Rewiring the connectome: Evidence and effects

Abstract

Neuronal connections form the physical basis for communication in the brain. Recently, there has been much interest in mapping the "connectome" to understand how brain structure gives rise to brain function, and ultimately, to behaviour. These attempts to map the connectome have largely assumed that connections are stable once formed. Recent studies, however, indicate that connections in mammalian brains may undergo rewiring during learning and experience-dependent plasticity. This suggests that the connectome is more dynamic than previously thought. To what extent can neural circuitry be rewired in the healthy adult brain? The connectome has been subdivided into multiple levels of scale, from synapses and microcircuits through to long-range tracts. Here, we examine the evidence for rewiring at each level. We then consider the role played by rewiring during learning. We conclude that harnessing rewiring offers new avenues to treat brain diseases.

Keywords: Axon; Behaviour; Cognition; Computational; Cortex; Dendrite; Learning; Memory; Network; Neuron; Neuropsychiatric; Stroke; Synapse; fMRI.

Fig. 1

Different effects of rewiring. Each neuron is depicted as a node in a network diagram (Holme and Saramäki, 2012). The connection between a pair of neurons is represented by a line termed and edge, which is drawn between the nodes. Information about the number and strength of synapses forming the connection between neurons can be included in the model by varying the thickness of the edge between the pair of nodes representing the connected neurons. (A) Synaptic rewiring. The thickness of the edge for the i → iii connection increases from t0 to t1 to represent more synapses at this connection. Note that the overall pattern of the edges and the connected nodes in the network do not change. (B) Rewiring of entire connections. The diagram represents loss of the connection, i → ii and formation of a new connection, i → iv. Both the pattern of the edges and the connected nodes in network have changed as a result of this type of rewiring.

Fig. 2

Scales of the connectome. Macro-connectome: Post-mortem human brain. Meso-connectome: Z-stack reconstruction of synaptically-connected pyramidal neurons imaged with confocal microscopy (Cheetham et al., 2007). Micro-connectome: electron microscope section through an asymmetric, presumably excitatory synapse (Cheetham et al., 2014).

Fig. 3

Investigating the connectome. Examples of techniques used to investigate the connectome at different scales. (A) Micro-connectome: electron microscope section through an asymmetric, presumably excitatory synapse (Cheetham et al., 2014). (B) Meso-connectome: Z-stack reconstruction of synaptically-connected pyramidal neurons imaged with confocal microscopy (Cheetham et al., 2007). (C) Human post-mortem brain. (1) Functional connectivity is inferred from brain regions exhibiting temporally-correlated fluctuations in blood-oxygen-level-dependent (BOLD) functional MRI signal (Biswal et al., 1995). (2) White matter tracts can be studied with diffusion tensor imaging (DTI) tractography (Basser et al., 2000). (3) Tracing methods based on molecular tracers or viruses encoding a fluorophore show the long-range axonal projections between brain regions. These tracing studies do not identify both the presynaptic and postsynaptic neurons that form each connection. Transsynaptic tracing technologies have been developed to achieve this e.g. by modifying viruses, such as the rabies virus (DeNardo et al., 2015; Miyamichi et al., 2011; Wickersham et al., 2007; Zingg et al., 2017, 2014). More recently, viral vectors used for tracing have been engineered to encode a fluorophore for tracing and a second protein, such as the calcium indicator GCaMP6s. This facilitates study of the structure of neurons whose activity has been studied in vivo (Wertz et al., 2015). (4) High-resolution two-photon imaging of neurons filled with a fluorophore has been used to follow structural changes in dendritic spines and axonal boutons during learning and experience-dependent plasticity (Holtmaat and Svoboda, 2009). (5) The gold standard for the structural study of synapses is electron microscopy (EM) (Bock et al., 2011; Knott et al., 2002). Focused Ion Beam scanning electron microscopy (FIBSEM) gives serial-section EM images through tissue several micrometres thick (Helmstaedter et al., 2013; Knott et al., 2011). (6) mGRASP (mammalian green fluorescent protein (GFP) reconstitution across synaptic partners) combines tracing with the identification of pre-synaptic and post-synaptic structures. It has been used to study excitatory inputs from CA3 onto CA1 pyramidal neurons in the hippocampus (Druckmann et al., 2014). (7) Array tomography combines ultrathin sectioning with fluorescent labelling of synaptic proteins (Micheva and Smith, 2007). It has been proposed that combining mGRASP with array tomography would facilitate investigation of the meso-connectome with synaptic level resolution (Rah et al., 2015). (8) Calcium imaging uses a calcium-sensitive fluorophore inside the neuron to report when the neuron fires an action potential e.g. (Margolis et al., 2012). (9) Optogenetics can be combined with a variety of techniques to investigate functional circuitry at all three scales of the connectome (Kim et al., 2017). (10–12) Electrophysiological methods are used to investigate the neural activity in the macro-, meso- and micro-connectome. (13) Changes in neural activity within the macro-, meso- or micro-connectome scale circuit may have consequences for behaviour.