Multifunctional bioelectronics for brain-body circuits

Abstract

The brain continuously receives, integrates, and responds to an influx of sensory signals emerging from the internal organs. This is mediated not only through direct neuronal connections defined by the peripheral nervous system, but also endocrine, humoral, metabolic, and immune pathways. Despite being predominantly imperceptible, the complex brain-body cross-talk is essential to maintaining physiological homeostasis. Moreover, it is increasingly recognized to play a critical role in cognitive and behavioral functions as well as in disorders of the nervous system. The functional and anatomical diversity of brain-body pathways necessitates the development of multifunctional implantable neurotechnologies that can facilitate causal studies during behavior. Although ubiquitous in studies of brain function, electrical, optical, and chemical interrogation of organ-brain circuits remains a challenge. In this review, we discuss recent developments in multifunctional implantable neurotechnologies, highlighting material selection, device architectures, integration challenges, and power and data transfer approaches necessary to establish robust bioelectronic interfaces to brain and peripheral organs suitable for long-term studies of brain-body signaling.

Introduction

The bidirectional crosstalk between the central and peripheral nervous systems (CNS and PNS) allows us to perceive, interpret, and respond to the visual, auditory, olfactory, gustatory, and tactile stimuli. This environmental sensory awareness is termed as exteroception. Besides the consciously perceived signals, the brain and the spinal cord are bombarded by subliminal sensory cues from visceral organs of the body. Even though most interoceptive signals are not consciously detected, these interoceptive signals are critical for maintaining homeostasis and constructing models of internal bodily state^1–3. Aberrations in bidirectional brain-body communication have been linked to neurological, psychiatric, metabolic, and neuroimmune disorders^4–6. Thus, understanding neural circuits, cell types, receptors, and signaling molecules that mediate brain-body crosstalk is essential to uncovering the influences of organ physiology on brain function and vice-versa⁷.

Despite the scientific and clinical significance, causal studies of brain-body circuits in freely-behaving model organisms remain challenging. Electrophysiological recording of neuronal activity, imaging with fluorescent indicators, and optogenetic neuromodulation have empowered studies of brain circuits⁸. However, the application of these methods to peripheral organs and neural innervation has remained limited. This is, in part, due to technological complexity; unlike the brain which boasts significant functional redundancy, benefits from an immunologically privileged status, undergoes microscale motion and is surrounded by the skull, the peripheral organs are delicate, mobile, difficult to access surgically, and subject to extensive immunosurveillance. These differences in anatomy and function preclude direct deployment of brain neurotechnology in the PNS, and thus innovations in neural interfaces are urgently needed to extend powerful neuroscience approaches to enable multi-organ, multi-region studies of brain-body communication.

In this review we discuss the innovations in multifunctional neurotechnologies in light of their design criteria, device architectures, material choices, and powering/communication schemes that are prerequisite to developing long-term bioelectronic interfaces for the brain, spinal cord, peripheral nerves, and organs such as the gastrointestinal (GI) tract, bladder, and the heart. Our primary focus will be on technologies for neuroscience research in animal models, with a limited discussion of translational bioelectronics. Furthermore, we predominantly highlight tools suitable for experiments in behaving subjects, with minimal consideration of proof-of-principle validations in terminal preparations. We also discuss engineering and biological failure modes of devices wherever applicable.

Although much research in the fields of neural engineering and bioelectronics is dedicated to expanding device capabilities and improving biointegration with the target organs, the device utility is critically dependent on its backend – the electronics that enable the data and power transmission between the implanted interface and external control hardware. While rarely discussed in academic literature, the systems-level implementation of multifunctional neurotechnology is often a critical deciding factor for its adoption in fundamental research.

The organization of this review is as follows: 1) we first discuss challenges of designing devices for brain and visceral organs with a focus on neuroanatomy, organ physiology, materials selection, and device mechanics; 2) second, we consider pros and cons of different neural interrogation modalities, and the need for multifunctional devices; 3) third, we highlight recent studies that demonstrate multifunctional devices capable of recording and modulation of brain functions; 4) we then focus on devices for visceral organs and PNS with specific examples of spinal cord, peripheral nerve, heart, bladder, and GI tract interfaces; 5) in Box-1 we spotlight challenges of systems-level integration; 6) and finally, we offer an outlook on the field of multifunctional neural interfaces and highlight the technological and translational (Box-2) opportunities that can advance the understanding of brain-body signaling in health and disease.

Box-1: System Integration.

An essential aspect of multifunctional device integration involves the design and implementation of backend configurations, which can be either tethered or wireless. Tethered configurations provide a robust connection between the implanted interface and external hardware and high data rate capacity but come with limitations such as compromised animal mobility and potential tissue damage. These systems are challenging for mobile organs like the spinal cord and peripheral nerves because they need to be fixed to stable body regions, and the forces exerted by tethers can cause tissue damage, compromising chronic stability and signal fidelity. Conversely, wireless configurations offer greater freedom of movement by making possible fully concealed implants, but may face challenges related to power consumption, signal fidelity, and signal interference¹⁷³.

Wireless power transfer.

Remote power transmission technologies are based on ultrasonic, capacitive, optical, radio frequency, and inductive links. Among them, resonant inductive coupling is well-established in clinical and pre-clinical use due to resonant circuits that enhance coupling and power transfer efficiencies between external and implanted coils¹⁷⁴. Magnetoelectric devices, which transduce mechanical strain generated in a magnetostrictive layer under an external alternating magnetic field into a voltage output from a coupled piezoelectric layer, have also emerged as a promising modality for power transfer. Besides electromagnetic fields, power delivery through ultrasound waves has been shown to bidirectionally control miniature peripheral nerve interfaces¹⁷⁵ as well penetrating deep-brain electrodes¹⁷⁶. In these platforms, ultrasound energy is transduced into a voltage output by a piezoelectric material. While miniaturization of the implantable device owing to the significantly smaller wavelength of sound waves in tissue is an obvious advantage of this method, it is currently impeded by scattering of ultrasound at tissue interfaces and poor tolerance to source-antennae misalignment. In Table 2 we summarize the key characteristics of the most common power transfer methods.

Data transmission.

Wireless data transmission methods include infrared-encoded signals, which provide high data transfer rates at relatively low powers, but are prone to interference and require direct line-of-sight with the receiver¹⁷⁷. Bluetooth radios (~2.4 GHz) offer simple and robust pairing¹⁷⁸, but have a limited bandwidth of ~ <10 Mbps and require higher power usage. Near field communication (NFC) offers low power communication but is restricted by short range and lower data transfer rates¹⁷⁹. Data communication has also been implemented with ultrasound backscattered signals as it provides better tissue penetration than RF but is affected by tissue inhomogeneity and mechanical movements¹⁷⁵.

Interconnects.

Interconnects are typically flexible or stretchable, with a dielectric insulating substrate that maintains low leakage current, allowing the transmission of electrical (or optical) signals without any disruption. Additionally, they can incorporate microfluidic channels, making them suitable for both electrical communication and drug delivery.²⁵ Flexible ribbon cables employing polymer substrates maintain low impedance, which is crucial for an optimal SNR, without significantly increasing the device size. While flexible cables are advantageous for their conformability, they lack the inherent ability to withstand significant deformation, risking fracture of thin-film metal traces during substrate deformation. Serpentines and wavy-buckled films deliver additional stretchability although they may experience microcracks when repetitively strained^180,181. The use of nanostructures in interconnects has enabled high conductivity even after repeated elongation. Examples include microcracked gold interconnects on elastomer substrates allowing substantial stretching without fatigue³² and interfacial self-assembly of metal nanowire and elastomer composite inks that generate highly conductive and stretchable nanomembranes ¹⁸².

Box-2: Translational considerations.

Co-opting body’s intrinsic sensory and regulatory neural circuits is poised to deliver novel non-pharmacological treatments for visceral conditions and nervous system disorders. In this context, the multifunctional brain-organ neurotechnology can play two roles: (1) Discovery of neural circuits that can be co-opted for therapeutic interventions in relevant disease models; and (2) translation of optimized therapies to translational animal models and humans.

The devices covered in this review are designed to improve our understanding of brain-body neurophysiology in pre-clinical rodent models and pave the way for their translational applications. A current challenge in realizing devices that are translation friendly is incorporation of robust physical and biochemical recording capabilities that can allow longitudinal experiments to help identify disease biomarkers, decode the underlying neural activity, and formulate approaches for personalized engineered medicines¹⁸³. While the challenges associated with electrical recording and stimulation of neural activity are well-documented^184,185, it is equally critical to address the obstacles to translation of optogenetics, optical imaging, and local chemical neuromodulation. Optogenetics and optical imaging, in particular, face significant barriers owing to the need for safe and effective gene delivery, as well as the invasive implantation of light sources. Similarly, for intracranial pharmacology the implantation of cannulas can lead to complications including tissue fibrosis and blockage of the microfluidic channels. These issues not only highlight the difficulties in translating the advanced research techniques to practical and safe therapeutic interventions, but also present opportunities for collaborative innovation.

Translational brain-body neurotechnology holds immense promise for human health. However, unlike many other frontier technologies, it has direct access to the neural data with the ability to “read” and “write” activity. In combination with increasingly powerful computational approaches, these devices may raise a range of ethical issues pertaining to person’s agency and security. Moreover, neurotechnology that goes a step beyond treating disorders and augments human capabilities can spark additional concerns over equitable access. It thus imperative for stake holders including patients, researchers, clinicians, businesses, lawyers, and regulatory bodies to establish fundamental neuro-ethics principles and guarantee neuro-rights.

Challenges of designing devices for brain and visceral organs

Anatomical and physiological considerations

Modulating and recording neuronal activity in the brain and peripheral organs poses distinct anatomical and engineering challenges to device design. In both cases, it is desirable to achieve high spatiotemporal resolution and multimodality, while ensuring a chronically stable interface with minimal disruption to the underlying tissue⁹. Beyond this, morphology, tissue properties, vasculature, cell organization, and immune response play crucial roles in determining the device architecture. Below we delineate some of the most salient differences between brain and visceral organ anatomy/physiology that inform device design (Fig. 1, Table 1).

Figure 1. Mechanical and anatomical features of the brain and organ systems to guide device design.

Overview of mechanical and anatomical features influencing probe design for the brain^23,208 and peripheral organs (spinal cord^23,209, heart^{189,210–212}, stomach²¹³, intestine^214,215, bladder^216,217). Adjacent panels provide details on tissue mechanics, immune response types, innervation patterns, the nature of relative motion between devices and organs, and organ volume expansion and contraction. The data reported in the figure are related to human species. DRG: dorsal root ganglia.

Table 1.

Anatomical considerations for brain and peripheral organs

Organ	Species	Physical properties	Young’s modulus (E)	Nature of stress between device and organ
Brain	Human	• Surface area ~ 2500 cm223 • Weight: 1500 g186	E= 0.1-10 kPa23	Bending
	Rat	• Surface area ~ 6 cm223 • Weight: 1.8 g23
Spinal Cord	Human	• Length: 45 cm187 • Diameter: 6.4 mm (thoracic area) 13 mm (cervical and lumbar area)187	E= 0.1-10 kPa23	Bending, Stretching, Buckling
	Mouse	• Length: 25-35 mm188 • Diameter: 1-3 mm188
Heart	Human	• Length: 12 cm189 • Width: 8.5 cm189	E= 10-35 kPa190,191	Stretching
	Mouse	• Length: 10 mm192 • Width: 4.15 mm192
Stomach	Human	• Length: 20 cm193 • Width: 15 cm (at widest point)193	E= 1.9 kPa194	Stretching
	Rat	• Surface area195: 6.2 cm2 • Weight: 3.90-8.50 g196
Small Intestine	Human	• Length: 7 m197 • Diameter: 5 cm193	E= 20-40 kPa190	Bending, Stretching, Buckling
	Mouse	• Length: 35 cm197 • Diameter: 0.3-0.5 cm193
Bladder	Human	• Length: 13 cm198 • Width: 8 cm (at typical capacity)198	E= 0.25-1 kPa199	Stretching
	Rat	• Surface area ~ 82 mm2200 • Weight: 70 mg (when empty)200

The brain forms a densely packed, three-dimensional network of neurons. For instance, non-human primate (NHP) visual cortex (V1) exhibits a neuronal density between 99-176 million neurons/gram (~ 21 million neurons per cm² of cortical surface)¹⁰, demanding high-resolution neural interfaces with electrode dimensions commensurate with neuronal soma^11–13. Therefore, targeting large number of individual neurons becomes feasible with dense, penetrating electrode arrays tailored to match the typical size and density of cell bodies. Conversely, innervation in peripheral organs is much sparser and neurons often reside within thin sheets of plexi surrounded by epithelial or muscle layers. For instance, in the GI tract the enteric nervous system (ENS) is distributed concentrically around the gut lumen within myenteric and submucosal plexi¹⁴. As such, interfacing with the ENS requires thin, planar devices. Peripheral nerves such as the vagus or sacral nerves are miniature cylindrical information highways that split into branched networks throughout the body. Such anatomical organization often demands nerve interfaces with a cuff architecture. The network of peripheral nerves itself originates from sub-mm scale (in rodents) peripheral neuronal ganglia, which are distributed across the brainstem, spinal cord, and visceral sites. As such, the devices interfacing with peripheral ganglia require extreme miniaturization and mechanical softness while densely integrating with recording and stimulation functions.

The brain exhibits a reduced tissue reaction, with less pronounced inflammation and fibrosis compared to peripheral organs. This property is an evolutionary adaptation bestowed on vital organs with limited regenerative capacity that protects them from accidental self-inflicted damage by the immune system¹⁵. This unique environment permits long-term implantation of devices, such as deep brain stimulation electrodes, with mechanical properties vastly different from those of neural tissue¹⁶. Conversely, peripheral organs may mount a significant foreign body response, which imposes stricter requirements on device mechanics to minimize inflammation and overgrowth of a fibrous layer that may hinder device utility. Regardless of the CNS or PNS target, it is imperative that the tissue interfacing materials are biochemically inert and resistant to corrosion in physiological environments.

The brain experiences microscale motion (see Fig. 1) triggered by breathing, locomotion, and vascular pulsation, leading to relative periodic (1-4 Hz) displacement of the implanted device that is anchored to the skull¹⁷. In contrast, visceral organs such as the GI tract experience macroscale deformation, with peristalsis causing movements of several millimeters to centimeters (3-5 cycles per minute in humans)^18,19, and the heart undergoes a volumetric change of 8-9% during a typical cardiac cycle²⁰. While a brain probe that fails to accommodate microscale tissue motion can lead to early functional failure, a mechanically mismatched device in a peripheral organ may result in morbidity. Moreover, unlike the skull-encased brain, peripheral organs rarely have convenient sites to secure the probes. This necessitates additional mechanical engagement features such as suture eyelets, cuffs, or bio-adhesion layers on such devices^21,22.

Materials and mechanics considerations

Biocompatibility, stiffness, and ease of implantation of bioelectronic interfaces are influenced by the properties of the constituent materials that make up the device. Selecting a substrate to host the functional features of the probe requires consideration of the target organ anatomy, neuronal density, intended duration of implantation, and the desired use (recording, modulation) or interface modalities (electrical, optical, chemical). Particularly for chronic implants, it is important that the substrate does not trigger inflammatory response and maintains a stable atraumatic interface with tissue over time^23–25. Substrate requirements differ significantly for brain and peripheral organs. Despite their mechanical mismatch with brain tissue and the corresponding propensity for glial scarring, rigid substrates such as metal²⁶, glass²⁷, and silicon^13,28 remain extensively employed in brain interfaces owing to their mechanical strength, fluid barrier properties, and compatibility with established manufacturing methods. However, such rigid substrates are unsuitable for organs that undergo growth, deformation, and movement, such as the spinal cord or the bladder. This necessitates interfaces based on flexible, low-modulus plastics (e.g. polyimide (PI), polyetherimide (PEI), polycarbonate (PC), poly (methyl methacrylate) (PMMA), SU-8, parylene-C)^29–31 or elastomers (e.g. silicones, styrene-ethylene-butylene-styrene (SEBS), cyclic olefin copolymer elastomer (ECOC))^32–36. These flexible substrates additionally offer compliant alternatives for brain interfaces, aiming to reduce foreign-body response by minimizing mechanical mismatch with the soft brain tissue⁹. Most of these materials are compatible with traditional photolithographic processing^37,38, and some permit thermal drawing into monolithic fiber-based devices^39,40. Parylene-C and PC have also been deployed in clinical devices approved by the United States (US) Food and Drug Administration (FDA)^41,42, and their transparency in the visible range enables applications within imaging or optical neuromodulation platforms⁴³. Parylene-C can be processed at sub-micrometer thicknesses enabling conformal interfaces to anatomically challenging targets. Despite these advantages, flexible substrates still exhibit elastic moduli in the ~GPa range, whereas neural tissue is viscoelastic with moduli in the ~kPa range^23,44. To better match the mechanical properties of brain tissue, recent studies have leveraged soft and stretchable elastomers (Young’s modulus (E) = 1-5 MPa), and hydrogels (E = 0.1 Pa to 60 kPa)^32,45,46. However, ultra-low modulus materials create an insertion challenge which can be quantitatively understood from the Euler’s equation. The maximum amount of force, F_m, that a rectangular neural implant can endure before buckling during tissue penetration is given by $F_m=\frac{{\pi }^{2}Eb{t}^3}{12l^2}$; where E is the elastic modulus, and b, t, and l are the implant width, thickness, and length, respectively⁴⁷. Thus, softer probes buckle at the tissue interface and hence require temporary stiffening or guide shuttles for targeted deployment. Moreover, although promising in short-term studies, the use of soft materials as dielectrics over extended timescales remains impeded by their limited biofluid barrier (a dielectric coating around the device that protects the embedded electronics from the surrounding biological fluid.) properties ⁴⁴.

Although beneficial for brain interfaces, stretchable platforms are obligatory for chronically stable bioelectronic devices for the viscera. Such devices can exploit intrinsically stretchable elastomeric conductors, semiconductors, and insulators either in the form of nanocomposites or blends of conductive/semiconducting/insulating polymers with tunable cross-linking density or degree-of-conjugation⁴⁸. A complementary approach relies on inducing mechanical stretchability into thin-films of high-modulus materials through engineering of serpentines, out-of-plane wavy structures, and kirigami cuts⁴⁹. This approach decouples mechanics from function and allows integration of a wide range of polymers, inorganic semiconductors, and metals into thin, stretchable multifunctional devices.

Neural interface modalities and the need for multiple functions

Investigating neural circuits underlying complex behaviors demands tools capable of probing and modulating neural activity at scales ranging from individual cells to networks spanning the brain and visceral organs^50,51. Chemical, electrical, and optical approaches offer unique advantages for interrogating neural signaling with different degrees of biological specificity, spatial resolution, and temporal precision (Fig. 2) ^52,53. Below, we first describe individual neural interface modalities and then discuss the biological motivation for multimodal neural probes.

Figure 2. Overview of common neural interface modalities.

(a) Pharmacology involves direct infusion of drugs into the brain, that otherwise cannot cross the blood-brain barrier⁵⁴. While important clinically, this approach of neuromodulation suffers from poor spatiotemporal precision; (b) Spatially targeted chemogenetics involves delivery of a designer drug through an implanted cannula that targets cells which are genetically modified to express designer receptors⁵⁵. This approach allows activation or inhibition of genetically well-defined neuronal populations; (c) Electrical stimulation allows delivery of biphasic current directly into the brain tissue through an implanted bipolar electrode⁵⁹. While this approach of neuromodulation is used clinically, it cannot target genetically homogenous cellular population and shows off-target effects; (d) Optogenetics allows spatiotemporally precise activation or inhibition of genetically well-defined cellular populations through expression of light-sensitive bacterial opsins in neurons^64,65; (e) Electrophysiology permits recording of neuronal activity with high spatiotemporal resolution^74,75; (f) Photometry takes advantage of genetically encoded activity indicators in neurons to record calcium transients (proxy for neuronal activity) or neurotransmitter/neuromodulator dynamics with cell-type specificity⁹⁰.

Chemical neuromodulation enables the study of receptor contributions to physiology and behavior, and is indispensable for treatment of neurological and psychiatric disorders⁵⁴. Given the overlapping receptor profiles across multiple cell types, systemic delivery of drugs may produce side effects or confound data interpretation (Fig. 2a). Chemogenetics overcomes this challenge in model organisms by introducing transgenes for designer receptors exclusively activated by designer drugs (DREADDs) to specific cells (Fig. 2b)⁵⁵. To overcome the poor spatial and temporal resolution of systemically administered drugs, targeted pharmacology in the CNS has leveraged implantable cannulas and external micropumps⁵³. Larger volumes of tissue can also be accessed through convection-enhanced delivery at high-pressure⁵⁶. Alternatively, to avoid tissue displacement caused by high-pressure injection, ion pumps leverage electrophoresis to deliver drugs without solvents⁵⁷. Wireless head-mounted pumps outfitted with miniature replaceable drug reservoirs enable precise and repeatable drug delivery in moving subjects⁵⁸. Despite these innovations, chemical and chemogenetic neuromodulation operate on timescales ranging from seconds, when delivered locally, to minutes when administered systemically⁵³.

Electrical stimulation offers a temporally precise alternative to pharmacology in clinical practice and translational research (Fig. 2c). To deliver currents sufficient to modulate neural activity at voltages within the electrochemical stability window of aqueous electrolytes (1.23 V v/s standard hydrogen electrode), stimulation electrodes must possess low impedance, high charge storage, and charge injection capacities⁵⁹. Despite their comparatively poor charge injection properties, noble metals such as platinum and platinum-iridium alloys are commonly used in clinic due to their biochemical inertness⁸. Surface coatings of iridium oxide or titanium nitride enhance the performance of noble metal electrodes⁶⁰, whereas conducting polymers, including poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS), offer superior charge injection due to their combined ionic and electronic conductance. Similarly, carbon nanotubes, nanofibers, and porous graphene offer electrochemically inert, capacitive interfaces with large effective surface areas^60,61. To reduce mechanical mismatch with neural tissue, electrically conductive hydrogels, conductive nanomaterial-doped elastomer composites, and noble metal nanomembranes have also been explored for their ability to maintain electrical properties under strain^32,62.

Although electrical stimulation is spatially and temporally precise, it does not discriminate between electrogenic cell types⁶³. Optogenetics grants bi-directional control over specific neuronal populations by endowing them with light-sensitive proteins engineered from microbial rhodopsins (Fig. 2d)^64,65. Opsin spectra have been tuned to span the UV and visible, and far-red range, while their kinetics enable neuronal activation/inhibition at scales ranging from milliseconds to minutes^66–68. Although, recent advances in ultrasensitive opsins have permitted neuromodulation via light sources positioned on the thinned skull or the skin of mice⁶⁹, the majority of neuroscience experiments still rely on implanted light-emitting devices (LEDs) or optical fibers coupled to external lasers for light delivery. Microscale LEDs (μLEDs) are commonly based on rigid III-V semiconductors (e.g Gallium nitride (GaN) and alloys) and must be heterogeneously integrated into neural interfaces^70,71. Organic LEDs (OLEDs) present a promising alternative for intrinsically flexible probes, but the limited stability and efficiency of organic emitters in physiological conditions have so far precluded their broad adoption in chronically deployed devices⁷². Although optogenetics offers unmatched cell-type and temporal precision, its reliance on genetic manipulation has limited its application in translational studies, with notable recent exceptions in vision restoration^65,73.

In addition to neuromodulation, recording of neuronal activity is indispensable for studies of circuits underpinning behavior and physiology, as well as for identifying biomarkers of pathology. Electrophysiological recording devices convert ionic currents into electrical signals, enabling extracellular detection of individual neuron action potentials (spikes) and local field potentials from groups of cells (Fig. 2e)^74,75. Electrochemical impedance spectra are key determinants of the electrodes’ efficiency in converting ionic to electronic signals, and the impedance at 1 kHz is often used to compare electrode platforms as this frequency corresponds to millisecond time scales of action potentials. Passive recording can be performed with essentially any conductive microwires as shown in groundbreaking studies by Hubel & Wiesel⁷⁶ and Strumwasser⁷⁷. To date, recording electrodes have been produced from metals (iridium, platinum, tungsten, nickel-chromium, and gold), graphene, MXenes, and conductive polymer gels^46,78,79. Surface treatments ranging from nanoscale surface area enlargement to coatings with PEDOT:PSS and its blends reduce electrode impedance and improve signal-to-noise ratio (SNR) of the electrophysiological recordings⁸⁰. Implementation of active amplification at each electrode and multiplexed switching between channels reduces the interconnect burden and increases the SNR. Probes based on complementary metal-oxide-semiconductor (CMOS) silicon technology, such as Neuropixels, permit on-probe signal amplification and deliver 1000s of electrodes by embedding active electronics beneath the contacts²⁸. The silicon electronics can also be made flexible using silicon nanomembrane transistors integrated into electrode arrays, thereby increasing their biocompatibility as surface probes while featuring thousands of amplified and multiplexed sensors⁸¹. Despite these advances, CMOS-based probes still rely on passive transduction of ionic to electronic potentials. Conductive polymers with mixed ionic and electronic conductivity give rise to the organic electrochemical transistors (OECTs) that transduce ionic signals into electrical ones, while simultaneously providing local signal amplification^82,83. Fully flexible active multiplexers can be created by combining OECTs with organic electrochemical diodes or field-effect transistors, which act as switches in high-resolution brain interfaces^84,85. Although temporally-precise, electrophysiological recordings provide limited information on neuronal types and demand ultra-high density arrays coupled with sophisticated algorithms to isolate single-neuron voltages from other types of physiological potentials or noise⁸⁶.

Optical imaging with genetically encoded fluorescent reporters of calcium, voltage, and neurotransmitters reveals spatially-resolved signaling in specific cell populations, albeit at limited temporal resolution⁸⁷. Genetically encoded indicators are commonly based on chimeras of permuted fluorescent proteins and sensing domains. Upon binding to an analyte, the quantum yield of these indicators increases manifesting in fluorescent transients⁸⁸. Traditionally, optical imaging of neuronal activity relied on single- or multiphoton microscopy through objective lenses positioned above transparent windows over tissue of interest in head-fixed or anesthetized subjects⁸⁹. Fiber-photometry (Fig. 2f) relies on capture of the fluorescence through the same implanted fiber as used to deliver excitation light. This approach is compatible with experiments in moving subjects, but sacrifices the spatial information, summing the signals from all cells within the numerical aperture of the fiber⁹⁰. Micro-endoscopes rely on gradient index (GRIN) lenses to preserve spatial information and grant access to deeper brain structures. Although enabling, these devices remain bulky, costly, and pose significant impact on tissue due to the mm-scale dimensions of implanted glass lenses⁹¹. More recently CMOS-based lens-less subdural imagers achieved wide fields of view and high spatiotemporal resolution at a depth of 100 μm below the cortical surface⁹².⁹³

Measuring neurochemical dynamics is key to understanding synaptic communication and long-range modulation via diffusing neurotransmitters/neuromodulators. While powerful in rodent models, optical imaging with genetically-encoded neurochemical indicators is currently not suitable for translational studies⁹³. The most universal neurotransmitter sensing tool is microdialysis, which relies on fluid extraction followed by spectroscopic benchtop analysis, offering high accuracy but poor temporal precision⁹⁴. Electrochemical techniques such as amperometry and fast scan cyclic voltammetry apply voltage to drive oxidation/reduction of analytes at electrode surfaces and then detect changes in electrical currents. These methods offer high temporal precision, but are limited to detection of redox active neurotransmitters, such as dopamine, norepinephrine, and serotonin⁹⁵. Voltametric and amperometric sensing demands electrodes with high electrochemical stability within the water-window and low impedance – properties found in carbon electrodes including glassy carbon, carbon paste, screen-printed carbon, and carbon fiber^96,97. More recently, carbon nanotubes, graphene, and nanodiamonds have been used to improve detection through increased surface-area and faster charge-transfer⁹⁵. MXene-based electrodes, known for their conductivity and large surface area, showed promise in detecting neurotransmitters, but demand further optimization to eliminate fluoride residues that can affect the electrochemical performance⁹⁸.

The complexity of chemical and electrical signaling in neuronal circuits across the CNS and the viscera motivates the combinatorial deployment of the diverse recording and modulation technologies described above. Such multifaceted approach enhances the precision of neural interfaces and permits causal rather than correlative associations in the neurophysiology underlying behavior and disease⁹⁹. Additionally, integrating multiple neural interface modalities may permit future development of precise and patient-specific closed-loop therapies for neurological and psychiatric conditions.

However, the disparate design criteria of different neural interrogation modalities pose formidable challenges to their co-integration into biocompatible platforms for long-term use in moving subjects or human patients. Below we discuss multifunctional probes that combine different modulation and recording functions to investigate the brain, spinal cord, peripheral nerve, and visceral organ physiology.

Multifunctional probes for the brain

Multimodal devices for electrical stimulation and electrical recording

Electrical stimulation and recording are routinely combined in translational studies using a variety of electrode platforms as most electrodes appropriate for neuromodulation are in principle suitable for neural recording. This dual functionality allows for the decoding of behavioral or physiological information by simultaneously stimulating specific neural circuits and recording the resulting changes in electrical activity. This is particularly valuable in the context of diagnosing and managing neurological conditions such as epilepsy, where recording the effects of activating or inhibiting potential seizure foci can guide resection surgeries. In addition, the ability to adjust stimulation parameters based on real-time electrophysiological feedback is essential for closed-loop therapies including those explored for Parkinson’s disease and epilepsy.

The key challenge to combining electrical recording and stimulation are the stimulation artifacts that overwhelm the recorded signals. This problem is elegantly solved via circuits that employ on-board processing to minimize stimulation artifacts, compute neural biomarkers, and regulate stimulation parameters. Equipped with a low-power bidirectional wireless link, this platform permitted closed-loop recording and stimulation in non-human primates (Fig. 3a)¹⁰⁰. Alternatively, achieving concurrent epicortical recording and intracortical stimulation can be implemented by employing multiple standalone submillimeter-sized silicon circuits (0.1 mm³), Neurograins (Fig. 3b), each equipped with independent radio-frequency telemetry¹⁰¹.

Figure 3. Multifunctional probes for the brain.

Multimodal devices for electrical stimulation and electrical recording (a) Three-dimensional design model of wireless and artifact-free 128-channel neuromodulation device (WAND) for continuous neural biomarker monitoring and closed-loop stimulation¹⁰⁰; (b) Schematic of transcutaneous RF power and data link for an array of microimplants. The inset displays an illustration of a single, standalone submillimeter-sized silicon chip (Neurograin) which features an independent RF-based wireless telemetry channel for neural recording and electrical stimulation¹⁰¹; (c) Illustration of a flexible parylene C-based 3D neural probe with PEDOT:PSS electrodes for surface stimulation and single and multi-neuronal activity recording at surface and depth(left, right). Intracortical shank insertion is guided by a carrier glass pipette (center)¹⁰². Multimodal devices for chemical stimulation and electrical recording (d) Multifunctional probe created by the application of nanoelectronic coatings on the surface of micropipettes and cannulas for controlled infusion of cyanquixaline (CNQX) and simultaneous neural recording from 32 channels²⁷; (e) Illustration of a miniaturized neural drug delivery system (MiNDS) underlining the key device components (tungsten electrode, two borosilicate channels, polyimide templates) inside an etched stainless steel Hamilton needle for chronic EEG recordings and muscimol infusion¹⁰³; (f) A multifunctional fiber integrating four tungsten microelectrodes and two microfluidic channels within a poly(etherimide) (PEI) cladding for electrophysiology and modulation of neural activity during complex behavior in non-human primates⁹⁹; (g) Schematic of a microfluidic ion pump (μFIP) composed of two parylene sheets sandwiched with double-sided adhesive tape to form a microfluidic channel, featuring PEDOT:PSS microelectrodes for electrical recording and a polymeric ion bridge as an outlet for electrophoretic drug delivery¹⁰⁴. Multimodal devices for optogenetic stimulation and electrical recording (h) Illustration of a silicon-based electrophysiological probe with monolithically integrated InGaN μLEDs, designed to record and stimulate single-neuron potentials in the mouse hippocampus¹⁰⁷; (i) A single ZnO-based optoelectrode featuring an exposed active tip coated with indium-tin oxide (ITO) for electrical contact with neural tissue, an electrically isolated central structure with Parylene-C and lateral polymer adhesive for inter-shank insulation (left). A 4 × 4 array of the same optotetrodes as transparent waveguides and recording electrodes for optical stimulation and electrical recording in the mouse motor cortex (right)¹⁰⁸; (j) Schematic illustrating a single-step fabrication process of multifunctional fibers comprised of optical waveguides, electrodes, and microfluidic channels for simultaneous optogenetics, electrophysiology, and targeted drug and gene delivery in freely moving animals¹⁰⁵; (k) Illustration of an implantable shank featuring 1024 individually addressable CMOS-integrated organic light-emitting diodes (OLED) pixels that enables bidirectional neural interfacing in brain regions when paired with PEDOT:PSS electrodes⁷². Multimodal devices for chemical sensing and electrical recording (l) Schematic illustration of a neural probe integrating electrophysiological recording electrodes with a silicone-based microfluidic port channeled to 4 enzyme biosensor electrodes for real-time monitoring of neurochemical concentrations¹¹⁰; (m) Schematic of a dual-side fabricated shank tip featuring a three-electrode electrochemical sensor and neural recording electrodes, enabling simultaneous recording of electrophysiological activity and in situ dopamine detection in mice¹¹¹; (n) Schematic illustration showing a 2D stretchable film with 150 micropatterned sensors (left) that can be rolled into a monolithic 1D fiber (right) for electrophysiology, electrochemical detection and pressure sensing¹¹³. Multimodal devices for chemical and optogenetic stimulation (o) Cross section of a thermally drawn fiber for photopharmacology in behavioral studies with moving subjects¹¹⁴; (p) Illustration of a polyimide-based optofluidic probe system enabling simultaneous and independent control of blue and orange μ-ILEDs, equipped with silicone microfluidics, for delivery of pharmacological agents and optogenetic stimulation with two different light wavelengths¹¹⁵; (q) Optofluidic neural probe enabling simultaneous drug delivery and photostimulation to manipulate reward behavior in untethered mice. The inset is meant to compare the flexibility of this device (top) with that of a conventional metal cannula (bottom) ¹¹⁶. Multimodal devices for optical imaging and electrical recording (r) A transparent and flexible 64-channel graphene array used concurrently with two-photon calcium imaging of the mouse visual cortex, allowing for the correlation of cortical potential recordings with calcium waves⁷⁸; (s) Illustration of a flexible device combining an ECoG electrode array and 1024 GaN μLEDs laminated on the back of platinum nanorod grids (PtNRGrids) to form an intracranial electroencephalogram (iEEG) microdisplay for concurrent electrophysiological recordings and real-time visual representation of cortical activities¹¹⁷.

For high-resolution electrical stimulation, electrode size must be minimized while maintaining sufficient current densities. The high charge injection capacity and low impedance of PEDOT: PSS electrodes permits their miniaturization to 15 μm (in diameter) while delivering ~80 μA/mm² current densities and enabling recording of neuronal activity. These electrodes can be fabricated on flexible 4 μm Parylene-C substrates affording bidirectional interfaces to the mouse visual cortex (Fig. 3c)¹⁰².

Multimodal devices combining chemical stimulation and electrical recording

Probes combining electrophysiological recording with drug delivery enable study of neurochemical signaling and inform pharmacological therapies. Such systems also enable electrophysiological feedback for therapeutic drug delivery where dosing can be tailored based on each patient’s metabolism¹⁰³. For instance, oscillatory activity associated with Parkinson’s disease or epilepsy recorded via the integrated electrodes could inform patients’ responsiveness to multidrug therapies.

Integration of chemical and electrical modalities was achieved by applying electronic coatings (1 μm) comprising 32 recording electrodes to the surfaces of micropipettes or cannulas, delivering multifunctional probes ~30 μm in diameter (Fig. 3d)²⁷. In translational studies, 200 μm stainless steel needles outfitted with tungsten electrodes were combined with implanted micropumps (Fig. 3e) enabling electroencephalography and drug infusion in both rodents and NHPs¹⁰³. Thermally drawn fibers composed of polyetherimide and integrating 4 tungsten electrodes and 2 microfluidic channels reduced the mechanical mismatch (Fig. 3f). Implanted in the NHP cortex or putamen, these devices permitted recording and modulation of neural activity during complex behaviors⁹⁹. Electrophoretic drug delivery offers an alternative to pressure injection and can be similarly combined with electrophysiological recording. For instance, an electrophoretic pump was integrated with PEDOT:PSS microelectrodes which revealed the inhibitory and excitatory effects of gamma-aminobutyric acid (GABA) and potassium, respectively (Fig. 3g)¹⁰⁴. Albeit elegant, these systems are only applicable to charged species and cannot be re-loaded for repetitive use⁵⁷.

Multimodal devices for optogenetic stimulation and electrical recording

Integrating electrophysiology and optogenetics in a single device is essential for uncovering the roles of specific cells in brain circuits and understanding brain-organ communication. In addition to its cell-type specificity, temporal precision, and bidirectional control of neural activity, optogenetics is compatible with electrophysiological recording as optical and electrical modalities are decoupled. One particularly useful application of these combined functions is in “opto-tagging” which allows linking of spiking activity of neurons to their genetic identity¹⁰⁵. Although, opto-electrochemical artifacts have been observed in some electrode platforms, they can largely be avoided by minimizing the illumination intensity and selecting electrode materials with a high workfunction and no bandgap in the visible range¹⁰⁶.

Michigan-type silicon-based electrophysiological probes were augmented with optogenetics capabilities via heterointegration of 20 μm indium gallium nitride (InGaN) μLEDs, enabling recording and stimulation of single-neuron potentials in the mouse hippocampus (Fig. 3h)¹⁰⁷. Alternatively, optically transparent and electrically conducting zinc oxide (ZnO) pillars acted both as waveguides and recording electrodes permitting optical stimulation and electrophysiological recording of neural activity in the mouse motor cortex (Fig. 3i) ¹⁰⁸. Multimaterial fibers produced at a kilometer scale via thermal drawing can monolithically integrate optical waveguides, electrodes, and microfluidic channels in a one-step fabrication process, permitting optogenetics, electrophysiology, and drug and gene delivery in moving subjects (Fig. 3j)¹⁰⁵. These probes were recently extended to monolithically integrate μLEDs, enabling their wireless operation while retaining multifunctionality¹⁰⁹. However, high-density electronic integration in fibers remains challenging limiting the number of independently controlled μLEDs. Active electronics offer a potential solution to this challenge, and CMOS-integrated 20 μm microscale OLEDs have been combined with transparent PEDOT:PSS electrodes (Fig. 3k) to enable bidirectional neural interfacing across 1000 channels over > 6 mm² area of the mouse visual cortex⁷².

Multimodal devices for chemical sensing and electrical recording

Combining chemical sensing with electrophysiological recording permits correlation of changes in neuronal potentials in response to a neurochemical release and vice versa. A probe integrating electrophysiological recording electrodes with a silicone-based microfluidic port routed to an enzyme biosensor array permitted simultaneous probing of neuronal potentials and redox-based detection of glucose, lactate, choline, and glutamate in a mouse (Fig. 3l)¹¹⁰. Similar functionality can be achieved without the use of microfluidics. Simultaneous recording of electrophysiological activity and dopamine concentration has been shown with dual-sided probes comprising an electrochemical sensor and neural recording electrodes on an SU-8 substrate. This compact design (4 mm long, 33 μm thick, and 145 μm wide) enabled long-term experiments in mice (Fig. 3m)¹¹¹. Alternatively, a centimeter-scale sheet as thin as 1.5 μm, containing up to 150 micropatterned sensing components, can be rolled into a monolithic fiber with diameters as small as 150 μm, namely Spiral-NeuroString (Fig. 3n). This device enabled neuronal recording across 32 electrodes composed of a PEDOT:PSS hydrogel composite, while a predecessor of this technology, NeuroString, permitted electrochemical sensing of dopamine and serotonin in the brain of behaving mice by leveraging laser-induced graphene nanofiber electrodes transferred onto elastomer substrates^112,113.

Multimodal devices for chemical and optogenetic stimulation

The integration of in vivo pharmacology with optogenetics is poised to advance understanding of drug-adaptive/maladaptive neuronal dynamics as well as to connect fundamental cell-specific insights to therapeutic interventions. Multifunctional fibers enable fluid and light delivery, which not only allowed for the delivery of opsin genes¹⁰⁵ but also enabled chemical modulation of optogenetically-triggered anxiety-like behavior in mice, and could further be extended to photopharmacology with light-activated receptor agonists or antagonists (Fig. 3o)¹¹⁴. Alternatively, silicone microfluidics were outfitted with polyimide based flexible probes that hosted InGaN μLEDs. Integrated with a wireless data and power module, this system enabled long-term studies in freely moving mice, providing both pharmacological and optogenetic access to the target brain region^115,116. Notably, these systems could independently administer four compounds and two wavelengths of light (Fig. 3p)¹¹⁵. Such probes were further leveraged to alter gene expression, administer neuropeptides, and simultaneously deliver photostimulation and neurochemicals to manipulate reward behavior in untethered mice (Fig. 3q)¹¹⁶.

Multimodal devices for optical imaging and electrical recording

While electrophysiology recordings offer excellent temporal precision, they lack cellular specificity and structural insight into neuronal circuits. In contrast, imaging with activity indicators reveals both dynamics and location of genetically-identifiable neurons albeit with lower temporal precision. The two complementary modalities access diverse spatial and temporal scales, enabling real-time tracking of behaviorally relevant or pathological brain activity and circuit connectivity.

Transparent electrophysiological recording arrays enable simultaneous multiphoton imaging. For example, transparent PEDOT:PSS electrodes were harnessed to record and electrically stimulate neuronal activity, while simultaneously imaging cortical dynamics, as marked by GCaMP6f fluorescence, in mice¹⁰². Similarly, transparent graphene electrodes were employed concomitantly with GCaMP6s imaging in mice allowing for correlation of surface potentials recorded via electrocorticography (ECoG) with calcium waves (Fig. 3r)⁷⁸. Alternatively, a device combining an ECoG array and an intracranial electroencephalogram microdisplay, featuring 2048 LEDs, provided concurrent electrophysiological recordings and real-time visual representation of cortical activities (Fig. 3s). This was achieved by projecting spatially-defined light patterns onto the brain surfaces of rats and pigs¹¹⁷.

Multifunctional probes for the spinal cord, peripheral nerves, and organs

Earlier in the article, we summarized the neuroanatomical, physiological, materials, and mechanical challenges that accompany the development of long-term bioelectronic interfaces for the viscera. In this section we consider specific organs and discuss recent strategies that have delivered promising solutions to pertinent challenges.

Spinal cord

Although the spinal cord is a part of the CNS, its fibrous anatomy and propensity to elongate pose engineering challenges that are similar to those observed with peripheral neural interfaces¹¹⁸. The spinal cord neural circuits transduce sensorimotor signals and are involved in physiological processes ranging from pain and immune response to motivation¹¹⁸. Patients suffering from traumatic injury or degeneration of the spinal cord exhibit loss of limb and/or organ function, and the development of therapies demands deeper understanding of spinal neurophysiology¹¹⁹. The design of multifunctional spinal neurotechnology is challenging. Unlike the skull-encased brain, spinal cord undergoes substantial deformation¹²⁰ during normal movement, while the probe anchoring sites are limited to the mobile vertebrae of the spinal column.

Soft neurotechnology that combines translational electrical modalities with precise genetic approaches is poised to advance our understanding of spinal circuits¹¹⁹. One such example is that of multifunctional eDura spinal implant. To match the viscoelastic properties of the spinal tissue (~10 kPa for humans), eDura comprised of a 120 μm silicone substrate equipped with a microfluidic channel and soft platinum-polydimethylsiloxane (PDMS) composite stimulation electrodes patterned onto stretchable contacts of micro-cracked gold³². eDura exhibited modulus of ~2 MPa and could accommodate ~20 % strain, which allowed for its implantation below the dura-matter. This subdural placement established intimate contact between the electrodes and the spinal tissue without altering the limb kinematics during skilled locomotion tasks in rats. eDura was additionally applied for combined electrical stimulation and serotonin delivery in a contusion model of the lumbar spinal cord injury and was shown to facilitate functional locomotor recovery in rats. Epidural electrical stimulation with a chronically implanted soft electrode array similar to eDura has also led to improvements in gait impairments in Parkinson’s disease (PD). While previous interventions in PD have focused on brain regions affected by loss of dopamine neurons, the spinal neuroprosthesis allowed targeting of the lumbosacral region of the spinal cord that produces walking and is spared in motor deficits underlying PD (Fig. 4a)¹²¹. In a NHP model of PD-induced gait impairment, an ensemble of implanted devices (in the motor cortex, spinal cord, and leg muscles) created a digital bridge between the brain and spinal motor neurons. Collectively these devices reduced gait deficits through closed-loop spinal electrical stimulation triggered by motor intentions recorded wirelessly¹²¹.

Figure 4. Bioelectronic devices for peripheral organs.

Spinal cord (a) A soft epidural electrode array implanted over the lumbar and sacral spinal cord along with a distally placed pulse generator in a NHP model of Parkinson induced gait deficiency¹²¹; (b) A soft and stretchable spinal cord optogenetic device with independently controlled μLEDs for dorsoventral optogenetic control of spinal circuits in freely moving mice¹²². The mice also carry a wearable wireless controller; (c) A soft, stretchable multifunctional fiber that integrates optical waveguide for light delivery and concentric Ag-NW electrode for simultaneous electrophysiology³⁶; (d) An ultrathin (~1 um) and flexible microelectrode array for single-unit electrophysiology of the motor neurons in the ventral aspect of the spinal cord in freely moving mice¹²³. Peripheral nerves (e) A soft viscoelastic sciatic nerve interface that integrates electrodes and strain sensors and maintains robust contact with a growing nerve in mice¹²⁹; (f) A soft hydrogel based optical waveguide and hydrogel cuff for studying pain behavior using optogenetics in the sciatic nerve of behaving mice¹²⁸; (g) A nerve cuff that is entirely composed of bioresorbable components including the electrodes and encapsulation layers and delivers high frequency (kHz) electrical stimulation to block nerve conduction as a form of a dissolvable pain management device¹³². Innervated Cardiovascular system (h) A multisite cardiac optogenetics device that allows wireless optical control of heart pacing in a chronically implanted mice¹³⁷; (i) An intrinsically stretchable electrode mesh composed of Ag-Au/SEBS composite that allows volumetric interfacing with the heart surface¹³⁸; (j) A bioresorbable leadless wireless cardiac pacemaker¹³⁹; (k) A wearable optical jacket for minimally invasive optogenetic control of cardiac rhythms that allows studies of heart-to-brain interoceptive circuits that modulate affective behaviors in mice⁶⁸. Innervated Bladder (l) A closed loop wireless optogenetic device that integrates μLEDs for optogenetic inhibition of bladder muscle and a strain sensor that informs about status of bladder filling for management of overactive bladder pathology in rats³¹; (m) An optoelectronic thread that integrates μLEDs, strain sensors, temperature sensors, electrodes, and a soft web that engages on the bladder to allow management of under active bladder pathology in mice¹⁴⁶; (n) A wireless soft strain sensor allows longitudinal studies of urodynamics in mice and baboons for several weeks¹⁴⁸. Innervated Gastrointestinal tract. (o) A soft multifunctional microelectronic fiber allows wireless optogenetic modulation of various components of gut neural circuitry such as the vagal afferents and enteroendocrine cells in the proximal and distal small intestine of mice¹⁰⁹; (p) A stomach interfacing soft, flexible wireless optogenetic device allows organ specific modulation of Calca+ stomach vagal afferents in mice¹⁵⁶; (q) An ingestible electronic pill that unfurls into S-shaped electrodes and delivers electrical stimulation to re-animate the intestine in a porcine model of post-operative ileus¹⁵⁸; (r) A gastrointestinal pill, that combines genetically engineered bacteria which emit light in response to heme sensing and a miniature electronic circuit that detects emitted fluorescence, serves as a ingestible biochemical sensor to detect blood leaks in the GI tract of a porcine model¹⁵⁹.

Although electrical spinal stimulation coupled with physical therapy has been shown to re-establish coordinated motor activity, the underlying neural mechanisms remain poorly understood¹¹⁹. Cell-type specific optogenetic neuromodulation combined with electrophysiological recording in the spinal cord has the potential to inform the refinement of therapies following spinal cord injury. For this purpose, multicolor μLEDs were integrated into polymer-based probes (Fig. 4b) that leveraged stretchable serpentine interconnects to accommodate ~15% strain observed in the lumbar spinal cord extension¹²². The device was wirelessly controlled via a battery-powered head-mounted module, and the ability to tune μLED emission wavelength permitted optogenetic excitation and inhibition in chronically implanted mice. Furthermore, closed-loop control was achieved by triggering light pulses upon detection of tibialis anterior EMG activity that effectively reduced step length.

Understanding spinal neurophysiology also benefits from devices that can target deeply located spinal neurons, such as those of the ventral horn. As a step towards that goal, fiber-based fabrication was applied to a transparent elastomer (~34 MPa)³⁶. The resulting fiber was outfitted with a silicone cladding to form a soft, stretchable optical waveguide, and a mesh of silver nanowires was deposited onto the probe surface to serve as a concentric electrode that was insulated by an additional layer of silicone. The resulting probes withstood repeated 20% strain while maintaining low impedance and high optical transmission, which enabled optogenetic neuromodulation and electrophysiological recording in the lumbar (L1) spinal cord of moving mice (Fig. 4c)³⁶. Stable, long-term recording of high-resolution electrical activity from the motor-circuits in the ventral horn of the spinal cord is particularly challenging due to the large relative motion between the spinal vertebrae and the spinal tissue. This was recently addressed by developing flexible electrode arrays in the form of ultra-thin (1 μm) polymer-based nanoelectronic threads which enabled laminar recordings of multiple neuronal units in the ventral aspect of the lumbar cord in unrestrained mice (Fig. 4d). These ultraflexible 32-channel spinal probes successfully tracked single motor neuron potentials for up to a week during locomotion¹²³.

Peripheral nerves

The PNS can be divided into somatic and autonomic branches. The somatic branch, comprising of sensory and motor neurons, controls voluntary actions and senses, while the autonomic branch consisting of the sympathetic and parasympathetic fibers controls involuntary functions including heart rate and digestion. Peripheral neuromodulation has recently emerged as a powerful approach for reversing the disrupted homeostatic state underlying the pathobiology of autoimmune, cardiovascular, and metabolic diseases. For instance, vagus and splenic nerve stimulation are being explored as therapies for rheumatoid arthritis¹²⁴, drug resistant depression, epilepsy¹²⁵, and inflammatory conditions¹²⁶. These pioneering bioelectronic therapies rely on commercial nerve interfaces with limited spatial resolution (~2 channels, ~2-4 contacts, 1 anchor point), which may yield off-target effects due to the heterogeneity of intertwined fibers within larger nerves¹²⁷. Consequently, future bioelectronic therapies demand further innovation in miniaturized and targeted nerve interfaces.

Peripheral nerves exhibit miniature cross sections (10 to 500 μm in rodents), delicate mechanics (54 kPa, ulnar nerve in humans), and undergo substantial repeated deformation (up to 10% strain, 10⁴ cycles per week in a mouse sciatic nerve)¹²⁸. Consequently, stable peripheral neural interfaces benefit from soft stretchable materials and miniature form factors. As an example, a viscoelastic conductor composed of PEDOT:PSS and glycerol was developed to maintain low impedance (~2-3 kΩ) even at 100% strain¹²⁹. This material was then leveraged in electrodes and strain sensors integrated on low-modulus (0.62 MPa) viscoelastic substrates enabling intimate interfaces with peripheral nerves. These devices monitored compound potentials and strain in a sciatic nerve of rats during their growth between the postnatal weeks 4 and 8 (Fig. 4e).

To interrogate nerve physiology with cell-type precision, soft hydrogel based optical-waveguides were integrated with the sciatic nerve cuffs in mice¹²⁸. To enable light confinement within the hydrogel fiber core, the latter was annealed to promote formation of nanocrystalline domains leading to an increased refractive index. These devices could withstand deformation during vigorous locomotion (Fig. 4f) and permitted selective inhibition of pain-transducing fibers in a mouse model of hyperalgesia. Combining optogenetics with pharmacology enabled simultaneous activation of nociceptors and their inhibition with an anesthetic drug bupivacaine in a contextual pain-induced aversion paradigm in mice. In these studies, sciatic nerve cuffs equipped with μLEDs and microfluidic interfaces were further augmented with implanted pumps and wireless power transfer antennae eliminating the need for cumbersome electrical and fluidic tethers to external hardware¹³⁰.

Although powerful in fundamental studies in rodents, optogenetic modulation of pain is currently inaccessible in the clinic, where there is an urgent need for non-pharmacological pain management solutions. High-frequency electrical stimulation offers a clinically viable approach to blocking nerve conduction and blunting of sensory signaling¹³¹. While some patients require chronic pain management, others benefit from a temporary pain relief following a major surgery/injury. Such cases could benefit from a wireless bioresorbable electrical cuff that employed molybdenum (300 μm) contact electrodes with a slow dissolution rate (0.02 μm/day) and magnesium (50 μm) interconnects encapsulated within a polyanhydride dielectric (200 μm) (Fig. 4g)¹³². This device enabled a complete sensory block at stimulation frequencies ≥25 kHz, and its complete time-dependent bioresorption was confirmed through computed tomography.

Cardiovascular system

The cardiovascular system comprising of the heart and the peripheral blood vessels is a complex electromechanical syncytium that requires synchronous function of cardiac muscles, sympathetic and parasympathetic nerves, and the intracardiac ganglia^133,134. Monitoring and modulation of cardiac physiology over extended timescales is essential to treatment and prevention of cardiovascular pathology. While studies of ex vivo and in vitro systems have delivered indispensable insights into cardiac function, they fall short of recapitulating biological complexity found in behaving subjects. Historically dogs, sheep, and swine were preferred as model organisms for developing cardiovascular bioelectronics as they exhibit anatomy, physiology, and biomechanics closely resembling that of a human cardiovascular system¹³⁵. However, recent progress in genetic tools has motivated the use of rodent models in cardiac research¹³⁶. Coupled with multifunctional bioelectronics, these genetic approaches are advancing our understanding and treatment of cardiac pathology.

To enable cardiac optogenetics and electrophysiology, μLEDs and platinum electrodes were integrated into polymer-insulated serpentine meshes¹³⁷. Equipped with wireless control circuit, these stretchable interfaces could be sutured directly onto the pericardium, enabling optogenetic studies of cardiac physiology in untethered mice over 12 days (Fig. 4h). This platform improved tissue integration and facilitated conformal interfacing, enabling multisite cardiac pacing.

Multisite cardiac pacing can also be accomplished electrically using epicardial meshes composed of elastomeric metal nanowire composites. Unlike traditional pacemakers that employ 2 rigid electrodes, nanocomposite meshes intimately interfaced with the heart and enabled synchronized stimulation of the ventricles that improved their systolic function in a rat (Fig. 4i)¹³⁸. A temporary form of cardiac pacing is often used during post-operative recovery after a major cardiac surgery to increase cardiac output. The current clinical standard relies on pacemakers with leads externalized through transcutaneous ports. Besides being prone to infection, this approach carries a risk of myocardial perforation during device removal. To address this, a bioresorbable leadless cardiac pacemaker was demonstrated by introducing thin tungsten-coated magnesium electrodes into biodegradable dielectric substrates¹³⁹. These devices incorporated wireless power antennas, allowing them to be fully implanted in rodents, where chronic pacing was demonstrated up to 4 days, and complete biodegradation was observed within 3 months (Fig. 4j).

In addition to pacing, spatiotemporal mapping of cardiac physiology demands intimately integrated electrical interfaces. Active 18×16 electrode arrays incorporating silicon nanomembrane transistors were developed on 25 μm-thick polyimide substrates, enabling on-board amplification and multiplexing of spatially-resolved cardiac potentials recorded from perfused rabbit hearts¹⁴⁰. Besides flexibility, device stretchability accommodates cardiac expansion. To minimize the stress concentration on the cardiac tissue, bioelectronic cardiac interfaces have been developed from intrinsically stretchable organic conductors, semiconductors, and insulators arranged into architectures with mechanical moduli ~10-100 kPa matching that of cardiac tissue (~10-35 kPa)¹⁴¹. Leveraging mixed ionic and electronic conduction in PEDOT:PSS based electrochemical transistors permitted amplification of cardiac recordings and achieved SNR of >50 dB with conformal 1.2 μm thick devices¹⁴².

The heart receives both parasympathetic (pre-ganglionic) and sympathetic (post-ganglionic) innervation from the brainstem and thoracic ganglia, respectively. Clinical research indicates that pathologies such as arrhythmias can arise in healthy cardiac tissue due to brain activity associated with maladaptive stress. However, until recently it was not possible to study if the aberrant cardiac signaling can similarly affect emotional states. To enable such studies, Hsueh et. al. developed a minimally invasive approach to optogenetically modulate heart rate and assessed the effect of tachycardia on anxiety in mice (Fig. 4k)⁶⁸. The authors expressed a red-light activated opsin, ChRmine, in cardiomyocytes of both ventricular and atrial walls in mice, and then outfitted these subjects with fabric vests housing LEDs. Optically induced ventricular tachycardia using this system produced an anxiogenic effect, which was then linked to the activity in the posterior insular cortex. While effective in few-minute long assays, wearable platforms are prone to damage inflicted by the animal itself and may restrict naturalistic behavior in longer studies, posing the need for bioelectronic cardiac interfaces that leverage advanced optogenetic tools in complex behavioral paradigms.

Bladder

The bladder functions as a temporary storage for urine formed in the kidneys. Under physiological conditions, bladder voiding is under voluntary control via the signals received from the CNS. The bladder also receives somatic afferent, sympathetic, and parasympathetic innervation from the thoracic and lumbosacral regions of the spinal cord¹⁴³, which provide an entry point for developing neuromodulation therapies for the disorders such as the overactive bladder and nonobstructive urinary retention¹⁴⁴. Patients unresponsive to behavioral and pharmacological interventions are considered for electrical stimulation of bladder innervating nerves via bioelectronic interfaces such as sacral nerve stimulators¹⁴⁵. Since rigid electrodes are ineffective in targeting smaller nerves closer to the organ, they are typically placed onto larger nerve bundles which branch out to multiple organs; thereby reducing specificity of stimulation. This invites innovation in miniature peripheral nerve interfaces targeting the bladder.

To inform the design of clinical bladder interfaces, optogenetic neuromodulation has been achieved in devices that can directly laminate onto the bladder wall. These flexible tools were applied to optically inhibit sensory nerves while continuously monitoring bladder activity with the integrated soft strain-gauges (Fig. 4l). These devices reduced the number of micturition events in a rat model of abnormally high voiding frequency. In a related example, optogenetic neuromodulation was applied to rescue an under-active bladder in a diabetic mouse model³¹. This multifunctional interface comprised a stretchable ribbon integrated with μLEDs for optogenetics, electrodes for bladder electromyography, a thermal sensor, and a strain-gauge (Fig. 4m)¹⁴⁶.

Major lower-urinary tract surgeries commonly require follow-up urodynamic studies¹⁴⁷. Measurements of bladder filling and voiding efficacy can only be performed intermittently as they require catheter insertions through the urethra. The chronically indwelling catheters create risks for urinary tract infections which may be fatal for vulnerable patients. To address these challenges, a fully implantable catheter-free wireless bioelectronic device (Fig. 4n) was developed to interface with the bladder wall and provide real-time continuous measurement of bladder volume in rats and NHPs¹⁴⁸. The device exploited a piezo- resistive strain-gauge composed of silicone-carbon composite. In rats, the device wirelessly recorded large and small voiding events over 24 h periods even in the presence of ambient electrical noise for 30 days. The platform was translated to partial cystectomy model in a NHP, and wireless real-time urodynamics recordings were performed over 8 weeks.

Gastrointestinal tract

The gastrointestinal (GI) tract is a luminal organ that includes oral cavity, esophagus, stomach, duodenum, jejunum, ileum in the upper section and cecum, colon, rectum in the lower section. The key functions of the GI tract are the digestion of food and absorption of nutrients, and expulsion of undigested waste. In addition to its critical role in metabolism, GI tract plays a major role in immunity, and all of the GI functions are aided by the commensal microbiota and extensive innervation^149,150. The GI tract receives dense vagal sensory and parasympathetic innervation from the nodose ganglia and brainstem respectively, as well as sympathetic and somatic innervation from the spinal cord^151–153. Besides this extrinsic innervation, the gut wall also harbors a vast autonomous enteric nervous system (ENS) consisting of concentric myenteric and submucosal plexi – an intricate network comprising ~1 billion sensory, motor, and interneurons, as well as glia^14,154. The ENS performs neural computations that facilitate digestion, absorption, and peristalsis without the mandatory involvement of the CNS.

Akin to studies of heart and bladder, optogenetics and physiological sensing in rodent models have the potential to reveal specific cellular contributions to various aspects of GI physiology and pathology. Soft, multifunctional devices that can chronically reside in luminal cavities or organ surfaces of the GI tract have recently emerged as promising tools for studies of neurogastroenterology. Thermally drawn ~ 230 μm diameter flexible polymer fibers were designed to exhibit low optical losses (0.93 dB/cm) invariant to mechanical deformations¹⁵⁵. Their low bending stiffness permitted chronic implantation in the mouse duodenum. These intraluminal devices delivered optogenetic neuromodulation to sensory enteroendocrine cholecystokinin (Cck) cells revealing their role in sugar sensing. While useful, passive optical fibers are limited to tip illumination and require tethers to external light sources which can restrict animal movement in long assays of metabolism. These challenges were overcome by monolithic integration of μLEDs into multifunctional fibers (Fig. 4o) enabling multi-site, multi-color illumination as well as fluid delivery controlled via a wireless module¹⁰⁹. Optogenetic studies with these microelectronic fibers revealed the role of enteroendocrine cells producing peptide YY and glucagon-like peptide-1 (GLP-1) in mediating ileal brake and demonstrated that reward behaviors can be controlled from the gut by leveraging vagal innervation in the duodenum. Similarly, a thin, flexible tether comprising a polymer substrate and insulation for metal interconnects and μLEDs was designed to curve along the stomach cavity simplifying intragastric implantation and reducing strains on metal traces¹⁵⁶. The device was further equipped with subdermal control electronics and wireless power transfer. These probes were applied for optogenetic excitation of chemosensory Calca⁺ vagal fibers in the gastric corpus, which resulted in robust reduction in food intake in fasted mice (Fig. 4p). Optogenetic excitation of Calca⁺ afferents was also shown to be aversive.

The GI tract provides a uniquely straightforward access to its luminal surfaces via the oral cavity which has led to an emergence of multimodal ingestible electronic capsules. Advances in microfabrication and miniaturized electronics permitted packaging of a camera, illumination source, supporting electronics, data-transmission antenna, and a power supply into ~2.5 cm long and ~1.3 cm wide capsules (akin to a large vitamin tablet) that can be ingested and then expelled arboreally by the patient¹⁵⁷. Such capsule endoscopes are routinely used in clinic for high resolution imaging of the entire gut lumen, and additional functions are under development in animal models. Ingestible electrical stimulators have been demonstrated for intestinal reanimation in post-operative ileus. These devices employed S-shaped intraluminal electrodes that ensure conformal contact with the gut wall without obstructing GI transit (Fig. 4q)¹⁵⁸. To achieve controlled deployment of the device from the gelatin capsule body into the small intestine, the platform featured a pH-sensitive coating which dissolved upon entry into the small intestine. In porcine models of paralyzed intestinal motility these devices accelerated net transit and increased local intestinal contractile rate. An ingestible diagnostic pill was also designed to host heme-sensing pro-biotic bacteria engineered to emit light upon detection of blood (Fig. 4r). This biohybrid sensing unit was juxtaposed over a sensitive miniature photodiode, microcontroller, and wireless communication circuitry which transduced the optical signal from the bacteria into a digital output. The ingestible devices were applied for detection of upper GI bleeding in a porcine model, which would otherwise require an endoscopic observation¹⁵⁹.

Ingestible devices have also been configured to deliver drugs directly into the gastric mucosa for rapid and efficient uptake. For instance, sensitive macromolecular drugs, such as insulin, were directly injected into the gastric mucosa using ingestible capsules equipped with an insulin needle activated via the dissolution of a biodegradable stopper. This platform was further extended to self-orienting gastric autoinjectors that could deliver liquid formulations of monoclonal antibodies and peptides with improved dosing efficiency (upto ~80% bioavailability) and rapid pharmacokinetics^160,161.

Challenges and Outlook

Implantable neurotechnologies have captured the imagination of academia and industry alike for their potential to improve and augment human health through functional body-machine interfaces¹⁶². The increasingly advanced multifunctional bioelectronics platforms demonstrate bidirectional interfaces with the central and peripheral nervous systems in pre-clinical animal models. This intense innovation foreshadows a future where any organ in the body can be manipulated or monitored with minimal invasiveness, thus promising to revolutionize therapies for a host of conditions ranging from spinal cord injury to metabolic diseases. An aspirational goal that directly follows from continuous monitoring will be deciphering the neural mechanisms of brain-body crosstalk that can then define both intervention timepoints and strategies for preempting pathology; thereby transforming clinical healthcare from reactive response to proactive prevention. This neurotechnological utopia, however, cannot be realized without deeper understanding of fundamental neurobiological mechanisms underlying physiological and pathological functions of the brain-body connectome. Even in their current forms, multifunctional bioelectronics are indispensable tools in hands of physiologists, neuroscientists, and physicians^163–165. Combined with advances in genetics, disease models, and powerful computational techniques for animal behavior, brain-body neurotechnology promises to provide new biological insights in studies of brain-body communication and unlock solutions to previously intractable questions.

Despite their promise, bioelectronic interfaces face technological challenges in their transition from proof-of-concept demonstrations to robust tools of discovery and therapy. We envision the following critical areas of innovation and impact: (1) Integration of chronically-stable biochemical recording capabilities into spinal and peripheral neurotechnologies; (2) Robust encapsulation strategies that extend the functional lifetime of implanted device to at least 1 year in rodents; (3) Efficient low-power circuits integrating edge-artificial intelligence for on-board processing of diverse data and closed-loop operation.

Mapping the dynamics of biochemical signatures in a physiological state or as a disease biomarker can provide insights into the neural mechanisms of homeostasis and help assess the efficacy of a therapeutic intervention. Electrochemical sensors¹¹² outfitted with polymer coatings that slow-down or prevent biofouling offer a promising avenue for innovation. The rapidly growing libraries of genetically encoded neurochemical indicators^166,167 fueled by advances in genetic screens and directed-evolution pipelines underscore the importance of optical methods of chemical recording¹⁶⁸. While such optical approaches have been successful in the brain, extending them to peripheral organs and nerves demands innovation in optoelectronic materials and device architectures, as well as advanced computational imaging methodologies^169–171.

Longitudinal recording and neuromodulation studies in rodent models, such as those centered on metabolism and related disorders, would benefit from neurotechnologies performing reliably over the course of months to years. Realizing (semi)hermetic device encapsulation while retaining flexibility and stretchability in multimodal probes for peripheral organs remains challenging. High-throughput accelerated aging test platforms should be implemented into bioelectronics engineering workflows to facilitate the design and evaluation of encapsulation strategies that can withstand strains associated with nerve and organ movement.

A subset of the more mature technological platforms discussed here has transitioned from academic research to commercial distribution¹⁷² and clinical practice¹⁶³. Such translation is expected to accelerate the pace of fundamental discovery and therapeutic intervention for disorders of the central and peripheral nervous systems. Brain-organ communication is a frontier of neuroscience and neurotechnology that requires synergistic collaboration between engineers, neuroscientists, and clinicians. By embracing biologically-informed design and by learning to communicate across disciplinary silos, the “intractable” problems in biology will find imaginative solutions in engineering and vice-versa.

Table 2.

Comparison of different wireless power-transfer techniques for brain–body neurotechnologies

Wireless power transfer method	Transmitter metrics	Field properties	Implant metrics	Advantages	Disadvantages
Near field Inductive coupling 174	• Coil geometry • Transmitter distance ~10 cm	• ~13 MHz Magnetic field • No tissue attenuation • $B_z=\frac{\mu NI{R}^2}{2\pi {(z^2+R^2)}^{3/2}}$	• Power harvested, P ~10mW • $\text{P}=\frac{{\pi }^{2}NB{R}^3}{4\mu }\times \frac{dB}{dt}$ • Depth ~ 10 cm • Angular misalignment > 45° • ~mm scale implant • Resonant frequency, $f_r=\frac{1}{2\pi {\left(LC\right)}^{1/2}}$	Miniature implants that can be high powered and located at deeper regions	The power coupling depends on area of the coil and is very sensitive to angular misalignments
Far-field 201,202	• Loop, Dipole, Horn • Transmitter distance ~ few meters	• 1-5 GHz electromagnetic radiation • Tissue attenuation	• Power harvested, P ~ 1mW • Depth ~ 1cm • Misalignment lower than inductive coupling ~25%	Long range of power transfer and particular efficient for data communication such as Bluetooth and Wi-fi radio	Low power transmission and shallow implants due to attenuation of radiative EM energy in tissue
Ultrasound 175,176	• Ultrasonic horn • Transmitter distance = 0 cm	• ~1.8 MHz sound waves • Moderate attenuation in tissue, bones	• Power harvested, P ~ several mW • Depth ~ 1-2 cm • Sub mm scale implant • $f_r=\frac{1}{2l}({\frac{Y}{\rho })}^{1/2}$	Acoustic resonances allow extreme device miniaturization	Transmitter has to be in contact with the tissue
Magnetoelectric 203,204	• Coil geometry • Transmitter distance ~10 cm	• ~300 kHz magnetic field • No tissue attenuation $B_z=\frac{\mu NI{R}^2}{2\pi {(z^2+R^2)}^{3/2}}$	• Power harvested, P ~ several mW • Depth ~10cm • Implant scale ~ several mm • $f_r=\frac{1}{2l}({\frac{Y}{\rho })}^{1/2}$	Combines advantages of ultrasound and near-field power coupling	Transmitter has to be custom built for each application
Photovoltaic 205	• Light source • Transmitter distance ~ 10 cm	• 450 nm to 1100 nm electromagnetic radiation • Strong tissue attenuation (wavelength dependent)	• Power harvested, P < 1mW • Depth ~ few mm • ~ sub mm scale implant	Good for low power application and miniaturized implants	Implants cannot be deeply located due to strong light attenuation by tissue
Ionic communication 206,207	• Geometry of TX and RX electrodes • Ion concentration • Transmitter distance ~ 0.5 cm	• Frequency range defined by electrode impedance • $f_c~\frac{R_s}{C_e}$	• Power transmitted, P ~1 mW • Depth ~ 0.5 cm • μm scale implant	Highly flexible and biocompatible	Tx and RX electrodes need to be firmly in contact with the tissue

B_z = magnetic field strength along the z-axis of the coil; N = number of loops in the coil; R = coil radius; μ = permeability of the medium; I = current; z = distance from center of the coil; t = time; L = inductance; C = capacitance; Y = Young’s modulus; ρ = density; TX = transmitter; RX = receiver; R_s= solution resistance; C_e= electrode capacitance.

Key Points.

• The bidirectional crosstalk between the brain and visceral organs is essential to maintaining homeostasis, and is additionally implicated in neurological, metabolic, and immune disorders.

• Discovery of neural pathways underlying brain-organ communication may reveal novel targets for autonomic neuromodulation therapies or allow us to modulate brain function from the viscera.

• Deciphering brain-body neural circuits is challenging, in part, due to the dearth of neurotechnology for multisite, multimodal investigation of brain and organ physiology in awake behaving model organisms for extended periods of time.

• Integration of multiple recording or stimulation modalities in a neural probe should not compromise device miniaturization, biocompatibility, and mechanical flexibility to ensure reliable long-term function in vivo.

• Besides the tissue interfacing front-ends, equal consideration should be devoted to developing complete functional systems which include interconnects, encapsulation, control electronics, data transfer protocols, and power delivery routes.