Scientific Understanding of Consciousness
Consciousness as an Emergent Property of Thalamocortical Activity

Axonal Synapse Sorting in Medial Entorhinal Cortex


Nature  549, 469–475 (28 September 2017)

Axonal synapse sorting in medial entorhinal cortex

Helene Schmidt,

Department of Connectomics, Max Planck Institute for Brain Research, D-60438 Frankfurt, Germany

Bernstein Center for Computational Neuroscience, Humboldt University, D-10115 Berlin, Germany

NeuroCure Cluster of Excellence, Humboldt University, D-10115 Berlin, Germany


Research on neuronal connectivity in the cerebral cortex has focused on the existence and strength of synapses between neurons, and their location on the cell bodies and dendrites of postsynaptic neurons. The synaptic architecture of individual presynaptic axonal trees, however, remains largely unknown. Here we used dense reconstructions from three-dimensional electron microscopy in rats to study the synaptic organization of local presynaptic axons in layer 2 of the medial entorhinal cortex, the site of grid-like spatial representations. We observe path-length-dependent axonal synapse sorting, such that axons of excitatory neurons sequentially target inhibitory neurons followed by excitatory neurons. Connectivity analysis revealed a cellular feedforward inhibition circuit involving wide, myelinated inhibitory axons and dendritic synapse clustering. Simulations show that this high-precision circuit can control the propagation of synchronized activity in the medial entorhinal cortex, which is known for temporally precise discharges.

Ultrastructural analysis of cortical synaptic connectivity by electron microscopy has typically been limited to small volumes of tens of micrometres in extent. Similarly, connectivity analysis using multiple intracellular electrical recordings in brain slices is typically limited to testing small numbers of connections within an individual brain slice. Only recently, larger-scale high-resolution three-dimensional imaging of neuronal circuits using electron microscopy has become feasible for volumes extending to several hundred micrometres for at least two dimensions, whereas this was previously unique to electrical recordings. These approaches allow the study of locally complete synaptic in- and output maps. Especially for mapping synapses along axons, the path length of the reconstructed axon is the key constraining factor, and this is limited by the smallest of the three imaged and reconstructed dimensions (40–52 μm in previous studies in the cortex.

Here we used serial block-face scanning electron microscopy (SBEM) and skeleton-based connectomic data analysis to investigate the neuronal circuitry in layer 2 (L2) of the medial entorhinal cortex (MEC) of rats in three-dimensional electron microscopy datasets for which the smallest dimensions were 274 μm (juvenile, 25-day-old rat (P25)) and 101 μm (adult, 90-day-old rat (P90)). The second dataset was acquired and analysed after analysis of the first, yielding a reproducibility control of the results presented here.

Previous electrical recording studies of the MEC have found that connectivity between excitatory neurons to is absent or sparse, suggesting types of attractor models in the MEC that are based on purely inhibitory connectivity between excitatory neurons.

We find that at least 30% of the output synapses of excitatory neurons are made onto other excitatory targets. Notably, this excitatory connectivity had a peculiar distance dependence: when investigating the output synapses along the axons of excitatory neurons. we find that inhibitory neurons are targeted first, offset by about 120 μm along the path length of the axon to the innervation of excitatory neurons (path-length-dependent axonal synapse sorting (PLASS)). We furthermore find that axons frequently provide multiple closely spaced synapses on the same postsynaptic dendrites, further enhancing the ability of the excitatory neuron to activate the postsynaptic neurons at high temporal precision in a cellular feedforward inhibition (cFFI) circuit. Our results reveal a level of synaptic specialization in the cerebral cortex that is beyond average cell-to-cell connectivity and emphasize the need for high-resolution connectomic circuit mapping. Using numerical simulations, we show that this circuit could enhance spike timing precision, and could control the propagation of synchronized activity.

We discovered PLASS of output synapses along the axons of excitatory neurons in the rat MEC, a previously undescribed level of specificity in neuronal circuits of the mammalian cerebral cortex. We found that PLASS acts in a cFFI circuit, in which synapses are frequently clustered, in particular on the dendrites of postsynaptic interneurons. The inhibitory branch of the circuit appears to be shaped for fast transmission of action potentials using myelinated, large-diameter axons. This high-precision circuit in the MEC may be in place to sharpen the timing of action potentials, and to control the propagation of highly synchronous activity in a cortex that is occupied with spatial sequence analysis.

Our data are the first, to our knowledge, to demonstrate the positional sorting of output synapses in the mammalian cortex. However, in hindsight, data from the mouse visual cortex and hippocampus can be interpreted as an indication that PLASS may operate in various cortices. In these studies, it had been noted that the fraction of synapses targeting interneurons was higher than expected on average. Because these studies were limited in electron-microscopy-reconstructed volume, the reconstructed axons were necessarily only very proximal. The fact that our volume was at least threefold larger in the third dimension made it possible to detect PLASS as a transition from interneuron-dominated to ExN-dominated targeting within the same excitatory axons, and to determine the properties of the local cFFI circuit.

Data on axonal conduction velocity and latencies of local inhibition in the cortex make it plausible that the described circuit can prevent the propagation of highly synchronous activity in L2 of the rat MEC (for this, propagation of action potentials along the axon would have to be between 120–240 μm ms−1 and local inhibitory action potential-to-PSP latencies 0.5–1 ms). This is especially noteworthy because so far, feedforward inhibition circuits have been interpreted to selectively propagate synchronous, but not asynchronous activity. By contrast, the PLASS–cFFI circuit can act as a synchrony block. At the same time, if the postsynaptic neuron was to receive additional excitatory input or an additional modulation of the underlying membrane potential, such as the theta-frequency oscillation that can be found in the MEC, the PLASS–cFFI circuit could allow a controlled, predictive gating of synchronous activity propagation. The same would be possible by disinhibition of the involved interneurons.

Our data show a high-precision wiring motif in the cortex, revealing an unexpected level of structural specialization in local cortical circuits. We explored possible functional implications of these structural findings, pointing towards an effect on spike timing precision and the control of synchronized activity propagation. Connectomic analysis of other cortices will allow us to determine if PLASS constitutes a general cortical wiring principle in mammals.

[end of paraphrase]


Return to — Neural Network