Adaptive self-healing electronic epineurium for chronic bidirectional neural interfaces

Realizing a clinical-grade electronic medicine for peripheral nerve disorders is challenging owing to the lack of rational material design that mimics the dynamic mechanical nature of peripheral nerves. Electronic medicine should be soft and stretchable, to feasibly allow autonomous mechanical nerve adaptation. Herein, we report a new type of neural interface platform, an adaptive self-healing electronic epineurium (A-SEE), which can form compressive stress-free and strain-insensitive electronics-nerve interfaces and enable facile biofluid-resistant self-locking owing to dynamic stress relaxation and water-proof self-bonding properties of intrinsically stretchable and self-healable insulating/conducting materials, respectively. Specifically, the A-SEE does not need to be sutured or glued when implanted, thereby significantly reducing complexity and the operation time of microneurosurgery. In addition, the autonomous mechanical adaptability of the A-SEE to peripheral nerves can significantly reduce the mechanical mismatch at electronics-nerve interfaces, which minimizes nerve compression-induced immune responses and device failure. Though a small amount of Ag leaked from the A-SEE is observed in vivo (17.03 ppm after 32 weeks of implantation), we successfully achieved a bidirectional neural signal recording and stimulation in a rat sciatic nerve model for 14 weeks. In view of our materials strategy and in vivo feasibility, the mechanically adaptive self-healing neural interface would be considered a new implantable platform for a wide range application of electronic medicine for neurological disorders in the human nervous system.

, respectively. Even though the initial applied stress of SHP and A-SEE were higher than that of PDMS, effective stress relaxation properties of SHP and A-SEE allowed to reach much lower stress than that of PDMS over time. At 37 °C, the initial stress and the final stress after 1 h of relaxation under 30% tensile strain of the materials are 57 kPa and 53 kPa (PDMS), 37 kPa and 0.003 kPa (SHP), 94 kPa and 3.6 kPa (Composite), and 64 kPa and 1.3 kPa (A-SEE), respectively. In addition to the stress relaxation performance, when the temperature increased, initial stress of SHP, composite, and A-SEE under 30 % strain significantly reduced owing to the thermoplasticity of SHP. Supplementary Fig. 5. Stress-strain characteristics (a) and the calculated Young's modulus (b) of the materials. The 0.3 mm thick SHP film, the 0.3 mm thick AuNM-composite film, the 0.9 mm thick encapsulated AuNM-composite (A-SEE) film, and the 0.3 mm thick PDMS (Sylgard 184, PDMS:crosslinking agent weight ratio of 10:1 and 20:1) films were each cut to have a length of 10 mm and a width of 5 mm. Each sample was stretched at a strain rate of 50% per minute using Instron 5565 instrument. The calculated Young's modulus of the materials were 896.6 ± 16.0 kPa (PDMS with a PDMS:crosslinking agent weight ratio of 10:1), 131.8 ± 13.8 kPa (PDMS with a PDMS:crosslinking agent weight ratio of 20:1), 163.3 ± 42.1 kPa (SHP), 1109.0 ± 24.8 kPa (Composite), and 272.7 ± 34.5 kPa (A-SEE). We used PDMS with a PDMS:crosslinking agent weight ratio of 20:1 as the control material in all experiments due to its value of Young's modulus (131.8 ± 13.8 kPa) being comparable with that of SHP (163.3 ± 42.1 kPa).
Supplementary Fig. 6. Ex vivo shear stress measurement set-up (a) and the measured shear stress-strain characteristics (b) of the materials. To examine the shear strength at the materialnerve interfaces induced by the movement of the nerve tissue, we designed an ex vivo shear force measurement set-up by using an isometric force transducer and an automatic unidirectional movement stage. A sciatic nerve tissue extracted from a rat was immersed in Krebs-Henseleit buffer solution (2000 mg/L Glucose, without calcium chloride and sodium bicarbonate) and fixed tightly by suturing. Then, the PDMS and SHP cuff structures which are connected to the force transducer were applied to the nerve tissue. The shear force was measured while the nerve tissue was moved by the automatic stage at a speed of 50% strain per minute. The measured maximum shear stress at PDMS-nerve interfaces and SHP-nerve interfaces was 5.06 kPa and 0.83 kPa, respectively, which are both much lower compared to the nerve tissue modulus (~100 kPa). After the maximum stress point, the stress values of ~2.2 kPa (PDMS) and ~0.6 kPa (SHP) were saturated, respectively. The remaining shear stress may indicate the kinetic friction between the materials and the nerve tissue in wet conditions. A much higher value of shear stress applied at PDMS-nerve interfaces (5.06 kPa) than that at SHP-nerve interfaces (0.83 kPa) may be due to the difference in the compressive stress induced by the fixed cuff structures. Effectively relaxed compressive stress at SHP-nerve interfaces facilitate the shear movement of the nerve tissue while maintaining conformal contact. Such a low level of the shear stress generated at SHP-nerve interfaces could be beneficial for safe and stable chronic implantation. Owing to the potential toxicity of the remaining PDMS precursor, we compared two different PDMS sterilization methods. We first prepared the PDMS blocks with three different PDMS:crosslinking agent ratios (10:1, 20:1, and 30:1). Then, a group of PDMS was kept in PBS for 7 days with PBS replacement and the other group was immersed in 70% ethanol solution and washed twice with PBS right before the viability test 34,35 . We evaluated the cell viability 7 days after the incubation of cells with PDMS. We used a THP-1 cell that is a human monocyte cell grown in suspension, because the adherent cells are not able to attach to the surface of PDMS, which results in inaccurate viability. The result shows that the effect of the PDMS washing method on the cell viability is negligible. In all cases, the cell maintained a viability degree of over 85%, which means that all conditions are bio-compatible. Data are represented as mean ± S.D. (n=10 4 cells/mL examined over 5 independent experiments). Representative images from n=3 biologically independent samples. H&E staining was used to confirm morphological changes and scar tissue formation. TUNEL assay was able to stain the apoptotic cells brown. e,f, Quantification of relative fluorescence intensity of CD68 and CTGF (e) and the relative intensity of TUNEL staining (f) from the immunostained nerve cross-section images shown in b-d. The results for H&E staining showed that the nerve tissue was deformed by PDMS while the nerve construct maintained in SHP. Furthermore, TUNEL staining results showed that PDMS cuff structure induced the apoptotic cell death due to the mechanical modulus mismatch at the biotic-abiotic interface. The immunofluorescence images for CTGF and CD68 represented that PDMS actively induced the CTGF over-expression and immune responses. Sham sample was used as a control ( b, SEM images of the AuNM-composite before and after stretching at 100% strain. Representative images from n=5 independent samples. c, Cell viability test after seven days culture. C2C12 cells were used for the test. The result clearly supports that cell death was not induced by leaked silver ions from the stretched AuNM-composite (c, green), compared to the composite sample with no AuNM (c, red). Data are represented as mean ± S.D. Statistical analysis was performed using oneway ANOVA with Tukey's multiple comparison test (n=10 4 cells/mL examined over 5 independent experiments. d, CDC values of the AuNM-composite before and after stretching at 100% strain. The CDC values were calculated from the obtained CV curves (20 cycles) shown in Fig. 2g (see Methods). Data are represented as mean ± S.D. Statistical analysis was performed using one-way ANOVA with Tukey's multiple comparison test (n=17 independent experiments, P=0.0001, *** P<0.001). When stretched to 100% strain, the CV curves shifted slightly (Fig. 2g) and the corresponding CDC values increased. These results may be correlated with cracks formed in the AuNM spreading to the surface of the composite. In the pristine state (before stretching), some AuNM wrinkles that may have been generated during the detachment from the oxide wafer were observed, but no cracks existed. However, at 100% strain state, cracks in the AuNM were generated because only the composite and SHP substrate were intrinsically stretchable and not the AuNM. Thus, the stretched electrode has a different surface nature compared with that of the pristine electrode due to the exposed Ag flake-SHP composite, which resulted in changes in the CV characteristics. at 800% strain) was much lower than that of the transferred AuNM-composite (502 S cm -1 at 200% strain and 558 S cm -1 at 800% strain). This difference in electrical performance may be due to 1) poor interfacial stability of the directly deposited AuNM with many cracks as shown in Supplementary Fig. 13, and 2) thermal degradation of the Ag flake-SHP composite during the deposition process 21 . and LabVIEW program code (right) for neural signal monitoring. c, Photographs of the recorded sensory neural signals induced by mechanical stimulation. The recorded neural signals were displayed in the second row of the monitoring system (The first and the third row indicated the system noise of the analog-to-digital converter board). Since A-SEE was implanted to the right side sciatic nerve of the rat, the sensory neural signals could be recorded when the mechanical stimuli were applied on mechanoreceptors of the right side hind paw of the rat.

Supplementary Fig. 19. Mechanical stability of the A-SEE for neural signal recordings under the tensile force applied to the interconnection wire. a, Photographs showing surgical sites of
A-SEE implantation after 0 and 4 weeks. b, Experimental setup for confirming the stability of neural signal recordings under tensile force applied to the interconnection wire. c, The recorded sensory neural signals from mechanical stimulation under different tensile forces (from 0.01 N to 0.1 N) applied to the interconnection wire after 4 weeks implantation of A-SEE. We could achieve stable neural signal measurement (SNR of 1.78 ~ 2.08) regardless of the applied force to the wire. The result shows that the self-bonding assembly of the A-SEE (see Supplementary Fig. 1) allows the contact interface to be electrically and mechanically passivated from external stimuli. In addition to our material uniqueness, the interconnection wire connected to our neural device was strongly fixed by wounded muscle.
Supplementary Fig. 20. SNR color mapping of recorded neural signals for six weeks of implantation (n=5). a,b, The neural signals recorded from strong (a) and weak (b) mechanical stimulation during six weeks of the A-SEE implantation were shown as color mapping images. These color maps clearly show that the magnitudes of the recorded neural signals from the sciatic nerve become different depending on the mechanical stimulation induced on the hind paw.
Supplementary Fig. 21. Denoising process of the recorded sensory neural signals. a,b, Raw signal data compared with denoised data for week 0 (a) and week 6 (b). c-f, Variance of the decomposed signals as a function of a time-window scale employed in the wavelet decomposition for the raw data (c,d) and the denoised data (e,f). g,h, The bottom two plots are the SNR of the raw data (g) and SNR ratio between the signals under stimulation and at the rest state (h), respectively. All data are represented as mean ± S.D. (n=5 data from independent animals).
Supplementary Fig. 22. Histological and immunofluorescence analyses of rat sciatic nerve cross-sections after one week and six weeks of A-SEE implantation. a, H&E and TUNEL staining images and immunofluorescence images of the rat sciatic nerves after one week and six weeks of A-SEE implantation. Representative images from n=3 biologically independent samples. b, Quantification of relative fluorescence intensity of CD68 from the immunostained nerve crosssection images shown in (a). After one week of implantation, A-SEE-induced nerve tissue deformation and damages were negligible in H&E and TUNEL staining images. Additionally, CTGF (green) and CD68 (red) protein expression levels in the A-SEE implanted sample were similar to those observed in the sham control sample. After six weeks, the thickness of the fibrotic scar tissue around the nerve bundle slightly increased compared with the nerve tissue after one week. However, the cellular morphology and the nerve construct were well maintained, which indicates that the neurons functioned normally during long-term monitoring.  (a, b), the recorded sensory neural signals (c, d), and the photographs of the explanted sciatic nerve interfaced with the A-SEEs for 6 weeks (e, f) are compared. We assumed that the small protruded fibrotic tissue formed at the SHP regions in the A-SEE with protruded electrodes (e) might originate from the physical height gap between the SHP regions and the nerve.
The composite electrodes of our neural interface were designed to be protruded on the SHP substrate (a), and thus SHP substrate regions between the electrodes might not be intimately contacted to the nerve tissue. Generally, any biological fluids can easily fill up the gap when implanted. The insufficiently contacted interfaces cause micro scar formation and the migration of immune-related cells, thereby resulting in the scar tissue formation (e). To better understand the complex interplay at the interface, we prepared a control experiment in which the A-SEE with a modified structure was fabricated to remove the protruding portion of the electrodes (b). We implanted the modified A-SEE to a rat sciatic nerve and recorded peripheral neural sensory signals for 6 weeks (c, d). As a result, we confirmed that the modified A-SEE could also achieve chronical stable neural signal recording. Furthermore, the explanted nerve tissue after 6 weeks of the modified A-SEE implantation showed no scar tissue formation at the whole interface areas (f), confirming our hypothesis.
Supplementary Fig. 25. Simultaneous neural signal recording and joint kinematic analysis from an awake rat during treadmill walking. The joint kinematic analysis system consists of a neural signal recording amplifier, a high-speed digital camera, and a grabber board with control software. The digital camera was positioned perpendicular to the walking trace to record the images. The anatomical landmarks include the hip (H), knee (K), ankle (A), and fifth metatarsal (F) joints on the lateral side of the hind limb with the reflect marker positioned on the defined anatomical landmarks.
Supplementary Fig. 26. Chronic recordings of peripheral sensory neural signal. The sensory neural signals from sciatic nerve were successfully recorded after 14 weeks of A-SEE implantation (N=1, we did not sacrifice one of the implanted rats (N=5) up to now). Especially, the signals from a rat are recorded after repetitive electrical stimulation (over 200,000 pulses) for neuromodulation and treadmill gait experiments to confirm the stability for chronic bidirectional interfacing.