November 19, 2018
Precise synaptic organization of a memory network
In order to examine how synaptic inputs to individual neurons are organized in a distributed memory network, researchers from the Friedrich group performed electrophysiological recordings in the zebrafish homolog of the olfactory cortex. They found that large excitatory inputs were paired with large inhibitory inputs, which may seem counterintuitive because inhibition cancels excitation. However, theoretical models predicted such a ‘precise balance’ of synaptic inputs because it stabilizes recurrent networks that store memories.
Two antagonistic forces are necessary to build a working nervous system: excitation and inhibition. Excitatory/inhibitory signaling from one cell to the next makes the latter cell more/less likely to fire. The balance between neural excitation and neural inhibition is crucial to healthy cognition and behavior. Insights into the tuning of excitatory and inhibitory inputs to individual neurons will help understand how a neural circuit works.
Another important concept in neuroscience is the ‘balanced state’. That is what defines a neuronal circuit when excitation and inhibition are both strong, and establish an average membrane potential near spike threshold. In this state, only a weak input is necessary to get them to fire, which allows a network to react very fast to changing inputs. However, such networks are often difficult to reconcile with memory functions because they tend to show chaotic behavior. Moreover, memory storage is thought to involve specific excitatory connections that amplify learned inputs, which may create ‘runaway’ activity as in an avalanche. In theory, these problems could be resolved if excitatory and inhibitory synaptic activity is matched – or ‘balanced’ – not only on average, across a population of neurons, but in every single neuron. In fact, such a 'precise balance' should make for very powerful memory networks. However, this idea has been difficult to test experimentally because recordings of synaptic inputs in individual neurons requires the voltage-clamp technique. Unfortunately, physical constraints often make it difficult or impossible to apply this technique in large neurons such as cortical neurons of rodents.
Peter Rupprecht, a PhD student in the Friedrich lab, realized that these limitations could be overcome in the smaller brain of zebrafish. He thus performed voltage-clamp recordings in the zebrafish homolog of olfactory cortex (called telencephalic area Dp) to analyze the synaptic inputs to these neurons during odor responses. This network is not topographically organized and has been proposed to function as a memory network; but the organization of its synaptic interactions is not well understood. Rupprecht's results – which have been published in Neuron – revealed that excitatory and inhibitory synaptic inputs to individual neurons in Dp were large, and for a short time established a balanced state during an odor response. Moreover, excitation and inhibition were co-tuned in individual neurons, showing that the balance is ‘precise’.
"We observed that excitation and inhibition go together for every single neuron during odor stimulation; basically, we obtained direct experimental evidence of what the theoreticians have been hypothesizing, namely that excitation and inhibition are strong and mirror-symmetric," says first author Rupprecht. "It would now be interesting to find out how the highly specific connectivity required to create such a network is established by developmental mechanisms and by experience."
A precise synaptic balance was not observed in another olfactory brain region. “We therefore hypothesize that the synaptic architecture of olfactory cortex is specialized for memory functions,” says Friedrich. "What should be explored next is whether the principle of precise synaptic balance exists also in other (memory) networks and how it is implemented in network connectivity."
Peter Rupprecht, Rainer Friedrich (2018). Precise Synaptic Balance in the Zebrafish Homolog of Olfactory Cortex. Neuron. doi.org/10.1016/j.neuron.2018.09.013