Voltage-dependent dynamics of the BK channel cytosolic gating ring are coupled to the membrane-embedded voltage sensor

Thank you for submitting your article "The BK channel gating ring is strongly coupled to the voltage sensor" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by Richard Aldrich as the Senior and Reviewing Editor. The following individual involved in review of your submission has agreed to reveal his identity: Christopher J Lingle (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Essential revisions:

1) Modify the Title

Reviewers thought the title was vague and could be modified to better convey the message of the manuscript. The term "strongly coupled" also needs to be modified or justified since it implies a quantitative assessment of coupling that was not addressed.

2) Address previous work and novelty

The authors should provide a more complete comparison with previous work, especially with regards to VSD/gating-ring coupling or interaction, to help clarify "what is new" and whether the results are consistent or inconsistent with previous results.

3) Provide more guidance for interpreting the results

Indicating the expected status of voltage sensors under different conditions and manipulations would help readers to interpret the corresponding FRET data. Reviewers also desired more guidance on how to interpret the quantitative relationship between VSD activation and the voltage-dependence of FRET as well as several interesting or puzzling features of the data including the effect of Ba²⁺, the remarkably steep Ca²⁺ dependence of FRET, and the lack of voltage-dependence of FRET in 0 Ca²⁺. Mechanistic or quantitative interpretation is lacking, and ideally should be improved, if only to provide some framework for future evaluation and tests. Proposing a model (i.e. a kinetic scheme) to account for the major features of the relation between Ca²⁺, voltage, and FRET efficiency (i.e. lack of FRET in 0 Ca²⁺ or high Ca²⁺, and correlations observed in the different mutants) would greatly strengthen the conclusions and enhance the significance of the results. While developing a model to account for all features of the data may be beyond the scope of this study, the authors should at minimum indicate which features of the data can or cannot be accounted for by previously proposed models (i.e. Miranda 2013) and whether their data provide any new insight into mechanism that must be included in gating models. This should be possible to do without incorporating additional experiments.

4) Reconcile discrepancies and apparent inconsistencies in the data

The authors should address the apparent inconsistency between the effect of the γ1 subunit on FRET and its predicted effect on VSD activation, as well as discrepancies between some of the results and data previously published.

Results which seem to contradict earlier results include: a) GV shift magnitude of F315A in 100 Ca²⁺; b) Lack of a GV shift for the β3Nβ1 construct; c) Whether they have adequate expression of β3b.

Because the individual reviews include several important points they are included here for you reference.

Reviewer #1:

The manuscript by Miranda et al. examines energetic coupling in BK channels between its modular domains that underlie voltage sensing and calcium sensing, using patch-clamp fluorometry. The authors show here that the FRET efficiency (E) depends strongly on the integrity of the Ca²⁺ binding site in RCK1 (determined by D362/D367), and relatively much less on integrity of the Ca²⁺ bowl site. It is further shown that movements in the gating ring that are tracked by the authors' FRET measurements are not directly linked to channel opening, but instead to voltage-sensor dynamics. The greater relative contribution of the RCK1 site to VSD coupling is corroborated by selective activation of the site by Cd²⁺. However (interestingly), selective activation of the Ca²⁺ bowl site by Ba²⁺ also exhibits voltage-dependence in the FRET response, which is attributed energetic coupling with the movement of the gate. These results thus substantially extend previous work on preferential coupling of Ca²⁺ binding at the RCK1 site with voltage-sensor activation.

Overall, the manuscript is written in a logical format and the data, as presented, are of high quality. This is an outstanding contribution to our understanding of BK channels, forming a bridge between structural and functional studies. I have only a few comments aimed at improving the presentation.

1) The title seems a little vague; maybe the authors could think of one that more strongly conveys the message of this paper, which appears to be that the coupling is preferential, or selective, among the Ca²⁺-activation sites.

2) Unfortunately the authors measurement of FRET efficiency is represented by "E", and the allosteric coupling between the gating ring and voltage sensor from the Horrigan model is also "E". This makes things a bit confusing, but I'm not sure what can be done about it.

3) In Discussion (paragraph 1):

• "inducing additive rearrangements" should probably read "inducing energetically-additive rearrangements".

• " Further, the effects of Ca²⁺, Cd²⁺ and Ba²⁺ on the E values showed a voltage-dependent component, for which we could not provide a structural basis." This statement seems a little misleading; there seems to be some structural basis for voltage-dependence of Ca²⁺ and Cd²⁺ effects, because the RCK1 sight is physically somewhat close to the VSD (as the authors reason later on), and it is the Ba²⁺ effect that is more puzzling.

4) It seems like the interpretation of the Ba²⁺ effect may require more study, or more comment; if Ba²⁺ really is acting solely through the Ca²⁺ bowl, then the E-V relation observed with Ba and the S4 mutants (Figure 6D and F) should resemble the E-V relations with an RCK1 site knockout (D362A/D367A) combined with S4 mutation. Is this the case? If not, then this might argue against Ba²+ acting solely through the Ca²⁺ bowl, and suggest the Ba²⁺ is having other effects.

Reviewer #2:

This team has for several years been probing conformational changes in the cytosolic gating ring of the BK channel using sets of genetic encoded fluorescent tags to generate FRET signals that change with conformational status. Previous results had revealed an unexplained result that the FRET-reported signals exhibited voltage-dependence, despite the fact that the gating ring is not within the electric field across the membrane. This tended to contradict a standard model of BK activation in which coupling between voltage-sensors (VSDs) and the cytosolic domain (CTD) was thought to be rather weak. Furthermore, previous results from this team indicated that the two high affinity Ca²⁺ binding sites in the gating ring may behave differently in regards to voltage-sensitivity. The present paper nicely extends these earlier results with a number of functional tests that lead to the conclusion that, in the presence of Ca²⁺, the status of the voltage-sensors (VSDs) can influence the conformational status of the gating ring. Furthermore, this largely involves Ca²⁺ ligation at the RCK1 domain, but not the Ca²⁺ bowl. In sum, the authors find that the previously reported voltage-dependent conformational change of the BK gating ring induced by Ca²⁺-binding to the RCK1 high-affinity Ca²⁺ binding site is strongly correlated with the activation of the BK VSD. Manipulations that shift the activation-voltage relationship of the BK VSD, including mutations of key residues in the BK VSD or co-expression with β-1 accessory subunits, also shift the fluorescence-voltage relationship of the BK gating ring toward same direction. Manipulations that change other aspect of BK channel gating, such as F315A mutation that reduces the intrinsic BK gating equilibrium constant, the coupling between the BK VSD and BK gate, and the coupling the BK VSD and BK gating ring, or co-expression with the γ-1 subunit that enhances the coupling between the BK VSD and BK gate, do not significantly change the fluorescence-voltage relationship of the BK gating ring.

Overall, the paper is interesting, addresses an important, unresolved question, and is clearly presented (albeit with some suggestions below). There are some details on particular experiments which may be somewhat inconsistent with some earlier results and may require clarification. This particularly applies to some of the β subunit tests.

Guidance regarding status of voltage sensors under different conditions. Some readers might also wonder about what the voltage-sensor status is under various manipulations for which GVs and EVs are generated. Although explicit measurements of gating currents would be too much to ask, there are situations where it might be useful for the readers to be reminded whether VSDs are likely to be entirely resting or mostly activated, for example. The authors are essentially assuming that the E/V measurements are a surrogate for VSD activation, but unfortunately this is largely an untested assumption, although not unreasonable. For the F315A construct, the earlier Carrasquel, 2015 paper does provide comparable Q/V curves for human F380A, which support the view that the VSD equilibrium is not altered by this mutation. However, the Vh for activation of the human F380A construct in that paper at 100 μM Ca²⁺ is about 232 mV, much more positive than what is reported (less than +100 mV) in the present manuscript for mouse F315A. Reasons for this rather major discrepancy are not clear and not mentioned.

Are CTD and VSD totally uncoupled in absence of ligand? There is one aspect of the data that seems to pose an interesting limiting condition regarding the overall mechanism that the authors might consider discussing. Specifically, in the absence of Ca²⁺, any conformational change involving VSDs and pore-gate-domain (PGD) activation seems entirely uncoupled from the gating ring. It seems impressive that for every construct that is examined, at 0 Ca²⁺, even when full activation is shifted to potentials negative to +100 mV (e.g., with γ1), there is never any detectable FRET signal. This seems to say that Ca²⁺ is required for conformational coupling between VSD and CTD, or, to say it another way, that VSD movements do not couple to the CTD in the absence of Ca²⁺, or Ca²⁺ is obligatory for VSD-CTD coupling.

Is there a way to compare Ca²⁺-dependence of EV measurements to other approaches? Another interesting aspect of the data about which little is said is the rather astonishingly steep Ca²⁺-dependence of FRET signal generation at any single voltage. This seems consistent across constructs, and is reduced in the single site constructs with Cd²⁺ and Ba²⁺. Maybe this fits with what we know about Ca²⁺-dependence of activation, but it seems unusually steep. Although it may be difficult to extract from the E./V data, it seems like there is some information regarding apparent voltage-dependence of ligand affinity embedded in the data, that might be interesting to contrast to earlier results, e.g., from Sweet and Cox.

Various aspects of the β subunit experiments are confusing. First, in Figure 3D, the β3bNβ1 construct is producing no effect on GVs at all. Yet, in the Supplementary Information of Castillo et al., 2015, Vh is reported to shift negatively about 50 mV for presumably the same construct. Part of their argument was that the positive residues in the β1 N-terminus are responsible not only for the effect on voltage-sensors, but also the gating shift. Mutation of the K3K4 charged residues in the N-terminus abolishes the β1 gating shift, while mutation of K10R11 retained the gating shift.

Figure 3—figure supplement 1. Presumably the β3b construct is the human form, but this should be clarified, since the mouse form does not cause rapid inactivation. The results in Figure 3—figure supplement 1 are only partially supportive of the idea that adequate expression of β3b has been achieved. The main argument the authors make is apparently the roll-over in the GV relationship seen at 12 and 22 microM Ca²⁺. But it is very odd that this roll-over is not seen at 95 and 4 microM. The Materials and methods state that all GVs were generated from tail currents. Because the β3b inactivation is very low affinity, it has been shown that recovery from block is essentially complete at the peak of the usual tail currents, so tail current GVs are not particularly affected by this block (Xia et al., 2000. Figure 6C vs. 6D.). Yet steady-state conductance with adequate β3b expression shows marked block, which is really not so readily apparent in the example traces in panel a. The β3 subunit, of all β subunits, is known to be a bit challenging to express and the results here are not convincing in that regards. This is critical, when the observed results are really not any different from BK667CY α alone. The fact that no gating shift was observed with the chimeric β3bNβ1 construct also raises some concern in that regard. Positive controls that these β subunit constructs are being adequately expressed are lacking.

The scale bars in panel A of Figure 3—figure supplement 1 are also likely to be in error. And the time base of these recordings really doesn't provide any chance of seeing the distinctive features of patches with β3b containing channels.

Resolution of the issues raised above would impact on the discussion about these experiments presented in the subsection “The dynamics of the VSD are directly reflected in the gating ring conformation”.

Reviewer #3:

This study examines the interaction between transmembrane and cytoplasmic domains of BK channels by determining the effects of mutations in each domain, or co-assembly with regulatory subunits, on conformational changes in the cytoplasmic gating ring. The authors have previously demonstrated that gating ring conformation can be monitored by measuring FRET efficiency between CFP- and YFP-tagged Slo1 subunits. FRET efficiency is increased by the binding of Ca²⁺ and other divalent cations to two previously identified high-affinity binding sites, and at intermediate Ca²⁺ concentrations by membrane depolarization. The effect of voltage on FRET is consistent with the suggestion, most recently based on BK channel structure, that functional interactions occur between the RCK1 domain of the gating ring and VSD. Here, this hypothesis is supported by showing that (a) voltage-dependent FRET is abolished by mutating the RCK1 Ca²⁺ binding site and (b) changes in voltage-dependent FRET induced by mutations or regulatory subunits are correlated with VSD activation rather than channel opening.

The manuscript is well written and the data clearly described. However, the conclusions are not particularly novel and I have concerns about the lack of comparison to previous work, and the extent to which the results provide new insight.

The idea that functional interactions occur between the VSD and gating ring in BK channels has already been addressed by several previous studies such as Horrigan and Aldrich, 2002, Sweet and Cox, 2008, and Savalli et al., 2012. Only the first of these is referenced in the manuscript, and that one is mischaracterized. The authors seem to suggest that an interaction between VSD and Ca²⁺-binding sites was included in the Horrigan Aldrich model simply for completeness (i.e. because it couldn't be ruled out). In fact, that study observed clear evidence for an interaction based on the effects of Ca²⁺ on VSD activation (gating current). These results led to an estimate of the interaction energy between Ca²⁺- and V-sensors and the conclusion that the interaction was not altered by channel opening. In addition, Horrigan and Aldrich concluded that the interaction was weak, not because they couldn't measure it, but because the interaction energy was small compared to that between Ca²⁺- or V-sensors and channel opening. Sweet and Cox 2008 then tested a prediction that the interaction is bi-directional, by confirming that the apparent Kd for Ca²⁺-binding is reduced by altered by membrane depolarization. Furthermore, they concluded that the interaction occurred with RCK1 rather than RCK2, based on the effect of mutating the different Ca²⁺-binding sites. Finally, Savalli et al., 2012 used patch clamp fluorometry (on the VSD) to confirm that Ca²⁺ binding to the gating ring alters VSD activation. Thus, while the detailed approach used in the manuscript is new, neither the conclusions or even the use of patch clamp fluorometry appear to be novel. That said, the results could be valuable if they provide new insight into the nature of the interaction or its properties. But here too I have doubts. As the authors point out, the results show a strong correlation between effects on voltage-sensor activation and changes in the voltage-dependence of FRET. However, the results are described in very qualitative terms that don't seem to provide much insight beyond the basic conclusion. At minimum, I think the authors need to discuss whether the results are consistent or inconsistent with previous results, and make clear what is new.

One specific concern is whether the lack of effect of the γ1 subunit on FRET efficiency in Figure 2F is indeed consistent with the hypothesis that voltage-dependent FRET reflects VSD activation. The authors typically reference the effect of various mutations and regulatory subunits on the half-activation voltage of the VSD, V_h(j). Strictly speaking, V_h(j) is the half-activation voltage when channels are closed, and for most of the experiments may be an appropriate parameter to look at since voltage-dependent FRET is being measured under condition where a large fraction of channels are closed. However, in the case of γ1, voltage-dependent FRET is occurring over a range where channels are open (i.e. rel. conduct. In Figure 2F is saturated). While the authors are correct that Yan and Aldrich, 2010 proposed that V_h(j) is unaltered by γ1, their model predicts a large shift in VSD half activation voltage when channels are open (because when channels are open, VSD activation depends on VSD-gate coupling, which is enhance by γ1). Therefore, is seems like one should expect γ1 to shift v-dependent FRET to more negative voltages if it was simply detecting VSD activation.