Chemical Benzethonium Chloride cross-linking of AG22 S16D/S411D and 14-3-3 clearly showed that these two proteins are able to interact directly, albeit weakly. We also used small angle X-ray scattering from the cross-linked complex to generate a model of the solution structure of the AG22:14-3-32 complex. Although many crystal structures of 14-33 bound to peptide ligands have been determined, there is only one crystal structure of 14-3-3 bound to a target protein, that of serotonin N-acetyltransferase. Thus, our structure of AG22:14-3-32 provides new information on the way in which 14-3-3 can interact with partner proteins, in this case showing that a weak or transient interaction occurs when one rather than two of the 14-3-32 peptide binding sites is occupied. This weak interaction was captured by chemical cross-linking. Why is the AG22 complex with 14-3-3 relatively weak? One possibility is that a weak or transient interaction of these two proteins is required in the context of cellular function. Analysis of phosphorylation effects on protein-protein interactions in the human proteome, showed that phosphorylation sites are often located on binding interfaces in weak or transient complexes and for the majority of complexes, phosphorylation was predicted to have minimal effect on stability and binding affinity. An example of another weak interaction involving 14-3-3 is the interaction with AS160. AS160 is also an Akt substrate, a RabGAP that mediates insulin-stimulated GLUT4 translocation. Phosphorylation of AS160 by PKB/Akt leads to 14-3-3 binding. However the binding affinity of AS160 to 14-3-3 is weak and the complex dissociates during immunoprecipitation wash steps. The weak/transient interaction between AS160 and 14-3-3 was also stabilized by chemical cross-linking allowing detection by coimmunoprecipitation from cell lysates. A weak or transient interaction between AG22 and 14-3-3 may thus be a characteristic of the complex in vivo, or perhaps additional partner proteins may be required to reinforce the interaction. Alternatively, a strong interaction may be necessary in vivo, but may not be generated under the experimental conditions we used. For example, if the coiled coil domain is important for the 14-3-3 interaction, then removing it to produce the soluble, stable form of the AG22 protein that we used might reduce 14-3-3 binding affinity. Similarly, the aspartate Butenafine hydrochloride mutants we used to mimic phosphorylated AG22 may not be optimal mimics of phosphorylation in this system. However, the fact that native AG22 can also be cross-linked to 14-3-3 suggests that AG22 phosphorylation is not an essential component of the 14-3-3 interaction. Indeed, there are precedents for phosphorylation-independent binding of 14-3-3 to target proteins. We found that the RhoGAP domain of AG22 was not sufficient to interact with 14-3-3. This conclusion is supported by recent findings that binding of 14-3-3 to the pSer411 binding site is dependent on binding to pSer16 preceding the PH domain. Although AG22 has been identified as an important binding partner of Rac1, we were unable to detect an interaction between the two, even with cross-linking. This interaction may therefore require the C-terminal coiled-coil domain of AG22, because full-length AG22 was used in the in vivo binding studies. As described above, we were unable to use recombinant fulllength AG22 due to rapid degradation. It is also possible that additional post-translational modifications or regulatory proteins are required for binding of AG22 to Rac1; these would be present in vivo but not in our in vitro experiments. It is noteworthy that the intrinsic GTPase activity of FilGAP, a related GAP protein which like AG22 has PH, GAP and coiled-coil domains, requires phosphorylation by ROCK.