The brain is the final frontier of modern medicine. Being able to peer into someone’s mind and understand their inner motivations and struggles is something that many of us desire on a daily basis. In recent years, neuroscience has come exceedingly close to capturing patient brain states often enough to inform clinical practice. For example, electroencephalography (EEG) and electrocorticography (ECoG), which record waves of brain activity asvoltage measurements, have helped us treat neurological diseases and improve mental health outcomes. Moreover, patient-derived tissues have enabled more invasive studies of neuronal network signaling using microelectrode arrays (MEAs) to understand connectivity changes during stress, disease, or aging. Unfortunately, it turns out that just like humans, what a neuron says isn’t necessarily what it thinks. We now know that beneath the “spikes” that a neuron fires to communicate with other neurons, there exists an intricate system of subcellular inputs and processing tasks that each neuron computes before it decides to send a message along to a particular subset of its downstream neighbors.
To measure these subcellular inputs, one needs to perform the Heisenbergian task of poking a neuron with a probe small enough to puncture a minimally invasive patch of the neuron’s membrane, yet large enough to be compatible with modern fabrication techniques. Currently, the gold standard technique, called patch clamping, for studying these subcellular dynamics can only record from up to 10 neurons at a time. Patch clamping harnesses the biophysical laws of adhesion between a glass micropipette and the protein-studded plasma membrane to create a fluid-tight physicalseal between the recording electrode and the interior of a neuron. Given that the human brain contains over 80 billion neurons, many neuroengineers are developing more scalable technologies. The new holy grail of neuroscience has shifted towards creating multiscale, topological maps of neuronal activity, from subcellular synaptic potential patterns (sub-millisecond, sub-micron), to neuronal circuitry (millisecond, micron - millimeters), to sweeping changes in brain homeostasis (seconds - hours, centimeters) throughout a person’s life. In this paper, we present a step forward in this direction by validating a nanopillar electrode array platform for recording sub-threshold (synaptic potentials) and supra-threshold (spikes) activity in 2D and 3D neuronal architecturesin vitro.
While neurons typically employ electrical signaling to communicate with other neurons, they are actually multilingual entities. Neurons, like other cells in the body, interact with their microenvironment using mechanical and chemical signaling as well. Neurons possess several mechanosensitive pathways which help guide their morphology and physical presence in the brain to facilitate synapse formation homeostatic regulation. When designing a bioelectronic device to interface with single neurons, neuroengineers often begin with known mechanical regulators of neuronal guidance, patterning, and synaptogenesis in order to design a welcoming microenvironment for neurons to grow and form connections. Recent research has revealed that micro- and nano-topographical structures, such as nanopillars, can trigger curvature-sensitive biological pathways in several cell types, such as increased adhesion, endocytosis, and directed growth. Sure enough, when we construct nanopillars out of glass using standard nanofabrication procedures, we observe both morphological patterning and synaptic guidance within neuronal cultures grown on top of our devices.
Figure 1. Morphological patterning of single neurons on flat and nanopillar arrays.
**Figure 2.**Synaptic guidance of neurites within single neurons on nanopillar arrays. (a) Merged fluorescence image, with the white dashed circles indicating 4 different micropillars (scale bar: 3 μm), showing synapsin-1 punctae curving around separate micropillars. (b) Another merged fluorescence image showing synapsin-1 punctae curving around a single micropillar (scale bar: 2 μm). The white dashed circles indicate the tip and base of the nanopillar. This same pillar is depicted schematically in (c), demonstrating putative microscale curvature achieved by synapsin-1 distribution across the presynaptic region of the neurite. (d) Zoomed in SEM images of ROI shown in (b) with white dashed circles corresponding to the same four micropillars (scale bars: 3 μm).
Based on these data, we constructed bioelectronic devices containing nanopillar electrodes designed to improve the physicalseal with overlying single neurons. We drew from previous designs of microelectrode arrays and nanostructured arrays to create nanopillar electrode arrays (NEAs) intended to interface with several different subcellular elements of single neurons, including the neuronal soma and the neurites. Then, we cultured neuron-like cells from several different sources, including human induced pluripotent stem cell-derived (iPSC) neurons, brain organoids, and primary neurons obtained from mouse hippocampi. We wanted to validate our platform across the most commonly studied models of neuronal networks accessible to neuroscientists around the world. Additionally, we wanted to show that our devices could record signals from neurons arranged in various cytoarchitectures, such as a monolayer of neurons, or a 3D construct of neurons arranged around a central focal point. Across all neuronal subtypes, we were able to obtain intracellular-like recordings of both supra-threshold signals, called action potentials or “spikes,” and sub-threshold signals, called postsynaptic potentials or subcellular oscillations. These two kinds of signals are typically distinguished by performing frequency analysis, since action potentials result from rapid sodium influx from voltage-gated sodium channels, while sub-threshold potentials have various but slower biological origins.
We performed additional validation experiments using pharmacological interrogation of neurons cultured on our NEAs. Electrophysiologists typically introduce well-established drugs, including ion channel blockers and antagonists, to their neuronal cultures in order to precisely modulate the ionic fluctuations that determine the membrane voltage of a neuron. The response of a neuron to having certain subsets of its ion channels blocked can reveal insights into its biological identity. For example, as shown below, the iPSC-derived neurons grown in our experiments were expected to be glutamatergic, and responded transiently to glutamate stimulation, as shown below. Alternatively, when we added two different drugs in quick succession to primary hippocampal neurons from mice, we were able to discern sub-threshold from supra-threshold spikes by first increasing all activity with kainate, and then suppressing mostly the supra-threshold action potentials using tetrodotoxin (TTX).
**Figure 3. Modulation of network activity using pharmacology to validate intracellular recordings.**Glutamate addition to iPSC-derived neuronal network temporarily increases spiking.
In summary, we demonstrate the successful recording of supra- and sub-threshold intracellular-like signals from neuronal networks cultured on nanopillar electrode arrays. We showed that the pillar shape can induce biological changes in subcellular neuronal compartments that likely improve their physical seal with nanoelectrodes, allowing highly sensitive recordings. In the future, we aim to (1) build even denser arrays of nanoelectrodes to enable accurate estimates of the excitatory to inhibitory ratio within various cultures, (2) explore nanostructures containing a broad spectrum of curvatures to enable patterning of synaptic subtypes, and (3) perform systematic pharmacological interrogation of healthy and diseased neuronal networks on NEAs to identify novel drug targets for combatting neurological disease.