Pore loops: An emerging theme in ion channel structure

Recent studies using mutagenesis to alter ion channel function have shown that the selectivity filter in many ion channels is formed by pore loops, relatively short polypep-tide segments that extend into an aqueous pore from one side of the membrane. Why do ion channels use pore loops to form their selective ion-binding sites? The purpose of this perspective is to address this question in the context of other well-characterized proteins. To begin, Figure 1 summarizes the ion channels that are thought to contain pore loops. The voltage-gated K ÷, Na ÷, and Ca 2+ channels and cation channels gated by in-tracellular cyclic nucleotides contain pore loops formed by the linker connecting the fifth and sixth membrane-spanning regions of each subunit or domain (Figure 1A) In the case of homotetra-meric K ÷ channels, four identical loops, one from each subunit, extend into the pore. The same is probably true for the cyclic nucleotide-gated channels, but their subunit stoichiometry has not been established. In voltage-gated Na ÷ and Ca 2÷ channels, each of the four homologous domains contributes a loop to the ion conduction pore: loop residues determine whether the channel conducts Na ÷ or Ca 2÷ (Heinemann et al., 1992). Pore loops are not limited to those ion channels with membrane topologies described in Figure 1A. Inward recti-fier K ÷ channels (Ho et al., 1993; Kubo et al., 1993a, 1993b) and ATP-gated cation channels (Brake et al., 1994; Valera et al., 1994) have not been studied as extensively, but they probably also have a pore loop (homologous to that in voltage-gated K ÷ channels) in the setting of a much simpler overall transmembrane topology (Figure 1 B). Very recently, studies from several laboratories have completely revised our view of the transmembrane topology of glutamate receptor ion channels, and in so doing have added to the growing list of ion channels thought to contain In the new model for glutamate receptor ion channels, a region known to be important for ion conduction (and previously thought to be a membrane-spanning segment) very nicely fits the description of a pore loop. In the absence of direct structural information, what do we know about the structure of any pore loop? Mutagene-sis studies on a Shaker voltage-gated K ÷ channel have shed light on some overall structural features of its pore Rather than extending completely across the lipid bilayer, a large portion of the pore loop …