AbstractBiological tissues, such as muscle, can increase their mechanical strength after swelling due to the existence of many biological membrane barriers that can regulate the transmembrane transport of water molecules and ions. Oppositely, typical synthetic materials show a swelling-weakening behavior, which always suffers from a sharp decline in mechanical strength after swelling, because of the dilution of the network. Here, we describe a swelling-strengthening phenomenon of polymer materials achieved by a bioinspired strategy. Liposomal membrane nanobarriers are covalently embedded in a crosslinked network to regulate transmembrane transport. After swelling, the stretched network deforms the liposomes and subsequently initiates the transmembrane diffusion of the encapsulated molecules that can trigger the formation of a new network from the preloaded precursor. Thanks to the tough nature of the double-network structure, the swelling-strengthening phenomenon is achieved to polymer hydrogels successfully. Swelling-triggered self-strengthening enables the development of various dynamic materials. IntroductionBiological tissues can increase their mechanical strength when needed1,2,3, although they are essentially hydrogel materials and are immersed in body fluids4,5. For example, muscles become much stronger after swelling with blood6,7. The contraction of skeletal muscle can activate its associated muscle by increasing blood flow. The specific release of potassium and calcium ions from the activated muscle is able to regulate contractile tension, which can further enhance the hyperemia. Consequently, the enhanced hyperemia results in a large increase in muscle stiffness (Fig.聽1a). Central to the swelling-strengthening nature of these systems is the existence of many biological membrane barriers that can regulate the transmembrane transport of water molecules and ions8,9,10. Namely, selective transport across these barriers during swelling maintains a steady structure of the network, which mainly determines the mechanical strength of the system11,12,13,14. By contrast, synthetic materials present a typical swelling-weakening phenomenon because of the dilution of the network15,16,17,18, which always suffers from a sharp decline in mechanical strength after swelling and largely limits the application, particularly when a given mechanical strength is required, such as biological glues or artificial tissues19,20,21,22,23,24. Although studies have shown that a few specially designed networks with a hydrophilic鈥搇ipophilic balance can resist swelling25,26,27,28, the preparation of polymer materials capable of reinforcing their mechanical strength after swelling remains difficult. The key challenge is that during swelling, the major change that occurs is the network being stretched, and no additional triggers can be exploited to enhance the mechanical strength.Fig. 1: Schematic illustration of the swelling-strengthening behavior. a Biological membrane barriers existed in living tissues, such as muscle. Transmembrane transport that is responsible for the hyperemia-mediated increasement of muscle stiffness depends on the ionic channel, transporters, diffusion, etc. b Variation of mechanical strength after swelling between typical synthetic materials and the swelling-strengthening hydrogels (SSHs) described in this work. c Development of SSHs via a biological membrane barrier-inspired strategy. Transport across the synthetic liposomal membranes mainly relies on diffusion.Full size imageHere, we describe a set of swelling-strengthening hydrogels (SSHs) achieved by virtue of a biological membrane barrier-inspired strategy (Fig.聽1b, c). Liposomal membrane nanobarriers that are covalently embedded in a cross-linked network are used to regulate transmembrane transport. After swelling, the stretched network deforms the embedded membrane nanobarriers and subsequently initiates the transmembrane diffusion of the encapsulated molecules, which can trigger the formation of a new network from the preloaded precursor and form a double-network structure. Thanks to the tough nature of double-network hydrogels, SSHs demonstrate their swelling-strengthening behavior without the assistance of external stimuli and additional additives. Swelling-triggered self-strengthening enables the preparation of various dynamic materials, which may greatly expand their application in physiological environments.Results and discussionDesign and preparation of SSHsTo build an artificial yet robust system that possesses the features of natural membrane barriers, liposomes were embedded within hydrogel materials to simultaneously isolate trigger molecules and regulate their transmembrane transport29,30,31. Our strategy was shown in Fig.聽1c. The liposomes decorated with double bonds were covalently incorporated into the first network by aqueous radical polymerization. The precursor of the second network was preloaded within the hydrogel, and the clickable crosslinker was prestored inside the liposomes, respectively. The swelling-triggered transmembrane transport of the crosslinker led to a catalyst-free yet highly efficient click reaction32,33,34 between the preloaded precursor and the diffused crosslinker, forming the second polymer network. Consequently, the mechanical strength after swelling could exceed the initial SSHs when the increment of the strength resulted from the formation of the double-network structure transcended the strength loss caused by swelling35,36,37,38,39. As a proof-of-concept study, we chose water-soluble 1,11-diazido-3,6,9-trioxaundecane (N3鈥揚EG3鈥揘3) and dibenzocyclooctyne end-capped 4-arm polyethylene glycol (Tetra-PEG-DBCO) as the clickable cross-linker and the precursor of the second network, respectively. Liposomes decorated with double bonds were prepared by co-assembly of hydrogenated soybean phospholipids with cholesterol and distearoyl phosphoethanolamine-PEG2000-acrylamide, while Tetra-PEG-DBCO was synthesized via the amidation of DBCO acid with Tetra-PEG-amine (Supplementary Figs.聽1 and 2). The decorated liposomes loaded with N3鈥揚EG3鈥揘3 (Supplementary Fig.聽3 and Supplementary Table聽1) were covalently linked to the first network by aqueous radical copolymerization of acrylamide and N,N鈥?methylenebisacrylamide. SSHs were obtained by preloading Tetra-PEG-DBCO inside the polyacrylamide (PAM) hydrogel (Supplementary Fig.聽4).Swelling-strengthening behavior of SSHsThe optimized SSHs showed excellent swelling-strengthening behavior. As shown in Fig.聽2a, b and Supplementary Fig.聽5, the compressive modulus of the SSH increased by 15.6%鈥壜扁€?.5 (mean鈥壜扁€塖D, three parallel samples) at a swelling ratio of 25%. Even with the swelling ratio increasing to 75%, the SSH could retain its initial mechanical strength. However, a typical swelling-weakening phenomenon was observed for the control hydrogels. The compressive moduli of both the corresponding PAM single-network and PAM/PEG double-network hydrogels decreased continuously during swelling, which declined by 32.1%鈥壜扁€?.7 and 39.7%鈥壜扁€?.4 at the swelling ratio of 75%, respectively. Further, the compressive stress of the SSH at 90% strain increased from 1.32 to 1.71鈥塎Pa as the swelling ratio increased by 25% (Fig.聽2c, d). With the swelling ratio further increasing up to 75%, the compressive stress was able to maintain at the same level of the gel before swelling. As expected, the compressive stress of both control hydrogels at 90% strain reduced largely and showed a reduction of 52.8%鈥壜扁€?.7 and 57.7%鈥壜扁€?.2 at the swelling ratio of 75%, respectively. In addition, the compressive strength of the swelled PAM/PEG hydrogel was lower than that of SSH. This could be explained by the presence of a large number of crosslinked liposomes in SSH, which increased the mechanical strength of the gel.Fig. 2: Swelling-strengthening behavior of swelling-strengthening hydrogels (SSHs) in vitro.a Compressive modulus and b variation of compressive modulus versus swelling ratio. c Compressive stress and d variation of compressive stress at 90% strain versus swelling ratio. The strain rate was 5鈥塵m/min, and the dimension of the cylindrical samples was 5鈥壝椻€? mm. All error bars represent the mean鈥壜扁€塖D (n鈥?鈥? independent experiments). Significance was assessed using one-way ANOVA with Tukey鈥檚 multiple comparisons test, giving P values, ***P鈥?lt;鈥?.001, ****P鈥?lt;鈥?.0001.Full size imageSSHs were then subcutaneously implanted in the back of rats to investigate the swelling-strengthening behavior in a physiological environment (Fig.聽3a, b and Supplementary Fig.聽6). After implantation in the body, the SSHs were infiltrated by the body fluids (Fig.聽3c, d). Swelling ratios at the predetermined time points were calculated by measuring the weight change of the implanted sample. Similar to the PAM single-network hydrogel, the swelling ratio of the SSH increased by ~50% after implantation for 30鈥塵in, which further increased to ~85% with the time extending to 1鈥塰. Remarkably, the compressive modulus of the SSH grew from 16.2鈥壜扁€?.5 to 18.9鈥壜扁€?.9鈥塎Pa (P鈥?lt;鈥?.01) after the swelling ratio approximated 50% (Fig.聽3e, f). With the swelling degree further increasing to ~85%, the modulus could maintain at 1.09-fold higher than that of the gel before implantation. Oppositely, the compressive modulus of the control PAM gel lowered gradually after implantation, which dropped by 43.3%鈥壜扁€?.8 at the swelling degree of ~85%. Moreover, a similar tendency was observed for the corresponding compressive stress of SSHs at 90% strain (Fig.聽3g, h), demonstrating the swelling-strengthening behavior under a complex in vivo condition. In addition, in vitro cell viability assay indicated negligible cytotoxicity of N3鈥揚EG3鈥揘3 and Tetra-PEG-DBCO (Supplementary Fig.聽7). The treated cells showed almost unaffected viability even with the concentrations of N3鈥揚EG3鈥揘3 and Tetra-PEG-DBCO increasing up to 1 and 60鈥塵g/ml, respectively, which far exceeded their loading contents in SSH. The self-strengthening property of SSHs was further elucidated by mechanical deformation given that the stretched network could deform the embedded liposomal membrane nanobarriers. We speculated that the mechanical deformation of SSHs could similarly trigger the release of the loaded crosslinker and ultimately reinforce the strength. As displayed in Fig.聽4a鈥揹 and Supplementary Figs.聽8 and 9, both the tensile and compressive stress of the SSH climbed significantly after prestretching or precompressing, while the corresponding PAM single-network and PAM/PEG double-network hydrogels weakened apparently. The tensile stress of the SSH enhanced by 37.4%鈥壜扁€?6.5 after prestretching to 100% its initial length for 5鈥塵in. Following prestretching to 200% its initial length, the tensile stress further increased by 51.7%鈥壜扁€?.8. In addition, the compressive stress of the SSH at 70% strain raised by 33.5%鈥壜扁€?1.6 after compression to 70% its initial height for 5鈥塵in, which could enlarge by 47.9%鈥壜扁€?5.1 with the loading time extending to 10鈥塵in. Similar strengthening was observed for both the tensile and compressive moduli of the SSH (Supplementary Fig.聽10). In contrast, the weakening behavior of both control gels was disclosed under the same experimental condition. Representatively, the PAM gel appeared a maximal decrease of 29.8%鈥壜扁€?.1 and 18.0%鈥壜扁€?2.9 in tensile stress and modulus as well as 30.5%鈥壜扁€?.0 and 32.5%鈥壜扁€?.9 in compressive stress and modulus, respectively. We further investigated the influence of the rate of deformation on the mechanical strength of SSH. Interestingly, the tensile stress of the gel increased with the decrease of the rate of deformation, which could be attributed to the insufficient release of N3鈥揚EG3鈥揘3 under a short period of time (Supplementary Fig.聽11).Fig. 3: Swelling-strengthening behavior of swelling-strengthening hydrogels (SSHs) in vivo.a Schematic diagram of the experimental procedure. b Photographs of the surgical implantation of SSHs in rats. Left: incision; middle: implantation; right: conglutination. A roughly 1-cm incision was made in the mediodorsal skin, and a lateral subcutaneous pocket was prepared. c Photographs and d swelling ratio of the hydrogel samples after implantation for the predetermined time points. e Compressive modulus at 90% strain versus swelling ratio. f Variation of compressive modulus versus swelling ratio. g Compressive stress at 90% strain versus swelling ratio. h Variation of compressive stress versus swelling ratio. The strain rate was 5鈥塵m/min, and the dimension of the cylindrical samples was 5鈥壝椻€?鈥塵m. All error bars represent the mean鈥壜扁€塖D (n鈥?鈥? independent experiments). Significance was assessed using unpaired two-tailed Student鈥檚 t test, giving P values, **P鈥?lt;鈥?.01, ***P鈥?lt;鈥?.001. NS no significance.Full size imageFig. 4: Swelling-triggered transmembrane transport and click reaction.a Tensile stress and b variation of tensile stress under different tensile conditions. (1) direct test; (2) test after prestretching at 100% strain for 5鈥塵in; (3) test after prestretching at 200% strain for 5鈥塵in. c Compressive stress and d variation of compressive stress at 70% strain under different compression conditions. (1) direct test; (2) test after compression at 70% strain for 5鈥塵in; (3) test after compression at 70% strain for 10鈥塵in. e Relationship between released yet unreacted N3-PEG3-N3 and swelling ratio of the gels with or without Tetra-PEG-DBCO. f Average size of the liposomes embedded in the gels before and after swelling, respectively. Scanning electron microscopy (SEM) images of the lyophilized swelling-strengthening hydrogels (SSHs) (g) before and (h) after fully swelling. Scale bar, 500鈥塶m. Each test was repeated three times independently with similar results. All error bars represent the mean鈥壜扁€塖D (n鈥?鈥? independent experiments). Significance was assessed using one-way ANOVA with Tukey鈥檚 multiple comparisons test, giving P values, ***P鈥?lt;鈥?.001, ****P鈥?lt;鈥?.0001.Full size imageCriteria required for swelling-strengthening behaviorTo realize swelling-strengthening behavior, the key challenge is to ensure that the mechanical strength of the eventually formed double-network structure is higher than that of the initial gel, which depends on the amount of the diffused crosslinker, the content of the preloaded precursor, and the efficiency of the click reaction. Therefore, achieving swelling-strengthening requires three correlative effects. First, the crosslinker should be triggered to diffuse across the liposomal membrane nanobarriers efficiently by the stretched network after swelling. Second, the released crosslinker should react with the preloaded precursor via a highly efficient catalyst-free reaction. Lastly, the mechanical strength of the formed double-network structure should be optimized to exceed the initial single-network gel. The rational design of SSHs was on the basis of three criteria: (1) the liposomes must have high stability and limited leakage of the loaded crosslinker, while can be deformed to initiate rapid transmembrane transport; (2) the ratio of the diffused crosslinker to the preloaded precursor must be matched to form a crosslinking network by click reaction; (3) both the concentration and crosslinking density of the second network must be optimized to strengthen the ultimate gel.We investigated the effect of liposomal composition on the stability and triggered transmembrane diffusion of the N3鈥揚EG3鈥揘3 loaded liposomes. The liposomes optimized with a component of hydrogenated soybean phospholipids/cholesterol/distearoyl phosphoethanolamine-PEG2000-acrylamide at a molar ratio of 50/45/5 exhibited extraordinary stability and had negligible leakage even with storage time increasing up to 30 days (Supplementary Fig.聽12). The release of N3鈥揚EG3鈥揘3 from the embedded liposomes was initiated by swelling and could increase with the swelling degree of the gel (Fig.聽4e). The release ratio approached 4.8%鈥壜扁€?.7, 7.2%鈥壜扁€?.4, and 13.6%鈥壜扁€?.4 with the swelling degree increasing to 25%, 50%, and 75%, respectively. We speculated that the mechanoresponsive release of N3鈥揚EG3鈥揘3 resulted from the stretched network, which enlarged the embedded liposomes and loosened the lipid bilayer arrangement after swelling. This hypothesis was verified by scanning electron microscopy (SEM) measurements. SEM images validated that the average size of the embedded liposomes increased from 97 to 230鈥塶m after swelling (Fig.聽4f鈥揾). In contrast to dynamic light scattering (DLS) measurement, the decrease of a size determined by SEM could be simply attributed to the dehydration of the liposomes after lyophilization. The triggered release of N3鈥揚EG3鈥揘3 was further confirmed by stretching, and the accumulated release could reach up to 30.9%鈥壜扁€?9.2 after prestretching to 300% its initial length for 5鈥塵in (Supplementary Fig.聽13). Similarly, the liposomes were deformed into a flat shape in the same direction that the gel was stretched (Supplementary Fig.聽14). We then studied the click reaction by preloading with a given amount of Tetra-PEG-DBCO inside the gel. As illustrated in Fig.聽4e, N3鈥揚EG3鈥揘3 reacted with Tetra-PEG-DBCO upon release. The conversion of N3鈥揚EG3鈥揘3 was determined by calculating the ratio of unreacted/released. Importantly, the efficiency of the click reaction remained consistent and could reach up to ~80% during swelling. In view of the high yield of the click reaction, the unattached polymer was statistically negligible as each polymer containing four DBCO reactive groups. A set of hydrogels with various concentrations of the second network were synthesized to further explore the effect of the double-network composition on their strength. All gels were transparent, and no phase separation was observed during the experiments (Supplementary Fig.聽15e). This might be ascribed to the low polymer concentration of the gels, in which the total contents were 20鈥墂t%. The compressive stress at 90% strain increased from 0.89 to 3.43鈥塎Pa, with the loading of the PEG network, increasing to 10鈥墂t% (Supplementary Fig.聽15). With the assistant of these correlative effects, the optimized SSH consisting of 4.5鈥墂t% of Tetra-PEG-DBCO and a feed ratio of 22.4 碌mol/ml of N3鈥揚EG3鈥揘3 successfully achieved the swelling-strengthening behavior. Namely, the increment of the strength resulted from the formation of the double-network structure exceeded the loss of strength at a swelling degree of 75%. Taken together, on the basis of the kinetics of both the transmembrane diffusion and the click reaction at a given swelling degree, the mechanical strength of SSHs after swelling could be regulated to accomplish the swelling-strengthening phenomenon by adjusting the loading of the second network.Verifying the formation of the second networkTo further confirm the structure of the second network responsible for the mechanical behavior of SSHs, we designed and synthesized a triggerable PAM network that could enable a comprehensive characterization of the newly formed network (Fig.聽5a). Namely, the first network was replaced by a disulfide-crosslinked PAM network, which was reduction-responsive. Therefore, the initial network could be dissolved by glutathione (GSH), a reducing agent, after the formation of the double-network structure. Once the first network was removed, the newly formed network could be obtained. As shown in Fig.聽5b, the disulfide-crosslinked PAM network was triggered to dissolution by GSH. As expected, the swelled hydrogel could retain its initial solid shape after treating with GSH (Fig.聽5b鈥揺), verifying the formation of the second network. The purified second network was obtained by complete removal of the de-crosslinked PAM via diffusion. Mechanical testing showed that the formed second network had a tensile stress of 28.4鈥塳Pa at 600% strain and compressive stress of 915鈥塳Pa at 90% strain, respectively (Fig.聽5f, g). SEM images evidenced the morphologic difference between the double-network and the second network hydrogels (Fig.聽6a, b). Peaks at 61.4 ppm (assigned to 鈥揘CH2CH2O鈥? and 57.7 ppm (assigned to 鈥揘CH2CH2O鈥? in the solid-state 13C nuclear magnetic resonance (13C NMR) spectrum reflected that the second network was crosslinked by N3鈥揚EG3鈥揘3 via click reaction (Fig.聽6c). Meanwhile, in comparison with the solid-state 13C NMR spectrum of the PAM network40, the disappearance of peaks around 183.1 ppm (assigned to C鈥?鈥塐) and 42.3 ppm (assigned to 鈥?i>CH鈥?i>CH2鈥? suggested the removal of the primary network. Absorbances of the ring stretching and plane bending of 1,2,3-triazole unit at 1450, 1250, 1050, and 750鈥塩m鈭? in the Fourier-transform infrared (FTIR) spectrum further demonstrated the crosslinking of the precursor by click reaction (Fig.聽6d). The absence of a signal at 3340 and 3168鈥塩m鈭? that was assigned to the N鈥揌 stretching of the amide group in the PAM network further demonstrated the disappearance of the primary network (Supplementary Fig.聽16)41. The absorbance peak at 3400鈥塩m鈭? revealed that the two networks were intertwined by hydrogen bonds. These results well confirmed the mechanical behavior, microscopic morphology, and molecular structure of SSHs.Fig. 5: Design, preparation, and mechanical testing of the second network.a Schematic illustration for the preparation of the second network via GSH-triggered dissolution of the disulfide-crosslinked polyacrylamide (PAM) network. The reduction-responsive PAM network was synthesized by using N,N鈥?bis(acryloyl)cystamine as a crosslinker. b鈥?b>e Photographs of different samples. b PAM single-network hydrogel (left) and triggerable swelling-strengthening hydrogels (SSH) (right) samples after soaking in 250鈥塵M GSH for 1 day. c A triggerable SSH strip stretched to its maximal tensile strain. d A triggerable SSH strip (25鈥壝椻€?鈥壝椻€?.5鈥塵m) bearing a load of 100鈥塯. e A triggerable SSH cylinder compressed to 90% strain and then released. Hydrogel samples were fully hydrated throughout the experiment. Curves of (f) tensile and (g) compressive stress versus strain of the purified second network hydrogels.Full size imageFig. 6: Physical characterization of the second network.Scanning electron microscopy (SEM) images of the triggerable swelling-strengthening hydrogels (SSHs) (a) before and (b) after treating with GSH. Samples were lyophilized before SEM measurement. Scale bar, 50鈥壩糾. Each test was repeated three times independently with similar results. c Solid-state 13C NMR spectra of the triggerable SSHs and the purified second network. The samples were lyophilized before analysis. d Fourier-transform infrared (FTIR) spectra of N3-PEG3-N3, blank liposomes, PEG hydrogel, and the purified second network.Full size imageIn summary, we report the swelling-strengthening behavior of synthetic hydrogel materials achieved by a biological membrane barrier-inspired strategy. The designed SSHs automatically switch from a single-network to a double-network structure via a catalyst-free click reaction without the help of external triggers and present increased mechanical strength after swelling. Central to the swelling-strengthening nature of SSHs is the existence of many artificial biological membrane nanobarriers that enable a swelling-triggered transmembrane transport. In addition to this swelling-strengthening property, other spontaneous and dynamic behaviors such as swelling-induced destroying, color-switching, or biological reaction could be incorporated into synthetic hydrogels by the isolation of the corresponding trigger molecules, studies that are currently ongoing.MethodsMaterialsHydrogenated soybean phospholipids (HSPC), cholesterol (Chol), and distearoyl phosphoethanolamine-PEG2000-acrylamide (DSPE-PEG2000-ACA) were purchased from Shanghai Ponsure Biotech, Inc. Four-arm poly(ethylene glycol) amine (Tetra-PEG-amine) was provided by Xiamen Sinopeg company. Dibenzocyclooctyne-acid (DBCO acid, 鈮?5%) was supplied by Xi鈥檃n ruixi Biological Technology Co., Ltd. O-(7-azabenzotriazol-1-yl)-N,N,N鈥?N鈥?tetramethyluronium hexafluorophosphate (HATU, 99%) and glutathione (GSH, 99%) were purchased from J K Scientific Ltd. N,N-diisopropylethylamine (鈮?9%, DIEA), N,N鈥?methylenebisacrylamide (99%, MBAA), N,N,N鈥?N鈥?tetramethylethylenediamine (99%, TEMED), and anhydrous N,N-dimethylformamide (99.8%, DMF) were provided by Sigma. Ammonium persulfate (99.99%, APS) and 1,11-diazido-3,6,9-trioxaundecane (鈮?5%, N3-PEG3-N3) were supplied by Aladdin. Acrylamide (99%, AM) was purchased from Adamas. Ethyl ether (鈮?9.7%, Et2O) and chloroform (鈮?9%, CHCl3) were provided by Sinopharm chemical reagent Co., Ltd. Ammonium sulfate (鈮?9%, (NH4)2SO4) and PBS buffer (10脳, pH 7.4) were supplied by Sangon Biotech company. The PBS buffer used in all experiments was diluted ten times. Deionized water was used to prepare all aqueous solutions. Other reagents and solvents were used as received.Synthesis and characterization of Tetra-PEG-DBCOThe preloaded precursor Tetra-PEG-DBCO was synthesized by a one-step amidation reaction between DBCO acid and Tetra-PEG-amine42 (Supplementary Fig.聽1). DBCO acid (171鈥塵g, 0.56鈥塵mol), HATU (213鈥塵g, 0.56鈥塵mol), and 3鈥塵l anhydrous DMF were added into a dry flask to give a light-yellow solution under stirring. Tetra-PEG-amine (Mn = 40鈥塳Da, 2.8鈥塯, 0.07鈥塵mol) in 12鈥塵l of anhydrous DMF was then added at 0鈥壜癈. After the addition of 200鈥壩糒of anhydrous DIEA, the reaction mixture was stirred at 25鈥壜癈 for 36鈥塰. Afterward, the mixture was dropwise added into cold Et2O with stirring to afford a white solid powder. The product was further washed by Et2O for three times and dried under vacuum to give a light-yellow powder (10.5鈥塯, yield 91.3%).The structure of the obtained compound was confirmed by proton nuclear magnetic resonance (1H NMR) spectrum (Bruker Avance 500 spectrometer). Chemical shifts (未) were expressed in ppm. According to 1H NMR (Supplementary Fig.聽2), the end-functionalization was 91.2%. 1H NMR (400鈥塎Hz, CDCl3): 7.67鈥?.32 (m, 32H), 6.19 (br, 4H), 5.17 (d, J鈥?鈥?3.6鈥塇z, 4H), 3.86鈥?.31 (m, 3678H), 2.80鈥?.93 (m, 16H) ppm.Preparation of NALipThe N3鈥揚EG3鈥揘3-loaded acrylamide decorated liposomes (NALip) were prepared via the conventional thin-film hydration method43. Briefly, HSPC/Chol/DSPE-PEG2000-ACA (molar ratio: 50/45/5) was added in a round-bottom flask and dissolved by CHCl3. Lipid films were acquired after removing CHCl3 by rotary evaporation. Then the lipids were hydrated by (NH4)2SO4 aqueous solution (155鈥塵M) at 50鈥壜癈 under ultrasonic vibration for 20鈥塵in. The generated white solution was extruded through polycarbonate membranes (pore size: 200鈥塶m) with the help of a mini extruder (Avavti Polar Lipids, USA) to homogenize the size of liposomes44. Acrylamide decorated liposomes (ALip) were obtained after removing (NH4)2SO4 solution through ultrafiltration and further dispersed in 1脳 PBS solution.N3鈥揚EG3鈥揘3 was incubated together with the obtained ALip in an N3鈥揚EG3鈥揘3/HSPC molar ratio of 3/1 at 56鈥壜癈 for 20鈥塵in. Finally, the resultant NALip was purified by repeated ultrafiltration to remove the unencapsulated N3鈥揚EG3鈥揘3, and further decentralized into 1脳 PBS solution.Morphology and size measurements of ALipThe morphology of ALip was observed by transmission electron microscopy (TEM, HT7700 Exalens, Hitachi, Japan) at an accelerating voltage of 10鈥塳V. In total, 10鈥壩糒 of ALip dilute solution was dripped on the surface of a carbon-coated copper grid (400 mesh). After deposition for 5鈥塵in, the excess solution was sucked away by filter paper. Then ultrapure water was repeatedly added to the surface of the grid and absorbed by filter paper to remove the residual salts. The obtained grid was negatively stained by uranyl acetate solution (1鈥墂t%, 5鈥壩糒) and dried completely prior to the test. The size and polydispersity index (PDI) of ALip were determined by DLS (Malvern Zetasizer Nano ZSP). In all, 2鈥塵l of ALip dilute solution was added into a quartz cuvette for the DLS measurements. Samples were tested in parallel for three times. TEM (Supplementary Fig.聽3a) and DLS (Supplementary Fig.聽3b) results show that the prepared ALip presented a uniform size with an average diameter of 173.9鈥壜扁€?.01鈥塶m.Quantification of entrapment efficiency of N3鈥揚EG3鈥揘3 in NALipThe entrapment efficiency of N3鈥揚EG3鈥揘3 was evaluated in the virtue of ultraviolet (UV) spectrophotometer. The standard curve of absorbance versus concentration of N3鈥揚EG3鈥揘3 was obtained at an absorption wavelength of 278鈥塶m (Supplementary Fig.聽3c). After incubation with ALip, a portion of N3鈥揚EG3鈥揘3 was encapsulated in Alip to acquire NALip, unencapsulated N3鈥揚EG3鈥揘3 was completely transferred to the subnatant by ultrafiltration. The amount of unencapsulated N3鈥揚EG3鈥揘3 in the subnatant can be calculated according to the standard curve. Finally, the encapsulated ratio and loading amount of N3鈥揚EG3鈥揘3 in NALip can be obtained. The characteristic peaks of N3鈥揚EG3鈥揘3, lyophilized NALip, and ALip were recorded via FTIR spectroscopy to verify the successful encapsulation of N3鈥揚EG3鈥揘3 in ALip (Supplementary Fig.聽3d). The molar feed ratio of N3鈥揚EG3鈥揘3 to HSPC showed a significant impact on the entrapment efficiency of N3鈥揚EG3鈥揘3 in the process of incubation. The loading amount and encapsulation efficiency of N3鈥揚EG3鈥揘3 in NALip under different feed ratios are listed in Supplementary Table聽S1. The optimized ratio of 3:1 was employed in the preparation of NALip for the following experiments.Preparation of SSHs, PAM, and PAM/PEG hydrogelsUsing concentrated NALip solution as a solvent, AM (284鈥塵g, 4鈥塵mol), crosslinker MBAA (40鈥壩糒, 0.1鈥塎, 4鈥壩糾ol), initiator APS (5鈥?w/w), catalyst TEMED (5鈥?w/w), and Tetra-PEG-DBCO were dissolved to give a pre-gel solution at room temperature in a nitrogen glove box. After 30鈥塵in polymerization in PTFE molds, SSHs were obtained and ready for further measurements. Noted that the amount of Tetra-PEG-DBCO depended on the encapsulation efficiency and released ratio of N3鈥揚EG3鈥揘3 in NALip. Tetra-PEG-DBCO was added in the stoichiometric proportion (molar ratio, 1:1) according to the release ratio of N3鈥揚EG3鈥揘3. Both PAM and PAM/PEG hydrogels were similarly prepared as controls. Briefly, PAM was prepared by mixing AM, MBAA, APS, and TEMED in aqueous solution, and PAM/PEG was synthesized via blending AM, MBAA, APS, TEMED, Tetra-PEG-DBCO, and free N3-PEG3-N3 in ultrapure water. The photographs of the above hydrogels are shown in Supplementary Fig.聽4.Swelling ratio (SR) of the hydrogelsHydrogels were weighed immediately after polymerization to determine the weight (WD), and then incubated in a predetermined amount of ultrapure water (WH) at 37鈥壜癈 for absorption. A period of 8鈥塰 was taken to reach a complete swelling ratio of 25, 50, and 75%. SR of the hydrogels was calculated according to the following Eq (1)45.$${mathrm{SR}} = W_{mathrm{H}}/W_D.$$ In vivo swelling experimentsAll the animal procedures complied with the guidelines of the Shanghai Medical Experimental Animal Care. Animal protocols were approved by the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University School of Medicine. Male rats (250鈥?00鈥塯) were used for in vivo swelling studies. A roughly 1-cm incision was made in the mediodorsal skin, and a lateral subcutaneous pocket was prepared. The SSH and PAM samples (n鈥?鈥?; 5鈥壝椻€?-mm cylinders) were implanted inside, and the incision was bonded with the help of tissue glue under sterile conditions. At designated time intervals (0.5鈥塰 and 1鈥塰), the rats were sacrificed, and then the implanted samples were retrieved for the subsequent compression tests.In vitro cell viability assayA cell counting kit-8 (CCK-8, Beyotime Institute of Biotechnology, China) was used to investigate the cytotoxicity of N3鈥揚EG3鈥揘3 and Tetra-PEG-DBCO against 4T1 cells. Briefly, cells were seeded at a density of 1鈥壝椻€?04 cells per well in a 96-well plate, and cultured overnight in RMPI 1640 (Hyclone, Thermo Scientific) supplemented with 10% fetal bovine serum (Hyclone, Thermo Scientific) and 1% penicillin streptomycin (GIBICO, Invitrogen) at 37鈥壜癈 in a humidified atmosphere containing 5% CO2. Subsequently, 100鈥壩糽 of fresh medium containing N3鈥揚EG3鈥揘3 at different concentrations (100, 500, or 1000鈥壩糶/ml) or Tetra-PEG-DBCO at different concentrations (6000, 10000, or 60000鈥壩糶/ml) was added to replace the culture medium for 24 and 48鈥塰, respectively. After incubation, the culture medium containing N3鈥揚EG3鈥揘3 or Tetra-PEG-DBCO was removed, 100鈥壩糽 of fresh medium and 10鈥壩糽 of CCK-8 solution were added to each well. After incubation for 3鈥塰, the OD value of cultures was recorded at 450鈥塶m with the help of a Multi-Detection Microplate reader (BioTek, USA). The well with medium and CCK-8 solution but without cells, N3鈥揚EG3鈥揘3, and Tetra-PEG-DBCO was used as a blank group. The well with medium, cells, and CCK-8 solution but without N3鈥揚EG3鈥揘3 and Tetra-PEG-DBCO was employed as a control group. The experiment was replicated for four parallel samples.Mechanical characterizationAll the mechanical measurements were conducted on the universal material testing machine (Instron-3342, 50鈥塏 sensor) and repeated three times. Compression tests after swelling: The compression tests were carried out for samples with different swelling ratios. The strain rate was set at 5鈥塵m/min, and the dimension of the cylindrical samples was 5鈥壝椻€?鈥塵m. The compression strain was set at 90%46. Tensile tests: The tensile properties were evaluated under three different tensile conditions. Test 1 was conducted directly. Test 2 was carried out after prestretching the sample at 100% strain for 5鈥塵in. Test 3 was initiated after prestretching the hydrogel at a strain of 200% for 5鈥塵in. The strain rate was set at 20鈥塵m/min, and the dimension of the rectangle samples was 25鈥壝椻€?鈥壝椻€?.5鈥塵m. Repeated compression tests: Three compression tests were carried out on an identical sample under different compression conditions. Test 1 was conducted directly. Test 2 was carried out after compression at 70% strain for 5鈥塵in. Test 3 was initiated after compression at 70% strain for 10鈥塵in. Cyclic tensile tests: Cyclic tensile tests were conducted to 250% strain at different tensile rates (5, 50, and 200鈥塵m/min), and the loading and unloading processes performed at the same rate. The tests were performed immediately following the initial loading for three times.SEM analysisThe gross morphology of lyophilized SSH was observed via the Raman imaging combined with emission scanning electron microscopy (RI-SEM, TESCAN-MAIA3) under an accelerating voltage of 5鈥塳V. The samples were completely swelled or stretched to 3 times their initial length before lyophilization. All the samples were gold-sputtered prior to the tests.Stability of NALipThe stability of NALip was evaluated by detecting the leakage of N3鈥揚EG3鈥揘3 encapsulated in NALip. The concentrated NALip solution (0.0224鈥壩糓) was stored in the dark at 4鈥壜癈 for 30 days and then diluted by 1鈥壝椻€塒BS for ultrafiltration. The amount of N3鈥揚EG3鈥揘3 in the subnatant was calculated by a UV spectrophotometer. There was no precipitation at the bottom of the vessel on the 30th day, and no N3鈥揚EG3鈥揘3 was detected in the subnatant.Quantification of the transmembrane transport and click reactionSwelling-triggered transmembrane transport: The sample without Tetra-PEG-DBCO was employed to determine the transmembrane transport of the encapsulated N3鈥揚EG3鈥揘3. The hydrogels were put in a given mass of ultrapure water at 37鈥壜癈 to reach different swelling ratios. Then they were immersed in 1鈥塵l of ultrapure water for 10鈥塵in. The concentration of N3鈥揚EG3鈥揘3 in the incubation solution was determined with the help of the UV standard curve. Swelling-triggered click reaction: The sample with Tetra-PEG-DBCO was utilized to demonstrate that catalyst-free click reaction occurred between the released N3鈥揚EG3鈥揘3 and the preloaded Tetra-PEG-DBCO. The experimental procedure was similar to the above Swelling-triggered transmembrane transport test. Tensile-triggered transmembrane transport: The sample without Tetra-PEG-DBCO was used to determine the transmembrane transport of N3鈥揚EG3鈥揘3 under different stretching strain and time. The samples were immersed in 1鈥塵l of water for 10鈥塵in after stretching, then the concentration of the released N3鈥揚EG3鈥揘3 was measured by three parallel tests.Preparation and characterization of triggerable SSHsA disulfide crosslinker, N,N鈥?bis(acryloyl)cystamine, was synthesized via nucleophilic substitution reaction between cystamine dihydrochloride and acryloyl chloride according to the reported method47. Triggerable SSHs were synthesized by using N,N鈥?bis(acryloyl)cystamine as a crosslinker. The resulted SSHs were first incubated in a predetermined amount of ultrapure water at 37鈥壜癈 until the samples were swelled completely, followed by soaking in 250鈥塵M GSH solution for ~3 days. After the dissolution of the reduction-responsive PAM network, the newly formed PEG network was purified by diffusion. The obtained samples were tailored to strips (20鈥壝椻€?鈥壝椻€?.5鈥塵m) and cylinders (7鈥壝椻€?鈥塵m) for mechanical testing. Lyophilized hydrogel samples were used for SEM, FTIR, and solid-state 13C NMR (Bruker AVANCE NEO, 600鈥塎Hz WB) measurements.Statistical analysisAll error bars represent the mean鈥壜扁€塻.d. A significant difference was assessed by Student鈥檚 t test between two groups or one-way analysis of variance with Tukey鈥檚 multiple comparisons test among three groups, giving P values, *P鈥?lt;鈥?.05, **P鈥?lt;鈥?.01, ***P鈥?lt;鈥?.001 and ****P鈥?lt;鈥?.0001. NS no significance.Reporting summaryFurther information on research design is available in the聽Nature Research Reporting Summary linked to this article. All data presented in the paper are available from the authors upon reasonable request. References1.van Oosten, A. S. G. et al. Emergence of tissue-like mechanics from fibrous networks confined by close-packed cells. Nature 573, 96鈥?01 (2019).ADS聽 PubMed聽Google Scholar聽 2.Wang, N., Butler, J. P. Ingber, D. E. Mechanotransduction across the cell surface and through the cytoskeleton. Science 260, 1124鈥?127 (1993).ADS聽 CAS聽 PubMed聽Google Scholar聽 3.Fung, Y. C. Biomechanics: Mechanical Properties of Living Tissues (Spinger Science Business Media, 2013).4.Burdick, J. A. Murphy, W. L. Moving from static to dynamic complexity in hydrogel design. Nat. Commun. 3, 1269 (2012).ADS聽 PubMed聽Google Scholar聽 5.Jatav, V. S., Singh, H. Singh, S. K. Recent trends on hydrogel in human body. Int. J. Res. Pharm. Biomed. Sci. 2, 442鈥?47 (2011). Google Scholar聽 6.Burns, W. R., Cohen, K. D. Jackson, W. F. K+-induced dilation of hamster cremasteric arterioles involves both the Na+/K+-ATPase and inward-rectifier K+ channels. Microcirculation 11, 279鈥?93 (2004).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 7.Laughlin, M. H. Korzick, D. H. Vascular smooth muscle: integrator of vasoactive signals during exercise hyperemia. Med. Sci. Sport. Exerc. 33, 81鈥?1 (2001).CAS聽Google Scholar聽 8.Gouaux, E. MacKinnon, R. Principles of selective ion transport in channels and pumps. Science 310, 1461鈥?465 (2005).ADS聽 CAS聽 PubMed聽Google Scholar聽 9.Lucy, J. A. The fusion of biological membranes. Nature 227, 815鈥?17 (1970).ADS聽 CAS聽 PubMed聽Google Scholar聽 10.Chernomordik, L. V. Kozlov, M. M. Mechanics of membrane fusion. Nat. Struct. Mol. Biol. 15, 675鈥?83 (2008).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 11.Lahiri, S., Toulmay, A. Prinz, W. A. Membrane contact sites, gateways for lipid homeostasis. Curr. Opin. Cell Biol. 33, 82鈥?7 (2015).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 12.Holthuis, J. C. M. Menon, A. K. Lipid landscapes and pipelines in membrane homeostasis. Nature 510, 48鈥?7 (2014).ADS聽 CAS聽 PubMed聽Google Scholar聽 13.Wenk, M. R. The emerging field of lipidomics. Nat. Rev. Drug Discov. 4, 594鈥?10 (2005).CAS聽 PubMed聽Google Scholar聽 14.Shevchenko, A. Simons, K. Lipidomics: coming to grips with lipid diversity. Nat. Rev. Mol. Cell Biol. 11, 593鈥?98 (2010).CAS聽 PubMed聽Google Scholar聽 15.Mahinroosta, M., Farsangi, Z. J., Allahverdi, A. Shakoori, Z. Hydrogels as intelligent materials: a brief review of synthesis, properties and applications. Mater. Today Chem. 8, 42鈥?5 (2018).CAS聽Google Scholar聽 16.Zhang, Y., Li, Y. Liu, W. Dipole鈥揹ipole and H鈥恇onding interactions significantly enhance the multifaceted mechanical properties of thermoresponsive shape memory hydrogels. Adv. Funct. Mater. 25, 471鈥?80 (2015).CAS聽Google Scholar聽 17.Alkayyali, L. B., Abu-Diak, O. A., Andrews, G. P. Jones, D. S. Hydrogels as drug-delivery platforms: physicochemical barriers and solutions. Ther. Deliv. 3, 775鈥?86 (2012).CAS聽 PubMed聽Google Scholar聽 18.Macdougall, L. J., Perez-Madrigal, M. M., Arno, M. C. Dove, A. P. Nonswelling thiol鈥搚ne cross-linked hydrogel materials as cytocompatible soft tissue scaffolds. Biomacromolecules 19, 1378鈥?388 (2018).CAS聽 PubMed聽Google Scholar聽 19.Taboada, G. M. et al. Overcoming the translational barriers of tissue adhesives. Nat. Rev. Mater. 5, 310鈥?29 (2020).ADS聽Google Scholar聽 20.Han, L. et al. Tough, self-healable and tissue-adhesive hydrogel with tunable multifunctionality. NPG Asia Mater. 9, e372 (2017).CAS聽Google Scholar聽 21.Chang, Y. et al. Tunable bioadhesive copolymer hydrogels of thermoresponsive poly(N-isopropyl acrylamide) containing zwitterionic polysulfobetaine. Biomacromolecules 11, 1101鈥?110 (2010).PubMed聽Google Scholar聽 22.Sui, Z. J., King, W. J. Murphy, W. L. Protein-based hydrogels with tunable dynamic responses. Adv. Funct. Mater. 18, 1824鈥?831 (2008).CAS聽Google Scholar聽 23.Yoshikawa, H. Y. et al. Quantitative evaluation of mechanosensing of cells on dynamically tunable hydrogels. J. Am. Chem. Soc. 133, 1367鈥?374 (2011).CAS聽 PubMed聽Google Scholar聽 24.Vats, K. Benoit, D. S. W. Dynamic manipulation of hydrogels to control cell behavior: a review. Tissue Eng. Part B Rev. 19, 455鈥?69 (2013).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 25.Kamata, H., Akagi, Y., Kayasuga-Kariya, Y., Chung, U. Sakai, T. 鈥淣onswellable鈥?hydrogel without mechanical hysteresis. Science 343, 873鈥?75 (2014).ADS聽 CAS聽 PubMed聽Google Scholar聽 26.Kamata, H., Kushiro, K., Takai, M., Chung, U. Sakai, T. Non鈥恛smotic hydrogels: a rational strategy for safely degradable hydrogels. Angew. Chem. Int. Ed. 55, 9282鈥?286 (2016).CAS聽Google Scholar聽 27.O鈥橲hea, T. M., Aimetti, A. A., Kim, E., Yesilyurt, V. Langer, R. Synthesis and characterization of a library of in-situ curing, nonswelling ethoxylated polyol thiol-ene hydrogels for tailorable macromolecule delivery. Adv. Mater. 27, 65鈥?2 (2015).PubMed聽Google Scholar聽 28.Calvert, P. Hydrogels for soft machines. Adv. Mater. 21, 743鈥?56 (2009).ADS聽 CAS聽Google Scholar聽 29.Grijalvo, S., Mayr, J., Eritja, R. Diaz, D. D. Biodegradable liposome-encapsulated hydrogels for biomedical applications: a marriage of convenience. Biomater. Sci. 4, 555鈥?74 (2016).CAS聽 PubMed聽Google Scholar聽 30.Lyu, D., Chen, S. Guo, W. Liposome crosslinked polyacrylamide/DNA hydrogel: a smart controlled鈥恟elease system for small molecular payloads. Small 14, e1704039 (2018).PubMed聽Google Scholar聽 31.Gao, W. et al. Hydrogel containing nanoparticle-stabilized liposomes for topical antimicrobial delivery. ACS Nano 8, 2900鈥?907 (2014).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 32.Truong, V. X. et al. In situ-forming robust chitosan-poly (ethylene glycol) hydrogels prepared by copper-free azide-alkyne click reaction for tissue engineering. Biomater. Sci. 2, 167鈥?75 (2014).CAS聽 PubMed聽Google Scholar聽 33.Zheng, J. et al. Strain-promoted cross-linking of PEG-based hydrogels via copper-free cycloaddition. ACS Macro Lett. 1, 1071鈥?073 (2012).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 34.Takahashi, A. et al. In situ cross-linkable hydrogel of hyaluronan produced via copper-free click chemistry. Biomacromolecules 14, 3581鈥?588 (2013).CAS聽 PubMed聽Google Scholar聽 35.Gong, J. P., Katsuyama, Y., Kurokawa, T. Osada, Y. Double鈥恘etwork hydrogels with extremely high mechanical strength. Adv. Mater. 15, 1155鈥?158 (2003).CAS聽Google Scholar聽 36.Zhao, Y. et al. Proteoglycans and glycosaminoglycans improve toughness of biocompatible double network hydrogels. Adv. Mater. 26, 436鈥?42 (2014).PubMed聽Google Scholar聽 37.Nakajima, T., Kurokawa, T., Ahmed, S., Wu, W. Gong, J. P. Characterization of internal fracture process of double network hydrogels under uniaxial elongation. Soft Matter. 9, 1955鈥?966 (2013).ADS聽 CAS聽Google Scholar聽 38.Jia, X. Kiick, K. L. Hybrid multicomponent hydrogels for tissue engineering. Macromol. Biosci. 9, 140鈥?56 (2009).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 39.Matsuda, T., Kawakami, R., Namba, R., Nakajima, T. Gong, J. P. Mechanoresponsive self-growing hydrogels inspired by muscle training. Science 363, 504鈥?08 (2019).ADS聽 CAS聽 PubMed聽Google Scholar聽 40.Garces, F. O., Sivadasan, K., Somasundaran, P. Turro, N. J. Interpolymer complexation of poly (acrylic acid) and polyacrylamide: structural and dynamic studies by solution-and solid-state NMR. Macromolecules 27, 272鈥?78 (1994).ADS聽 CAS聽Google Scholar聽 41.Adhikary, P., Tiwari, K. N. Singh, R. P. Synthesis, characterization, and flocculation characteristics of polyacrylamide鈥恎rafted glycogen. J. Appl. Polym. Sci. 103, 773鈥?78 (2007).CAS聽Google Scholar聽 42.Chen, M. et al. Living additive manufacturing: transformation of parent gels into diversely functionalized daughter gels made possible by visible light photoredox catalysis. ACS Cent. Sci. 3, 124鈥?34 (2017).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 43.Liang, Y. Kiick, K. L. Liposome-cross-linked hybrid hydrogels for glutathione-triggered delivery of multiple cargo molecules. Biomacromolecules 17, 601鈥?14 (2016).CAS聽 PubMed聽 PubMed Central聽Google Scholar聽 44.Hu, C. J. et al. Erythrocyte membrane-camouflaged polymeric nanoparticles as a biomimetic delivery platform. Proc. Natl Acad. Sci. USA 108, 10980鈥?0985 (2011).ADS聽 CAS聽 PubMed聽Google Scholar聽 45.Zhao, X. et al. Khademhosseini, photocrosslinkable gelatin hydrogel for epidermal tissue engineering. Adv. Healthc. Mater. 5, 108鈥?18 (2016).CAS聽 PubMed聽Google Scholar聽 46.Gao, F. et al. Direct 3D printing of high strength biohybrid gradient hydrogel scaffolds for efficient repair of osteochondral defect. Adv. Funct. Mater. 28, 1706644 (2018). Google Scholar聽 47.Pan, Y. J. et al. Redox/pH dual stimuli-responsive biodegradable nanohydrogels with varying responses to dithiothreitol and glutathione for controlled drug release. Biomaterials 33, 6570鈥?579 (2012).CAS聽 PubMed聽Google Scholar聽 Download referencesAcknowledgementsThis work was financially supported by the National Natural Science Foundation of China (21875135), the Recruitment Program of Global Youth Experts of China (D1410022), the Shanghai Municipal Education Commission-Gaofeng Clinical Medicine Grant Support (20181704, 20191820), and the Innovative research team of high-level local universities in Shanghai (SSMU-ZLCX20180701).Author informationAffiliationsShanghai Key Laboratory for Nucleic Acid Chemistry and Nanomedicine, Institute of Molecular Medicine, State Key Laboratory of Oncogenes and Related Genes, Shanghai Cancer Institute, Renji Hospital, School of Medicine, Shanghai Jiao Tong University, 200127, Shanghai, ChinaFeng Wu聽 聽Jinyao LiuDepartment of Ophthalmology, Shanghai Ninth People鈥檚 Hospital, School of Medicine, Shanghai Jiao Tong University, 200011, Shanghai, ChinaYan PangAuthorsFeng WuView author publicationsYou can also search for this author in PubMed聽Google ScholarYan PangView author publicationsYou can also search for this author in PubMed聽Google ScholarJinyao LiuView author publicationsYou can also search for this author in PubMed聽Google ScholarContributionsJ.L. and Y.P. conceived and designed the experiments. F.W. performed all experiments. All authors analyzed and discussed the data. F.W. and J.L. wrote the paper.Corresponding authorsCorrespondence to Yan Pang or Jinyao Liu.Ethics declarations Competing interests The authors declare no competing interests. Additional informationPeer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.Publisher鈥檚 note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary information CommentsBy submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate. Sign up for the Nature Briefing newsletter 鈥?what matters in science, free to your inbox daily.