Potassium channel gating: Not an open and shut case

Potassium channels are highly complex molecular systems that modulate the cell potential in response to extracellular or intracellular signals. In multicellular organisms, K⁺ currents are essential to electrical signaling, governing central nervous system, cardiac, renal, and a host of other organ functions. Publication of the first K⁺ channel structure in 1998 (1) represented a landmark in our understanding of the molecular basis for channel function, revealing for the first time the overall architecture of an ion channel and illuminating the basis of the channel's selectivity for K⁺ over other cations. The subject of that study was the pH-activated prokaryotic channel KcsA. Despite the determination of structures of other channel classes, and extensive supporting biophysical data, the means by which K⁺ conduction is switched on and off has remained elusive.

In a recent issue of PNAS, Imai et al. (2) describe elegant and illuminating studies of KcsA that enable them to distinguish different gating states of the channel by using NMR spectroscopy. The three major features of K⁺ channel activity are ion selectivity, conduction, and gating, all of which localize to a permeation pathway at the common interface of the four subunits comprising the pore (1). Although the first two functions have been ascribed to the highly conserved ion selectivity filter positioned in the outer leaflet of the bilayer, considerable functional evidence suggests that the main permeation gate is located at the inner-helix bundle on the intracellular face, where it governs K⁺ access by adopting discrete closed or open conformations. Recent reports have focused on the possibility that the selectivity filter also acts to gate ion current (3). The roles and interdependency of the two conduction gates remain to be elucidated.

The work of Imai et al. (2) makes a significant advance by providing direct NMR evidence for structural changes in the selectivity filter of KcsA, correlating it to open probability, and inferring coupling between events at the intracellular and extracellular gates. Although the nature of these conformational changes cannot be determined directly by the TROSY (transverse relaxation optimized spectroscopy) method employed (4), the correlations are nevertheless compelling. Although previous studies have used TROSY methods to examine these issues (5), the coup that sets the present study apart is its identification of signals in the spectra that not only distinguish resting and activated states of the channel, but also discriminate between activated and inactivated conformations of the selectivity filter.

Although structures of K⁺ channels have conventionally been determined by X-ray crystallography, Imai et al. (2) have taken advantage of the fact that each subunit of KcsA contains just two transmembrane helices and that a functional channel can be reconstituted in detergent micelles to yield a complex that is small enough to afford good quality solution-state NMR spectra. Even so, a combination of TROSY methods and selective labeling of methyl groups in the aliphatic residues valine, leucine, and isoleucine (6) was needed to achieve good resolution.

At neutral pH, KcsA is in a resting state, but under acidic conditions protonation-induced changes at the intracellular face trigger conduction (3). Methyl TROSY spectra of KcsA at low pH displayed a subset of peaks exhibiting two distinct signatures that exchange slowly, reflecting an equilibrium between subtly differing conformations (2) (Fig. 1). Signature resonances were assigned to Val-76, which is in the selectivity filter consensus sequence (VGYG), and Leu-59, which is involved in packing the pore helices at the extracellular face of the bilayer. To interrogate the transition between activated and inactivated states, the authors exploited two functional mutations affecting gating at the selectivity filter (3). Under identical experimental conditions, a noninactivating mutant (E71A) adopts exclusively one conformation, reflecting its measured open probability of 100%, whereas the signal for a rapidly inactivating mutant (Y82A) is split into two populations that approximate its open and closed probabilities. These observations enabled Imai et al. (2) to correlate specific NMR signals to discrete conformations of the extracellular gate.

Fig. 1.

Schematic depicting the relationship between the gating transitions described as resting, activated, and inactivated. At neutral pH, the intracellular gate is closed and the channel is in a resting state. At acidic pH, the intracellular gate opens and the extracellular gates exist in equilibrium between conducting (activated) and nonconducting (inactivated) states. The red question mark signifies that the state of the extracellular gate has not been defined unequivocally in the resting state of the channel.

Assuming that structural changes in the mutants equate to gating transitions in wild-type KcsA, the authors then set out to obtain comparable spectra using wild-type KcsA by varying the pH and temperature, successfully identifying conditions that favored the three key states (resting, activated, and inactivated). In this way, they could relate chemical shift perturbations of methyl groups to the processes of activation and inactivation at the selectivity filter. The two largest perturbations between activated and inactivated states mapped to Val-76. To further investigate the nature of possible conformational changes in the selectivity filter, NOE cross-peaks were then sought between Val-76 and Tyr-78 using a Tyr-labeled preparation. As anticipated, under experimental conditions promoting the resting and activated states of the channel, the specific cross-peaks appeared consistent with the placement of Val-76 and Tyr-78 in the 2.0-Å crystal structure of KcsA (7). In the inactivated state, however, one cross-peak decreased, consistent with an alternative conformation of the selectivity filter in this state. This represents the first direct structural evidence from a functional wild-type channel associating conformational change of the selectivity filter with inactivation.

KcsA is activated by low pH. By mutating Asp, Glu, and His residues, Imai et al. (2) identified a His25 mutant that exhibited both TROSY signatures at acidic pH, indicating that the selectivity filter was alternating between activated and inactivated conformations. The distribution between two forms may be a consequence of a packing void introduced by the side chain truncation. Recently, other groups investigating pH gating have focused on a cluster of charged and polar residues, including His-25, at the intracellular face of the membrane (8, 9). Although Imai et al. found that mutation of Glu-118 (<4 Å from His-25) had no effect on the TROSY signature, one cannot rule out compensatory effects by the network of side chain interactions in the vicinity.

The activation-coupled inactivation process may have a parallel in C-type inactivation of Kv channels (3). Slow C-type inactivation (10) is sensitive to extracellular cation concentrations and entails a cooperative conformational transition of the selectivity filter, predicted to coincide with specific changes in ion site occupancies (11). The response of the selectivity filter to K⁺ concentration was therefore probed, with the key finding that at physiological (intracellular) K⁺ concentrations the equilibrium favors the activated form, whereas as K⁺ levels are depleted it progressively shifts to an inactivated conformation. No intermediate forms are discernable under the conditions of the experiment, indicating threshold switching between states, rather than a gradual collapse of the selectivity filter.

These results may have general application to K⁺ channels, not just to C-type inactivation.

There is evidence that water molecules accompany K⁺ through the membrane, and in this study NOE data were used to probe the binding of water molecules to the selectivity filter of activated and inactivated channels. Taking into account the residence time of water molecules and typically rapid exchange with bulk solvent, a critical finding is that an NOE cross-peak was observed between water and Val-76 only in the inactivated state, indicating that one or more water molecules had a residence time of >300 ps. No such signal was observed for the activated state of the channel. Whereas the authors interpret this as the coordination of water in the selectivity filter by the carbonyls of Val-76, another possibility lies in changes in ordered water molecules packing around the selectivity filter. In the high- and low-K⁺ structures (7), differences in the molecular packing predict an NOE signal between Val-76 and water only in the K⁺-depleted or inactivated state, where an ordered water molecule connects the carbonyl oxygen of Val-76 to the amide N of Gly-77 of a neighboring subunit.

The study by Imai et al. (2) represents an important addition to a growing body of evidence implicating the selectivity filter in K⁺ channel gating. Moreover, it shows functional interdependency of the selectivity filter and the intracellular gate at the helix bundle crossing. The conservation of the canonical pore structure suggests that these results may have general application to K⁺ channels, not just to C-type inactivation. The combined forces of crystallography, NMR, and biophysics are finally revealing the secrets of gating in this important class of ion channels.