Predicting channel structures with genetics

No one doubts the power of genetics. Nevertheless, the use of genetic techniques in structural research — specifically, for predicting the exact three-dimensional structures of proteins — is not so widespread. Now, Minor, Masseling, Jan and Jan (Cell 96: 879- 891) have verified that genetic approaches can indeed contribute handsomely to structural knowledge, notably in the often intractable setting of the membrane. They predict the transmembrane structure of an inwardly rectifying potassium channel (K_ir), a member of a superfamily of eukaryotic channels that are responsible for regulating many cellular functions (for example, cell excitability, vascular tone, heart rate, renal salt flow, and insulin release) by modulating membrane potential. The results of Minor and colleagues indicate that the K_ir channel structure is distinct from that of the only known X-ray crystal structure of a potassium channel (KcsA) — despite contrary expectations, which arise from the fact that these channels, at first glance, have the same transmembrane topology.

K_ir channels are made up of four subunits, each with two transmembrane domains, M1 and M2. To address how these domains are arranged in the active channel, Minor and colleagues constructed libraries of K_ir proteins with randomly mutagenized M1 or M2 domains. From these libraries, they selected functional K_ir channels that could rescue a potassium transport-deficient strain of yeast (left image). Significantly, positive clones from these selection experiments displayed wild type electrophysiological properties, indicating that the wild-type channel structure was present. The results of these experiments show that, like other transmembrane domains, M1 and M2 are extremely tolerant to sequence changes, but only at specific positions, and preferences are observed with respect to side chain shape and chemistry.

The patterns of allowed amino acid substitutions suggest that both M1 and M2 are helices and hint at the protein–protein, protein–lipid, and protein–water interaction surfaces. The specific arrangement of transmembrane helices in the K_ir channel was determined by second-site suppressor experiments. In these studies, mutations that result in nonfunctional K_ir channels were made at conserved positions in one transmembrane domain, and suppressor mutations that restored channel activity were identified from libraries containing randomized mutations in the other domain. Together, the results suggest that each M1 helix interacts with two M2 helices, thus yielding a helical bundle model for the K_ir transmembrane channel structure, with M1 and M2 forming the outer and inner helices, respectively (right image). This arrangement appears to be shared by all K_ir family members, based on sequence comparisons. Such a clear result indicates that genetic approaches should be added to every structural biologist's list of tools.