In This Issue

Proteins residing in biological membranes are among the most attractive targets of biomedical research, but their native forms are notoriously resistant to structural studies, as most of the traditional techniques cannot be applied in the natural lipid environment. In this review, we summarize the latest advances in solid-state Nuclear Magnetic Resonance (NMR) spectroscopy, the emerging powerful technique for the investigation of structure and dynamics of membrane proteins. While solid-state NMR has been successfully applied to purified membrane proteins reconstituted into lipid bilayers for many years, the most recent developments extended its reach to native biological cell membranes, giving it an unprecedented advantage in studying this important category of proteins.

In order to represent variations and/or uncertainties in structure, some protein structure determination and prediction approaches yield a set of models (i.e. an ensemble) rather than a single model. When comparing such sets of models, each ensemble is typically replaced with a single representative model even though this results in a loss of information. The ENSEMBLATOR introduced here is a novel strategy that carries out direct comparisons of protein ensembles to sensitively identify systematic differences between them on both a global and a local level. The locally-overlaid-dipeptide-residual (LODR) is introduced as a novel measure of local conformational similarity.

As a precursor to a range of biologically active compounds such as nitric oxide, arginine is involved in processes as diverse as neural signaling and wound repair. However, imaging arginine in tissue and in real time is a major challenge. Whitfield et al., utilize ancestral protein reconstruction (APR) in this issue to create a robust and sensitive fluorescent reporter for arginine, allowing for measurement of arginine concentrations in intact brain tissue. This work illustrates the power of APR as a tool for creating stable protein scaffolds for protein engineering.

As sequence databases continue to expand exponentially, automated methods to identify protein function are essential. One such method involves creating similarity networks—networks in which the proteins are related by numerical edge metrics. Functionally related clusters are identified as subnetworks within the larger network. A significant assumption underlying such approaches is that the edge metric used to create the network correlates with function. This assumption is evaluated. Similarity networks created using full sequence, structure and active site similarity as edge metrics were evaluated at different edge thresholds; the resulting subnetwork hierarchies were compared to highly curated protein function hierarchies. In some superfamilies, different edge metrics identify similar subnetworks; in other cases, the subnetwork topology differs. Detailed analysis of the differences clarifies what can be learned from each type of network.