendency, with a high correlation coefficient of 0.99 (SI Appendix, Fig. S21). These results demonstrated that our DNA hydrogel was potential for the semiquantitative detection of exosomal miR-21 with high accuracy.

To further evaluate the feasibility of the clinical diagnosis of breast cancer based on this DNA hydrogel, we measured the relative expression levels of miR-21 in 30 serum samples, from 15 BCs and 15 healthy donors (HDs). Each sample was treated to form DNA hydrogel. By fluorescent microscopy, two fluorescence images of DNA hydrogel (Fig. 4E) and the corresponding fluorescence intensity ratios of 30 samples (SI Appendix, Fig. S22) were obtained. As shown in Fig. 4F, the average ratio of the samples from BCs was 2.12-fold higher than that from HDs, indicating the significantly higher level of exosomal miR-21 in the samples of BCs. Receiver operating characteristic analysis showed the reliable discrimination of BC versus HD (area under the curve = 1, Fig. 4G); meanwhile, the confusion matrix showed 100% sensitivity and 100% specificity for the classification of BC versus HD by our DNA hydrogel–based separation technology (Fig. 4H). These results demonstrated the potential of our DNA hydrogel–based detection method in clinical diagnosis of breast cancer.

We next applied DNA hydrogel to the therapeutics toward MI. Previous studies have proven that BMSC-derived EXOs were able to effectively improve the hypoxic state of myocardia and repair infarcted cardiomyocytes (38, 39). To make EXOs play a better role of repair in the region of infarcted myocardia, the localized administration and efficient release of EXOs were crucial. DNA hydrogel was able to be enzymatically degraded to gradually release EXOs in serum-containing environment (SI Appendix, Fig. S8). The swelling degree of DNA hydrogel in fetal bovine serum was −51.97% in 8 h (SI Appendix, Fig. S23). The DNA hydrogel was also proved to be injectable due to its shear-thinning property (SI Appendix, Fig. S24), and showed stable storage modulus after 250 cycles under alternating strain between 1% and 500% (SI Appendix, Fig. S25), indicating excellent stability against shearing. On the other hand, the results of cholecystokinin octapeptide assay demonstrated the low cytotoxicity of DNA hydrogel, which was essential for the in vivo experiments (SI Appendix, Fig. S26). Consequently, our DNA hydrogel provided an ideal delivery method to achieve the localized administration and efficient release of EXOs. BMSC-derived EXOs were separated from culture medium of BMSCs and encapsulated in DNA hydrogel. To investigate whether these separated EXOs had the biological activity to protect cardiomyocytes from hypoxia damage, an in vitro hypoxia model was established by incubating rat cardiomyocytes (H9C2) under hypoxic condition (5% O₂) for 24 h. The results of living/dead cell assay showed that the numbers of dead cells in hypoxia group were significantly higher than those in the normoxia group, increasing from 7.16 to 24.09% (SI Appendix, Fig. S27), demonstrating the successful establishment of the in vitro hypoxia model. To verify the internalization of the EXOs released from DNA hydrogel by cardiomyocytes, the released EXOs were stained by CM-DiI and incubated with human cardiomyocytes (AC16) and rat cardiomyocytes (H9C2), respectively. With the increase in incubation time, the fluorescence intensity of CM-DiI increased significantly in the cytoplasm of cardiomyocytes (SI Appendix, Figs. S28 and S29), indicating the efficient internalization of EXOs. We then studied the influence of BMSC-derived EXOs on H9C2 cells under hypoxic condition. With the increase in the EXO concentration, the viability of H9C2 cells significantly increased, reaching 140.91% of that in normoxic control H9C2 cells at 30 μg/mL EXOs (SI Appendix, Fig. S30). Fluorescence microscopy images also proved that the number of proliferating H9C2 cells (labeled by Alexa Fluor 488) gradually increased with increased EXO concentration (SI Appendix, Fig. S31). The protection effect toward hypoxia damage was further investigated at molecular level. Hypoxia inducible factor protein (HIF-1α) mediated the cellular response to hypoxia, which was overexpressed in cells under hypoxia condition (40). The results of western blot showed that the expression of HIF-1α in hypoxic H9C2 cells gradually decreased with increased EXO concentration; the HIF-1α level in hypoxic H9C2 cells treated with 10% EXOs was close to that in normoxic control H9C2 cells (SI Appendix, Fig. S32). These results demonstrated that the BMSC-derived EXOs separated by our DNA hydrogel possessed the biological activity of protecting cardiomyocytes from hypoxia damage.

To verify the sustained degradation of DNA hydrogel in vivo, GelRed-stained hydrogel was injected in the thoracic cavity of rat, and the change of fluorescence intensity was monitored. The fluorescence of GelRed on the 4th day decreased to 5.22% of that at the beginning (SI Appendix, Fig. S33). Based on these results, we applied the EXO-contained DNA hydrogel as a therapeutic agent in the rat MI model (Fig. 5A). The rat MI model was established by the ligation of the left anterior descending branch of coronary artery (LAD), and 100 μL DNA hydrogel encapsulating ~2 × 10⁸ BMSC-derived EXOs was injected onto the rat myocardium immediately following LAD ligation. At the 28-d end point, the heart function of rats was analyzed via echocardiography (Fig. 5B). Compared with the rats in normal group, the rats treated with PBS buffer showed significantly increased left ventricular internal dimensions (LVIDs) (SI Appendix, Fig. S34), as well as significantly decreased left ventricular ejection fraction (LVEF) (Fig. 5C) and left ventricular fraction shortening (LVFS) (Fig. 5D), indicating the severe ischemic necrosis of the myocardium. After being treated with EXO-contained hydrogel, the diastolic and systolic LVIDs significantly decreased from 12.13 to 8.78% and from 11.21 to 7.10%, respectively; while LVEF and LVFS significantly increased from 18.80 to 45.01% and from 9.15 to 23.48%, respectively, demonstrating the significant therapeutic effect of improving the functions of the infarcted myocardium by our DNA hydrogel.

To further evaluate the therapeutic effect of DNA hydrogel, histology was performed at the 28-d end point. To quantitively analyze the cardiac collagen morphology, the paraffin sections of rat hearts were treated by Masson’s trichrome staining, in which the areas of collagen fibrosis were stained blue (Fig. 5E). The results showed that the average fractions of the collagen fibrosis area to total left ventricle (LV) area were 0.07% (normal), 22.41% (PBS buffer), 7.30% (only DNA hydrogel) and 2.62% (DNA hydrogel containing BMSC-derived EXOs, hydrogel+EXO), respectively (Fig. 5F), indicating the significant therapeutic effect in blocking collagen fiber accumulation after treatment by our DNA hydrogel. On the other hand, to evaluate the apoptosis of myocardial cells, the paraffin sections of rat hearts were treated by terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL). The groups without EXO treatment (PBS and hydrogel) showed stronger green fluorescence in cell nuclei, indicating a higher level of apoptosis (Fig. 5G). In contrast, the apoptotic cells in the hydrogel+EXO group was 5.46% of total cells, which was significantly lower than that in the PBS group (17.96%) and hydrogel group (16.17%) (Fig. 5H). These results demonstrated that our DNA hydrogel was able to effectively improve heart function and prevent myocardial apoptosis in the rat MI model.

In RCA system, circ-DNA (circ-DNA-1 and circ-DNA-2, respectively, 50 nmol/L), dNTPs (1 mmol/L) (TIANGEN, CD117), bovine serum albumin (0.2 mg/mL) (Solarbio, 9048-46-8), NaCl (80 mmol/L) (YUANLI, 7647-14-5), and phi29 polymerase (200 U/mL) [Biometa Life Science (Ningbo), BMT001L] were mixed in the buffer (50 mM Tris-HCl, 10 mM MgCl₂, 10 mM (NH)₄SO₄, 4 mM dithiothreitol, pH 7.5) [Biometa Life Science (Ningbo), BMT001L] and incubated at 37 °C, 450 rpm. Then the products were incubated at 65 °C for 10 min to inactivate phi29 polymerase. Two RCA products (DNA-chain-1 and DNA-chain-2) were mixed with equal volume at 37 °C, 450 rpm to form DNA hydrogel.

In this experiment, the BMSC-derived EXOs were obtained by ultracentrifugation. The protein concentration of EXOs was measured by the BCA protein quantification kit (Yeasen, 20201ES76*). EXOs with protein concentration of 0.3 mg/mL were stained by 3 μM CM-DiI (Yeasen, 40718ES50) at 37 °C, 450 rpm for 30 min. Then, the EXOs were incubated with 0.1 μM FAM-labeled AptCD63 and FAM-labeled NC for 30 min at 37 °C, respectively. The samples were observed by using an inverted fluorescence microscope (Nikon, Ti-E).

In this experiment, the BMSC EXOs were obtained by ultracentrifugation. DNA-chain-1 (SYBR Green I staining) was incubated with EXOs (CM-DiI staining) at 37 °C for 30 min. DNA-chain-2 was then added and continued to be mixed at 37 °C for 30 min. The captured EXOs and hydrogel were observed by using a fluorescence microscope (Nikon, Ti-E).

DNA hydrogel was degraded by DNase I (Beyotime, BYT-D7073) with different concentration. The degradation rate of DNA hydrogel was calculated by measuring the ssDNA concentration in supernatant with a microvolume spectrophotometer (Quawell, Q5000).

The BMSC EXOs were obtained by ultracentrifugation. In PAGE test, three DSs (10 μM) were mixed with equal amounts (molar quantity) of AptCD63 and AptCD63+EXOs at 37 °C for 30 min under 1.25 mM Mg²⁺, respectively. In fluorescence microscopy imaging, DS₃₂ was added in DNA hydrogel and incubated at 37 °C for 30 min to release EXOs. The final concentration of DS₃₂ was 2.5 μM. The displacement effect was observed by using an inverted fluorescence microscope (Nikon, Ti-E).

Four Cy3-labeled MBs (MB-21, MB-27a, MB-375 and MB-NC) (100 nM) were mixed with three synthetic miRNAs (miR-21, miR-27a, and miR-375, 100 nM) at 37 °C, 450 rpm for 30 min, respectively. The fluorescence intensities of samples were measured by using a microplate reader (BioTek, SYNERGY H1). Excitation/emission: 530 nm/570 nm.

In this experiment, MCF-7 cells were cultured in complete DMEM-H media for 48 h, and the media were then treated by gradient centrifugation at 2,000 × g for 10 min, 10,000 × g for 30 min. In a 100 μL system, 10 μL media were mixed with 100 nM Cy3-labeled MB-21 at 37 °C for 30 min. Cy3-labeled MB-NC was used as control to be mixed with media. The fluorescence intensities of samples were measured by using a microplate reader (BioTek, SYNERGY H1). Excitation: 500 nm; emission: 530 to 600 nm.

In this experiment, the cells (including MCF-7, MDA-MB-231 and BEAS-2B) were cultured in complete cell media for 48 h, and the media were then treated by gradient centrifugation at 2,000 × g for 10 min, 10,000 × g for 30 min. DNA-chain-1 and DNA-chain-2 were synthesized by RCA for 4 h. Phi29 polymerase was inactivated at 65 °C for 10 min. Cy3-labeled MB-21 (final concentration, 100 nM) and FAM-labeled SP (final concentration, 60 nM) were added into DNA-chain-1 and incubated at 37 °C, 450 rpm for 15 min. Then, 7 μL medium was added in 100 μL DNA-chain-1 and incubated at 37 °C, 450 rpm for 30 min. Next, 100 μL DNA-chain-2 was added and incubated at 37 °C, 450 rpm for 30 min to form DNA hydrogel. DNA hydrogel was washed once by PBS buffer. The green fluorescence and red fluorescence of DNA hydrogel were observed by using an inverted fluorescence microscope (Nikon, TI-E). The parameters were consistent under the same channel.

The clinical serum samples gradually melted by gradient warming. DNA-chain-1 and DNA-chain-2 were synthesized by RCA for 4 h. Phi29 polymerase was inactivated at 65 °C for 10 min. Cy3-labeled MB-21 (final concentration, 100 nM) and FAM-labeled SP (final concentration, 60 nM) were added into DNA-chain-1 and incubated at 37 °C, 450 rpm for 15 min. Then, the 10 μL of clinical serum sample was added to 100 μL of DNA-chain-1 and incubated at 37 °C, 450 rpm for 30 min. The 100 μL of DNA-chain-2 was added and incubated at 37 °C, 450 rpm for 30 min to form DNA hydrogel. DNA hydrogel was washed once by PBS buffer. The green fluorescence and red fluorescence of DNA hydrogels were observed by using an inverted fluorescence microscope (Nikon, TI-E). The parameters were consistent under the same channel.

All rats were measured by transthoracic echocardiography under anesthesia after 4 wk by ultrasound imaging system (VisualSonics, Vevo 2100). Hearts were imaged 2D in long-axis views at the level of the greatest left ventricular diameter. LVEF and LVFS were determined by measurement from views taken from the infarcted area.

All rats were sacrificed after treatments for 4 wk. Hearts were harvested and stored in 4% paraformaldehyde (Solarbio, P1110). The paraffin sections of these hearts were treated by Masson’s trichrome stain kit (Solarbio, G1340). Image analysis was performed with ImageJ software.

The paraffin sections of rat hearts were treated by the TUNEL apoptosis detection kit (Alexa Fluor 488) (Yeasen, 40307ES20*).

J.T., X.J., D.Y., and C.Y. designed research; J.T., X.J., Q.L., Z.C., A.L., and B.K. performed research; B.K. contributed new reagents/analytic tools; J.T., X.J., Q.L., Z.C., A.L., D.Y., and C.Y. analyzed data; and J.T., Q.L., D.Y., and C.Y. wrote the paper.

The authors declare no competing interest.