Potassium channels: a computer prediction of structure and selectivity

Abstract

Model structures for the pore of the potassium channels Shaker and ROMK1 are predicted. The models arise from computer simulations and suggest reasons for the striking selectivity of these channels for K+ and the blocking of ROMK1 by internal Mg2+. The modelled structure of the Shaker pore is supported by mutagenesis data. The mutagenesis experiments indicate the side chains responsible for binding to blocking agents [tetraethylammonium (TEA) and charybdotoxin (CTX)] and the model has these side chains suitably oriented for binding. An aromatic K+ binding site part way down the pore is also predicted by the Shaker pore model.

A new "hydrophobic template" method detects segments forming transmembrane α-helical bundles in ion channels

“…Thus, TM a-helix bundles and residues involved in channel functioning were proposed for M2 segments of AChR [5] and GABA A receptors [15], the a5 segment of d-endotoxin [6] and the in¯uenza virus M2 protein (VMT2) [16]. We also examined the H5 regions of three K + channels ± ROMK1 [17], Shaker [17], and Kv2.1 [18] ± which are believed to oligomerize in membrane-forming channels with b-barrel architecture [19] (although monomeric H5 segments are able to form a-helices in micelles [20]). Corresponding 1D-MHP plots are presented in Fig.…”

Shaker pore structure as predicted by annealed atomic simulation using symmetry and novel geometric restraints

“…A consequence of the distance and side-chain orientation restraints imposed on Y445 (Table 2) is the formation of a closely packed tyrosine quartet (see Fig. 5 A), a configuration that had also been proposed before (Lee, 1992;Bradley and Richards, 1993). Interestingly, as can be seen from Fig.…”

“…However, this advantage may not apply to K+ or the closely related Na+ and Ca2+ channels because increasing evidence points to an irregular loop structure for their ionconducting pore lining, thus annotating a structural notion to the term "pore loops" (MacKinnon, 1995). Consequently, structural models of these channels (Bogusz et al, 1992;Durell and Guy, 1992;Bradley and Richards, 1993;Lipkind and Fozzard, 1994;Lipkind et al, 1995;Sansom and Kerr, 1995;Guy and Durell, 1995) have been largely the products of extensive maneuvers of interactive molecular graphics mingled with the modeler's discretion and/or presupposed secondary structures.…”

The pore-lining region of shaker voltage-gated potassium channels: comparison of beta-barrel and alpha-helix bundle models

“…It is noteworthy that these models, which were generated using restraints derived from only a handful of experimental studies, identified a plausible /3-barrel model that would correspond to our S8-839-LA ensemble (Bogusz et al, 1992), a configuration that we have shown still possesses a high degree of agreement with the more extensive mutagenesis data. Additional eight-stranded X3-barrels have been constructed by homology modeling to the S = 8 /3-barrel of superoxide dismutase (Bradley and Richards, 1994). However, from the published data it is not possible to determine precisely the thread of this model, although the pattern of pore-lining residues indicates that it is equivalent to our 9-2-9 or 9-4-7 threads.…”

“…Both long-pore (Bogusz et al, 1992;Jin and Weaver, 1993;Bradley and Richards, 1994;Lipkind et al, 1995) and short-pore (Durell and Guy, 1996) models for the H5 region have been proposed. The central H5 region of these has been modeled variously as an antiparallel (3-hairpin (Bogusz et al, 1992;Bradley and Richards, 1994), an a-helix-turn-random coil conformation (Durell and Guy, 1996), and a (3-strand-turn-random coil conformation (Lipkind et al, 1995). Although all of these models have their merits, it is evident that a somewhat more wide-ranging and systematic comparison of possible models is called for.…”