Portal de Periódicos da CAPES

Unlocking Family Secrets

1999; Cell Press; Volume: 97; Issue: 5 Linguagem: Inglês

10.1016/s0092-8674(00)80765-5

1097-4172

David E. Clapham,

Lipid Membrane Structure and Behavior

Ion channels originate and orchestrate the signals that drive the contracting muscle, beating heart, and thinking brain. As macromolecular protein tunnels, ion channels span the cell membrane’s lipid bilayer. Approximately one-third of the energy generated in cells is expended to maintain the ionic gradient across the cell membrane that makes the cell a battery. The energy collected and stored by the cell is spent in short bursts by the ion channel. Collectively, they make complex temporal patterns, such as action potentials, to coordinate the entry of calcium. Ion channels are classified by whether they pass sodium, potassium, calcium, or chloride ions, although some are indiscriminate. They are opened and closed (or gated) by either extracellular ligands, transmembrane voltage, or intracellular second messengers (for review, see 7Hille B Ionic Channels of Excitable Membranes. Sinauer Associates, Sunderland, MA1992Google Scholar, 1Ackerman M Clapham D.E N. Engl. J. Med. 1997; 336: 1575-1586Crossref PubMed Scopus (295) Google Scholar). In this brief review, I will summarize some recent advances in our understanding of the function of the transmembrane domains of highly cation-selective ion channels. The deepest, darkest secrets of ion channels are locked in their transmembrane (TM) domains. The 40 Å–thick lipid bilayer membrane presents a formidable barrier to a charged ∼1 Å cation, with only one in 1000 billion ions moving through a square centimeter of lipid bilayer by simple diffusion. In contrast, when an ion channel opens, as many as 10 million ions move across this single open pore per second. The α-helical, hydrophobic helices tunnel through the plasma membrane’s sea of lipid to permit charged, hydrated ions to cross from one side to the other. Highly selective ion channels enable the cell membrane potential to swing well above and below 0 mV. These selective ion channels allow only a very specific subset of ions to transit the membrane. Remarkably, for very selective ion channels the error rate is only 0.1%. The loss of selectivity by a single amino acid substitution in the pore of K+ channels can lead to cell death. Selectivity, is now much more clearly understood thanks to the near atomic resolution structure of a bacterial K+ channel (KcsA) elucidated last year by Rod MacKinnon and colleagues (6Doyle D.A Cabral J.M Pfuetzner R.A Kuo A Gulbis J.M Cohen S.L Chait B.T MacKinnon R Science. 1998; 280: 69-77Crossref PubMed Scopus (5468) Google Scholar). The crystal structure of the KcsA channel revealed that it is a tetramer of two transmembrane-spanning α helices (Figure 1; 6Doyle D.A Cabral J.M Pfuetzner R.A Kuo A Gulbis J.M Cohen S.L Chait B.T MacKinnon R Science. 1998; 280: 69-77Crossref PubMed Scopus (5468) Google Scholar). The M1 helices face the lipid membrane, the shorter connecting pore helices lie near the outer membrane, and the M2 helices face the central channel. At both inner and outer membrane faces, layers of aromatic amino acids form a cuff around the pore to keep the opening taut. At the cytoplasmic mouth of KcsA, negatively charged amino acids attract a cloud of cations, effectively concentrating them above the level of surrounding anions. When the channel opens, hydrated cations rapidly accelerate by electrodiffusion into an 18 Å–long hydrophobic canal that leads into a wider, 10 Å–long cavity. In this cavity, dozens of H2O molecules are attracted to the K+ ion to help shield its charge. A negative field generated by the pore helical dipoles also focuses on the cavity, which may help neutralize the charge of the cation. The crucial selectivity filter is a narrow 12 Å region near the extracellular face of the membrane that is lined by carbonyl oxygens along the four backbones of conserved amino acids. In theory, the rings of carbonyl oxygens act as surrogate waters to coordinate dehydrated K+ ions in the channel. In the key selectivity step, K+ ions waiting in the antechamber exchange their waters of hydration for the carbonyl oxygens along the backbones of the selectivity filter amino acids. Once in the 12 Å chute, mutual repulsion by stripped K+ ions helps move them on through to the extracellular solution. One mystery not yet resolved by direct structural observation is why the dehydrated K+ ion coordinates these carbonyl oxygens while the dehydrated Na+ ion cannot. Presumably the pore is stretched open by the surrounding aromatic amino acids to just the right size to pull H2O from the 1.3 Å K+ but not the hydrated 0.9 Å Na+ (6Doyle D.A Cabral J.M Pfuetzner R.A Kuo A Gulbis J.M Cohen S.L Chait B.T MacKinnon R Science. 1998; 280: 69-77Crossref PubMed Scopus (5468) Google Scholar). It is likely that a variant of the K+ channel theme is present in other selective cationic channels. For Ca2+-selective channels, a ring of negative amino acids transiently traps Ca2+ ions in the selectivity filter while a less charged ring selects for Na+. Since single helical TM domains surrounding a pore could theoretically form an ion channel, why do cation-selective channels occur with two, four, or six helices per functional subunit? Presumably the second M2 helix described above positions the carbonyl backbone and pore helix to impart selectivity, while M1 also shields the pore with a lipid-interfacing surface. It may also provide a handle to impart torsion on the pore helices. For the bacterial outward rectifier K+ channels, the helices pack in a very symmetrical pattern. Shown from above (Figure 2A), the outer and inner helices slant as they move across the membrane—they are clearly not straight pillars. A simplified conceptual diagram is shown in Figure 2B in which the outer helix lies adjacent and distal to the inner helix, the main point being that the M1 helix does not contact adjacent helices. Now a pair of papers add to our understanding of the packing of other K+ channel TM helices (13Minor Jr., D.L Masseling S.J Jan Y.N Jan L.Y Cell. 1999; 96: 879-891Abstract Full Text Full Text PDF PubMed Scopus (129) Google Scholar, 14Monks S.A Needleman D.J Miller C J. Gen. Physiol. 1999; 113: 415-423Crossref PubMed Scopus (119) Google Scholar). 13Minor Jr., D.L Masseling S.J Jan Y.N Jan L.Y Cell. 1999; 96: 879-891Abstract Full Text Full Text PDF PubMed Scopus (129) Google Scholar used a yeast genetic screen to identify functional channels from libraries of an inward rectifier K+ channel (Kir 2.1) containing mutagenized M1 and M2 domains. Surfaces of the helices that faced either the lipid or water were verified by sequence minimization experiments in which specific amino acids were replaced, the channel expressed, and function determined. The only necessary requirement for the M1 protein–lipid interface was found to be side chain hydrophobicity, not shape or size. Not surprisingly, one surface of the M2 helix was found to require a hydrophilic interface for function. These experiments essentially verified the placement of the M2 segment around the central pore for these K+ channels as observed in the bacterial channel structure (6Doyle D.A Cabral J.M Pfuetzner R.A Kuo A Gulbis J.M Cohen S.L Chait B.T MacKinnon R Science. 1998; 280: 69-77Crossref PubMed Scopus (5468) Google Scholar). Then, in a clever set of experiments, Minor et al. examined the packing of the helices through second-site suppressor mutations. Changes in side chains that disrupted function were made at conserved positions in the M1/M2 interfaces and screened against libraries of the other helix of the pair. For example, a serine in M1 (S95) and a glutamine in M2 (Q164) are conserved in the Kir family and in the selection experiments. The authors reasoned that the M1 and M2 helices were linked by polar residues buried in the membrane and, assuming that conserved amino acids are the most relevant, showed that only amino acids substitutions that preserved side chain length and hydrogen bonding potential made functional channels. Further work with other suppressor mutations and a subunit interaction assay led the authors to speculate that Kir M1 helix contacts two M2 helices, one from its own subunit and one from an adjacent subunit (Figure 3). As Doyle et al. pointed out, the mammalian family of voltage-gated K+ channels (Kv) have more homology in their helices to the KcsA than to the Kir family, suggesting that its helical packing order may be a clue to the Kv channels. 14Monks S.A Needleman D.J Miller C J. Gen. Physiol. 1999; 113: 415-423Crossref PubMed Scopus (119) Google Scholar took a somewhat similar approach to Minor et al., using the more complex Kv class, Shaker K+ channel as a starting point. Focusing on the putative S2 TM helix, Monks et al. employed tryptophan as a kind of clumsy oaf amino acid that would disrupt protein–protein interactions but be tolerated at lipid interfaces (see also 4Choe S Stevens C.F Sullivan J.M Proc. Natl. Acad. Sci. USA. 1995; 92: 12046-12049Crossref PubMed Scopus (40) Google Scholar). Surprisingly, their laborious mutagenesis efforts yielded only a whimper of protest from the molecule, with only one mutation (R297W) failing to contribute to a functional channel. Looking more closely at gating parameters culled from their voltage-clamp recordings, they were able to separate mutations into tolerant, low impact mutations (no different from wild type) and higher impact mutations. They concluded that S2 is indeed α helical with a lipid-facing surface and conjecture that two conserved glutamates in the core of the protein–protein interface may participate in a salt bridge with S4 as previously suggested for S3 and S4 (17Papazian D.M Shao X.M Seoh S.A Mock A.F Huang Y Wainstock D.H Neuron. 1995; 14: 1293-1301Abstract Full Text PDF PubMed Scopus (317) Google Scholar). Figure 4 incorporates various aspects of proposed interactions for the Shaker K+ channel, including the S6-facing pore and the position of S5 corresponding to M1 in KcsA. Like the blade of a propeller affixed to the center of the S2 and S3 helices, the S4 helix may twirl very slightly with voltage changes, perhaps opening crevices in the bilayer that allow counterions from the solution to offset positive charges on the S4 helix (for reviews see 16Papazian D.M Bezanilla F News Physiol. Sci. 1997; 12: 203-210Google Scholar, 22Yellen G Quart. Rev. Biophys. 1998; 31: 239-296Crossref PubMed Scopus (388) Google Scholar). S2 and S3 are linked to S4 via salt bridges with their lipophilic backs to the bilayer. Unfortunately, there is no information on the placement of the S1 helix. In fact, loose ends still abound in the Shaker model with many unanswered questions on the detailed placement of the helices (for an alternative packing model see 14Monks S.A Needleman D.J Miller C J. Gen. Physiol. 1999; 113: 415-423Crossref PubMed Scopus (119) Google Scholar). Furthermore, given the tight interactions between helices it is difficult to imagine where to place accessory subunits, such as minK, that have been proposed to border the pore of a related K+ channel. The selectivity of ion channels is not enough to allow organisms to move, sense, and think. Evolutionary tinkering has attached various valves and levers onto the channel in order that voltage changes, various molecules and ions, and time can all turn on and off the flow of ions. Although there are various cytoplasmic domains whose secrets will soon be revealed by crystallography (some functions, such as intersubunit and accessory protein binding, already have; e.g.5Doyle D.A Lee A Lewis J Kim E Sheng M MacKinnon R Cell. 1996; 85: 1067-1076Abstract Full Text Full Text PDF PubMed Scopus (929) Google Scholar, 9Kreusch A Pfaffinger P.J Stevens C.F Choe S Nature. 1998; 392: 945-948Crossref PubMed Scopus (261) Google Scholar, 15Morais J.H Lee A Cohen S.L Chait B.T Li M MacKinnon R Cell. 1998; 95: 649-655Abstract Full Text Full Text PDF PubMed Google Scholar), little is known about the transmembrane structural components that control gating. In the case of the pH-gated bacterial K+ channel, proton binding at the extracellular face of KcsA induces a large conformational change in the C-terminal end of TM2, increasing the diameter of the internal vestibule with channel opening (18Perozo E Cortes D.M Cuello L.G Nat. Struct. Biol. 1998; 5: 459-469Crossref PubMed Scopus (276) Google Scholar). In KcsA, the four M2 helices cross in a bundle, like the top of a teepee, leaving a small aperture at the bundle crossing. This crossing site may be the channel gate, but of course one crystal structure is static and alone cannot answer the gating question. In fact, KcsA was crystallized at pH 7.5, most likely in its closed gating conformation. In recent experiments 8Holmgren M Shin K.S Yellen G Neuron. 1998; 21: 617-621Abstract Full Text Full Text PDF PubMed Scopus (179) Google Scholar substituted cysteine into the Shaker K+ channel S6 to create high-affinity Cd2+-binding sites on the intracellular end of the helical bundle. The metal formed a bridge between the engineered cysteine in one S6 subunit and a native histidine in an adjacent subunit, trapping the channel in an open state. Holmgren et al. suggest that gating occurs at this bundle crossing by slightly twisting (or perhaps scissoring) the helices to expand the ring of the pore defined by the intersection of the S6 helices (see 22Yellen G Quart. Rev. Biophys. 1998; 31: 239-296Crossref PubMed Scopus (388) Google Scholar). For the simple Kir channels, gating is primitive and in part controlled by Mg2+ (10Logothetis D.E Kurachi Y Galper J Neer E.J Clapham D.E Nature. 1987; 325: 321-326Crossref PubMed Scopus (837) Google Scholar) or polyamine block (11Lopatin A Makhina E Nichols C Nature. 1994; 372: 366-369Crossref PubMed Scopus (713) Google Scholar) of the cytoplasmic mouth of the pore. In keeping with the proposed slightly different structural classification for inward rectifier K+ channels, indirect structural data from 12Lu L Nguyen B Zhang X Jian Y Neuron. 1999; 22: 571-580Abstract Full Text Full Text PDF PubMed Scopus (89) Google Scholar suggest that the cytoplasmic opening of Kir 2.1 channels is 10 Å in diameter, widening to more than 12 Å in the center of the membrane. At a position known to be important for inward rectification, the pore may be sufficiently wide to bind three Mg2+ ions or polyamine molecules simultaneously. For another large pore channel, the nonselective, mechanically gated bacterial channel (MscL), crystallography reveals a pentameric structure surrounding a large water-filled opening that narrows through a hydrophilic pore to an occluded hydrophobic gate (3Chang G Spencer R.H Lee A.T Barclay M.T Rees D.C Science. 1998; 282: 2220-2226Crossref PubMed Scopus (825) Google Scholar). For the ligand-gated, nonselective, pentameric nicotinic channels, Unwin has provided structural data at 9 Å resolution on the putative open and closed conformations (20Unwin N Nature. 1995; 373: 37-43Crossref PubMed Scopus (884) Google Scholar). Binding of acetylcholine initiates small rotations of the protein subunits that trigger a change in configuration of α helices lining the pore (21Unwin N J. Struct. Biol. 1998; 121: 181-190Crossref PubMed Scopus (74) Google Scholar), but higher resolution is needed to understand the most interesting aspects of the nicotinic receptor channel gating. The next hurdle in the field of voltage-gated channels will be to determine how movements of S4 are translated into twisting and opening of the S6 helical bundle. Recent creative fluorescent molecule tagging of residues (19Siegel M.S Isacoff E.Y Neuron. 1997; 19: 735-741Abstract Full Text Full Text PDF PubMed Scopus (328) Google Scholar, 2Cha A Bezanilla F J. Gen. Physiol. 1998; 112: 391-408Crossref PubMed Scopus (116) Google Scholar), fluorescence resonance energy transfer, and simultaneous gating charge movement measurements will certainly be brought to bear. Alternatively, this second central problem of ion channels may be solved in one fell swoop by crystallography of voltage-gated ion channels trapped in open and in closed states.