Let’s Twist (the S4) Again

When Noda et al. (1) cloned the voltage-gated sodium channel, they predicted that a stretch containing charged residues might adopt a 3₁₀ helix conformation, which was considered rare in proteins. This structure is more tightly wound, longer, and thinner than an α-helix with the same number of residues (2). The charged region Noda et al. (1) referred to is in fact the S4 transmembrane segment, whose movement across the electrical field underlies function of the voltage-sensing domains (VSD) to control the gating of the central pore in voltage-gated ion channels (3). Understanding the structural basis for S4 movement has been the focus of great debate, and several models have been put forward with different degrees of translation and/or rotation. It has been proposed that the positive S4 gating charges are paired with negatively charged amino acids at neighboring transmembrane segments and that, as S4 moves, the ion pairs are sequentially exchanged (3).

Some excitement came into the field when, as envisioned by Noda et al. (1), 3₁₀ conformations were observed in the available ion channel crystal structures. The characteristics of 3₁₀-helices in this structural context may bring some harmony into the S4 controversy, because they would make the S4 tighter-wound, aligning gating charges and favoring their transfer (2) (Fig. 1). Thus, a new model was proposed where the S4 would adopt partial and transient 3₁₀-helix conformations (2). In this issue of the Biophysical Journal, Kubota et al. (4) elegantly test the proposed dynamic switch between S4 α and 3₁₀ conformations of the VSD by using fluorescent spectroscopic approaches combined with electrophysiology. The study is done in the context of the Ciona intestinalis voltage-sensing phosphatase (Ci-VSP), where the VSD regulates enzymatic function, but is thought to retain voltage-sensing mechanisms similar to those in voltage-gated ion channels (5). In this protein, the VSD is monomeric (6), which simplifies the interpretation of the spectroscopic measurements.

Figure 1.

Constructs used by Kubota et al. (4) for LRET and FRET experiments. Schematic representation of Ci-VSP S4 helix (green) flanked by the lanthanide-binding tag that binds Tb³⁺ as LRET donor (yellow circle), and the red fluorescent protein mCherry as LRET acceptor (red). S4 segment length can be deduced from the total distance between fluorophores, after subtracting the size of lanthanide-binding tag (12.1 Å), and the total distance from the end of Ci-VSP S4 to the mCherry chromophore (27.7 Å) (4). (Blue) Charged residues in Ci-VSD S4. The hypothetical α to 3₁₀ transition of the whole S4 that is tested in this study by Kubota et al. (4) is represented. In response to voltage, the α-helix (left) would stretch 9.5 Å, from 28.5 to 38 Å, adopting a 3₁₀ conformation (right). The 3₁₀ conformation shows a better alignment of charged residues, because the 3₁₀ helix has three amino acids per turn (as opposed to 3.6 in the α-helix). In FRET experiments, the donor (ATTO) is positioned at residue 222 into the S4 helix (red circle). The acceptor (mCherry) is the same as in LRET experiments. In this scenario, a transitional α-3₁₀ switch should lead to a decrease in real-time FRET measurements; however, contrary to this hypothesis, an increase in FRET is observed. See Kubota et al. (4) for details. To see this figure in color, go online.

The relevance of the experiments performed by Kubota et al. (4) resides in the fact that, because no crystal structures have been obtained of the resting and intermediate states that would reveal the VSD conformations during gating, the transitional 3₁₀ hypothesis is based mainly on theoretical models. Until now, no experimental evidence has been obtained to physically validate the α-3₁₀ dynamic transition. By using lanthanide-based resonance energy transfer (LRET) and fluorescence resonance energy transfer (FRET), the authors address different questions about the role of α-to-3₁₀ transitions in the S4 movement. Does the interconversion of the helix occur in the whole S4, leading to simultaneous alignment of all sensing charges or, alternatively, does it occur with stepwise α to 3₁₀ transitions restricted to small regions of S4 (2, 4, 7, 8)? What is the duration of such transitions? LRET allows Kubota et al. (4) to estimate absolute changes in distance between a donor (Tb³⁺) and an acceptor (mCherry) that tightly flank Ci-VSP S4 in order to calculate the extension of the α-(or 3₁₀-)helix and test for long-lived transitions; FRET is used to further explore the possibility that 3₁₀ conformations are solely associated to S4 short-lived transitional states.

The first observation by Kubota et al. (4) is that Ci-VSP S4 adopts an α-helical conformation that does not change with voltage (Fig. 1), showing that a complete α-3₁₀ switch to simultaneously align all gating changes is not essential for gating to occur. Does this mean that no α-3₁₀ transitions are associated to Ci-VSP S4 movement? Before reaching this conclusion, there are two considerations to be taken into account:

First, it must be noted that S4 movement has been modeled including three states: resting, activated, and relaxed. In this model, transition from a resting or an activated state to the relaxed state involves an α-3₁₀ conversion (9). In the case of Ci-VSP, where resting and activated states seem to adopt mainly α-helix conformations, an α-3₁₀ switch may occur at the relaxed state. However, this question remains unanswered because, as well noted by the authors, the voltage sensor in their fluorescent constructs seems not to reach the relaxed state (4).

Second, LRET would only detect long-lived interconversions between helices. Molecular dynamics studies in peptide helices have shown that α-to-3₁₀ transitions take place very rapidly, on the nanosecond timescale (2). Using time-resolved FRET (Fig. 1), Kubota et al. (4) show that reversible submillisecond α-to-3₁₀ transitions seem not to occur. Somewhat surprisingly, their results indicate the existence of S4 tilt or rotation after the gating charges have been transferred. Although this conclusion needs further experiments to be confirmed, it is supported by the relatively high anisotropy values of both donor and acceptor. Altogether, Kubota et al. (4) conclude that Ci-VSD S4 does not undergo complete α-to-3₁₀ interconversions during gating. However, their results are compatible with a model in which the α-3₁₀ switch is restricted to a fraction of the transmembrane helix (2, 4, 7, 8).

The observation that Ci-VSD does not undergo large α-3₁₀ conversions during gating may be a unique feature of Ci-VSP. In fact, multiple scenarios might be expected, because 3₁₀ conformations observed in available ion-channel crystal structures cover variable extents. Thus, the prokaryotic NavAb sodium channel shows almost the entire S4 in 3₁₀ conformation, whereas a hybrid α/3₁₀ conformation is observed in the Kv1.2/2.1 chimera structure (2). Similarly to the VSD of Ci-VSP, the Kv7.1 channel S4 shows a pure α-helix conformation (10). In an attempt to characterize the possible conformational switch of a voltage-gated ion channel VSD, the authors push the technique further to test such conformational change for the NavAb channel S4 (11), which is inserted in the structural context of Ci-VSP in two different versions, according to two possible alignments of charged residues. Whereas both constructs produce gating currents, LRET measurements with NavAb S4 are consistent with a pure α-helix without long-lived transitions to a 3₁₀-helix.

It could be concluded that, in a native conformation, the VSD of NavAb does not adopt a 3₁₀ conformation. Nevertheless, although it is incorporated into a functional protein in the context of the plasma membrane, NavAb does not necessarily adopt its native conformation in the experiments presented by Kubota et al. (4). The authors are conscious of this limitation, because the isolated NavAb-S4 is transplanted into Ci-VSP and thus might lack the specific stereochemical environment required for the NavAb-S4 to adopt its 3₁₀ structure. Another relevant difference might be the fact that Ci-VSP is monomeric, whereas NavAb channels are tetramers. However, regardless of the helical conformation that NavAb S4 adopts in the Ci-VSP protein context, gating currents are observed, suggesting again that an α-3₁₀ transition of the whole S4 is not a requirement for gating to occur. Interestingly, 3₁₀-helices are also found in a ligand-gated channel in which the S4 lacks the crucial charges and is thought not to function as a voltage sensor (2).

The contribution of Kubota et al. (4) constitutes the first attempt to physically measure dynamic S4 α-to-3₁₀ transitions from functional VSDs in the membrane. Their main finding is that an α-3₁₀ switch of the whole S4 is not mandatory for gating to occur. Nonetheless, many exciting questions remain about the role of 3₁₀ helices in the S4 structural movement. In fact, a single model may not apply to all voltage-gated ion channels, as hinted by the varying α/3₁₀ conformations observed in the crystal structures. Kubota et al. (4) give the existing models a new twist, setting the stage for further experiments to explore the role of dynamic S4 α-3₁₀ transitions in voltage-gated ion channels function.

Editor: Michael Pusch.