A nanofluidic spiking synapse.

Currently, the nanofluidic synapse can only perform basic neuromorphic pulse patterns. One immediate problem that needs to be addressed to further its capability of brain-like computing is the realization of a nanofluidic spiking device. Here, we report the use of a poly(3,4-ethylenedioxythiophene) polystyrene sulfonate membrane to achieve bionic ionic current-induced spiking. In addition to the simulation of various electrical pulse patterns, our synapse could produce transmembrane ionic current-induced spiking, which is highly analogous to biological action potentials with similar phases and excitability. Moreover, the spiking properties could be modulated by ions and neurochemicals. We expect that this work could contribute to biomimetic spiking computing in solution.

Prefrontal cortex neurons encode ambient light intensity differentially across regions and layers.

While light can affect emotional and cognitive processes of the medial prefrontal cortex (mPFC), no light-encoding was hitherto identified in this region. Here, extracellular recordings in awake mice revealed that over half of studied mPFC neurons showed photosensitivity, that was diminished by inhibition of intrinsically photosensitive retinal ganglion cells (ipRGCs), or of the upstream thalamic perihabenular nucleus (PHb). In 15% of mPFC photosensitive neurons, firing rate changed monotonically along light-intensity steps and gradients. These light-intensity-encoding neurons comprised four types, two enhancing and two suppressing their firing rate with increased light intensity. Similar types were identified in the PHb, where they exhibited shorter latency and increased sensitivity. Light suppressed prelimbic activity but boosted infralimbic activity, mirroring the regions' contrasting roles in fear-conditioning, drug-seeking, and anxiety. We posit that prefrontal photosensitivity represents a substrate of light-susceptible, mPFC-mediated functions, which could be ultimately studied as a therapeutical target in psychiatric and addiction disorders.

Local field potential sharp waves with diversified impact on cortical neuronal encoding of haptic input.

Cortical sensory processing is greatly impacted by internally generated activity. But controlling for that activity is difficult since the thalamocortical network is a high-dimensional system with rapid state changes. Therefore, to unwind the cortical computational architecture there is a need for physiological 'landmarks' that can be used as frames of reference for computational state. Here we use a waveshape transform method to identify conspicuous local field potential sharp waves (LFP-SPWs) in the somatosensory cortex (S1). LFP-SPW events triggered short-lasting but massive neuronal activation in all recorded neurons with a subset of neurons initiating their activation up to 20 ms before the LFP-SPW onset. In contrast, LFP-SPWs differentially impacted the neuronal spike responses to ensuing tactile inputs, depressing the tactile responses in some neurons and enhancing them in others. When LFP-SPWs coactivated with more distant cortical surface (ECoG)-SPWs, suggesting an involvement of these SPWs in global cortical signaling, the impact of the LFP-SPW on the neuronal tactile response could change substantially, including inverting its impact to the opposite. These cortical SPWs shared many signal fingerprint characteristics as reported for hippocampal SPWs and may be a biomarker for a particular type of state change that is possibly shared byboth hippocampus and neocortex.

Extreme value statistics of nerve transmission delay.

Delays in nerve transmission are an important topic in the field of neuroscience. Spike signals fired or received by the dendrites of a neuron travel from the axon to a presynaptic cell. The spike signal then triggers a chemical reaction at the synapse, wherein a presynaptic cell transfers neurotransmitters to the postsynaptic cell, regenerates electrical signals via a chemical reaction through ion channels, and transmits them to neighboring neurons. In the context of describing the complex physiological reaction process as a stochastic process, this study aimed to show that the distribution of the maximum time interval of spike signals follows extreme-order statistics. By considering the statistical variance in the time constant of the leaky Integrate-and-Fire model, a deterministic time evolution model for spike signals, we enabled randomness in the time interval of the spike signals. When the time constant follows an exponential distribution function, the time interval of the spike signal also follows an exponential distribution. In this case, our theory and simulations confirmed that the histogram of the maximum time interval follows the Gumbel distribution, one of the three forms of extreme-value statistics. We further confirmed that the histogram of the maximum time interval followed a Fréchet distribution when the time interval of the spike signal followed a Pareto distribution. These findings confirm that nerve transmission delay can be described using extreme value statistics and can therefore be used as a new indicator of transmission delay.

GABAergic synaptic scaling is triggered by changes in spiking activity rather than AMPA receptor activation.

Homeostatic plasticity represents a set of mechanisms that are thought to recover some aspect of neural function. One such mechanism called AMPAergic scaling was thought to be a likely candidate to homeostatically control spiking activity. However, recent findings have forced us to reconsider this idea as several studies suggest AMPAergic scaling is not directly triggered by changes in spiking. Moreover, studies examining homeostatic perturbations in vivo have suggested that GABAergic synapses may be more critical in terms of spiking homeostasis. Here, we show results that GABAergic scaling can act to homeostatically control spiking levels. We found that perturbations which increased or decreased spiking in cortical cultures triggered multiplicative GABAergic upscaling and downscaling, respectively. In contrast, we found that changes in AMPA receptor (AMPAR) or GABAR transmission only influence GABAergic scaling through their indirect effect on spiking. We propose that GABAergic scaling represents a stronger candidate for spike rate homeostat than AMPAergic scaling.

A Continuous Attractor Model with Realistic Neural and Synaptic Properties Quantitatively Reproduces Grid Cell Physiology.

Computational simulations with data-driven physiological detail can foster a deeper understanding of the neural mechanisms involved in cognition. Here, we utilize the wealth of cellular properties from Hippocampome.org to study neural mechanisms of spatial coding with a spiking continuous attractor network model of medial entorhinal cortex circuit activity. The primary goal is to investigate if adding such realistic constraints could produce firing patterns similar to those measured in real neurons. Biological characteristics included in the work are excitability, connectivity, and synaptic signaling of neuron types defined primarily by their axonal and dendritic morphologies. We investigate the spiking dynamics in specific neuron types and the synaptic activities between groups of neurons. Modeling the rodent hippocampal formation keeps the simulations to a computationally reasonable scale while also anchoring the parameters and results to experimental measurements. Our model generates grid cell activity that well matches the spacing, size, and firing rates of grid fields recorded in live behaving animals from both published datasets and new experiments performed for this study. Our simulations also recreate different scales of those properties, e.g., small and large, as found along the dorsoventral axis of the medial entorhinal cortex. Computational exploration of neuronal and synaptic model parameters reveals that a broad range of neural properties produce grid fields in the simulation. These results demonstrate that the continuous attractor network model of grid cells is compatible with a spiking neural network implementation sourcing data-driven biophysical and anatomical parameters from Hippocampome.org. The software (version 1.0) is released as open source to enable broad community reuse and encourage novel applications.

A 3D neuronal network read-out interface with high recording performance using a neuronal cluster patterning on a microelectrode array.

In recent years, in vitro three-dimensional (3D) neuronal network models utilizing extracellular matrices have been advancing. To understand the network activity from these models, attempts have been made to measure activity in multiple regions simultaneously using a microelectrode array (MEA). Although there hve been many attempts to measure the activity of 3D networks using 2-dimensional (2D) MEAs, the physical coupling between the 3D network and the microelectrodes was not stable and needed to be improved. In this study, we proposed a neuronal cluster interface that improves the active channel ratio of commercial 2D MEAs, enabling reliable measurement of 3D network activity. To achieve this, neuronal clusters, which consist of a small number of neurons, were patterned on microelectrodes and used as mediators to transmit the signal between the 3D network and the microelectrodes. We confirmed that the patterned neuronal clusters enhanced the active channel ratio and SNR(signal-to-noise-ratio) about 3D network recording and stimulation for a month. Our interface was able to functionally connect with 3D networks and measure the 3D network activity without significant alternation of activity characteristics. Finally, we demonstrated that our interface can be used to analyze the differences in the dynamics of 3D and 2D networks and to construct the 3D clustered network. This method is expected to be useful for studying the functional activity of various 3D neuronal network models, offering broad applications for the use of these models.

Neuropeptide Modulation Enables Biphasic Internetwork Coordination via a Dual-Network Neuron.

Linked rhythmic behaviors, such as respiration/locomotion or swallowing/chewing, often require coordination for proper function. Despite its prevalence, the cellular mechanisms controlling coordination of the underlying neural networks remain undetermined in most systems. We use the stomatogastric nervous system of the crab Cancer borealis to investigate mechanisms of internetwork coordination, due to its small, well-characterized feeding-related networks (gastric mill [chewing, ∼0.1 Hz]; pyloric [filtering food, ∼1 Hz]). Here, we investigate coordination between these networks during the Gly1-SIFamide neuropeptide modulatory state. Gly1-SIFamide activates a unique triphasic gastric mill rhythm in which the typically pyloric-only LPG neuron generates dual pyloric-plus gastric mill-timed oscillations. Additionally, the pyloric rhythm exhibits shorter cycles during gastric mill rhythm-timed LPG bursts, and longer cycles during IC, or IC plus LG gastric mill neuron bursts. Photoinactivation revealed that LPG is necessary to shorten pyloric cycle period, likely through its rectified electrical coupling to pyloric pacemaker neurons. Hyperpolarizing current injections demonstrated that although LG bursting enables IC bursts, only gastric mill rhythm bursts in IC are necessary to prolong the pyloric cycle period. Surprisingly, LPG photoinactivation also eliminated prolonged pyloric cycles, without changing IC firing frequency or gastric mill burst duration, suggesting that pyloric cycles are prolonged via IC synaptic inhibition of LPG, which indirectly slows the pyloric pacemakers via electrical coupling. Thus, the same dual-network neuron directly conveys excitation from its endogenous bursting and indirectly funnels synaptic inhibition to enable one network to alternately decrease and increase the cycle period of a related network.

Towards biologically plausible model-based reinforcement learning in recurrent spiking networks by dreaming new experiences.

Humans and animals can learn new skills after practicing for a few hours, while current reinforcement learning algorithms require a large amount of data to achieve good performances. Recent model-based approaches show promising results by reducing the number of necessary interactions with the environment to learn a desirable policy. However, these methods require biological implausible ingredients, such as the detailed storage of older experiences, and long periods of offline learning. The optimal way to learn and exploit world-models is still an open question. Taking inspiration from biology, we suggest that dreaming might be an efficient expedient to use an inner model. We propose a two-module (agent and model) spiking neural network in which "dreaming" (living new experiences in a model-based simulated environment) significantly boosts learning. Importantly, our model does not require the detailed storage of experiences, and learns online the world-model and the policy. Moreover, we stress that our network is composed of spiking neurons, further increasing the biological plausibility and implementability in neuromorphic hardware.

Unravelling the operation of organic artificial neurons for neuromorphic bioelectronics.

Organic artificial neurons operating in liquid environments are crucial components in neuromorphic bioelectronics. However, the current understanding of these neurons is limited, hindering their rational design and development for realistic neuronal emulation in biological settings. Here we combine experiments, numerical non-linear simulations, and analytical tools to unravel the operation of organic artificial neurons. This comprehensive approach elucidates a broad spectrum of biorealistic behaviors, including firing properties, excitability, wetware operation, and biohybrid integration. The non-linear simulations are grounded in a physics-based framework, accounting for ion type and ion concentration in the electrolytic medium, organic mixed ionic-electronic parameters, and biomembrane features. The derived analytical expressions link the neurons spiking features with material and physical parameters, bridging closer the domains of artificial neurons and neuroscience. This work provides streamlined and transferable guidelines for the design, development, engineering, and optimization of organic artificial neurons, advancing next generation neuronal networks, neuromorphic electronics, and bioelectronics.

Fueling the heartbeat: Dynamic regulation of intracellular ATP during excitation-contraction coupling in ventricular myocytes.

The heart beats approximately 100,000 times per day in humans, imposing substantial energetic demands on cardiac muscle. Adenosine triphosphate (ATP) is an essential energy source for normal function of cardiac muscle during each beat, as it powers ion transport, intracellular Ca2+ handling, and actin-myosin cross-bridge cycling. Despite this, the impact of excitation-contraction coupling on the intracellular ATP concentration ([ATP]i) in myocytes is poorly understood. Here, we conducted real-time measurements of [ATP]i in ventricular myocytes using a genetically encoded ATP fluorescent reporter. Our data reveal rapid beat-to-beat variations in [ATP]i. Notably, diastolic [ATP]i was <1 mM, which is eightfold to 10-fold lower than previously estimated. Accordingly, ATP-sensitive K+ (KATP) channels were active at physiological [ATP]i. Cells exhibited two distinct types of ATP fluctuations during an action potential: net increases (Mode 1) or decreases (Mode 2) in [ATP]i. Mode 1 [ATP]i increases necessitated Ca2+ entry and release from the sarcoplasmic reticulum (SR) and were associated with increases in mitochondrial Ca2+. By contrast, decreases in mitochondrial Ca2+ accompanied Mode 2 [ATP]i decreases. Down-regulation of the protein mitofusin 2 reduced the magnitude of [ATP]i fluctuations, indicating that SR-mitochondrial coupling plays a crucial role in the dynamic control of ATP levels. Activation of ß-adrenergic receptors decreased [ATP]i, underscoring the energetic impact of this signaling pathway. Finally, our work suggests that cross-bridge cycling is the largest consumer of ATP in a ventricular myocyte during an action potential. These findings provide insights into the energetic demands of EC coupling and highlight the dynamic nature of ATP concentrations in cardiac muscle.

KCNQ1 is an essential mediator of the sex-dependent perception of moderate cold temperatures.

Low temperatures and cooling agents like menthol induce cold sensation by activating the peripheral cold receptors TRPM8 and TRPA1, cation channels belonging to the TRP channel family, while the reduction of potassium currents provides an additional and/or synergistic mechanism of cold sensation. Despite extensive studies over the past decades to identify the molecular receptors that mediate thermosensation, cold sensation is still not fully understood and many cold-sensitive peripheral neurons do not express the well-established cold sensor TRPM8. We found that the voltage-gated potassium channel KCNQ1 (Kv7.1), which is defective in cardiac LQT1 syndrome, is, in addition to its known function in the heart, a highly relevant and sex-specific sensor of moderately cold temperatures. We found that KCNQ1 is expressed in skin and dorsal root ganglion neurons, is sensitive to menthol and cooling agents, and is highly sensitive to moderately cold temperatures, in a temperature range at which TRPM8 is not thermosensitive. C-fiber recordings from KCNQ1-/- mice displayed altered action potential firing properties. Strikingly, only male KCNQ1-/- mice showed substantial deficits in cold avoidance at moderately cold temperatures, with a strength of the phenotype similar to that observed in TRPM8-/- animals. While sex-dependent differences in thermal sensitivity have been well documented in humans and mice, KCNQ1 is the first gene reported to play a role in sex-specific temperature sensation. Moreover, we propose that KCNQ1, together with TRPM8, is a key instrumentalist that orchestrates the range and intensity of cold sensation.

The neuron as a direct data-driven controller.

In the quest to model neuronal function amid gaps in physiological data, a promising strategy is to develop a normative theory that interprets neuronal physiology as optimizing a computational objective. This study extends current normative models, which primarily optimize prediction, by conceptualizing neurons as optimal feedback controllers. We posit that neurons, especially those beyond early sensory areas, steer their environment toward a specific desired state through their output. This environment comprises both synaptically interlinked neurons and external motor sensory feedback loops, enabling neurons to evaluate the effectiveness of their control via synaptic feedback. To model neurons as biologically feasible controllers which implicitly identify loop dynamics, infer latent states, and optimize control we utilize the contemporary direct data-driven control (DD-DC) framework. Our DD-DC neuron model explains various neurophysiological phenomena: the shift from potentiation to depression in spike-timing-dependent plasticity with its asymmetry, the duration and adaptive nature of feedforward and feedback neuronal filters, the imprecision in spike generation under constant stimulation, and the characteristic operational variability and noise in the brain. Our model presents a significant departure from the traditional, feedforward, instant-response McCulloch-Pitts-Rosenblatt neuron, offering a modern, biologically informed fundamental unit for constructing neural networks.

Juxtacellular recordings from identified neurons in the mouse locus coeruleus.

The locus coeruleus (LC) is the primary source of noradrenergic transmission in the mammalian central nervous system. This small pontine nucleus consists of a densely packed nuclear core-which contains the highest density of noradrenergic neurons-embedded within a heterogeneous surround of non-noradrenergic cells. This local heterogeneity, together with the small size of the LC, has made it particularly difficult to infer noradrenergic cell identity based on extracellular sampling of in vivo spiking activity. Moreover, the relatively high cell density, background activity and synchronicity of LC neurons have made spike identification and unit isolation notoriously challenging. In this study, we aimed at bridging these gaps by performing juxtacellular recordings from single identified neurons within the mouse LC complex. We found that noradrenergic neurons (identified by tyrosine hydroxylase, TH, expression; TH-positive) and intermingled putatively non-noradrenergic (TH-negative) cells displayed similar morphologies and responded to foot shock stimuli with excitatory responses; however, on average, TH-positive neurons exhibited more prominent foot shock responses and post-activation firing suppression. The two cell classes also displayed different spontaneous firing rates, spike waveforms and temporal spiking properties. A logistic regression classifier trained on spontaneous electrophysiological features could separate the two cell classes with 76% accuracy. Altogether, our results reveal in vivo electrophysiological correlates of TH-positive neurons, which can be useful for refining current approaches for the classification of LC unit activity.

Neural Drive and Motor Unit Characteristics of the Serratus Anterior in Individuals With Scapular Dyskinesis.

OBJECTIVE: Scapular dyskinesis is one of the causes of shoulder disorders and involves muscle weakness in the serratus anterior. This study investigated whether motor unit (MU) recruitment and firing property, which are important for muscle exertion, have altered in serratus anterior of the individuals with scapular dyskinesis. METHODS: Asymptomatic adults with (SD) and without (control) scapular dyskinesis were analyzed. Surface electromyography (sEMG) waveforms were collected at submaximal voluntary contraction of the serratus anterior. The sEMG waveform was decomposed into MU action potential amplitude (MUAPAMP), mean firing rate (MFR), and recruitment threshold. MUs were divided into low, moderate, and high thresholds, and MU recruitment and firing properties of the groups were compared. RESULTS: High-threshold MUAPAMP was significantly smaller in the SD group than in the control group. The control group also exhibited recruitment properties that reflected the size principle, however, the SD group did not. Furthermore, the SD group had a lower MFR than the control group. CONCLUSIONS: Individuals with scapular dyskinesis exhibit altered MU recruitment properties and lower firing rates of the serratus anterior; this may be detrimental to muscle performance. Thus, it may be necessary to improve the neural drive of the serratus anterior when correcting scapular dyskinesis.

Two Distinct Neuronal Populations in the Rat Parafascicular Nucleus Oppositely Encode the Engagement in Stimulus-driven Reward-seeking.

BACKGROUND: The thalamus is a phylogenetically well-preserved structure. Known to densely contact cortical regions, its role in the transmission of sensory information to the striatal complex has been widely reconsidered in recent years. METHODS: The parafascicular nucleus of the thalamus (Pf) has been implicated in the orientation of attention toward salient sensory stimuli. In a stimulus-driven reward-seeking task, we sought to characterize the electrophysiological activity of Pf neurons in rats. RESULTS: We observed a predominance of excitatory over inhibitory responses for all events in the task. Neurons responded more strongly to the stimulus compared to lever-pressing and reward collecting, confirming the strong involvement of the Pf in sensory information processing. The use of long sessions allowed us to compare neuronal responses to stimuli between trials when animals were engaged in action and those when they were not. We distinguished two populations of neurons with opposite responses: MOTIV+ neurons responded more intensely to stimuli followed by a behavioral response than those that were not. Conversely, MOTIV- neurons responded more strongly when the animal did not respond to the stimulus. In addition, the latency of excitation of MOTIV- neurons was shorter than that of MOTIV+ neurons. CONCLUSION: Through this encoding, the Pf could perform an early selection of environmental stimuli transmitted to the striatum according to motivational level.

Extracellular electrophysiology on clonal human ß-cell spheroids.

Background: Pancreatic islets are important in nutrient homeostasis and improved cellular models of clonal origin may very useful especially in view of relatively scarce primary material. Close 3D contact and coupling between ß-cells are a hallmark of physiological function improving signal/noise ratios. Extracellular electrophysiology using micro-electrode arrays (MEA) is technically far more accessible than single cell patch clamp, enables dynamic monitoring of electrical activity in 3D organoids and recorded multicellular slow potentials (SP) provide unbiased insight in cell-cell coupling. Objective: We have therefore asked whether 3D spheroids enhance clonal ß-cell function such as electrical activity and hormone secretion using human EndoC-ßH1, EndoC-ßH5 and rodent INS-1 832/13 cells. Methods: Spheroids were formed either by hanging drop or proprietary devices. Extracellular electrophysiology was conducted using multi-electrode arrays with appropriate signal extraction and hormone secretion measured by ELISA. Results: EndoC-ßH1 spheroids exhibited increased signals in terms of SP frequency and especially amplitude as compared to monolayers and even single cell action potentials (AP) were quantifiable. Enhanced electrical signature in spheroids was accompanied by an increase in the glucose stimulated insulin secretion index. EndoC-ßH5 monolayers and spheroids gave electrophysiological profiles similar to EndoC-ßH1, except for a higher electrical activity at 3 mM glucose, and exhibited moreover a biphasic profile. Again, physiological concentrations of GLP-1 increased AP frequency. Spheroids also exhibited a higher secretion index. INS-1 cells did not form stable spheroids, but overexpression of connexin 36, required for cell-cell coupling, increased glucose responsiveness, dampened basal activity and consequently augmented the stimulation index. Conclusion: In conclusion, spheroid formation enhances physiological function of the human clonal ß-cell lines and these models may provide surrogates for primary islets in extracellular electrophysiology.

Persistent spiking activity in neuromorphic circuits incorporating post-inhibitory rebound excitation.

Objective. This study introduces a novel approach for integrating the post-inhibitory rebound excitation (PIRE) phenomenon into a neuronal circuit. Excitatory and inhibitory synapses are designed to establish a connection between two hardware neurons, effectively forming a network. The model demonstrates the occurrence of PIRE under strong inhibitory input. Emphasizing the significance of incorporating PIRE in neuromorphic circuits, the study showcases generation of persistent activity within cyclic and recurrent spiking neuronal networks.Approach. The neuronal and synaptic circuits are designed and simulated in Cadence Virtuoso using TSMC 180 nm technology. The operating mechanism of the PIRE phenomenon integrated into a hardware neuron is discussed. The proposed circuit encompasses several parameters for effectively controlling multiple electrophysiological features of a neuron.Main results. The neuronal circuit has been tuned to match the response of a biological neuron. The efficiency of this circuit is evaluated by computing the average power dissipation and energy consumption per spike through simulation. The sustained firing of neural spikes is observed till 1.7 s using the two neuronal networks.Significance. Persistent activity has significant implications for various cognitive functions such as working memory, decision-making, and attention. Therefore, hardware implementation of these functions will require our PIRE-integrated model. Energy-efficient neuromorphic systems are useful in many artificial intelligence applications, including human-machine interaction, IoT devices, autonomous systems, and brain-computer interfaces.

Cell-specific models of hiPSC-CMs developed by the gradient-based parameter optimization method fitting two different action potential waveforms.

Parameter optimization (PO) methods to determine the ionic current composition of experimental cardiac action potential (AP) waveform have been developed using a computer model of cardiac membrane excitation. However, it was suggested that fitting a single AP record in the PO method was not always successful in providing a unique answer because of a shortage of information. We found that the PO method worked perfectly if the PO method was applied to a pair of a control AP and a model output AP in which a single ionic current out of six current species, such as IKr, ICaL, INa, IKs, IKur or IbNSC was partially blocked in silico. When the target was replaced by a pair of experimental control and IKr-blocked records of APs generated spontaneously in human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs), the simultaneous fitting of the two waveforms by the PO method was hampered to some extent by the irregular slow fluctuations in the Vm recording and/or sporadic alteration in AP configurations in the hiPSC-CMs. This technical problem was largely removed by selecting stable segments of the records for the PO method. Moreover, the PO method was made fail-proof by running iteratively in identifying the optimized parameter set to reconstruct both the control and the IKr-blocked AP waveforms. In the lead potential analysis, the quantitative ionic mechanisms deduced from the optimized parameter set were totally consistent with the qualitative view of ionic mechanisms of AP so far described in physiological literature.

GluN2A- and GluN2B-containing pre-synaptic N-methyl-d-aspartate receptors differentially regulate action potential-evoked Ca2+ influx via modulation of SK channels.

N-methyl-d-aspartate receptors (NMDARs) play a pivotal role in synaptic plasticity. While the functional role of post-synaptic NMDARs is well established, pre-synaptic NMDAR (pre-NMDAR) function is largely unexplored. Different pre-NMDAR subunit populations are documented at synapses, suggesting that subunit composition influences neuronal transmission. Here, we used electrophysiological recordings at Schaffer collateral-CA1 synapses partnered with Ca2+ imaging and glutamate uncaging at boutons of CA3 pyramidal neurones to reveal two populations of pre-NMDARs that contain either the GluN2A or GluN2B subunit. Activation of the GluN2B population decreases action potential-evoked Ca2+ influx via modulation of small-conductance Ca2+-activated K+ channels, while activation of the GluN2A population does the opposite. Critically, the level of functional expression of the subunits is subject to homeostatic regulation, bidirectionally affecting short-term facilitation, thus providing a capacity for a fine adjustment of information transfer. This article is part of a discussion meeting issue 'Long-term potentiation: 50 years on'.