You have full access to this article via Jozef Stefan Institute.

Synapses are highly organized complex molecular machines which facilitate the rapid and precise spread of information between cells. These structures are widespread in extant lineages of animals and are part of the biological foundation for the traits that make animals unique among life on earth, such as complex sensory perception, complex behaviours and cognition. Although synapses are highly specialized structures, the majority of the components that represent the synaptic toolkit -- those involved in vesicle exocytosis (for example, SNAREs, Unc18, Unc13, complexin, synaptotagmins and Rab3 interacting molecule (RIM)), receptors and ion channels (for example, G-protein-coupled receptors (GPCRs), ionotropic glutamate receptors (iGluRs) and voltage-gated calcium channels (Ca)), adhesion proteins (for example, cadherins, neurexins and neuroligins) and postsynaptic scaffolding proteins (for example, Shank, discs large (DLG) and Homer) (Fig. 1a) -- predate the emergence of true neurons and even the origin of animals. The ancient origin of these major parts raises the question of how the first true synapses evolved. Did they emerge gradually, through the stepwise addition and integration of components over evolutionary time, or were pre-assembled functional units -- originally serving other roles -- co-opted and integrated, eventually giving rise to a specialized function? The bulk of animal diversity is found in bilaterians, and although there has been a great deal of modification and specialization, it is well established that the last common ancestor of bilaterians had a complex nervous system with defined chemical synapses. For this reason, we centre our Review on general components of bilaterian chemical synapses. However, electrical synapses also play an important role in nervous systems and are likely to have had a strong influence on their early evolution. Electrical synapses are built around gap junctions, composed of innexins (in invertebrates) or connexins (in vertebrates), which electrically couple discrete cells. Although gap junctions are found and function in various cell types, innexins are intriguingly found only in animals with nervous systems. Apart from forming gap junctions, innexins can form ion channels, based on N-glycosylation, making it difficult to infer function from gene presence.

Although there is ambiguity in the term 'synapse', at its core a true synapse forms a bridge between two cells (or two parts of the same cell in the case of autapses) to facilitate the rapid transmission of information. Although both presynaptic and postsynaptic components can be found in various cell types in animals, synapses are most often associated with neurons. Further linking synapses to neurons, synaptic structures have not been described in the major aneural animal lineages of animals (sponges or placozoans), suggesting a link between synaptic origins and the emergence of nervous systems. Neurons are found broadly, but not entirely, across extant lineages of animals and are potentially fundamental animal cell types (cell types that were present in the last common ancestor of all of today's animals). However, the homology of all nervous systems remains contentious owing to the widely varying morphologies and large evolutionary time separating the crown groups of early branching animal lineages. In order to address these questions and build hypotheses about the historical events, we must look broadly at how these proteins function in, and their interactions with, the lineages that bracket the origin of true synapses, as well as animal multicellularity. In this Review, we aim to reconcile the current state of knowledge on the evolutionary trajectories of various 'synaptic' proteins to raise hypotheses about when and how these became integrated into a functional system and elucidate the foundation of the nervous system.

There are four animal lineages that diverged before the advent of bilateral symmetry, during the time when specialized cell types such as neurons first arose (Fig. 1b). These non-bilaterian lineages are composed of ctenophores, sponges, placozoans and cnidarians. Although each lineage contains diversity, ctenophores are generally free-living gelatinous zooplankton, which primarily move through the beating of combs composed of rows of large, specialized cilia. Sponges, found throughout the oceans and in freshwater environments, are generally sessile filter feeders, which generate flow through the beating of flagella on specialized feeding cells. Placozoans are small, free-living marine animals that move through ciliary beating and show low levels of morphological complexity. Cnidarians are the most abundant and morphologically diverse lineage of non-bilaterian animals, composed of jellyfish, corals, sea anemones and hydrozoans as well as marine parasites.

It remains unclear whether the lineages leading to modern sponges or ctenophores branched first, but there is stronger consensus that this was followed by placozoans and then cnidarians. With respect to synapses and neurons, these are present in ctenophores and cnidarians, whereas both sponges and placozoans lack them. This could be explained by independent or convergent evolution of synapses and/or neurons in ctenophores and cnidarians, or a single origin before the divergence of ctenophores, followed by secondary loss in sponges and placozoans. Animals are a member of the holozoan clade, which also includes ichthyosporeans, filastereans and choanoflagellates. These are generally considered unicellular, although there are many instances of simple multicellularity in each of these lineages (Box 1). As genetic information from representatives of these lineages builds, the list of 'animal-specific' genes dwindles, which also holds true for those of the synaptic toolkit. This further complicates the question of when these structures first arose, as everything necessary to build a minimal chemical synapse was present before the first animal.

At its core, the synaptic toolkit assembles into distinct functional units, making up the presynaptic zone, where the signal is released, and the postsynaptic zone, where the signal is received (Fig. 1a). The presynapse is built around discrete modules involved in regulated secretion, vesicle packaging and recycling, as well as processes that govern cell excitability and ion clearance. On the postsynaptic side, distinct modules are built around scaffolding proteins that cluster receptors and coordinate signalling molecules necessary for processing the incoming signal. Crucially, the synaptic cleft, the space between the presynaptic and postsynaptic cells, is regulated by cell adhesion machinery, which determines its size and positioning, directly influencing synaptic function.

It is important to note that although this classification may be helpful for identifying the assembly of the synaptic machinery across large evolutionary distances, it represents an artificial boundary placed on dynamically intertwined pieces that have no a priori membership in a specific 'class'. However, this can still serve as the framework for defining a minimal chemical synapse in order to identify the essential evolutionary steps from specialization events that have given rise to the diversity of synapses we see today.

Considering the various components involved in chemical synaptic function, we can define a minimal chemical synapse as one that includes polarized trafficking of secretory vesicles, SNARE-mediated regulated secretion, electrical excitability, stable adhesion between the presynaptic and postsynaptic structures, and clustering of receptors and signalling molecules in the postsynaptic cell (Fig. 1a). These elements form the basis for defining true chemical synapses, although the first synapses in evolution may have been more complex. However, in extant lineages, none of these modules has functions that are restricted to synapses. These parts can be found elsewhere -- what makes a chemical synapse unique is the localized combination of them all. By understanding the modular components of synaptic function, we can begin to trace the evolutionary steps that led to the synapses we observe today and identify which components were co-opted or gradually assembled over time.

So how are we able to tease out the sequence of events that led to the occurrence of the first true synapse? The fossil record does not preserve structures as fine as synapses, or contain molecular detail, so we must make comparisons across extant lineages (Fig. 1c). Our current perspective builds on earlier frameworks such as the chemical brain hypothesis and the neurosecretory network hypothesis, which underscored the ancient roles of secretory and synaptic proteins. One potential scenario consists of localized regulated secretion being established in a polarized cell before the first animal arose. This intrinsically implies not just a mechanism for spatiotemporal control of exocytosis but also an upstream trigger and method of propagation of the signal to the release site. In a maximal integrated system, this would rely on other synaptic modules, that is, clustered receptors localized by scaffolding proteins such as DLG, Homer and Shank as seen in glutamatergic postsynapses, or such as gephyrin and collybistin in GABAergic postsynapses, and action potential-based propagation of a receptor site to a release site. In a minimal sense, these would be independent mechanisms and modules. In this case, one functional module such as regulated secretion, functioning in cell-cell communication, would serve as the basis or open evolutionary space for the assembly of the other components.

Although multicellularity is a somewhat common occurrence on earth, complex multicellularity has occurred as few as five times (Box 1). With simple multicellularity, although colonies exploit different niches and produce collective behaviours, the component cells often retain individual identity and do not differentiate into specialized cell types with distinct functions. By contrast, complex multicellularity, as seen in animals, involves highly specialized cells that work together to form tissues and organs. Complex multicellularity is characterized by its organization into distinct levels of structure, from individual cells to tissues, organs and whole organisms. This hierarchical organization opens the space for more refined deployment of cellular modules and specialization of function.

There are examples of simple multicellularity in each lineage of the closest relatives of animals; ichthyosporeans, filastereans and choanoflagellates. These include coenocytic development (nuclear division without cytokinesis, followed by cellularization), which is common in ichthyosporeans; aggregative colonies, which are common in filastereans but also seen in choanoflagellates; and colonies established through clonal division, which are common in choanoflagellates but also seen in ichthyosporeans (Fig. 2a-c). Apart from these transitions, the organisms can show complex transitions between cell states, such as from a flagellated swimming state to an amoeboid one. There is also some evidence for potential heterogeneity within colonies, suggesting differentiation of cell types. This suggests that the common ancestor of holozoans had a complex life cycle of temporal differentiation, potentially more so than the general characters seen in the individual extant lineages today. From this standpoint, the major innovation that occurred in animals is a shift from temporal to spatial differentiation within a clonal form (Box 1). This is essential to our understanding of the first neurons and specialized structures such as synapses, because it gives insight into the selective pressures as well as the novel niches these types of development allow the organism to fill. Of the synaptic modules discussed above, some were assembled into similar functional units prior to this event, although the purpose they served was tuned to that given organism's life history. Stable or obligate multicellularity opens the space for deployment and development of specialized structures, such as synapses, but at the same time these structures open new space by providing mechanisms to overcome limitations such as those of communication and size seen in simple multicellularity.