A mechanistically interpretable model of the retinal neural code for natural scenes with multiscale adaptive dynamics

The visual system processes stimuli over a wide range of spatiotemporal scales, with individual neurons receiving input from tens of thousands of neurons whose dynamics range from milliseconds to tens of seconds. This poses a challenge to create models that both accurately capture visual computations and are mechanistically interpretable. Here we present a model of salamander retinal ganglion cell spiking responses recorded with a multielectrode array that captures natural scene responses and slow adaptive dynamics. The model consists of a three-layer convolutional neural network (CNN) modified to include local recurrent synaptic dynamics taken from a linear-nonlinear-kinetic (LNK) model \cite{ozuysal2012linking}. We presented alternating natural scenes and uniform field white noise stimuli designed to engage slow contrast adaptation. To overcome difficulties fitting slow and fast dynamics together, we first optimized all fast spatiotemporal parameters, then separately optimized recurrent slow synaptic parameters. The resulting full model reproduces a wide range of retinal computations and is mechanistically interpretable, having internal units that correspond to retinal interneurons with biophysically modeled synapses. This model allows us to study the contribution of model units to any retinal computation, and examine how long-term adaptation changes the retinal neural code for natural scenes through selective adaptation of retinal pathways.

Microcircuit synchronization and heavy tailed synaptic weight distribution in preBötzinger Complex contribute to generation of breathing rhythm

The preBötzinger Complex, the mammalian inspiratory rhythm generator, encodes inspiratory time as motor pattern. Spike synchronization throughout this sparsely connected network generates inspiratory bursts albeit with variable latencies after preinspiratory activity onset in each breathing cycle. Using preBötC rhythmogenic microcircuit minimal models, we examined the variability in probability and latency to burst, mimicking experiments. Among various physiologically plausible graphs of 1000 point neurons with experimentally determined neuronal and synaptic parameters, directed Erdős-Rényi graphs best captured the experimentally observed dynamics. Mechanistically, preBötC (de)synchronization and oscillatory dynamics are regulated by the efferent connectivity of spiking neurons that gates the amplification of modest preinspiratory activity through input convergence. Furthermore, to replicate experiments, a lognormal distribution of synaptic weights was necessary to augment the efficacy of convergent coincident inputs. These mechanisms enable exceptionally robust yet flexible preBötC attractor dynamics that, we postulate, represent universal temporal-processing and decision-making computational motifs throughout the brain.

SynapsEM: Computer-Assisted Synapse Morphometry

The structural features of a synapse help determine its function. Synapses are extremely small and tightly packed with vesicles and other organelles. Visualizing synaptic structure requires imaging by electron microscopy, and the features in micrographs must be quantified, a process called morphometry. Three parameters are typically assessed from each specimen: (1) the sizes of individual vesicles and organelles; (2) the absolute number and densities of organelles; and (3) distances between organelles and key features at synapses, such as active zone membranes and dense projections. For data to be meaningful, the analysis must be repeated from hundreds to thousands of images from several biological replicates, a daunting task. Here we report a custom computer program to analyze key structural features of synapses: SynapsEM. In short, we developed ImageJ/Fiji macros to record x,y-coordinates of segmented structures. The coordinates are then exported as text files. Independent investigators can reload the images and text files to reexamine the segmentation using ImageJ. The Matlab program then calculates and reports key synaptic parameters from the coordinates. Since the values are calculated from coordinates, rather than measured from each micrograph, other parameters such as locations of docked vesicles relative to the center of an active zone can be extracted in Matlab by additional scripting. Thus, this program can accelerate the morphometry of synapses and promote a more comprehensive analysis of synaptic ultrastructure.

SynapsEM: computer-assisted synapse morphometry

The structural features of a synapse, in part, determine its output. Synapses are extremely small and tightly packed with vesicles and other organelles. Visualizing synaptic structure requires imaging by electron microscopy, and the features in micrographs must be quantified using morphometry. Three parameters are typically assessed from each specimen: 1) the sizes of individual vesicles and organelles, 2) the absolute number and densities of organelles, and 3) distances between organelles and key features at synapses such as active zone membranes and dense projections. For data to be valid, the analysis must be repeated from hundreds or thousands of images from several biological replicates, a daunting task. Here we report a custom computer program to analyze these features: SynapsEM. In short, we developed macros for ImageJ/Fiji to record x,y-coordinates of segmented structures; the coordinates are then exported as text files. Independent investigators can reload the images and text files to confirm or re-evaluate the segmentation using ImageJ. The Matlab program calculates and reports key synaptic parameters from the coordinates. Since the values are calculated, rather than measured from each micrograph, other parameters can be extracted in Matlab by additional scripting. Thus, this program can accelerate morphometry of synapses and promote a more comprehensive analysis of synaptic ultrastructure.

NMDA-receptor antibodies alter cortical microcircuit dynamics

NMDA-receptor antibodies (NMDAR-Abs) cause an autoimmune encephalitis with a diverse range of EEG abnormalities. NMDAR-Abs are believed to disrupt receptor function, but how blocking this excitatory synaptic receptor can lead to paroxysmal EEG abnormalities—or even seizures—is poorly understood. Here we show that NMDAR-Abs change intrinsic cortical connections and neuronal population dynamics to alter the spectral composition of spontaneous EEG activity and predispose brain dynamics to paroxysmal abnormalities. Based on local field potential recordings in a mouse model, we first validate a dynamic causal model of NMDAR-Ab effects on cortical microcircuitry. Using this model, we then identify the key synaptic parameters that best explain EEG paroxysms in pediatric patients with NMDAR-Ab encephalitis. Finally, we use the mouse model to show that NMDAR-Ab–related changes render microcircuitry critically susceptible to overt EEG paroxysms when these key parameters are changed, even though the same parameter fluctuations are tolerated in the in silico model of the control condition. These findings offer mechanistic insights into circuit-level dysfunction induced by NMDAR-Ab.

Weak and equally balanced synaptic inputs to interneurons in the CA1 hippocampus characterize in vivo rhythmic states

Brain coding strategies are enabled by the balance of synaptic inputs that individual neurons receive as determined by the networks in which they reside. Inhibitory cell types contribute to brain function in distinct ways but recording from specific, inhibitory cell types during behaviour to determine their contributions is difficult. In particular, the in vivo activities of vasoactive intestinal peptide-expressing interneuron specific 3 (IS3) cells in the hippocampus that only target other inhibitory cells are unknown at present. We perform a massive, computational exploration of possible synaptic inputs to IS3 cells using multi-compartment models and optimized synaptic parameters. We find that asynchronous in vivo-like states that are sensitive to additional theta-timed inputs exist when excitatory and inhibitory synaptic conductances are equally balanced and there are low amounts of correlated inputs. Thus, using a generally applicable computational approach we predict the existence of balanced states in hippocampal circuits during rhythmic activities.

The Enhanced Rise and Delayed Fall of Memory in a Model of Synaptic Integration: Extension to Discrete State Synapses

Integrate-and-express models of synaptic plasticity propose that synapses may act as low-pass filters, integrating synaptic plasticity induction signals in order to discern trends before expressing synaptic plasticity. We have previously shown that synaptic filtering strongly controls destabilizing fluctuations in developmental models. When applied to palimpsest memory systems that learn new memories by forgetting old ones, we have also shown that with binary-strength synapses, integrative synapses lead to an initial memory signal rise before its fall back to equilibrium. Such an initial rise is in dramatic contrast to nonintegrative synapses, in which the memory signal falls monotonically. We now extend our earlier analysis of palimpsest memories with synaptic filters to consider the more general case of discrete state, multilevel synapses. We derive exact results for the memory signal dynamics and then consider various simplifying approximations. We show that multilevel synapses enhance the initial rise in the memory signal and then delay its subsequent fall by inducing a plateau-like region in the memory signal. Such dynamics significantly increase memory lifetimes, defined by a signal-to-noise ratio (SNR). We derive expressions for optimal choices of synaptic parameters (filter size, number of strength states, number of synapses) that maximize SNR memory lifetimes. However, we find that with memory lifetimes defined via mean-first-passage times, such optimality conditions do not exist, suggesting that optimality may be an artifact of SNRs.