The design of functional membrane proteins from first principles represents a grand challenge in chemistry and structural biology. Here, we report the design of a membrane-spanning, four-helical bundle that transports first-row transition metal ions Zn 2+ and Co 2+ , but not Ca 2+ , across membranes. The conduction path was designed to contain two di-metal binding sites that bind with negative cooperativity. X-ray crystallography and solid-state and solution nuclear magnetic resonance indicate that the overall helical bundle is formed from two tightly interacting pairs of helices, which form individual domains that interact weakly along a more dynamic interface. Vesicle flux experiments show that as Zn 2+ ions diffuse down their concentration gradients, protons are antiported. These experiments illustrate the feasibility of designing membrane proteins with predefined structural and dynamic properties.
The transmembrane protein 16 (TMEM16) family of membrane proteins includes both lipid scramblases and ion channels involved in olfaction, nociception, and blood coagulation. The crystal structure of the fungal Nectria haematococca TMEM16 (nhTMEM16) scramblase suggested a putative mechanism of lipid transport, whereby polar and charged lipid headgroups move through the low-dielectric environment of the membrane by traversing a hydrophilic groove on the membrane-spanning surface of the protein. Here, we use computational methods to explore the membrane–protein interactions involved in lipid scrambling. Fast, continuum membrane-bending calculations reveal a global pattern of charged and hydrophobic surface residues that bends the membrane in a large-amplitude sinusoidal wave, resulting in bilayer thinning across the hydrophilic groove. Atomic simulations uncover two lipid headgroup-interaction sites flanking the groove. The cytoplasmic site nucleates headgroup–dipole stacking interactions that form a chain of lipid molecules that penetrate into the groove. In two instances, a cytoplasmic lipid interdigitates into this chain, crosses the bilayer, and enters the extracellular leaflet, and the reverse process happens twice as well. Continuum membrane-bending analysis carried out on homology models of mammalian homologs shows that these family members also bend the membrane—even those that lack scramblase activity. Sequence alignments show that the lipid-interaction sites are conserved in many family members but less so in those with reduced scrambling ability. Our analysis provides insight into how large-scale membrane bending and protein chemistry facilitate lipid permeation in the TMEM16 family, and we hypothesize that membrane interactions also affect ion permeation.
8 June 2014 in Nature Structural & Molecular Biology
Deep learning has advanced the design of static protein structures, but the controlled conformational changes that are hallmarks of natural signaling proteins have remained inaccessible to de novo design. Here, we describe a general deep learning–guided approach for de novo design of dynamic changes between intradomain geometries of proteins, similar to switch mechanisms prevalent in nature, with atomic-level precision. We solve four structures that validate the designed conformations, demonstrate modulation of the conformational landscape by orthosteric ligands and allosteric mutations, and show that physics-based simulations are in agreement with deep-learning predictions and experimental data. Our approach demonstrates that new modes of motion can now be realized through de novo design and provides a framework for constructing biology-inspired, tunable, and controllable protein signaling behavior de novo.