In the tree of life, sponges (Porifera), generally recognized as the oldest surviving metazoan phyletic lineage (Fig. 1.1B), occupy a highly informative position for understanding the evolution of features that uniquely characterize animals (Srivastava et al., 2010). The synapse, a cellular machine formed through the dynamic assembly of multiple proteins that together perform a specific biological function, is one such metazoan specialization. The synaptic machinery delivers a chemical signal via vesicle fusion at the presynaptic neuronal membrane to postsynaptic receptors, which convert that signal back to an electrical impulse in the postsynaptic neuronal cell. Surprisingly, the genome of the Poriferan demosponge, Amphimedon queenslandica, contains an almost complete set of genes homologous to those found in mammalian synapses (Fig. 1.1A), although the organism does not assemble any structure morphologically resembling a synapse (Sakarya et al., 2007; Srivastava et al., 2010). Although limited gene innovation and the invention of new protein interaction sites can partially explain how preexisting genes came together to form the synaptic complex (Sakarya et al., 2010), the multiple evolutionary steps involved in building a cellular machine through the assembly of an interaction network that can operate as a unit with a discrete biological function remains unknown.
Changes in conserved transcriptional programs arising from modification of instructions encoded in the genome have contributed to our understanding of animal evolution (Barabási and Oltvai, 2004; Oldham et al., 2006, 2008; Brawand et al., 2011). Specific patterns of expression can define discrete tissues, cell types, and even functional protein complexes. Genes with similar expression patterns often have similar function (Eisen et al., 1998). Furthermore, when comparing orthologues across divergent species, highly conserved coexpression is a strong predictor of shared function in similar pathways (Quackenbush, 2003; Stuart et al., 2003; van Noort et al., 2003). These results suggest that functionally related genes might be under similar expression constraints (Carlson et al., 2006). Thus, changes in coexpression relationships for any group of genes may contain information on the assembly and evolution of cellular machines. To understand the evolutionary transition leading to the emergence of a functional synapse, we used network analysis to identify unique patterns of synaptic gene coexpression in representative species from diverse phylogenetic positions. We show that “protosynaptic” genes have an inherent modular structure and that the coregulatory links between these modules characterize species with functional synapses. In contrast, ancient eukaryotic cellular machines, such as the proteasome and nuclear pore, already operate in early metazoans, and their associated genes display highly correlated expression patterns over development. These findings