Final Report Summary - VOLTSENS (Probing the sequence determinants of ion channel voltage sensing via computation: towards the design of custom-tailored voltage-sensing modules)
Voltage-sensing domains (VSD) are ubiquitous transmembrane protein domains that confer voltage sensitivity to proteic assemblies such as ion channels or phosphatases. Computational methods such as molecular dynamics (MD) simulations have proven that they can shed light on the molecular-level processes involved in voltage-sensing modules of voltage-gated channels. This project aimed to characterize fully the thermodynamic and kinetics of VSD activation, to identify of the molecular determinants that are essential to voltage sensing and discover the role played by these determinants on the various properties of the VSD function. Specifically, an emphasis was placed on distinguishing the determinants that are conserved amongst the superfamily of voltage sensing modules, and therefore constitute obligatory markers of the voltage-sensing function and the determinants that are specific to a VSD subfamily or even individual protein, conferring thereby specific voltage-gating properties.
The three main objectives of the VOLTSENS project were:
1- to produce a framework that will enable to characterize fully the thermodynamics and kinetics of activation/deactivation of VSDs of known structures.
2- to characterize the effect of the mutation of specific residues within the VSD module.
3- to produce a predictive modeling strategy to infer the effect of the mutation of the different VSD activation properties, using amongst other techniques evolutionary modeling.
VSDs are membrane-bound protein modules that transduces electrical signals into mechanical work. They sense changes in the transmembrane voltage and convert the electrical signal into a conformational change called activation, which consists in a reorganization of the membrane protein charges that is detected experimentally as transient currents. These so-called gating currents (of which the time-integral amounts to the gating charge) have been investigated extensively within the theoretical framework of so-called Discrete-state Markov Models (DMM), whereby activation is conceptualized as a series of transitions across a discrete set of states.
In the first part of this project, we have used atomistic level modeling and well-tempered metadynamics to calculate the configurational free energy along a single transition of the Kv1.2 VSD from first principles (Delemotte et al. 2015). We have shown that this transition is intrinsically multi-dimensional and described by a rough free-energy landscape. Remarkably, however, a coarse-grained description of the system, based on the use of the gating charge as reaction coordinate, has revealed a smooth profile with a single barrier, consistent with phenomenological models. Our results have shown that choosing the gating charge as reaction coordinate masks the topological complexity of the network of microstates participating in the transition. Because they can often not be transferred to other applications, however, the applicability of discrete state Markov models is often limited. Accordingly, we have designed a bottom-up strategy to compute directly gating currents from first principles: we first use molecular dynamics simulations to compute to the potential of mean force associated with the complete activation mechanism of the voltage sensor domain of Kv1.2 before following the time evolution of a charged particle along this potential of mean force profile. (Delemotte et al. 2016) By directly accessing complete VSD free energy profiles, our study bridges microscopic and macroscopic views of ion channel dynamics and confirms the critical role of rate-determining barriers controlling channel gating long suspected from electrophysiological experiments.
Using the wealth of data gathered in the MD simulations carried out in the first part of this project has enabled us to characterize better the role played by individual residues in the activation cycle and to therefore focus on mutations of physiological interest. Indeed, in collaboration with the Chahine lab, CIURSMQ, Quebec, Canada, we have characterized the effect of specific VSD residues mutations implicated in cardiac pathologies and the appearance of anomalous leak currents through the VSD responsible for the pathologic phenotype (Moreau et al. 2015). We have also turned our attention to a member of the VSD family, the voltage-gated proton channel (Hv1), in which several S4 arginines are endogenously mutated to uncharged residues and whose role is to transport protons across the membrane in a voltage-dependent manner. We have used structural modeling and MD simulations to characterize the most relevant channel conformations along the activation cycle and then performed computational docking of a known Hv1 inhibitor, 2-gunaidiniumbenzimidazole (2GBI). We have shown that water molecules localize in intermittent hydrogen-bonded clusters that are replaced by potential drug moieties upon 2GBI binding. The entropic gain resulting from releasing these tightly restrained waters to the bulk solvent is likely a major contributor to the binding free energy (Gianti et al. 2016). We have also investigated the proton transport mechanism in Hv1 via a multiscale approach. We have shown that protons localize in three binding sites along the channel lumen, formed by three pairs of conserved negatively charged residues lining the pore: D174/E153, D112/D185 and E119/D123. Local rearrangements allow protons to hop from one acidic residue to the next through a bridging water molecule. This mechanism challenges the more traditional view of Hv1 proton transport hopping along a pre-oriented water file, so-called Grotthuss hopping (van Keulen et al. 2016).
Finally, in the third part of the project, we have investigated sequence-function relationships in VSDs with using information theory and probabilistic modeling, using the notion according to which VSDs can be envisioned as modular biomolecular machines that transduce electrical signals in cells through a highly-conserved activation mechanism (Palovcak et al. 2014). Specifically, we have collected over 6600 unique VSD sequences from diverse, long-diverged phylogenetic lineages and related the statistical properties of this ensemble to functional constraints imposed by evolution and have applied an information-theoretic approach to describe site-specific frequency distributions of residues and built a probabilistic model accounting for both site-specific residue propensity and pairwise residue-residue statistical couplings. In both cases, we have found that evolutionary constraints are exerted on specific sets of residues or pairs of residues, suggesting a sparse network of residue-residue interactions in which evolution attends to only certain nodes and edges. For the first time, we have provided a comprehensive picture of the design principles of the VSD: (i) state-dependent coevolving contacts between a conserved position on S2 and another one on S4 tune the relative stability of the resting state; (ii) absence of specific, tuned interactions along the hydrophobic interfaces of S4 and neighboring S1 and S3 helices enables S4 translation; (iii) networks of hydrophilic and hydrophobic residues lining the lumen of the VSD tune the access of penetrating water and gating charges; and (iv) conserved residue positions, such as the counter-charges or selected hydrophobic position facing the lumen of the VSD are surrounded by coevolving networks of hydrophobic residues stabilizing their orientation. With the publication of the high resolution structures of two non voltage gated channels (TRPV1 and TRPA1) of the same superfamily as voltage gated potassium and sodium channels, it has become extremely interesting to compare the features of the ensemble of sequences of Kv and TRP channels. The transient receptor potential (TRP) channel superfamily plays a central role in the transduction of diverse sensory stimuli in eukaryotes. Though dissimilar in sequence and domain organization, all known TRP channels serve as polymodal cellular sensors and form tetrameric assemblies similar to their distant relatives, the voltage-gated potassium (Kv) channels. Is the allosteric mechanism underlying polymodal gating shared amongst TRP channels and how does this mechanism differ from that underpinning voltage sensitivity in Kv channels? To provide insights into these questions, we have performed comparative sequence analysis on large, comprehensive ensembles of TRP and Kv channel sequences, contextualizing the patterns of conservation and correlation observed in the TRP channel sequences in light of the well-studied Kv channels (Palovcak et al. 2015). We report sequence features that are specific to TRP channels and, thanks to the insight from recent TRPV1 structures, we suggest a model of TRP channel gating.
The three main objectives of the VOLTSENS project were:
1- to produce a framework that will enable to characterize fully the thermodynamics and kinetics of activation/deactivation of VSDs of known structures.
2- to characterize the effect of the mutation of specific residues within the VSD module.
3- to produce a predictive modeling strategy to infer the effect of the mutation of the different VSD activation properties, using amongst other techniques evolutionary modeling.
VSDs are membrane-bound protein modules that transduces electrical signals into mechanical work. They sense changes in the transmembrane voltage and convert the electrical signal into a conformational change called activation, which consists in a reorganization of the membrane protein charges that is detected experimentally as transient currents. These so-called gating currents (of which the time-integral amounts to the gating charge) have been investigated extensively within the theoretical framework of so-called Discrete-state Markov Models (DMM), whereby activation is conceptualized as a series of transitions across a discrete set of states.
In the first part of this project, we have used atomistic level modeling and well-tempered metadynamics to calculate the configurational free energy along a single transition of the Kv1.2 VSD from first principles (Delemotte et al. 2015). We have shown that this transition is intrinsically multi-dimensional and described by a rough free-energy landscape. Remarkably, however, a coarse-grained description of the system, based on the use of the gating charge as reaction coordinate, has revealed a smooth profile with a single barrier, consistent with phenomenological models. Our results have shown that choosing the gating charge as reaction coordinate masks the topological complexity of the network of microstates participating in the transition. Because they can often not be transferred to other applications, however, the applicability of discrete state Markov models is often limited. Accordingly, we have designed a bottom-up strategy to compute directly gating currents from first principles: we first use molecular dynamics simulations to compute to the potential of mean force associated with the complete activation mechanism of the voltage sensor domain of Kv1.2 before following the time evolution of a charged particle along this potential of mean force profile. (Delemotte et al. 2016) By directly accessing complete VSD free energy profiles, our study bridges microscopic and macroscopic views of ion channel dynamics and confirms the critical role of rate-determining barriers controlling channel gating long suspected from electrophysiological experiments.
Using the wealth of data gathered in the MD simulations carried out in the first part of this project has enabled us to characterize better the role played by individual residues in the activation cycle and to therefore focus on mutations of physiological interest. Indeed, in collaboration with the Chahine lab, CIURSMQ, Quebec, Canada, we have characterized the effect of specific VSD residues mutations implicated in cardiac pathologies and the appearance of anomalous leak currents through the VSD responsible for the pathologic phenotype (Moreau et al. 2015). We have also turned our attention to a member of the VSD family, the voltage-gated proton channel (Hv1), in which several S4 arginines are endogenously mutated to uncharged residues and whose role is to transport protons across the membrane in a voltage-dependent manner. We have used structural modeling and MD simulations to characterize the most relevant channel conformations along the activation cycle and then performed computational docking of a known Hv1 inhibitor, 2-gunaidiniumbenzimidazole (2GBI). We have shown that water molecules localize in intermittent hydrogen-bonded clusters that are replaced by potential drug moieties upon 2GBI binding. The entropic gain resulting from releasing these tightly restrained waters to the bulk solvent is likely a major contributor to the binding free energy (Gianti et al. 2016). We have also investigated the proton transport mechanism in Hv1 via a multiscale approach. We have shown that protons localize in three binding sites along the channel lumen, formed by three pairs of conserved negatively charged residues lining the pore: D174/E153, D112/D185 and E119/D123. Local rearrangements allow protons to hop from one acidic residue to the next through a bridging water molecule. This mechanism challenges the more traditional view of Hv1 proton transport hopping along a pre-oriented water file, so-called Grotthuss hopping (van Keulen et al. 2016).
Finally, in the third part of the project, we have investigated sequence-function relationships in VSDs with using information theory and probabilistic modeling, using the notion according to which VSDs can be envisioned as modular biomolecular machines that transduce electrical signals in cells through a highly-conserved activation mechanism (Palovcak et al. 2014). Specifically, we have collected over 6600 unique VSD sequences from diverse, long-diverged phylogenetic lineages and related the statistical properties of this ensemble to functional constraints imposed by evolution and have applied an information-theoretic approach to describe site-specific frequency distributions of residues and built a probabilistic model accounting for both site-specific residue propensity and pairwise residue-residue statistical couplings. In both cases, we have found that evolutionary constraints are exerted on specific sets of residues or pairs of residues, suggesting a sparse network of residue-residue interactions in which evolution attends to only certain nodes and edges. For the first time, we have provided a comprehensive picture of the design principles of the VSD: (i) state-dependent coevolving contacts between a conserved position on S2 and another one on S4 tune the relative stability of the resting state; (ii) absence of specific, tuned interactions along the hydrophobic interfaces of S4 and neighboring S1 and S3 helices enables S4 translation; (iii) networks of hydrophilic and hydrophobic residues lining the lumen of the VSD tune the access of penetrating water and gating charges; and (iv) conserved residue positions, such as the counter-charges or selected hydrophobic position facing the lumen of the VSD are surrounded by coevolving networks of hydrophobic residues stabilizing their orientation. With the publication of the high resolution structures of two non voltage gated channels (TRPV1 and TRPA1) of the same superfamily as voltage gated potassium and sodium channels, it has become extremely interesting to compare the features of the ensemble of sequences of Kv and TRP channels. The transient receptor potential (TRP) channel superfamily plays a central role in the transduction of diverse sensory stimuli in eukaryotes. Though dissimilar in sequence and domain organization, all known TRP channels serve as polymodal cellular sensors and form tetrameric assemblies similar to their distant relatives, the voltage-gated potassium (Kv) channels. Is the allosteric mechanism underlying polymodal gating shared amongst TRP channels and how does this mechanism differ from that underpinning voltage sensitivity in Kv channels? To provide insights into these questions, we have performed comparative sequence analysis on large, comprehensive ensembles of TRP and Kv channel sequences, contextualizing the patterns of conservation and correlation observed in the TRP channel sequences in light of the well-studied Kv channels (Palovcak et al. 2015). We report sequence features that are specific to TRP channels and, thanks to the insight from recent TRPV1 structures, we suggest a model of TRP channel gating.