In vivo assembled, transient, and bioresorbable bioelectronics have the potential to supplement pharmaceuticals in nonchronic treatments (e.g. immunotherapy, pain, and cancer). Conventional drugs typically solely address biochemical processes, whereas bioelectronics tackle the dysfunction of bioelectric circuits. Research on conductive structures in vivo mainly focuses on chronic applications. In transient bioelectronics, minimally invasive methods for implantation and bioresorbable materials are central requirements in addition to function. 1 The latter is because there is no need for revision surgery after treatment using transient materials. This would, for example, be important in the electrotherapy of solid tumors, especially in the brain, such as glioblastoma. 2,3
Various strategies have been undertaken to design electrodes that seamlessly connect to the structures of the nervous system in animals. 4-8 Most depend on external cues, such as electrons/electric fields, chemicals (including proteins), or genetic engineering. In pioneering work by Martin and co-workers, conducting polymers were formed within rodents' brains. A solution of the monomer (3,4-ethylenedioxythiophene [EDOT]) and the larger polymer polystyrene sulfonate (PSS), acting as a template, was injected and electropolymerized in vivo to derive a massive spheric cloud protruding from the electrode. 9 To further advance the concept of in vivo polymerization, the genetic engineering of animals was carried out to make enzymes expressed in specific cells that promote the local polymerization of aniline near the target tissue in nematode C. elegans and mice. 6 However, from this study, it is not easy to evaluate the importance of the expression specificity of the enzymes in vivo. After the syringe injection of aniline, a dense and comparably large hemisphere of polymers was formed in the mouse brain. This is probably because of diffusion-controlled polymerization. Although elegant, the dependence on genetic engineering is not optimal for nonchronic therapies and complicates its use in humans. To avoid genetic manipulation, the enzymes that catalyze polymerization could be endogenously expressed. Endogenous peroxidase enzymes polymerize 4-(2-(2,5-bis(2,3-dihydrothieno[3,4-b][1,4]dioxin-5-yl)thiophene-3-yl)ethoxy)butane-1-sulfonate (ETE-S, Fig. 1) in plants10 and the tiny freshwater organism Hydra vulgaris. 11 However, this promising approach is limited to tissues that exhibit sufficient peroxidase activity, and no mammalian peroxidases have been reported to form conductive polymers in vivo. In 2023, we solved this problem by in vivo injecting a cocktail including an oxidase enzyme and horseradish peroxidase (HRP), where the oxidase consumed endogenous metabolites while generating hydrogen peroxide for HRP-mediated polymerization of an ETE-S analog, ETE-COOH, into a conductive structure. 8 However, these enzymatic in vivo formed conductive polymers are used in passive mode; they are not affected by direct external stimuli. Thus, combining the formation of seamless cell-integrating organic microstructures with direct external stimuli has proven difficult.
Our exploratory research used vertebrate zebrafish (Danio rerio) as a model. Zebrafish have proven to be translationally relevant; they possess evolutionary conserved features, ranging from gross brain structure to behavior. The use of zebrafish is cost-efficient and regarded as a 3R (reduce, refine, and replace) alternative, making ethical considerations more defendable than exploratory studies in mammals. Allometric scaling to larger animals and humans is possible in compliance with traditional drug development. 12
Recently, we reported on the discovery of A5, a self-doped water-soluble mixed ion–electron conductor poly(3,4-ethylenedioxythiophene)butoxy-1-sulfonate (PEDOT-S) derivative (Fig. 1). 13 Among the PEDOT-S derivatives, A5 is unique because it self-assembles in an agarose gel cast with a physiological buffer and generates a highly conductive hydrogel (1-5 S cm-1) stable for several months. Furthermore, on average, A5 comprises small polymers (i.e., oligomers) of 7–8 monomers. Thus, A5 is smaller than the antisense oligonucleotide drugs (about 20 nucleotides) and is expected to have better bioresorption properties than PEDOT:PSS, where the PSS part is a large polymer of 200–300 monomers (Mn ~70 000 g mol-1).
Here, we report a general approach that is not dependent on specific external or endogenous triggers to assemble bioresorbable high-performing electrode structures within the central nervous system (CNS) implanted using a minimally invasive method. We show that highly water-dispersed nanoparticles—comprising the conductive polymer A5—injected into a zebrafish brain using a small-diameter capillary self-organize into a mixed ion–electron conducting hydrogel. The conductive hydrogel is transient, and the initial inflammation in the brain caused by the injection resolves, leaving no tissue damage from the electrode behind. To modify the initial conductive structure after injection, ETE-R derivatives (e.g., ETE-S and ETE-PC, Fig. 1) were mixed into the A5 solution, injected into the brains, and electropolymerized. ETE-Rs provide the functionalization of choice and facilitate reaching neurons where the initially formed electrodes would not. The ETE-Rs are designed to have a lower oxidation potential than the EDOT monomer: 0.3–0.5 V and 1.2 V, respectively, which is significant for minimal damage to the tissue during electropolymerization. In addition to the structural modification (shape and functionality) and increased in vivo stability, conductivities of two to three orders of magnitude higher than tissue were measured.
A5 self-assembles to form a conductive electrode in tissue.
Low-concentration agarose gel cast with physiological buffer (Ringer’s solution) mimics brain tissue and is an excellent introduction to understanding the aggregation behaviors of self-organizing conductive polymers. Divalent ions can enable A5 to self-assemble into a long-term stable hydrogel in Ringer’s solution–agarose gel showing significantly higher conductivity than the surrounding environment. 13 This inspired exploration of the use of A5 as an injectable in vivo electrode. Although tissue mimics and cell cultures are beneficial and ethically ideal for establishing basic properties, highly dynamic homeostasis and cellular diversity in vivo are challenging to study in vitro. In this exploratory study, the 3R vertebrate zebrafish model was used for in vivo studies; it is ethically sound and translationally relevant.
The zebrafish caudal fin is a model system for limb regeneration and neuropathy—it is highly dynamic. It is transparent, enabling direct optical access to an injected polymer. A5 (20 mg mL− 1) was injected into the caudal fin of zebrafish between the fin rays (Fig. 1B). To facilitate the assembly of the soft electrodes in the fin, a 25% Ringer’s solution was used to dissolve the A5. This use of a formulation with additional ion strength, compared to Milli-Q water, is not necessary when injecting into the zebrafish brain, as will be shown below. However, this exemplifies the adaptive nature of A5, matching the nanoparticle formulation with specific tissue and making it possible to inject into lower and higher ionic strength regions. Directly after injections, dark-blue coherent structures were formed between the fin rays (Fig. 1B).
The A5 was contacted using two tungsten microelectrodes connected to a Keithley sourcemeter to measure the gel’s resistance (Fig. S1, Fig. S8). This proved challenging as the A5 gel was pushed around by the W-electrodes leading to poor W–gel contact. Microampere currents were measured (resistance 0.16 MΩ) in the A5 gel; this is about twice the current compared to the reference rays without A5 (0.32 MΩ), Fig. S1. The difference was more pronounced after drying the fin, reaching a difference of two orders of magnitude. This is because of a decreased resistance of the A5 (0.02 MΩ) and an increased resistance of the reference fin rays (1 MΩ), which was not surprising since the conductive A5 oligomers were expected to come closer together upon drying, thereby allowing for higher conductivity.
The conductive structure had partly disappeared after one week for fish left to swim with A5 in their caudal fins. It was completely gone after four weeks (Fig. 1B). Throughout the experiment, no changes in zebrafish behavior or fin damage were observed. Thus, the electrodes were transient and bioresorbed, leaving healthy tissue behind.
The brain is the most complex and delicate biological system and challenging to correctly mimic in surrogate models. Therefore, A5 was injected into the brain for direct evaluation in the targeted environment (Fig. 1C). For studies in zebrafish brains, A5 (20 mg mL− 1) was injected using the columnar injection technique, a method used for intracerebral cell therapy in humans. Cannula diameters of 700 µm have been used for cell therapy where injected cells are in the order of 10–20 µm in diameter. However, to avoid blood vessel rupture and bleeding in the brain, it was recently suggested that a syringe diameter below 25 µm should be used to implant bioelectronics. 14 A5 nanoparticles are highly water-dispersed and have an average diameter of 80 nm, 13 making it possible to use cannulas in the suggested size range. In this study, we used a diameter of 30 µm, which was within the margin of error of the recommended size. Immediately after injection, the fish were revived and allowed to swim for up to nine days, showing no detrimental effects from the injected polymer.
A redox histochemistry method was used to evaluate the response to the oxidative insults of injections and the soft electrode formed on brain tissue, a technique that allows for sensitive detection of changes in the intracellular redox state in situ; thiol groups in glutathione and proteins sense the cellular redox environment. 15 Two hours post-injection, an extensive inflammatory response was observed around the injection area (Fig. 1D). The inflammation was resolved nine days after injection, with A5 still present, thus indicating that the inflammation was mainly induced by tissue damage from the columnar injection. At the latter point, the conductive structure started to degrade and lacked long-distance conductivity. However, no change in the redox state compared to normal tissue was observed in the degradation and bioresorption process.
After establishing the biocompatibility, geometry, and bioresorbable properties of the injected polymer, the conductive properties within the brain tissue were investigated. Fish with polymer were left to swim for up to seven days, after which the brains were excised, sectioned, and placed on interdigitated gold electrodes (Fig. 1E). In consecutive order, two interdigitated electrodes were biased (Keithley) and the current was registered. Although small, defined patterns of conductive polymer were visible, about 50 µm track widths and several hundred µm in length could easily be seen in bright field microscopy as dark regions. Enabling the detection of low currents reliably.
A5 has previously shown an excellent in vitro conductivity of more than 30 S cm− 1. 13 However, the conductivity is expected to be much lower in complex biological systems, such as the brain, where the conductive polymer can diffuse, be sequestered by endogenous mechanisms, removed by active transport, intercalated with endogenous chemical species, or combinations thereof. Further technical challenges added additional complexity, such as variabilities in Au electrode–tissue section contact resistance and poorly defined geometries. Still, we could clearly observe a linear current–voltage dependence between close-lying contacts (15 µm) on the brain slices, thus indicative of a highly conductive polymer. A5 presented currents in the range of a few nA under biases of up to 0.2 V (resistances of 50–100 MΩ). For brain regions without A5 or non-continuous A5, see the blue trace in Fig. 1F; as expected, we could not measure any electrical conductivity. We typically could not measure longer distance conductivities, making it difficult to extract a value for the conductivity. However, our resistance values compare favorably to a recently published study on genetically targeted in vivo conductive polymer assembly where Liu et al. used up to 100 V (we used 0.2 V) to measure polymers with a resistance of more than 10 GΩ. 6
Electrofunctionalizing A5.
Using the columnar injection technology gives essential control over the positioning and pattern of the soft electrode in the tissue, perhaps more so than the assembly controlled by the genetic expression of enzymes. To add modularity to the A5 electrode–tissue interface, an auxiliary method that enables flexibility in the electrode surface area, functionality, and further protrusion for seamless extension into the tissue was developed. Co-injecting the aforementioned ETE-R derivatives (e.g. ETE-S) with A5 would position the soft electrode at the targeted site. Because of the concentration gradient and the electrostatic repulsion between the negatively charged ETE-S and the negatively charged A5, the former diffused from the formed A5 electrode. After a delay, electropolymerization using A5 as an electrode attached the ETE-S to the surface of A5 and intercalated within the A5 backbone.
The ETE-R derivatives have significantly lower oxidation potentials than EDOT, ~ 0.3–0.5 V and 1.2 V vs. Ag/AgCl, respectively; the latter is close to the oxidation potential of water. 16 Thus, the electropolymerization of ETE-R derivatives is mild and not potentially toxic to tissues. In addition to increasing the surface area and tissue reach, different substituents (R) on the ETE-R derivatives would change the properties of the A5 electrode, as exemplified herein by ETE-S and the zwitterionic ETE-phosphatidylcholine (ETE-PC, Fig. 1A). The required high solubility of the injectable electrode imposes requirements on the nanoparticles forming the soft electrode, namely that they should be highly water-soluble and still be able to self-organize when injected into the tissue and thereafter bioresorbable without harming tissue. Thus, instead of redesigning the oligomers forming A5, this modular approach takes advantage of unique A5 properties and then adds soluble monomers in situ to customize the electrode–tissue interface.
The auxiliary A5 modular concept was evaluated in an in vitro agarose gel (0.5%) cast with Ringer’s solution, which mimics brain tissue. A5 (20 mg mL− 1) was dissolved in ETE-S or ETE-PC (40 mg mL− 1), without the formation of any precipitate. The dark solution was injected into agarose using a Hamilton syringe (Fig. 2A) where the A5 material instantly formed an aggregate. The diffusion of the ETE-R derivatives from the A5 injection track was monitored using UV light (365nm, Fig. 2B). After 2 h, ETE-R diffused from the A5 at a distance approximately twice the thickness of the A5 hydrogel electrode (Fig. S9). The A5 was then the connected electrode, and the ETE-Rs were electropolymerized at an applied bias of 1.5 V. This enabled electropolymerization, even with possible contact resistance. The applied bias can be further optimized, but it was selected for further use to have a high tolerance to variances in contact resistances during electropolymerization.
The thickness of the A5 electrode increased, confirming the successful polymerization of ETE-S (Fig. 2B). Further image analysis showed dendritic structures growing from the A5 core (Fig. 2C and Movie S6).The electropolymerized regions also displayed increased conductivity (Fig. 2D,E). With one contact on the A5 (and the other in the agarose) during electropolymerization, the comparably high resistance in agarose ensures a limited, constant current flow. Instead, by contacting both electrodes on the A5 during electropolymerization, the decreased resistance can be directly mapped as an increase in the current flow (Fig. 2D). The current increased by an order of magnitude during polymerization over a time course of 12 min. No change in the geometry, except for dendrite formation, of the A5 core was observed; the decreased resistance was explained by the polymerized ETE-S intercalating between the A5 nanoparticles. Although we optically observed electropolymerization (A5 darkening) when applying the voltage, we did not register a corresponding increase in current flow for the first 5 min. The initial incubation time could arise from one or more resistive bottlenecks that limit the current. Once these bottlenecks were removed by electropolymerization, a gradual current increase was observed as electropolymerization continued uniformly along the A5. Keeping both ends of the A5 contacted allowed for cyclic voltametric measurements which showed that electropolymerization increased the current 100 times (Fig. 2E).
A5–ETE-Rs electropolymerized in agarose was used as a basis for evaluating the mechanical properties (Fig. S7). The brain is very soft with shear modulus of about 0.5–1kPa17 which has been challenging to match for conventional inorganic electrodes. Electropolymerized A5–ETE-PC closely mimicked brain tissue with a static shear modulus of 0.57 ± 0.1kPa, see supplemental information for details.
The A5–ETE-S was electropolymerized in an agarose gel incorporating live cells to evaluate biocompatibility. We molded lung adenocarcinoma cells, the A549 cell line, into an agarose gel without (Fig. 2F) and with (Fig. 2G) contrast-enhancing DiI cell labeling. After injection of the A5–ETE-S solution, one end of the polymer electrode was contacted, and the grounded counter electrode was kept in agarose (outside the A5). During electropolymerization, the newly formed dendritic structures of ETE-S extended from the A5 and achieved cell contact and, in some cases, embedded the cells, creating close contact with the cells without any observable detrimental effects such as loss of cell integrity (Fig. 2F,G). A close connection between the electrode and cells has been shown to be necessary for efficient and precise low-voltage electrical stimulation–recording. 18 A5 and ETE-R toxicity was also evaluated in a limiting-dilution assay (Fig. 2H–J). Neither A5 nor ETE-PC showed cell toxicity after 1 day exposure in up to 1 mg mL− 1 A5 or ETE-PC. On the other hand, ETE-S showed some toxicity at high concentrations. The toxicity study covered a wide range reaching up to more than 1000x more ETE-R than we injected during in vivo experiments (around 200 µg vs. 400 ng), and a high dilution occurred when injecting into the tissue. For the quantities used in the in vivo setting, no cell toxicity from any of the compounds was observed. We summarize the in vitro experiments concluding that A5 mediated electropolymerization enabled flexible surface modification, close contact with cells, and significantly decreased electrical resistance.
Electropolymerization in vivo
Transferring the approach to an in vivo setting (Fig. 3A) puts severe constraints on the experimental settings: 1) small diameter injection capillary to avoid blood vessel rupture; 2) A5, ETE-R, and A5–ETE-R all need to be highly soluble and biocompatible; 3) applied voltages and currents used for the electropolymerization need to be kept low to not damage the brain tissue; and 4) the procedure needs to be quick to avoid anesthesia related damage. The 30 µm diameter capillaries, precoated with 50 nm iridium (see details in Supplemental), were used to inject an A5–ETE-S solution into the brain of anesthetized zebrafish (Fig. 3, S2A). After injection, the ETE-S was left to diffuse into the tissue for 1 min. The coated capillary was then used as the biased electrode to establish seamless contact with the injected A5 (Fig. S2A). Placing the counter electrode onto the zebrafish skin at the nostril allowed electropolymerization in the sedated fish with a low current (1–7 µA, Fig. S5), mimicking the agarose setup. The procedure, injection, and electropolymerization lasted about 10 min and after an additional 5–10 min, the fish were awake and displayed normal behavior. Typically, no fast darting or odd swim patterns were observed, nor were any difficulties with buoyancy and balance indicative of injury to the brain or stress behavior due to discomfort (e.g. pain) observed. We occasionally observed stress behaviors such as rolling or upside-down swimming in fish that required longer presurgery anesthesia than normal. However, this behavior was not linked to the injection per se. The lack of adverse events shows that the minimally invasive approach was sound and well tolerated by the fish.
When excising the brains, it was evident that the electropolymerization rendered a darker polymer from the addition of ETE-S. In sections from brains injected with A5 and ETE-S with a short electropolymerization duration (1 min), the dark polymer was primarily contained within a dense and discrete injection track surrounded by a halo of ETE-S monomer (Fig. 3B,C). When injection was followed by longer electropolymerization (5 min), the polymer contained a darker middle region, the A5, with polymerized ETE-S dendrites extending outwards in a radial fashion (Fig. 3D,E). The similarities to electropolymerization in agarose further corroborated the generality of the method and comparable results were also observed for ETE-PC (Fig. S3). The nonpolymerized ETE-S (seen by its green fluorescence) followed the injection track and extended outwards by about 50 µm without visibly affecting tissue integrity, as evident from nearby blood vessels with blood cells and neuronal cell bodies (Fig. 3E).
Histological staining of the A5–ETE-S containing sagittal brain sections (30 µm thick) showed close contact between cells and the polymer at the polymer–cell interface. Figure 3F–H, depicts an A5–ETE-S injection (dark blue) extending deep into the brain between corpus cerebelli (C) and optic tectum (OT) with ETE-S dendrites extending radially from the A5 reaching into the granular and molecular layers of C and superficial layer of OT. As in the in vitro cell–agarose model, the A5–ETE-S wrapped around the neurons, and some were even entirely surrounded by the conductive polymer electrode without any visible damage to the cells. This highlights the benefits of a gel-like microstructural electrode that allows for the exchange of metabolites and ions through the electrode to sustain cell homeostasis.
In the methylene blue–thionine staining (Fig. 3I,J), which also stains A5–ETE-S, seamless integration with no signs of tissue damage was observed of the A5–ETE-S being positioned between OT and C where the dendrites extend into the vagal lobes (VL). In Fig. 3K–N, A5–ETE-S is seen to extend dorsally into both the molecular and granular layer of C further highlighting the possibility of reaching any subregions of the brain.
After the A5 injections, an inflammatory response (oxidation of tissue) from the tissue damage mainly induced by the mechanical injection was observed, which was resolved after a few days (Fig. 1D). Introducing the auxiliary module, it was important to elucidate whether adding ETE-Rs to A5 followed by electropolymerization would cause damage to the tissue. As with the A5 injection, significant inflammatory signatures were found around the injection track within a few hours of the injection of both A5–ETE-S and A5–ETE-PC (Fig. 3O–V). Zebrafish that were left to swim with the polymers for days showed a significant decrease, with ETE-S after 7–9 days and complete removal with ETE-PC after 3 days, of inflammation. Thus, the applied voltage used in electropolymerization did not cause further tissue oxidation. Maybe not surprisingly, the A5–ETE-S had basically the same oxidation profile as A5 itself; they share the negatively charged sulfonate functionality. The zwitterionic phosphatidylcholine A5–ETE-PC showed a slightly better profile, and the oxidation was completely resolved after 3 days. Immunohistological staining did not show any signs of gliosis (Fig. S10). This corroborates that the phosphatidylcholine, as a constituent of cell membranes, has improved in vivo biocompatibility. 19 These data, that the electropolymerization of monomers (ETE-Rs) with low oxidation potential did not cause additional damage to the tissue combined with the normal behavior observed in the fish, strengthens the methodology as being minimally invasive and a suitable strategy to install soft conductive dendritic electrodes in brains.
A5–ETE-R electrical properties in zebrafish brains
The conductivity of A5 was improved by ETE-R functionalization (Fig. 4). For fish that underwent electropolymerization, more than 10 times higher currents (resistances of 5–10 MΩ) were observed under the same applied bias (for both ETE-S and ETE-PC). It was also possible to map conductivity over long electrode distances, allowing for the deduction of a value on the conductivity. Adjusting to the increased diameter of the conductive polymer upon electropolymerization, we estimated that the conductivity was around 3 S cm− 1 for both A5–ETE-S and A5–ETE-PC. This is 2 to 3 orders higher than most tissues (< 10− 2 S cm− 1).
Some fish were also left to swim around with the conductive polymer in the brain for 7 days, showing normal fish behavior (Fig. 4C,D). For A5–ETE-S and A5–ETE-PC, distinct polymers were clearly observed in the brains of two out of three fish but with lower conductivities than in the one-day experiments. Despite the clear polymer presence, only low currents with no clear voltage dependence were measured for the A5–ETE-S. This is due to the active bioresorption of some oligomers, thereby introducing high-resistance bottlenecks in the conductive structure. The A5–ETE-PC showed a linear voltage-dependent current in the low nA regime (resistance around 1 GΩ), thereby presenting a polymer still exhibiting long-distance conductivity. Interestingly, the higher 7-day conductivity of A5–ETE-PC indicates that this modification to the polymer rendered it more stable.
A5-mediated brain stimulation
After establishing the possibility of installing and modifying bioresorbable electrodes in vivo, attention turned to using them to actively modulate the electrical activity in live brains. After injecting the mixture of A5 and ETE-S, followed by electropolymerization, the fish were left to swim for one day and then sacrificed. Brains were excised and vibratome sectioned, resulting in 300–400 µm brain slices containing the A5–ETE-S electrodes (Fig. 5). The procedure kept the brain slices viable with functional neuronal signaling. For these experiments, we used adult zebrafish Casper mutants (Tg(elav3:GCaMP6f)) to visualize, by GFP fluorescence, increases in intracellular Ca2+ in neurons. 20 Action potential firing was seen through an increase in green fluorescence, which is a convenient and non-invasive way to track neural activity.
We used a thin tungsten microelectrode connected to a Keithley sourcemeter to contact and stimulate the A5–ETE-S in the brain slice (Fig. 5B). Train sequences of square voltage pulses were applied to modulate the electric activity in the brain slice with settings like those used previously. 21,22 Within a second after turning on the electrical stimulation, a response was observed in which regions close to the A5–ETE-S lit up (Fig. 5C, Movies S2 and S3). The external voltage pulses were applied only for 5 s to initiate a signaling cascade. The signaling cascade lasted up to 20 s and then reverted to the initial (prestimulation) situation (Fig. 5F). The slices were then left to recover for a couple of minutes before new pulse trains were applied to initiate new signaling (Fig. 5G–J). As the brain slices were used for stimulation for more than 1 h, we noticed a general decrease both in fluorescence intensity and in the extension of signaling. Additionally, as the slices were re-stimulated, we saw a slower onset, however still starting within the first few seconds of stimulation.
Interestingly, a response from the tissue was observed in several regions, all located around the polymer and extending far away from the microelectrode that was in contact with the polymer. Reference brains without A5–ETE-S did not show similar long-distance stimulation effects, showing that the conductive polymer transmits the electrical pulses to distant regions. After several rounds of stimulation, an increased spontaneous electrical activity was typically observed within the brain slices. The spontaneous activity most often appeared in the previously stimulated regions but had a different appearance in smaller, less intense firing with a stochastic appearance across the slice (Movie S3). In comparison, the externally induced excitations were more pronounced and synchronized with the onset of the applied voltage pulses.
The excited regions, mainly centered around the cerebellum and torus longitudinalis, were separated by regions without any increased neural activity, indicating the possibility of targeting specific regions in the brain. This was confirmed by adding the GABA-A receptor antagonist pentylenetetrazole (PTZ) to the imaging medium (Fig. S4). PTZ stimulates unspecific neural activity and upon addition, many more regions in the brain slices initiated neuronal firing. These regions were not close enough to the soft electrodes to be stimulated by the applied electric fields. The repeated cycling shows that stimulation was not acutely toxic to the cells, and the regional specificity highlights that the method can be implemented in an in vivo setting.
Leaving part of the microcapillary inside the brain after injection allowed to establish external contact with the soft electrode (Fig. 5K–M, S3, Movie S4). The fish were left to swim around with the capillary for one day (Movie S5) before the brains were excised, sectioned with the capillary remaining in the slices, and analyzed in the electrical setup. The stimulating electrode was positioned on the capillary outside of the brain rather than directly on the polymer in the slice (Fig. 5O). This configuration offers a solution to how one could contact the polymer electrodes in the brains of live fish. When adding the stimulating pulse trains, the same regions became excited as when the A5–ETE-S itself was contacted. This experiment proves that not only can we use the conductive polymer to alter brain activity, but it was also possible to contact it outside of the brain, which is a critical need for most clinical applications. Performing the same workflow in freshly excised mouse brains (Fig. S6) demonstrated that the methodologies could be directly transferred to larger animals without any alteration to the protocol.
Herein, we have presented a novel technology that establishes functional and well-tolerated organic bioelectronics in the brain. To meet the demand for transient bioelectronic therapies, implantation was performed using a minimally invasive injection technique, and the formed structures were bioresorbable. The latter is a desired property in, for example, electrotherapeutic cancer treatments, making revision surgery obsolete. This was made possible by the discovery of thiophene oligomers (A5) that form nanoparticles. These nanoparticles are highly water-dispersed, which makes it possible to have them in high concentrations in a solution without aggregation. However, by injecting them into the tissue and following interaction with endogenous ions, a stable soft electrode is formed. Thus, no specific triggers are necessary. It was also possible to ion-match the nanoparticle solution with endogenous ion strengths in quite different tissues, as shown by establishing conductive structures in the zebrafish caudal fin and brain. This makes the technology general to several tissues and cross-species. Also, because the nanoparticles comprise oligomers that are the size of conventional drugs; they are bioresorbable.
Instead of redesigning the oligomers forming A5 to generate functional flexibility at the electrode–tissue interface, the difficult-to-copy unique properties of A5 inspired us to develop an auxiliary modular methodology. This modular approach takes advantage of A5’s properties and then adds soluble monomers (ETE-Rs) that attach to A5 in situ to customize the electrode–tissue interface. The electropolymerization of ETE-Rs with low oxidation potential in situ increased conductivity, formed a close connection to cells and established functional group modifications to the A5 electrodes. This was demonstrated using both ETE-S and ETE-PC, where the latter had higher long-term stability and less toxicity. Furthermore, we showed that electropolymerization did not cause additional oxidation damage to brain tissue. This modular approach opens up for ETE-Rs with different functional groups and potentially other monomers with low oxidation potential.
Despite the small zebrafish brain, we solved the critical challenge of contacting soft neural electrodes to allow efficient external interaction. With external contact, neuronal signaling was modulated in live brain slices—excised from fish with implanted bioelectronics—by applying electrical pulses.
The methodologies and workflows presented here are general and not confined to zebrafish (Fig. S6). A reasonable speculation is that the procedures would be more straightforward with larger brains, e.g. rodents and primates, especially external connections.
In summary, these findings initiate the discovery of in vivo assembled, fully integrated, bioresorbable electronics within nervous systems and other tissues that can be used for nonchronic treatments.