br Conclusions br Acknowledgement Work in the authors labs w

Conclusions
Acknowledgement Work in the authors’ labs was funded by Natural Sciences and Engineering Research Council of Canada (OGP0194652 and OGP0041653) and Diabète Québec.
Introduction Fifty years ago, Sydney Brenner proposed Caenorhabditis elegans as a useful model for the study of animal behavior, and how behavior is generated from its cells and synapses [1], [2]. Brenner and his technician, Nichol Thomson tested many nematode species for suitability in high resolution studies by transmission electron microscopy (TEM), since Brenner expected in a small simple specimen one could identify every cell in the body to produce a “parts list”, trace all 2'-O-Methyl-ATP and dendrites in serial thin sections, and perhaps list every synapse to generate a complete wiring diagram. C. elegans had other advantages, being cheap to raise, having pliable genetics for producing and crossing behavioral mutants, and an almost transparent body. Beginning from TEM surveys in the 1960s, it would take 2 decades for the first major papers to emerge, including circuitry for pharynx [3], ventral nerve cord [4] and then the whole hermaphrodite adult [5]. They also produced a partial circuit for adult male tail [6]. However, complete accounting of complexities of the male tail emerged from another decade of study of archival data in the Emmons and Hall labs at Albert Einstein College of Medicine [7]. The male tail was harder to trace in detail since its neurons were more highly branched than in the hermaphrodite. Brenner’s original premise has come to fruition, and a complete comparison of hermaphrodite and male adult nervous systems is close to publication. The hermaphrodite adult has 302 neurons and 138 muscles (depending on one’s definitions, cf. [8], [15]), connected by thousands of synapses. The male adult has almost 400 neurons and 150 muscles, with twice as many synapses. When all connections are accounted for, the prominence of electrical synapses is impressive, as the total “ weight” of synapses is equally divided between chemical and electrical contacts, where synapses are judged not by their numbers, but by their collective size in TEM images [7]. Gap junctions are numerically fewer, but are often physically larger. While many aspects of the wiring diagram utilize both types of connections in parallel, there is a clear preference for electrical signaling to coordinate activity within groups of muscles, within groups of sensory neurons (especially left/right pairs), and for chains of homologous motor neurons along the length of the body. In this review we concentrate on patterns of gap junction (GJ) usage, though the complete accounting is not published yet. Similar to what was learned by mathematical analysis of male tail circuits, the whole adult nervous system constitutes a “sparse network”, where neurons only make synaptic connections with some of their neighbors, the number of contacts per cell pair is variable, and there are few separate contacts per “edge” in graphical network parlance (cf. [7], [8], [9]). In the hermaphrodite, most neurons lack many branches, and generally find partners “en passant”. Neurons in the adult male tail are more branchy, and have access to more potential partners, yet connections remain sparse.
The connectome
Discussion − how does the connectome operate? When viewing wiring patterns across the whole connectome, one can try to identify cell groups involved in separate behaviors, as was done by White et al. [5], but it has often proven difficult to verify the exact behaviors produced within each group. Faumont et al. [18] detailed the organization of 4 layers of neurons for the bodywall motor circuit, where GJs play major roles in much the manner as described above within certain layers, and often in parallel to chemical synapses connecting the same sets of neurons. Chen et al. [36] analysed neurons based upon their positions along the long body axis, hoping to see economies in the total length of “wires” per cell as a measure of their connectivity. There was strong evidence for economies in the layout, as most cells seemed optimally placed. That effort helped to identify some neurons with cell bodies placed at extreme locales, which best served roles in pioneer axon outgrowth, rather than in any behavioral circuit. Varshney et al. [21] tried to fit all the neurons into layers to show steps in sequential processing, with sensory neurons feeding to interneurons, which then fed onto motor neurons and then to muscles. This exercise demonstrated that in the hermaphrodite, the distinct layers are shallow, with few processing steps, but still did not highlight particular behavioral circuits. Jarrell et al. [7] used a mathematical approach to group male tail cells into “communities” based upon their relative strengths of synaptic contact, fitting closely linked neurons and muscles into smaller subgroups, which could then be matched more clearly to individual steps in the male mating ritual. The male circuits also showed a shallow set of synaptic processing steps, with some “sensory neurons” producing direct NMJs onto sex muscles. The sparseness of the network, and small number of neuron classes within male tail circuits suggested that neurons could participate in several different steps in behavior, rather than being dedicated to just one action. Progress of a behavior might then depend on the relative level of activity in a few driver neurons that could tip the balance to switch the whole connectome’s output instantaneously from behavior #1 to #2, and then to #3, and so forth. Kawano et al. [24] showed experimental evidence for the role of GJs in this type of behavioral switch for the bodywall motor circuit. This switching feature is similar to what was predicted by Hopfield [37] for what has come to be known as a “Hopfield network”. The capacity of the network to store multiple different stable behavioral actions depends on the number of nodes (cells) and edges (synapses). Given the relatively small number of nodes and edges in the nematode connectome, one expects a modest set of different possible behaviors. As current work to detail the complete network reaches fruition, it will be interesting to test models for nematode behavior, and to determine where GJs provide unique features that are not easily accomplished by chemical signaling alone.