Fabrication and characterization of NC@C
A composite conduit (NC@conduit, NC@C) integrating features of mesoporous carbon sheets and protein-permeable elastomers to promote peripheral nerve regeneration under magnetoelectric stimulation was developed in this study. As shown in Fig. 1a, a bioink composed of mesoporous carbon layers on a GO sheet (sandwich NCs) and 4-HBA/PU-EO-PO monomer was placed in a 3D printer resin bath. Then, UV light possessing a channel pattern in a ring was applied to cure the bioink to form a flexible composite NC@C with internal channels through DLP 3D printing. As shown in Fig. 1b, 3D printed NC@C loaded with a large payload of NGF was implanted to implement the spatial distribution of NGF for the promotion of axon outgrowth after a peripheral nerve was transected 10 mm from the bifurcation point of the sciatic nerve of SD rats. Upon receiving MF irradiation, the NFG encapsulated in NC@C showed excellent permeability and on-demand release to induce the differentiation and proliferation of nerve cells in vivo, while simultaneously supplying electromagnetic stimulation to cells. Via UV pattern design, conduits with various channels can be prepared and loaded with drugs or proteins in the matrix (Supplementary Fig. S1). After embedding NCs, the printed conduit and composite maintained good elastic and stretchable properties (Fig. 1c, d). A particular success of this strategy is the controllable roughness of the surface on a conduit by exposing NCs to 3D printing, which not only increases the roughness for cell adhesion, but also manifests as a direct physical stimulus to cells (Fig. 1e). Furthermore, through exquisite fabrication, the microchannels of NC conduits effectively promote the cell alignment (Fig. 1f).
Synthesis and characterization of NCs
The robust composite NCs can be fabricated by employing GO as the substrate of the structure, as schematically illustrated in Fig. 2a. To prepare GO, the modified Hummers method was used. Similar to our previous work, a cationic surfactant (CTAB) was adsorbed onto the surface of synthesized GO via electrostatic interactions to form the micelle template. Under weakly basic conditions, the silica pore was constructed via the hydrolysis of tetraethyl orthosilicate and polymerization of styrene on the packed CTAB micelles on GO. Then, the particles were carbonized at high temperature under nitrogen to observe the NCs. TEM revealed that the mesoporous constructs were deposited on GO (Fig. 2b). With CTAB as the pore template, the pore size ranged from 3 to 6 nm in diameter, with an approximately random arrangement on the GO (Supplementary Fig. S2). Moreover, the chemical composition of graphite, GO, and NCs was estimated by XPS, and the analysis of the XPS patterns is displayed in Supplementary Fig. S3. A new, broad peak observed for GO at binding energies of 532 eV was attributed to O1s (ref. 37) and NCs at 154 and 174 eV for the Si2p and Si2s components, respectively (Supplementary Fig. S3a). Furthermore, to provide insight regarding the surface-decorated functional groups, deconvoluted XPS spectra of C1s O1s and Si2p are shown in Supplementary Fig. S3b–d in the Supporting Information. In addition to displaying the peaks of GO, the NCs exhibited the formation of C–Si and C–O–Si bonds in the structures. These results confirmed the successful exfoliation from graphite to GO by the Hummers method, and the attachment of carbon-containing silica onto the surface of the NCs. Furthermore, Raman spectroscopy was applied to evaluate the structure of graphite, GO, and NCs (Fig. 2c), using an excitation wavelength of 532 nm. All samples were deposited in powder form on silicon wafers without using any solvent. The peak of the G band is ~1614 cm−1 (in-phase sp2 bond vibration), and the D band is ~1360 cm−1 (disorder band of the graphene edge functional group). Two characteristic peaks were observed in all products. The relative peak ratio of G/D in graphene is higher than that in GO and NCs, which is probably caused by the low proportion of sp3 carbon centers (high content of sp2-hybridized C–C bonds)38.
The synthesized GO with a mean size of ~2 μm was exfoliated from graphite, resulting in slow sedimentation of GO in water. After the formation of cookie structures on GO, the NCs were separated by centrifugation at 4000 r.p.m. to reduce their sizes, and the size distribution of the resulting NCs ranged from ~300–550 nm, as measured by DLS (Fig. 2d). Then, the BET method was utilized to investigate the pore structures through nitrogen absorption–desorption isotherms, which suggest that the surface area of the NCs was ~550 m2g−1; a pore size of 3.6 nm was determined by BJH analysis. The absorption–desorption curve exhibits typical Langmuir type IV characteristics with a hysteresis loop, which is the inherent property of mesoporous materials. Furthermore, the main hysteresis loop in the medium-pressure region indicates capillary condensation and multilayer adsorption in the mesopores. Compared to the NCs, both graphite and GO exhibited relatively low surface areas due to the nonporous structure (Fig. 2e and Supplementary Fig. S4).
NFG release from NCs actuated by MF
The magnetometry result of the NCs determined by a SQUID is given in Fig. 2f. The saturation magnetizations of the NCs were 0.001 and 0.4 e.m.u. g−1 at 300 K and 6 K, respectively. The weak magnetic properties were induced by defects on the graphene. Through adsorption, the loading capacity of NGF in the NCs is 290 µg g−1 NC. The high loading efficiency mostly reflects an affinity between NGF and carbon surfaces, which NCs possess in abundance. Furthermore, to evaluate the effects of the MF on NGF release, the NGF-loaded NCs (NGF@NCs) were subjected to an MF for 15 min at a power of 3.2 kW (1 MHz). A similar instrument developed by MagForce was approved for application in Europe to treat glioblastoma multiforme by local hyperthermia of iron oxide nanoparticles in brain tumors39. As shown in Fig. 2g, the amount of NGF released reached ~45% after 15 min of MF treatment. Without MF treatment, the natural release of NGF was ~14%. The main reason for the triggered NGF release is the electrostatic repulsive force between the protein and NCs. According to Lenz’s law, the direction of current in a conductor is induced by the external MF, which causes a MF that opposes the change that produced it. The magnetic flux causes the electromotive force in the conductive materials, and eddy currents are induced by the MF in closed loops within the stationary conductors40. As the NCs are a carbon-based conductive material, the MF potentially produces eddy currents in them. To investigate the MF effects on the NCs, the NCs were exposed to an MF for 15 min. After the MF treatment, the surface charge of the NCs estimated by a zeta potential analyzer changed to −26.3 mV from −6.5 mV due to the increased electrons on the particles. In addition, the NCs were slightly conductive, with a resistance of 1.64 kΩ cm−1, which is similar to that of graphene and its derivatives41,42, and the eddy current could be induced in the bulk NCs (compact NCs formed by hydraulic press) by MF to light up an LED (Supplementary Fig. S5a). To evaluate the eddy current, NCs were placed in DMEM supplemented with 10% FBS to mimic the physiological environment, and then subjected to an MF at various powers and distances with a frequency of 1 MHz. When subjected to an MF with a distance of 1 cm, currents of 0.78, 0.19, and 0.1 μA were measured at powers of 3.2, 1.92, and 0.32 kW, respectively (Supplementary Fig. S5b). Once the distance between the MF and NCs was increased to 2 cm, the induced currents were significantly decreased (Supplementary Fig. S5c).
Electromagnetized NCs for PC12 cytotoxicity and differentiation
The cytotoxicity of NCs with and without MF treatment was evaluated by incubating the NCs with PC12 cells, which were derived from a pheochromocytoma of a rat adrenal medulla line at different concentrations. Briefly, after coculturing the NCs and PC12 cells for 4 h, the cells were subjected to 5 and 15 min of MF; then, cell viability was examined at 24 h posttreatment. To monitor the heat induced by the eddy current, as shown in Supplementary Fig. S6, at 120 µg ml−1 NCs, the temperature of the solution was observed to be slightly changed (7 °C) under 35 °C, and the viability of the cells with 5 min of MF treatment was ~92%, indicating that the short MF treatment did not exhibit an obvious influence on the cells. When the treatment time was increased to 15 min, cell viability was maintained at 86%, suggesting that the long treatment caused slight cell death due to energy generation. The cellular uptake of the NCs was examined by flow cytometry, as shown in Fig. 2h, which reveals significant fluorescence of the NCs after 0.5, 1, and 2 h of incubation.
To evaluate the MF effect on NGF@NCs in vitro, the PC12 cells were treated with an MF for 15 min at 1 h postincubation of the particles. Without the MF, NGF@NCs were observed on the cells (Fig. 2i), and the PC12 cells exhibited a short neurite length (<15 µm) and a low differentiation percentage (<4%). However, once the PC12 cells were treated with the MF, the average neurite length and differentiation percentage increased to 245 µm and 18%, respectively, indicating effective PC12 cell differentiation when combining the NGF@NCs and MF (Fig. 2j). To clarify the MF and NGF effects, both the PC12 cell neurite length and number of cells with neurites were studied under various conditions, as shown in Fig. 2k. With the assistance of the MF, the average neurite lengths of the PC12 cells reached 46 and 116 µm for NCs and NGF@NCs, respectively. Generally, the neurite outgrowth of PC12 cells can be induced by NGF through activating the receptor tyrosine kinase, G protein-coupled receptors, and heterotrimeric G proteins. The results showed that the NCs were also able to induce PC12 cell differentiation without NGF; in addition, Fig. 2k (right panel) shows 8% of the cells with neurites under an MF. As documented in the literature, the electrical stimulation generated by locally charging objects is able to influence neuronal differentiation, biological processes, nerve regeneration processes, and neurite outgrowth through the upregulation of calcium signaling and phosphorylation of the cAMP-response element-binding pathway. Previous studies also documented that energy does not affect most cell types, but activates intracellular signaling to neurogenesis43.
As shown in Supplementary Fig. S7a, PC12 cells with and without attached NCs were placed together, and an MF was applied. PC12 cells attached to NCs exhibited clear enhancement of neurite outgrowth. Furthermore, to evaluate the effects of electromagnetized NCs on a biomaterial, NGF-NCs were coated on silk through layer-by-layer PLL/NC adsorption. As shown in Supplementary Fig. S7b, the surface of the silk displayed a rough morphology after NC coating, indicating a high density of NCs on the silk. Then, the PC12 cells were seeded on the substrate for 24 h of incubation, and the spherical morphology of PC12 cells was obtained because the cells were not differentiated with neurites (Supplementary Fig. S7c). While subjected to the MF for 15 min at a power of 3.2 kW, increased differentiation and neurite outgrowth were apparently observed on NC@silk (Supplementary Fig. S7d), and the synergistic effects of the NCs and MF were demonstrated on the PC12 cells (Supplementary Fig. S7e). This observation indicates that the electromagnetization of NC can be executed on a substrate and that cell fate can be manipulated, using an external physical energy source.
3D printing of NC@C
A light-curable bioink composed of 4-HBA, PU-EO-PO monomer, and Irgacure 819 (an initiator) could be polymerized by light with absorbance at a wavelength of 260 nm through 3D printing (Supplementary Fig. S8), and each curing layer in the DLP 3D printing process was ~50 µm. 4-HBA has a primary hydroxyl group at the end of a long alkyl chain, providing stretchable and flexible characteristics. To enhance the water retention and protein permeability, the EO and PO segments served as hydrophilic and hydrophobic parts in the elastomer, which was capable of incorporating water-soluble molecules. With the loading of NCs or red dye, the resulting printed NC-loaded elastomer (NC@elastomer) exhibited solid morphology and excellent stability in various architectures (Fig. 3a). The image in Fig. 3b depicts a dog-bone tensile specimen of the 3D printed elastomer. On the specimen, the appearance of the grid on the surface in the SEM image (the lower panel of Fig. 3b) resulted from the layer-by-layer printing and DLP light sources. The color and surface morphology of the sample would also be affected by adding NCs to the bioink, as shown in Fig. 3c, d. Once the concentration of NCs increased to 1%, large amounts of NCs were exposed on the surface via the 3D printing process (Fig. 3e). This phenomenon reflects the local curing of DLP 3D printing, where the particles between light-emitting and unexposed areas were fixed on the resulting matrix surface (Supplementary Fig. S9). However, when traditional molding was used to prepare the composite, the particles preferred to be embedded in the gel. Thus, by using this printing technology, the surface roughness can be controlled, which has been demonstrated to improve cell adhesion44.
The precision of the 3D printed matrix was evaluated by Zen Desk software by comparing the resulting 3D printed matrix and the original CAD image (Supplementary Fig. S10). The precision was 87.55% and 82.5% for the elastomer and NC@elastomer, respectively. Furthermore, the resulting NC@elastomer displayed an elongation of ~150% (Fig. 4a). While adding 10% of the PU-EO-PO monomer in the bioink, the elastic modulus could be ~2.5 times (Fig. 4b) because the aromatic urethane group of PU exhibits a strong intermolecular force. Moreover, by adding 1% NCs to the elastomer, the elastic modulus was increased to 0.34 MPa from 0.22 MPa (Fig. 4c and Supplementary Fig. S11). However, further addition of NCs to the bioink would lower the elasticity, so 1% NCs was used in the elastomer in the following drug/protein release, cell, and animal studies. Furthermore, to evaluate swelling behavior, the resulting NC@elastomer was placed in water and ethanol for 2 h. As shown in Fig. 4d, with the assistance of a partial hydrophilic segment of PU-EO-PO and NCs, the swelling ratio of the matrix was ~15% and 46% in water and ethanol, respectively. The NC@C could still maintain the elasticity and microchannels after 2 weeks of immersion (Fig. 4e). After swelling, protein penetration was investigated by immersing Texas Red™-conjugated ovalbumin (ThermoFisher, USA) in the NC@C; then, the cross section of the conduit was examined by confocal laser scanning microscopy (CLSM, Zeiss LSM800, Germany). As shown in Fig. 4f, a significant fluorescence signal of ovalbumin was observed on the surface of the conduits after 2 h of diffusion, where the green fluorescence was caused by the autofluorescence of the conduit. After increasing the time to 48 h, ovalbumin was detected in the whole conduit, and strong signals were also observed.
NGF penetration and release from NC@C under MF treatment
The protein penetration was potentially caused by the PU-EO-PO amphiphilic polymer chains and the NCs. Indeed, the NC amounts in the conduit reflected the protein-loading capacity of NC@C. As shown in Fig. 5a, the protein-loading capacity of 1%NC@C was approximately three times greater than that of the conduit alone. However, 1% NCs in the conduit did not obviously affect the loading capacity of small molecules when compared to that of the conduit alone, where doxorubicin (Dox) was used as a model drug. The mechanism was probably induced by the strong affinity between the carbon porous structure and the proteins, where the porous surface provided the binding sites to the proteins45. The external MF enhanced the penetration of Dox and ovalbumin into the NC@elastomer, which was monitored by CLSM from the top of the elastomer (Supplementary Fig. S12). The profiles of penetration depth under the influence of 15 min of MF treatment at the initial time point display several important features (Fig. 5b). (a) The penetration rate with an MF is much faster than that without an MF, as evident from the deep penetration of Dox and ovalbumin under MF treatment compared to that without. (b) The penetration rate of ovalbumin was slower than that of the small molecules, which could be understood in terms of the large molecular size of the protein. Similar triggered release results are displayed in Fig. 2g.
Since the MF is known to cause the local current to be mediated through the conductor of NCs, physical energy may enhance the chain flexibility of the elastomer, further aided by the affinity of the NCs to water, realizing the fast penetration of protein. Furthermore, the release patterns of Dox and NGF (a small and large model molecule, respectively) from NC@C under MF treatment were also investigated, as shown in Fig. 5c, d. For all the release profiles, there was an initial burst release, which manifested the rapid desorption of Dox and NGF from the surface of the conduit. Similar to the influence of the MF on penetration, the release rates were obviously increased while subjected to the MF due to the induced eddy current in the NCs. Although fast Dox and NGF release was obtained for the first hour to 3 days, the cumulative release of NGF reached only 60%, whereas the Dox release was ~100% due to the low diffusion rate of large molecules. Furthermore, Dox in conduits exhibited faster release than that in NC@C, likely due to the strong affinity of NC for Dox. These results suggest that the short MF exposure caused a permanent conformation change that sustained the release of the cargo for a few days.
Cell adhesion and PC12 differentiation on NC@C
The in vitro cell attachment and compatibility of NC@C were determined using fibroblast cells (HIG-82), SCs, and N2a incubated for 24 h. As shown in Fig. 6a, the three types of cells effectively attached and developed on the surface, where DAPI stained the nucleus and phalloidin stained F-actin. However, the conduit demonstrated weak cell adhesion. Compared to that on the cell culture dish, cell adhesion on the NC@C was ~80%, but cell adhesion on the conduit alone was only ~40%, as shown in Supplementary Fig. S13. This observation is reminiscent of the effect of roughness through particle exposure, which is the basis of the 3D printing composite process. As expected, 1% NCs increased the surface roughness of the conduits after the printing process, as shown in Supplementary Fig. S14. The profile roughness parameter (Ra) increased from 2.93 to 36.32 when adding 1% NCs to the bioinks during 3D printing. Having demonstrated cell adhesion, we followed ISO 10993-5 (tests for in vitro cytotoxicity) to determine the cell viability of the 3D printed materials before removal of unreacted agents. As shown in Fig. 6b, higher cell viability of the 3D printed materials was observed after 4 h postcuring in HIG-82 cells. The results indicated that polymers containing residual initiators and monomers would lead to considerable toxicity46. Therefore, the postcuring and removal of unreacted agents in the polymer by ethanol were apparently critical in the application of photocured NC@C. Figure 6c also reveals that the NCs slightly decreased the cell viability of the conduits, suggesting the low toxicity of the latter. Therefore, before NGF loading and animal studies were carried out, these agents were removed to mitigate toxicity.
For effective nerve outgrowth, differentiation should be triggered by an external physical cue and growth factors. For NC@C, we investigated the percentage of PC12 cells differentiated with neurites and the length of axonal outgrowth induced by NGF and MF (Supplementary Fig. S15a). Similar to a previous study, NCs treated with the MF impacted β-III-tubulin and the neuron-specific cytoskeletal protein in PC12 cells through the synergistic effect of NGF and electrical stimulation (Fig. 6d). As expected, the fluorescence images of PC12 cells with MF stimulation exhibited longer neurite extensions than those cells without treatment. To estimate the NGF (protein) activity after MF treatment, NGF was loaded on NCs in advance and released by applying an MF. The released NGF was then added to PC12 cells. As the results showed that the released NGF was still able to induce PC12 cell differentiation, indicating the activity of NGF (Fig. 6d). Statistical analysis of neurite length and neurite number over various treatments is also demonstrated in Fig. 6e, f. In the NGF-NC@C group, the average neurite length reached 40 µm, and >60% of the PC12 cells were differentiated after 15 min of MF treatment; for the PC12 cells without MF treatment, the average neurite length reached only 22 µm, and 31% of the cells were differentiated. The results confirmed the significant enhancement in the differentiation of PC12 cells under MF treatment through electromagnetic stimulation of the cells and intense NGF release. To further evaluate the effects of NC exposure on the conduits, the NCs were embedded in a gel to model PC12 cell differentiation. As shown in Supplementary Fig. S15b, c, while subjected to the MF, the NGF-NC-embedded gel could induce cell differentiation by triggering NGF release. However, compared to that for the exposed NC on the conduit, the neurite length or percentage of cells with neurites for the NGF-NC-embedded gel was decreased, indicating that the amplification of PC12 cells was stimulated by directly contacting the NCs to cells. Therefore, 3D printing to expose NCs on the conduits can significantly enhance cell differentiation. To ascertain the effects of the microchannels on cell orientation, HIG-82 cells were cultured on 3D printed conduits containing microgrooves with and without NCs. As shown in Fig. 6g, the cells on the conduit without channels had a random orientation. However, in the groups of conduits with channels, the HIG-82 cells were highly aligned after 4 days of incubation. This orientation was also quantified by a Gaussian distribution (oriented at a 0° basis angle), as shown in the right column of Fig. 6g, suggesting the effective improvement of in vitro cell orientation via these channels. Compared with the conduit with the channel (without NC), NC@C with the channel exhibited less orientation, which was potentially attributed to the surface roughness, facilitating cell adhesion47,48. To clarify the differences in the generated eddy currents in vitro and in vivo, the NCs were placed in DMEM supplemented with 10% FBS to mimic the cell culture conditions and biological tissue (animal muscle), respectively. The generated currents in vitro and in vivo were ~28 and ~38 μA, respectively (Supplementary Fig. S16). The differences may be attributed to the ions in the microenvironments.
Functional recovery of a regenerated sciatic nerve in vivo
To implement the spatial distribution of NGF for the promotion of axon outgrowth, 3D printed NC@C loaded with NGF was implanted after a peripheral nerve was transected 10 mm from the bifurcation point of the sciatic nerve in the popliteal fossa of SD rats. In the animal studies, five groups, including autograft, NC@C, NGF-NC@C, and NGF-NC@C+MF, were implanted in vivo in the sciatic nerve transection, where identical amounts of NGF (29 µg g−1 conduit) were applied in the NGF-loaded groups (Fig. 7a). The distance between the MF and the incision at the posterior of the left hind was maintained at 1 cm (Supplementary Fig. S17). The MF treatment was carried out for 15 min at a power of 3.2 kW and frequency of 1 MHz. The toe spreading of an injured hind paw and the SFI were measured 30 days postimplantation, which is an atypical walking footprint analysis of toe spreading. The SFI is 0 for healthy animals and −100 for animals with a transected sciatic nerve. After 1 month of treatment, a significantly wide toe spreading of the injured hind paw was observed in the NGF-NC@C+MF group, indicating excellent axonal regeneration (Fig. 7b, c), where 15 min of MF was locally applied at 4 days postimplantation at a power of 3.2 kW (1 MHz). After analysis, we observed that the SFI value of the NGF-NC@C+MF group was higher than those in the other groups. Furthermore, once sciatic nerve injury occurs without treatment, it causes atrophy of the gastrocnemius muscle (decreases in muscle weight and muscle fiber size), leading to a decreased weight ratio of the wet gastrocnemius muscles between the left and right sides. However, in the NGF-NC@C+MF group, the weight ratio of the gastrocnemius muscles was close to that in the autograft group, suggesting excellent recovery of the gastrocnemius muscles (Fig. 7d).
For peripheral nerve regeneration, the morphologies of atrophied muscle were used to estimate limb recovery. At 1 month postimplantation, the gastrocnemius muscles were harvested and stained by hematoxylin and eosin, as shown in Supplementary Fig. S18. Then, the diameter and distribution of the muscle fibers were calculated (Fig. 7e). The atrophied muscle in the conduit alone group demonstrated obvious degradation of muscles. Even though the weight ratio of the gastrocnemius muscles of the NGF-NC@C group was close to that of the autograft, weak recovery was observed in the muscle fibers, i.e., small fiber diameter. However, combining the MF treatment, the diameter of the muscle fibers was regained after nerve reinnervation and was close to that in the autograft (gold standard in nerve gap repair), as observed in the NGF-NC@C+MF group.
Regenerated myelinated nerve fibers
The myelinization process of SCs is a key factor in axonal regeneration after PNI because the developing nervous system is guided by a myelin sheath. Furthermore, distal SCs undergo atrophy due to the loss of axonal contact for a prolonged period, resulting in an inhibitory extracellular environment for nerve regeneration. Generally, stronger distal axons and myelination lead to better functional recovery for the injured nerve. As shown in Fig. 7f, the cross section of distal regenerative nerves evaluated by TEM at 1 month postimplantation suggests that the lamination of myelin sheathes of the regenerated nerve fibers in NGF-NC@C+MF was successful compared to that of the gold standard (autograft). However, in the NC@C and NGF-NC@C groups, the number of layers of myelin sheaths was smaller. After myelination analysis, the axon densities in each group were similar, but only NGF-NC@C+MF exhibited a similar number of myelin layers as that of the autograft (Fig. 7g, h). These findings indicated that NGF-NC@C+MF improved the diameter of the regenerative axon by controlling biocues and magnetoelectric stimulation.
To understand the growth and orientation of regenerated axons, the whole regenerated nerve from the proximal to distal section in the conduit was harvested, and stained with ß-III-tubulin and DAPI to track the regenerated axons and cell nuclei, respectively. Figure 8 displays a large area of CLSM images of immunohistochemically stained sciatic nerve defects from various treatments at 1 month postimplantation. Several conclusions could be drawn from these results. First, although the autograft group had a thick nerve, curled and random fiber-like morphologies of regenerated axons were observed because the direct surgical suture led to the mismatch of axons. Second, NGF-NC@C+MF exhibited a thicker nerve bundle than NGF-NC@C and NC@C, indicating that nerve growth was promoted by magnetoelectric stimulation and NGF release (inset picture). Third, a small distribution of axon orientation was detected in the conduit groups. This observation confirmed the effective regeneration of the MF treatment (i.e., magnetoelectric stimulation induced nerve differentiation mediated by NCs in the conduit, and NGF release enhanced cell growth). In addition, judging from the limb muscle weights and functions, the nerve recovery effects in NGF-NC@C+MF appear to be excellent. Electrophysiological analysis includes nerve conduction velocity (NCV) compound muscle action potential (CMAP) in Supplementary Fig. S19a, which is a classical electrophysiological method with diagnostic and prognostic value for evaluating PNI49,50. NCV and CMAP are highly influenced by axons and myelin, which are correlated with the maturation of regenerative nerve fibers17,51. A normal nerve was found to exhibit an NCV of 47.5 m s−1 and a CMAP of 46.2 mV. At 1 month postsurgery, the NGF-NC@C+MF group exhibited an NCV of 44.7 m s−1 and a CMAP of 44.9 mV, both of which were higher than those of the NGF-NC@C group (Supplementary Fig. S19b, c).
To estimate the immune response for a long period, the autograft nerve and NC@C were harvested and compared after 2 months of surgery, where the numbers and distribution of macrophages were stained by CD68 antibody for tracking purposes (Supplementary Fig. S20a, b). The CLSM images revealed that there was no significant difference in the numbers of macrophages between the normal and NC@C groups, indicating low inflammation in the tissue. Furthermore, the weight changes of mice for autograft, conduit, NC@C, and NC@C+MF displayed almost identical patterns due to the biocompatibility of the implanted materials (Supplementary Fig. S20c). Next, six indices of liver as well as kidney functions, including albumin, alkaline phosphatase, alanine aminotransferase, aspartate aminotransferase, creatinine, and total bilirubin, and two strong predictors of all-cause mortality includingserum C-reactive protein and plasma D-dimers, were also evaluated after 60 days postimplantation of high-density polyene (control group), conduit, and NC@C (Supplementary Fig. S21). The differences between these groups were relatively small, implying the low toxicity of various treatments and safety of the materials. On the other hand, no severe immune response or tissue damage was observed once NC@C was implanted in muscle tissue, following the ISO 10993-6 test (tests for local effects after implantation, Supplementary Fig. S22). These results are consistent with previous studies exhibiting tolerable bioresponses for a long time.
In summary, we have developed a 3D printing method to prepare an NC-exposed conduit with stretchable and biocompatible characteristics. The NC, a nanocomposite composed of reduced GO, mesoporous silica, and a thin carbon layer, possesses superb attributes of magnetoelectric conversion, protein adsorption, and cell stimulation. After integrating the NCs into a bioink, the printed NC@C was able to carry large amounts of NGF with high efficiency and exhibited molecule/protein permeability. On-demand NGF release and cell stimulus were achieved in vitro using a high magnetoelectric treatment to effectively induce cell proliferation and differentiation, regaining the diameter of animal muscle fibers, as well as improving layers of myelin sheathes in vivo and directing the axon orientation. This printed NC@C with exposed particles, excellent mechanical properties, controllable NGF release, and magnetoelectric stimulus is expected to introduce a new approach for the regeneration of nerves and other tissues.