Synaptic promiscuity in brain development

Abstract

Precise synaptic connectivity is a prerequisite for the function of neural circuits, yet individual neurons, taken out of their developmental context, readily form unspecific synapses. How does the genome encode brain wiring in light of this apparent contradiction? Synaptic specificity is the outcome of a long series of developmental processes and mechanisms before, during and after synapse formation. How much promiscuity is permissible or necessary at the moment of synaptic partner choice depends on the extent to which prior development restricts available partners or subsequent development corrects initially made synapses. Synaptic promiscuity at the moment of choice can thereby play important roles in the development of precise connectivity, but also facilitate developmental flexibility and robustness. In this review, we assess the experimental evidence for the prevalence and roles of promiscuous synapse formation during brain development. Many well-established experimental approaches are based on developmental genetic perturbation and an assessment of synaptic connectivity only in the adult; this can make it difficult to pinpoint when a given defect or mechanism occurred. In many cases, such studies reveal mechanisms that restrict partner availability already prior to synapse formation. Subsequently, at the moment of choice, factors including synaptic competency, interaction dynamics and molecular recognition further restrict synaptic partners. The discussion of the development of synaptic specificity through the lens of synaptic promiscuity suggests an algorithmic process based on neurons capable of promiscuous synapse formation that are continuously prevented from making the wrong choices, with no single mechanism or developmental time point sufficient to explain the outcome.

Figure 1

Synaptic specificity is the outcome of development before, during and after the moment of synaptic partner choice. (A) The developmental timeline for synaptic specificity in the outcome includes pre-specification, the moment of synaptic partner choice, and post-specification. As described in the text, processes can overlap and cross-activate each other. (B) At the moment of choice, more or less promiscuous synapse formation as the limiting cases of total promiscuity (any two neurons can form a synapse if given the opportunity) and no promiscuity (some molecular or other specification mechanism ensures that exclusively the correct two partners can form a synapse). Both limiting cases are very unlikely based on experimental evidence, indicating that the level of promiscuity is a quantitative property at the moment of synapse formation.

Figure 2

Stepwise sieving out of incorrect partners as a conceptual model for the successive restriction of synaptic partnerships during development. (A) Pre-specification: development leading up to the moment of choice creates a local neighbourhood (encircled region) and selective adjacencies between neurons in this neighbourhood. Within this neighbourhood, all possible connections between five types of pre-synapses and five types of post-synapses (white, black, red, green, blue) are depicted. (B) Restriction at the moment of synaptic partner choice: at least three sieving mechanisms collaborate to restrict the choice: synaptic competency (represented by sieving out all connections between red/blue/green pre-synapses and black/white post-synapses), molecular recognition (represented by sieving out all connections between black/white pre-synapses and red/blue/green post-synapses) and interaction dynamics (represented by sieving out all connections where pre- and post-synapses have the same color). (C) Post-specification includes all developmental mechanisms following initial synapse formation that further restrict specificity, most prominently pruning (fine tuning; represented by sieving out multiples of any specific connection type). The output state of post-specification represents synaptic specificity in the outcome, i.e. all possible synaptic partnerships that were not actively prevented (sieved out) during preceding development.

Figure 3

Developmental consequences of genetically induced defects at different times in the developmental decision tree. (A) Genetic perturbation of a mechanism specific to the moment of choice (marked by semi-transparent red rectangle) may affect subsequent developmental processes minimally (blue line) or have a cascading effect to change all subsequent developmental processes (red lines). Assessment of the synaptic specificity phenotype in the outcome may not easily distinguish between these possibilities. (B) Genetic perturbation of a mechanism prior to synapse formation is likely to have a cascading effect on both the moment of choice as well as subsequent development. Assessment of the synaptic specificity defect in the outcome may reveal little about the primary defect at the beginning of the cascading developmental changes. (C) Perturbation of a single gene often causes multiple hits throughout the developmental decision tree, leading to compound cascading developmental changes that alter synaptic specificity in the outcome.

Figure 4

Outcomes of perturbation experiments of interacting cell adhesion molecules in synaptically connected neurons in the Drosophila visual system. In all examples, the molecular pairs are expressed specifically in neurons that form synapses in a distinct layer with subcellular specificity, yet loss of the inter-cellular interaction does not lead to a loss of synapses where they normally occur (blue regions) but instead to ectopic synapses in a nearby region (red regions; ectopic branches and ectopic synapses also marked in red). See text for possible mechanistic explanations. (A) Genetic perturbation of the Beat II–Side IV interaction or DIP-β causes synaptic specificity defects between the interneurons L2 and L4 in the Drosophila lamina. In both cases, synapses in the proximal layer of the lamina are still present, while ectopic synapses form in the distal lamina. (B) Genetic perturbation of the DIP-α–Dpr6/10 interaction does not lead to a loss of synapses in medulla layer M3 but to ectopic branching and synapse formation in M8 where Dm12 neurons do not normally branch or form synapses. (C) Genetic perturbation of the Side II–Beat VI interaction does not lead to a loss of synapses between T4d/T5d and LLPC3 neurons in the lobula plate layer 4 but to ectopic branching and synapse formation in the neighboring lobula plate layer 3.