To capitalize on these initial advances, new efforts must continue to focus on three primary tasks: forcing new interactions between signaling proteins, creating new mechanisms of regulation, and controlling the input-output processing behavior of the assembled circuit. Additional layers of regulation could be achieved by using targeting signals to control subcellular localization, oligomerization, and proteolysis (Devit et al., 2005
; Grilly et al., 2007
; Corson et al., 2008
; Haruki et al., 2008
). Synthetic derivatives of interaction motifs such as leucine zippers could help minimize unwanted interactions with endogenous proteins and could also promote standardization by building collections of interacting parts with pre-determined specificity and affinity (Acharya et al., 2002
; Bashor et al., 2008
; Bromley et al., 2008
). Further control over signal detection and processing can be achieved via multicellular networks that propagate responses from cell to cell (Basu et al., 2004
; Andrianantoandro et al., 2006
One avenue warranting further exploration is whether binding interactions may be generally amenable to synthetically-imposed regulation through phosphorylation, by analogy to the disruptive effects of PKA phosphorylation on PDZ-peptide binding () (Yeh et al., 2007
). To increase the variety of usable kinases, it may suffice to place the phosphorylated sites adjacent to (rather than within) the target peptide; then, electrostatic effects of phosphorylation could either inhibit or promote binding, depending on whether the peptide-binding partner motif is flanked by electronegative or electropositive surfaces, respectively. Conceivably, varying the number of phosphorylation sites and their distance from the target peptide could adjust the strength of the effect and the sensitivity to kinase concentration (Serber and Ferrell, 2007
). In principle these strategies could be applied to many protein-protein interaction pairs.
Eventually, the ability to predict pathway behavior will benefit from computational modeling and the use of pre-characterized circuit motifs (Papin et al., 2005
; Brandman and Meyer, 2008
). Nevertheless, substantial advances have already been accomplished through adventurous experimentation, and these studies also highlight how systematic, trial-and-error strategies can identify parameters that are critical yet unpredictable. For instance, only 5 of 34 artificial N-WASP chimeras showed the desired “AND gate” behavior in which activation required two simultaneous inputs (Dueber et al., 2003
); while the majority showed some form of regulation (and some interesting surprises), the desired behavior required rather subtle variations (e.g., in module geometry, domain affinity, and linker lengths) that seem unlikely to be predictable by computational approaches anytime soon. A related issue emerges from recent studies in which unexpected features such as the subcellular location where signaling is initiated (i.e., cytoplasm, plasma membrane, or internal membranes) were found to have a strong influence on whether the input-output response behavior is graded or switch-like (Inder et al., 2008
; Takahashi and Pryciak, 2008
). Thus, despite our deep understanding of some pathways, unanticipated subtleties can have dramatic effects on the overall system behavior. Ideally, theoretical and empirical approaches will work together to help eliminate these lingering blind spots, some of which may actually become exposed as a byproduct of synthetic research.
It seems a given that the next decade will witness increasingly sophisticated examples of custom-designed signaling pathways. As the successful strategies are refined and their applications are expanded, it will be fascinating to watch whether these efforts coalesce into a unified discipline. Will cellular engineers be able to develop new devices as routinely and rigorously as their mechanical or electronic counterparts, or is biology inherently too messy and unpredictable? Time will tell.