Hebbian depression of isolated neuromuscular synapses in vitro

Emerging memristive artificial neuron and synapse devices for the neuromorphic electronics era

Growth of data eases the way to access the world but requires increasing amounts of energy to store and process. Neuromorphic electronics has emerged in the last decade, inspired by biological neurons and synapses, with in-memory computing ability, extenuating the 'von Neumann bottleneck' between the memory and processor and offering a promising solution to reduce the efforts both in data storage and processing, thanks to their multi-bit non-volatility, biology-emulated characteristics, and silicon compatibility. This work reviews the recent advances in emerging memristive devices for artificial neuron and synapse applications, including memory and data-processing ability: the physics and characteristics are discussed first, i.e., valence changing, electrochemical metallization, phase changing, interfaced-controlling, charge-trapping, ferroelectric tunnelling, and spin-transfer torquing. Next, we propose a universal benchmark for the artificial synapse and neuron devices on spiking energy consumption, standby power consumption, and spike timing. Based on the benchmark, we address the challenges, suggest the guidelines for intra-device and inter-device design, and provide an outlook for the neuromorphic applications of resistive switching-based artificial neuron and synapse devices.

STDP-based adaptive graph convolutional networks for automatic sleep staging

Automatic sleep staging is important for improving diagnosis and treatment, and machine learning with neuroscience explainability of sleep staging is shown to be a suitable method to solve this problem. In this paper, an explainable model for automatic sleep staging is proposed. Inspired by the Spike-Timing-Dependent Plasticity (STDP), an adaptive Graph Convolutional Network (GCN) is established to extract features from the Polysomnography (PSG) signal, named STDP-GCN. In detail, the channel of the PSG signal can be regarded as a neuron, the synapse strength between neurons can be constructed by the STDP mechanism, and the connection between different channels of the PSG signal constitutes a graph structure. After utilizing GCN to extract spatial features, temporal convolution is used to extract transition rules between sleep stages, and a fully connected neural network is used for classification. To enhance the strength of the model and minimize the effect of individual physiological signal discrepancies on classification accuracy, STDP-GCN utilizes domain adversarial training. Experiments demonstrate that the performance of STDP-GCN is comparable to the current state-of-the-art models.

Evidence That the Central Nervous System Can Induce a Modification at the Neuromuscular Junction That Contributes to the Maintenance of a Behavioral Response

Neurons within the spinal cord are sensitive to environmental relations and can bring about a behavioral modification without input from the brain. For example, rats that have undergone a thoracic (T2) transection can learn to maintain a hind leg in a flexed position to minimize exposure to a noxious electrical stimulation (shock). Inactivating neurons within the spinal cord with lidocaine, or cutting communication between the spinal cord and the periphery (sciatic transection), eliminates the capacity to learn, which implies that it depends on spinal neurons. Here we show that these manipulations have no effect on the maintenance of the learned response, which implicates a peripheral process. EMG showed that learning augments the muscular response evoked by motoneuron output and that this effect survives a sciatic transection. Quantitative fluorescent imaging revealed that training brings about an increase in the area and intensity of ACh receptor labeling at the neuromuscular junction (NMJ). It is hypothesized that efferent motoneuron output, in conjunction with electrical stimulation of the tibialis anterior muscle, strengthens the connection at the NMJ in a Hebbian manner. Supporting this, paired stimulation of the efferent nerve and tibialis anterior generated an increase in flexion duration and augmented the evoked electrical response without input from the spinal cord. Evidence is presented that glutamatergic signaling contributes to plasticity at the NMJ. Labeling for vesicular glutamate transporter is evident at the motor endplate. Intramuscular application of an NMDAR antagonist blocked the acquisition/maintenance of the learned response and the strengthening of the evoked electrical response.SIGNIFICANCE STATEMENT The neuromuscular junction (NMJ) is designed to faithfully elicit a muscular contraction in response to neural input. From this perspective, encoding environmental relations (learning) and the maintenance of a behavioral modification over time (memory) are assumed to reflect only modifications upstream from the NMJ, within the CNS. The current results challenge this view. Rats were trained to maintain a hind leg in a flexed position to avoid noxious stimulation. As expected, treatments that inhibit activity within the CNS, or disrupt peripheral communication, prevented learning. These manipulations did not affect the maintenance of the acquired response. The results imply that a peripheral modification at the NMJ contributes to the maintenance of the learned response.

Brain experiments imply adaptation mechanisms which outperform common AI learning algorithms

Attempting to imitate the brain's functionalities, researchers have bridged between neuroscience and artificial intelligence for decades; however, experimental neuroscience has not directly advanced the field of machine learning (ML). Here, using neuronal cultures, we demonstrate that increased training frequency accelerates the neuronal adaptation processes. This mechanism was implemented on artificial neural networks, where a local learning step-size increases for coherent consecutive learning steps, and tested on a simple dataset of handwritten digits, MNIST. Based on our on-line learning results with a few handwriting examples, success rates for brain-inspired algorithms substantially outperform the commonly used ML algorithms. We speculate this emerging bridge from slow brain function to ML will promote ultrafast decision making under limited examples, which is the reality in many aspects of human activity, robotic control, and network optimization.

On Practical Issues for Stochastic STDP Hardware With 1-bit Synaptic Weights

In computational neuroscience, synaptic plasticity learning rules are typically studied using the full 64-bit floating point precision computers provide. However, for dedicated hardware implementations, the precision used not only penalizes directly the required memory resources, but also the computing, communication, and energy resources. When it comes to hardware engineering, a key question is always to find the minimum number of necessary bits to keep the neurocomputational system working satisfactorily. Here we present some techniques and results obtained when limiting synaptic weights to 1-bit precision, applied to a Spike-Timing-Dependent-Plasticity (STDP) learning rule in Spiking Neural Networks (SNN). We first illustrate the 1-bit synapses STDP operation by replicating a classical biological experiment on visual orientation tuning, using a simple four neuron setup. After this, we apply 1-bit STDP learning to the hidden feature extraction layer of a 2-layer system, where for the second (and output) layer we use already reported SNN classifiers. The systems are tested on two spiking datasets: a Dynamic Vision Sensor (DVS) recorded poker card symbols dataset and a Poisson-distributed spike representation MNIST dataset version. Tests are performed using the in-house MegaSim event-driven behavioral simulator and by implementing the systems on FPGA (Field Programmable Gate Array) hardware.

General differential Hebbian learning: Capturing temporal relations between events in neural networks and the brain

Learning in biologically relevant neural-network models usually relies on Hebb learning rules. The typical implementations of these rules change the synaptic strength on the basis of the co-occurrence of the neural events taking place at a certain time in the pre- and post-synaptic neurons. Differential Hebbian learning (DHL) rules, instead, are able to update the synapse by taking into account the temporal relation, captured with derivatives, between the neural events happening in the recent past. The few DHL rules proposed so far can update the synaptic weights only in few ways: this is a limitation for the study of dynamical neurons and neural-network models. Moreover, empirical evidence on brain spike-timing-dependent plasticity (STDP) shows that different neurons express a surprisingly rich repertoire of different learning processes going far beyond existing DHL rules. This opens up a second problem of how capturing such processes with DHL rules. Here we propose a general DHL (G-DHL) rule generating the existing rules and many others. The rule has a high expressiveness as it combines in different ways the pre- and post-synaptic neuron signals and derivatives. The rule flexibility is shown by applying it to various signals of artificial neurons and by fitting several different STDP experimental data sets. To these purposes, we propose techniques to pre-process the neural signals and capture the temporal relations between the neural events of interest. We also propose a procedure to automatically identify the rule components and parameters that best fit different STDP data sets, and show how the identified components might be used to heuristically guide the search of the biophysical mechanisms underlying STDP. Overall, the results show that the G-DHL rule represents a useful means to study time-sensitive learning processes in both artificial neural networks and brain.

Reversal of Cocaine-Associated Synaptic Plasticity in Medial Prefrontal Cortex Parallels Elimination of Memory Retrieval

Addiction is characterized by abnormalities in prefrontal cortex that are thought to allow drug-associated cues to drive compulsive drug seeking and taking. Identification and reversal of these pathologic neuroadaptations are therefore critical for treatment of addiction. Previous studies using rodents reveal that drugs of abuse cause dendritic spine plasticity in prelimbic medial prefrontal cortex (PL-mPFC) pyramidal neurons, a phenomenon that correlates with the strength of drug-associated memories in vivo. Thus, we hypothesized that cocaine-evoked plasticity in PL-mPFC may underlie cocaine-associated memory retrieval, and therefore disruption of this plasticity would prevent retrieval. Indeed, using patch clamp electrophysiology we find that cocaine place conditioning increases excitatory presynaptic and postsynaptic transmission in rat PL-mPFC pyramidal neurons. This was accounted for by increases in excitatory presynaptic release, paired-pulse facilitation, and increased AMPA receptor transmission. Noradrenergic signaling is known to maintain glutamatergic plasticity upon reactivation of modified circuits, and we therefore next determined whether inhibition of noradrenergic signaling during memory reactivation would reverse the cocaine-evoked plasticity and/or disrupt the cocaine-associated memory. We find that administration of the β-adrenergic receptor antagonist propranolol before memory retrieval, but not after (during memory reconsolidation), reverses the cocaine-evoked presynaptic and postsynaptic modifications in PL-mPFC and causes long-lasting memory impairments. Taken together, these data reveal that cocaine-evoked synaptic plasticity in PL-mPFC is reversible in vivo, and suggest a novel strategy that would allow normalization of prefrontal circuitry in addiction.

An Event-Driven Classifier for Spiking Neural Networks Fed with Synthetic or Dynamic Vision Sensor Data

This paper introduces a novel methodology for training an event-driven classifier within a Spiking Neural Network (SNN) System capable of yielding good classification results when using both synthetic input data and real data captured from Dynamic Vision Sensor (DVS) chips. The proposed supervised method uses the spiking activity provided by an arbitrary topology of prior SNN layers to build histograms and train the classifier in the frame domain using the stochastic gradient descent algorithm. In addition, this approach can cope with leaky integrate-and-fire neuron models within the SNN, a desirable feature for real-world SNN applications, where neural activation must fade away after some time in the absence of inputs. Consequently, this way of building histograms captures the dynamics of spikes immediately before the classifier. We tested our method on the MNIST data set using different synthetic encodings and real DVS sensory data sets such as N-MNIST, MNIST-DVS, and Poker-DVS using the same network topology and feature maps. We demonstrate the effectiveness of our approach by achieving the highest classification accuracy reported on the N-MNIST (97.77%) and Poker-DVS (100%) real DVS data sets to date with a spiking convolutional network. Moreover, by using the proposed method we were able to retrain the output layer of a previously reported spiking neural network and increase its performance by 2%, suggesting that the proposed classifier can be used as the output layer in works where features are extracted using unsupervised spike-based learning methods. In addition, we also analyze SNN performance figures such as total event activity and network latencies, which are relevant for eventual hardware implementations. In summary, the paper aggregates unsupervised-trained SNNs with a supervised-trained SNN classifier, combining and applying them to heterogeneous sets of benchmarks, both synthetic and from real DVS chips.

Energetic Constraints Produce Self-sustained Oscillatory Dynamics in Neuronal Networks

Overview: We model energy constraints in a network of spiking neurons, while exploring general questions of resource limitation on network function abstractly.

Background: Metabolic states like dietary ketosis or hypoglycemia have a large impact on brain function and disease outcomes. Glia provide metabolic support for neurons, among other functions. Yet, in computational models of glia-neuron cooperation, there have been no previous attempts to explore the effects of direct realistic energy costs on network activity in spiking neurons. Currently, biologically realistic spiking neural networks assume that membrane potential is the main driving factor for neural spiking, and do not take into consideration energetic costs.

Methods: We define local energy pools to constrain a neuron model, termed Spiking Neuron Energy Pool (SNEP), which explicitly incorporates energy limitations. Each neuron requires energy to spike, and resources in the pool regenerate over time. Our simulation displays an easy-to-use GUI, which can be run locally in a web browser, and is freely available.

Results: Energy dependence drastically changes behavior of these neural networks, causing emergent oscillations similar to those in networks of biological neurons. We analyze the system via Lotka-Volterra equations, producing several observations: (1) energy can drive self-sustained oscillations, (2) the energetic cost of spiking modulates the degree and type of oscillations, (3) harmonics emerge with frequencies determined by energy parameters, and (4) varying energetic costs have non-linear effects on energy consumption and firing rates.

Conclusions: Models of neuron function which attempt biological realism may benefit from including energy constraints. Further, we assert that observed oscillatory effects of energy limitations exist in networks of many kinds, and that these findings generalize to abstract graphs and technological applications.

Calcium dynamics predict direction of synaptic plasticity in striatal spiny projection neurons

The striatum is a major site of learning and memory formation for sensorimotor and cognitive association. One of the mechanisms used by the brain for memory storage is synaptic plasticity - the long-lasting, activity-dependent change in synaptic strength. All forms of synaptic plasticity require an elevation in intracellular calcium, and a common hypothesis is that the amplitude and duration of calcium transients can determine the direction of synaptic plasticity. The utility of this hypothesis in the striatum is unclear in part because dopamine is required for striatal plasticity and in part because of the diversity in stimulation protocols. To test whether calcium can predict plasticity direction, we developed a calcium-based plasticity rule using a spiny projection neuron model with sophisticated calcium dynamics including calcium diffusion, buffering and pump extrusion. We utilized three spike timing-dependent plasticity (STDP) induction protocols, in which postsynaptic potentials are paired with precisely timed action potentials and the timing of such pairing determines whether potentiation or depression will occur. Results show that despite the variation in calcium dynamics, a single, calcium-based plasticity rule, which explicitly considers duration of calcium elevations, can explain the direction of synaptic weight change for all three STDP protocols. Additional simulations show that the plasticity rule correctly predicts the NMDA receptor dependence of long-term potentiation and the L-type channel dependence of long-term depression. By utilizing realistic calcium dynamics, the model reveals mechanisms controlling synaptic plasticity direction, and shows that the dynamics of calcium, not just calcium amplitude, are crucial for synaptic plasticity.

A Hebbian feedback covariance learning paradigm for self-tuning optimal control

We propose a novel adaptive optimal control paradigm inspired by Hebbian covariance synaptic adaptation, a preeminent model of learning and memory as well as other malleable functions in the brain. The adaptation is driven by the spontaneous fluctuations in the system input and output, the covariance of which provides useful information about the changes in the system behavior. The control structure represents a novel form of associative reinforcement learning in which the reinforcement signal is implicitly given by the covariance of the input-output (I/O) signals. Theoretical foundations for the paradigm are derived using Lyapunov theory and are verified by means of computer simulations. The learning algorithm is applicable to a general class of nonlinear adaptive control problems. This on-line direct adaptive control method benefits from a computationally straightforward design, proof of convergence, no need for complete system identification, robustness to noise and uncertainties, and the ability to optimize a general performance criterion in terms of system states and control signals. These attractive properties of Hebbian feedback covariance learning control lend themselves to future investigations into the computational functions of synaptic plasticity in biological neurons.

Losing the battle but winning the war: game theoretic analysis of the competition between motoneurons innervating a skeletal muscle

The fibers in a skeletal muscle are divided into groups called "muscle units" whereby each muscle unit is innervated by a single neuron. It was found that neurons with low activation thresholds have smaller muscle units than neurons with higher activation thresholds. This results in a fixed recruitment order of muscle units, from smallest to largest, called the "size principle." It is thought that the size principle results from a competitive process-taking place after birth-between the neurons innervating the muscle. The underlying mechanism of the competition was not understood. Moreover, the results in the majority of experiments that manipulated the activity during the competition period seemed to contradict the size principle. Experiments at the isolated muscle fibers showed that the competition is governed by a Hebbian-like rule, whereby neurons with low activation thresholds have a competitive advantage at any single muscle fiber. Thus neurons with low activation thresholds are expected to have larger muscle units in contradiction to what is seen empirically. This state of affairs was termed "paradoxical." In the present study we developed a new game theoretic framework to analyze such competitive biological processes. In this game, neurons are the players competing to innervate a maximal number of muscle fibers. We showed that in order to innervate more muscle fibers, it is advantageous to win (as the neurons with higher activation thresholds do) later competitions. This both explains the size principle and resolves the seemingly paradoxical experimental data. Our model establishes that the competition at each muscle fiber may indeed be Hebbian and that the size principle still emerges from these competitions as an overall property of the system. Thus, the less active neurons "lose the battle but win the war." Our model provides experimentally testable predictions. The new game-theoretic approach may be applied to competitions in other biological systems.

Learning optimisation by high firing irregularity

In a network of leaky integrate-and-fire (LIF) neurons, we investigate the functional role of irregular spiking at high rates. Irregular spiking is produced by either employing the partial somatic reset mechanism on every LIF neuron of the network or by using temporally correlated inputs. In both the benchmark problem of XOR (exclusive-OR) and in a general-sum game, it is shown that irrespective of the mechanism that is used to produce it, high firing irregularity enhances the learning capability of the spiking neural network trained with reward-modulated spike-timing-dependent plasticity. These results suggest that the brain may be utilising high firing irregularity for the purposes of learning optimisation.

Spike-timing-dependent plasticity and reliability optimization: the role of neuron dynamics

Plastic changes in synaptic efficacy can depend on the time ordering of presynaptic and postsynaptic spikes. This phenomenon is called spike-timing-dependent plasticity (STDP). One of the most striking aspects of this plasticity mechanism is that the STDP windows display a great variety of forms in different parts of the nervous system. We explore this issue from a theoretical point of view. We choose as the optimization principle the minimization of conditional entropy or maximization of reliability in the transmission of information. We apply this principle to two types of postsynaptic dynamics, designated type I and type II. The first is characterized as being an integrator, while the second is a resonator. We find that, depending on the parameters of the models, the optimization principle can give rise to a wide variety of STDP windows, such as antisymmetric Hebbian, predominantly depressing or symmetric with one positive region and two lateral negative regions. We can relate each of these forms to the dynamical behavior of the different models. We also propose experimental tests to assess the validity of the optimization principle.

Pro-BDNF-induced synaptic depression and retraction at developing neuromuscular synapses

Postsynaptic cells generate positive and negative signals that retrogradely modulate presynaptic function. At developing neuromuscular synapses, prolonged stimulation of muscle cells induces sustained synaptic depression. We provide evidence that pro-brain-derived neurotrophic factor (BDNF) is a negative retrograde signal that can be converted into a positive signal by metalloproteases at the synaptic junctions. Application of pro-BDNF induces a dramatic decrease in synaptic efficacy followed by a retraction of presynaptic terminals, and these effects are mediated by presynaptic pan-neurotrophin receptor p75 (p75(NTR)), the pro-BDNF receptor. A brief stimulation of myocytes expressing cleavable or uncleavable pro-BDNF elicits synaptic potentiation or depression, respectively. Extracellular application of metalloprotease inhibitors, which inhibits the cleavage of endogenous pro-BDNF, facilitates the muscle stimulation-induced synaptic depression. Inhibition of presynaptic p75(NTR) or postsynaptic BDNF expression also blocks the activity-dependent synaptic depression and retraction. These results support a model in which postsynaptic secretion of a single molecule, pro-BDNF, may stabilize or eliminate presynaptic terminals depending on its proteolytic conversion at the synapses.

2D co-ordinate transformation based on a spike timing-dependent plasticity learning mechanism

In order to plan accurate motor actions, the brain needs to build an integrated spatial representation associated with visual stimuli and haptic stimuli. Since visual stimuli are represented in retina-centered co-ordinates and haptic stimuli are represented in body-centered co-ordinates, co-ordinate transformations must occur between the retina-centered co-ordinates and body-centered co-ordinates. A spiking neural network (SNN) model, which is trained with spike-timing-dependent-plasticity (STDP), is proposed to perform a 2D co-ordinate transformation of the polar representation of an arm position to a Cartesian representation, to create a virtual image map of a haptic input. Through the visual pathway, a position signal corresponding to the haptic input is used to train the SNN with STDP synapses such that after learning the SNN can perform the co-ordinate transformation to generate a representation of the haptic input with the same co-ordinates as a visual image. The model can be applied to explain co-ordinate transformation in spiking neuron based systems. The principle can be used in artificial intelligent systems to process complex co-ordinate transformations represented by biological stimuli.

Role for the skeletal muscle action potential in non-Hebbian long-term depression at the amphibian (Bufo marinus) neuromuscular junction

Retrograde signaling from skeletal muscle cells to motor nerve terminals is a recognized mechanism for modulating the strength of neuromuscular transmission. We recently described a form of long-term depression of transmitter release at the mature neuromuscular junction that is dependent on the production of nitric oxide, most likely by the muscle cell (Etherington and Everett 2004 J Physiol (Lond) 559:507-517). We now show that the depression is blocked by treating neuromuscular preparations with mu-conotoxin G111A, an antagonist of skeletal muscle voltage gated sodium channels, indicating that the depression requires postsynaptic action potential firing. Experiments on dually-innervated sartorius muscles revealed that propagation of action potentials generated by low-frequency stimulation of one nerve branch gives rise to nitric-oxide mediated depression at unstimulated nerve terminals located many millimetres away on the same muscle fiber. The non-Hebbian pattern of expression of the depression, as well as its reliance on postsynaptic action potential firing, distinguish it from forms of synaptic depression described at immature neuromuscular synapses and may provide a mechanism for coregulation of the strength of motoneurons innervating the same postsynaptic cell.

Reinforcement learning through modulation of spike-timing-dependent synaptic plasticity

The persistent modification of synaptic efficacy as a function of the relative timing of pre- and postsynaptic spikes is a phenomenon known as spike-timing-dependent plasticity (STDP). Here we show that the modulation of STDP by a global reward signal leads to reinforcement learning. We first derive analytically learning rules involving reward-modulated spike-timing-dependent synaptic and intrinsic plasticity, by applying a reinforcement learning algorithm to the stochastic spike response model of spiking neurons. These rules have several features common to plasticity mechanisms experimentally found in the brain. We then demonstrate in simulations of networks of integrate-and-fire neurons the efficacy of two simple learning rules involving modulated STDP. One rule is a direct extension of the standard STDP model (modulated STDP), and the other one involves an eligibility trace stored at each synapse that keeps a decaying memory of the relationships between the recent pairs of pre- and postsynaptic spike pairs (modulated STDP with eligibility trace). This latter rule permits learning even if the reward signal is delayed. The proposed rules are able to solve the XOR problem with both rate coded and temporally coded input and to learn a target output firing-rate pattern. These learning rules are biologically plausible, may be used for training generic artificial spiking neural networks, regardless of the neural model used, and suggest the experimental investigation in animals of the existence of reward-modulated STDP.

Formation of feedforward networks and frequency synchrony by spike-timing-dependent plasticity

Spike-timing-dependent plasticity (STDP) with asymmetric learning windows is commonly found in the brain and useful for a variety of spike-based computations such as input filtering and associative memory. A natural consequence of STDP is establishment of causality in the sense that a neuron learns to fire with a lag after specific presynaptic neurons have fired. The effect of STDP on synchrony is elusive because spike synchrony implies unitary spike events of different neurons rather than a causal delayed relationship between neurons. We explore how synchrony can be facilitated by STDP in oscillator networks with a pacemaker. We show that STDP with asymmetric learning windows leads to self-organization of feedforward networks starting from the pacemaker. As a result, STDP drastically facilitates frequency synchrony. Even though differences in spike times are lessened as a result of synaptic plasticity, the finite time lag remains so that perfect spike synchrony is not realized. In contrast to traditional mechanisms of large-scale synchrony based on mutual interaction of coupled neurons, the route to synchrony discovered here is enslavement of downstream neurons by upstream ones. Facilitation of such feedforward synchrony does not occur for STDP with symmetric learning windows.

Both A chain and B chain of β-bungarotoxin are functionally involved in the facilitation of spontaneous transmitter release in Xenopus nerve–muscle cultures

In the present study, Xenopus nerve-muscle cultures were used to explore the functional roles of A chain (a phospholipase A(2) subunit) and B chain (a non-phospholipase A(2) subunit) of Bungarus multicinctus beta-bungarotoxin. It was found that beta-bungarotoxin induced an increment of the frequency of spontaneous synaptic currents (SSCs) in the nerve-muscle cultures. Modification of beta-bungarotoxin with pyridoxal-5'-phosphate or substitution of Ca(2+) with Ba(2+) in buffer abolished the phospholipase A(2) activity of beta-bungarotoxin and the facilitatory phase of SSC as well. Antibodies that were directed specifically against A chain or B chain effectively inhibited phospholipase A(2) activity, and as a consequence the SSC frequency was not greatly different from the control rate. These results suggest that both A and B chains are indispensable parts of beta-bungarotoxin for inducing the facilitation of SSC frequency with Xenopus nerve-muscle cultures.

Potentiation of quantal secretion by insulin-like growth factor-1 at developing motoneurons in Xenopus cell culture

Although evidence suggests that insulin-like growth factor (IGF) plays an important role in the development and growth of the nervous system, the effect of IGF-1 in the regulation of neurotransmitter release in the peripheral nervous system remains unknown. Here we show that acute application of IGF-1, a factor widely expressed in developing myocytes, dose-dependently enhances the spontaneous acetylcholine (ACh) secretion at developing neuromuscular synapses in Xenopus cell culture using whole-cell patch clamp recording. We studied the role of endogenously released IGF-1 by examining the effect of IGF-1 antibody on the frequency of spontaneous synaptic currents (SSCs) at high-activity synapses, and found SSC frequency was markedly reduced at these high-activity synapses. The IGF-1-induced synaptic potentiation was not abolished when Ca2+ was eliminated from the culture medium or there was bath-application of the pharmacological Ca2+ channel inhibitor Cd2+, indicating that Ca2+ influxes through voltage-activated Ca2+ channels are not required. Application of membrane-permeable inhibitors of inositol 1,4,5-trisphosphate (IP3) or ryanodine receptors effectively occluded the increase of SSC frequency elicited by IGF-I. Treating cells with the phosphoinositide-3 kinase (PI3-K) inhibitors wortmannin or LY294002, and with phospholipase Cgamma (PLCgamma) inhibitor U73122, but not the inhibitor of mitogen-activated protein (MAP) kinase PD98059, abolished IGF-1-induced synaptic potentiation. Taken collectively, these results suggest that endogenously released IGF-1 from myocytes elicits Ca2+ release from IP3- and/or ryanodine-sensitive intracellular Ca2+ stores of the presynaptic nerve terminal. This is done via PI3-K and PLCgamma signalling cascades, leading to an enhancement of spontaneous transmitter release.

Long-term depression of synaptic inhibition is expressed postsynaptically in the developing auditory system

Inhibitory transmission is critically involved in the functional maturation of neural circuits within the brain. However, the mechanisms involved in its plasticity and development remain poorly understood. At an inhibitory synapse of the developing auditory brain stem, we used whole cell recordings to determine the site of induction and expression of long-term depression (LTD), a robust activity-dependent phenomenon that decreases inhibitory synaptic gain and is postulated to underlie synapse elimination. Recordings were obtained from lateral superior olivary (LSO) neurons, and hyperpolarizing inhibitory potentials were evoked by stimulation of the medial nucleus of the trapezoid body (MNTB). Both postsynaptic glycine and GABAA receptors could independently display LTD when isolated pharmacologically. Focal application of GABA, but not glycine, on the postsynaptic LSO neuron was sufficient to induce depression of the amino acid-evoked response, or MNTB-evoked inhibitory postsynaptic potentials. This GABA-mediated depression, in the absence of MNTB stimulation, was blocked by a GABAB receptor antagonist. To assess whether a change in neurotransmitter release is associated with the LTD, the polyvalent cation, ruthenium red, was used to increase the frequency of miniature inhibitory synaptic events. Consistent with a postsynaptic locus of expression, we found that the mean amplitude of miniature events decreased after LTD with no change in their frequency of occurrence. Furthermore, there was no change in the paired-pulse ratio or release kinetics of evoked inhibitory responses. Together, these results provide direct evidence that activity-dependent LTD of inhibition has a postsynaptic locus of induction and alteration, and that GABA but not glycine plays a pivotal role.

Spike-timing-dependent plasticity and relevant mutual information maximization

Synaptic plasticity was recently shown to depend on the relative timing of the pre- and postsynaptic spikes. This article analytically derives a spike-dependent learning rule based on the principle of information maximization for a single neuron with spiking inputs. This rule is then transformed into a biologically feasible rule, which is compared to the experimentally observed plasticity. This comparison reveals that the biological rule increases information to a near-optimal level and provides insights into the structure of biological plasticity. It shows that the time dependency of synaptic potentiation should be determined by the synaptic transfer function and membrane leak. Potentiation consists of weight-dependent and weight-independent components whose weights are of the same order of magnitude. It further suggests that synaptic depression should be triggered by rare and relevant inputs but at the same time serves to unlearn the baseline statistics of the network's inputs. The optimal depression curve is uniformly extended in time, but biological constraints that cause the cell to forget past events may lead to a different shape, which is not specified by our current model. The structure of the optimal rule thus suggests a computational account for several temporal characteristics of the biological spike-timing-dependent rules.

Sodium imaging of climbing fiber innervation fields in developing mouse Purkinje cells

Maturation of specific neuronal connections in the mature nervous system includes elimination of redundant synapses formed earlier during development. In the cerebellum of adult animals, each Purkinje cell (PC) is innervated by a single climbing fiber (CF). In early postnatal development each PC is innervated by multiple CFs and elimination of synapses formed by supernumerary CFs occurs until monoinnervation is established at around postnatal day 20 (P20) in mice. It is not clear whether multiple CFs, or only a single CF, translocate from the cell body of immature PCs to the developing dendrite and, in case several CFs translocate, whether they share or segregate their innervation fields. To localize CF innervation fields, we imaged changes in postsynaptic sodium concentration resulting from CF-mediated postsynaptic currents. We found that more than one CF translocates from an innervation field on the cell body of the PC to the developing dendrite and that these CFs share rather than segregate their innervation fields. We concluded that both the soma and the proximal dendrite of the PC are territories of competition for the developing CFs and that the overlapping of their termination fields may be the prerequisite for a local process of elimination of all but one CF, as previously demonstrated in the developing neuromuscular junction.

Activity-dependent synaptic plasticity: insights from neuromuscular junctions

Experience-dependent editing shapes synaptic connections throughout the developing nervous system, but the underlying cellular mechanisms remain poorly understood. A useful model synapse for addressing these mechanisms is the neuromuscular junction, the connection between spinal motor neurons and skeletal muscle fibers. Here the authors review current ideas about the role of activity in editing neuromuscular synaptic connections. A variety of new tools are being used to address some unanswered questions in vivo and in vitro. Understanding activity-dependent plasticity at developing neuromuscular synapses may reveal how neural circuits in the central nervous system are altered by experience throughout life.

Pre- and postsynaptic mechanisms in Hebbian activity-dependent synapse modification

We have used a three compartment tissue culture system that involved two separate populations of cholinergic neurons in the side compartments that converged on a common target population of myotubes in the center compartment. Activation of the axons from one population of neurons produced selective down-regulation of the synaptic inputs from the other neuronal population (when the two inputs innervated the same myotubes). The decrease in heterosynaptic inputs was mediated by protein kinase C (PKC). An activity-dependent action of protein kinase A (PKA) was associated with the stimulated input and this served to selectively stabilize this input. These changes associated with PKA and PKC activation were mediated by alterations in the number of acetylcholine receptors at the neuromuscular junction. These results suggest that neuromuscular electrical activity produces postsynaptic activation of both PKA and PKC, with the latter producing generalized synapse weakening and the former a selective synapse stabilization. Treatment of the neuronal cell body and axon to increase PKC activity by putting phorbal ester (PMA) in the side chamber did not affect synaptic transmission (with or without stimulation). By contrast, PKA blockade in the side compartment did produce an activity-dependent decrease in synaptic efficacy, which was due to a decrease in quantal release of neurotransmitter. Thus, when the synapse is activated, it appears that presynaptic PKA action is necessary to maintain transmitter output.

Synaptic precedence during synapse formation between reciprocally connected neurons involves transmitter-receptor interactions and AA metabolites

The cellular mechanisms that determine specificity of synaptic connections between mutually connected neurons in the nervous system have not yet been fully examined in vertebrate and invertebrate species. Here we report on a novel form of synaptic interaction during early stages of synapse formation between reciprocally connected Lymnaea neurons. Specifically, using soma-soma synapses between an identified dopaminergic neuron (also known as the giant dopamine cell), right pedal dorsal 1 (RPeD1), and a FMRFamidergic neuron, visceral dorsal 4 (VD4), we demonstrate that although reciprocal inhibitory synapses re-form between the somata after 24-36 h of pairing, VD4 is, however, the first cell to establish synaptic contacts with RPeD1 (within 12-18 h). We show that VD4 "captures" RPeD1 first as a postsynaptic cell by suppressing its transmitter secretory machinery during early stages of cell-cell pairing. The VD4-induced suppression of transmitter release from RPeD1 was transient, and it required transcription and de novo protein synthesis dependent step in VD4 but not in RPeD1. The VD4-induced effects on RPeD1 were mimicked by a FMRFamide-like peptide. Perturbation of FMRFamide-activated metabolites of the arachidonic acid pathway in RPeD1 not only prevented FMRFamide-induced suppression of transmitter release from the giant dopamine cell but also shifted the synaptic balance in favor of RPeD1, thus making it the first cell to begin synaptic transmission with VD4 within 12-18 h. A single RPeD1 that had developed dopamine secretory capabilities overnight and was subsequently paired with VD4 for 12-18 h was, however, immune to VD4-induced suppression of transmitter release. Under these experimental conditions, both cells developed mutual inhibitory synapses concurrently. Taken together, our data provide evidence for novel synaptic interaction between reciprocally connected neurons and underscore the importance of transmitter-receptor interplay in regulating the timing of synapse formation in the nervous system.

Electrical activity and development of neural circuits

A distinct feature of the nervous system is the intricate network of synaptic connections among neurons of diverse phenotypes. Although initial connections are formed largely through molecular mechanisms that depend on intrinsic developmental programs, spontaneous and experience-driven electrical activities in the developing brain exert critical epigenetic influence on synaptic maturation and refinement of neural circuits. Selective findings discussed here illustrate some of our current understanding of the effects of electrical activity on circuit development and highlight areas that await further study.

Developmental depression of glutamate neurotransmission by chronic low-level activation of NMDA receptors

Slabs of slow-release plastic (Elvax) containing NMDA or solvent were implanted over the rat colliculus beginning on postnatal day 8 (P8). Whole-cell patch clamping in the superficial superior collicular layers (sSCs) from P10 to P21 demonstrated a severe decrease in spontaneous EPSC frequency after chronic NMDA treatment. The decrease was not attributable to an increase in GABA(A) receptor-mediated inhibition and was present only when NMDA receptor (NMDAR) current was blocked by Mg(2+). Analysis of miniature EPSCs indicated that many active sites on NMDA-treated neurons lacked functional AMPA and kainate receptor (AMPA/KAR) currents, and AMPA/KAR:NMDAR current ratios of evoked EPSCs were also significantly reduced. In addition, the normal downregulation of NMDAR decay time in sSC neurons at P11 was absent after NMDA treatment. Nevertheless, neither AMPA nor NMDA receptor subunit expression was altered by NMDA treatment, and experiments with the NMDAR antagonist ifenprodil suggested that incorporation of NR2A-containing NMDARs at the sSC synapses was unperturbed. Thus, disrupting but not blocking NMDARs suppresses the development of AMPA/KAR currents. The absence of the P11 NMDAR current downregulation is likely a secondary effect resulting from the reduction of AMPA/KAR function. Chronic agonist application reduces but does not eliminate NMDAR conductances. Therefore these data support an active role for NMDAR currents in synaptic development. Prolonged NMDA treatment in vivo, which couples reduced postsynaptic Ca(2+) responses with normally developing afferent activity, produces a long-lasting synaptic depression and stalls glutamatergic synaptogenesis, suggesting that the correlation between robust NMDAR activation and afferent activity is an essential component during normal development.

Activity-dependent editing of neuromuscular synaptic connections

Work over the past four decades has suggested that neural activity edits synaptic connections throughout the developing nervous system. Synaptic editing is shaped in large part by competitive interactions among different inputs innervating the same target cell that profoundly influence synaptic strength and structure. While competition plays out among presynaptic inputs that anterogradely influence their targets, postsynaptic target cells also modulate competition, in part through retrograde interactions that modulate presynaptic neurotransmitter release. One of the most useful synapses for studying how neural activity mediates synaptic editing is the connections between spinal motor neurons and skeletal muscle fibers, called neuromuscular junctions. Here we review current ideas about the role of activity in editing neuromuscular synaptic connections. The mechanisms by which activity mediates synaptic competition at these peripheral synapses are relevant to understanding how neural circuits in the central nervous system are continually altered by experience throughout life.

Plasticity of cardiorespiratory neural processing: classification and computational functions

Neural plasticity, or malleability of neuronal structure and function, is an important attribute of the mammalian forebrain and is generally thought to be a kernel of biological intelligence. In this review, we examine some reported manifestations of neural plasticity in the cardiorespiratory system and classify them into four functional categories, integral; differential; memory; and statistical-type plasticity. At the cellular and systems level the myriad forms of cardiorespiratory plasticity display emergent and self-organization properties, use- and disuse-dependent and pairing-specific properties, short-term and long-term potentiation or depression, as well as redundancy in series or parallel structures, convergent pathways or backup and fail-safe surrogate pathways. At the behavioral level, the cardiorespiratory system demonstrates the capability of associative and nonassociative learning, classical and operant conditioning as well as short-term and long-term memory. The remarkable similarity and consistency of the various types of plasticity exhibited at all levels of organization suggest that neural plasticity is integral to cardiorespiratory control and may subserve important physiological functions.

Modeling synaptic plasticity in conjuction with the timing of pre- and postsynaptic action potentials

We present a spiking neuron model that allows for an analytic calculation of the correlations between pre- and postsynaptic spikes. The neuron model is a generalization of the integrate-and-fire model and equipped with a probabilistic spike-triggering mechanism. We show that under certain biologically plausible conditions, pre- and postsynaptic spike trains can be described simultaneously as an inhomogeneous Poisson process. Inspired by experimental findings, we develop a model for synaptic long-term plasticity that relies on the relative timing of pre- and post-synaptic action potentials. Being given an input statistics, we compute the stationary synaptic weights that result from the temporal correlations between the pre- and postsynaptic spikes. By means of both analytic calculations and computer simulations, we show that such a mechanism of synaptic plasticity is able to strengthen those input synapses that convey precisely timed spikes at the expense of synapses that deliver spikes with a broad temporal distribution. This may be of vital importance for any kind of information processing based on spiking neurons and temporal coding.

Activity-induced potentiation of developing neuromuscular synapses

Electrical activity plays a critical role in shaping the structure and function of synaptic connections in the nervous system. In Xenopus nerve-muscle cultures, a brief burst of action potentials in the presynaptic neuron induced a persistent potentiation of neuromuscular synapses that exhibit immature synaptic functions. Induction of potentiation required an elevation of postsynaptic Ca2+ and expression of potentiation appeared to involve an increased probability of transmitter secretion from the presynaptic nerve terminal. Thus, activity-dependent persistent synaptic enhancement may reflect properties characteristic of immature synaptic connections, and bursting activity in developing spinal neurons may promote functional maturation of the neuromuscular synapse.

Brain waves and brain wiring: the role of endogenous and sensory-driven neural activity in development

Neural activity is critical for sculpting the intricate circuits of the nervous system from initially imprecise neuronal connections. Disrupting the formation of these precise circuits may underlie many common neurodevelopmental disorders, ranging from subtle learning disorders to pervasive developmental delay. The necessity for sensory-driven activity has been widely recognized as crucial for infant brain development. Recent experiments in neurobiology now point to a similar requirement for endogenous neural activity generated by the nervous system itself before sensory input is available. Here we use the formation of precise neural circuits in the visual system to illustrate the principles of activity-dependent development. Competition between the projections from lateral geniculate nucleus neurons that receive sensory input from the two eyes shapes eye-specific connections from an initially diffuse projection into ocular dominance columns. When the competition is altered during a critical period for these changes, by depriving one eye of vision, the normal ocular dominance column pattern is disrupted. Before ocular dominance column formation, the highly ordered projection from retina to lateral geniculate nucleus develops. These connections form before the retina can respond to light, but at a time when retinal ganglion cells spontaneously generate highly correlated bursts of action potentials. Blockade of this endogenous activity, or biasing the competition in favor of one eye, results in a severe disruption of the pattern of retinogeniculate connections. Similar spontaneous, correlated activity has been identified in many locations in the developing central nervous system and is likely to be used during the formation of precise connections in many other neural systems. Understanding the processes of activity-dependent development could revolutionize our ability to identify, prevent, and treat developmental disorders resulting from disruptions of neural activity that interfere with the formation of precise neural circuits.

Regulation of quantal secretion from developing motoneurons by postsynaptic activity-dependent release of NT-3

Neurotrophic factors derived from postsynaptic muscle cells may play important roles in the development of presynaptic neuronal functions. In 3-d-old Xenopus nerve-muscle cultures, embryonic spinal neurons that had made natural contact with co-cultured myocytes exhibited spontaneous release of larger packets of acetylcholine (ACh) quanta than those released by the isolated neurons having no contact with any myocyte. Treatment of isolated neurons with neurotrophin-3 (NT-3) for 2 d increased the average sizes of quantal ACh packets at newly formed nerve-muscle synapses, whereas treatment with antibody against NT-3 or with K252a, a specific inhibitor of tyrosine kinase receptors, decreased the quantal size at existing synapses, which suggests that NT-3 supplied by the postsynaptic muscle cell may be responsible for the development and maintenance of the quantal packets. The muscle effect seems to depend on synaptic activities mediated by postsynaptic ACh receptor channels, because chronic treatment of the culture with D-tubocurarine (D-Tc) for 2 d resulted in a marked reduction of the quantal sizes, when assayed after extensive washing of the culture with Ringer's solution. The curare treatment did not affect the postsynaptic ACh receptor sensitivity, because iontophoretically applied ACh induced current responses similar to those of control. Finally, co-treatment of the culture with NT-3 and D-Tc reversed the effect of D-Tc on the quantal size, and this reversal effect was abolished when K252a was also applied concomitantly. Our results suggest that muscle-derived NT-3 participates in the maturation of normal transmitter packets in developing neurons, and the secretion of NT-3 depends on spontaneous synaptic activity.

Modulation of synaptic efficacy and synaptic depression by glial cells at the frog neuromuscular junction

The ability of perisynaptic glial cells to modulate transmitter release and synaptic depression was studied at the frog neuromuscular junction (nmj). Injection of GTPgammaS in perisynaptic Schwann cells (PSCs), glial cells at this synapse, induced a reduction in the amplitude of nerve-evoked synaptic responses but had no effect on the frequency, the amplitude, or the duration of the miniature endplate currents (MEPCs). Also, paired pulse facilitation was not affected. The reduction in transmitter release was mediated by pertussis toxin-(PTX) sensitive and insensitive G proteins. Blockade of G proteins in PSCs with GDPbetaS reduced synaptic depression induced by high frequency trains of stimuli, whereas activation of G proteins occluded it. Hence, the activation by endogenous neurotransmitters of G proteins in PSCs induced a profound depression in neurotransmitter release.

Actin disassembles reversibly during electrically induced recycling of synaptic vesicles in cultured neurons

We have studied depolarization-induced regulation of actin assembly in exocytotically active areas of dissociated chick sympathetic neurons. Active areas were identified with the fluorescent dye FM1-43 which labels synaptic vesicles that recycle in these regions. Exocytosis (electrically stimulated) was monitored in real time through depletion of FM1-43 fluorescence. To study depolarization-induced disassembly of actin in the FM1-43-stained regions, the cells were fixed after different periods of depolarization and stained with rhodamine phalloidin, which binds preferentially to the filamentous form of actin. In active regions, actin disassembles and reassembles during continuous 2 min depolarization. Actin disassembly that occurs after the first 25 s of depolarization was detected by a reduction in rhodamine phalloidin staining and confirmed by correlative electron microscopy. Immunogold staining revealed that actin is abundant throughout resting terminals. In some experiments, actin filaments were stabilized by loading cells with unlabelled phalloidin before stimulating secretion. Stabilizing the filaments does not alter the initial release but strongly reduces the release rate at later stages. These data are consistent with a model in which partial disassembly of actin filaments is necessary for facilitating the transport of vesicles within the terminal and reassembly is necessary for limiting that movement.

Retrograde signaling in the development and modification of synapses

Retrograde signaling from the postsynaptic cell to the presynaptic neuron is essential for the development, maintenance, and activity-dependent modification of synaptic connections. This review covers various forms of retrograde interactions at developing and mature synapses. First, we discuss evidence for early retrograde inductive events during synaptogenesis and how maturation of presynaptic structure and function is affected by signals from the postsynaptic cell. Second, we review the evidence that retrograde interactions are involved in activity-dependent synapse competition and elimination in developing nervous systems and in long-term potentiation and depression at mature synapses. Third, we review evidence for various forms of retrograde signaling via membrane-permeant factors, secreted factors, and membrane-bound factors. Finally, we discuss the evidence and physiological implications of the long-range propagation of retrograde signals to the cell body and other parts of the presynaptic neuron.

Developmental synaptic depression underlying reorganization of visceral reflex pathways in the spinal cord

During development, neuronal connectivity has a remarkable plasticity. Synaptic refinement in the spinal autonomic nucleus might be involved in the elimination of primitive segmental reflexes and the emergence of mature spinobulbospinal reflexes, which occurs a few weeks after birth. To address this possibility, we examined the postnatal changes of segmental excitatory synaptic transmission by applying the whole-cell recording technique to parasympathetic preganglionic neurons in slice preparations of the rat lumbosacral spinal cord. The mean magnitude of unitary excitatory synaptic currents evoked in preganglionic neurons by stimulation of single interneurons remained unchanged during the first two postnatal weeks but was reduced by 50% during the third postnatal week. This reduction in synaptic efficacy was associated with a decrease in the amount of transmitter release from interneurons. Moreover, this developmental depression of segmental synaptic transmission was prevented by spinal cord transection at the thoracic level on postnatal day 14. Thus, developmental modification of excitatory synapses on preganglionic neurons appears to be attributable to competition between segmental interneuronal and descending bulbospinal inputs, which results in the developmental reorganization of parasympathetic excretory reflex pathways.

A new twist in an old story: the role for crosstalk of neuronal and trophic activity

A number of recent findings suggest a reciprocal interaction between neurotransmitters and neurotrophins functioning at the level of the synapse, which may be relevant not only for plasticity changes in the mature nervous system, but also for the development of synaptic connectivity and for survival or maturation of neurons prior to target contact. Thus, neurotrophin-induced attenuation of frequency-dependent depletion of releasable synaptic vesicle pools of neurotransmitter at synapses may participate in Hebbian and non-Hebbian forms of LTP, as a characteristic of mature synaptic contacts. Subsequent to nerve/target contact, neurotrophins also appear to mediate contact-induced enhancement of neurotransmitter release; this may participate in a developmental improvement of synapse efficacy, stabilization of synaptic contacts, and maturation of "conductive" functional synapses. Coincident with a transmitter-induced elevation of cytosolic Ca2+ levels within growth cones, a local neurotrophin-mediated increase in released neurotransmitter occurring subsequent to stabilization of a distinct synaptic contact may then participate in the refinement of synapses with retention of those neurites affected by neurotrophins and withdrawal of those neurites not affected by neurotrophins. Finally, prior to nerve/target contact, Ca2+ channel-generated spontaneous neuronal activity as well as co-expression of neurotrophins and their receptors may play a role in maturational changes.

Potentiation of developing synapses by postsynaptic release of neurotrophin-4

The hypothesis that synaptic functions can be regulated by neurotrophins secreted from the postsynaptic cell was examined in Xenopus nerve-muscle cultures. Neuromuscular synapses formed on myocytes overexpressing neurotrophin-4 (M+ synapses) exhibited a higher level of spontaneous synaptic activity and enhanced evoked synaptic transmission as compared to those formed on normal control myocytes (M- synapses). The NT-4 effects involve a potentiation of presynaptic transmitter secretion as well as a lengthening of the mean burst duration of postsynaptic low conductance acetylcholine channels. Repetitive stimulation of either the presynaptic neuron or the postsynaptic myocyte led to a potentiation of synaptic transmission at M+ synapses. All potentiation effects of NT-4 overexpression were abolished by the extracellular presence of TrkB-IgG but not by the presence of TrkA-IgG, indicating that postsynaptic secretion of NT-4 was responsible for the synaptic modification.

Temporal correlations between functional and molecular changes in NMDA receptors and GABA neurotransmission in the superior colliculus

Activation of the NMDA subtype of glutamate receptor is required for activity-dependent structural plasticity in many areas of the young brain. Previous work has shown that NMDA receptor currents decline approximately at the time that developmental synaptic plasticity ends, and in situ hybridization studies have suggested that receptor subunit changes may be occurring during the same developmental interval. To establish a system in which the relationship between these properties of developing synapses can be explored, we have combined patch-clamp recordings with mRNA- and protein-level biochemical analyses to study the developmental regulation of NMDA receptors in the superficial layers of the rat superior colliculus. These experiments document an abrupt decrease in the NMDA receptor contribution to synaptic currents that occurs before eye opening and is closely associated with changes in NR1 protein, rapidly rising levels of the NMDA receptor subunit NR2A, and decreasing levels of NR2B. The functional and molecular changes also are correlated with the developmental decline in structural plasticity in these layers. In addition, both physiological and biochemical methods show evidence of GABA-mediated inhibition in the superficial collicular layers beginning after eye opening. This may provide an additional heterosynaptic mechanism for controlling excitation and plasticity in this neuropil by pattern vision. Thus our findings lend support to the idea that high levels of NMDA receptor function are associated with the potential for structural rearrangement in CNS neuropil and that the functional downregulation of this molecule results, at least partially, from changes in its subunit composition.

Propagation of activity-dependent synaptic depression in simple neural networks

Triple whole-cell recordings from simple networks of cultured hippocampal neurons show that Induction of long-term depression at glutamatergic synapses is accompanied by a back propagation of depression to Input synapses on the dendrite of the presynaptic neuron. The depression also propagates laterally to divergent outputs of the presynaptic neuron and to convergent inputs on the postsynaptic neuron. There is no forward propagation of depression to the output of the postsynaptic neuron and no presynaptic propagation accompanying long-term depression at GABAergic synapses. Activity-induced synaptic modification is therefore not restricted to the activated synapse, but selectively propagates throughout the neural network.

Memory, sleep and the evolution of mechanisms of synaptic efficacy maintenance

The origin of both sleep and memory appears to be closely associated with the evolution of mechanisms of enhancement and maintenance of synaptic efficacy. The development of activity-dependent synaptic plasticity apparently was the first evolutionary adaptation of nervous systems beyond a capacity to respond to environmental stimuli by mere reflexive actions. After the origin of activity-dependent synaptic plasticity, whereby single activations of synapses led to short-term efficacy enhancement, lengthy maintenance of enhancements probably was achieved by repetitive activations ("dynamic stabilization"). One source of selective pressure for the evolutionary origin of neurons and neural circuits with oscillatory firing capacities may have been a need for repetitive spontaneous activations to maintain synaptic efficacy in circuits that were in infrequent use. This process is referred to as "non-utilitarian" dynamic stabilization. Dynamic stabilization of synapses in "simple" invertebrates occurs primarily through frequent use. In complex, locomoting forms, it probably occurs through both frequent use and non-utilitarian activations during restful waking. With the evolution of increasing repertories and complexities of behavioral and sensory capabilities--with vision usually being the vastly pre-eminent sense brain complexity increased markedly. Accompanying the greater complexity, needs for storage and maintenance of hereditary and experiential information (memories) increased greatly. It is suggested that these increases led to conflicts between sensory input processing during restful waking and concomitant non-utilitarian dynamic stabilization of infrequently used memory circuits. The selective pressure for the origin of primitive sleep may have been a resulting need to achieve greater depression of central processing of sensory inputs largely complex visual information than occurs during restful waking. The electrical activities of the brain during sleep (aside from those that subserve autonomic activities) may function largely to maintain sleep and to dynamically stabilize infrequently used circuitry encoding memories. Sleep may not have been the only evolutionary adaptation to conflicts between dynamic stabilization and sensory input processing. In some ectothermic vertebrates, sleep may have been postponed or rendered unnecessary by a more readily effected means of resolution of the conflicts, namely, extensive retinal processing of visual information during restful waking. By this means, processing of visual information in central regions of the brain may have been maintained at a sufficiently low level to allow adequate concomitant dynamic stabilization. As endothermy evolved, the skeletal muscle hypotonia of primitive sleep may have become insufficient to prevent sleep-disrupting skeletal muscle contractions during non-utilitarian dynamic stabilization of motor circuitry at the accompanying higher body temperatures and metabolic rates. Selection against such disruption during dynamic stabilization of motor circuitry may have led to the inhibition of skeletal muscle tone during a portion of primitive sleep, the portion designated as rapid-eye-movement sleep. Many marine mammals that are active almost continuously engage only in unihemispheric non-rapid-eye-movement sleep. They apparently do not require rapid-eye-movement sleep and accompanying non-utilitarian dynamic stabilization of motor circuitry, because this circuitry is in virtually continuous use. Studies of hibernation by arctic ground squirrels suggest that each hour of sleep may stabilize brain synapses for as long as 4 h. Phasic irregularities in heart and respiratory rates during rapid-eye-movement sleep may be a consequence of superposition of dynamic stabilization of motor circuitry on the rhythmic autonomic control mechanisms. Some information encoded in circuitry being dynamically stabilized during sleep achieves unconscious awareness in authentic and var

Specificity and development of innervation pattern of Xenopus pectoralis muscle

Multielectrode recordings were used to identify and measure the axonal inputs to each end plate on contiguous surface fibers covering about 25% of the Xenopus pectoralis muscle in mature and developing animals. The mature innervation pattern was remarkably precise. Individual axons tended to innervate fibers of similar input resistance (R(in)) in compact motor units restricted to only a portion of the region studied. Motor units comprising fibers of similar R(in) overlapped mainly near their borders. Most fibers had two end plates. In more than 80% of these fibers, both end plates received input from the same axon. In 57%, this was the only input to both end plates. This implies a powerful mechanism for excluding or eliminating inputs from other axons. About 16% of the mature junctions showed focal polyneuronal innervation, with the weaker end plate potential component often less than 1 mV in noncurarized preparation. However, we have no evidence that the weaker inputs were being eliminated. During development, motor units became more compact, which was associated with synapse elimination; but from the earliest times studied, soon after metamorphosis when many fibers were adding second end plates, a majority of those that had two end plates were innervated at both sites by the same axon.

Regeneration of specific innervation in Xenopus pectoralis muscle

We investigated the motor unit organization and precision of reinnervation in the Xenopus pectoralis muscle following different manipulations, including crush or section of the posterior pectoralis nerve, foreign nerve innervation, and crush coupled with activity modulation or block. Most fibers have two neuromuscular junctions, and multielectrode recordings were used to identify the axonal origin of all inputs to both junctions on most or all fibers covering about 25% of the muscle surface. Following simple nerve crush, a highly organized innervation pattern was restored, indistinguishable from the normal pattern, including selective innervation of fibers of similar input resistance (R(in)), compact motor unit organization, and high incidence of exclusive innervation of both end plates on each fiber by the same axon (distributed mononeuronal innervation, or a/a pattern). Initial reinnervation was equally precise when nerve conduction in the regenerating nerve was blocked by tetrodotoxin. More distant or repeated nerve crush or nerve section delayed and reduced the precision of reinnervation, but the majority of fibers still received input to both end plates by the same axon, often in combination with others. A foreign nerve, the pectoralis sternalis, which in its own muscle forms only single end plates, showed less precise reinnervation, but still had an incidence of a/a innervation far above chance. These data imply the expression and recognition of remarkably precise chemospecific cues even in mature animals, superimposed on which is a further refinement by synapse elimination, probably based on an activity-dependent process.