Preparation of self-deployable and X-ray detectable substrate
PLCL-PLGA was created by mixing PLGA (lactide:glycolide; 85:15, Sigma-Aldrich) and PLCL (lactide:caprolactone; 50:50, PolySciTech) in a 1:1 weight ratio and dissolving it in a chloroform solution (20 wt%). A sheet of PLCL-PLGA (~ 50 µm) was spin-casted on a silane (trichloro(octadecyl)silane, Sigma-Aldrich) treated glass substrate and dried under ambient conditions. The substrate consisted of two patterned PLCL-PLGA layers; the bottom layer formed the overall tent structure for axis-transformation and the top layer with a through hole was aligned on each chassis for electrical connection. To create X-ray detectable substrate, PBAT (S-EnPol) was dissolved in chloroform (30 wt%), and iodixanol solution (Visipaque, GE Healthcare) was mixed into the PBAT/chloroform solution in a 3:1 weight ratio and stirred for at least 24 hours to achieve uniformly mixed PBAT-iodixanol solution. A sheet of PBAT-iodixanol (~ 50 µm) was fabricated following the same procedure as PLCL-PLGA sheet fabrication. PBAT-iodixanol layers were stacked on the substrate PLCL-PLGA layer for detection using a medical imaging instrument.
Fabrication of electronics on self-deployable substrate
Fabrication of silicon electronics: The fabrication of silicon-based sensors (strain and pH sensors) began with doping phosphorus for pH sensor (P509, Filmtronics) or boron for strain sensor (B153, Filmtronics) on the top silicon Si NM of a SOI wafer (top Si layer, ~ 400 nm, SOITEC) with thermal diffusion (phosphorus, 1050 ℃, 5 min; boron, 1050 ℃, 20 min). 5 µm\(\times\)5 µm size holes were photopatterned (AZ5214, MicroChemicals) on the top doped Si NM layer followed by reactive ion etching (RIE, J Vacuum Technology) and the underlying oxide layer was removed by immersion in hydrofluoric acid (HF, 49%, electronic grade, SAMCHUN) for 30 min. The released Si NM was transferred onto a temporary substrate of polyimide (PI, ~ 1 µm) / Si wafer by PDMS holder (sylgard 184, 10:1, DOW) and the sensing region was defined by photolithography and RIE. A thin layer of Mg (~ 500 nm) was sputtered to create interconnect, which were defined by lift-off process with negative photoresist (nLOF2070, MicroChemicals). Thereafter, a thin layer of PBAT-iodixanol (~ 20 µm) was coated on the fabricated device for protection. The fabrication of active multiplexed electrodes began with deposition of SiO2 layer as a doping barrier (~ 600 nm; Plasma enhanced chemical vapor deposition, PECVD, Oxford instruments). The doping region was photolithographically defined on the barrier oxide layer followed by oxide etching (Buffered oxide etchant 6:1, SAMCHUN Chemicals). The exposed doping region was then doped with phosphorus by thermal diffusion, and the doped Si NM was released from the SOI wafer. Doping and transfer printing of doped Si NM on the temporary substrate (PI/Si wafer) and isolation of the Si NM active region were conducted as described previously. SiO2 (~ 100 nm) for the gate oxide and Mo (~ 500 nm) for source, drain, and gate electrode were sputtered and photopatterned to form transistor-based active electrode. Thereafter, a thin layer of PBAT-iodixanol (~ 20 µm) was coated on the fabricated device for protection. The PBAT-iodixanol/Si device/PI triple layer was carefully cut into the desired size using a razor blade and then peeled off from Si wafer by PDMS holder. Bottom PI layer was removed by reactive ion etching (RIE; J Vacuum Technology). The PBAT-iodixanol/device layer was transferred onto the prepared PBAT-iodixanol sheet (~ 20 µm) with heat sealing (80 ℃) and stack another PLCL-PLGA layer (~ 50 µm) onto PBAT-iodixanol/device/PBAT-iodixanol layer. The PLCL-PLGA/PBAT-iodixanol/Si device/PBAT-iodixanol were patterned into the individual chassis geometry by UV-laser (MD-U1000C, Keyence Corp.). Individual chassis devices were integrated onto the fully patterned PLCL-PLGA substrate (~ 50 µm) with heat sealing (110 ℃). The sequential fabrication process is illustrated in Supplementary Fig. 4.
Fabrication of metal-based electronics
The fabrication of metal-based electronics (a passive electrode, a temperature sensor, and an inductor coil) began with spin-casting a temporary substrate of PI (~ 1 µm) on a Si wafer, and Mg (temperature sensor, ~ 500 nm; coil, ~ 1.5 µm) or Mo (passive electrode, ~ 500 nm) was sputtered onto the temporary substrate. A PBAT (~ 10 µm) was coated on the deposited metal layer for protection. The PBAT/metal/PI triple layer was carefully cut into the desired size using a razor blade and then peeled off from Si wafer by PDMS holder. The PBAT/metal/PI pieces were patterned into device (resistor or coil) by UV-laser and bottom PI layer was removed by RIE. The patterned PBAT/metal layer was transferred onto the prepared PBAT-iodixanol sheet (~ 20 µm) with heat sealing (80 ℃) and stack another PBAT-iodixanol/PLCL-PLGA layer (~ 20 µm / ~50 µm) on them. Stacked layer was patterned to the individual chassis geometry by laser cutting. Individual chassis devices were integrated onto the fully patterned PLCL-PLGA substrate (~ 50 µm) with heat sealing (110 ℃).
NFC based wireless module
The fabrication of a wireless module began with the formation of a lab-made printed circuit board (diameter ~ 8 mm). A Cu foil (~ 10 µm, KRTLAB) was placed on a PDMS slab and ablated by UV-laser to define the circuit for NFC. A patterned Cu circuit was then transferred onto the adhesive side of Kapton tape (~ 50 µm) as a flexible substrate. Electrical components (NFC chip, w 4 mm\(\times\)h 4 mm, RF430FRL152H, Texas Instruments; capacitor, w 600 µm\(\times\)h 300 µm, GRM033, Murata) were electrically soldered on the fabricated circuit and encapsulated with PBAT, leaving the contact pad exposed. The NFC circuit was electrically connected with a Mg coil (~ 1.5 µm) stacked on the preformed PLCL-PLGA substrate for communication with an NFC reader (TRF7970A, Texas Instruments).
Injection process with controller
The injecting process was controlled using a customized injection module. The injection module is consisted of a syringe pump (TYD01-01, Lead Fluid Technology) for precise control of injection speed (1-100 mL/min), a customized stand for injection height adjustment, and a transparent injection tube for monitoring of the shape of the electronic tent during injection. The injection was performed under controlled condition (injection speed: 40 mL/min) for characterization of the injection behavior.
Mechanical characterization with finite element analysis (FEA)
A three-dimensional FEA simulation captured the axis-transformation behavior of the device using commercial software (ABAQUS 2016, Dassault Systèmes). The explicit mode was employed to make the motion of the substrate more similar to the experimental result. Additionally, the mesh size and C3D8R of the element type were refined sufficiently to saturate the simulation result for more accurate simulation result. In the simulation, the invasive tube was set as a rigid body, and the viscoelastic properties were for the substrate material properties. The stress relaxation test was conducted to evaluate the viscoelastic property of the PLCL-PLGA, and the result was fitted with Prony series for input into the finite element model. For the simulation, the initial elastic modulus was set at 350 MPa, Poisson’s ratio was set at 0.3, and the density was set at 1.3 g/cm3.
Animal surgery procedure
All animal experiments were conducted in accordance with approved protocol by the KBIO Osong Medical Innovation Foundation (KBIO-IACUC-2021-272 and KBIO-IACUC-2022-103) and the ASAN Institute for Life Sciences at the Asan Medical Center (2022-14-012) Institutional Animal Care and Use Committee (IACUC). Both male (Beagle, male, eight 8–10 months, 10–13 kg, n = 2) and female (Beagle, female, 6–8 months, 7–8 kg, n = 2) canine models were used for the verification of in vivo shape recovery and device functionality test. The canine models were anaesthetized with Zoletil (5 mg/kg, I.M) and xylazine (2 mg/Kg, I.M) and maintained under anaesthesia with isoflurane (< 3%). The craniotomy was done in two different ways and a surgical drill (REF 7020-001, CONMED) was used for both procedures: (i) Skull removal for replacement with a transparent skull replica (replica dimension, ~ 50 mm\(\times\)~40 mm); and (ii) a hole drilling for minimally invasive deployment (hole diameter, ~ 8 mm). The sciatic nerve was exposed after a skin incision and splitting the underlying muscles for electrical stimulation to induce SSEP. Rat models (Sprague Dawley rat, male, 8 weeks, 200–250 g) were used for the verification of in vivo biodegradation and biocompatibility. The rats were anaesthetized with Zoletil (50 mg/kg, I.M.) and xylazine (10 mg/kg, I.M) and a 10 mm diameter craniotomy was conducted for biocompatibility and biodegradation verification of the implants.
In vivo imaging for self-deployment and biodegradability
In vivo detection of self-deployment in canine model was conducted using the C-arm X-ray system (Ziehm Vision RFD, Ziehm Imaging). The X-ray images were taken after deployment process with the following parameters: 80 kV voltage, 4 mA current, 90 s exposure time and a dose of 2.50 mGy. In vivo biodegradation was monitored in rat model using X-ray microcomputed tomography (Skyscan 1176, Bruker). The CT images were captured at specific time at day 0, 14, 70 and 140. The images were acquired with the following settings: 50 kV source voltage, 50 µA source current, Al 0.5 mm filter, and camera pixel size of 12.64 µm. The acquired images were reconstructed into 3D images using the NRecon software (reconstruction of cross-section view, Bruker) and the CTvox software (reconstruction of 3D images, Bruker).
In vivo acute ECoG recording with leg stimulation
The entire ECoG data was acquired from anaesthetized canine models using a multichannel recorder (IX-RA-834, iWorx, USA). A needle-type reference and ground electrode were placed on the frontal skull after the device deployment. Additionally, a hook-type electrode was carefully positioned on the sciatic nerve to provide leg stimulation. An electrical input (20 mVpp, 1 kHz, 100 ms interval) was delivered through the leg stimulation electrode to induce trembling motion at the leg. The ECoG signals were recorded with a bandpass filter between 0.1 and 2000 Hz (sampling rate of 2000 sample/s) and the acquired signals were analyzed by LabScribe software (iWorx). To preprocess the recorded data, a low-pass filter (< 100 Hz) and a notch filter (60 Hz) were used to eliminate interferences.
In vivo chronic ECoG recording
Chronic ECoG recording was carried in the same manner as the acute ECoG recording procedure without leg stimulation. An additional Pt electrode was placed adjacent to the Mo electrode for reference. Before the recording, a C-arm image was taken to verify the precise electrode location. I/O interfaces of the device were securely fixed on the skull using a dental adhesive, and the exposed part of the I/O interfaces was positioned over the scalp for easy connection with a recording device. The beagle was medicated with a pain reliever (metacam, 0.2 mg/kg) twice daily for three days to ensure the comfort of the beagle during the recording period.
Immunohistochemistry
For immunohistological analysis, brain tissue was collected from euthanized rats and divided into four equal parts in a sagittal plane based on the center. Each tissue was fixed by immersion in a 4% paraformaldehyde solution, and then embedded in paraffin blocks and sliced into 10 µm thick sections using a microtome. The paraffin-embedded sections were subjected to deparaffinization with xylene three times and hydration with decreasing amounts of ethyl alcohol. For immunofluorescence analysis, primary and secondary antibodies were applied to observe the target proteins according to the previous study46. The sections were first pretreated for antibody staining and then incubated overnight at 4°C with the following primary antibodies: chicken anti-GFAP (ab4674, 1:1000, Abcam) and anti-Iba1 (ab178846, 1:1000, Abcam). Subsequently, sections were incubated with the Alexa Fluor 488 goat anti-chicken IgY secondary antibody (ab4784, 1:1000, Abcam) and Alexa Fluor 568 goat anti-rabbit IgG (ab175471, 1:1000, Abcam) for 1 h. The stained sections were mounted with the mounting medium and observed using a confocal microscope (FV3000, Olympus). For the biocompatibility study, the specimens were embedded in optimal cutting temperature compound with sucrose infiltration, sectioned, and processed by haematoxylin and eosin staining (BBC Biochemical). The stained sections were observed under a microscope with a 10× objective lens (BX71, Olympus). TUNEL assay (ab206386, TUNEL Assay Kit – HRP-DAB (horseradish peroxidase- 2,3-diaminophenazine), Abcam) was performed according to the manufacturer’s protocol. In brief, the sectioned samples were washed with 1× PBS for 30 min and then incubated in a permeabilization solution for 2 min at 4°C. A 50 mL of DAB mixture solution was added to the sectioned samples and incubated in a humidified chamber for 15 min at 37°C in a dark place. After washing the samples with dH2O and ethanol, the samples were mounted with glass coverslip using mounting media after several immersions in xylene. The stained sections were observed under a microscope with a 10× objective lens.
Blood analysis
For blood cell analysis, 300 µL of blood samples were collected from the tail vein of both sham and implanted rats at 2, 4 and 6 weeks. The collected blood samples were placed in a 0.5 mL EDTA tubes (Becton-Dickinson) and mixed for 5 min to prevent blood coagulation. Overall haematology from collected blood cells was monitored by a haematology analyzer (ADVIA 2120i, Siemens). For blood chemistry analysis, 500 µL of blood samples were collected from tail vein of both sham and implanted rats at 2, 4 and 6 weeks. The collected blood samples were placed in 5 mL STT tubes (Becton-Dickinson) and allowed for coagulation for 20 min. After coagulation, the blood samples were centrifuged for 10 min at 300 rpm and − 4 ℃ to separate the cells and serum. Chemistry analysis of the collected serum was monitored by a biochemistry analyzer (7180, Hitachi).