Portal de Periódicos da CAPES

Dynamic Selectivity Filters in Ion Channels

1999; Cell Press; Volume: 23; Issue: 4 Linguagem: Inglês

10.1016/s0896-6273(01)80025-8

1097-4199

Baljit S. Khakh, Henry A. Lester,

Neuroscience and Neuropharmacology Research

Membrane ion channels contain integral pores that precisely select their permeant ions. This selectivity anchors our concepts of transmembrane signaling in many tissues, including the nervous system. Excitable cells have a rich repertoire of dynamically regulated channels, and this richness and plasticity allows them to use channels beautifully to suit their needs. The selectivity of ion channel pores has generally been viewed as fixed. However, recent studies on disparate classes of ion channels challenge the generality of this idea and show that some ion channels can change their ion selectivity such that normally impermeant ions do in fact permeate under some circumstances. In no case is the mechanism fully understood, but the phenomenon represents both a new functional aspect of ion channels and a suggestion about novel ways in which channels may process information in the nervous system. This review seeks to highlight studies on ion channels that show selectivity changes, point to possible mechanisms, and draw on common themes. We consider P2X and proton-gated channels from the superfamily of transmitter-gated ion channels, Kv and cardiac sodium channels from the superfamily of voltage-gated ion channels, and cyclic nucleotide–gated channels from the family of channels that are gated by intracellular messengers (Figure 1).Figure 1Channels that Show Ion Selectivity ChangesShow full caption(A) P2X channels.(B) Proton-gated channels.(C) Cardiac sodium channels.(D) Kv channels.Upper panels show representations of the membrane disposition and key structural features of the various ion channels (A–D). The lower panels show traces that illustrate ion selectivity changes for ion channels. The traces have been reproduced from original papers ([A]Khakh et al. 1999; [B]Lingueglia et al. 1997; [C]Nuss et al. 1999; [D]Kiss et al. 1999). Acid-sensing channels and P2X channels have the same overall membrane topology but are unrelated to each other at the level of amino acid sequence. CNG channels also change their ion selectivity (Hackos and Korenbrot 1999) and are structurally similar to voltage-gated ion channels (C and D).View Large Image Figure ViewerDownload Hi-res image Download (PPT) (A) P2X channels. (B) Proton-gated channels. (C) Cardiac sodium channels. (D) Kv channels. Upper panels show representations of the membrane disposition and key structural features of the various ion channels (A–D). The lower panels show traces that illustrate ion selectivity changes for ion channels. The traces have been reproduced from original papers ([A]Khakh et al. 1999; [B]Lingueglia et al. 1997; [C]Nuss et al. 1999; [D]Kiss et al. 1999). Acid-sensing channels and P2X channels have the same overall membrane topology but are unrelated to each other at the level of amino acid sequence. CNG channels also change their ion selectivity (Hackos and Korenbrot 1999) and are structurally similar to voltage-gated ion channels (C and D). In 1979, Cockcroft and Gomperts showed that mast cells express a receptor for extracellular ATP (termed the P2Z receptor) that when activated allowed a time-dependent membrane permeability to large molecules (4Cockcroft S. Gomperts B.D. ATP induces nucleotide permeability in rat mast cells.Nature. 1979; 279: 541-542Crossref PubMed Scopus (192) Google Scholar). Large molecules permeated mast cells only during prolonged ATP application. Seemingly, the P2Z receptor in mast cells could sense how long it was activated and open an increasingly large hole in the plasma membrane. It was unclear how this occurred at a mechanistic level or whether the ATP receptor was an ion channel. Nevertheless, these lovely experiments focused attention on a receptor that may have important implications for signaling and for cytolytic activity in immune cells of the brain and periphery. Similar P2Z receptors exist on many cell types. Of the groups that attempted to clone this channel, one made particular progress (22Nuttle L.C. Dubyak G.R. Differential activation of cation channels and non-selective pores by macrophage P2z purinergic receptors expressed in Xenopus oocytes.J. Biol. Chem. 1994; 269: 13988-13996Abstract Full Text PDF PubMed Google Scholar). However, it was 1996 when the Glaxo group cloned the P2X7 receptor channel (32Surprenant A. Rassendren F. Kawashima E. North R.A. Buell G. The cytolytic P2Z receptor for extracellular ATP identified as a P2X receptor (P2X7).Science. 1996; 272: 735-738Crossref PubMed Scopus (1412) Google Scholar). The P2X7 channel seemed to reproduce many of the phenotypes of the P2Z receptor, including its time-dependent change in permeability. The P2X7 channel had two forms of functioning: for brief application of agonist, it opened to a channel pore that was permeable to small cations, whereas, for longer activation, a larger pore opened that was permeable to molecules as large as the 630 Da YO-PRO-1 dye. This experiment demonstrated, for the first time, real time changes in ion selectivity for a molecularly defined, single ion channel species (but see 23Petrou S. Ugur M. Drummond R.M. Singer J.J. Walsh J.V.J. P2X7 purinoceptor expression in Xenopus oocytes is not sufficient to produce a pore-forming P2Z-like phenotype.FEBS Lett. 1997; 411: 339-345Abstract Full Text Full Text PDF PubMed Scopus (83) Google Scholar). Seven distinct P2X receptor subunits are known (numbered P2X1 through to P2X7). Each P2X subunit is between 379 and 595 amino acids in length (P2X6, smallest; P2X7, longest), and amino acid identity varies between 38% and 48% in pairwise comparisons. P2X subunits have two hydrophobic segments of sufficient length to cross the membrane, and the amino and carboxy termini are both intracellular. The largest part of the protein is extracellular, and this loop has ten conserved cysteines (not all shown in Figure 1A). This view of the P2X channel topology (Figure 1) is supported by much data, including substituted cysteine accessibility mutagenesis, identification, introduction and mutagenesis of glycosylation sites, and the functional expression of tail-to-head tandem subunit channels (24Rassendren F. Buell G. Newbolt A. North R.A. Surprenant A. Identification of amino acid residues contributing to the pore of a P2X receptor.EMBO J. 1997; 16: 3446-3454Crossref PubMed Scopus (176) Google Scholar, 6Egan T.M. Haines W.R. Voigt M.M. A domain contributing to the ion channel of ATP-gated P2X2 receptors identified by the substituted cysteine accessibility method.J. Neurosci. 1998; 18: 2350-2359Crossref PubMed Google Scholar, 19Newbolt A. Stoop R. Virginio C. Surprenant A. North R.A. Buell G. Rassendren F. Membrane topology of an ATP-gated ion channel (P2X receptor).J. Biol. Chem. 1998; 273: 15177-15182Crossref PubMed Scopus (102) Google Scholar, 33Torres G.E. Egan T.M. Voigt M.M. N-Linked glycosylation is essential for the functional expression of the recombinant P2X2 receptor.Biochemistry. 1998; 37 (a): 4845-4851Crossref Scopus (53) Google Scholar, 34Torres G.E. Egan T.M. Voigt M.M. Topological analysis of the ATP-gated ionotropic P2X2 receptor subunit.FEBS Lett. 1998; 425 (b): 19-23Abstract Full Text Full Text PDF PubMed Scopus (54) Google Scholar). The pore is lined by the second transmembrane region, and its narrowest part, the channel gate, is likely to include a conserved glycine in this hydrophobic stretch (G342 of P2X2; 6Egan T.M. Haines W.R. Voigt M.M. A domain contributing to the ion channel of ATP-gated P2X2 receptors identified by the substituted cysteine accessibility method.J. Neurosci. 1998; 18: 2350-2359Crossref PubMed Google Scholar). A conserved aspartate (position 349 in P2X2) is internal to the gate (24Rassendren F. Buell G. Newbolt A. North R.A. Surprenant A. Identification of amino acid residues contributing to the pore of a P2X receptor.EMBO J. 1997; 16: 3446-3454Crossref PubMed Scopus (176) Google Scholar). The greatest divergence among P2X channels is in the carboxy tail, which varies in length between 31 residues for P2X4 and 242 residues for P2X7 (32Surprenant A. Rassendren F. Kawashima E. North R.A. Buell G. The cytolytic P2Z receptor for extracellular ATP identified as a P2X receptor (P2X7).Science. 1996; 272: 735-738Crossref PubMed Scopus (1412) Google Scholar). We remain naive about the agonist-binding site, the region responsible for the binding-gating link, or the role of TM1 in the pore. The cysteine-rich extracellular loop forms up to 60% of the protein, and we know little about its structure or physiological function. 4Cockcroft S. Gomperts B.D. ATP induces nucleotide permeability in rat mast cells.Nature. 1979; 279: 541-542Crossref PubMed Scopus (192) Google Scholar suggested that the less selective state of the P2Z channel may be due to successive oligomerization of existing monomeric channels, so that the pore would grow in diameter as oligomerization proceeded. Recent efforts to determine the P2X channel stoichiometry indicate that the channel is a trimer, but the authors also suggest that the channel may exist in the hexamer state (20Nicke A. Baumert H.G. Rettinger J. Eichele A. Lambrecht G. Mutschler E. Schmalzing G. P2X1 and P2X3 receptors form stable trimers a novel structural motif of ligand-gated ion channels.EMBO J. 1998; 17: 3016-3028Crossref PubMed Scopus (466) Google Scholar). However, other studies on the extracellular domain of P2X2 channels show that they assemble as tetramers (13Kim M. Yoo O.J. Choe S. Molecular assembly of the extracellular domain of P2X2, an ATP-gated ion channel.Biochem. Biophys. Res. Commun. 1997; 240: 618-622Crossref PubMed Scopus (60) Google Scholar). Thus, the issue of stoichiometry is still not settled, but, whatever the number, no example has arisen in the past 20 years of a channel from higher eukaryotes whose subunit stoichiometry changes on a time scale of seconds. The synthetic channel formed by the fungal antibiotic alamethicin displays subconductance states that differ in selectivity among ions of various sizes (10Hanke W. Boheim G. The lowest conductance state of the alamethicin pore.Biochim. Biophys. Acta. 1980; 596: 456-462Crossref PubMed Scopus (107) Google Scholar). In one interpretation, this arises from a single channel with a variably sized pore: the pore accretes alamethicin molecules (nine to ten; 10Hanke W. Boheim G. The lowest conductance state of the alamethicin pore.Biochim. Biophys. Acta. 1980; 596: 456-462Crossref PubMed Scopus (107) Google Scholar). Nevertheless, present knowledge about the atomic scale structure of the selectivity filter (for instance5Doyle D.A. Cabral J.M. Pfuetzner R.A. Kuo A. Gulbis J.M. Cohen S.L. Chait B.T. MacKinnon R. The structure of the potassium channel molecular basis of K+ conduction and selectivity.Science. 1998; 280: 69-77Crossref PubMed Scopus (5475) Google Scholar) prepares us for the possibility that the selectivity filter could by reshaped by angstrom level motions of side chains in and near the pore. Unfortunately, we still lack detailed structural information for dynamic changes in any selectivity filter. When the P2X7 channel was identified, ion selectivity changes were thought unique to this ion channel. A clear difference between P2X7 and other P2X channels was its longer C tail, and this prompted the obvious question: does this long tail contribute to the observed changes in ion selectivity for P2X7? Indeed, truncating the tail of the P2X7 channel does block the change in ion selectivity (32Surprenant A. Rassendren F. Kawashima E. North R.A. Buell G. The cytolytic P2Z receptor for extracellular ATP identified as a P2X receptor (P2X7).Science. 1996; 272: 735-738Crossref PubMed Scopus (1412) Google Scholar). Two recent papers change this view: neuronal P2X channels can change their ion selectivities as well (12Khakh B. Bao X. Labarca C. Lester H. Neuronal P2X receptor-transmitter-gated cation channels change their ion selectivity in seconds.Nat. Neurosci. 1999; 2: 322-330Crossref PubMed Scopus (306) Google Scholar, 35Virginio C. MacKenzie A. Rassendren F.A. North R.A. Surprenant A. Pore dilation of neuronal P2X receptor channels.Nat. Neurosci. 1999; 2: 315-321Crossref PubMed Scopus (343) Google Scholar). For instance, brain P2X4 channels show a change in ion selectivity, despite their small 31 residue tail (Figure 2). Furthermore, two neuronal P2X2 channel splice variants change their selectivity identically: the full-length P2X2-long channel with a tail of 120 residues, and the P2X2-short channel, which has only 51 residues in its tail (35Virginio C. MacKenzie A. Rassendren F.A. North R.A. Surprenant A. Pore dilation of neuronal P2X receptor channels.Nat. Neurosci. 1999; 2: 315-321Crossref PubMed Scopus (343) Google Scholar) (Figure 2). Changes in selectivity seem to be a general mechanism for P2X channels. We continue to lack information about the function of the cytoplasmic C-terminal tails of P2X receptors, but we suspect that they function in sorting or regulation, like those of amiloride-sensitive sodium channels (29Shimkets R.A. Lifton R.P. Canessa C.M. The activity of the epithelial sodium channel is regulated by clathrin-mediated endocytosis.J. Biol. Chem. 1997; 272: 25537-25541Crossref PubMed Scopus (236) Google Scholar). On the other hand, two recent papers show that time-dependent ion selectivity changes are at least partially governed by the pore-lining second transmembrane domain in P2X channels (12Khakh B. Bao X. Labarca C. Lester H. Neuronal P2X receptor-transmitter-gated cation channels change their ion selectivity in seconds.Nat. Neurosci. 1999; 2: 322-330Crossref PubMed Scopus (306) Google Scholar, 35Virginio C. MacKenzie A. Rassendren F.A. North R.A. Surprenant A. Pore dilation of neuronal P2X receptor channels.Nat. Neurosci. 1999; 2: 315-321Crossref PubMed Scopus (343) Google Scholar). In P2X channels, it is possible to dramatically up- and downregulate the ability to change ion selectivity by mutating residues that line the channel pore. It has been appreciated for about a decade that one obtains large effects on ion selectivity by mutating residues that line the permeation pathway of ion channels. The new studies verify that the ability of P2X channels to change their ion selectivity depends on the same permeation pathway as the flow of ions for the more conventional, initial, high-selectivity state. The mutations that up- and downregulate ion selectivity changes are located at G347 of P2X4/T339 of P2X2 or very close to the narrowest part of the "normal" channel pore (24Rassendren F. Buell G. Newbolt A. North R.A. Surprenant A. Identification of amino acid residues contributing to the pore of a P2X receptor.EMBO J. 1997; 16: 3446-3454Crossref PubMed Scopus (176) Google Scholar, 6Egan T.M. Haines W.R. Voigt M.M. A domain contributing to the ion channel of ATP-gated P2X2 receptors identified by the substituted cysteine accessibility method.J. Neurosci. 1998; 18: 2350-2359Crossref PubMed Google Scholar, 12Khakh B. Bao X. Labarca C. Lester H. Neuronal P2X receptor-transmitter-gated cation channels change their ion selectivity in seconds.Nat. Neurosci. 1999; 2: 322-330Crossref PubMed Scopus (306) Google Scholar, 35Virginio C. MacKenzie A. Rassendren F.A. North R.A. Surprenant A. Pore dilation of neuronal P2X receptor channels.Nat. Neurosci. 1999; 2: 315-321Crossref PubMed Scopus (343) Google Scholar). In the simplest interpretation, ion selectivity changes occur because the selectivity filter opens, perhaps on the order of angstroms, near the channel gate. Remarkably, P2X channels remain cation selective (35Virginio C. MacKenzie A. Rassendren F.A. North R.A. Surprenant A. Pore dilation of neuronal P2X receptor channels.Nat. Neurosci. 1999; 2: 315-321Crossref PubMed Scopus (343) Google Scholar). As noted, the time scale of these transitions to the low selectivity state is slower (hundreds of milliseconds to seconds) than opening to the initial state (milliseconds at most) and can be triggered by sustained ATP applications or by repetitive pulses of ATP (12Khakh B. Bao X. Labarca C. Lester H. Neuronal P2X receptor-transmitter-gated cation channels change their ion selectivity in seconds.Nat. Neurosci. 1999; 2: 322-330Crossref PubMed Scopus (306) Google Scholar, 35Virginio C. MacKenzie A. Rassendren F.A. North R.A. Surprenant A. Pore dilation of neuronal P2X receptor channels.Nat. Neurosci. 1999; 2: 315-321Crossref PubMed Scopus (343) Google Scholar). Ion selectivity changes occur for natively expressed P2X channels in neurons. What is the biological and pathophysiological significance of the fact that neuronal P2X receptor channel proteins have an intrinsic ability to detect the duration of their activation, and that they respond by decreasing their ionic selectivity? We expect that researchers in this field will employ the full range of modern physiological, pharmacological, and genetic approaches to attack this problem. Bevan and Yeats showed that acid pH could evoke biphasic inward cation currents in sensory neurons from dorsal root ganglion; the second phase had a less positive reversal potential than the first (2Bevan S. Yeats J. Protons activate a cation conductance in a sub-population of rat dorsal root ganglion neurones.J. Physiol. (Lond.). 1991; 433: 145-161Crossref Scopus (295) Google Scholar). Acid pH can cause pain, and the second sustained phase of these currents is an important candidate as a mediator of nonadaptive pain evoked by acid pH. Work on acid-sensing channels was propelled forward immensely when Michel Lazdunski's group cloned a cDNA for a mammalian channel that responded to acid pH changes with fast inward currents (36Waldmann R. Champigny G. Bassilana F. Heurteaux C. Lazdunski M. A proton-gated cation channel involved in acid-sensing.Nature. 1997; 386: 173-177Crossref PubMed Scopus (1049) Google Scholar); these channels are related to degenerins of C. elegans and to the amiloride-sensitive sodium channels. Subsequently, many acid-sensing channels have been identified and one of these, DRASIC (orsal oot cid-ensing on hannel), is found only in sensory ganglia, implying an important role in the transduction of pain (15Lingueglia E. de Weille J.R. Bassilana F. Heurteaux C. Sakai H. Waldmann R. Lazdunski M. A modulatory subunit of acid sensing ion channels in brain and dorsal root ganglion cells.J. Biol. Chem. 1997; 272: 29778-29783Crossref PubMed Scopus (416) Google Scholar). When DRASIC is coexpressed with another acid-sensing channel subunit, MDEG2 (ammalian enerin related channel ), the novel heteromeric channel shows biphasic currents evoked by acid pH (Figure 1B; 15Lingueglia E. de Weille J.R. Bassilana F. Heurteaux C. Sakai H. Waldmann R. Lazdunski M. A modulatory subunit of acid sensing ion channels in brain and dorsal root ganglion cells.J. Biol. Chem. 1997; 272: 29778-29783Crossref PubMed Scopus (416) Google Scholar). After the first, transient phase of the current, the second sustained phase is associated with a selectivity change of the channel from sodium-selective to nonselective (15Lingueglia E. de Weille J.R. Bassilana F. Heurteaux C. Sakai H. Waldmann R. Lazdunski M. A modulatory subunit of acid sensing ion channels in brain and dorsal root ganglion cells.J. Biol. Chem. 1997; 272: 29778-29783Crossref PubMed Scopus (416) Google Scholar). The biphasic current, with a decreased ion selectivity of the second phase, thus recalls the proton-gated currents described in sensory neurons (2Bevan S. Yeats J. Protons activate a cation conductance in a sub-population of rat dorsal root ganglion neurones.J. Physiol. (Lond.). 1991; 433: 145-161Crossref Scopus (295) Google Scholar). A component of acid pain is nonadaptive, and the switch in ion selectivity may mediate this type of pain. Definitive experiments are eagerly awaited, but if this is true MDEG2/DRASIC channels act as primary afferent "activity detectors." Furthermore, homologs of acid-sensing channels, the degenerins, cause severe neurodegeneration in C. elegans, and this heightens interest in acid-sensing channels as possible mediators of neurodegeneration in mammals. One acid-sensing channel is found abundantly in the brain (ASIC1; 36Waldmann R. Champigny G. Bassilana F. Heurteaux C. Lazdunski M. A proton-gated cation channel involved in acid-sensing.Nature. 1997; 386: 173-177Crossref PubMed Scopus (1049) Google Scholar); a human homolog has been identified that is widely expressed in the brain (hASIC3), and it shows biphasic currents and ion selectivity changes evoked by acid pH (1Babinski K. Le K.T. Seguela P. Molecular cloning and regional distribution of a human proton receptor subunit with biphasic functional properties.J. Neurochem. 1999; 72: 51-57Crossref PubMed Scopus (150) Google Scholar). Although synaptic vesicles have an acidic interior that would in turn acidify the synaptic cleft upon exocytosis, it is unlikely that such a pH decrease would last long enough to produce the second phase of channel selectivity during synaptic transmission. But metabolic activity in general results in proton extrusion, and perhaps conditions exist when the buffering capacity of extracellular fluid is exceeded to the extent that acid-sensing ion channels undergo ion selectivity changes and play a role in pathophysiology. Do all transmitter-gated ion channels change their selectivity? No. Work over the last 15 years has shown that many transmitter-gated ion channels have distinct conductance states, but where researchers have studied permeation in detail, these subconductance states do not generally differ in their ion selectivities (7Fox J.A. Ion channel subconductance states.J. Mem. Biol. 1987; 97: 1-8Crossref PubMed Scopus (103) Google Scholar). Recent studies confirm that recombinant α4β2 nicotinic and 5−HT3 serotonin channels do not change their ionic selectivity in a manner similar to P2X channels (12Khakh B. Bao X. Labarca C. Lester H. Neuronal P2X receptor-transmitter-gated cation channels change their ion selectivity in seconds.Nat. Neurosci. 1999; 2: 322-330Crossref PubMed Scopus (306) Google Scholar, 35Virginio C. MacKenzie A. Rassendren F.A. North R.A. Surprenant A. Pore dilation of neuronal P2X receptor channels.Nat. Neurosci. 1999; 2: 315-321Crossref PubMed Scopus (343) Google Scholar). However, a mutant (N598Q) at a well-studied position in the permeation pathway of the NMDA receptor NR1 subunit (27Schneggenburger R. Ascher P. Coupling of permeation and gating in an NMDA-channel pore mutant.Neuron. 1997; 18: 167-177Abstract Full Text Full Text PDF PubMed Scopus (84) Google Scholar) displays two open states with different ion selectivity. To our knowledge, subconductance states of native and wild-type NMDA channels do not differ in ion selectivity. Nevertheless, the data from the mutant NMDA channel buttresses our argument, namely that a single ion channel can have distinct states each with different ion selectivity (27Schneggenburger R. Ascher P. Coupling of permeation and gating in an NMDA-channel pore mutant.Neuron. 1997; 18: 167-177Abstract Full Text Full Text PDF PubMed Scopus (84) Google Scholar). The importance of sodium channels to the cardiac action potential is obvious. Classically, one thinks of the fast inward sodium current as selective for sodium ions, and much work supports this view. However, this view of the cardiac sodium channel was changed last year by the work of W. J. Lederer's group (25Santana L.F. Gomez A.M. Lederer W.J. Ca2+ flux through promiscuous cardiac Na+ channels slip-mode conductance.Science. 1998; 279: 1027-1033Crossref PubMed Scopus (150) Google Scholar). Sodium channels in cardiac myocytes enter a state that is dependent on the presence of intracellular cyclic AMP and phosphorylation of the channel (25Santana L.F. Gomez A.M. Lederer W.J. Ca2+ flux through promiscuous cardiac Na+ channels slip-mode conductance.Science. 1998; 279: 1027-1033Crossref PubMed Scopus (150) Google Scholar). The cAMP-dependent state was dubbed "slip-mode conductance," and, under these conditions, calcium ions become permeant at these channels. It is hard to imagine a cell type where such a mechanism could potentially have a bigger physiological impact than in cardiac myocytes: in these cells, calcium is key to the timing and force of contraction. The authors presented evidence based on electrophysiological measurements of reversal potentials and on measuring the entry of calcium into cells by direct imaging (25Santana L.F. Gomez A.M. Lederer W.J. Ca2+ flux through promiscuous cardiac Na+ channels slip-mode conductance.Science. 1998; 279: 1027-1033Crossref PubMed Scopus (150) Google Scholar). That the channel was a sodium channel is supported by the finding that slip-mode conductance could be blocked by tetrodotoxin, the classic sodium channel blocker that blocks myocyte sodium channels with lower affinity than those in neurons. Overall, these findings demonstrate that cardiac sodium channels permeate calcium ions under some circumstances, and this is similar to findings where atrionatriuretic peptide allows calcium ions to permeate through sodium channels (30Sorbera L.A. Morad M. Atrionatriuretic peptide transforms cardiac sodium channels into calcium-conducting channels.Science. 1990; 247: 969-973Crossref PubMed Scopus (60) Google Scholar; but see 28Sheets M.F. Hanck D.A. Sorbera L.A. Morad M. Atrionatiuretic peptide and calcium-conducting sodium channels.Science. 1991; 252: 449-452Crossref PubMed Scopus (8) Google Scholar). In light of the potential importance of calcium permeability of sodium channels, it was important to show that cardiac sodium channels show an increase in permeability to Ca2+ ions when expressed in a simple null cell. Two groups have addressed this question. Nuss and Marbàn show that sodium channels do not change their ion selectivity when the α subunit is expressed alone, or with the β1 subunit, in CHO cells (21Nuss H.B. Marban E. Balke C.W. Goldman L. Aggarwal R. Shorofsky S.R. Cruz J.S. Santana L.F. Frederick C.A. Isom L.L. et al.Whether "slip-mode conductance" occurs.Science. 1999; 284 (711–711a)Crossref Google Scholar). This led the authors to conclude that the original observation was perhaps an experimental artifact; indeed, cardiac myocytes are subject to various voltage-clamp artifacts (21Nuss H.B. Marban E. Balke C.W. Goldman L. Aggarwal R. Shorofsky S.R. Cruz J.S. Santana L.F. Frederick C.A. Isom L.L. et al.Whether "slip-mode conductance" occurs.Science. 1999; 284 (711–711a)Crossref Google Scholar). Lederer's group performed similar physiological manipulations after expressing sodium channel subunits in HEK cells (21Nuss H.B. Marban E. Balke C.W. Goldman L. Aggarwal R. Shorofsky S.R. Cruz J.S. Santana L.F. Frederick C.A. Isom L.L. et al.Whether "slip-mode conductance" occurs.Science. 1999; 284 (711–711a)Crossref Google Scholar). This study shows that the ion selectivity change in sodium channels is most pronounced when three subunits are expressed together (αβ1β2), less so when the α subunit is expressed with any one β subunit, and not observable if any subunit is expressed in isolation. In these experiments, artifactual results are less of a concern, because HEK cells can be voltage clamped with high fidelity, and the authors performed experiments to measure calcium entry directly into cells (Figure 1C). Overall, these studies with the αβ1β2 heteromultimer extend earlier studies in cardiac myocytes (25Santana L.F. Gomez A.M. Lederer W.J. Ca2+ flux through promiscuous cardiac Na+ channels slip-mode conductance.Science. 1998; 279: 1027-1033Crossref PubMed Scopus (150) Google Scholar), but they also raise many more questions. It is not clear why there is a such a dramatic difference between expressing the channels in HEK cells or CHO cells (21Nuss H.B. Marban E. Balke C.W. Goldman L. Aggarwal R. Shorofsky S.R. Cruz J.S. Santana L.F. Frederick C.A. Isom L.L. et al.Whether "slip-mode conductance" occurs.Science. 1999; 284 (711–711a)Crossref Google Scholar). Perhaps either cell type expresses modulatory subunits, or perhaps the basal kinase and/or phosphatase activity of the cells is a key factor. For the present, it appears clear that, when all three subunits (αβ1β2) are coexpressed in HEK cells, there is marked calcium entry through a sodium channel. Voltage-gated potassium channels undergo inactivation by at least two distinct mechanisms during membrane depolarization. N type fast inactivation is mediated by an N-terminal cytoplasmic blocking particle that moves to block the channel during gating; this is, of course, the classic ball and chain model for fast inactivation. The usually slower mode of C type inactivation is not due to a blocking particle but involves changes in the conformation of the channel outer vestibule (37Yellen G. The moving parts of voltage-gated ion channels.Q. Rev. Biophys. 1998; 31: 239-295Crossref PubMed Scopus (390) Google Scholar). Recent studies show that some C type inactivation actually consists of a rather more specific change in the selectivity filter: C type inactivated Kv channels cannot conduct K+, but they acquire permeability for Na+ (31Starkus J.G. Kuschel L. Rayner M.D. Heinemann S.H. Ion conduction through C-type inactivated Shaker channels.J. Gen. Physiol. 1997; 110: 539-550Crossref PubMed Scopus (151) Google Scholar, 14Kiss L. LoTurco J. Korn S.J. Contribution of the selectivity filter to inactivation in potassium channels.Biophys. J. 1999; 76: 253-263Abstract Full Text Full Text PDF PubMed Scopus (148) Google Scholar). The structure of the KCSA potassium channel suggests a detailed mechanism: C type inactivation would involve inward angstrom scale movements in the planar network of aromatic side chains that buttress the selectivity filter; the newly constricted filter would allow Na+, but not K+, ions to permeate (37Yellen G. The moving parts of voltage-gated ion channels.Q. Rev. Biophys. 1998; 31: 239-295Crossref PubMed Scopus (390) Google Scholar). Although the change in permeability for Kv channels during C type inactivation is thus far the only example of such a change that allows a reasonable structural hypothesis, this change in selectivity is subtly different from the others discussed above. For P2X, acid-sensing, and cardiac sodium channels, the ion selectivity change results in permeation by previously impermeable, larger cations, as though the selectivity filter dilates. In the case of Kv channels, the opposite occurs because the channel becomes more permeable to a smaller ion. Beautiful experiments performed by Gary Yellen's group show that the outer vestibule can undergo dynamic rearrangements, such that the distance between two engineered cysteines (in the vestibule) decreases during gating (16Liu Y. Jurman M.E. Yellen G. Dynamic rearrangement of the outer mouth of a K+ channel during gating.Neuron. 1996; 16: 859-867Abstract Full Text Full Text PDF PubMed Scopus (391) Google Scholar). These experiments show that the diameter of the outer vestibule is dynamic, and not fixed. In the case of P2X channels, acid-sensing channels, and cardiac sodium channels, we think it probable that the ion selectivity change occurs because of similar mechanisms, namely dynamic rearrangement in the channel pore, and the resulting appearance of a distinct state of the channel with altered ionic selectivity. Presumably, the experimental strategy that Yellen adopted is generic and can be applied to these other channels. Shaker channels carrying the T442S mutation in the potassium channel signature sequence (TXXXGYGD) display fully open and subconductance states each with different ion preferences during activation (38Zheng J. Sigworth F.J. Selectivity changes during activation of mutant Shaker potassium channels.J. Gen. Physiol. 1997; 110: 101-117Crossref PubMed Scopus (112) Google Scholar). It has been suggested that these conductances represent voltage-dependent transitions of the tetrameric channel through different states during activation. In such a scheme, the fully conductive state has all four subunits in the "permissive" conformation, whereas intermediate states conceivably represent channels with only one, two, or three permissive subunits (3Chapman M.L. VanDongen H.M.A. VanDongen A.M.J. Activation-dependent subconductance levels in the drk 1 channel suggest a subunit basis for ion permeation and gating.Biophys. J. 1997; 72: 708-719Abstract Full Text PDF PubMed Scopus (99) Google Scholar, 38Zheng J. Sigworth F.J. Selectivity changes during activation of mutant Shaker potassium channels.J. Gen. Physiol. 1997; 110: 101-117Crossref PubMed Scopus (112) Google Scholar). Implicitly, in this model, the permissive transitions that ultimately lead to channel opening can occur somewhat independently in each subunit. This implies that the T442S mutation stabilizes intermediate states of Shaker whose ion preferences are distinct from the fully open state (38Zheng J. Sigworth F.J. Selectivity changes during activation of mutant Shaker potassium channels.J. Gen. Physiol. 1997; 110: 101-117Crossref PubMed Scopus (112) Google Scholar). Thus, these studies provide perhaps the most detailed discussion of how a single ion channel species may possess different ion selectivities, and provide a quantitative framework for research on other ion channels. Cyclic nucleotide–gated (CNG) channels are designated as members of the voltage-gated ion channel superfamily because of structural similarities (Figure 1), but they are also strikingly different—CNG channels are directly gated by intracellular cAMP and cGMP. CNG channels are found in retinal photoreceptors and in olfactory neurons; in both cases, they act as transducers linking changes in intracellular cyclic nucleotide to membrane potential. Our current molecular level understanding of these channels indicates that the CNG channels of the retina and olfactory sensory neurons are formed by α subunits and modulatory β subunits. Because, first, CNG channels are calcium permeable and, second, a basal concentration of cGMP in photoreceptors keeps CNG channels open, the dark current includes a calcium component (8Frings S. Tuning of permeation in cyclic nucleotide-gated channels.J. Gen. Physiol. 1999; 113: 795-797Crossref PubMed Scopus (4) Google Scholar). However, light triggers cGMP hydrolysis, and CNG channels close, causing membrane hyperpolarization and a fall in photoreceptor calcium levels. Hackos and Korenbrot studied the retinal photoreceptor CNG channel at low (physiological) concentrations of cGMP and also at higher concentrations (9Hackos D.H. Korenbrot J.I. Divalent cation selectivity is a function of gating in native and recombinant cyclic-nucleotide-gated ion channels from retinal photoreceptors.J. Gen. Physiol. 1999; 113: 799-817Crossref PubMed Scopus (38) Google Scholar). Remarkably, they report that increasing cGMP not only increases the open probability of CNG channels, as expected, but also affects their calcium permeability. It turns out that CNG channels have lower calcium permeability at low cGMP concentrations. The size of the cGMP-evoked current, and therefore the open probability, is described by Hill functions, and the same functions also describe the change in calcium permeability as a function of cGMP concentration. The inescapable conclusion: channel gating and ion selectivity are linked (9Hackos D.H. Korenbrot J.I. Divalent cation selectivity is a function of gating in native and recombinant cyclic-nucleotide-gated ion channels from retinal photoreceptors.J. Gen. Physiol. 1999; 113: 799-817Crossref PubMed Scopus (38) Google Scholar). As a result of the cloning efforts of many groups, we know the molecular identity of CNG channels: Hackos and Korenbrot next exploited this knowledge by expressing rod photoreceptor α and β subunits in Xenopus oocytes and tested for calcium permeability changes as a function of cGMP concentration. They report that both α and β subunits are necessary for recombinant rod CNG channels to change their calcium permeability as a function of cGMP concentration: the α subunit alone is not enough. This appears similar to findings with cardiac sodium channels: they become calcium permeable only when β and α subunits are expressed together, in the presence of intracellular cAMP (see above). CNG channels have relatives. The growing family of hyperpolarization-activated cation channels (Ih) are similar to CNG channels in many respects (17Ludwig A. Zong X. Jeglitsch M. Hofmann F. Biel M. A family of hyperpolarization-activated mammalian cation channels.Nature. 1998; 393: 587-591Crossref PubMed Scopus (757) Google Scholar, 18Ludwig A. Zong X. Stieber J. Hullin R. Hofmann F. Biel M. Two pacemaker channels from human heart with profoundly different activation kinetics.EMBO J. 1999; 18: 2323-2329Crossref PubMed Scopus (304) Google Scholar, 26Santoro B. Liu D.T. Yao H. Bartsch D. Kandel E.R. Siegelbaum S.A. Tibbs G.R. Identification of a gene encoding a hyperpolarization-activated pacemaker channel of brain.Cell. 1998; 93: 717-729Abstract Full Text Full Text PDF PubMed Scopus (561) Google Scholar). There is considerable similarity between CNG channels and Ih channels: both channels have a CNG binding site, both are profoundly modulated by cyclic nucleotides, and both are mixed cationic channels. In light of these findings, it might be fruitful to test whether Ih channels change their ionic permeability as a function of intracellular cyclic nucleotide concentration, like their cousins the CNG channels. Hodgkin and Huxley defined and shaped the basis for understanding and investigating integral membrane structures with high selectivity for permeation of individual ion species (11Hodgkin A.L. Huxley A.F. A quantitative description of membrane current and its application to conduction and excitation in nerve.J. Physiol. (Lond.). 1952; 117: 500-544Crossref Scopus (13126) Google Scholar). It was only 1 year ago that the first 3D structure of a potassium channel was found, some secrets of permeation and selectivity confirmed, and yet new ones revealed (5Doyle D.A. Cabral J.M. Pfuetzner R.A. Kuo A. Gulbis J.M. Cohen S.L. Chait B.T. MacKinnon R. The structure of the potassium channel molecular basis of K+ conduction and selectivity.Science. 1998; 280: 69-77Crossref PubMed Scopus (5475) Google Scholar). A common theme to the hundreds of known ion channels is a central cavity or pore(s) through which, during gating, ions flow depending on their charge, size, and concentration. Which ions flow is determined along the permeation pathway by the selectivity filter (5Doyle D.A. Cabral J.M. Pfuetzner R.A. Kuo A. Gulbis J.M. Cohen S.L. Chait B.T. MacKinnon R. The structure of the potassium channel molecular basis of K+ conduction and selectivity.Science. 1998; 280: 69-77Crossref PubMed Scopus (5475) Google Scholar). In this review, we have discussed recent studies showing that some (but clearly not all) ion channels can change their selectivity, such that normally impermeant ions can also flow or that normally permeant ions can flow with greater ease. Thus, the selectivity filter is a dynamic structure; and these dynamics occur on the time scale of milliseconds to tens of seconds. These new findings challenge us both to determine the dynamic aspects of channel structure and also to understand the physiological sequelae of changes in ionic selectivity during synaptic transmission and impulse firing. This work was supported by a Wellcome Trust (UK) International Prize Travelling Fellowship and the NIH (NS-11756).