Uncorrelated Neural Firing in Mouse Visual Cortex during Spontaneous Retinal Waves.

Matthew T Colonnese,Yasunobu Murata,Jing Shen

doi:10.3389/fncel.2017.00289

Matthew T Colonnese, Yasunobu Murata + Show 1 more

Open Access

https://doi.org/10.3389/fncel.2017.00289

Copy DOI

Abstract

Synchronous firing among the elements of forming circuits is critical for stabilization of synapses. Understanding the nature of these local network interactions during development can inform models of circuit formation. Within cortex, spontaneous activity changes throughout development. Unlike the adult, early spontaneous activity occurs in discontinuous population bursts separated by long silent periods, suggesting a high degree of local synchrony. However, whether the micro-patterning of activity within early bursts is unique to this early age and specifically tuned for early development is poorly understood, particularly within the column. To study this we used single-shank multi-electrode array recordings of spontaneous activity in the visual cortex of non-anesthetized neonatal mice to quantify single-unit firing rates, and applied multiple measures of network interaction and synchrony throughout the period of map formation and immediately after eye-opening. We find that despite co-modulation of firing rates on a slow time scale (hundreds of ms), the number of coactive neurons, as well as pair-wise neural spike-rate correlations, are both lower before eye-opening. In fact, on post-natal days (P)6–9 correlated activity was lower than expected by chance, suggesting active decorrelation of activity during early bursts. Neurons in lateral geniculate nucleus developed in an opposite manner, becoming less correlated after eye-opening. Population coupling, a measure of integration in the local network, revealed a population of neurons with particularly strong local coupling present at P6–11, but also an adult-like diversity of coupling at all ages, suggesting that a neuron’s identity as locally or distally coupled is determined early. The occurrence probabilities of unique neuronal “words” were largely similar at all ages suggesting that retinal waves drive adult-like patterns of co-activation. These findings suggest that the bursts of spontaneous activity during early visual development do not drive hyper-synchronous activity within columns. Rather, retinal waves provide windows of potential activation during which neurons are active but poorly correlated, adult-like patterns of correlation are achieved soon after eye-opening.

Highlights

Connectivity during development is achieved by synapse formation under the control of molecular guidance cues, and modification of these synapses by neural activity (Katz and Shatz, 1996)
We used multi-electrode array based spike-sorting in very young, head-fixed mice to quantify features of local network interaction within a visual cortical column during the period of spontaneous retinal waves and compared these to spontaneous activity in the week after eye-opening, when many aspects of ongoing cortical activity emerge (Rochefort et al, 2009, 2011; Hoy and Niell, 2015)
Our primary finding is that—despite the macro-patterning of activity present in cortex during the period of retinal waves, in which long silent periods are interrupted by large oscillations that control firing times (Hanganu et al, 2006; Colonnese and Khazipov, 2010), suggesting hypersynchrony of the developing network—the firing of cortical neurons is remarkably uncorrelated and adult-like patterns of network interaction are achieved remarkably early, either during the period of glutamatergic retinal waves or immediately after eye-opening

Summary

Introduction

Connectivity during development is achieved by synapse formation under the control of molecular guidance cues, and modification of these synapses by neural activity (Katz and Shatz, 1996). One possibility (refinement model) is that all neurons respond synchronously to low-frequency maps such as topography before network fractionation into local microcircuits, such as for orientation or direction selectivity (White and Fitzpatrick, 2007; Butts and Kanold, 2010). Such synchronization would be driven by anatomical hyperconnectivity, and by circuit properties such as weak inhibition, excitatory GABAA currents, long channel decay-time, abundant electrical connectivity, and high neuron excitability which increase synchronization and reduce the specificity of neuronal responses (Blankenship and Feller, 2010; Cossart, 2011; Dehorter et al, 2012). A constructionist perspective would predict that synchronization emerges gradually and in parallel with the specialization of neuronal function by cell class and/or response properties (Erwin and Miller, 1998; Crowley and Katz, 2002)

Methods

Results

Conclusion