BluePyOpt: Leveraging Open Source Software and Cloud Infrastructure to Optimise Model Parameters in Neuroscience

Cell-class-specific electric field entrainment of neural activity

Electric fields affect the activity of neurons and brain circuits, yet how this happens at the cellular level remains enigmatic. Lack of understanding of how to stimulate the brain to promote or suppress specific activity significantly limits basic research and clinical applications. Here, we study how electric fields impact subthreshold and spiking properties of major cortical neuronal classes. We find that neurons in the rodent and human cortex exhibit strong, cell-class-dependent entrainment that depends on stimulation frequency. Excitatory pyramidal neurons, with their slower spike rate, entrain to both slow and fast electric fields, while inhibitory classes like Pvalb and Sst (with their fast spiking) predominantly phase-lock to fast fields. We show that this spike-field entrainment is the result of two effects: non-specific membrane polarization occurring across classes and class-specific excitability properties. Importantly, these properties are present across cortical areas and species. These findings allow for the design of selective and class-specific neuromodulation.

Impact of Extracellular Current Flow on Action Potential Propagation in Myelinated Axons

Myelinated axons conduct action potentials, or spikes, in a saltatory manner. Inward current caused by a spike occurring at one node of Ranvier spreads axially to the next node, which regenerates the spike when depolarized enough for voltage-gated sodium channels to activate, and so on. The rate at which this process progresses dictates the velocity at which the spike is conducted and depends on several factors including axial resistivity and axon diameter that directly affect axial current. Here we show through computational simulations in modified double-cable axon models that conduction velocity also depends on extracellular factors whose effects can be explained by their indirect influence on axial current. Specifically, we show that a conventional double-cable model, with its outside layer connected to ground, transmits less axial current than a model whose outside layer is less absorptive. A more resistive barrier exists when an axon is packed tightly between other myelinated fibers, for example. We show that realistically resistive boundary conditions can significantly increase the velocity and energy efficiency of spike propagation, while also protecting against propagation failure. Certain factors like myelin thickness may be less important than typically thought if extracellular conditions are more resistive than normally considered. We also show how realistically resistive boundary conditions affect ephaptic interactions. Overall, these results highlight the unappreciated importance of extracellular conditions for axon function.

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.

Complex biophysical changes and reduced neuronal firing in an SCN8A variant associated with developmental delay and epilepsy

Mutations in the SCN8A gene, encoding the voltage-gated sodium channel NaV1.6, are associated with a range of neurodevelopmental syndromes. The p.(Gly1625Arg) (G1625R) mutation was identified in a patient diagnosed with developmental epileptic encephalopathy (DEE). While most of the characterized DEE-associated SCN8A mutations were shown to cause a gain-of-channel function, we show that the G1625R variant, positioned within the S4 segment of domain IV, results in complex effects. Voltage-clamp analyses of NaV1.6G1625R demonstrated a mixture of gain- and loss-of-function properties, including reduced current amplitudes, increased time constant of fast voltage-dependent inactivation, a depolarizing shift in the voltage dependence of activation and inactivation, and increased channel availability with high-frequency repeated depolarization. Current-clamp analyses in transfected cultured neurons revealed that these biophysical properties caused a marked reduction in the number of action potentials when firing was driven by the transfected mutant NaV1.6. Accordingly, computational modeling of mature cortical neurons demonstrated a mild decrease in neuronal firing when mimicking the patients' heterozygous SCN8A expression. Structural modeling of NaV1.6G1625R suggested the formation of a cation-π interaction between R1625 and F1588 within domain IV. Double-mutant cycle analysis revealed that this interaction affects the voltage dependence of inactivation in NaV1.6G1625R. Together, our studies demonstrate that the G1625R variant leads to a complex combination of gain and loss of function biophysical changes that result in an overall mild reduction in neuronal firing, related to the perturbed interaction network within the voltage sensor domain, necessitating personalized multi-tiered analysis for SCN8A mutations for optimal treatment selection.

Network-neuron interactions underlying sensory responses of layer 5 pyramidal tract neurons in barrel cortex

Neurons in the cerebral cortex receive thousands of synaptic inputs per second from thousands of presynaptic neurons. How the dendritic location of inputs, their timing, strength, and presynaptic origin, in conjunction with complex dendritic physiology, impact the transformation of synaptic input into action potential (AP) output remains generally unknown for in vivo conditions. Here, we introduce a computational approach to reveal which properties of the input causally underlie AP output, and how this neuronal input-output computation is influenced by the morphology and biophysical properties of the dendrites. We demonstrate that this approach allows dissecting of how different input populations drive in vivo observed APs. For this purpose, we focus on fast and broadly tuned responses that pyramidal tract neurons in layer 5 (L5PTs) of the rat barrel cortex elicit upon passive single whisker deflections. By reducing a multi-scale model that we reported previously, we show that three features are sufficient to predict with high accuracy the sensory responses and receptive fields of L5PTs under these specific in vivo conditions: the count of active excitatory versus inhibitory synapses preceding the response, their spatial distribution on the dendrites, and the AP history. Based on these three features, we derive an analytically tractable description of the input-output computation of L5PTs, which enabled us to dissect how synaptic input from thalamus and different cell types in barrel cortex contribute to these responses. We show that the input-output computation is preserved across L5PTs despite morphological and biophysical diversity of their dendrites. We found that trial-to-trial variability in L5PT responses, and cell-to-cell variability in their receptive fields, are sufficiently explained by variability in synaptic input from the network, whereas variability in biophysical and morphological properties have minor contributions. Our approach to derive analytically tractable models of input-output computations in L5PTs provides a roadmap to dissect network-neuron interactions underlying L5PT responses across different in vivo conditions and for other cell types.

In-silico testing of new pharmacology for restoring inhibition and human cortical function in depression

Reduced inhibition by somatostatin-expressing interneurons is associated with depression. Administration of positive allosteric modulators of α5 subunit-containing GABAA receptor (α5-PAM) that selectively target this lost inhibition exhibit antidepressant and pro-cognitive effects in rodent models of chronic stress. However, the functional effects of α5-PAM on the human brain in vivo are unknown, and currently cannot be assessed experimentally. We modeled the effects of α5-PAM on tonic inhibition as measured in human neurons, and tested in silico α5-PAM effects on detailed models of human cortical microcircuits in health and depression. We found that α5-PAM effectively recovered impaired cortical processing as quantified by stimulus detection metrics, and also recovered the power spectral density profile of the microcircuit EEG signals. We performed an α5-PAM dose-response and identified simulated EEG biomarker candidates. Our results serve to de-risk and facilitate α5-PAM translation and provide biomarkers in non-invasive brain signals for monitoring target engagement and drug efficacy.

Human Purkinje cells outperform mouse Purkinje cells in dendritic complexity and computational capacity

Purkinje cells in the cerebellum are among the largest neurons in the brain and have been extensively investigated in rodents. However, their morphological and physiological properties remain poorly understood in humans. In this study, we utilized high-resolution morphological reconstructions and unique electrophysiological recordings of human Purkinje cells ex vivo to generate computational models and estimate computational capacity. An inter-species comparison showed that human Purkinje cell had similar fractal structures but were larger than those of mouse Purkinje cells. Consequently, given a similar spine density (2/μm), human Purkinje cell hosted approximately 7.5 times more dendritic spines than those of mice. Moreover, human Purkinje cells had a higher dendritic complexity than mouse Purkinje cells and usually emitted 2-3 main dendritic trunks instead of one. Intrinsic electro-responsiveness was similar between the two species, but model simulations revealed that the dendrites could process ~6.5 times (n = 51 vs. n = 8) more input patterns in human Purkinje cells than in mouse Purkinje cells. Thus, while human Purkinje cells maintained spike discharge properties similar to those of rodents during evolution, they developed more complex dendrites, enhancing computational capacity.

Cell-Type Specific Connectivity of Whisker-Related Sensory and Motor Cortical Input to Dorsal Striatum

The anterior dorsolateral striatum (DLS) is heavily innervated by convergent excitatory projections from the primary motor (M1) and sensory cortex (S1) and considered an important site of sensorimotor integration. M1 and S1 corticostriatal synapses have functional differences in their connection strength with striatal spiny projection neurons (SPNs) and fast-spiking interneurons (FSIs) in the DLS and, as a result, exert distinct influences on sensory-guided behaviors. In the present study, we tested whether M1 and S1 inputs exhibit differences in the subcellular anatomical distribution of striatal neurons. We injected adeno-associated viral vectors encoding spaghetti monster fluorescent proteins (sm.FPs) into M1 and S1 in male and female mice and used confocal microscopy to generate 3D reconstructions of corticostriatal inputs to single identified SPNs and FSIs obtained through ex vivo patch clamp electrophysiology. We found that M1 and S1 dually innervate SPNs and FSIs; however, there is a consistent bias towards the M1 input in SPNs that is not found in FSIs. In addition, M1 and S1 inputs were distributed similarly across the proximal, medial, and distal regions of SPN and FSI dendrites. Notably, closely localized M1 and S1 clusters of inputs were more prevalent in SPNs than FSIs, suggesting that cortical inputs are integrated through cell-type specific mechanisms. Our results suggest that the stronger functional connectivity from M1 to SPNs compared to S1, as previously observed, is due to a higher quantity of synaptic inputs. Our results have implications for how sensorimotor integration is performed in the striatum through cell-specific differences in corticostriatal connections.

Altered synaptic plasticity at hippocampal CA1-CA3 synapses in Alzheimer's disease: integration of amyloid precursor protein intracellular domain and amyloid beta effects into computational models

Alzheimer's disease (AD) is a progressive memory loss and cognitive dysfunction brain disorder brought on by the dysfunctional amyloid precursor protein (APP) processing and clearance of APP peptides. Increased APP levels lead to the production of AD-related peptides including the amyloid APP intracellular domain (AICD) and amyloid beta (Aβ), and consequently modify the intrinsic excitability of the hippocampal CA1 pyramidal neurons, synaptic protein activity, and impair synaptic plasticity at hippocampal CA1-CA3 synapses. The goal of the present study is to build computational models that incorporate the effect of AD-related peptides on CA1 pyramidal neuron and hippocampal synaptic plasticity under the AD conditions and investigate the potential pharmacological treatments that could normalize hippocampal synaptic plasticity and learning in AD. We employ a phenomenological N-methyl-D-aspartate (NMDA) receptor-based voltage-dependent synaptic plasticity model that includes the separate receptor contributions on long-term potentiation (LTP) and long-term depression (LTD) and embed it into the a detailed compartmental model of CA1 pyramidal neuron. Modeling results show that partial blockade of Glu2NB-NMDAR-gated channel restores intrinsic excitability of a CA1 pyramidal neuron and rescues LTP in AICD and Aβ conditions. The model provides insight into the complex interactions in AD pathophysiology and suggests the conditions under which the synchronous activation of a cluster of synaptic inputs targeting the dendritic tree of CA1 pyramidal neuron leads to restored synaptic plasticity.

Different responses of mice and rats hippocampus CA1 pyramidal neurons to in vitro and in vivo-like inputs

The fundamental role of any neuron within a network is to transform complex spatiotemporal synaptic input patterns into individual output spikes. These spikes, in turn, act as inputs for other neurons in the network. Neurons must execute this function across a diverse range of physiological conditions, often based on species-specific traits. Therefore, it is crucial to determine the extent to which findings can be extrapolated between species and, ultimately, to humans. In this study, we employed a multidisciplinary approach to pinpoint the factors accounting for the observed electrophysiological differences between mice and rats, the two species most used in experimental and computational research. After analyzing the morphological properties of their hippocampal CA1 pyramidal cells, we conducted a statistical comparison of rat and mouse electrophysiological features in response to somatic current injections. This analysis aimed to uncover the parameters underlying these distinctions. Using a well-established computational workflow, we created ten distinct single-cell computational models of mouse CA1 pyramidal neurons, ready to be used in a full-scale hippocampal circuit. By comparing their responses to a variety of somatic and synaptic inputs with those of rat models, we generated experimentally testable hypotheses regarding species-specific differences in ion channel distribution, kinetics, and the electrophysiological mechanisms underlying their distinct responses to synaptic inputs during the behaviorally relevant Gamma and Sharp-Wave rhythms.

Modeling Synaptic Integration of Bursty and β Oscillatory Inputs in Ventromedial Motor Thalamic Neurons in Normal and Parkinsonian States

The ventromedial motor thalamus (VM) is implicated in multiple motor functions and occupies a central position in the cortico-basal ganglia-thalamocortical loop. It integrates glutamatergic inputs from motor cortex (MC) and motor-related subcortical areas, and it is a major recipient of inhibition from basal ganglia. Previous in vitro experiments performed in mice showed that dopamine depletion enhances the excitability of thalamocortical (TC) neurons in VM due to reduced M-type potassium currents. To understand how these excitability changes impact synaptic integration in vivo, we constructed biophysically detailed mouse VM TC model neurons fit to normal and dopamine-depleted conditions, using the NEURON simulator. These models allowed us to assess the influence of excitability changes with dopamine depletion on the integration of synaptic inputs expected in vivo We found that VM neuron models in the dopamine-depleted state showed increased firing rates with the same synaptic inputs. Synchronous bursting in inhibitory input from the substantia nigra pars reticulata (SNR), as observed in parkinsonian conditions, evoked a postinhibitory firing rate increase with a longer duration in dopamine-depleted than control conditions, due to different M-type potassium channel densities. With β oscillations in the inhibitory inputs from SNR and the excitatory inputs from cortex, we observed spike-phase locking in the activity of the models in normal and dopamine-depleted states, which relayed and amplified the oscillations of the inputs, suggesting that the increased β oscillations observed in VM of parkinsonian animals are predominantly a consequence of changes in the presynaptic activity rather than changes in intrinsic properties.

Controlling morpho-electrophysiological variability of neurons with detailed biophysical models

Variability, which is known to be a universal feature among biological units such as neuronal cells, holds significant importance, as, for example, it enables a robust encoding of a high volume of information in neuronal circuits and prevents hypersynchronizations. While most computational studies on electrophysiological variability in neuronal circuits were done with single-compartment neuron models, we instead focus on the variability of detailed biophysical models of neuron multi-compartmental morphologies. We leverage a Markov chain Monte Carlo method to generate populations of electrical models reproducing the variability of experimental recordings while being compatible with a set of morphologies to faithfully represent specifi morpho-electrical type. We demonstrate our approach on layer 5 pyramidal cells and study the morpho-electrical variability and in particular, find that morphological variability alone is insufficient to reproduce electrical variability. Overall, this approach provides a strong statistical basis to create detailed models of neurons with controlled variability.

A universal workflow for creation, validation, and generalization of detailed neuronal models

Detailed single-neuron modeling is widely used to study neuronal functions. While cellular and functional diversity across the mammalian cortex is vast, most of the available computational tools focus on a limited set of specific features characteristic of a single neuron. Here, we present a generalized automated workflow for the creation of robust electrical models and illustrate its performance by building cell models for the rat somatosensory cortex. Each model is based on a 3D morphological reconstruction and a set of ionic mechanisms. We use an evolutionary algorithm to optimize neuronal parameters to match the electrophysiological features extracted from experimental data. Then we validate the optimized models against additional stimuli and assess their generalizability on a population of similar morphologies. Compared to the state-of-the-art canonical models, our models show 5-fold improved generalizability. This versatile approach can be used to build robust models of any neuronal type.

Online interoperable resources for building hippocampal neuron models via the Hippocampus Hub

To build biophysically detailed models of brain cells, circuits, and regions, a data-driven approach is increasingly being adopted. This helps to obtain a simulated activity that reproduces the experimentally recorded neural dynamics as faithfully as possible, and to turn the model into a useful framework for making predictions based on the principles governing the nature of neural cells. In such a context, the access to existing neural models and data outstandingly facilitates the work of computational neuroscientists and fosters its novelty, as the scientific community grows wider and neural models progressively increase in type, size, and number. Nonetheless, even when accessibility is guaranteed, data and models are rarely reused since it is difficult to retrieve, extract and/or understand relevant information and scientists are often required to download and modify individual files, perform neural data analysis, optimize model parameters, and run simulations, on their own and with their own resources. While focusing on the construction of biophysically and morphologically accurate models of hippocampal cells, we have created an online resource, the Build section of the Hippocampus Hub -a scientific portal for research on the hippocampus- that gathers data and models from different online open repositories and allows their collection as the first step of a single cell model building workflow. Interoperability of tools and data is the key feature of the work we are presenting. Through a simple click-and-collect procedure, like filling the shopping cart of an online store, researchers can intuitively select the files of interest (i.e., electrophysiological recordings, neural morphology, and model components), and get started with the construction of a data-driven hippocampal neuron model. Such a workflow importantly includes a model optimization process, which leverages high performance computing resources transparently granted to the users, and a framework for running simulations of the optimized model, both available through the EBRAINS Hodgkin-Huxley Neuron Builder online tool.

Structural and functional specializations of human fast-spiking neurons support fast cortical signaling

Fast-spiking interneurons (FSINs) provide fast inhibition that synchronizes neuronal activity and is critical for cognitive function. Fast synchronization frequencies are evolutionary conserved in the expanded human neocortex despite larger neuron-to-neuron distances that challenge fast input-output transfer functions of FSINs. Here, we test in human neurons from neurosurgery tissue, which mechanistic specializations of human FSINs explain their fast-signaling properties in human cortex. With morphological reconstructions, multipatch recordings, and biophysical modeling, we find that despite threefold longer dendritic path, human FSINs maintain fast inhibition between connected pyramidal neurons through several mechanisms: stronger synapse strength of excitatory inputs, larger dendrite diameter with reduced complexity, faster AP initiation, and faster and larger inhibitory output, while Na+ current activation/inactivation properties are similar. These adaptations underlie short input-output delays in fast inhibition of human pyramidal neurons through FSINs, explaining how cortical synchronization frequencies are conserved despite expanded and sparse network topology of human cortex.

Heterogeneous off-target impact of ion-channel deletion on intrinsic properties of hippocampal model neurons that self-regulate calcium

How do neurons that implement cell-autonomous self-regulation of calcium react to knockout of individual ion-channel conductances? To address this question, we used a heterogeneous population of 78 conductance-based models of hippocampal pyramidal neurons that maintained cell-autonomous calcium homeostasis while receiving theta-frequency inputs. At calcium steady-state, we individually deleted each of the 11 active ion-channel conductances from each model. We measured the acute impact of deleting each conductance (one at a time) by comparing intrinsic electrophysiological properties before and immediately after channel deletion. The acute impact of deleting individual conductances on physiological properties (including calcium homeostasis) was heterogeneous, depending on the property, the specific model, and the deleted channel. The underlying many-to-many mapping between ion channels and properties pointed to ion-channel degeneracy. Next, we allowed the other conductances (barring the deleted conductance) to evolve towards achieving calcium homeostasis during theta-frequency activity. When calcium homeostasis was perturbed by ion-channel deletion, post-knockout plasticity in other conductances ensured resilience of calcium homeostasis to ion-channel deletion. These results demonstrate degeneracy in calcium homeostasis, as calcium homeostasis in knockout models was implemented in the absence of a channel that was earlier involved in the homeostatic process. Importantly, in reacquiring homeostasis, ion-channel conductances and physiological properties underwent heterogenous plasticity (dependent on the model, the property, and the deleted channel), even introducing changes in properties that were not directly connected to the deleted channel. Together, post-knockout plasticity geared towards maintaining homeostasis introduced heterogenous off-target effects on several channels and properties, suggesting that extreme caution be exercised in interpreting experimental outcomes involving channel knockouts.

Bayesian inference of a spectral graph model for brain oscillations

The relationship between brain functional connectivity and structural connectivity has caught extensive attention of the neuroscience community, commonly inferred using mathematical modeling. Among many modeling approaches, spectral graph model (SGM) is distinctive as it has a closed-form solution of the wide-band frequency spectra of brain oscillations, requiring only global biophysically interpretable parameters. While SGM is parsimonious in parameters, the determination of SGM parameters is non-trivial. Prior works on SGM determine the parameters through a computational intensive annealing algorithm, which only provides a point estimate with no confidence intervals for parameter estimates. To fill this gap, we incorporate the simulation-based inference (SBI) algorithm and develop a Bayesian procedure for inferring the posterior distribution of the SGM parameters. Furthermore, using SBI dramatically reduces the computational burden for inferring the SGM parameters. We evaluate the proposed SBI-SGM framework on the resting-state magnetoencephalography recordings from healthy subjects and show that the proposed procedure has similar performance to the annealing algorithm in recovering power spectra and the spatial distribution of the alpha frequency band. In addition, we also analyze the correlations among the parameters and their uncertainty with the posterior distribution which cannot be done with annealing inference. These analyses provide a richer understanding of the interactions among biophysical parameters of the SGM. In general, the use of simulation-based Bayesian inference enables robust and efficient computations of generative model parameter uncertainties and may pave the way for the use of generative models in clinical translation applications.

Automated customization of large-scale spiking network models to neuronal population activity

Understanding brain function is facilitated by constructing computational models that accurately reproduce aspects of brain activity. Networks of spiking neurons capture the underlying biophysics of neuronal circuits, yet the dependence of their activity on model parameters is notoriously complex. As a result, heuristic methods have been used to configure spiking network models, which can lead to an inability to discover activity regimes complex enough to match large-scale neuronal recordings. Here we propose an automatic procedure, Spiking Network Optimization using Population Statistics (SNOPS), to customize spiking network models that reproduce the population-wide covariability of large-scale neuronal recordings. We first confirmed that SNOPS accurately recovers simulated neural activity statistics. Then, we applied SNOPS to recordings in macaque visual and prefrontal cortices and discovered previously unknown limitations of spiking network models. Taken together, SNOPS can guide the development of network models and thereby enable deeper insight into how networks of neurons give rise to brain function.

Overexpression of UCP4 in astrocytic mitochondria prevents multilevel dysfunctions in a mouse model of Alzheimer's disease

Alzheimer's disease (AD) is becoming increasingly prevalent worldwide. It represents one of the greatest medical challenges as no pharmacologic treatments are available to prevent disease progression. Astrocytes play crucial functions within neuronal circuits by providing metabolic and functional support, regulating interstitial solute composition, and modulating synaptic transmission. In addition to these physiological functions, growing evidence points to an essential role of astrocytes in neurodegenerative diseases like AD. Early-stage AD is associated with hypometabolism and oxidative stress. Contrary to neurons that are vulnerable to oxidative stress, astrocytes are particularly resistant to mitochondrial dysfunction and are therefore more resilient cells. In our study, we leveraged astrocytic mitochondrial uncoupling and examined neuronal function in the 3xTg AD mouse model. We overexpressed the mitochondrial uncoupling protein 4 (UCP4), which has been shown to improve neuronal survival in vitro. We found that this treatment efficiently prevented alterations of hippocampal metabolite levels observed in AD mice, along with hippocampal atrophy and reduction of basal dendrite arborization of subicular neurons. This approach also averted aberrant neuronal excitability observed in AD subicular neurons and preserved episodic-like memory in AD mice assessed in a spatial recognition task. These findings show that targeting astrocytes and their mitochondria is an effective strategy to prevent the decline of neurons facing AD-related stress at the early stages of the disease.

Strong and reliable synaptic communication between pyramidal neurons in adult human cerebral cortex

Synaptic transmission constitutes the primary mode of communication between neurons. It is extensively studied in rodent but not human neocortex. We characterized synaptic transmission between pyramidal neurons in layers 2 and 3 using neurosurgically resected human middle temporal gyrus (MTG, Brodmann area 21), which is part of the distributed language circuitry. We find that local connectivity is comparable with mouse layer 2/3 connections in the anatomical homologue (temporal association area), but synaptic connections in human are 3-fold stronger and more reliable (0% vs 25% failure rates, respectively). We developed a theoretical approach to quantify properties of spinous synapses showing that synaptic conductance and voltage change in human dendritic spines are 3-4-folds larger compared with mouse, leading to significant NMDA receptor activation in human unitary connections. This model prediction was validated experimentally by showing that NMDA receptor activation increases the amplitude and prolongs decay of unitary excitatory postsynaptic potentials in human but not in mouse connections. Since NMDA-dependent recurrent excitation facilitates persistent activity (supporting working memory), our data uncovers cortical microcircuit properties in human that may contribute to language processing in MTG.

Thalamic control of sensory processing and spindles in a biophysical somatosensory thalamoreticular circuit model of wakefulness and sleep

Thalamoreticular circuitry plays a key role in arousal, attention, cognition, and sleep spindles, and is linked to several brain disorders. A detailed computational model of mouse somatosensory thalamus and thalamic reticular nucleus has been developed to capture the properties of over 14,000 neurons connected by 6 million synapses. The model recreates the biological connectivity of these neurons, and simulations of the model reproduce multiple experimental findings in different brain states. The model shows that inhibitory rebound produces frequency-selective enhancement of thalamic responses during wakefulness. We find that thalamic interactions are responsible for the characteristic waxing and waning of spindle oscillations. In addition, we find that changes in thalamic excitability control spindle frequency and their incidence. The model is made openly available to provide a new tool for studying the function and dysfunction of the thalamoreticular circuitry in various brain states.

Reduced oriens-lacunosum/moleculare cell model identifies biophysical current balances for in vivo theta frequency spiking resonance

Conductance-based models have played an important role in the development of modern neuroscience. These mathematical models are powerful "tools" that enable theoretical explorations in experimentally untenable situations, and can lead to the development of novel hypotheses and predictions. With advances in cell imaging and computational power, multi-compartment models with morphological accuracy are becoming common practice. However, as more biological details are added, they make extensive explorations and analyses more challenging largely due to their huge computational expense. Here, we focus on oriens-lacunosum/moleculare (OLM) cell models. OLM cells can contribute to functionally relevant theta rhythms in the hippocampus by virtue of their ability to express spiking resonance at theta frequencies, but what characteristics underlie this is far from clear. We converted a previously developed detailed multi-compartment OLM cell model into a reduced single compartment model that retained biophysical fidelity with its underlying ion currents. We showed that the reduced OLM cell model can capture complex output that includes spiking resonance in in vivo-like scenarios as previously obtained with the multi-compartment model. Using the reduced model, we were able to greatly expand our in vivo-like scenarios. Applying spike-triggered average analyses, we were able to to determine that it is a combination of hyperpolarization-activated cation and muscarinic type potassium currents that specifically allow OLM cells to exhibit spiking resonance at theta frequencies. Further, we developed a robust Kalman Filtering (KF) method to estimate parameters of the reduced model in real-time. We showed that it may be possible to directly estimate conductance parameters from experiments since this KF method can reliably extract parameter values from model voltage recordings. Overall, our work showcases how the contribution of cellular biophysical current details could be determined and assessed for spiking resonance. As well, our work shows that it may be possible to directly extract these parameters from current clamp voltage recordings.

Introducing the Dendrify framework for incorporating dendrites to spiking neural networks

Computational modeling has been indispensable for understanding how subcellular neuronal features influence circuit processing. However, the role of dendritic computations in network-level operations remains largely unexplored. This is partly because existing tools do not allow the development of realistic and efficient network models that account for dendrites. Current spiking neural networks, although efficient, are usually quite simplistic, overlooking essential dendritic properties. Conversely, circuit models with morphologically detailed neuron models are computationally costly, thus impractical for large-network simulations. To bridge the gap between these two extremes and facilitate the adoption of dendritic features in spiking neural networks, we introduce Dendrify, an open-source Python package based on Brian 2. Dendrify, through simple commands, automatically generates reduced compartmental neuron models with simplified yet biologically relevant dendritic and synaptic integrative properties. Such models strike a good balance between flexibility, performance, and biological accuracy, allowing us to explore dendritic contributions to network-level functions while paving the way for developing more powerful neuromorphic systems.

Multi-modal characterization and simulation of human epileptic circuitry

Temporal lobe epilepsy is the fourth most common neurological disorder, with about 40% of patients not responding to pharmacological treatment. Increased cellular loss is linked to disease severity and pathological phenotypes such as heightened seizure propensity. While the hippocampus is the target of therapeutic interventions, the impact of the disease at the cellular level remains unclear. Here, we show that hippocampal granule cells change with disease progression as measured in living, resected hippocampal tissue excised from patients with epilepsy. We show that granule cells increase excitability and shorten response latency while also enlarging in cellular volume and spine density. Single-nucleus RNA sequencing combined with simulations ascribes the changes to three conductances: BK, Cav2.2, and Kir2.1. In a network model, we show that these changes related to disease progression bring the circuit into a more excitable state, while reversing them produces a less excitable, "early-disease-like" state.

Computational models of neurotransmission at cerebellar synapses unveil the impact on network computation

The neuroscientific field benefits from the conjoint evolution of experimental and computational techniques, allowing for the reconstruction and simulation of complex models of neurons and synapses. Chemical synapses are characterized by presynaptic vesicle cycling, neurotransmitter diffusion, and postsynaptic receptor activation, which eventually lead to postsynaptic currents and subsequent membrane potential changes. These mechanisms have been accurately modeled for different synapses and receptor types (AMPA, NMDA, and GABA) of the cerebellar cortical network, allowing simulation of their impact on computation. Of special relevance is short-term synaptic plasticity, which generates spatiotemporal filtering in local microcircuits and controls burst transmission and information flow through the network. Here, we present how data-driven computational models recapitulate the properties of neurotransmission at cerebellar synapses. The simulation of microcircuit models is starting to reveal how diverse synaptic mechanisms shape the spatiotemporal profiles of circuit activity and computation.

The EBRAINS Hodgkin-Huxley Neuron Builder: An online resource for building data-driven neuron models

In the last decades, brain modeling has been established as a fundamental tool for understanding neural mechanisms and information processing in individual cells and circuits at different scales of observation. Building data-driven brain models requires the availability of experimental data and analysis tools as well as neural simulation environments and, often, large scale computing facilities. All these components are rarely found in a comprehensive framework and usually require ad hoc programming. To address this, we developed the EBRAINS Hodgkin-Huxley Neuron Builder (HHNB), a web resource for building single cell neural models via the extraction of activity features from electrophysiological traces, the optimization of the model parameters via a genetic algorithm executed on high performance computing facilities and the simulation of the optimized model in an interactive framework. Thanks to its inherent characteristics, the HHNB facilitates the data-driven model building workflow and its reproducibility, hence fostering a collaborative approach to brain modeling.

Single-neuron models linking electrophysiology, morphology, and transcriptomics across cortical cell types

Which cell types constitute brain circuits is a fundamental question, but establishing the correspondence across cellular data modalities is challenging. Bio-realistic models allow probing cause-and-effect and linking seemingly disparate modalities. Here, we introduce a computational optimization workflow to generate 9,200 single-neuron models with active conductances. These models are based on 230 in vitro electrophysiological experiments followed by morphological reconstruction from the mouse visual cortex. We show that, in contrast to current belief, the generated models are robust representations of individual experiments and cortical cell types as defined via cellular electrophysiology or transcriptomics. Next, we show that differences in specific conductances predicted from the models reflect differences in gene expression supported by single-cell transcriptomics. The differences in model conductances, in turn, explain electrophysiological differences observed between the cortical subclasses. Our computational effort reconciles single-cell modalities that define cell types and enables causal relationships to be examined.

Combining hypothesis- and data-driven neuroscience modeling in FAIR workflows

Modeling in neuroscience occurs at the intersection of different points of view and approaches. Typically, hypothesis-driven modeling brings a question into focus so that a model is constructed to investigate a specific hypothesis about how the system works or why certain phenomena are observed. Data-driven modeling, on the other hand, follows a more unbiased approach, with model construction informed by the computationally intensive use of data. At the same time, researchers employ models at different biological scales and at different levels of abstraction. Combining these models while validating them against experimental data increases understanding of the multiscale brain. However, a lack of interoperability, transparency, and reusability of both models and the workflows used to construct them creates barriers for the integration of models representing different biological scales and built using different modeling philosophies. We argue that the same imperatives that drive resources and policy for data - such as the FAIR (Findable, Accessible, Interoperable, Reusable) principles - also support the integration of different modeling approaches. The FAIR principles require that data be shared in formats that are Findable, Accessible, Interoperable, and Reusable. Applying these principles to models and modeling workflows, as well as the data used to constrain and validate them, would allow researchers to find, reuse, question, validate, and extend published models, regardless of whether they are implemented phenomenologically or mechanistically, as a few equations or as a multiscale, hierarchical system. To illustrate these ideas, we use a classical synaptic plasticity model, the Bienenstock-Cooper-Munro rule, as an example due to its long history, different levels of abstraction, and implementation at many scales.

Pareto optimality, economy-effectiveness trade-offs and ion channel degeneracy: improving population modelling for single neurons

Neurons encounter unavoidable evolutionary trade-offs between multiple tasks. They must consume as little energy as possible while effectively fulfilling their functions. Cells displaying the best performance for such multi-task trade-offs are said to be Pareto optimal, with their ion channel configurations underpinning their functionality. Ion channel degeneracy, however, implies that multiple ion channel configurations can lead to functionally similar behaviour. Therefore, instead of a single model, neuroscientists often use populations of models with distinct combinations of ionic conductances. This approach is called population (database or ensemble) modelling. It remains unclear, which ion channel parameters in the vast population of functional models are more likely to be found in the brain. Here we argue that Pareto optimality can serve as a guiding principle for addressing this issue by helping to identify the subpopulations of conductance-based models that perform best for the trade-off between economy and functionality. In this way, the high-dimensional parameter space of neuronal models might be reduced to geometrically simple low-dimensional manifolds, potentially explaining experimentally observed ion channel correlations. Conversely, Pareto inference might also help deduce neuronal functions from high-dimensional Patch-seq data. In summary, Pareto optimality is a promising framework for improving population modelling of neurons and their circuits.

Modernizing the NEURON Simulator for Sustainability, Portability, and Performance

The need for reproducible, credible, multiscale biological modeling has led to the development of standardized simulation platforms, such as the widely-used NEURON environment for computational neuroscience. Developing and maintaining NEURON over several decades has required attention to the competing needs of backwards compatibility, evolving computer architectures, the addition of new scales and physical processes, accessibility to new users, and efficiency and flexibility for specialists. In order to meet these challenges, we have now substantially modernized NEURON, providing continuous integration, an improved build system and release workflow, and better documentation. With the help of a new source-to-source compiler of the NMODL domain-specific language we have enhanced NEURON's ability to run efficiently, via the CoreNEURON simulation engine, on a variety of hardware platforms, including GPUs. Through the implementation of an optimized in-memory transfer mechanism this performance optimized backend is made easily accessible to users, providing training and model-development paths from laptop to workstation to supercomputer and cloud platform. Similarly, we have been able to accelerate NEURON's reaction-diffusion simulation performance through the use of just-in-time compilation. We show that these efforts have led to a growing developer base, a simpler and more robust software distribution, a wider range of supported computer architectures, a better integration of NEURON with other scientific workflows, and substantially improved performance for the simulation of biophysical and biochemical models.

Scaling and Benchmarking an Evolutionary Algorithm for Constructing Biophysical Neuronal Models

Single neuron models are fundamental for computational modeling of the brain's neuronal networks, and understanding how ion channel dynamics mediate neural function. A challenge in defining such models is determining biophysically realistic channel distributions. Here, we present an efficient, highly parallel evolutionary algorithm for developing such models, named NeuroGPU-EA. NeuroGPU-EA uses CPUs and GPUs concurrently to simulate and evaluate neuron membrane potentials with respect to multiple stimuli. We demonstrate a logarithmic cost for scaling the stimuli used in the fitting procedure. NeuroGPU-EA outperforms the typically used CPU based evolutionary algorithm by a factor of 10 on a series of scaling benchmarks. We report observed performance bottlenecks and propose mitigation strategies. Finally, we also discuss the potential of this method for efficient simulation and evaluation of electrophysiological waveforms.

A calcium-based plasticity model for predicting long-term potentiation and depression in the neocortex

Pyramidal cells (PCs) form the backbone of the layered structure of the neocortex, and plasticity of their synapses is thought to underlie learning in the brain. However, such long-term synaptic changes have been experimentally characterized between only a few types of PCs, posing a significant barrier for studying neocortical learning mechanisms. Here we introduce a model of synaptic plasticity based on data-constrained postsynaptic calcium dynamics, and show in a neocortical microcircuit model that a single parameter set is sufficient to unify the available experimental findings on long-term potentiation (LTP) and long-term depression (LTD) of PC connections. In particular, we find that the diverse plasticity outcomes across the different PC types can be explained by cell-type-specific synaptic physiology, cell morphology and innervation patterns, without requiring type-specific plasticity. Generalizing the model to in vivo extracellular calcium concentrations, we predict qualitatively different plasticity dynamics from those observed in vitro. This work provides a first comprehensive null model for LTP/LTD between neocortical PC types in vivo, and an open framework for further developing models of cortical synaptic plasticity.

Exploring Parameter and Hyper-Parameter Spaces of Neuroscience Models on High Performance Computers With Learning to Learn

Neuroscience models commonly have a high number of degrees of freedom and only specific regions within the parameter space are able to produce dynamics of interest. This makes the development of tools and strategies to efficiently find these regions of high importance to advance brain research. Exploring the high dimensional parameter space using numerical simulations has been a frequently used technique in the last years in many areas of computational neuroscience. Today, high performance computing (HPC) can provide a powerful infrastructure to speed up explorations and increase our general understanding of the behavior of the model in reasonable times. Learning to learn (L2L) is a well-known concept in machine learning (ML) and a specific method for acquiring constraints to improve learning performance. This concept can be decomposed into a two loop optimization process where the target of optimization can consist of any program such as an artificial neural network, a spiking network, a single cell model, or a whole brain simulation. In this work, we present L2L as an easy to use and flexible framework to perform parameter and hyper-parameter space exploration of neuroscience models on HPC infrastructure. Learning to learn is an implementation of the L2L concept written in Python. This open-source software allows several instances of an optimization target to be executed with different parameters in an embarrassingly parallel fashion on HPC. L2L provides a set of built-in optimizer algorithms, which make adaptive and efficient exploration of parameter spaces possible. Different from other optimization toolboxes, L2L provides maximum flexibility for the way the optimization target can be executed. In this paper, we show a variety of examples of neuroscience models being optimized within the L2L framework to execute different types of tasks. The tasks used to illustrate the concept go from reproducing empirical data to learning how to solve a problem in a dynamic environment. We particularly focus on simulations with models ranging from the single cell to the whole brain and using a variety of simulation engines like NEST, Arbor, TVB, OpenAIGym, and NetLogo.

Cell type-specific mechanisms of information transfer in data-driven biophysical models of hippocampal CA3 principal neurons

The transformation of synaptic input into action potential output is a fundamental single-cell computation resulting from the complex interaction of distinct cellular morphology and the unique expression profile of ion channels that define the cellular phenotype. Experimental studies aimed at uncovering the mechanisms of the transfer function have led to important insights, yet are limited in scope by technical feasibility, making biophysical simulations an attractive complementary approach to push the boundaries in our understanding of cellular computation. Here we take a data-driven approach by utilizing high-resolution morphological reconstructions and patch-clamp electrophysiology data together with a multi-objective optimization algorithm to build two populations of biophysically detailed models of murine hippocampal CA3 pyramidal neurons based on the two principal cell types that comprise this region. We evaluated the performance of these models and find that our approach quantitatively matches the cell type-specific firing phenotypes and recapitulate the intrinsic population-level variability in the data. Moreover, we confirm that the conductance values found by the optimization algorithm are consistent with differentially expressed ion channel genes in single-cell transcriptomic data for the two cell types. We then use these models to investigate the cell type-specific biophysical properties involved in the generation of complex-spiking output driven by synaptic input through an information-theoretic treatment of their respective transfer functions. Our simulations identify a host of cell type-specific biophysical mechanisms that define the morpho-functional phenotype to shape the cellular transfer function and place these findings in the context of a role for bursting in CA3 recurrent network synchronization dynamics.

The hippocampus is comprised of numerous types of neurons, which constitute the cellular substrate for its rich repertoire of network dynamics. Among these are sharp waves, sequential activations of ensembles of neurons that have been shown to be crucially involved in learning and memory. In the CA3 area of the hippocampus, two types of excitatory cells, thorny and a-thorny neurons, are preferentially active during distinct phases of a sharp wave, suggesting a differential role for these cell types in phenomena such as memory consolidation. Using a strictly data-driven approach, we built biophysically realistic models of both thorny and a-thorny cells and used them to investigate the integrative differences between these two cell types. We found that both neuron classes have the capability of integrating incoming synaptic inputs in a supralinear fashion, although only a-thorny cells respond with bursts of action potentials to spatially and temporally clustered synaptic inputs. Additionally, by using a computational approach based on information theory, we show that, owing to this propensity for bursting, a-thorny cells can encode more information in their spiking output than their thorny counterpart. These results shed new light on the computational capabilities of two types of excitatory neurons and suggest that thorny and a-thorny cells may play distinct roles in the generation of hippocampal network synchronization.

Computational synthesis of cortical dendritic morphologies

Neuronal morphologies provide the foundation for the electrical behavior of neurons, the connectomes they form, and the dynamical properties of the brain. Comprehensive neuron models are essential for defining cell types, discerning their functional roles, and investigating brain-disease-related dendritic alterations. However, a lack of understanding of the principles underlying neuron morphologies has hindered attempts to computationally synthesize morphologies for decades. We introduce a synthesis algorithm based on a topological descriptor of neurons, which enables the rapid digital reconstruction of entire brain regions from few reference cells. This topology-guided synthesis generates dendrites that are statistically similar to biological reconstructions in terms of morpho-electrical and connectivity properties and offers a significant opportunity to investigate the links between neuronal morphology and brain function across different spatiotemporal scales. Synthesized cortical networks based on structurally altered dendrites associated with diverse brain pathologies revealed principles linking branching properties to the structure of large-scale networks.

Statistical Emulation of Neural Simulators: Application to Neocortical L2/3 Large Basket Cells

Many scientific systems are studied using computer codes that simulate the phenomena of interest. Computer simulation enables scientists to study a broad range of possible conditions, generating large quantities of data at a faster rate than the laboratory. Computer models are widespread in neuroscience, where they are used to mimic brain function at different levels. These models offer a variety of new possibilities for the neuroscientist, but also numerous challenges, such as: where to sample the input space for the simulator, how to make sense of the data that is generated, and how to estimate unknown parameters in the model. Statistical emulation can be a valuable complement to simulator-based research. Emulators are able to mimic the simulator, often with a much smaller computational burden and they are especially valuable for parameter estimation, which may require many simulator evaluations. This work compares different statistical models that address these challenges, and applies them to simulations of neocortical L2/3 large basket cells, created and run with the NEURON simulator in the context of the European Human Brain Project. The novelty of our approach is the use of fast empirical emulators, which have the ability to accelerate the optimization process for the simulator and to identify which inputs (in this case, different membrane ion channels) are most influential in affecting simulated features. These contributions are complementary, as knowledge of the important features can further improve the optimization process. Subsequent research, conducted after the process is completed, will gain efficiency by focusing on these inputs.

Hippocampal sharp wave-ripples and the associated sequence replay emerge from structured synaptic interactions in a network model of area CA3

Hippocampal place cells are activated sequentially as an animal explores its environment. These activity sequences are internally recreated (‘replayed’), either in the same or reversed order, during bursts of activity (sharp wave-ripples [SWRs]) that occur in sleep and awake rest. SWR-associated replay is thought to be critical for the creation and maintenance of long-term memory. In order to identify the cellular and network mechanisms of SWRs and replay, we constructed and simulated a data-driven model of area CA3 of the hippocampus. Our results show that the chain-like structure of recurrent excitatory interactions established during learning not only determines the content of replay, but is essential for the generation of the SWRs as well. We find that bidirectional replay requires the interplay of the experimentally confirmed, temporally symmetric plasticity rule, and cellular adaptation. Our model provides a unifying framework for diverse phenomena involving hippocampal plasticity, representations, and dynamics, and suggests that the structured neural codes induced by learning may have greater influence over cortical network states than previously appreciated.

NeuroGPU: Accelerating multi-compartment, biophysically detailed neuron simulations on GPUs

BACKGROUND

The membrane potential of individual neurons depends on a large number of interacting biophysical processes operating on spatial-temporal scales spanning several orders of magnitude. The multi-scale nature of these processes dictates that accurate prediction of membrane potentials in specific neurons requires the utilization of detailed simulations. Unfortunately, constraining parameters within biologically detailed neuron models can be difficult, leading to poor model fits. This obstacle can be overcome partially by numerical optimization or detailed exploration of parameter space. However, these processes, which currently rely on central processing unit (CPU) computation, often incur orders of magnitude increases in computing time for marginal improvements in model behavior. As a result, model quality is often compromised to accommodate compute resources.

NEW METHOD

Here, we present a simulation environment, NeuroGPU, that takes advantage of the inherent parallelized structure of the graphics processing unit (GPU) to accelerate neuronal simulation.

RESULTS & COMPARISON WITH EXISTING METHODS

NeuroGPU can simulate most biologically detailed models 10-200 times faster than NEURON simulation running on a single core and 5 times faster than GPU simulators (CoreNEURON). NeuroGPU is designed for model parameter tuning and best performs when the GPU is fully utilized by running multiple (> 100) instances of the same model with different parameters. When using multiple GPUs, NeuroGPU can reach to a speed-up of 800 fold compared to single core simulations, especially when simulating the same model morphology with different parameters. We demonstrate the power of NeuoGPU through large-scale parameter exploration to reveal the response landscape of a neuron. Finally, we accelerate numerical optimization of biophysically detailed neuron models to achieve highly accurate fitting of models to simulation and experimental data.

CONCLUSIONS

Thus, NeuroGPU is the fastest available platform that enables rapid simulation of multi-compartment, biophysically detailed neuron models on commonly used computing systems accessible by many scientists.

Reconstruction of the Hippocampus

The hippocampus is a widely studied brain region thought to play an important role in higher cognitive functions such as learning, memory, and navigation. The amount of data on this region increases every day and delineates a complex and fragmented picture, but an integrated understanding of hippocampal function remains elusive. Computational methods can help to move the research forward, and reconstructing a full-scale model of the hippocampus is a challenging yet feasible task that the research community should undertake.

In this chapter, we present strategies for reconstructing a large-scale model of the hippocampus. Based on a previously published approach to reconstruct and simulate brain tissue, which is also explained in Chap. 10, we discuss the characteristics of the hippocampus in the light of its special anatomical and physiological features, data availability, and existing large-scale hippocampus models. A large-scale model of the hippocampus is a compound model of several building blocks: ion channels, morphologies, single cell models, connections, synapses. We discuss each of those building blocks separately and discuss how to merge them back and simulate the resulting network model.

Realistic retinal modeling unravels the differential role of excitation and inhibition to starburst amacrine cells in direction selectivity

Retinal direction-selectivity originates in starburst amacrine cells (SACs), which display a centrifugal preference, responding with greater depolarization to a stimulus expanding from soma to dendrites than to a collapsing stimulus. Various mechanisms were hypothesized to underlie SAC centrifugal preference, but dissociating them is experimentally challenging and the mechanisms remain debatable. To address this issue, we developed the Retinal Stimulation Modeling Environment (RSME), a multifaceted data-driven retinal model that encompasses detailed neuronal morphology and biophysical properties, retina-tailored connectivity scheme and visual input. Using a genetic algorithm, we demonstrated that spatiotemporally diverse excitatory inputs–sustained in the proximal and transient in the distal processes–are sufficient to generate experimentally validated centrifugal preference in a single SAC. Reversing these input kinetics did not produce any centrifugal-preferring SAC. We then explored the contribution of SAC-SAC inhibitory connections in establishing the centrifugal preference. SAC inhibitory network enhanced the centrifugal preference, but failed to generate it in its absence. Embedding a direction selective ganglion cell (DSGC) in a SAC network showed that the known SAC-DSGC asymmetric connectivity by itself produces direction selectivity. Still, this selectivity is sharpened in a centrifugal-preferring SAC network. Finally, we use RSME to demonstrate the contribution of SAC-SAC inhibitory connections in mediating direction selectivity and recapitulate recent experimental findings. Thus, using RSME, we obtained a mechanistic understanding of SACs’ centrifugal preference and its contribution to direction selectivity.

Retinal direction selectivity is a canonical example for a computation undertaken by the retina. Starburst amacrine cells (SACs), interneurons in the retina, mediate direction selectivity via two mechanisms: they form asymmetric inhibitory connections with direction selective ganglion cells (DSGCs); and their processes are themselves direction selective, displaying a centrifugal preference. Various hypotheses were raised to account for this centrifugal preference, including the arrangement of SAC excitatory inputs, their kinetics, as well as reciprocal inhibition between SACs. To address this, we developed the Retinal Stimulation Modeling Environment (RSME)–a modeling environment for highly detailed, biologically plausible simulations, tailored to the exploration of neuronal dynamic and visual processing in retinal circuits. We started with exploring the excitation to a single SAC, and found that a precise organization of the input kinetics along SAC processes can generate a centrifugal preference that matched our experimental recordings. We then generated a network of SACs and found that reciprocal inhibition between SACs further enhances the centrifugal preference. Finally, we embedded a DSGC in the network, and dissected the contribution of SAC-DSGC asymmetric connections and SAC centrifugal preference to direction selectivity in DSGC.

Structure and function of axo-axonic inhibition

Inhibitory neurons in mammalian cortex exhibit diverse physiological, morphological, molecular, and connectivity signatures. While considerable work has measured the average connectivity of several interneuron classes, there remains a fundamental lack of understanding of the connectivity distribution of distinct inhibitory cell types with synaptic resolution, how it relates to properties of target cells, and how it affects function. Here, we used large-scale electron microscopy and functional imaging to address these questions for chandelier cells in layer 2/3 of the mouse visual cortex. With dense reconstructions from electron microscopy, we mapped the complete chandelier input onto 153 pyramidal neurons. We found that synapse number is highly variable across the population and is correlated with several structural features of the target neuron. This variability in the number of axo-axonic ChC synapses is higher than the variability seen in perisomatic inhibition. Biophysical simulations show that the observed pattern of axo-axonic inhibition is particularly effective in controlling excitatory output when excitation and inhibition are co-active. Finally, we measured chandelier cell activity in awake animals using a cell-type-specific calcium imaging approach and saw highly correlated activity across chandelier cells. In the same experiments, in vivo chandelier population activity correlated with pupil dilation, a proxy for arousal. Together, these results suggest that chandelier cells provide a circuit-wide signal whose strength is adjusted relative to the properties of target neurons.

Neuronal population models reveal specific linear conductance controllers sufficient to rescue preclinical disease phenotypes

Preclinical drug candidates are screened for their ability to ameliorate in vitro neuronal electrophysiology, and go/no-go decisions progress drugs to clinical trials based on population means across cells and animals. However, these measures do not mitigate clinical endpoint risk. Population-based modeling captures variability across multiple electrophysiological measures from healthy, disease, and drug phenotypes. We pursued optimizing therapeutic targets by identifying coherent sets of ion channel target modulations for recovering heterogeneous wild-type (WT) population excitability profiles from a heterogeneous Huntington's disease (HD) population. Our approach combines mechanistic simulations with population modeling of striatal neurons using evolutionary optimization algorithms to design 'virtual drugs'. We introduce efficacy metrics to score populations and rank virtual drug candidates. We found virtual drugs using heuristic approaches that performed better than single target modulators and standard classification methods. We compare a real drug to virtual candidates and demonstrate a novel in silico triaging method.

Dopaminergic and Cholinergic Modulation of Large Scale Networks in silico Using Snudda

Neuromodulation is present throughout the nervous system and serves a critical role for circuit function and dynamics. The computational investigations of neuromodulation in large scale networks require supportive software platforms. Snudda is a software for the creation and simulation of large scale networks of detailed microcircuits consisting of multicompartmental neuron models. We have developed an extension to Snudda to incorporate neuromodulation in large scale simulations. The extended Snudda framework implements neuromodulation at the level of single cells incorporated into large-scale microcircuits. We also developed Neuromodcell, a software for optimizing neuromodulation in detailed multicompartmental neuron models. The software adds parameters within the models modulating the conductances of ion channels and ionotropic receptors. Bath application of neuromodulators is simulated and models which reproduce the experimentally measured effects are selected. In Snudda, we developed an extension to accommodate large scale simulations of neuromodulation. The simulator has two modes of simulation – denoted replay and adaptive. In the replay mode, transient levels of neuromodulators can be defined as a time-varying function which modulates the receptors and ion channels within the network in a cell-type specific manner. In the adaptive mode, spiking neuromodulatory neurons are connected via integrative modulating mechanisms to ion channels and receptors. Both modes of simulating neuromodulation allow for simultaneous modulation by several neuromodulators that can interact dynamically with each other. Here, we used the Neuromodcell software to simulate dopaminergic and muscarinic modulation of neurons from the striatum. We also demonstrate how to simulate different neuromodulatory states with dopamine and acetylcholine using Snudda. All software is freely available on Github, including tutorials on Neuromodcell and Snudda-neuromodulation.

The Role of Hub Neurons in Modulating Cortical Dynamics

Many neurodegenerative diseases are associated with the death of specific neuron types in particular brain regions. What makes the death of specific neuron types particularly harmful for the integrity and dynamics of the respective network is not well understood. To start addressing this question we used the most up-to-date biologically realistic dense neocortical microcircuit (NMC) of the rodent, which has reconstructed a volume of 0.3 mm³ and containing 31,000 neurons, ∼37 million synapses, and 55 morphological cell types arranged in six cortical layers. Using modern network science tools, we identified hub neurons in the NMC, that are connected synaptically to a large number of their neighbors and systematically examined the impact of abolishing these cells. In general, the structural integrity of the network is robust to cells’ attack; yet, attacking hub neurons strongly impacted the small-world topology of the network, whereas similar attacks on random neurons have a negligible effect. Such hub-specific attacks are also impactful on the network dynamics, both when the network is at its spontaneous synchronous state and when it was presented with synchronized thalamo-cortical visual-like input. We found that attacking layer 5 hub neurons is most harmful to the structural and functional integrity of the NMC. The significance of our results for understanding the role of specific neuron types and cortical layers for disease manifestation is discussed.

The EBRAINS NeuroFeatureExtract: An Online Resource for the Extraction of Neural Activity Features From Electrophysiological Data

The description of neural dynamics, in terms of precise characterizations of action potential timings and shape and voltage related measures, is fundamental for a deeper understanding of the neural code and its information content. Not only such measures serve the scientific questions posed by experimentalists but are increasingly being used by computational neuroscientists for the construction of biophysically detailed data-driven models. Nonetheless, online resources enabling users to perform such feature extraction operation are lacking. To address this problem, in the framework of the Human Brain Project and the EBRAINS research infrastructure, we have developed and made available to the scientific community the NeuroFeatureExtract, an open-access online resource for the extraction of electrophysiological features from neural activity data. This tool allows to select electrophysiological traces of interest, fetched from public repositories or from users’ own data, and provides ad hoc functionalities to extract relevant features. The output files are properly formatted for further analysis, including data-driven neural model optimization.

Active Dendrites and Local Field Potentials: Biophysical Mechanisms and Computational Explorations

Neurons and glial cells are endowed with membranes that express a rich repertoire of ion channels, transporters, and receptors. The constant flux of ions across the neuronal and glial membranes results in voltage fluctuations that can be recorded from the extracellular matrix. The high frequency components of this voltage signal contain information about the spiking activity, reflecting the output from the neurons surrounding the recording location. The low frequency components of the signal, referred to as the local field potential (LFP), have been traditionally thought to provide information about the synaptic inputs that impinge on the large dendritic trees of various neurons. In this review, we discuss recent computational and experimental studies pointing to a critical role of several active dendritic mechanisms that can influence the genesis and the location-dependent spectro-temporal dynamics of LFPs, spanning different brain regions. We strongly emphasize the need to account for the several fast and slow dendritic events and associated active mechanisms - including gradients in their expression profiles, inter- and intra-cellular spatio-temporal interactions spanning neurons and glia, heterogeneities and degeneracy across scales, neuromodulatory influences, and activitydependent plasticity - towards gaining important insights about the origins of LFP under different behavioral states in health and disease. We provide simple but essential guidelines on how to model LFPs taking into account these dendritic mechanisms, with detailed methodology on how to account for various heterogeneities and electrophysiological properties of neurons and synapses while studying LFPs.

Predicting Synaptic Connectivity for Large-Scale Microcircuit Simulations Using Snudda

Simulation of large-scale networks of neurons is an important approach to understanding and interpreting experimental data from healthy and diseased brains. Owing to the rapid development of simulation software and the accumulation of quantitative data of different neuronal types, it is possible to predict both computational and dynamical properties of local microcircuits in a ‘bottom-up’ manner. Simulated data from these models can be compared with experiments and ‘top-down’ modelling approaches, successively bridging the scales. Here we describe an open source pipeline, using the software Snudda, for predicting microcircuit connectivity and for setting up simulations using the NEURON simulation environment in a reproducible way. We also illustrate how to further ‘curate’ data on single neuron morphologies acquired from public databases. This model building pipeline was used to set up a first version of a full-scale cellular level model of mouse dorsal striatum. Model components from that work are here used to illustrate the different steps that are needed when modelling subcortical nuclei, such as the basal ganglia.

On the Use of a Multimodal Optimizer for Fitting Neuron Models. Application to the Cerebellar Granule Cell

This article extends a recent methodological workflow for creating realistic and computationally efficient neuron models whilst capturing essential aspects of single-neuron dynamics. We overcome the intrinsic limitations of the extant optimization methods by proposing an alternative optimization component based on multimodal algorithms. This approach can natively explore a diverse population of neuron model configurations. In contrast to methods that focus on a single global optimum, the multimodal method allows directly obtaining a set of promising solutions for a single but complex multi-feature objective function. The final sparse population of candidate solutions has to be analyzed and evaluated according to the biological plausibility and their objective to the target features by the expert. In order to illustrate the value of this approach, we base our proposal on the optimization of cerebellar granule cell (GrC) models that replicate the essential properties of the biological cell. Our results show the emerging variability of plausible sets of values that this type of neuron can adopt underlying complex spiking characteristics. Also, the set of selected cerebellar GrC models captured spiking dynamics closer to the reference model than the single model obtained with off-the-shelf parameter optimization algorithms used in our previous article. The method hereby proposed represents a valuable strategy for adjusting a varied population of realistic and simplified neuron models. It can be applied to other kinds of neuron models and biological contexts.

Cellular Classes in the Human Brain Revealed In Vivo by Heartbeat-Related Modulation of the Extracellular Action Potential Waveform

Determining cell types is critical for understanding neural circuits but remains elusive in the living human brain. Current approaches discriminate units into putative cell classes using features of the extracellular action potential (EAP); in absence of ground truth data, this remains a problematic procedure. We find that EAPs in deep structures of the brain exhibit robust and systematic variability during the cardiac cycle. These cardiac-related features refine neural classification. We use these features to link bio-realistic models generated from in vitro human whole-cell recordings of morphologically classified neurons to in vivo recordings. We differentiate aspiny inhibitory and spiny excitatory human hippocampal neurons and, in a second stage, demonstrate that cardiac-motion features reveal two types of spiny neurons with distinct intrinsic electrophysiological properties and phase-locking characteristics to endogenous oscillations. This multi-modal approach markedly improves cell classification in humans, offers interpretable cell classes, and is applicable to other brain areas and species.

Stellate cell computational modeling predicts signal filtering in the molecular layer circuit of cerebellum

The functional properties of cerebellar stellate cells and the way they regulate molecular layer activity are still unclear. We have measured stellate cells electroresponsiveness and their activation by parallel fiber bursts. Stellate cells showed intrinsic pacemaking, along with characteristic responses to depolarization and hyperpolarization, and showed a marked short-term facilitation during repetitive parallel fiber transmission. Spikes were emitted after a lag and only at high frequency, making stellate cells to operate as delay-high-pass filters. A detailed computational model summarizing these physiological properties allowed to explore different functional configurations of the parallel fiber-stellate cell-Purkinje cell circuit. Simulations showed that, following parallel fiber stimulation, Purkinje cells almost linearly increased their response with input frequency, but such an increase was inhibited by stellate cells, which leveled the Purkinje cell gain curve to its 4 Hz value. When reciprocal inhibitory connections between stellate cells were activated, the control of stellate cells over Purkinje cell discharge was maintained only at very high frequencies. These simulations thus predict a new role for stellate cells, which could endow the molecular layer with low-pass and band-pass filtering properties regulating Purkinje cell gain and, along with this, also burst delay and the burst-pause responses pattern.