Ensemble learning and ground-truth validation of synaptic connectivity inferred from spike trains.

Probing the architecture of neuronal circuits and the principles that underlie their functional organization remains an important challenge of modern neurosciences. This holds true, in particular, for the inference of neuronal connectivity from large-scale extracellular recordings. Despite the popularity of this approach and a number of elaborate methods to reconstruct networks, the degree to which synaptic connections can be reconstructed from spike-train recordings alone remains controversial. Here, we provide a framework to probe and compare connectivity inference algorithms, using a combination of synthetic ground-truth and in vitro data sets, where the connectivity labels were obtained from simultaneous high-density microelectrode array (HD-MEA) and patch-clamp recordings. We find that reconstruction performance critically depends on the regularity of the recorded spontaneous activity, i.e., their dynamical regime, the type of connectivity, and the amount of available spike-train data. We therefore introduce an ensemble artificial neural network (eANN) to improve connectivity inference. We train the eANN on the validated outputs of six established inference algorithms and show how it improves network reconstruction accuracy and robustness. Overall, the eANN demonstrated strong performance across different dynamical regimes, worked well on smaller datasets, and improved the detection of synaptic connectivity, especially inhibitory connections. Results indicated that the eANN also improved the topological characterization of neuronal networks. The presented methodology contributes to advancing the performance of inference algorithms and facilitates our understanding of how neuronal activity relates to synaptic connectivity.

A Multimodal Fitting Approach to Construct Single-Neuron Models With Patch Clamp and High-Density Microelectrode Arrays.

In computational neuroscience, multicompartment models are among the most biophysically realistic representations of single neurons. Constructing such models usually involves the use of the patch-clamp technique to record somatic voltage signals under different experimental conditions. The experimental data are then used to fit the many parameters of the model. While patching of the soma is currently the gold-standard approach to build multicompartment models, several studies have also evidenced a richness of dynamics in dendritic and axonal sections. Recording from the soma alone makes it hard to observe and correctly parameterize the activity of nonsomatic compartments. In order to provide a richer set of data as input to multicompartment models, we here investigate the combination of somatic patch-clamp recordings with recordings of high-density microelectrode arrays (HD-MEAs). HD-MEAs enable the observation of extracellular potentials and neural activity of neuronal compartments at subcellular resolution. In this work, we introduce a novel framework to combine patch-clamp and HD-MEA data to construct multicompartment models. We first validate our method on a ground-truth model with known parameters and show that the use of features extracted from extracellular signals, in addition to intracellular ones, yields models enabling better fits than using intracellular features alone. We also demonstrate our procedure using experimental data by constructing cell models from in vitro cell cultures. The proposed multimodal fitting procedure has the potential to augment the modeling efforts of the computational neuroscience community and provide the field with neuronal models that are more realistic and can be better validated.

Epileptogenic effects of NMDAR antibodies in a passive transfer mouse model.

Most patients with N-methyl D-aspartate-receptor antibody encephalitis develop seizures but the epileptogenicity of the antibodies has not been investigated in vivo. Wireless electroencephalogram transmitters were implanted into 23 C57BL/6 mice before left lateral ventricle injection of antibody-positive (test) or healthy (control) immunoglobulin G. Mice were challenged 48 h later with a subthreshold dose (40 mg/kg) of the chemo-convulsant pentylenetetrazol and events recorded over 1 h. Seizures were assessed by video observation of each animal and the electroencephalogram by an automated seizure detection programme. No spontaneous seizures were seen with the antibody injections. However, after the pro-convulsant, the test mice (n = 9) had increased numbers of observed convulsive seizures (P = 0.004), a higher total seizure score (P = 0.003), and a higher number of epileptic 'spike' events (P = 0.023) than the control mice (n = 6). At post-mortem, surprisingly, the total number of N-methyl D-aspartate receptors did not differ between test and control mice, but in test mice the levels of immunoglobulin G bound to the left hippocampus were higher (P < 0.0001) and the level of bound immunoglobulin G correlated with the seizure scores (R(2) = 0.8, P = 0.04, n = 5). Our findings demonstrate the epileptogenicity of N-methyl D-aspartate receptor antibodies in vivo, and suggest that binding of immunoglobulin G either reduced synaptic localization of N-methyl D-aspartate receptors, or had a direct effect on receptor function, which could be responsible for seizure susceptibility in this acute short-term model.

Mechanical stimulation and electrophysiological monitoring at subcellular resolution reveals differential mechanosensation of neurons within networks.

A growing consensus that the brain is a mechanosensitive organ is driving the need for tools that mechanically stimulate and simultaneously record the electrophysiological response of neurons within neuronal networks. Here we introduce a synchronized combination of atomic force microscopy, high-density microelectrode array and fluorescence microscopy to monitor neuronal networks and to mechanically characterize and stimulate individual neurons at piconewton force sensitivity and nanometre precision while monitoring their electrophysiological activity at subcellular spatial and millisecond temporal resolution. No correlation is found between mechanical stiffness and electrophysiological activity of neuronal compartments. Furthermore, spontaneously active neurons show exceptional functional resilience to static mechanical compression of their soma. However, application of fast transient (∼500 ms) mechanical stimuli to the neuronal soma can evoke action potentials, which depend on the anchoring of neuronal membrane and actin cytoskeleton. Neurons show higher responsivity, including bursts of action potentials, to slower transient mechanical stimuli (∼60 s). Moreover, transient and repetitive application of the same compression modulates the neuronal firing rate. Seemingly, neuronal networks can differentiate and respond to specific characteristics of mechanical stimulation. Ultimately, the developed multiparametric tool opens the door to explore manifold nanomechanobiological responses of neuronal systems and new ways of mechanical control.

A CMOS-based highly scalable flexible neural electrode interface.

Perception, thoughts, and actions are encoded by the coordinated activity of large neuronal populations spread over large areas. However, existing electrophysiological devices are limited by their scalability in capturing this cortex-wide activity. Here, we developed an electrode connector based on an ultra-conformable thin-film electrode array that self-assembles onto silicon microelectrode arrays enabling multithousand channel counts at a millimeter scale. The interconnects are formed using microfabricated electrode pads suspended by thin support arms, termed Flex2Chip. Capillary-assisted assembly drives the pads to deform toward the chip surface, and van der Waals forces maintain this deformation, establishing Ohmic contact. Flex2Chip arrays successfully measured extracellular action potentials ex vivo and resolved micrometer scale seizure propagation trajectories in epileptic mice. We find that seizure dynamics in absence epilepsy in the Scn8a+/- model do not have constant propagation trajectories.

Protosequences in human cortical organoids model intrinsic states in the developing cortex.

Neuronal firing sequences are thought to be the basic building blocks of neural coding and information broadcasting within the brain. However, when sequences emerge during neurodevelopment remains unknown. We demonstrate that structured firing sequences are present in spontaneous activity of human brain organoids and ex vivo neonatal brain slices from the murine somatosensory cortex. We observed a balance between temporally rigid and flexible firing patterns that are emergent phenomena in human brain organoids and early postnatal murine somatosensory cortex, but not in primary dissociated cortical cultures. Our findings suggest that temporal sequences do not arise in an experience-dependent manner, but are rather constrained by an innate preconfigured architecture established during neurogenesis. These findings highlight the potential for brain organoids to further explore how exogenous inputs can be used to refine neuronal circuits and enable new studies into the genetic mechanisms that govern assembly of functional circuitry during early human brain development.

Profound desensitization by ambient GABA limits activation of δ-containing GABAA receptors during spillover.

High-affinity extrasynaptic GABA(A) receptors (GABA(A)Rs) are a prominent feature of cerebellar granule neurons and thalamic relay neurons. In both cell types, the presence of synaptic glomeruli would be expected to promote activation of these GABA(A)Rs, contributing to phasic spillover-mediated currents and tonic inhibition. However, the precise role of different receptor subtypes in these two phenomena is unclear. To address this question, we made recordings from neurons in acute brain slices from mice, and from tsA201 cells expressing recombinant GABA(A)Rs. We found that δ subunit-containing GABA(A)Rs of both cerebellar granule neurons and thalamic relay neurons of the lateral geniculate nucleus contributed to tonic conductance caused by ambient GABA but not to spillover-mediated currents. In the presence of a low "ambient" GABA concentration, recombinant "extrasynaptic" δ subunit-containing GABA(A)Rs exhibited profound desensitization, rendering them insensitive to brief synaptic- or spillover-like GABA transients. Together, our results demonstrate that phasic spillover and tonic inhibition reflect the activation of distinct receptor populations.

Tracking axon initial segment plasticity using high-density microelectrode arrays: A computational study.

Despite being composed of highly plastic neurons with extensive positive feedback, the nervous system maintains stable overall function. To keep activity within bounds, it relies on a set of negative feedback mechanisms that can induce stabilizing adjustments and that are collectively termed "homeostatic plasticity." Recently, a highly excitable microdomain, located at the proximal end of the axon-the axon initial segment (AIS)-was found to exhibit structural modifications in response to activity perturbations. Though AIS plasticity appears to serve a homeostatic purpose, many aspects governing its expression and its functional role in regulating neuronal excitability remain elusive. A central challenge in studying the phenomenon is the rich heterogeneity of its expression (distal/proximal relocation, shortening, lengthening) and the variability of its functional role. A potential solution is to track AISs of a large number of neurons over time and attempt to induce structural plasticity in them. To this end, a promising approach is to use extracellular electrophysiological readouts to track a large number of neurons at high spatiotemporal resolution by means of high-density microelectrode arrays (HD-MEAs). However, an analysis framework that reliably identifies specific activity signatures that uniquely map on to underlying microstructural changes is missing. In this study, we assessed the feasibility of such a task and used the distal relocation of the AIS as an exemplary problem. We used sophisticated computational models to systematically explore the relationship between incremental changes in AIS positions and the specific consequences observed in simulated extracellular field potentials. An ensemble of feature changes in the extracellular fields that reliably characterize AIS plasticity was identified. We trained models that could detect these signatures with remarkable accuracy. Based on these findings, we propose a hybrid analysis framework that could potentially enable high-throughput experimental studies of activity-dependent AIS plasticity using HD-MEAs.

Inferring monosynaptic connections from paired dendritic spine Ca2+imaging and large-scale recording of extracellular spiking.

Objective: Techniques to identify monosynaptic connections between neurons have been vital for neuroscience research, facilitating important advancements concerning network topology, synaptic plasticity, and synaptic integration, among others.Approach: Here, we introduce a novel approach to identify and monitor monosynaptic connections using high-resolution dendritic spine Ca2+imaging combined with simultaneous large-scale recording of extracellular electrical activity by means of high-density microelectrode arrays.Main results: We introduce an easily adoptable analysis pipeline that associates the imaged spine with its presynaptic unit and test it onin vitrorecordings. The method is further validated and optimized by simulating synaptically-evoked spine Ca2+transients based on measured spike trains in order to obtain simulated ground-truth connections.Significance: The proposed approach offers unique advantages as (a) it can be used to identify monosynaptic connections with an accurate localization of the synapse within the dendritic tree, (b) it provides precise information of presynaptic spiking, and (c) postsynaptic spine Ca2+signals and, finally, (d) the non-invasive nature of the proposed method allows for long-term measurements. The analysis toolkit together with the rich data sets that were acquired are made publicly available for further exploration by the research community.

Predicting in vitro single-neuron firing rates upon pharmacological perturbation using Graph Neural Networks.

Modern Graph Neural Networks (GNNs) provide opportunities to study the determinants underlying the complex activity patterns of biological neuronal networks. In this study, we applied GNNs to a large-scale electrophysiological dataset of rodent primary neuronal networks obtained by means of high-density microelectrode arrays (HD-MEAs). HD-MEAs allow for long-term recording of extracellular spiking activity of individual neurons and networks and enable the extraction of physiologically relevant features at the single-neuron and population level. We employed established GNNs to generate a combined representation of single-neuron and connectivity features obtained from HD-MEA data, with the ultimate goal of predicting changes in single-neuron firing rate induced by a pharmacological perturbation. The aim of the main prediction task was to assess whether single-neuron and functional connectivity features, inferred under baseline conditions, were informative for predicting changes in neuronal activity in response to a perturbation with Bicuculline, a GABA A receptor antagonist. Our results suggest that the joint representation of node features and functional connectivity, extracted from a baseline recording, was informative for predicting firing rate changes of individual neurons after the perturbation. Specifically, our implementation of a GNN model with inductive learning capability (GraphSAGE) outperformed other prediction models that relied only on single-neuron features. We tested the generalizability of the results on two additional datasets of HD-MEA recordings-a second dataset with cultures perturbed with Bicuculline and a dataset perturbed with the GABA A receptor antagonist Gabazine. GraphSAGE models showed improved prediction accuracy over other prediction models. Our results demonstrate the added value of taking into account the functional connectivity between neurons and the potential of GNNs to study complex interactions between neurons.

Cortical Up states induce the selective weakening of subthreshold synaptic inputs.

Slow-wave sleep is thought to be important for retuning cortical synapses, but the cellular mechanisms remain unresolved. During slow-wave activity, cortical neurons display synchronized transitions between depolarized Up states and hyperpolarized Down states. Here, using recordings from LIII pyramidal neurons from acute slices of mouse medial entorhinal cortex, we find that subthreshold inputs arriving during the Up state undergo synaptic weakening. This does not reflect a process of global synaptic downscaling, as it is dependent on presynaptic spiking, with network state encoded in the synaptically evoked spine Ca2+ responses. Our data indicate that the induction of synaptic weakening is under postsynaptic control, as it can be prevented by correlated postsynaptic spiking activity, and depends on postsynaptic NMDA receptors and GSK3ß activity. This provides a mechanism by which slow-wave activity might bias synapses towards weakening, while preserving the synaptic connections within active neuronal assemblies.Slow oscillations between cortical Up and Down states are a defining feature of deep sleep, but their function is not well understood. Here the authors study Up/Down states in acute slices of entorhinal cortex, and find that Up states promote the weakening of subthreshold synaptic inputs, while suprathreshold inputs are preserved or strengthened.