Concept: Action potential
Information in a computer is quantified by the number of bits that can be stored and recovered. An important question about the brain is how much information can be stored at a synapse through synaptic plasticity, which depends on the history of probabilistic synaptic activity. The strong correlation between size and efficacy of a synapse allowed us to estimate the variability of synaptic plasticity. In an EM reconstruction of hippocampal neuropil we found single axons making two or more synaptic contacts onto the same dendrites, having shared histories of presynaptic and postsynaptic activity. The spine heads and neck diameters, but not neck lengths, of these pairs were nearly identical in size. We found that there is a minimum of 26 distinguishable synaptic strengths, corresponding to storing 4.7 bits of information at each synapse. Because of stochastic variability of synaptic activation the observed precision requires averaging activity over several minutes.
Progressive multiple sclerosis (MS) is a severely disabling neurological condition, and an effective treatment is urgently needed. Recently, high-dose biotin has emerged as a promising therapy for affected individuals. Initial clinical data have shown that daily doses of biotin of up to 300 mg can improve objective measures of MS-related disability. In this article, we review the biology of biotin and explore the properties of this ubiquitous coenzyme that may explain the encouraging responses seen in patients with progressive MS. The gradual worsening of neurological disability in patients with progressive MS is caused by progressive axonal loss or damage. The triggers for axonal loss in MS likely include both inflammatory demyelination of the myelin sheath and primary neurodegeneration caused by a state of virtual hypoxia within the neuron. Accordingly, targeting both these pathological processes could be effective in the treatment of progressive MS. Biotin is an essential co-factor for five carboxylases involved in fatty acid synthesis and energy production. We hypothesize that high-dose biotin is exerting a therapeutic effect in patients with progressive MS through two different and complementary mechanisms: by promoting axonal remyelination by enhancing myelin production and by reducing axonal hypoxia through enhanced energy production.
BackgroundCharcot-Marie-Tooth type 1A disease (CMT1A) is a rare orphan inherited neuropathy caused by an autosomal dominant duplication of a gene encoding for the structural myelin protein PMP22, which induces abnormal Schwann cell differentiation and dysmyelination, eventually leading to axonal suffering then loss and muscle wasting. We favour the idea that diseases can be more efficiently treated when targeting multiple disease-relevant pathways. In CMT1A patients, we therefore tested the potential of PXT3003, a low-dose combination of three already approved compounds (baclofen, naltrexone and sorbitol). Our study conceptually builds on preclinical experiments highlighting a pleiotropic mechanism of action that includes downregulation of PMP22. The primary objective was to assess safety and tolerability of PXT3003. The secondary objective aimed at an exploratory analysis of efficacy of PXT3003 in CMT1A, to be used for designing next clinical development stages (Phase 2b/3).Methods80 adult patients with mild-to-moderate CMT1A received in double-blind for 1 year Placebo or one of the three increasing doses of PXT3003 tested, in four equal groups. Safety and tolerability were assessed with the incidence of related adverse events. Efficacy was assessed using the Charcot-Marie-Tooth Neuropathy Score (CMTNS) and the Overall Neuropathy Limitations Scale (ONLS) as main endpoints, as well as various clinical and electrophysiological outcomes.ResultsThis trial confirmed the safety and tolerability of PXT3003. The highest dose (HD) showed consistent evidence of improvement beyond stabilization. CMTNS and ONLS, with a significant improvement of respectively of 8% (0.4% - 16.2%) and 12.1% (2% - 23.2%) in the HD group versus the pool of all other groups, appear to be the most sensitive clinical endpoints to treatment despite their quasi-stability over one year under Placebo. Patients who did not deteriorate over one year were significantly more frequent in the HD group.ConclusionsThese results confirm that PXT3003 deserves further investigation in adults and could greatly benefit CMT1A-diagnosed children, usually less affected than adults.Trial registrationEudraCT Number: 2010-023097-40. ClinicalTrials.gov Identifier: NCT01401257. The Committee for Orphan Medicinal Products issued in February 2014 a positive opinion on the application for orphan designation for PXT3003 (EMA/OD/193/13).
- Proceedings of the National Academy of Sciences of the United States of America
- Published almost 3 years ago
Deep networks are now able to achieve human-level performance on a broad spectrum of recognition tasks. Independently, neuromorphic computing has now demonstrated unprecedented energy-efficiency through a new chip architecture based on spiking neurons, low precision synapses, and a scalable communication network. Here, we demonstrate that neuromorphic computing, despite its novel architectural primitives, can implement deep convolution networks that (i) approach state-of-the-art classification accuracy across eight standard datasets encompassing vision and speech, (ii) perform inference while preserving the hardware’s underlying energy-efficiency and high throughput, running on the aforementioned datasets at between 1,200 and 2,600 frames/s and using between 25 and 275 mW (effectively >6,000 frames/s per Watt), and (iii) can be specified and trained using backpropagation with the same ease-of-use as contemporary deep learning. This approach allows the algorithmic power of deep learning to be merged with the efficiency of neuromorphic processors, bringing the promise of embedded, intelligent, brain-inspired computing one step closer.
Background The senses of touch and proprioception evoke a range of perceptions and rely on the ability to detect and transduce mechanical force. The molecular and neural mechanisms underlying these sensory functions remain poorly defined. The stretch-gated ion channel PIEZO2 has been shown to be essential for aspects of mechanosensation in model organisms. Methods We performed whole-exome sequencing analysis in two patients who had unique neuromuscular and skeletal symptoms, including progressive scoliosis, that did not conform to standard diagnostic classification. In vitro and messenger RNA assays, functional brain imaging, and psychophysical and kinematic tests were used to establish the effect of the genetic variants on protein function and somatosensation. Results Each patient carried compound-inactivating variants in PIEZO2, and each had a selective loss of discriminative touch perception but nevertheless responded to specific types of gentle mechanical stimulation on hairy skin. The patients had profoundly decreased proprioception leading to ataxia and dysmetria that were markedly worse in the absence of visual cues. However, they had the ability to perform a range of tasks, such as walking, talking, and writing, that are considered to rely heavily on proprioception. Conclusions Our results show that PIEZO2 is a determinant of mechanosensation in humans. (Funded by the National Institutes of Health Intramural Research Program.).
We often perform movements and actions on the basis of internal motivations and without any explicit instructions or cues. One common example of such behaviors is our ability to initiate movements solely on the basis of an internally generated sense of the passage of time. In order to isolate the neuronal signals responsible for such timed behaviors, we devised a task that requires nonhuman primates to move their eyes consistently at regular time intervals in the absence of any external stimulus events and without an immediate expectation of reward. Despite the lack of sensory information, we found that animals were remarkably precise and consistent in timed behaviors, with standard deviations on the order of 100 ms. To examine the potential neural basis of this precision, we recorded from single neurons in the lateral intraparietal area (LIP), which has been implicated in the planning and execution of eye movements. In contrast to previous studies that observed a build-up of activity associated with the passage of time, we found that LIP activity decreased at a constant rate between timed movements. Moreover, the magnitude of activity was predictive of the timing of the impending movement. Interestingly, this relationship depended on eye movement direction: activity was negatively correlated with timing when the upcoming saccade was toward the neuron’s response field and positively correlated when the upcoming saccade was directed away from the response field. This suggests that LIP activity encodes timed movements in a push-pull manner by signaling for both saccade initiation towards one target and prolonged fixation for the other target. Thus timed movements in this task appear to reflect the competition between local populations of task relevant neurons rather than a global timing signal.
Behavioral output of neural networks depends on a delicate balance between excitatory and inhibitory synaptic connections. However, it is not known whether network formation and stability is constrained by the sign of synaptic connections between neurons within the network. Here we show that switching the sign of a synapse within a neural circuit can reverse the behavioral output. The inhibitory tyramine-gated chloride channel, LGC-55, induces head relaxation and inhibits forward locomotion during the Caenorhabditis elegans escape response. We switched the ion selectivity of an inhibitory LGC-55 anion channel to an excitatory LGC-55 cation channel. The engineered cation channel is properly trafficked in the native neural circuit and results in behavioral responses that are opposite to those produced by activation of the LGC-55 anion channel. Our findings indicate that switches in ion selectivity of ligand-gated ion channels (LGICs) do not affect network connectivity or stability and may provide an evolutionary and a synthetic mechanism to change behavior.
Current methods for studying central nervous system myelination necessitate permissive axonal substrates conducive to myelin wrapping by oligodendrocytes. We have developed a neuron-free culture system in which electron-spun nanofibers of varying sizes substitute for axons as a substrate for oligodendrocyte myelination, thereby allowing manipulation of the biophysical elements of axonal-oligodendroglial interactions. To investigate axonal regulation of myelination, this system effectively uncouples the role of molecular (inductive) cues from that of biophysical properties of the axon. We use this method to uncover the causation and sufficiency of fiber diameter in the initiation of concentric wrapping by rat oligodendrocytes. We also show that oligodendrocyte precursor cells display sensitivity to the biophysical properties of fiber diameter and initiate membrane ensheathment before differentiation. The use of nanofiber scaffolds will enable screening for potential therapeutic agents that promote oligodendrocyte differentiation and myelination and will also provide valuable insight into the processes involved in remyelination.
An unusual feature of the cerebellar cortex is that its output neurons, Purkinje cells, release GABA (γ-aminobutyric acid). Their high intrinsic firing rates (50 Hz) and extensive convergence predict that their target neurons in the cerebellar nuclei would be largely inhibited unless Purkinje cells pause their spiking, yet Purkinje and nuclear neuron firing rates do not always vary inversely. One indication of how these synapses transmit information is that populations of Purkinje neurons synchronize their spikes during cerebellar behaviours. If nuclear neurons respond to Purkinje synchrony, they may encode signals from subsets of inhibitory inputs. Here we show in weanling and adult mice that nuclear neurons transmit the timing of synchronous Purkinje afferent spikes, owing to modest Purkinje-to-nuclear convergence ratios (∼40:1), fast inhibitory postsynaptic current kinetics (τ(decay) = 2.5 ms) and high intrinsic firing rates (∼90 Hz). In vitro, dynamically clamped asynchronous inhibitory postsynaptic potentials mimicking Purkinje afferents suppress nuclear cell spiking, whereas synchronous inhibitory postsynaptic potentials entrain nuclear cell spiking. With partial synchrony, nuclear neurons time-lock their spikes to the synchronous subpopulation of inputs, even when only 2 out of 40 afferents synchronize. In vivo, nuclear neurons reliably phase-lock to regular trains of molecular layer stimulation. Thus, cerebellar nuclear neurons can preferentially relay the spike timing of synchronized Purkinje cells to downstream premotor areas.
Local supra-linear summation of excitatory inputs occurring in pyramidal cell dendrites, the so-called dendritic spikes, results in independent spiking dendritic sub-units, which turn pyramidal neurons into two-layer neural networks capable of computing linearly non-separable functions, such as the exclusive OR. Other neuron classes, such as interneurons, may possess only a few independent dendritic sub-units, or only passive dendrites where input summation is purely sub-linear, and where dendritic sub-units are only saturating. To determine if such neurons can also compute linearly non-separable functions, we enumerate, for a given parameter range, the Boolean functions implementable by a binary neuron model with a linear sub-unit and either a single spiking or a saturating dendritic sub-unit. We then analytically generalize these numerical results to an arbitrary number of non-linear sub-units. First, we show that a single non-linear dendritic sub-unit, in addition to the somatic non-linearity, is sufficient to compute linearly non-separable functions. Second, we analytically prove that, with a sufficient number of saturating dendritic sub-units, a neuron can compute all functions computable with purely excitatory inputs. Third, we show that these linearly non-separable functions can be implemented with at least two strategies: one where a dendritic sub-unit is sufficient to trigger a somatic spike; another where somatic spiking requires the cooperation of multiple dendritic sub-units. We formally prove that implementing the latter architecture is possible with both types of dendritic sub-units whereas the former is only possible with spiking dendrites. Finally, we show how linearly non-separable functions can be computed by a generic two-compartment biophysical model and a realistic neuron model of the cerebellar stellate cell interneuron. Taken together our results demonstrate that passive dendrites are sufficient to enable neurons to compute linearly non-separable functions.