A binding site for phosphoinositide modulation of voltage-gated sodium channels described by multiscale simulations

Reviewer #1 (Public Review):

Summary: Here, the authors were attempting to use molecular simulation or probe the nature of how lipids, especially PIP lipids, bind to a medically-important ion channel. In particular, they look at how this binding impact the function of the channel.

Strengths: the study is very well written and composed. The techniques are used appropriately, with plenty of sampling and analysis. The findings are compelling and provide clear insights into the biology of the system.

Weaknesses: a few of the analyses are hard to understand/follow, and rely on "in house" scripts. This is particularly the case for the lipid binding events, which can be difficult to compute accurately. Additionally, a lack of experimental validation, or coupling to existing experimental data, limits the study.

It is my view that the authors have achieved their aims, and their findings are compelling and believable. Their findings should have impacts on how researchers understand the functioning of the Nav1.4 channel, as well as on the study of other ion channels and how they interact with membrane lipids.

Reviewer #2 (Public Review):

Summary:
Y., Tao E., et al. used multiscale MD simulations to show that PI(4,5)P2 binds stably to an inactivated state of Nav channels at a conserved site within the DIV S4-S5 linker, which couples the voltage sensing domain (VSD) to the pore. The authors hypothesized that PI(4,5)P2 prolongs inactivation by binding to the same site where the C-terminal tail is proposed to bind during recovery from inactivation. They convincingly showed that PI(4,5)P2 reduces the mobility of both the DIV S4-S5 linker and the DIII-IV linker, thus slowing the conformational changes required for the channel to recover to the resting state. They also conducted MD simulations to show that phosphoinositides bind to VSD gating charges in the resting state of Nav channels. These interactions may anchor VDS at the resting state and impede its activation. Their results provide a mechanism by which phosphoinositides alter the voltage dependence of activation and the recovery rate from inactivation, an important step for developing novel therapies to treat Nav-related diseases. However, the study is incomplete and lacks the expected confirmatory studies which are relevant to such proposals.

Strengths:
The authors identified a novel binding between phosphoinositides and the VSD of Nav and showed that the strength of this interaction is state-dependent. Based on their work, the affinity of PIPs to the inactivated state is higher than the resting state. This work will help pave the way for designing novel therapeutics that may help relieve pain or treat diseases like arrhythmia, which may result from a leftward shift of the channel's activation.

Weaknesses:
However, the study lacks the expected confirmatory studies which are relevant to such proposals. For example, one would expect that the authors would mutate the positive residues that they claim to make interactions with phosphoinositides to show that there are much fewer interactions once they make these mutations. Another point is that the authors found that the main interaction site of PIPs with Nav1.4 is the VSD-DIV and DIII-DIV linker, an interaction that is expected to delay fast inactivation if it happens at the resting state. The authors should make a resting state model of the Nav1.4 channel to explain the recent experimental data showing that PIP2 delays the activation of Nav1.4, with almost no effect on the voltage dependence of fast inactivation.

Major concern:
1- Lack of confirmatory experiments, e.g., mutating the positive residues that show a high affinity towards PIPs to a neutral and negative residue and assessing the effect of mutagenesis on binding.
2- Nav1.4 is the only channel that has been studied in terms of the effect of PIPs on it, therefore the authors should build a resting state model of Nav1.4 and study the effect of PIPs on it.
Minor points:

There are a lot of incorrect statements in many areas, e.g., "These diseases 335 are associated with accelerated rates of channel recovery from inactivation, consistent with our observations that an interaction between PI(4,5)P2 and the residue corresponding to R1469 in other Nav 337 subtypes could be important for prolonging the fast-inactivated state." Prolonging the fast inactivated state would actually reduce recovery from inactivation and not accelerate it.

Reviewer #3 (Public Review):

Summary:
This work uses multiscale molecular dynamics simulations to demonstrate molecular mechanism(s) for phosphatidylinositol regulation of voltage gated sodium channel (Nav1.4) gating. Recent experimental work by Gada et al. JGP 2023 showed altered Nav1.4 gating when Nav1.4 current was recorded with simultaneous application of PI(4,5)P2 dephosphorylate. Here the authors revealed probable molecular mechanism that can explain PI(4,5)P2 modulation of Nav1.4 gating. They found PIP lipids interacting with the gating charges - potentially making it harder to move the voltage sensor domain and altering the channels voltage sensitivity. They also found a stable PIP binding site that reaches the D_IV S4-S5 linker, reducing the mobility of the linker and potentially competing with the C-terminal domain.

Strengths:
Using multiscale simulations with course-grained simulations to capture lipid-protein interactions and the overall protein lipid fingerprint and then all-atom simulations to verify atomistic details for specific lipid-protein interactions is extremely appropriate for the question at hand. Overall, the types of simulation and their length are suitable for the questions the authors pose and a thorough set of analysis was done which illustrates the observed PIP-protein interactions.

Weaknesses:
Although the set of current simulations and analysis supports the conclusions drawn nicely, there are some limitations imposed by the authors on the course-grained simulations. If those were not imposed, it would have allowed for an even richer set and more thorough exploration of the protein-lipid interactions. The Martini 2 force field indeed cannot change secondary structure but if run with a properly tuned elastic network instead of backbone restraints, the change in protein configuration can be sampled and/or some adaptation of the protein to the specific protein environment can be observed. Additionally, with the 4to1 heavy atoms to a bead mapping some detailed chemical specificity is averaged out but parameters for different PIP family members do exist - including specific PIP(4,5)P2 vs PIP(3,4)P2, and could have been explored.

Author Response

We thank the reviewers for their helpful comments and thorough assessment of our manuscript which will allow us to improve the work in a subsequent revision. Many suggestions, such as mutating residues to help validate the proposed site will be included in a future revision. Below we clarify three aspects that led to confusion in the initial review

The comment of reviewer 2 that “... the main interaction site of PIPs with Nav1.4 is the VSD-DIV and DIII-DIV linker, an interaction that is expected to delay fast inactivation if it happens at the resting state." is true. However, as explained in our manuscript (Fig. 7), we don’t expect binding at this position to happen in the resting state as the C-terminal domain is bound to this region, impeding PIP binding.

Reviewer 2 also suggests that we produce a resting state model of Nav1.4 to replace/supplement the results we obtained using our resting Nav1.7 model. We chose to model Nav1.7 due to the availability of structures with different VSDs in the deactivated conformation, something that is not true for Nav1.4. While we plan to explore a Nav1.4 resting state based on the reviewer's suggestion, we note that this introduces an extra layer of uncertainty. However, due to sequence conservation of the gating charges and proposed binding site residues between Nav subtypes, we propose very similar modes of PIP binding among the Nav subtypes across the different conformations.

Finally, we strongly disagree with the reviewer’s assessment that there are ‘There are a lot of incorrect statements in many areas’ and this may have come from a misreading of the mentioned sentence. The sentence in question reads "These diseases 335 are associated with accelerated rates of channel recovery from inactivation, consistent with our observations that an interaction between PI(4,5)P2 and the residue corresponding to R1469 in other Nav 337 subtypes could be important for prolonging the fast-inactivated state." To which the reviewer 2 states ‘Prolonging the fast inactivated state would actually reduce recovery from inactivation and not accelerate it.’ The statement quoted is not incorrect – from the original experiments we know that the presence of PIP prolongs the time spent in the fast inactivated state. Mutations at the PIP binding site are likely to reduce PIP binding, and with less PIP present the channel will recover from inactivation more quickly. We appreciate that this sentence could be reworded for clarity and will address this in our revision to prevent such misreading.