Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Scalable batch fabrication of ultrathin flexible neural probes using a bioresorbable silk layer

Scalable batch fabrication of ultrathin flexible neural probes using a bioresorbable silk layer Flexible intracerebral probes for neural recording and electrical stimulation have been the focus of many research works to achieve better compliance with the surrounding tissue while minimizing rejection. Strategies have been explored to find the best way to insert flexible probes into the brain while maintaining their flexibility once positioned. Here, we present a novel and versatile scalable batch fabrication approach to deliver ultrathin and flexible probes consisting of a silk-parylene bilayer. The biodegradable silk layer, whose degradation time is programmable, provides a temporary and programmable stiffener to allow the insertion of ultrathin parylene-based flexible devices. Our innovative and robust batch fabrication technology allows complete freedom over probe design in terms of materials, size, shape, and thickness. We demonstrate successful ex vivo insertion of the probe with acute high-fidelity recordings of epileptic seizures in field potentials as well as single-unit action potentials in mouse brain slices. Our novel technological solution for implanting ultraflexible devices in the brain while minimizing rejection risks shows high potential for use in both brain research and clinical therapies. Introduction stability at the interface between conventional electrodes 7,8 and brain tissue Chronically implanted microelectrodes have been a key . This is partly due to the mechanical tool in neuroscience research by allowing the recording of mismatch between the stiffness of the materials, e.g., electrical brain activity at the level of a small population of silicon, glass, platinum, or iridium (Young’s modulus E ≈ neurons (local field potential (LFP), multiunit spiking 150 GPa), constituting such probes and the softness of the activity) and of individual neurons (single-unit activity). cerebral tissue (E ≈ 10 kPa) . This mechanical mismatch, The past decades have seen impressive technological which can be as large as seven orders of magnitude, leads developments of neural implants incorporating electrodes to irreversible tissue damage and glial scar formation, at the micrometer scale, e.g., silicon-based pin (Utah), flat resulting in failure of the device within months or even 10,11 (Michigan), or wire (floating microwire arrays), for the weeks after implantation . 1–3 characterization of neuronal activity . Such devices are To improve the brain tissue-electrode interface, research now routinely used in animal studies . Although long- has focused on the use of flexible probes, which would lasting recording is sometimes achieved using these achieve better compliance with the surrounding neural 5 10,12 probes , large variations in electrical recording capabilities tissue and minimal rejection . The fabrication of these have often been reported . The implementation of long- compliant devices typically involves either the use of soft 13–15 lasting intracerebral recordings is limited by the lack of polymeric materials as substrates, e.g., parylene , 16,17 18 polyimide , polydimethylsiloxane (PDMS) , hydro- gels , and/or the use of significantly thinner stiff materi- 20,21 Correspondence: Ali Maziz (ali.maziz@laas.fr) als . However, an important issue with flexible probes is LAAS-CNRS, 7 Avenue du Colonel Roche, F-31400 Toulouse, France that they have a tendency to fail penetrating the brain CerCo, Université Toulouse 3, CNRS, Pavillon Baudot, CHU Purpan, BP 25202, meninges and reach their location goal. Indeed, a device 31052 Toulouse, France Full list of author information is available at the end of the article © The Author(s) 2022 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to theCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 2 of 11 that is too soft tends to bend when pressed against a rigid film, allowing the degradation of the stiffening layer surface, such as the pia mater . Strategies have been through proteolytic reactions within a programmable explored to find the best way to implant flexible probes time lapse. Furthermore, insertion trials in artificial brain into the brain while still maintaining their flexibility once phantoms are achieved without buckling of the probe or positioned. Some teams have focused on the use of a stiff undesired alteration of the electrical properties of the 20,22,23 shuttle that is removed immediately after implan- electrodes. We also demonstrate successful ex vivo insertion of the probe with high-fidelity acute recordings tation, while others have promoted the integration of a stiff bioresorbable coating that is not removed but dissolves of epileptic activity as well as single-unit action potentials 24,25 inside the brain on a time scale of minutes to days . in mouse brain slices. The possibilities offered by our Due to the tissue trauma caused by the implantation and approach are very promising for the development of withdrawal of a stiff shuttle, the integration of bioresorb- ultraflexible probes for application in research and brain able coatings as a temporary stiffener has been shown to therapy. better address both mechanical and biological failures . Various bioresorbable polymers, e.g., poly(ethylene glycol) Results and discussion (PEG), polylactic acid (PLA), chitosan, and silk fibroin, Development of the bilayered probes have been reported to be excellent candidates to add to Our ultrathin flexible probes are developed using stan- polymeric implants for facilitating insertion into the dard microsystem techniques. The probes are made of a brain . In addition, they benefit to some extent from silk-parylene bilayer that comprises four gold microelec- common attributes, such as a high Young’s modulus, trodes coated with the conducting polymer poly(3,4-ethy- proven biocompatibility for in vivo application purposes, lenedioxythio-phene):poly(styrene-sulfonate) (PEDOT:PSS) and resorption when in contact with biological tissues . to lower the impedance and obtain a better signal-to-noise 13,30 Although the reported flexible implants incorporating a ratio for neuron recording . The schematics of the fab- biodegradable coating have shown successful short-term rication procedures and the probe structure are shown in electrical recordings, their fabrication process is incom- Fig. 1a. Briefly, there are three main fabrication stages. In patible with standard microfabrication techniques. Man- stage I, a cellulose acetate (CA)-coated glass substrate is ual handling is required, which limits further downscaling coated with a silk fibroin layer before (stage II) an ultrathin (to less than 10 µm) and makes it difficult and time- parylene-based structure is microfabricated on top of the 12,28,29 consuming to create many devices in parallel . Fur- silk layer using standard top-down lithography techniques (discussed in more detail later in the paper). Finally, in stage thermore, existing flexible probes require the additional preparation of a carrier support for the biodegradable III, the whole silk-parylene probe undergoes microscale coating, with a manual assembly procedure that increases shaping by a reactive ion etching (RIE) process before being the difficulty of reducing the dimensions and volume of released from the glass substrate. the probe. These coated probes may generate surgical There are a couple of key considerations in this design. footprints of a volume considerably larger than the elec- First, the silk fibroin material is chosen among other trode itself, which may induce trauma to brain tissue in bioresorbable materials on the basis of its excellent bio- 24,25 the range of hundreds of micrometers to millimeters . compatibility, tunable biodegradability, and high Cellular and/or vascular damage may then elicit sustained mechanical strength . Indeed, silk fibroin possesses a inflammation and tissue responses. A global rethinking of Young’s modulus of E ≈ 3 GPa so that the thickness of the production framework is therefore needed to effec- parylene-C can be drastically reduced to the µm level tively achieve the simple integration of biodegradable while allowing for the handling and implantation of the coatings to allow further development of minimally probe. The biodegradable silk layer is described in detail invasive neural probes. below because of its specific chemical structure and its To achieve this goal, we report here a versatile fabri- importance in the final device. Active research is currently cation framework utilizing a bioresorbable silk fibroin underway on the use of silk biomaterials, particularly to layer that can be integrated in a microfabrication process obtain a biodegradable coating for implantable neural for preparing ultrathin parylene-based penetrating probes. recording devices . However, no work has yet been The probes consist of a silk-parylene bilayer, which is reported on the integration of degradable biopolymers in obtained by successively depositing layers on top of each the batch fabrication of neural recording devices, allowing other. The first layer is obtained using a degradable silk for further scaling down (less than 10 μm thick) and fibroin coating as a temporary stiffener that allows for the preventing de facto any manual handling procedure and insertion of ultrathin parylene-based flexible devices deep buckling issues during implantation. into the brain. The additional subtlety of the process The thickness of the silk layer can be tuned by adjusting derives from the exposure of the silk fibroin layer to the polymer dilution percentage or the volume of the methanol, which increases the crystallized domains in the solution on the casting area. As an example, a dose of Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 3 of 11 Au/Ti Cellulose acetate Parylene C Glass wafer Silk fibroin Ø 40 µm 250 µm d 5 mm e f 3h 6h 12h 24h 1 cm Au/Ti 4 µm 1 cm Silk fibroin 30 µm 0 250 µm Parylene C Days 0.8 0.6 0.4 30 µm 500 µm 0.2 0 20406080 % CF in O plasma 4 2 Fig. 1 Fabrication of the bilayered silk-parylene probes. a Schematic illustration of the main fabrication steps on a glass substrate: (1) cellulose acetate deposition, (2) silk fibroin deposition, (3) parylene base layer deposition, (4) gold microelectrode patterning, (5) parylene top layer deposition and (6) final shaping of the silk-parylene probe by RIE and release from the substrate. b Picture of the 4-inch glass substrate after batch microfabrication. The substrate contains 80 elements (less than 1 cm each). c Schematic illustration of the bilayered silk-parylene probe with characteristic dimensions. The device contains four recording microelectrodes with diameters of 40 μm patterned on a 3 mm-long and 250 μm-wide shank. d Picture of the microfabricated implant. The inset shows the corresponding magnified view of the 4 gold microelectrodes. e Picture of the silk-parylene probe highlighting the contact pads bonded with the FFC cable for the following electrochemical and electrophysiological measurements. f Enzymatic degradation of silk fibroin after different immersion times in methanol. g Etching rate of silk fibroin vs. percent of CF in O plasma at 500 W and 20 mTorr. An optimal ratio of 25/75 was found. h SEM image of the final bilayered silk-parylene shank, and i the corresponding magnified SEM image 0.1 mL/cm of 7 wt.% silk fibroin solution produces films fibroin samples degrade at different rates depending on with a thickness of ∼30 ± 5 μm (± SD). the methanol treatment time. For instance, we find that Silk degradation time is another critical parameter. films treated for 3 and 6 h degrade within a few hours Depending on the implantation strategy, the probe posi- while those treated for 12 and 24 h last up to one week tion may need to be adjusted for a prescribed time of (Fig. 1f). This control over the degradation time is con- several minutes or more, during which it must remain sistent with the literature, where reports have shown stiff. This critical time is determined by the degradation proteolytic degradation in vitro of water-stable silk films 25,32 rate of the bioresorbable coating. The lifetime of the after approximately two weeks . bioresorbable layer can be specifically adjusted via the The second step consists of fabricating an ultraflexible crystallinity of the silk protein, i.e., the β-sheet content . parylene-based probe on top of the silk fibroin substrate Indeed, treating silk by immersion in methanol increases (Fig. 1a) (more details are available in the Materials and the β-sheet content in the film. In a proteolytic medium Methods). Parylene is chosen as the substrate for its well- (protease XIV (PXIV) in PBS solution (1 U/ml)), silk documented biocompatibility, chemical inertness, and Flexible cable Parylene C (4µm) Silk Fibroin (30µm) -1 RIE rate (µm.min ) Silk Weight (%) Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 4 of 11 high electrical and moisture insulation properties. the manipulator approaching and (2) contacting the gel, Figure 1c shows the design of the ultraflexible parylene- (3) the probe successfully perforating the gel, (4) the based probe, which consists of three layers: a top parylene probe inserted in the gel, and (5 and 6) the probe deeply layer (thickness: 1 μm) for encapsulation, a middle Au inserted into the gel. The bilayered silk-parylene probes layer (thickness: 200 nm) for the electrophysiological thus succeed in penetrating the brain phantom without measurement, and a bottom parylene layer (thickness: any buckling or bending, which clearly indicates that the insertion shuttle is sufficiently stiff. From the insertion 3 μm) for mechanical support. The total thickness of the probe is only 4 μm. experiment (Fig. 2a), the first peak force (stage 3) is 0.7 The shaping of the silk-parylene probe is achieved by mN, corresponding to the minimum force required to dry etching, which is performed through RIE using a 75/ penetrate the tissue-mimicking gel. Other works show a 25,33 25 O /CF gas ratio. Under these conditions, the micro- similar minimum force . As Movie S1 (Supporting 2 4 fabrication of bilayered silk-parylene microstructures can Information) shows, the probe is explanted from the gel be performed with precise control of the geometry, size, without mechanical damage and successfully reinserted and shape, as illustrated in Fig. 1d, h and i. The proof of into the gel a second time. It is worth mentioning that the concept presented in Fig. 1b contains 80 elements, and bare parylene shank itself, being only 4 µm thick, curls each device contains four recording gold microelectrodes and cannot be handled or manipulated without the with diameters of 40 μm patterned on 3 mm-long and bioresorbable silk polymer support. 250 μm-wide probes. In other words, large-scale batch The type of brain phantom we used has been widely fabrication of precisely defined silk-parylene probes can used as a model of implantation in gray and white mat- 25,34 be performed directly on the glass substrate. Finally, the ter , yet it does not take into consideration other brain cellulose acetate sacrificial layer is dissolved in acetone to features. Indeed, the minimum insertion force also release the bilayered silk-parylene probes from the sub- depends on animal species, on the biological tissue (pia or strate (Fig. 1d). The probe is robust enough to easily bond dura mater, gray or white matter), on the size and shape of directly with external electrical connections (Fig. 1e). the shank, on the speed of approach, etc . For example, higher insertion forces are needed for dura mater pene- Insertion testing in brain phantom tration . Our fabrication technology allows complete Our bilayered probes are designed as insertable ultra- freedom in terms of materials, size, shape, and thickness. thin flexible devices that do not buckle during insertion in Thus, the probe buckling strength can be easily tuned the brain. An important parameter lies in the mechanical depending on the biological target. stability of the stiffening resorbable coating during the We next performed in vivo implantation in an anes- implantation/explantation process. To test the rigidity thetized mouse to evaluate the ability of the silk-stiffened afforded by the coating, we performed cyclic insertions of probe to achieve more difficult tasks of penetration, e.g., the probe into a brain tissue-mimicking phantom (1 wt.% through the pia mater. The dura mater is resected prior to agarose gel, Young’s modulus of 40 kPa). Figure 2b shows implantation. The probe is lowered with a manual the whole process of implanting the probe into the brain micromanipulator through the pia mater and is left in the tissue-mimicking gel, showing (1) the probe mounted on somatosensory cortex for one hour after insertion, such a b 1.5 2 mm 3 2 mm 2 mm Approaching Contacting Dimpling 4 5 2 mm 2 mm 6 2 mm 0.5 Perforating Inserting Deep insertion 0 0.5 1 1.5 2 Travel distance (mm) Fig. 2 Insertion test of the silk-parylene probe in brain phantoms. a Force profile during insertion of the shank in 1 wt.% agarose gel (Young’s modulus of 40 kPa), and the numbers correspond with those in (b). b Optical pictures showing the different stages of insertion in correlation with the evolution of the force Force (mN) Force (mN) Probe length (mm) Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 5 of 11 that the silk coating has enough time to dissolve. linked to its Young’s modulus E; the area moment of Figure S4 shows images of the silk-coated probe during inertia I (I = wt /12) along the x-axis; the length L, width x x insertion in the mouse brain. Insertion of the probe w, and thickness t of the probe; and the effective length occurs with no sign of buckling, which is consistent with factor of the column K. The cross-section of the device is what we observed in the in vitro insertion tests on the gel considered a constant rectangle, and the shanks are beams brain phantoms (Fig. 2). We do not notice any visible fixed at one side (K = 0.7). Assuming that the E values for 29,33 damage to the electrodes after insertion and retraction. parylene and silk fibroin are 3 GPa , the calculation These results indicate that our probe design is compatible yields a theoretical buckling force of 7.2 ± 2.3 mN for the with in vivo implantation. probe (Fig. 3c). The calculated theoretical buckling force predicts the trend and order of magnitude of the experi- Standard compression tests mental values (Fig. 3d). We further studied the mechanical properties of our In addition to axial compression tests, the mechanical bilayered probes against a hard substrate. Axial com- stability of our probes under long-term cyclic compres- pression of beams eventually results in their buckling. The sion was tested under cyclic loading with a bending radius highest force that a sample can withstand before bending of 0.3 mm. As shown in Fig. 3e, the original buckling force is called the buckling force F . As depicted in Fig. 3a, (first cycle) is 9.1 ± 0.5 mN and shows minimal change buckling compression tests on a hard substrate show the clear (only a 10% decrease) after nearly 1000 bending cycles. superiority of the bilayered silk-parylene assembly in After prolonged axial compression tests, the bilayered terms of its mechanical strength. The silk coating silk-parylene probes do not show any peeling or sign of improves the probe strength, with an average buckling delamination. This indicates that these engineered strength of 10.9 ± 1.3 mN (Fig. 3d), which is 15 times bilayered silk-parylene probes can endure a very large higher than the force required to penetrate the brain- number of contact cycles without any damage or mimicking gel (Fig. 2a). These experimental values were delamination. compared with theoretical analysis predictions (Fig. 3c). Our probes are modeled as single beams, whose buckling Electrical and electrochemical characterization 2 2 force is defined by Euler’s formula: F = π I E/(KL) . Possible alteration of the electrical and/or electro- buckling x The buckling force along the x-axis of a clamped beam is chemical properties of the probe should also be ab c Force (mN) 3 4 1 2 39.40 35.46 Buckling 35 31.52 2 mm 2 mm 27.58 23.64 2 25 19.70 2 mm 15.76 2 mm 0 0.2 0.4 0.6 11.82 Travel distance (mm) 7.880 de 10 10 3.940 8 0.000 2.2 40 2.4 2.6 2.8 3.2 20 3.4 2 3.6 2 3.8 510 15 20 0 200 400 600 800 1000 Buckling force (mN) Number of cycles Fig. 3 Compression tests of the bilayered silk-parylene probes against a hard substrate. a Force profile during compression, numbers and buckling position, which correspond with those in (b). b Series of optical pictures showing the corresponding stages of shank compression in correlation with the evolution of the force. c Theoretical influence of the shank length (in mm) and thickness (in µm) on the buckling force. The model was clamped-pinned, the cross-section of the device was considered a constant rectangle, and the shanks were beams fixed on one side and pinned on the other side (K = 0.7). We assumed that E for parylene C and silk fibroin was 3 GPa. d Electrode-buckling force histogram of the silk- parylene probes (N = 20). e Monitoring of the average buckling force as a function of the number of bending cycles (1000 cycles) with a bending radius of 0.3 mm Probe thickness (µ µm) Force (mN) Counts Buckling force (mN) Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 6 of 11 4 bc a 0 PEDOT:PSS PEDOT:PSS PEDOT:PSS -10 Au Au Au 3 -20 -30 -40 -50 -60 1 4 -70 -80 10 0 1 2 3 4 1 2 3 4 10 10 10 10 0 2 4 6 8 10 12 14 10 10 10 10 Frequency (Hz) Re(Z) (kOhm) Frequency (Hz) ef -4 15 2x10 250 35 PEDOT:PSS PEDOT:PSS Au Au -4 1x10 25 0 15 -4 -1x10 0 0 Z Z CSC CSC 02345 -0.5 0 0.5 Au PEDOT:PSS Au PEDOT:PSS Days Potential (V) Fig. 4 Electrochemical characterization of the microelectrodes. a EIS measurements of the gold- and PEDOT:PSS-coated microelectrodes in PBS at room temperature from 10 Hz to 7 MHz. b Corresponding phase vs. frequency plotting of gold- and PEDOT:PSS-coated microelectrodes. c Nyquist diagram at frequencies ranging from 10 Hz to 7 MHz of gold- and PEDOT:PSS-coated microelectrodes. d CV in the PBS buffer at room temperature by potential sweeping between −0.6 and +0.6 V at 200 mV/s vs. the Ag/AgCl reference electrode of gold- and PEDOT:PSS-coated microelectrodes. e Comparison of the electrochemical characteristics (Z at 1 kHz and CSC) between the gold- and PEDOT:PSS-coated microelectrodes. f Impedance evolution at 1 kHz of the PEDOT:PSS-coated microelectrode in PBS-soaked gel, obtained from the EIS measurements We next evaluated the charge transfer capabilities of the investigated to ensure their inherent electrode perfor- mance. We evaluated the electrical properties of the microelectrodes. CV from −0.6 and 0.6 V at a scan rate of microelectrodes by electrochemical impedance spectro- 0.2 V/s was performed, and the cathodal CSC was calcu- scopy (EIS), cyclic voltammetry (CV), and calculations of lated by the time integral of the cathodal currents within the charge storage capacity (CSC). Figure 4a, b shows the the cycled region (Fig. 4d). The cathodic CSC increases −2 Bode plots across the frequencies of interest from an average value of 1.95 ± 0.3 mC cm for the gold −2 (10 Hz–7 MHz) for a gold microelectrode with a diameter microelectrodes to 31.5 ± 1.3 mC cm for the PEDOT- of 40 µm before and after the electrochemical deposition coated microelectrodes. A higher charge capacity results of PEDOT:PSS. The mean impedances at 1 kHz are used in higher charge injection, which is desirable for electrical for comparison, as action potentials have a characteristic stimulation. frequency band centered at that frequency (Fig. 4e). In addition, to track the integrity of the probe structure Before PEDOT:PSS deposition, the average impedance is and electrical connections, in vitro EIS measurements 210 ± 8.2 kΩ (n = 5) in PBS, while the mean impedance were carried out daily for a 5-day period in saline brain decreases to 9.4 ± 0.9 kΩ after polymer deposition. This phantoms (gels soaked in PBS) with four different probes. well-known phenomenon is due to an increase in the Figure 4f shows the impedance measured at 1 kHz after effective surface area with the formation of PEDOT:PSS the insertion of the silk-parylene probe in PBS-based gel material, leading to a decrease in impedance of the as a function of immersion time. EIS measurements over microelectrode. The corresponding phase plot of the the 5-day period demonstrate a small decrease (~11 ± impedance reveals that the PEDOT:PSS microelectrode is 1.2–8 ± 0.5 kΩ) followed by a stabilization of the impe- capacitive in the low frequency range (10 Hz) and more dances. We observe no distortion of the probe upon resistive at higher frequencies (Fig. 4b). The Nyquist plot repeated insertion into the gel, which indicates that the recorded in PBS is presented in Fig. 4c. The deposition of thickness of the silk layer remains sufficient to carry the PEDOT:PSS produces a very small radius of the semicircle full ultrathin structure without compromising the elec- on the Nyquist plot with a charge transfer resistance of trical integrity of the parylene probe. The stable electrical approximately 7.8 kΩ, revealing the low electron-transfer impedance throughout the process proves the robustness resistance associated with the polymer coating. of our design and protocol for in vivo implantation. Current (mA) Impedance (KOhm) Impedance (KOhm) Phase (deg) Charge storage capacity -Im(Z) (KOhm) (mC/cm ) Impedance Z (KOhm) Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 7 of 11 In vitro proteolytic degradation depends on both implant-related factors (molecular To track the integrity of the structure, we performed weight, molecular structure, crystallinity, etc.) as well as incubation studies in PBS containing the enzyme protease on host-related factors (i.e., immune response at the XIV (Fig. 5). For this experiment, probes are crystallized implant site) . The silk biodegradability of the silk- for 6 h in methanol. Most naturally biodegradable poly- parylene probes is measured as the loss of weight of the mers degrade in vivo via enzymatic degradation, owing to implant during continuous incubation at 37 °C in pro- the stability of their backbone structure. It is generally teolytic solution for one week. In the proteolytic envir- accepted that silk protein is degraded mainly by pro- onment, the silk-parylene probes demonstrate a gradual tease . Protease XIV has been widely used to mimic decrease in mass, corresponding to slow protein frag- extracellular degradation mediated by proteolytic mentation during incubation. This experiment shows the enzymes. However, in vivo, the degradation of silk importance of using a biodegradable silk coating as a temporary stiffener to deliver ultrathin parylene-based flexible devices in deep tissue. It is worth mentioning that the bare parylene shank, which is only 4 µm thick, curls and cannot be handled or manipulated without the mechanical silk support. Electrophysiological recordings from mouse brain slices Electrophysiological experiments were conducted on mouse brain slices to assess the recording quality of the probes. The recordings were performed after the silk coating had completely dissolved to ensure that the 4 µm- thick flexible parylene probes were still functional for 1 2 34 5 acute recordings. The silk coating process was tuned so Days that the coating would completely dissolve in the slice within 30 min (corresponding to a 1 h long immersion of Fig. 5 In vitro proteolytic degradation of the silk layer. Enzymatic the probe in methanol for crystallization). Probe insertion degradation of the silk in the bilayered silk-parylene probes by protease XIV (PXIV) solution (1 U/ml) for 5 days at 37 °C was aimed at layer 2 of the piriform cortex (Fig. 6b). This 300–3000 Hz a c 0–100 Hz 50 µV 1 s b d e 10 µV 10 µV 2 s 0.5 ms Fig. 6 Electrophysiological recording in the piriform cortex of a mouse brain slice. a Photograph of a 35 μm-thick silk parylene probe inserted into a mouse brain slice: insertion occurs with no sign of buckling. b Scheme of the placement of the silk-parylene neural probe. c Traces show two consecutive epileptic seizures. The upper trace corresponds to the bandpass filtered signal (300–3000 Hz) used to reveal spiking activity, and the bottom trace corresponds to the low pass filtered signal (cutoff at 100 Hz) used to reveal slow changes in the LFP. d Traces show spontaneous spiking activity occurring between epileptic seizures. The dashed line indicates the threshold used for sorting action potentials. It is set at −4xRMS of the voltage trace (RMS calculated outside of the epileptic seizure). The action potentials that cross the threshold are examined through cluster analysis and interspike interval analysis. e Overlap of the action potentials recorded over 5 min at high temporal resolution. The constancy of the spike shape and refractory period (6.4 ms in that case) definitively ensures that the blue action potentials belong to one single unit. The action potentials in black correspond to multiunit activity Weight (%) Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 8 of 11 layer is characterized by a high density of neuronal cell mouse brain in vivo. Insertion in ex vivo mouse brain bodies. As spontaneous activity is low or null in most slices allowed acute recordings of LFP, as well as spon- regions of the mouse brain ex vivo, we added taneous multi and single-unit activities. These results 4-aminopyridine (4AP, 100 µM) to the superfusion solu- suggest that our design provides a versatile technological tion to activate the slices. 4AP infusion usually leads to solution for producing ultraflexible and ultrathin devices the appearance of spiking activity and most often induces that might be implanted in the brain with minimal rejection. In future works, we will directly address this epileptiform activity . The latter provides an opportunity to report on the capability of the probe to record LFPs in issue by performing in vivo neural recording and by addition to single- and multiunit spiking activity. Exam- correlating the quality of the recording with the presence ples of epileptic and spiking activities are presented in or lack of inflammatory reaction and glial scar formation. Fig. 6c–e. Epileptic activity is characterized by the occurrence of epileptic seizures in the LFP, which are Materials and methods evidenced by low pass filtering (0–100 Hz) of the voltage Chemicals trace (Fig. 6c, bottom trace). Bandpass filtering Parylene C dimer (PXC) was purchased from Comelec (300–3000 Hz) of the same original trace reveals the SA. ECI 3012 photoresist was purchased from Micro- action potentials of neurons (Fig. 6c, top trace). Large chemicals GmbH. MF CD26 developer was purchased bursts of action potentials can be observed at the time of from MicroChem. 3,4-Ethylene dioxythiophene (EDOT), epileptic seizures. Action potentials can also be seen poly(sodium 4-styrenesulfonate) (NaPSS), and acetone between epileptic seizures (Fig. 6c, d). We examined were purchased from Sigma Aldrich. Phosphate-buffered whether single-unit activity, i.e., action potentials that can saline (PBS, Gibco DPBS 1X) was purchased from Fisher be attributed to the activity of one single neuron, could be Scientific. Platinum (Pt) counter electrodes and silver/ observed in these periods. For this purpose, portions of silver chloride (Ag/AgCl) reference electrodes were pur- traces (2 ms wide, Fig. 6e) were extracted using a chased from WPI. Solutions were prepared with deionized threshold set at −4xRMS of the voltage trace (dashed water (18 MΩ). lines in Fig. 6d, e). These were then submitted to PCA and clustering (not illustrated). This analysis allows for seg- Bioresorbable silk fibroin solution preparation regation of different spike shapes. The spike shapes Silk fibroin aqueous solution was prepared from Bom- belonging to a given cluster are attributed to one single byx mori cocoons following a protocol detailed pre- unit if the interspike interval distribution histogram (not viously . The fibroin protein was first extracted from the illustrated) shows a clear refractory period, i.e., no interval silk fibers by boiling the silk cocoons (5 g) in a solution of <1 ms. Otherwise, the cluster corresponds to multiunit 0.02 M Na CO for 30 min. The regenerated silk fibroin 2 3 activity, that is, action potentials that are issued from at was then recovered and rinsed thoroughly in deionized least two neurons and that cannot be segregated. The blue water before being dried overnight under ambient con- action potentials in Fig. 6e correspond to the activity of ditions. The dried silk fibroin (3.6 g) was dissolved in a one single unit, identified as such by its constant spike 9.3 M LiBr solution at 60 °C for 4 h. The salt was then shape and a refractory period of 6.4 ms. The action removed by dialyzing the solution against deionized water potentials in black correspond to multiunit activity. for 24 h at room temperature using a dialysis membrane Overall, multiunit activity is recorded in all trials, and (MWCO 3.5 KD, Spectra/PorTM) and regularly changing single units are recorded in 7/12 trials. the water. Centrifugation was performed to remove impurities. The recovered silk fibroin solution had a final Conclusion concentration of 7 wt.%. This work proposes a novel and versatile approach to fabricate, pattern, and deliver ultrathin probes consisting Probe microfabrication of a silk-parylene bilayer. The biodegradable silk layer A 4-inch glass wafer was used to prepare the overall provides a temporary stiffener that can be used to deliver process. The glass substrate was first cleaned by MW- ultrathin parylene-based flexible devices in deep tissue. oxygen plasma (800 W, 10 min, O ) prior to processing. Our innovative and robust batch fabrication technology The fabrication began with the deposition of a cellulose allows complete design freedom of the probe in terms of acetate layer (~2 μm) by spin-coating at 1000 rpm for 30 s materials, size, shape, and thickness. We systematically (5% w/vol in acetone). It acted as a sacrificial layer to studied the behavior of the bilayered structure in gel brain release the final device from the substrate. The silk fibroin phantoms and demonstrated that parylene probes as thin aqueous solution (7 wt.%) was deposited by drop casting as 4 μm could be delivered accurately to a desired depth and left to dry under ambient conditions overnight, with intact geometry and electrical functionality. We also resulting in an ~30 µm-thick silk film. The thickness of demonstrated successful insertion of the probe in the the resulting film was controlled by adjusting the volume Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 9 of 11 of the silk fibroin solution. Then, a 3 μm-thick layer of the test, the implants were held between two glass plates, Parylene C (PXC) was deposited onto the silk-coated and compression was carried out over a tip length of substrate through CVD using C30S Comelec equipment. approximately 3.4 mm. The silk-parylene probes were A 50/200 nm layer of Ti/Au was then deposited by eva- fixed on a MARK-10 ESM303 test bench coupled with a poration and patterned through an electroplated nickel- MARK-10 M5-012 force sensor. An internally developed based shadow mask. Another 1.3 μm top layer of Parylene LabVIEW program allowed the equipment to be con- trolled and the force to be monitored according to the C was deposited onto the processed metal layer. The shape of the electrode pads, contacts and device body displacement of the implant. Compression tests were −1 were defined by photolithography steps followed by RIE in performed at a slow speed of 2 mm min for optimal O /CF plasma (75/25) at 500 W and 20 mT. Finally, the monitoring of the buckling force. A video recording 2 4 cellulose acetate sacrificial layer was dissolved with acet- (Movie S1) of the compression test using a Dino-Lite Edge one to release the bilayered silk-parylene probes. camera was made to complete the experiment. Electrochemical characterization In vitro insertion into the gel brain phantoms Electrochemical characterization was performed with a Insertion tests of the probes were performed using 1% 3-electrode system that included a Pt wire as the counter w/v agarose gel brain phantoms imitating the mechanical electrode, a Ag/AgCl wire as the reference electrode and the properties of brain tissue. In the same way as in the gold microelectrodes from the silk-parylene probe as the compression tests, the silk-parylene probes were held working electrodes. EIS and cyclic voltammetry (CV) were between two glass plates and fixed to the MARK-10 test performed using a Bio-Logic VMP3 potentiostat. CV was bench. To limit damage during the insertion tests and to performed in PBS at room temperature by potential have optimal monitoring of the forces involved, the sweeping between −0.6 and +0.6V at 200mV/s vs. Ag/ experiment was carried out at a very low speed of 0.5 mm −1 AgCl reference, allowing cathodal charge storage capacity min . A video recording was made using the Dino-Lite 37,38 (CSCc) evaluation . EIS was also performed in PBS at Edge camera to observe the different stages of insertion in room temperature by applying a 10 mV sine wave at fre- correlation with the evolution of the force. quencies ranging from 10 Hz to 7 MHz. Improved electrical properties were achieved by PEDOT:PSS deposition. CV Dissolution tests was performed in EDOT:NaPSS solution (10 mM:34 mM) To assess the biodegradation of the silk fibroin layer, an at room temperature by potential sweeping between −0.7 enzymatic degradation test was carried out over several and +1 V at 10 mV/s vs. Ag/AgCl reference. CV and EIS days. Briefly, five implants (~2 mg) were selected and were then performed again to compare the evolution of the incubated in 1 ml of protease (Proteas XIV from S. griseus, CSCc, impedance, phase, and Nyquist results. 3.5 U/mg, Sigma-Aldrich) and PBS (1 U/ml of PBS buffer) at 37 °C. The implants were photographed and weighed Crystallization procedure of the silk fibroin every day after cleaning with DI water and dried at 60 °C Water-insoluble silk films were prepared by methanol for 10 min. The enzyme solution was changed after each treatment, which increased the crystallized domains in the weighing to maintain enzymatic activity. film, allowing degradation within a programmable time window. The bilayered silk-parylene probes were posi- Brain slice preparation tioned between two flexible filter papers and held in All procedures were conducted in accordance with the position with clips to prevent any deformation of the guidelines from the European Community (directive devices during annealing. The probes were then immersed 2010/63/UE) and from the French Ministry of Agri- in 80% methanol solution at room temperature for a fixed culture, Agri-food and Forestry (décret 2013–118) and time to increase the β-sheet crystal content. After were approved by the Ministère de l’Enseignement methanol annealing, the filter paper was removed, and the Supérieur, de la Recherche et de l’Innovation (N° 15226- implants were dried under vacuum at 40 °C for 1 h to 2018052417151228). Two- to 4-month-old C57BL/6 wild- remove all traces of methanol. type female mice were used for brain slice preparation. The protocol has been detailed previously and is briefly Standard compression tests summarized here. Mice were decapitated after deep Standard compression tests against a hard substrate anesthetization with isoflurane. The brain was removed were performed to assess the mechanical properties of the and prepared for slicing in ice-cold modified ACSF bilayered silk-parylene probes. The axial compression of (mACSF). The composition of the mACSF was (in mM): the implant allowed us to evaluate its buckling force, NaCl 124, NaHCO 26, KCl 3.2, MgSO 1, NaH PO 0.5, 3 4 2 4 which corresponds to the highest force that a sample can MgCl 9, Glucose 10. The mACSF was bubbled with a gas withstand before bending. To guarantee the stability of mixture of 95% O and 5% CO . Then, 400 µm-thick 2 2 Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 10 of 11 slices were cut with a vibratome in the presence of cold, for the implantation tests in vivo. For anesthesia, subjects oxygenated mACSF. The slices were allowed to recover received a chemical induction of ketamine/xylazine (125/ for at least one hour at room temperature in a holding 5 mg/kg) by intraperitoneal injection followed by gas induc- chamber filled with oxygenated, in vivo-like ACSF, whose tion with 3% vaporized isoflurane (TEM SEGA’s MiniHUB composition was (in mM): NaCl 124, NaHCO 26, KCl V3.Pessac).Oncethe mousewas placed on thesurgical 3.2, MgSO 1, NaH PO 0.5, CaCl 1.1, and glucose 10. platform, anesthesia was maintained with isoflurane (1.5% v/v 4 2 4 2 in O This ACSF was continuously bubbled with a 95% O /5% ). The level of anesthesia was checked by the lack of the 2 2 CO gas mixture (pH 7.4). withdrawal reflex after pinching the hindlimbs. A heating pad was positioned underneath the animal to keep the body Ex vivo electrophysiological recordings of the brain slices temperature at 37 °C. Eye dehydration was prevented by For the ex vivo recordings, a slice was transferred to a topical application of ophthalmic gel. An incision was made submersion-type chamber that was continuously gravity- through all skin layers at the top of the skull, the skin flaps fed with oxygenated (95% O /5% CO ) in vivo-like ACSF were retracted, and the tissues detached from the bone. The 2 2 at a flow rate of 3–3.5 ml/min. All recordings were per- skull was cleaned and dried with sterile cotton swabs, and a formed at 34–35 °C. For recording, two probes were used, hole was drilled using a Harvard Apparatus MicroDrill and 2 electrodes were tested for each probe. After deep (0.5 mm diameter: Les Ulis, France), allowing the dura mater insertion, one of the two probes was slowly pulled up in to be carefully removed to expose the brain. The silk-parylene steps of 50–100 µm, yielding five pairs of recordings. probes were inserted in the somatosensory (SS) cortex or Electrophysiological signals were amplified (×1000) and visual (V1) cortex (SS stereotaxic coordinates: AP:−0.9, ML: filtered (0.1 Hz–10 kHz) with a NeuroLog system (Digi- −3.0; right V1 stereotactic coordinates: AP,−3.5 mm; ML, timer, UK). Notably, 50 Hz noise was eliminated with a 2.4 mm; at a depth of 2 mm from the brain surface). All Humbug system (Quest Scientific, Canada). Low-pass procedures above complied with the guidelines of State Sci- filtering (0–100 Hz) and bandpass filtering (300–3000 Hz) entific and Technological Commission Animal procedures were used offline to isolate LFP and spiking activity, and were approved by the national Animal Care and Ethics respectively. All signals were digitized with a digitization Committee (CE2A122 protocol number 2020051519454596) rate of 20 kHz (power1401, CED, UK). The real-time following European Directive 2010/63/EU. display of the signals was achieved with an oscilloscope Acknowledgements and Spike2 software (CED, UK). Spike sorting was con- We thank Dr. Denis Calise (Institut des Maladies Métaboliques et ducted using Spike2. First, we computed the root mean Cardiovasculaires, Toulouse, France) for his help regarding animal facilities. The authors acknowledge funding from the Agence Nationale de la Recherche square (RMS) of the voltage trace (running average of (ANR-19-CE19-0002-01 and ANR ANR-15-CE19-0006). This work was supported over 5 s). Next, 2 ms wide segments of voltage traces by the French RENATECH network. (“wavemarks”) were generated whenever negative deflec- tions, likely corresponding to action potentials, exceeded a Author details 1 2 LAAS-CNRS, 7 Avenue du Colonel Roche, F-31400 Toulouse, France. CerCo, threshold set at −4 times the RMS. The wavemarks Université Toulouse 3, CNRS, Pavillon Baudot, CHU Purpan, BP 25202, 31052 extended from −0.5 to +1.5 ms relative to the peak of the 3 Toulouse, France. UMR Institut National de la Santé et de la Recherche negative deflection. The minimal wavemark overlap was Médicale 1048, Institut des Maladies Métaboliques et Cardiovasculaires, Toulouse, France set at 0.5 ms. The wavemarks were then submitted to principal component analysis (PCA) and cluster analysis. Conflict of interest Single units corresponded to clusters containing spikes of The authors declare no conflict of interest. constant shape and displaying a refractory period of >1 ms Supplementary information The online version contains supplementary were determined from interspike interval distribution material available at https://doi.org/10.1038/s41378-022-00353-7. histograms computed with a bin width of 0.2 ms. Received: 19 February 2021 Revised: 26 November 2021 Accepted: 19 Packaging December 2021 Prior to recording, the flexible device was bonded to a customized flexible ribbon cable with golden traces (AXO-00021, pro-POWER, China) by using epoxy silver References and photosensitive glue. The overall connection system 1. Jog, M. et al. Tetrode technology: Advances in implantable hardware, neu- was then passivated with insulating biocompatible glue roimaging, and data analysis techniques. J. Neurosci. Methods 117,141–152 (Polytec EP 601). (2002). 2. Kipke, D. R., Vetter, R. J., Williams, J. C. & Hetke, J. F. Silicon-substrate intracortical microelectrode arrays for long-term recording of neuronal spike activity in In vivo implantation into a mouse brain cerebral cortex. IEEE Trans. Neural Syst. Rehabilitation Eng. 11,151–155 (2003). Male C57BL/6J mice aged 15–17 weeks, purchased from a 3. Normann, R. A.,Maynard,E.M., Rousche, P. J. & Warren, D. J. A neural interface for a cortical vision prosthesis. Vis. Res. 39,2577–2587 (1999). commercial supplier (Charles River, Eculy, France), were used Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 11 of 11 4. Nicolelis, M. A. et al. Chronic, multisite, multielectrode recordings in macaque Conference of the IEEE Engineering in Medicine and Biology Society 871–874 monkeys. Proc. Natl Acad. Sci. USA 100, 11041–11046 (2003). (IEEE, 2012). 5. Okun, M., Lak, A., Carandini, M. & Harris, K. D. Long term recordings with 23. Williamson, A. et al. Localized neuron stimulation with organic electrochemical immobile silicon probes in the mouse cortex. PLoS One 11, e0151180 (2016). transistors on delaminating depth probes. Adv. Mater. 27, 4405–4410 (2015). 6. Rousche, P. J. & Normann, R. A. Chronic recording capability of the Utah 24. Pas, J. et al. A bilayered PVA/PLGA-bioresorbable shuttle to improve the Intracortical Electrode Array in cat sensory cortex. J. Neurosci. Methods 82,1–15 implantation of flexible neural probes. J. Neural Eng. 15, 065001 (2018). (1998). 25. Lecomte, A. et al. Silk and PEG as means to stiffen a parylene probe for 7. Lacour, S. P., Courtine, G. & Guck, J. Materials and technologies for soft insertion in the brain: toward a double time-scale tool for local drug delivery. J. implantable neuroprostheses. Nat. Rev. Mater. 1,1–14 (2016). Micromech. Microeng. 25, 125003 (2015). 8. Chen, R., Canales, A. & Anikeeva, P. Neural recording and modulation tech- 26. Lewitus,D., Smith, K. L.,Shain,W.&Kohn,J.Ultrafast resorbingpolymersfor nologies. Nat. Rev. Mater. 2,1–16 (2017). use as carriers for cortical neural probes. Acta Biomater. 7,2483–2491 (2011). 9. Muthuswamy, J., Saha, R. & Gilletti, A. Tissue micromotion induced stress 27. Feig, V. R.,Tran, H. &Bao,Z.Biodegradable polymeric materials in degradable around brain implants. In 3rd IEEE/EMBS Special Topic Conference on Micro- electronic devices. ACS Cent. Sci. 4,337–348 (2018). technology in Medicine and Biology, 2005 102–103 (IEEE, 2005). 28. Felix, S. H. et al. Insertion of flexible neural probes using rigid stiffeners 10. Jeong, J.-W. et al. Soft materials in neuroengineering for hard problems in attached with biodissolvable adhesive. J. Vis. Exp.: JoVE https://doi.org/10.3791/ neuroscience. Neuron 86,175–186 (2015). 50609 (2013). 11. Polikov,V.S., Tresco,P.A. & Reichert,W.M. Responseofbrain tissue to 29. Weltman, A., Yoo, J. & Meng, E. Flexible, penetrating brain probes enabled by chronically implanted neural electrodes. J. Neurosci. Methods 148,1–18 (2005). advances in polymer microfabrication. Micromachines 7, 180 (2016). 12. Lecomte, A., Descamps, E. & Bergaud, C. A review on mechanical considera- 30. Maziz,A., Özgür, E., Bergaud,C.& Uzun,L. Progress in conductingpolymersfor tions for chronically-implanted neural probes. J. Neural Eng. 15, 031001 (2017). biointerfacing and biorecognition applications. Sensors Actuators Rep. 3, 13. Castagnola, V. et al. Parylene-based flexible neural probes with PEDOT coated 100035 (2021). surface for brain stimulation and recording. Biosens. Bioelectron. 67,450–457 31. Maziz, A., Leprette,O., Boyer, L.,Blatchè,C.&Bergaud, C. Tuning theproperties (2015). of silk fibroin biomaterial via chemical cross-linking. Biomed. Phys. Eng. Express 14. Rodger, D. C. et al. Flexible parylene-based multielectrode array technology for 4, 065012 (2018). high-density neural stimulation and recording. Sens. Actuators B: Chem. 132, 32. Hu, X. et al. Regulation of silk material structure by temperature-controlled 449–460 (2008). water vapor annealing. Biomacromolecules 12,1686–1696 (2011). 15. Khodagholy, D. et al. Highly conformable conducting polymer electrodes for 33. Seo,K. J.etal. Transparent, flexible, penetrating microelectrode arrays with in vivo recordings. Adv. Mater. 23,H268–H272 (2011). capabilities of single‐unit electrophysiology. Adv. Biosyst. 3, 1800276 (2019). 16. Rousche, P. J. et al. Flexible polyimide-based intracortical electrode arrays with 34. Hosseini, N. H. et al. Comparative study on the insertion behavior of cerebral bioactive capability. IEEE Trans. Biomed. Eng. 48,361–371 (2001). microprobes In 2007 29th Annual International Conference of the IEEE Engi- 17. Boehler, C., Stieglitz, T. & Asplund, M. Nanostructured platinum grass enables neering in Medicine and Biology Society 4711–4714 (IEEE, 2007). superior impedance reduction for neural microelectrodes. Biomaterials 67, 35. Li, C. et al. Design of biodegradable, implantable devices towards clinical 346–353 (2015). translation. Nat. Rev. Mater. 5,61–81 (2020). 18. Minev,I.R.etal. Electronic dura mater for long-term multimodal neural 36. Galvan, M., Grafe, P.& TenBruggencate,G.Convulsant actions of interfaces. Science 347, 159–163 (2015). 4-aminopyridine on the guinea-pig olfactory cortex slice. Brain Res. 241,75–86 19. Goding, J. et al. A living electrode construct for incorporation of cells into (1982). bionic devices. MRS Commun. 7, 487 (2017). 37. Saunier, V., Flahaut, E., Blatché, C., Bergaud, C. & Maziz, A. Carbon nanofiber- 20. Hong, G. et al. Syringe injectable electronics: Precise targeted delivery with PEDOT composite films as novel microelectrode for neural interfaces and quantitative input/output connectivity. Nano Lett. 15, 6979–6984 (2015). biosensing. Biosens. Bioelectron. 165, 112413 (2020). 21. Xie, C. et al. Three-dimensional macroporous nanoelectronic networks as 38. Saunier, V., Flahaut, E., Blatché, M.-C., Bergaud, C. & Maziz, A. Microelectrodes minimally invasive brain probes. Nat. Mater. 14, 1286–1292 (2015). from PEDOT-carbon nanofiber composite for high performance neural 22. Felix, S. et al. Removable silicon insertion stiffeners for neural probes using recording, stimulation and neurochemical sensing. MethodsX 7, 101106 polyethylene glycol as a biodissolvable adhesive. In 2012 Annual International (2020). http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Microsystems & Nanoengineering Springer Journals

Scalable batch fabrication of ultrathin flexible neural probes using a bioresorbable silk layer

Download PDF
11 pages

Loading next page...
 
/lp/springer-journals/scalable-batch-fabrication-of-ultrathin-flexible-neural-probes-using-a-5NZCG0jadt

References (44)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2022
eISSN
2055-7434
DOI
10.1038/s41378-022-00353-7
Publisher site
See Article on Publisher Site

Abstract

Flexible intracerebral probes for neural recording and electrical stimulation have been the focus of many research works to achieve better compliance with the surrounding tissue while minimizing rejection. Strategies have been explored to find the best way to insert flexible probes into the brain while maintaining their flexibility once positioned. Here, we present a novel and versatile scalable batch fabrication approach to deliver ultrathin and flexible probes consisting of a silk-parylene bilayer. The biodegradable silk layer, whose degradation time is programmable, provides a temporary and programmable stiffener to allow the insertion of ultrathin parylene-based flexible devices. Our innovative and robust batch fabrication technology allows complete freedom over probe design in terms of materials, size, shape, and thickness. We demonstrate successful ex vivo insertion of the probe with acute high-fidelity recordings of epileptic seizures in field potentials as well as single-unit action potentials in mouse brain slices. Our novel technological solution for implanting ultraflexible devices in the brain while minimizing rejection risks shows high potential for use in both brain research and clinical therapies. Introduction stability at the interface between conventional electrodes 7,8 and brain tissue Chronically implanted microelectrodes have been a key . This is partly due to the mechanical tool in neuroscience research by allowing the recording of mismatch between the stiffness of the materials, e.g., electrical brain activity at the level of a small population of silicon, glass, platinum, or iridium (Young’s modulus E ≈ neurons (local field potential (LFP), multiunit spiking 150 GPa), constituting such probes and the softness of the activity) and of individual neurons (single-unit activity). cerebral tissue (E ≈ 10 kPa) . This mechanical mismatch, The past decades have seen impressive technological which can be as large as seven orders of magnitude, leads developments of neural implants incorporating electrodes to irreversible tissue damage and glial scar formation, at the micrometer scale, e.g., silicon-based pin (Utah), flat resulting in failure of the device within months or even 10,11 (Michigan), or wire (floating microwire arrays), for the weeks after implantation . 1–3 characterization of neuronal activity . Such devices are To improve the brain tissue-electrode interface, research now routinely used in animal studies . Although long- has focused on the use of flexible probes, which would lasting recording is sometimes achieved using these achieve better compliance with the surrounding neural 5 10,12 probes , large variations in electrical recording capabilities tissue and minimal rejection . The fabrication of these have often been reported . The implementation of long- compliant devices typically involves either the use of soft 13–15 lasting intracerebral recordings is limited by the lack of polymeric materials as substrates, e.g., parylene , 16,17 18 polyimide , polydimethylsiloxane (PDMS) , hydro- gels , and/or the use of significantly thinner stiff materi- 20,21 Correspondence: Ali Maziz (ali.maziz@laas.fr) als . However, an important issue with flexible probes is LAAS-CNRS, 7 Avenue du Colonel Roche, F-31400 Toulouse, France that they have a tendency to fail penetrating the brain CerCo, Université Toulouse 3, CNRS, Pavillon Baudot, CHU Purpan, BP 25202, meninges and reach their location goal. Indeed, a device 31052 Toulouse, France Full list of author information is available at the end of the article © The Author(s) 2022 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to theCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 2 of 11 that is too soft tends to bend when pressed against a rigid film, allowing the degradation of the stiffening layer surface, such as the pia mater . Strategies have been through proteolytic reactions within a programmable explored to find the best way to implant flexible probes time lapse. Furthermore, insertion trials in artificial brain into the brain while still maintaining their flexibility once phantoms are achieved without buckling of the probe or positioned. Some teams have focused on the use of a stiff undesired alteration of the electrical properties of the 20,22,23 shuttle that is removed immediately after implan- electrodes. We also demonstrate successful ex vivo insertion of the probe with high-fidelity acute recordings tation, while others have promoted the integration of a stiff bioresorbable coating that is not removed but dissolves of epileptic activity as well as single-unit action potentials 24,25 inside the brain on a time scale of minutes to days . in mouse brain slices. The possibilities offered by our Due to the tissue trauma caused by the implantation and approach are very promising for the development of withdrawal of a stiff shuttle, the integration of bioresorb- ultraflexible probes for application in research and brain able coatings as a temporary stiffener has been shown to therapy. better address both mechanical and biological failures . Various bioresorbable polymers, e.g., poly(ethylene glycol) Results and discussion (PEG), polylactic acid (PLA), chitosan, and silk fibroin, Development of the bilayered probes have been reported to be excellent candidates to add to Our ultrathin flexible probes are developed using stan- polymeric implants for facilitating insertion into the dard microsystem techniques. The probes are made of a brain . In addition, they benefit to some extent from silk-parylene bilayer that comprises four gold microelec- common attributes, such as a high Young’s modulus, trodes coated with the conducting polymer poly(3,4-ethy- proven biocompatibility for in vivo application purposes, lenedioxythio-phene):poly(styrene-sulfonate) (PEDOT:PSS) and resorption when in contact with biological tissues . to lower the impedance and obtain a better signal-to-noise 13,30 Although the reported flexible implants incorporating a ratio for neuron recording . The schematics of the fab- biodegradable coating have shown successful short-term rication procedures and the probe structure are shown in electrical recordings, their fabrication process is incom- Fig. 1a. Briefly, there are three main fabrication stages. In patible with standard microfabrication techniques. Man- stage I, a cellulose acetate (CA)-coated glass substrate is ual handling is required, which limits further downscaling coated with a silk fibroin layer before (stage II) an ultrathin (to less than 10 µm) and makes it difficult and time- parylene-based structure is microfabricated on top of the 12,28,29 consuming to create many devices in parallel . Fur- silk layer using standard top-down lithography techniques (discussed in more detail later in the paper). Finally, in stage thermore, existing flexible probes require the additional preparation of a carrier support for the biodegradable III, the whole silk-parylene probe undergoes microscale coating, with a manual assembly procedure that increases shaping by a reactive ion etching (RIE) process before being the difficulty of reducing the dimensions and volume of released from the glass substrate. the probe. These coated probes may generate surgical There are a couple of key considerations in this design. footprints of a volume considerably larger than the elec- First, the silk fibroin material is chosen among other trode itself, which may induce trauma to brain tissue in bioresorbable materials on the basis of its excellent bio- 24,25 the range of hundreds of micrometers to millimeters . compatibility, tunable biodegradability, and high Cellular and/or vascular damage may then elicit sustained mechanical strength . Indeed, silk fibroin possesses a inflammation and tissue responses. A global rethinking of Young’s modulus of E ≈ 3 GPa so that the thickness of the production framework is therefore needed to effec- parylene-C can be drastically reduced to the µm level tively achieve the simple integration of biodegradable while allowing for the handling and implantation of the coatings to allow further development of minimally probe. The biodegradable silk layer is described in detail invasive neural probes. below because of its specific chemical structure and its To achieve this goal, we report here a versatile fabri- importance in the final device. Active research is currently cation framework utilizing a bioresorbable silk fibroin underway on the use of silk biomaterials, particularly to layer that can be integrated in a microfabrication process obtain a biodegradable coating for implantable neural for preparing ultrathin parylene-based penetrating probes. recording devices . However, no work has yet been The probes consist of a silk-parylene bilayer, which is reported on the integration of degradable biopolymers in obtained by successively depositing layers on top of each the batch fabrication of neural recording devices, allowing other. The first layer is obtained using a degradable silk for further scaling down (less than 10 μm thick) and fibroin coating as a temporary stiffener that allows for the preventing de facto any manual handling procedure and insertion of ultrathin parylene-based flexible devices deep buckling issues during implantation. into the brain. The additional subtlety of the process The thickness of the silk layer can be tuned by adjusting derives from the exposure of the silk fibroin layer to the polymer dilution percentage or the volume of the methanol, which increases the crystallized domains in the solution on the casting area. As an example, a dose of Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 3 of 11 Au/Ti Cellulose acetate Parylene C Glass wafer Silk fibroin Ø 40 µm 250 µm d 5 mm e f 3h 6h 12h 24h 1 cm Au/Ti 4 µm 1 cm Silk fibroin 30 µm 0 250 µm Parylene C Days 0.8 0.6 0.4 30 µm 500 µm 0.2 0 20406080 % CF in O plasma 4 2 Fig. 1 Fabrication of the bilayered silk-parylene probes. a Schematic illustration of the main fabrication steps on a glass substrate: (1) cellulose acetate deposition, (2) silk fibroin deposition, (3) parylene base layer deposition, (4) gold microelectrode patterning, (5) parylene top layer deposition and (6) final shaping of the silk-parylene probe by RIE and release from the substrate. b Picture of the 4-inch glass substrate after batch microfabrication. The substrate contains 80 elements (less than 1 cm each). c Schematic illustration of the bilayered silk-parylene probe with characteristic dimensions. The device contains four recording microelectrodes with diameters of 40 μm patterned on a 3 mm-long and 250 μm-wide shank. d Picture of the microfabricated implant. The inset shows the corresponding magnified view of the 4 gold microelectrodes. e Picture of the silk-parylene probe highlighting the contact pads bonded with the FFC cable for the following electrochemical and electrophysiological measurements. f Enzymatic degradation of silk fibroin after different immersion times in methanol. g Etching rate of silk fibroin vs. percent of CF in O plasma at 500 W and 20 mTorr. An optimal ratio of 25/75 was found. h SEM image of the final bilayered silk-parylene shank, and i the corresponding magnified SEM image 0.1 mL/cm of 7 wt.% silk fibroin solution produces films fibroin samples degrade at different rates depending on with a thickness of ∼30 ± 5 μm (± SD). the methanol treatment time. For instance, we find that Silk degradation time is another critical parameter. films treated for 3 and 6 h degrade within a few hours Depending on the implantation strategy, the probe posi- while those treated for 12 and 24 h last up to one week tion may need to be adjusted for a prescribed time of (Fig. 1f). This control over the degradation time is con- several minutes or more, during which it must remain sistent with the literature, where reports have shown stiff. This critical time is determined by the degradation proteolytic degradation in vitro of water-stable silk films 25,32 rate of the bioresorbable coating. The lifetime of the after approximately two weeks . bioresorbable layer can be specifically adjusted via the The second step consists of fabricating an ultraflexible crystallinity of the silk protein, i.e., the β-sheet content . parylene-based probe on top of the silk fibroin substrate Indeed, treating silk by immersion in methanol increases (Fig. 1a) (more details are available in the Materials and the β-sheet content in the film. In a proteolytic medium Methods). Parylene is chosen as the substrate for its well- (protease XIV (PXIV) in PBS solution (1 U/ml)), silk documented biocompatibility, chemical inertness, and Flexible cable Parylene C (4µm) Silk Fibroin (30µm) -1 RIE rate (µm.min ) Silk Weight (%) Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 4 of 11 high electrical and moisture insulation properties. the manipulator approaching and (2) contacting the gel, Figure 1c shows the design of the ultraflexible parylene- (3) the probe successfully perforating the gel, (4) the based probe, which consists of three layers: a top parylene probe inserted in the gel, and (5 and 6) the probe deeply layer (thickness: 1 μm) for encapsulation, a middle Au inserted into the gel. The bilayered silk-parylene probes layer (thickness: 200 nm) for the electrophysiological thus succeed in penetrating the brain phantom without measurement, and a bottom parylene layer (thickness: any buckling or bending, which clearly indicates that the insertion shuttle is sufficiently stiff. From the insertion 3 μm) for mechanical support. The total thickness of the probe is only 4 μm. experiment (Fig. 2a), the first peak force (stage 3) is 0.7 The shaping of the silk-parylene probe is achieved by mN, corresponding to the minimum force required to dry etching, which is performed through RIE using a 75/ penetrate the tissue-mimicking gel. Other works show a 25,33 25 O /CF gas ratio. Under these conditions, the micro- similar minimum force . As Movie S1 (Supporting 2 4 fabrication of bilayered silk-parylene microstructures can Information) shows, the probe is explanted from the gel be performed with precise control of the geometry, size, without mechanical damage and successfully reinserted and shape, as illustrated in Fig. 1d, h and i. The proof of into the gel a second time. It is worth mentioning that the concept presented in Fig. 1b contains 80 elements, and bare parylene shank itself, being only 4 µm thick, curls each device contains four recording gold microelectrodes and cannot be handled or manipulated without the with diameters of 40 μm patterned on 3 mm-long and bioresorbable silk polymer support. 250 μm-wide probes. In other words, large-scale batch The type of brain phantom we used has been widely fabrication of precisely defined silk-parylene probes can used as a model of implantation in gray and white mat- 25,34 be performed directly on the glass substrate. Finally, the ter , yet it does not take into consideration other brain cellulose acetate sacrificial layer is dissolved in acetone to features. Indeed, the minimum insertion force also release the bilayered silk-parylene probes from the sub- depends on animal species, on the biological tissue (pia or strate (Fig. 1d). The probe is robust enough to easily bond dura mater, gray or white matter), on the size and shape of directly with external electrical connections (Fig. 1e). the shank, on the speed of approach, etc . For example, higher insertion forces are needed for dura mater pene- Insertion testing in brain phantom tration . Our fabrication technology allows complete Our bilayered probes are designed as insertable ultra- freedom in terms of materials, size, shape, and thickness. thin flexible devices that do not buckle during insertion in Thus, the probe buckling strength can be easily tuned the brain. An important parameter lies in the mechanical depending on the biological target. stability of the stiffening resorbable coating during the We next performed in vivo implantation in an anes- implantation/explantation process. To test the rigidity thetized mouse to evaluate the ability of the silk-stiffened afforded by the coating, we performed cyclic insertions of probe to achieve more difficult tasks of penetration, e.g., the probe into a brain tissue-mimicking phantom (1 wt.% through the pia mater. The dura mater is resected prior to agarose gel, Young’s modulus of 40 kPa). Figure 2b shows implantation. The probe is lowered with a manual the whole process of implanting the probe into the brain micromanipulator through the pia mater and is left in the tissue-mimicking gel, showing (1) the probe mounted on somatosensory cortex for one hour after insertion, such a b 1.5 2 mm 3 2 mm 2 mm Approaching Contacting Dimpling 4 5 2 mm 2 mm 6 2 mm 0.5 Perforating Inserting Deep insertion 0 0.5 1 1.5 2 Travel distance (mm) Fig. 2 Insertion test of the silk-parylene probe in brain phantoms. a Force profile during insertion of the shank in 1 wt.% agarose gel (Young’s modulus of 40 kPa), and the numbers correspond with those in (b). b Optical pictures showing the different stages of insertion in correlation with the evolution of the force Force (mN) Force (mN) Probe length (mm) Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 5 of 11 that the silk coating has enough time to dissolve. linked to its Young’s modulus E; the area moment of Figure S4 shows images of the silk-coated probe during inertia I (I = wt /12) along the x-axis; the length L, width x x insertion in the mouse brain. Insertion of the probe w, and thickness t of the probe; and the effective length occurs with no sign of buckling, which is consistent with factor of the column K. The cross-section of the device is what we observed in the in vitro insertion tests on the gel considered a constant rectangle, and the shanks are beams brain phantoms (Fig. 2). We do not notice any visible fixed at one side (K = 0.7). Assuming that the E values for 29,33 damage to the electrodes after insertion and retraction. parylene and silk fibroin are 3 GPa , the calculation These results indicate that our probe design is compatible yields a theoretical buckling force of 7.2 ± 2.3 mN for the with in vivo implantation. probe (Fig. 3c). The calculated theoretical buckling force predicts the trend and order of magnitude of the experi- Standard compression tests mental values (Fig. 3d). We further studied the mechanical properties of our In addition to axial compression tests, the mechanical bilayered probes against a hard substrate. Axial com- stability of our probes under long-term cyclic compres- pression of beams eventually results in their buckling. The sion was tested under cyclic loading with a bending radius highest force that a sample can withstand before bending of 0.3 mm. As shown in Fig. 3e, the original buckling force is called the buckling force F . As depicted in Fig. 3a, (first cycle) is 9.1 ± 0.5 mN and shows minimal change buckling compression tests on a hard substrate show the clear (only a 10% decrease) after nearly 1000 bending cycles. superiority of the bilayered silk-parylene assembly in After prolonged axial compression tests, the bilayered terms of its mechanical strength. The silk coating silk-parylene probes do not show any peeling or sign of improves the probe strength, with an average buckling delamination. This indicates that these engineered strength of 10.9 ± 1.3 mN (Fig. 3d), which is 15 times bilayered silk-parylene probes can endure a very large higher than the force required to penetrate the brain- number of contact cycles without any damage or mimicking gel (Fig. 2a). These experimental values were delamination. compared with theoretical analysis predictions (Fig. 3c). Our probes are modeled as single beams, whose buckling Electrical and electrochemical characterization 2 2 force is defined by Euler’s formula: F = π I E/(KL) . Possible alteration of the electrical and/or electro- buckling x The buckling force along the x-axis of a clamped beam is chemical properties of the probe should also be ab c Force (mN) 3 4 1 2 39.40 35.46 Buckling 35 31.52 2 mm 2 mm 27.58 23.64 2 25 19.70 2 mm 15.76 2 mm 0 0.2 0.4 0.6 11.82 Travel distance (mm) 7.880 de 10 10 3.940 8 0.000 2.2 40 2.4 2.6 2.8 3.2 20 3.4 2 3.6 2 3.8 510 15 20 0 200 400 600 800 1000 Buckling force (mN) Number of cycles Fig. 3 Compression tests of the bilayered silk-parylene probes against a hard substrate. a Force profile during compression, numbers and buckling position, which correspond with those in (b). b Series of optical pictures showing the corresponding stages of shank compression in correlation with the evolution of the force. c Theoretical influence of the shank length (in mm) and thickness (in µm) on the buckling force. The model was clamped-pinned, the cross-section of the device was considered a constant rectangle, and the shanks were beams fixed on one side and pinned on the other side (K = 0.7). We assumed that E for parylene C and silk fibroin was 3 GPa. d Electrode-buckling force histogram of the silk- parylene probes (N = 20). e Monitoring of the average buckling force as a function of the number of bending cycles (1000 cycles) with a bending radius of 0.3 mm Probe thickness (µ µm) Force (mN) Counts Buckling force (mN) Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 6 of 11 4 bc a 0 PEDOT:PSS PEDOT:PSS PEDOT:PSS -10 Au Au Au 3 -20 -30 -40 -50 -60 1 4 -70 -80 10 0 1 2 3 4 1 2 3 4 10 10 10 10 0 2 4 6 8 10 12 14 10 10 10 10 Frequency (Hz) Re(Z) (kOhm) Frequency (Hz) ef -4 15 2x10 250 35 PEDOT:PSS PEDOT:PSS Au Au -4 1x10 25 0 15 -4 -1x10 0 0 Z Z CSC CSC 02345 -0.5 0 0.5 Au PEDOT:PSS Au PEDOT:PSS Days Potential (V) Fig. 4 Electrochemical characterization of the microelectrodes. a EIS measurements of the gold- and PEDOT:PSS-coated microelectrodes in PBS at room temperature from 10 Hz to 7 MHz. b Corresponding phase vs. frequency plotting of gold- and PEDOT:PSS-coated microelectrodes. c Nyquist diagram at frequencies ranging from 10 Hz to 7 MHz of gold- and PEDOT:PSS-coated microelectrodes. d CV in the PBS buffer at room temperature by potential sweeping between −0.6 and +0.6 V at 200 mV/s vs. the Ag/AgCl reference electrode of gold- and PEDOT:PSS-coated microelectrodes. e Comparison of the electrochemical characteristics (Z at 1 kHz and CSC) between the gold- and PEDOT:PSS-coated microelectrodes. f Impedance evolution at 1 kHz of the PEDOT:PSS-coated microelectrode in PBS-soaked gel, obtained from the EIS measurements We next evaluated the charge transfer capabilities of the investigated to ensure their inherent electrode perfor- mance. We evaluated the electrical properties of the microelectrodes. CV from −0.6 and 0.6 V at a scan rate of microelectrodes by electrochemical impedance spectro- 0.2 V/s was performed, and the cathodal CSC was calcu- scopy (EIS), cyclic voltammetry (CV), and calculations of lated by the time integral of the cathodal currents within the charge storage capacity (CSC). Figure 4a, b shows the the cycled region (Fig. 4d). The cathodic CSC increases −2 Bode plots across the frequencies of interest from an average value of 1.95 ± 0.3 mC cm for the gold −2 (10 Hz–7 MHz) for a gold microelectrode with a diameter microelectrodes to 31.5 ± 1.3 mC cm for the PEDOT- of 40 µm before and after the electrochemical deposition coated microelectrodes. A higher charge capacity results of PEDOT:PSS. The mean impedances at 1 kHz are used in higher charge injection, which is desirable for electrical for comparison, as action potentials have a characteristic stimulation. frequency band centered at that frequency (Fig. 4e). In addition, to track the integrity of the probe structure Before PEDOT:PSS deposition, the average impedance is and electrical connections, in vitro EIS measurements 210 ± 8.2 kΩ (n = 5) in PBS, while the mean impedance were carried out daily for a 5-day period in saline brain decreases to 9.4 ± 0.9 kΩ after polymer deposition. This phantoms (gels soaked in PBS) with four different probes. well-known phenomenon is due to an increase in the Figure 4f shows the impedance measured at 1 kHz after effective surface area with the formation of PEDOT:PSS the insertion of the silk-parylene probe in PBS-based gel material, leading to a decrease in impedance of the as a function of immersion time. EIS measurements over microelectrode. The corresponding phase plot of the the 5-day period demonstrate a small decrease (~11 ± impedance reveals that the PEDOT:PSS microelectrode is 1.2–8 ± 0.5 kΩ) followed by a stabilization of the impe- capacitive in the low frequency range (10 Hz) and more dances. We observe no distortion of the probe upon resistive at higher frequencies (Fig. 4b). The Nyquist plot repeated insertion into the gel, which indicates that the recorded in PBS is presented in Fig. 4c. The deposition of thickness of the silk layer remains sufficient to carry the PEDOT:PSS produces a very small radius of the semicircle full ultrathin structure without compromising the elec- on the Nyquist plot with a charge transfer resistance of trical integrity of the parylene probe. The stable electrical approximately 7.8 kΩ, revealing the low electron-transfer impedance throughout the process proves the robustness resistance associated with the polymer coating. of our design and protocol for in vivo implantation. Current (mA) Impedance (KOhm) Impedance (KOhm) Phase (deg) Charge storage capacity -Im(Z) (KOhm) (mC/cm ) Impedance Z (KOhm) Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 7 of 11 In vitro proteolytic degradation depends on both implant-related factors (molecular To track the integrity of the structure, we performed weight, molecular structure, crystallinity, etc.) as well as incubation studies in PBS containing the enzyme protease on host-related factors (i.e., immune response at the XIV (Fig. 5). For this experiment, probes are crystallized implant site) . The silk biodegradability of the silk- for 6 h in methanol. Most naturally biodegradable poly- parylene probes is measured as the loss of weight of the mers degrade in vivo via enzymatic degradation, owing to implant during continuous incubation at 37 °C in pro- the stability of their backbone structure. It is generally teolytic solution for one week. In the proteolytic envir- accepted that silk protein is degraded mainly by pro- onment, the silk-parylene probes demonstrate a gradual tease . Protease XIV has been widely used to mimic decrease in mass, corresponding to slow protein frag- extracellular degradation mediated by proteolytic mentation during incubation. This experiment shows the enzymes. However, in vivo, the degradation of silk importance of using a biodegradable silk coating as a temporary stiffener to deliver ultrathin parylene-based flexible devices in deep tissue. It is worth mentioning that the bare parylene shank, which is only 4 µm thick, curls and cannot be handled or manipulated without the mechanical silk support. Electrophysiological recordings from mouse brain slices Electrophysiological experiments were conducted on mouse brain slices to assess the recording quality of the probes. The recordings were performed after the silk coating had completely dissolved to ensure that the 4 µm- thick flexible parylene probes were still functional for 1 2 34 5 acute recordings. The silk coating process was tuned so Days that the coating would completely dissolve in the slice within 30 min (corresponding to a 1 h long immersion of Fig. 5 In vitro proteolytic degradation of the silk layer. Enzymatic the probe in methanol for crystallization). Probe insertion degradation of the silk in the bilayered silk-parylene probes by protease XIV (PXIV) solution (1 U/ml) for 5 days at 37 °C was aimed at layer 2 of the piriform cortex (Fig. 6b). This 300–3000 Hz a c 0–100 Hz 50 µV 1 s b d e 10 µV 10 µV 2 s 0.5 ms Fig. 6 Electrophysiological recording in the piriform cortex of a mouse brain slice. a Photograph of a 35 μm-thick silk parylene probe inserted into a mouse brain slice: insertion occurs with no sign of buckling. b Scheme of the placement of the silk-parylene neural probe. c Traces show two consecutive epileptic seizures. The upper trace corresponds to the bandpass filtered signal (300–3000 Hz) used to reveal spiking activity, and the bottom trace corresponds to the low pass filtered signal (cutoff at 100 Hz) used to reveal slow changes in the LFP. d Traces show spontaneous spiking activity occurring between epileptic seizures. The dashed line indicates the threshold used for sorting action potentials. It is set at −4xRMS of the voltage trace (RMS calculated outside of the epileptic seizure). The action potentials that cross the threshold are examined through cluster analysis and interspike interval analysis. e Overlap of the action potentials recorded over 5 min at high temporal resolution. The constancy of the spike shape and refractory period (6.4 ms in that case) definitively ensures that the blue action potentials belong to one single unit. The action potentials in black correspond to multiunit activity Weight (%) Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 8 of 11 layer is characterized by a high density of neuronal cell mouse brain in vivo. Insertion in ex vivo mouse brain bodies. As spontaneous activity is low or null in most slices allowed acute recordings of LFP, as well as spon- regions of the mouse brain ex vivo, we added taneous multi and single-unit activities. These results 4-aminopyridine (4AP, 100 µM) to the superfusion solu- suggest that our design provides a versatile technological tion to activate the slices. 4AP infusion usually leads to solution for producing ultraflexible and ultrathin devices the appearance of spiking activity and most often induces that might be implanted in the brain with minimal rejection. In future works, we will directly address this epileptiform activity . The latter provides an opportunity to report on the capability of the probe to record LFPs in issue by performing in vivo neural recording and by addition to single- and multiunit spiking activity. Exam- correlating the quality of the recording with the presence ples of epileptic and spiking activities are presented in or lack of inflammatory reaction and glial scar formation. Fig. 6c–e. Epileptic activity is characterized by the occurrence of epileptic seizures in the LFP, which are Materials and methods evidenced by low pass filtering (0–100 Hz) of the voltage Chemicals trace (Fig. 6c, bottom trace). Bandpass filtering Parylene C dimer (PXC) was purchased from Comelec (300–3000 Hz) of the same original trace reveals the SA. ECI 3012 photoresist was purchased from Micro- action potentials of neurons (Fig. 6c, top trace). Large chemicals GmbH. MF CD26 developer was purchased bursts of action potentials can be observed at the time of from MicroChem. 3,4-Ethylene dioxythiophene (EDOT), epileptic seizures. Action potentials can also be seen poly(sodium 4-styrenesulfonate) (NaPSS), and acetone between epileptic seizures (Fig. 6c, d). We examined were purchased from Sigma Aldrich. Phosphate-buffered whether single-unit activity, i.e., action potentials that can saline (PBS, Gibco DPBS 1X) was purchased from Fisher be attributed to the activity of one single neuron, could be Scientific. Platinum (Pt) counter electrodes and silver/ observed in these periods. For this purpose, portions of silver chloride (Ag/AgCl) reference electrodes were pur- traces (2 ms wide, Fig. 6e) were extracted using a chased from WPI. Solutions were prepared with deionized threshold set at −4xRMS of the voltage trace (dashed water (18 MΩ). lines in Fig. 6d, e). These were then submitted to PCA and clustering (not illustrated). This analysis allows for seg- Bioresorbable silk fibroin solution preparation regation of different spike shapes. The spike shapes Silk fibroin aqueous solution was prepared from Bom- belonging to a given cluster are attributed to one single byx mori cocoons following a protocol detailed pre- unit if the interspike interval distribution histogram (not viously . The fibroin protein was first extracted from the illustrated) shows a clear refractory period, i.e., no interval silk fibers by boiling the silk cocoons (5 g) in a solution of <1 ms. Otherwise, the cluster corresponds to multiunit 0.02 M Na CO for 30 min. The regenerated silk fibroin 2 3 activity, that is, action potentials that are issued from at was then recovered and rinsed thoroughly in deionized least two neurons and that cannot be segregated. The blue water before being dried overnight under ambient con- action potentials in Fig. 6e correspond to the activity of ditions. The dried silk fibroin (3.6 g) was dissolved in a one single unit, identified as such by its constant spike 9.3 M LiBr solution at 60 °C for 4 h. The salt was then shape and a refractory period of 6.4 ms. The action removed by dialyzing the solution against deionized water potentials in black correspond to multiunit activity. for 24 h at room temperature using a dialysis membrane Overall, multiunit activity is recorded in all trials, and (MWCO 3.5 KD, Spectra/PorTM) and regularly changing single units are recorded in 7/12 trials. the water. Centrifugation was performed to remove impurities. The recovered silk fibroin solution had a final Conclusion concentration of 7 wt.%. This work proposes a novel and versatile approach to fabricate, pattern, and deliver ultrathin probes consisting Probe microfabrication of a silk-parylene bilayer. The biodegradable silk layer A 4-inch glass wafer was used to prepare the overall provides a temporary stiffener that can be used to deliver process. The glass substrate was first cleaned by MW- ultrathin parylene-based flexible devices in deep tissue. oxygen plasma (800 W, 10 min, O ) prior to processing. Our innovative and robust batch fabrication technology The fabrication began with the deposition of a cellulose allows complete design freedom of the probe in terms of acetate layer (~2 μm) by spin-coating at 1000 rpm for 30 s materials, size, shape, and thickness. We systematically (5% w/vol in acetone). It acted as a sacrificial layer to studied the behavior of the bilayered structure in gel brain release the final device from the substrate. The silk fibroin phantoms and demonstrated that parylene probes as thin aqueous solution (7 wt.%) was deposited by drop casting as 4 μm could be delivered accurately to a desired depth and left to dry under ambient conditions overnight, with intact geometry and electrical functionality. We also resulting in an ~30 µm-thick silk film. The thickness of demonstrated successful insertion of the probe in the the resulting film was controlled by adjusting the volume Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 9 of 11 of the silk fibroin solution. Then, a 3 μm-thick layer of the test, the implants were held between two glass plates, Parylene C (PXC) was deposited onto the silk-coated and compression was carried out over a tip length of substrate through CVD using C30S Comelec equipment. approximately 3.4 mm. The silk-parylene probes were A 50/200 nm layer of Ti/Au was then deposited by eva- fixed on a MARK-10 ESM303 test bench coupled with a poration and patterned through an electroplated nickel- MARK-10 M5-012 force sensor. An internally developed based shadow mask. Another 1.3 μm top layer of Parylene LabVIEW program allowed the equipment to be con- trolled and the force to be monitored according to the C was deposited onto the processed metal layer. The shape of the electrode pads, contacts and device body displacement of the implant. Compression tests were −1 were defined by photolithography steps followed by RIE in performed at a slow speed of 2 mm min for optimal O /CF plasma (75/25) at 500 W and 20 mT. Finally, the monitoring of the buckling force. A video recording 2 4 cellulose acetate sacrificial layer was dissolved with acet- (Movie S1) of the compression test using a Dino-Lite Edge one to release the bilayered silk-parylene probes. camera was made to complete the experiment. Electrochemical characterization In vitro insertion into the gel brain phantoms Electrochemical characterization was performed with a Insertion tests of the probes were performed using 1% 3-electrode system that included a Pt wire as the counter w/v agarose gel brain phantoms imitating the mechanical electrode, a Ag/AgCl wire as the reference electrode and the properties of brain tissue. In the same way as in the gold microelectrodes from the silk-parylene probe as the compression tests, the silk-parylene probes were held working electrodes. EIS and cyclic voltammetry (CV) were between two glass plates and fixed to the MARK-10 test performed using a Bio-Logic VMP3 potentiostat. CV was bench. To limit damage during the insertion tests and to performed in PBS at room temperature by potential have optimal monitoring of the forces involved, the sweeping between −0.6 and +0.6V at 200mV/s vs. Ag/ experiment was carried out at a very low speed of 0.5 mm −1 AgCl reference, allowing cathodal charge storage capacity min . A video recording was made using the Dino-Lite 37,38 (CSCc) evaluation . EIS was also performed in PBS at Edge camera to observe the different stages of insertion in room temperature by applying a 10 mV sine wave at fre- correlation with the evolution of the force. quencies ranging from 10 Hz to 7 MHz. Improved electrical properties were achieved by PEDOT:PSS deposition. CV Dissolution tests was performed in EDOT:NaPSS solution (10 mM:34 mM) To assess the biodegradation of the silk fibroin layer, an at room temperature by potential sweeping between −0.7 enzymatic degradation test was carried out over several and +1 V at 10 mV/s vs. Ag/AgCl reference. CV and EIS days. Briefly, five implants (~2 mg) were selected and were then performed again to compare the evolution of the incubated in 1 ml of protease (Proteas XIV from S. griseus, CSCc, impedance, phase, and Nyquist results. 3.5 U/mg, Sigma-Aldrich) and PBS (1 U/ml of PBS buffer) at 37 °C. The implants were photographed and weighed Crystallization procedure of the silk fibroin every day after cleaning with DI water and dried at 60 °C Water-insoluble silk films were prepared by methanol for 10 min. The enzyme solution was changed after each treatment, which increased the crystallized domains in the weighing to maintain enzymatic activity. film, allowing degradation within a programmable time window. The bilayered silk-parylene probes were posi- Brain slice preparation tioned between two flexible filter papers and held in All procedures were conducted in accordance with the position with clips to prevent any deformation of the guidelines from the European Community (directive devices during annealing. The probes were then immersed 2010/63/UE) and from the French Ministry of Agri- in 80% methanol solution at room temperature for a fixed culture, Agri-food and Forestry (décret 2013–118) and time to increase the β-sheet crystal content. After were approved by the Ministère de l’Enseignement methanol annealing, the filter paper was removed, and the Supérieur, de la Recherche et de l’Innovation (N° 15226- implants were dried under vacuum at 40 °C for 1 h to 2018052417151228). Two- to 4-month-old C57BL/6 wild- remove all traces of methanol. type female mice were used for brain slice preparation. The protocol has been detailed previously and is briefly Standard compression tests summarized here. Mice were decapitated after deep Standard compression tests against a hard substrate anesthetization with isoflurane. The brain was removed were performed to assess the mechanical properties of the and prepared for slicing in ice-cold modified ACSF bilayered silk-parylene probes. The axial compression of (mACSF). The composition of the mACSF was (in mM): the implant allowed us to evaluate its buckling force, NaCl 124, NaHCO 26, KCl 3.2, MgSO 1, NaH PO 0.5, 3 4 2 4 which corresponds to the highest force that a sample can MgCl 9, Glucose 10. The mACSF was bubbled with a gas withstand before bending. To guarantee the stability of mixture of 95% O and 5% CO . Then, 400 µm-thick 2 2 Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 10 of 11 slices were cut with a vibratome in the presence of cold, for the implantation tests in vivo. For anesthesia, subjects oxygenated mACSF. The slices were allowed to recover received a chemical induction of ketamine/xylazine (125/ for at least one hour at room temperature in a holding 5 mg/kg) by intraperitoneal injection followed by gas induc- chamber filled with oxygenated, in vivo-like ACSF, whose tion with 3% vaporized isoflurane (TEM SEGA’s MiniHUB composition was (in mM): NaCl 124, NaHCO 26, KCl V3.Pessac).Oncethe mousewas placed on thesurgical 3.2, MgSO 1, NaH PO 0.5, CaCl 1.1, and glucose 10. platform, anesthesia was maintained with isoflurane (1.5% v/v 4 2 4 2 in O This ACSF was continuously bubbled with a 95% O /5% ). The level of anesthesia was checked by the lack of the 2 2 CO gas mixture (pH 7.4). withdrawal reflex after pinching the hindlimbs. A heating pad was positioned underneath the animal to keep the body Ex vivo electrophysiological recordings of the brain slices temperature at 37 °C. Eye dehydration was prevented by For the ex vivo recordings, a slice was transferred to a topical application of ophthalmic gel. An incision was made submersion-type chamber that was continuously gravity- through all skin layers at the top of the skull, the skin flaps fed with oxygenated (95% O /5% CO ) in vivo-like ACSF were retracted, and the tissues detached from the bone. The 2 2 at a flow rate of 3–3.5 ml/min. All recordings were per- skull was cleaned and dried with sterile cotton swabs, and a formed at 34–35 °C. For recording, two probes were used, hole was drilled using a Harvard Apparatus MicroDrill and 2 electrodes were tested for each probe. After deep (0.5 mm diameter: Les Ulis, France), allowing the dura mater insertion, one of the two probes was slowly pulled up in to be carefully removed to expose the brain. The silk-parylene steps of 50–100 µm, yielding five pairs of recordings. probes were inserted in the somatosensory (SS) cortex or Electrophysiological signals were amplified (×1000) and visual (V1) cortex (SS stereotaxic coordinates: AP:−0.9, ML: filtered (0.1 Hz–10 kHz) with a NeuroLog system (Digi- −3.0; right V1 stereotactic coordinates: AP,−3.5 mm; ML, timer, UK). Notably, 50 Hz noise was eliminated with a 2.4 mm; at a depth of 2 mm from the brain surface). All Humbug system (Quest Scientific, Canada). Low-pass procedures above complied with the guidelines of State Sci- filtering (0–100 Hz) and bandpass filtering (300–3000 Hz) entific and Technological Commission Animal procedures were used offline to isolate LFP and spiking activity, and were approved by the national Animal Care and Ethics respectively. All signals were digitized with a digitization Committee (CE2A122 protocol number 2020051519454596) rate of 20 kHz (power1401, CED, UK). The real-time following European Directive 2010/63/EU. display of the signals was achieved with an oscilloscope Acknowledgements and Spike2 software (CED, UK). Spike sorting was con- We thank Dr. Denis Calise (Institut des Maladies Métaboliques et ducted using Spike2. First, we computed the root mean Cardiovasculaires, Toulouse, France) for his help regarding animal facilities. The authors acknowledge funding from the Agence Nationale de la Recherche square (RMS) of the voltage trace (running average of (ANR-19-CE19-0002-01 and ANR ANR-15-CE19-0006). This work was supported over 5 s). Next, 2 ms wide segments of voltage traces by the French RENATECH network. (“wavemarks”) were generated whenever negative deflec- tions, likely corresponding to action potentials, exceeded a Author details 1 2 LAAS-CNRS, 7 Avenue du Colonel Roche, F-31400 Toulouse, France. CerCo, threshold set at −4 times the RMS. The wavemarks Université Toulouse 3, CNRS, Pavillon Baudot, CHU Purpan, BP 25202, 31052 extended from −0.5 to +1.5 ms relative to the peak of the 3 Toulouse, France. UMR Institut National de la Santé et de la Recherche negative deflection. The minimal wavemark overlap was Médicale 1048, Institut des Maladies Métaboliques et Cardiovasculaires, Toulouse, France set at 0.5 ms. The wavemarks were then submitted to principal component analysis (PCA) and cluster analysis. Conflict of interest Single units corresponded to clusters containing spikes of The authors declare no conflict of interest. constant shape and displaying a refractory period of >1 ms Supplementary information The online version contains supplementary were determined from interspike interval distribution material available at https://doi.org/10.1038/s41378-022-00353-7. histograms computed with a bin width of 0.2 ms. Received: 19 February 2021 Revised: 26 November 2021 Accepted: 19 Packaging December 2021 Prior to recording, the flexible device was bonded to a customized flexible ribbon cable with golden traces (AXO-00021, pro-POWER, China) by using epoxy silver References and photosensitive glue. The overall connection system 1. Jog, M. et al. Tetrode technology: Advances in implantable hardware, neu- was then passivated with insulating biocompatible glue roimaging, and data analysis techniques. J. Neurosci. Methods 117,141–152 (Polytec EP 601). (2002). 2. Kipke, D. R., Vetter, R. J., Williams, J. C. & Hetke, J. F. Silicon-substrate intracortical microelectrode arrays for long-term recording of neuronal spike activity in In vivo implantation into a mouse brain cerebral cortex. IEEE Trans. Neural Syst. Rehabilitation Eng. 11,151–155 (2003). Male C57BL/6J mice aged 15–17 weeks, purchased from a 3. Normann, R. A.,Maynard,E.M., Rousche, P. J. & Warren, D. J. A neural interface for a cortical vision prosthesis. Vis. Res. 39,2577–2587 (1999). commercial supplier (Charles River, Eculy, France), were used Cointe et al. Microsystems & Nanoengineering (2022) 8:21 Page 11 of 11 4. Nicolelis, M. A. et al. Chronic, multisite, multielectrode recordings in macaque Conference of the IEEE Engineering in Medicine and Biology Society 871–874 monkeys. Proc. Natl Acad. Sci. USA 100, 11041–11046 (2003). (IEEE, 2012). 5. Okun, M., Lak, A., Carandini, M. & Harris, K. D. Long term recordings with 23. Williamson, A. et al. Localized neuron stimulation with organic electrochemical immobile silicon probes in the mouse cortex. PLoS One 11, e0151180 (2016). transistors on delaminating depth probes. Adv. Mater. 27, 4405–4410 (2015). 6. Rousche, P. J. & Normann, R. A. Chronic recording capability of the Utah 24. Pas, J. et al. A bilayered PVA/PLGA-bioresorbable shuttle to improve the Intracortical Electrode Array in cat sensory cortex. J. Neurosci. Methods 82,1–15 implantation of flexible neural probes. J. Neural Eng. 15, 065001 (2018). (1998). 25. Lecomte, A. et al. Silk and PEG as means to stiffen a parylene probe for 7. Lacour, S. P., Courtine, G. & Guck, J. Materials and technologies for soft insertion in the brain: toward a double time-scale tool for local drug delivery. J. implantable neuroprostheses. Nat. Rev. Mater. 1,1–14 (2016). Micromech. Microeng. 25, 125003 (2015). 8. Chen, R., Canales, A. & Anikeeva, P. Neural recording and modulation tech- 26. Lewitus,D., Smith, K. L.,Shain,W.&Kohn,J.Ultrafast resorbingpolymersfor nologies. Nat. Rev. Mater. 2,1–16 (2017). use as carriers for cortical neural probes. Acta Biomater. 7,2483–2491 (2011). 9. Muthuswamy, J., Saha, R. & Gilletti, A. Tissue micromotion induced stress 27. Feig, V. R.,Tran, H. &Bao,Z.Biodegradable polymeric materials in degradable around brain implants. In 3rd IEEE/EMBS Special Topic Conference on Micro- electronic devices. ACS Cent. Sci. 4,337–348 (2018). technology in Medicine and Biology, 2005 102–103 (IEEE, 2005). 28. Felix, S. H. et al. Insertion of flexible neural probes using rigid stiffeners 10. Jeong, J.-W. et al. Soft materials in neuroengineering for hard problems in attached with biodissolvable adhesive. J. Vis. Exp.: JoVE https://doi.org/10.3791/ neuroscience. Neuron 86,175–186 (2015). 50609 (2013). 11. Polikov,V.S., Tresco,P.A. & Reichert,W.M. Responseofbrain tissue to 29. Weltman, A., Yoo, J. & Meng, E. Flexible, penetrating brain probes enabled by chronically implanted neural electrodes. J. Neurosci. Methods 148,1–18 (2005). advances in polymer microfabrication. Micromachines 7, 180 (2016). 12. Lecomte, A., Descamps, E. & Bergaud, C. A review on mechanical considera- 30. Maziz,A., Özgür, E., Bergaud,C.& Uzun,L. Progress in conductingpolymersfor tions for chronically-implanted neural probes. J. Neural Eng. 15, 031001 (2017). biointerfacing and biorecognition applications. Sensors Actuators Rep. 3, 13. Castagnola, V. et al. Parylene-based flexible neural probes with PEDOT coated 100035 (2021). surface for brain stimulation and recording. Biosens. Bioelectron. 67,450–457 31. Maziz, A., Leprette,O., Boyer, L.,Blatchè,C.&Bergaud, C. Tuning theproperties (2015). of silk fibroin biomaterial via chemical cross-linking. Biomed. Phys. Eng. Express 14. Rodger, D. C. et al. Flexible parylene-based multielectrode array technology for 4, 065012 (2018). high-density neural stimulation and recording. Sens. Actuators B: Chem. 132, 32. Hu, X. et al. Regulation of silk material structure by temperature-controlled 449–460 (2008). water vapor annealing. Biomacromolecules 12,1686–1696 (2011). 15. Khodagholy, D. et al. Highly conformable conducting polymer electrodes for 33. Seo,K. J.etal. Transparent, flexible, penetrating microelectrode arrays with in vivo recordings. Adv. Mater. 23,H268–H272 (2011). capabilities of single‐unit electrophysiology. Adv. Biosyst. 3, 1800276 (2019). 16. Rousche, P. J. et al. Flexible polyimide-based intracortical electrode arrays with 34. Hosseini, N. H. et al. Comparative study on the insertion behavior of cerebral bioactive capability. IEEE Trans. Biomed. Eng. 48,361–371 (2001). microprobes In 2007 29th Annual International Conference of the IEEE Engi- 17. Boehler, C., Stieglitz, T. & Asplund, M. Nanostructured platinum grass enables neering in Medicine and Biology Society 4711–4714 (IEEE, 2007). superior impedance reduction for neural microelectrodes. Biomaterials 67, 35. Li, C. et al. Design of biodegradable, implantable devices towards clinical 346–353 (2015). translation. Nat. Rev. Mater. 5,61–81 (2020). 18. Minev,I.R.etal. Electronic dura mater for long-term multimodal neural 36. Galvan, M., Grafe, P.& TenBruggencate,G.Convulsant actions of interfaces. Science 347, 159–163 (2015). 4-aminopyridine on the guinea-pig olfactory cortex slice. Brain Res. 241,75–86 19. Goding, J. et al. A living electrode construct for incorporation of cells into (1982). bionic devices. MRS Commun. 7, 487 (2017). 37. Saunier, V., Flahaut, E., Blatché, C., Bergaud, C. & Maziz, A. Carbon nanofiber- 20. Hong, G. et al. Syringe injectable electronics: Precise targeted delivery with PEDOT composite films as novel microelectrode for neural interfaces and quantitative input/output connectivity. Nano Lett. 15, 6979–6984 (2015). biosensing. Biosens. Bioelectron. 165, 112413 (2020). 21. Xie, C. et al. Three-dimensional macroporous nanoelectronic networks as 38. Saunier, V., Flahaut, E., Blatché, M.-C., Bergaud, C. & Maziz, A. Microelectrodes minimally invasive brain probes. Nat. Mater. 14, 1286–1292 (2015). from PEDOT-carbon nanofiber composite for high performance neural 22. Felix, S. et al. Removable silicon insertion stiffeners for neural probes using recording, stimulation and neurochemical sensing. MethodsX 7, 101106 polyethylene glycol as a biodissolvable adhesive. In 2012 Annual International (2020).

Journal

Microsystems & Nanoengineering – Springer Journals

Published: Feb 16, 2022

There are no references for this article.