Addition of new neurons and the emergence of a local neural circuit for precise timing

Abstract

1During development, neurons arrive at local brain areas in extended period of time, but how 2 they form local neural circuits is unknown. Here we computationally model the emergence of 3 a network for precise timing in the premotor nucleus HVC in songbird. We show that new 4 motor projection neurons, mostly added to HVC before and during song learning, are recruited 5 to the end of a growing feedforward network. High spontaneous activity of new neurons makes 6 them the prime targets for recruitment in a self-organized process via synaptic plasticity. Once 7 recruited, the new neurons fire readily at precise times, and they become mature. Neurons that 8 are not recruited become silent and replaced by new immature neurons. Our model incorporates 9 realistic HVC features such as interneurons, spatial distributions of neurons, and distributed 10 axonal delays. The model predicts that the birth order of the projection neurons correlates with 11 their burst timing during the song. 12 Significance Statement 13 Functions of local neural circuits depend on their specific network structures, but how the net-14 works are wired is unknown. We show that such structures can emerge during development 15 through a self-organized process, during which the network is wired by neuron-by-neuron re-16 cruitment. This growth is facilitated by steady supply of immature neurons, which are highly 17 excitable and plastic. We suggest that neuron maturation dynamics is an integral part of con-18 structing local neural circuits. 19 During development, the birth order of neurons plays a critical role in constructing the brain's 21 large-scale structures. In mammalian cortex, neurons that are destined to the deep cortical 22 layers are born earlier than those to the superficial layers [1, 2]. In rodent hippocampus, earlier 23 born neurons and late born neurons form distinctive parallel circuits through the hippocampal 24 pathway [3]. However, whether birth order is also important in constructing microcircuits in 25 local brain areas is unknown [4]. The premotor nucleus HVC (proper name) of the zebra finch 26 provides an excellent opportunity to investigate this issue. 27HVC is a premotor nucleus that drives singing of the courtship song in the zebra finch [5, 6]. 28 An adult zebra finch sings repetitions of a motif consisting of fixed sequence of syllables [7]. 29 Excitatory HVC neurons that project to the downstream premotor area RA (robust nucleus of 30 the arcopallium) encode the timing of acoustic features of the song [8]. Each HVCRA neuron 31 bursts once during the motif [8, 9]. As a population, HVCRA neurons sequentially burst through 32 the entire motif [10, 11]. 33 There is strong evidence that the sequential bursting of HVCRA neurons is generated within 34 HVC [12, 9, 13, 14]. Moreover, HVCRA neurons most likely form a feedforward synaptic chain 35 network, which supports propagation of burst spikes [15, 9]. Such a microcircuit in HVC acts as 36 an infrastructure for subsequent learning of the song, during whi...

Addition of new neurons and the emergence of a local neural circuit for precise timing

“…During development, immature neurons in many neural circuits across multiple species go through a period of depolarizing inhibition before switching to hyperpolarizing inhibition, which is caused by an elevated GABA reversal potential on immature neurons [ 51 ]. Our computational experiments with developmental switch in GABA resulted in the emergence of numerous connections between nearby HVC RA neurons [ 52 ]. This was because the dense local connectivity between HVC RA and HVC INT neurons promotes recruitment of nearby immature neurons through depolarizing local inhibition.…”

“…Our growth mechanism is robust with respect to the changes in the model parameter values. The use of different strength of inhibitory connections (varied between G ie = 0.015 mS / cm 2 and G ie = 0.060 mS / cm 2 ), different number of efferent supersynaptic connections ( N s = 10 and N s = 20), and different maximal strength of excitatory connections between HVC RA neurons (between G max = 1.5 nS and G max = 4 nS ) lead to the emergence of precisely timed neural sequences [ 52 ]. Thus our modeling results do not rely on fine-tuning of the model parameters.…”