Computational design of membrane proteins

Abstract

Membrane proteins are involved in a wide variety of cellular processes, and are typically part of the first interaction a cell has with extracellular molecules. As a result, these proteins comprise a majority of known drug targets. Membrane proteins are among the most difficult proteins to obtain and characterize, and a structure-based understanding of their properties can be difficult to elucidate. Notwithstanding, the design of membrane proteins can provide stringent tests of our understanding of these crucial biological systems, as well as introduce novel or targeted functionalities. Computational design methods have been particularly helpful in addressing these issues, and this review discusses recent studies that tailor membrane proteins to display specific structures or functions and examines how redesigned membrane proteins are being used to facilitate structural and functional studies.

Figure 1. Pentameric Ligand-Gated Ion Channel Chimera

Rendering of the chimera membrane protein structure based on the structure of GLIC (pdb accession code: 3EHZ). The extracellular domain (yellow) is from the prokaryotic proton-gated ion channel GLIC and the transmembrane domain (blue) is from eukaryotic anionic-selective α1 glycine receptor (Duret et al., 2011). Small modifications at the interface of the two domains are colored in magenta and orange. For clarity, the other subunits are colored gray.

Figure 2. Structure of the Redesigned Glycophorin A in Complex with the Cofactor Protoporphyrin IX

The designed bis-histidine binding site is depicted together with the protoporphyrin IX ligand (Cordova et al., 2007). The modified positions in the structure of glycophorin A are colored in blue.

Figure 3. Topology of the De Novo Designed Membrane Protein PRIME

The de novo designed membrane protein PRIME is depicted with two nonbiological iron diphenylporphyrin (Fe^IIIDPP) cofactors (in blue) (Korendovych et al., 2010). The cofactor binding site is displayed in more detail showing the axial interaction of the histidine residue and the iron metal. The second-shell hydrogen bond with threonine residue is also indicated.

Figure 4. CHAMP Transmembrane Peptides

Structural models of the CHAMP transmembrane peptides (blue) designed to bind (yellow) α_IIb (right panel) and (yellow) α_v (left panel) integrins with high specificity–both transmembrane motifs naturally bind integrin β₃ (Yin et al., 2007). The GxxxG motif is highlighted with space-filling representations in both cases.

Figure 5. Transmembrane Portion of the Bacterial Potassium Ion Channel KcsA and its Water-soluble Variant

Comparison of the structure of the bacterial potassium ion channel KcsA (yellow) (pdb accession code: 1K4C) (Zhou et al., 2001) and its water-soluble variant (blue) (pdb accession code: 2K1E) (Ma et al., 2008). The water-soluble variant was expressed in E. Coli and contains 29 computationally designed exterior mutations in each of the four 104-residue subunits. Depicted as orange spheres in the water-soluble structure (blue), are the Cα atoms for the exterior positions that were computationally designed (Slovic et al., 2004). In the right image all four subunits are depicted while in the left image (side view) only two subunits are rendered.

Figure 6. Cryo-EM Structure of a Transmembrane Domain from the Nicotinic Acetylcholine Receptor, NMR Structure of its Water-Soluble Analog (WSA) and X-ray structure of the Prokaryotic Homolog GLIC

Comparison of the 4-Å-resolution cryo-EM structure of the transmembrane domain of the α1 subunit from the nicotinic acetylcholine receptor (gray) (pdb accession code: 1OED) (Miyazawa et al., 2003), the NMR structure of a water-soluble analog from the same segment (blue) (pdb accession code: 2LKG) (Cui et al., 2012), and transmembrane domain of the prokaryotic homolog GLIC (yellow) (pdb accession code: 3EAM) (Bocquet et al., 2009). The water-soluble variant was expressed in E. Coli and contains 23 computationally designed exterior mutations (Cα atoms of these positions are depicted as orange spheres). To link the TM4 helix with the rest of the bundle, a polyglycine linker was inserted (magenta). Based on photoaffinity labeling studies, V31 (colored in green) was identified as potential binding site in WSA for general anesthetics (azi-propofol and azi-isoflorane). For comparison, the residues forming the anesthetic binding site in the co-crystal structure of GLIC (Nury et al., 2011) are shown in sticks representations.