Toll-like receptor 2 (TLR2) initiates potent immune responses by recognizing diacylated and triacylated lipopeptides. Its ligand specificity is controlled by whether it heterodimerizes with TLR1 or TLR6. We have determined the crystal structures of TLR2-TLR6-diacylated lipopeptide, TLR2-lipoteichoic acid, and TLR2-PE-DTPA complexes. PE-DTPA, 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N-diethylenetriaminepentaacetic acid, is a synthetic phospholipid derivative. Two major factors contribute to the ligand specificity of TLR2-TLR1 or TLR2-TLR6 heterodimers. First, the lipid channel of TLR6 is blocked by two phenylalanines. Simultaneous mutation of these phenylalanines made TLR2-TLR6 fully responsive not only to diacylated but also to triacylated lipopeptides. Second, the hydrophobic dimerization interface of TLR2-TLR6 is increased by 80%, which compensates for the lack of amide lipid interaction between the lipopeptide and TLR2-TLR6. The structures of the TLR2-lipoteichoic acid and the TLR2-PE-DTPA complexes demonstrate that a precise interaction pattern of the head group is essential for a robust immune response by TLR2 heterodimers.

Background: Cyclic di-AMP inactivates the potassium transport activity of KtrA. Results: Cyclic di-AMP binding to KrtA induced conformational changes. Conclusion: Cyclic di-AMP selectively binds to the KtrA RCK_C domain and signals the inactivation of potassium transport. Significance: The molecular basis for the role of cyclic di-AMP in potassium channel activity was investigated.

Building a sophisticated protein nano-assembly requires a method for linking protein components in a predictable and stable structure. Diabodies are engineered antibody fragments that are composed of two Fv domains connected by short peptide linkers. They are attractive candidates for mediators in assembling protein nano-structures because they can simultaneously bind to two different proteins and are rigid enough to be crystallized. However, comparison of previous crystal structures demonstrates that there is substantial structural diversity in the Fv interface region of diabodies and, therefore, reliable prediction of its structure is not trivial. Here, we present the crystal structures of ten mono- and bi-specific diabodies. We found that changing an arginine residue in the Fv interface to threonine greatly reduced the structural diversity of diabodies. We also found that one of the bispecific diabodies underwent an unexpected process of chain swapping yielding a non-functional monospecific diabody. In order to further reduce structural flexibility and prevent chain shuffling, we introduced disulfide bridges in the Fv interface regions. The disulfide-bridged diabodies have rigid and predictable structures and may have applications in crystallizing proteins, analyzing cryo-electron microscopic images and building protein nano-assemblies.

Generating artificial protein assemblies with complex shapes requires a method for connecting protein components with stable and predictable structures. Currently available methods for creating rigid protein assemblies rely on either complicated calculations or extensive trial and error. We describe a simple and efficient method for connecting two proteins via a fused alpha helix that is formed by joining two preexisting helices into a single extended helix. Because the end-to-end ligation of helices does not guarantee the formation of a continuous helix, we superimposed 1–2 turns of pairs of connecting helices by using a molecular graphics program. Then, we chose amino acids from the two natural sequences that would stabilize the connecting helix. This “shared helix method” is highly efficient. All the designed proteins that could be produced in Escherichia coli were readily crystallized and had the expected fusion structures. To prove the usefulness of this method, we produced two novel repeat proteins by assembling several copies of natural or artificial proteins with alpha helices at both termini. Their crystal structures demonstrated the successful assembly of the repeating units with the intended curved shapes. We propose that this method could dramatically expand the available repertoire of natural repeat proteins.