ORIGINAL RESEARCH published: 04 March 2021 doi: 10.3389/fnins.2021.615256 Novel Graphene Electrode for Retinal Implants: An in vivo Biocompatibility Study Diep Nguyen 1* , Manon Valet 1 , Julie Dégardin 1 , Leyna Boucherit 1 , Xavi Illa 2,3 , Jose de la Cruz 4 , Elena del Corro 4 , Jessica Bousquet 4 , Jose A. Garrido 4,5 , Clément Hébert 4 and Serge Picaud 1* 1 INSERM, CNRS, Institut de la Vision, Sorbonne Université, Paris, France, 2 Instituto de Microelectrónica de Barcelona, IMB-CNM (CSIC), Bellaterra, Spain, 3 Centro de Investigación Biomédica en Red en Bioingeniería, Biomateriales y Nanomedicina, Madrid, Spain, 4 Catalan Institute of Nanoscience and Nanotechnology, Barcelona, Spain, 5 Catalan Institution for Research and Advanced Studies, Barcelona, Spain Evaluating biocompatibility is a core essential step to introducing a new material as a candidate for brain-machine interfaces. Foreign body reactions often result in glial scars that can impede the performance of the interface. Having a high conductivity and large electrochemical window, graphene is a candidate material for electrical stimulation Edited by: with retinal prosthesis. In this study, non-functional devices consisting of chemical vapor Winfried Mayr, Medical University of Vienna, Austria deposition (CVD) graphene embedded onto polyimide/SU-8 substrates were fabricated Reviewed by: for a biocompatibility study. The devices were implanted beneath the retina of blind Bernhard Baumann, P23H rats. Implants were monitored by optical coherence tomography (OCT) and Medical University of Vienna, Austria Aaron J. Suminski, eye fundus which indicated a high stability in vivo up to 3 months before histology University of Wisconsin–Madison, studies were done. Microglial reconstruction through confocal imaging illustrates that United States the presence of graphene on polyimide reduced the number of microglial cells in the *Correspondence: retina compared to polyimide alone, thereby indicating a high biocompatibility. This study Serge Picaud serge.picaud@inserm.fr highlights an interesting approach to assess material biocompatibility in a tissue model Diep Nguyen of central nervous system, the retina, which is easily accessed optically and surgically. diep.nguyen@inserm.fr Keywords: graphene, prosthesis, biocompatibility, retina, implant Specialty section: This article was submitted to Neuroprosthetics, INTRODUCTION a section of the journal Frontiers in Neuroscience Brain machine interfaces are emerging technologies to restore perception and action following Received: 08 October 2020 different degenerative diseases and traumatic incidents (Slutzky, 2019). These interfaces can be Accepted: 02 February 2021 used directly on the brain or in more peripheral locations of the central nervous systems. In this Published: 04 March 2021 context, retinal prostheses have offered solutions to restore some useful vision in blind patients (da Citation: Cruz et al., 2016; Stingl et al., 2017; Palanker et al., 2020). The targeted vision impairment is due Nguyen D, Valet M, Dégardin J, to retinitis pigmentosa and age-related macular degeneration that leads to the progressive loss of Boucherit L, Illa X, de la Cruz J, photoreceptors in the retina. Though there exists no cure, rehabilitation therapy through electrical del Corro E, Bousquet J, Garrido JA, stimulation at the retina has enabled late-stage patients to experience visual sensation including the Hébert C and Picaud S (2021) Novel Graphene Electrode for Retinal ability to read albeit with a low visual acuity (Hornig et al., 2017; Hahn and Fine, 2018). The Argus II Implants: An in vivo Biocompatibility (Second Sight) (da Cruz et al., 2016) and the Alpha-AMS (Retina Implant AG) (Edwards et al., 2018) Study. Front. Neurosci. 15:615256. devices have received approval for market sale to patients with end stage retinitis pigmentosa by doi: 10.3389/fnins.2021.615256 either the US FDA and European CE or only European CE, respectively. More recently, the Prima Frontiers in Neuroscience | www.frontiersin.org 1 March 2021 | Volume 15 | Article 615256 Nguyen et al. Novel Graphene for Retinal Implants device was directly tested in patients with age-related macular graphene derivatives to many cell types such as cultured human degeneration due to the expected high visual acuity (Lorach et al., and mouse fibroblast and various in vivo organs like rodent liver 2015; Palanker et al., 2020; Prévot et al., 2020). In all implant and lung showing high cell viability and survival rates despite designs, the devices contain an array of electrodes either in the some visual changes in histological results (Pinto et al., 2013). subretinal space (Alpha IMS, Prima) (Stingl et al., 2017; Palanker Because tissue reactions can be very cell specific, there are few et al., 2020) or in the epiretinal position in the vitreous body (da studies evaluating the biocompatibility of graphene in the central Cruz et al., 2016). Electrical stimulation of the retina generates nervous system. It has been previously shown that single layer phosphenes (visual sensation) to patients who learn over time to graphene is capable of promoting an improved neurite growth associate them to form basic vision. However, current devices are and adhesion for retinal ganglion neurons compared to glass unable to provide facial recognition, independent locomotion, (Bendali et al., 2013) and that primary hippocampal neurons has a and complex text reading. They require improvements to provide significantly better adherence to graphene compared to insulating restored vision of higher resolution. polymers as well as CVD diamond in the absence of poly-L- For retinal prostheses, as for most brain machine interfaces, lysine (Bourrier et al., 2019). However, no in vivo study has been conductive electrode materials with a high double layer completed to show a chronic biocompatibility of graphene in the capacitance are expected to offer improved spatial resolution ocular environment. (Joucla and Yvert, 2009). Carbon related materials and specifically In this study, the retina was used to investigate an graphene and graphene derivatives are currently being studied in vivo biocompatibility of graphene due its highly structured for their electrical properties as biosensors (Pumera et al., organization in layers, which enables the identification of toxic 2010) and neural interfacing (Kostarelos et al., 2017). As changes. The biocompatibility of graphene was examined in neural interfacing materials, graphene and their derivatives the ocular environment of the P23H rat, a model of retinitis make up a class of electrode material that have advantageous pigmentosa. Soft and flexible subretinal implants with large properties for recording and stimulation. Graphene is a single- CVD graphene surfaces were designed and adapted to test layered, transparent but visible, 2-D material that has a large biocompatibility in rats. An in vivo monitoring of the implant electrochemical water window and high electron mobility (Ji was taken to assess the retinal state and the implant stability in et al., 2014). Graphene is mainly grown at high temperatures on the eye. Finally, a post mortem histological study was performed metallic substrates such as copper in chemical vapor deposition to quantify microglial cells of the retina, which are an excellent (CVD) reactors, and can be wet transferred onto flexible biomarker to monitor the reactive state (Polikov et al., 2005), polymer substrates (Lee et al., 2010). The high transparency and thus the biocompatibility of graphene compared to a control of CVD graphene has already been exploited in acute cortical implant made of soft insulating polymers. recordings simultaneously with calcium imaging (Kuzum et al., 2014) as well as measuring electroretinograms in non-human primates (Yin et al., 2018). Fabricating a 90% transparent MATERIALS AND METHODS device with graphene electrodes has demonstrated its capability to record brain signals during light dependent applications Microfabrication of Device such as florescence microscopy and optogenetics (Park et al., Implants were designed to maximize the exposed area of 2014). For neural stimulation, flat CVD graphene is often graphene on the circular head to the tissue with the underlying considered a poor material lacking the desired charge capacity metal contacts to validate the presence of graphene by a resistance of metals. However, some have demonstrated its stimulation measurement (Figure 1). They were fabricated using a 4-inch capabilities to active cortical neurons (Park et al., 2018). In thermally oxidized Si wafer as a carrier support. A 10-µm addition, 3D nanostructures of graphene using porous templates thick polyimide layer (PI-2611, HD MicroSystems) was spin- (Fan et al., 2013; Li et al., 2013), graphene foam (Dong coated and cured twice to obtain a substrate with a 20-µm et al., 2012), and graphene mesh (Jiang and Fan, 2014) have thickness. A lift-off process using image reversal photoresist demonstrated to maintain graphene material properties with AZ5214 (Clariant GmbH, Germany) was used to define the metal increased electrochemical surface area for an improved charge tracks (Ti/Pt, 10/100 nm) that were deposited using an electron- storage capacity that could be beneficial for neural stimulation. beam vapor deposition system. Single layer graphene was grown Additionally, reduced graphene oxide generally made by redox by CVD and transferred onto the polyimide substrate using a wet treatment of graphite have already been fabricated to include chemical method as previously described (Blaschke et al., 2016; porous structures intended for neural stimulation (Apollo et al., Hébert et al., 2017). Large graphene areas (1 mm in diameter) 2015; Lu et al., 2016). on top of the metal tracks were protected using a positive However, despite having ideal electrical properties, the safety photoresist (HiPR 6512, FujiFilm) that was used as a mask for and biocompatibility of new materials must be evaluated before etching the rest of the graphene layer by means of an oxygen- determining their relevance in stimulation applications. A host based reactive ion etching (RIE). After removing the photoresist, rejection to foreign material over time can render a device useless, the negative photoresist SU-8 (SU-8 2005, MicroChemCorp., causing complications and requiring additional surgery. For United States) was spin-coated and defined to passivate the metal instances, gold electrodes were observed to have degraded within leads while defining the uncovered graphene area. To define the a few months below the retina (Chow et al., 2001). Various studies final geometry of the implants, a final photolithographic process have provided evidence for the biocompatibility of graphene and using a thick positive photoresist (AZ9260, Clariant) was used Frontiers in Neuroscience | www.frontiersin.org 2 March 2021 | Volume 15 | Article 615256 Nguyen et al. Novel Graphene for Retinal Implants FIGURE 1 | Graphene-based implants. Orthogonal schemes of fabricated implants showing the side and top views. Graphene and platinum lines are passivated between polyimide and SU-8. Control implants have the same scheme without the graphene layer, exposing instead the bottom polyimide layer. to define the mask for structuring the 20-µm thick polyimide injected in the sclerotomy to generate a retinal detachment. layer in a deep-RIE process. The wafer was then cleaned in The implant was then inserted below the detached retina in isopropanol and rinsed with water before being peeled off from the subretinal space targeting an adjacent location to the optic the silicon wafer. disk. Observation under the surgical microscope confirmed the In parallel, a second wafer was processed to fabricate the placement in the subretinal position using a plastic coverslip control devices. The process was identical as the one described gently pressed on ophthalmic gel that was applied to the cornea. above but without the implementation of the graphene layer. Subcutaneous injection of atipamezole (antisedan) was then used to reanimate the animal in a chamber maintained 37◦ C. Subretinal Implantation Twelve homozygous P23H rats were implanted with the In vivo Monitoring fabricated implant devices: N = 7 with graphene and N = 5 In vivo monitoring of implant state consisted of eye fundus and without graphene as a sham. Seven non-implanted eyes were optical coherence tomography (OCT) imaging done at 7, 15, 30, used for a retina-only control. Animals were enclosed in a 60, and 90 days post implantation. Eye fundus images were taken controlled environment with a half-day dark/light cycle with on a Micron III and Micron IV machines (Pheonix Technology nutrition ad libitum. All animal experiments were completed Group, Pleasanton, CA, United States) and OCT imaging was in accordance to the Charles Darwin No5 Ethics Committee in done on a Bioptigen OCT system (Leica microsystems). Animal Animal Experimentation (agreement #15219) in the approved anesthesia, eye dilation, and reanimation were performed in animal facilities associated to the Institut de la Vision (Sorbonne the same manner as explained in the implantation section. Université). All experimental work were performed according to Immediately after the 90-day time point, the animals were the institutional policies on biosecurity and safety procedures. All euthanized by CO2 induction followed by cervical dislocation. animals were unilaterally implanted at 9-months of age. The short surgery consisted of placing the device in the subretinal space in Immunohistochemistry the central region next to the optic nerve as previously described Immediately after euthanasia, the implanted eye was removed (Salzmann et al., 2006). Briefly, anesthesia was provided by an and dissected. The cornea and lens were removed in phosphate intraperitoneal injection of a ketamine-medetomidine mixture buffer saline (PBS) solution (0.1 M, pH 7.4). A 3-mm biopsy (40 mg kg−1 ketamine, 0.14 mg kg−1 medetomidine) following punch in the eye cup was used to extract the implant between the a 5% gaseous induction of isoflurane. An additional ocular retina and the sclera and choroid. This eye fragment was fixed by anesthesia using oxybuprocaïn chlorohydrate eye drops was incubation overnight in paraformaldehyde solution (4% in PBS) administered. Eye dilatation was obtained by application of at 4◦ C and then washed using PBS. tropicamide (0.5%) solution. A heating platform maintained For immunolabelling, eye fragments were incubated in a body temperature at 37◦ C. A small sclerotomy was performed blocking solution (10% bovine serum albumin (Sigma, France), on the dorsal sclera tangential to the cornea. A gel of sodium 2% Triton X-100 (Sigma), 0.5% Tween 20 (Sigma, France) chondroitin sulfate – sodium hyaluronate (Viscoat Alcon) was and 0.1 g L−1 Thimerosal (Sigma, France) in PBS for 1 h at Frontiers in Neuroscience | www.frontiersin.org 3 March 2021 | Volume 15 | Article 615256 Nguyen et al. Novel Graphene for Retinal Implants room temperature. Afterward, a 3-day incubation at 4◦ C with factor for neural affinity (Veliev et al., 2016). Using the metallic slow stirring was done, followed by an incubation at room tracks designed on the device extending up to the circular head temperature for 2 h with primary antibodies in blocking solution. of the device, the successful transfer of graphene on this implant The antibodies used were polyclonal antibodies directed against head was verified by measuring the resistance between the tracks. Chicken anti Glial Fibrillary Acidic Protein (1:100, LifeSpan The resistance measured for all the implants was contained Biosciences, WA, United States), Rabbit anti Iba1 (1:500, Wako between 0.5 and 1 kOhms instead of MOhms in the absence of Sobioda, France). The fragments were rinsed and then incubated graphene indicating that the two tracks were indeed connected with secondary antibodies: goat anti-Chicken IgY Alexa 647 by the deposited graphene. and goat anti-rabbit IgG Alexa 488 (1:500, Molecular Probes, Invitrogen, Eugene, OR, United States) for 2 days at 4◦ C Implantation and in vivo Monitoring followed by incubation at room temperature for 1 h. The nuclear A total of 14 animals were used for this study. The animal stain, 40 ,6-diamidino-2-phenylindole (DAPI), was added to the weighed an average 348 ± 15 g at the time of implantation and incubating solution prior to washing steps. 393 ± 25 g at time of euthanasia 3 months later. To verify the The implant/retina ensemble was then rinsed and mounted, implant position, and in order to monitor any in vivo tissue in permanent mounting medium (MM France, France), on inflammation, the eye fundus and OCT image of an optical a microscope slide, for viewing under an upright confocal retinal section was taken at specified intervals of 7, 15, 30, 60, microscope (FV1000, Olympus). DAPI counterstaining, and 90 days. Figure 3 illustrates an example of the in vivo AlexaFluor-488, and AlexaFluor-647 were used to be detected by monitoring by eye fundus and OCT scanning of an animal with excitation with a 405 nm laser diode, a 488 nm argon ion laser, a graphene implant and an animal with a control polyimide and 635 nm laser diode lines, respectively. implant. It should be noted that eye fundus was taken during a transition from the Micron III to the Micron IV machine Microglial Count and Statistical Analysis therefore displaying a transition in image results between day 60 Confocal data was imported into Bitplane Imaris software for and day 90 for the graphene images and between day 15 and day visualization and analysis of microglial cells in the entire retina 30 for the polyimide control images. By eye fundus imaging, it using the Surface creation wizard. The Iba1 channel was selected can be seen that the implant was placed dorsal to the optic nerve and surfaces were generated from the signal using a set threshold and remained in the same location during the entire monitoring to accurately visualize the source signal, which was then modeled period. Additionally, no apparent delamination of the graphene, by the software. Cells were verified by colocalization using the passivation layer, nor the metal contacts can be identified. The DAPI signal marking cell nuclei. False counts, duplicates as well implant was confirmed to be in the subretinal space visualized as cells overlapping the boarder of the image were filtered before by the large retinal vasculature above the implant, the white line displaying the final result. Average cell volume was obtained by indicates the area where the OCT image was taken. dividing counted cells by the total area generated from the signal Optical coherence tomography imaging provided a second source. Significance test was done using an unpaired two-tailed confirmation of a subretinal implantation, showing that the Welch’s t-test utilizing a p < 0.05 significance level. implant was placed between the choroid and the retina. OCT images displayed in Figure 3 were averaged from five consecutive frames of the OCT scan at the location of the eye indicated RESULTS by the white line in the eye fundus images. The OCT imaging demonstrated that retinal reattachment had occurred at day 7 Microfabrication with no subsequent retinal detachment until the end of the The layer graphene grown for this study were characterized on monitoring for both types of implants. Looking closely at the copper foil after the graphene growth and before their transfer implant/retina OCT images (bottom row of Figure 3), the retina to the polymer substrate using SEM and Raman spectroscopy. seems to be uniform and in close contact with the implant, The SEM image (Figure 2a) shows that the graphene layer indicating no inflammation. Supplementary Figure 1 shows the is closed. Additionally some small second nucleation can be eye fundus and OCT scan of P23H rats at age of 9 and 12 months distinguished every 50 µm. Figure 2b shows a representative that had no implantation. Raman spectrum of the layer graphene. Figures 2c,d are mappings of 15 µm × 15 µm of the 2D/G ratio and 2D full Immunohistochemistry width at half maximum (FWHM). The G (around 1590 cm−1 ) After the in vivo monitoring, inflammation at the cellular level and 2D (around 2720 cm−1 ) bands of graphene can be clearly was examined. The eye fragment containing the implant was fixed observed while the By contrast, the D (around 1350 cm−1 ) band and treated to visualize glial cells. Histological examinations of cannot be distinguished suggesting that the layer has very few the tissue indicated the well-preserved retina above the implant, defects (Ferrari, 2007). Furthermore, the value of 2D/G ratio of as it can be seen at a 10x magnification in Figure 4. An around 4, and the 2D full width at half maximum (FWHM) immunolabelling was achieved to reveal the microglial cells (Iba1 of around 25 cm−1 combined to the good homogeneity of this antibody) and macroglial cells (GFAP antibody), while the DAPI characteristics over the mapped area. This provides proof of staining revealed all cell nuclei within the retinal tissue above the a good crystalline quality single graphene layer (Ferrari, 2007; implant. At the low magnification (10x), the implant could be Roscher et al., 2019) which has be shown to be an important seen below the retina by the green polyimide autofluorescence Frontiers in Neuroscience | www.frontiersin.org 4 March 2021 | Volume 15 | Article 615256 Nguyen et al. Novel Graphene for Retinal Implants FIGURE 2 | SEM and Raman spectroscopy of the single layer graphene (a) SEM of graphene on copper. The black arrows show the position of the second nucleation, (b) representative Raman spectrum, no D band can be observed, (c) 2D/G mapping of a 15 µm × 15 µm area, (d) FWHM mapping of a 15 µm × 15 µm area. when imaging at the bipolar cell layer of both types of devices the depth of the retina with no abnormal accumulation at the (Figure 4), and also when looking at the vertical retinal sections surface of the graphene implant direct contact. (Figures 5i,j). On the same field of view, the GFAP-macroglial cells were observed to remain at the surface of the retinal Statistical Analysis of Inflammatory tissue. In both types of implants (Figure 4), the classic dense Markers retinal network of astrocytes was visible at the low and high To further quantify any potential retinal reactive state at magnifications. Iba1-immunopositive microglial cells (green in the graphene implant interface, the microglial cells above the Figure 4, 5) were seen in the tissue depth to form a more discrete implants were counted. Figure 6a illustrates the strategy used network of cells among other cells revealed by DAPI-stained for quantifying microglial cells within the retinal volume with nuclei (blue in Figures 4, 5). The distribution of microglial cells a 0.1 mm2 base area. Using Iba1 data from confocal imaging, could be discretely defined inside the tissue depth by serial optical microglial cells could be reconstituted within the investigated confocal imaging allowing reconstruction of vertical sections volume for a semi-automated counting. Individual cells were showing the different layers of DAPI-stained nuclei (Figure 5). quantified for both graphene implants and control polyimide This cell type were observed in different horizontal levels within implants. A quantification on non-operated P23H rat eyes at the retinal tissue in the inner plexiform layer (IPL) below the the same age was also added for comparison to the degenerated ganglion cell layer containing dispersed DAPI-stained nuclei retina. This quantification showed that the number of microglial and in the outer plexiform layer (OPL) just below the implant. cells within the retina was significantly greater in the presence The inner nuclear layer (INL) composed by dense DAPI-stained of the control polyimide implant (123.6 ± 21.1, n = 5) than in nuclei was devoid of microglial cells as in the normal tissue. the control non-operated eyes (67.0 ± 8.4, n = 7, p = 0.044). Therefore, the microglial distribution appeared normal within This could be expected from the surgical stress and the implant Frontiers in Neuroscience | www.frontiersin.org 5 March 2021 | Volume 15 | Article 615256 Nguyen et al. Novel Graphene for Retinal Implants FIGURE 3 | In vivo imaging of devices with graphene (top set) and devices without graphene (bottom set) at different specified time intervals. Each set includes eye fundus (top row), OCT scans of device in the eye (middle row), and a magnified image of OCT scan showing only the retina and device (bottom row). Eye fundus provides evidence for the long-term stability of both types of devices up to 90 days. The white line indicates the location where the OCT scans were taken. The OCT scans shows the implant (red arrow) sandwiched between the retina (white arrow) and the choroid (yellow). Dotted white box is a zoom of each OCT scan showing in detail the retina and device. The magnified images of the OCT scans (bottom row) show the close contact of the retina and implant devices. Bottom row of each set is the zoom on the retina in contact with implant as indicated by dashed line of the OCT scan. All scale bars are 200 µm. presence (Figure 6b). However, this difference was reduced non-operated eye and operated eye (p < 0.5), the average volume for implants consisting graphene (79.86 ± 8.8, n = 7) with in the graphene implant and polymer implant is similar to each respect to the retina from non-operated eyes with no significant other and smaller than non-operated tissue. The results signify difference between the two. All differences between the graphene that graphene does not impact the biocompatibility of polyimide implant with either the polymer implants (p = 0.110) and the and SU-8 polymers. control P23H retina (p = 0.131) were not statistically significant. The results by microglial count signify that graphene is more biocompatible than the polymer implant. DISCUSSION In pathological conditions, microglial cells are known to change from a star-like shape with long processes indicating a Graphene is a new carbon-based material with a large relaxed state to a more amoeboid morphology. This change in water window providing exceptional electrical performances state alters the cell body volume, which can be quantified through for neurophysiological recording and capable charge injection the Iba1-immunolabeled microglial cells. In non-operated eyes, for neuronal stimulation. Despite this, free carbon-nanotubes the microglial cell volume (749.9 ± 118.8 µm3 ) was greater than were found to present some toxicity issues (Yan et al., 2019). that measured for the graphene implant (556.2 ± 251.8 µm3 ) and Therefore, it is necessary to determine whether a carbon- for the control polymer implants (554.8 ± 299.9 µm3 ). Though based electrode tethered to a device would present toxicity there are no significant differences in microglial volume between to nervous tissue. Previous cell culture studies have already Frontiers in Neuroscience | www.frontiersin.org 6 March 2021 | Volume 15 | Article 615256 Nguyen et al. Novel Graphene for Retinal Implants FIGURE 4 | Low (10x) and high (40x) magnifications of retinal tissues at the graphene and polyimide implant interface. Images show the different immunolabeled cells: green Iba1-positive microglia, and white GFAP-positive Müller and astrocyte cells with blue DAPI-stained nuclei. Note the green autofluorescent polyimide ring of the implants at the 10x magnification in the bipolar layer of the polyimide control and at both layer for the graphene implant. Scale bars are 50 µm (10x) and 200 µm (40x). demonstrated the high biocompatibility of CVD graphene electrode, which represents the largest possible graphene-retinal bounded to a hard material (glass) for retinal neurons interface without increasing the dimensions of the device. The (Bendali et al., 2013) as well as investigated its effect on graphene electrode was connected to two metal contacts used to peripheral neurons (Convertino et al., 2018). Additionally, verify the presence of graphene. However, the lack of functional CVD graphene coated onto intracortical probes that were electrode in these devices did not allow for the assessment chronically implanted in mice cortex showed that that the of tissue damage and inflammation that could result from presences of graphene had reduced the local density of astrocytes electrical stimulation. and microglia at the implanted site (Bourrier et al., 2019). Functional devices require a tail relayed to an external Following graphene transfer on soft biocompatible polymer headstage connector, which is not the case here. However, such material, the in vivo biocompatibility of graphene in the complete systems introduce an additional mechanical constraint retina was shown here. The retina represents a relatively within a moving eye could lead to potential shear stress at the accessible location to the central nervous system and could electrode array which could result in an inflammation. Such bypass complication of placing materials directly on the brain. functional devices had been for instance produced with titanium– These results are important for further use of graphene in platinum or diamond electrodes for monitoring the evolution retinal prostheses for restoring vision. However, the results of electrode impedance over time (Linderholm et al., 2013; have wider relevance for applications in any other part of the Hébert et al., 2016). Eye fundus examination and OCT imaging central nervous system. confirmed the absence of abnormal situations (edema and In the context of visual restoration, the study was performed hemorrhage, Linderholm et al., 2013) during the in vivo period. in the blind homozygous P23H rat, which is widely used as a This in vivo monitoring further demonstrated that graphene model of retinitis pigmentosa due to a complete degeneration of remained in tight contact with the retinal tissue for the whole photoreceptors occurring prior to 9 months (Pinilla et al., 2005; implantation period. Orhan et al., 2015). The 3 months in vivo follow-up showed Microglial cells act in the retina and more generally in a high stability of the large area of graphene placed between the central nervous system as resident macrophages assessing flexible polyimide and SU-8 substrate without any delamination. cell function and any intrusion. Biocompatibility of various Generally, retinal implant arrays contain microelectrodes (less materials could be investigated by assessing the reactive microglia than 100 µm in diameter) for electrical stimulation. Here, a status as indicated by microglial proliferation and amoeboid worst-case scenario was assessed by using a 1 mm diameter transformation into macrophages. In the degenerating retina, Frontiers in Neuroscience | www.frontiersin.org 7 March 2021 | Volume 15 | Article 615256 Nguyen et al. Novel Graphene for Retinal Implants FIGURE 5 | Microglial distributions within the retinal tissues at the graphene (a–d,j) and control polymer interface (e–i). Confocal images of the retinal tissues showing Iba1-positive microglia cells (green) at different depths. (a,e) Optic fiber layer showing GFAP-positive Müller and astrocyte cells (white) and no microglial cells. Iba1-positive microglial cells with DAPI-stained nuclei (blue) at the junction between the inner plexiform layers (IPL) and the ganglion cell layer (GCL) (b,f), in the inner nuclear layer (INL) (c,g) in the outer plexiform layer (OPL) (d,h). Confocal reconstruction of vertical sections for the retina above the control polymer implant (i) and the graphene implant (j) showing the different depths (1–4) for all above horizontal sections (a–h). Scale bars represent 100 µm in (a–h) and 50 µm in (i,j). FIGURE 6 | Modeling of microglia based on Iba1 marker on retinal tissue in contact with devices with graphene (a). Each different color represent a different microglial. Scale bars are 50 µm. Graph (b) is the statistical analysis of microglial count comparing retinal tissue in contact with the two types of implant and retinal tissue from non-operated eyes. Graph (c) is the comparison of microglial volume between the three conditions. ns indicates p > 0.05, * indicates p ≤ 0.05. microglial cells are expected to be already in a reactive state. in microglial features, microglial count in the retina in contact We therefore decided to use the Iba1 marker to label both with polymer implants was significantly greater than in non- reactive and relaxed microglial cells in the retina to visualize operated eyes. However, there was no significant difference the inflammatory response at the implant position. Assessing between that of graphene implants and non-operated eyes. cell volume gave an indication on morphology, which indirectly Additionally, average microglial volume showed no significant relates to the active/non-reactive state of the tissue. Although differences between graphene and control implants made with surgery and the presences of a retinal implant induced a change biocompatible polymers. In addition, there were no significant Frontiers in Neuroscience | www.frontiersin.org 8 March 2021 | Volume 15 | Article 615256 Nguyen et al. Novel Graphene for Retinal Implants differences between both implant and non-operated eyes. These methodology, writing-original draft, resources, investigation, and findings are consistent with a high biocompatibility of graphene- formal analysis. JD: methodology, resources, investigation, and based electrodes as previously suggested with in vitro studies writing-review and editing. LB: investigation, formal analysis, (Bendali et al., 2013; Convertino et al., 2018), and in vivo studies and writing-review and editing. XI: validation, investigation, (Bourrier et al., 2019). However, future works should assess and writing-original draft. JC and EC: validation, writing-review neuronal function to demonstrate further the biocompatibility and editing. JB: investigation, validation, data curation, and of the graphene material although this assessment is particularly writing-review and editing. JG: conceptualization, supervision, difficult in blind animals with a flat electroretinogram (Pinilla writing-review and editing, and funding acquisition. CH: et al., 2005; Orhan et al., 2015). conceptualization, investigation, supervision, visualization, validation, writing-original draft, and writing-review and editing. SP: conceptualization, supervision, funding acquisition, writing- CONCLUSION original draft, and writing-review and editing. All authors: contributed to the article and approved the submitted version. Assessing in vivo material biocompatibility is an essential step in developing new technologies for neuronal recording and stimulation in electronic devices. This study focused on CVD graphene material and its potential use in active electrodes. The FUNDING study demonstrates that the presences of graphene induces a This project has received funding from the European Union’s lesser inflammation quantified by microglial labeling compared Horizon 2020 research and innovation programme under to biocompatible polymers. Graphene can therefore be further grant agreement No. 785219 (GrapheneCore2) and No. investigated for use in electrodes for retinal prostheses or more 881603 (GrapheneCore3). DN has received funding from generally any electronic device for recording and stimulation the doctoral school of Cerveau, cognition, comportement of the central nervous system. Further studies will assess more (3C) of Sorbonne Université. SP was also supported by the developed graphene structures on chronic devices to compare French state funds managed by the Agence Nationale de la biocompatibility, safety and stimulation capabilities between Recherche within the Programme Investissements d’Avenir, classic materials and graphene-based electrodes. LABEX LIFESENSES (ANR-10-LABX-65) and IHU FOReSIGHT (ANR-18-IAHU-0001). This work has made use of the Spanish DATA AVAILABILITY STATEMENT ICTS Network MICRONANOFABS partially supported by MICINN and the ICTS ‘NANBIOSIS,’ more specifically by the The raw data supporting the conclusions of this article will be Micro-NanoTechnology Unit of the CIBER in Bioengineering, made available by the authors, without undue reservation. Biomaterials and Nanomedicine (CIBER-BBN) at the IMB-CNM. ETHICS STATEMENT SUPPLEMENTARY MATERIAL The animal study was reviewed and approved by Charles Darwin No5 Ethics Committee in Animal Experimentation (agreement The Supplementary Material for this article can be found #15219). Written informed consent was obtained from the online at: https://www.frontiersin.org/articles/10.3389/fnins. owners for the participation of their animals in this study. 2021.615256/full#supplementary-material Supplementary Figure 1 | Typical imaging of eye fundus and OCT scan of non-implanted P23H rat at 9 and 12 months of age. The white line on the eye AUTHOR CONTRIBUTIONS fundus indicates the location where the OCT scans were taken. The OCT scans shows the retina (white arrow), the choroid (yellow) and the sclera (blue). Dotted DN: conceptualization, methodology, investigation, writing- white box is a zoom of each OCT scan showing in detail the retina. All scale bars original draft, formal analysis, and visualization. MV: are 200 µm. REFERENCES Bourrier, A., Shkorbatova, P., Bonizzato, M., Rey, E., Barraud, Q., Courtine, G., et al. (2019). Monolayer graphene coating of intracortical probes for long- Apollo, N. V., Maturana, M. I., Tong, W., Nayagam, D. A. X., Shivdasani, M. N., lasting neural activity monitoring. Adv. Healthcare Mater. 8:1801331. doi: 10. Foroughi, J., et al. (2015). Soft, flexible freestanding neural stimulation and 1002/adhm.201801331 recording electrodes fabricated from reduced graphene oxide. Adv. Funct. Chow, A. Y., Pardue, M. T., Chow, V. Y., Peyman, G. A., Chanping, L., Perlman, Mater. 25, 3551–3559. doi: 10.1002/adfm.201500110 J. I., et al. (2001). Implantation of silicon chip microphotodiode arrays into Bendali, A., Hess, L. H., Seifert, M., Forster, V., Stephan, A. F., Garrido, the cat subretinal space. IEEE Trans. Neural Syst. Rehabil. Eng. 9, 86–95. doi: J. A., et al. (2013). Purified neurons can survive on peptide-free graphene 10.1109/7333.918281 layers. Adv. Healthcare Mater. 2, 929–933. doi: 10.1002/adhm.20120 Convertino, D., Luin, S., Marchetti, L., and Coletti, C. (2018). Peripheral neuron 0347 survival and outgrowth on graphene. Front. Neurosci. 12:1. doi: 10.3389/fnins. Blaschke, B. M., Lottner, M., Drieschner, S., Calia, A. B., Stoiber, K., Rousseau, L., 2018.00001 et al. (2016). Flexible graphene transistors for recording cell action potentials. da Cruz, L., Dorn, J. D., Humayun, M. S., Dagnelie, G., Handa, J., Barale, P.-O., 2D Mater. 3:025007. doi: 10.1088/2053-1583/3/2/025007 et al. (2016). Five-year safety and performance results from the argus II retinal Frontiers in Neuroscience | www.frontiersin.org 9 March 2021 | Volume 15 | Article 615256 Nguyen et al. Novel Graphene for Retinal Implants prosthesis system clinical trial. Ophthalmology 123, 2248–2254. doi: 10.1016/j. Palanker, D., Le Mer, Y., Mohand-Said, S., Muqit, M. M. K., and Sahel, J. A. (2020). ophtha.2016.06.049 Photovoltaic restoration of central vision in atrophic age-related macular Dong, X., Wang, X., Wang, L., Song, H., Zhang, H., Huang, W., et al. (2012). degeneration. Ophthalmology 127, 1097–1104. doi: 10.1016/j.ophtha.2020.02. 3D graphene foam as a monolithic and macroporous carbon electrode for 024 electrochemical sensing. ACS Appl. Mater. Interfaces 4, 3129–3133. doi: 10. Park, D.-W., Ness, J. P., Brodnick, S. K., Esquibel, C., Novello, J., Atry, F., et al. 1021/am300459m (2018). Electrical neural stimulation and simultaneous in vivo monitoring with Edwards, T. L., Cottriall, C. L., Xue, K., Simunovic, M. P., Ramsden, J. D., transparent graphene electrode arrays implanted in GCaMP6f mice. ACS Nano Zrenner, E., et al. (2018). Assessment of the electronic retinal implant alpha 12, 148–157. doi: 10.1021/acsnano.7b04321 AMS in restoring vision to blind patients with end-stage retinitis pigmentosa. Park, D.-W., Schendel, A. A., Mikael, S., Brodnick, S. K., Richner, T. J., Ness, Ophthalmology 125, 432–443. doi: 10.1016/j.ophtha.2017.09.019 J. P., et al. (2014). Graphene-based carbon-layered electrode array technology Fan, Z., Yan, J., Ning, G., Wei, T., Zhi, L., and Wei, F. (2013). Porous graphene for neural imaging and optogenetic applications. Nat. Commun. 5:5258. doi: networks as high performance anode materials for lithium ion batteries. Carbon 10.1038/ncomms6258 60, 558–561. doi: 10.1016/j.carbon.2013.04.053 Pinilla, I., Lund, R. D., and Sauvé, Y. (2005). Enhanced cone dysfunction in rats Ferrari, A. C. (2007). Raman spectroscopy of graphene and graphite: disorder, homozygous for the P23H rhodopsin mutation. Neurosci. Lett. 382, 16–21. electron–phonon coupling, doping and nonadiabatic effects. Solid State doi: 10.1016/j.neulet.2005.02.055 Commun. 143, 47–57. doi: 10.1016/j.ssc.2007.03.052 Pinto, A. M., Gonçalves, I. C., and Magalhães, F. D. (2013). Graphene-Based Hahn, P., and Fine, H. F. (2018). Practical concepts with the argus II retinal Materials Biocompatibility: A Review. Amsterdam: Elsevier. prosthesis. Ophthalmic Surg. Lasers Imaging Retina 49, 742–746. doi: 10.3928/ Polikov, V. S., Tresco, P. A., and Reichert, W. M. (2005). Response of brain tissue 23258160-20181002-01 to chronically implanted neural electrodes. J. Neurosci. Methods 148, 1–18. Hébert, C., Cottance, M., Degardin, J., Scorsone, E., Rousseau, L., Lissorgues, doi: 10.1016/j.jneumeth.2005.08.015 G., et al. (2016). Monitoring the evolution of boron doped porous diamond Prévot, P.-H., Gehere, K., Arcizet, F., Akolkar, H., Khoei, M. A., Blaize, K., et al. electrode on flexible retinal implant by OCT and in vivo impedance (2020). Behavioural responses to a photovoltaic subretinal prosthesis implanted spectroscopy. Mater. Sci. Eng. C 69, 77–84. doi: 10.1016/j.msec.2016.06.032 in non-human primates. Nat. Biomed. Eng. 4, 172–180. doi: 10.1038/s41551- Hébert, C., Masvidal-Codina, E., Suarez-Perez, A., Calia, A. B., Piret, G., Garcia- 019-0484-2 Cortadella, R., et al. (2017). Flexible graphene solution-gated field-effect Pumera, M., Ambrosi, A., Bonanni, A., Chng, E. L. K., and Poh, H. L. transistors: efficient transducers for micro-electrocorticography. Adv. Funct. (2010). Graphene for Electrochemical Sensing and Biosensing. Amsterdam: Mater. 28, 1–15. doi: 10.1002/adfm.201703976 Elsevier. Hornig, R., Dapper, M., Le Joliff, E., Hill, R., Ishaque, K., Posch, C., et al. Roscher, S., Hoffmann, R., and Ambacher, O. (2019). Determination of the (2017). “Pixium Vision: First Clinical Results and Innovative Developments,” in graphene–graphite ratio of graphene powder by Raman 2D band symmetry Artificial Vision: A Clinical Guide, ed. V. P. Gabel (Cham: Springer International analysis. Anal. Methods 11, 1224–1228. doi: 10.1039/C8AY02619J Publishing), 99–113. Salzmann, J., Linderholm, O. P., Guyomard, J. L., Paques, M., Simonutti, M., Ji, H., Zhao, X., Qiao, Z., Jung, J., Zhu, Y., Lu, Y., et al. (2014). Capacitance Lecchi, M., et al. (2006). Subretinal electrode implantation in the P23H rat for of carbon-based electrical double-layer capacitors. Nat. Commun. 5:3317. doi: chronic stimulations. Br. J. Ophthalmol. 90, 1183–1187. doi: 10.1136/bjo.2005. 10.1038/ncomms4317 089110 Jiang, L., and Fan, Z. (2014). Design of advanced porous graphene materials: Slutzky, M. W. (2019). Brain-Machine interfaces: powerful tools for clinical from graphene nanomesh to 3D architectures. Nanoscale 6, 1922–1945. doi: treatment and neuroscientific investigations. Neuroscientist 25, 139–154. doi: 10.1039/C3NR04555B 10.1177/1073858418775355 Joucla, S., and Yvert, B. (2009). Improved focalization of electrical Stingl, K., Schippert, R., Bartz-Schmidt, K. U., Besch, D., Cottriall, C. L., Edwards, microstimulation using microelectrode arrays: a modeling study. PLoS T. L., et al. (2017). Interim results of a multicenter trial with the new One 4:e4828. doi: 10.1371/journal.pone.0004828 electronic subretinal implant alpha ams in 15 patients blind from inherited Kostarelos, K., Vincent, M., Hebert, C., and Garrido, J. A. (2017). Graphene in the retinal degenerations. Front. Neurosci. 11:445. doi: 10.3389/fnins.2017. Design and Engineering of Next-Generation Neural Interfaces. Adv. Mater. 29, 00445 1–7. doi: 10.1002/adma.201700909 Veliev, F., Briançon-Marjollet, A., Bouchiat, V., and Delacour, C. (2016). Impact Kuzum, D., Takano, H., Shim, E., Reed, J. C., Juul, H., Richardson, A. G., of crystalline quality on neuronal affinity of pristine graphene. Biomaterials 86, et al. (2014). Transparent and flexible low noise graphene electrodes for 33–41. doi: 10.1016/j.biomaterials.2016.01.042 simultaneous electrophysiology and neuroimaging. Nat. Commun. 5:5259. doi: Yan, H., Xue, Z., Xie, J., Dong, Y., Ma, Z., Sun, X., et al. (2019). Toxicity of carbon 10.1038/ncomms6259 nanotubes as anti-tumor drug carriers. Int. J. Nanomed. 14, 10179–10194. doi: Lee, Y., Bae, S., Jang, H., Jang, S., Zhu, S.-E., Sim, S. H., et al. (2010). Wafer- 10.2147/IJN.S220087 scale synthesis and transfer of graphene films. Nano Lett. 10, 490–493. doi: Yin, R., Xu, Z., Mei, M., Chen, Z., Wang, K., Liu, Y., et al. (2018). Soft transparent 10.1021/nl903272n graphene contact lens electrodes for conformal full-cornea recording Li, N., Zhang, Q., Gao, S., Song, Q., Huang, R., Wang, L., et al. (2013). Three- of electroretinogram. Nat. Commun. 9:2334. doi: 10.1038/s41467-018- dimensional graphene foam as a biocompatible and conductive scaffold for 04781-w neural stem cells. Sci. Rep. 3:1604. doi: 10.1038/srep01604 Linderholm, P., Guyomard, J.-L., Djilas, M., Salzmann, J., Simonutti, M., Sahel, Conflict of Interest: SP is a funder and shareholder of Pixium Vision. J. A., et al. (2013). Long-term in vivo impedance changes of subretinal microelectrodes implanted in dystrophic P23H rats. Int. J. Artif. Organs 36, The remaining authors declare that the research was conducted in the absence of 612–619. doi: 10.5301/ijao.5000213 any commercial or financial relationships that could be construed as a potential Lorach, H., Goetz, G., Smith, R., Lei, X., Mandel, Y., Kamins, T., et al. (2015). conflict of interest. Photovoltaic restoration of sight with high visual acuity. Nat. Med. 21, 476–482. doi: 10.1038/nm.3851 Copyright © 2021 Nguyen, Valet, Dégardin, Boucherit, Illa, de la Cruz, del Corro, Lu, Y., Lyu, H., Richardson, A. G., Lucas, T. H., and Kuzum, D. (2016). Flexible Bousquet, Garrido, Hébert and Picaud. This is an open-access article distributed neural electrode array based-on porous graphene for cortical microstimulation under the terms of the Creative Commons Attribution License (CC BY). The use, and sensing. Sci. Rep. 6:33526. doi: 10.1038/srep33526 distribution or reproduction in other forums is permitted, provided the original Orhan, E., Dalkara, D., Neuillé, M., Lechauve, C., Michiels, C., Picaud, S., et al. author(s) and the copyright owner(s) are credited and that the original publication (2015). Genotypic and phenotypic characterization of P23H line 1 rat model. in this journal is cited, in accordance with accepted academic practice. No use, PLoS One 10:e0127319. doi: 10.1371/journal.pone.0127319 distribution or reproduction is permitted which does not comply with these terms. Frontiers in Neuroscience | www.frontiersin.org 10 March 2021 | Volume 15 | Article 615256