Computational methods of studying the binding of toxins from venomous animals to biological ion channels: theory and applications.

The discovery of new drugs that selectively block or modulate ion channels has great potential to provide new treatments for a host of conditions. One promising avenue revolves around modifying or mimicking certain naturally occurring ion channel modulator toxins. This strategy appears to offer the prospect of designing drugs that are both potent and specific. The use of computational modeling is crucial to this endeavor, as it has the potential to provide lower cost alternatives for exploring the effects of new compounds on ion channels. In addition, computational modeling can provide structural information and theoretical understanding that is not easily derivable from experimental results. In this review, we look at the theory and computational methods that are applicable to the study of ion channel modulators. The first section provides an introduction to various theoretical concepts, including force-fields and the statistical mechanics of binding. We then look at various computational techniques available to the researcher, including molecular dynamics, brownian dynamics, and molecular docking systems. The latter section of the review explores applications of these techniques, concentrating on pore blocker and gating modifier toxins of potassium and sodium channels. After first discussing the structural features of these channels, and their modes of block, we provide an in-depth review of past computational work that has been carried out. Finally, we discuss prospects for future developments in the field.

Inhibition of Voltage-Gated K+ Channel Kv1.5 by Antiarrhythmic Drugs.

Molecular dynamics simulations are employed to determine the inhibitory mechanisms of three drugs, 5-(4-phenoxybutoxy)psoralen (PAP-1), vernakalant, and flecainide, on the voltage-gated K+ channel Kv1.5, a target for the treatment of cardiac arrhythmia. At neutral pH, PAP-1 is neutral, whereas the other two molecules carry one positive charge. We show that PAP-1 forms stable dimers in water, primarily through hydrophobic interactions between aromatic rings. All three molecules bind to the cavity between the Ile508 and Val512 residues from the four subunits of the channel. Once bound, the drug molecules are flexible, with the average root-mean-square fluctuation being between 2 and 3 Å, which is larger than the radius of gyration of a bulky amino acid. The presence of a monomeric PAP-1 causes the permeating K+ ion to dehydrate, thereby creating a significant energy barrier. In contrast, vernakalant blocks the ion permeation primarily via an electrostatic mechanism and, therefore, must be in the protonated and charged form to be effective.

Molecular dynamics simulations of Na(+) and leucine transport by LeuT.

Molecular dynamics simulations are used to gain insight into the binding of Na(+) and leucine substrate to the bacterial amino acid transporter LeuT, focusing on the crystal structures of LeuT in the outward-open and inward-open states. For both conformations of LeuT, a third Na(+) binding site involving Glu290 in addition to the two sites identified from the crystal structures is observed. Once the negative charge from Glu290 in the inward-open LeuT is removed, the ion bound to the third site is ejected from LeuT rapidly, suggesting that the protonation state of Glu290 regulates Na(+) binding and release. In Cl(-)-dependent transporters where Glu290 is replaced by a neutral serine, a Cl(-) ion would be required to replace the role of Glu290. Thus, the simulations provide insights into understanding Na(+) and substrate transport as well as Cl(-)-independence of LeuT.

Mechanisms and energetics of potassium channel block by local anesthetics and antifungal agents.

Many drug molecules inhibit the conduction of several families of cation channels by binding to a small cavity just below the selectivity filter of the channel protein. The exact mechanisms governing drug-channel binding and the subsequent inhibition of conduction are not well understood. Here the inhibition of two K(+) channel isoforms, Kv1.2 and KCa3.1, by two drug molecules, lidocaine and TRAM-34, is examined in atomic detail using molecular dynamics simulations. A conserved valine-alanine-valine motif in the inner cavity is found to be crucial for drug binding in both channels, consistent with previous studies of similar systems. Potential of mean force calculations show that lidocaine in its charged form creates an energy barrier of ∼6 kT for a permeating K(+) ion when the ion is crossing over the drug, while the neutral form of lidocaine has no significant effect on the energetics of ion permeation. On the other hand, TRAM-34 in the neutral form is able to create a large energy barrier of ∼10 kT by causing the permeating ion to dehydrate. Our results suggest that TRAM-34 analogues that remain neutral and permeable to membranes under acidic conditions common to inflammation may act as possible drug scaffolds for combating local anesthetic failure in inflammation.

Conductance properties of the inwardly rectifying channel, Kir3.2: molecular and Brownian dynamics study.

Using the recently unveiled crystal structure, and molecular and Brownian dynamics simulations, we elucidate several conductance properties of the inwardly rectifying potassium channel, Kir3.2, which is implicated in cardiac and neurological disorders. We show that the pore is closed by a hydrophobic gating mechanism similar to that observed in Kv1.2. Once open, potassium ions move into, but not out of, the cell. The asymmetrical current-voltage relationship arises from the lack of negatively charged residues at the narrow intracellular mouth of the channel. When four phenylalanine residues guarding the intracellular gate are mutated to glutamate residues, the channel no longer shows inward rectification. Inward rectification is restored in the mutant Kir3.2 when it becomes blocked by intracellular Mg(2+). Tertiapin, a polypeptide toxin isolated from the honey bee, is known to block several subtypes of the inwardly rectifying channels with differing affinities. We identify critical residues in the toxin and Kir3.2 for the formation of the stable complex. A lysine residue of tertiapin protrudes into the selectivity filter of Kir3.2, while two other basic residues of the toxin form hydrogen bonds with acidic residues located just outside the channel entrance. The depth of the potential of mean force encountered by tertiapin is -16.1kT, thus indicating that the channel will be half-blocked by 0.4µM of the toxin.

Mechanism of tetrodotoxin block and resistance in sodium channels.

Tetrodotoxin (TTX) has been used for many decades to characterize the structure and function of biological ion channels. Yet, the precise mechanism by which TTX blocks voltage-gated sodium (NaV) channels is not fully understood. Here molecular dynamics simulations are used to elucidate how TTX blocks mammalian voltage-gated sodium (Nav) channels and why it fails to be effective for the bacterial sodium channel, NaVAb. We find that, in NaVAb, a sodium ion competes with TTX for the binding site at the extracellular end of the filter, thus reducing the blocking efficacy of TTX. Using a model of the skeletal muscle channel, NaV1.4, we show that the conduction properties of the channel observed experimentally are faithfully reproduced. We find that TTX occludes the entrance of NaV1.4 by forming a network of hydrogen-bonds at the outer lumen of the selectivity filter. The guanidine group of TTX adopts a lateral orientation, rather than pointing into the filter as proposed previously. The acidic residues just above the selectivity filter are important in stabilizing the hydrogen-bond network between TTX and NaV1.4. The effect of two single mutations of a critical tyrosine residue in the filter of NaV1.4 on TTX binding observed experimentally is reproduced using computational mutagenesis.

Multi-ion versus single-ion conduction mechanisms can yield current rectification in biological ion channels.

There is clear evidence that the net magnitude of negative charge at the intracellular end of inwardly rectifying potassium channels helps to generate an asymmetry in the magnitude of the current that will pass in each direction. However, a complete understanding of the physical mechanism that links these charges to current rectification has yet to be obtained. Using Brownian dynamics, we compare the conduction mechanism and binding sites in rectifying and non-rectifying channel models. We find that in our models, rectification is a consequence of asymmetry in the hydrophobicity and charge of the pore lining. As a consequence, inward conduction can occur by a multi-ion conduction mechanism. However, outward conduction is restricted, since there are fewer ions at the intracellular entrance and outwardly moving ions must cross the pore on their own. We pose the question as to whether the same mechanism could be at play in inwardly rectifying potassium channels.

Molecular dynamics simulations of scorpion toxin recognition by the Ca(2+)-activated potassium channel KCa3.1.

The Ca(2+)-activated channel of intermediate-conductance (KCa3.1) is a target for antisickling and immunosuppressant agents. Many small peptides isolated from animal venoms inhibit KCa3.1 with nanomolar affinities and are promising drug scaffolds. Although the inhibitory effect of peptide toxins on KCa3.1 has been examined extensively, the structural basis of toxin-channel recognition has not been understood in detail. Here, the binding modes of two selected scorpion toxins, charybdotoxin (ChTx) and OSK1, to human KCa3.1 are examined in atomic detail using molecular dynamics (MD) simulations. Employing a homology model of KCa3.1, we first determine conduction properties of the channel using Brownian dynamics and ascertain that the simulated results are in accord with experiment. The model structures of ChTx-KCa3.1 and OSK1-KCa3.1 complexes are then constructed using MD simulations biased with distance restraints. The ChTx-KCa3.1 complex predicted from biased MD is consistent with the crystal structure of ChTx bound to a voltage-gated K(+) channel. The dissociation constants (Kd) for the binding of both ChTx and OSK1 to KCa3.1 determined experimentally are reproduced within fivefold using potential of mean force calculations. Making use of the knowledge we gained by studying the ChTx-KCa3.1 complex, we attempt to enhance the binding affinity of the toxin by carrying out a theoretical mutagenesis. A mutant toxin, in which the positions of two amino acid residues are interchanged, exhibits a 35-fold lower Kd value for KCa3.1 than that of the wild-type. This study provides insight into the key molecular determinants for the high-affinity binding of peptide toxins to KCa3.1, and demonstrates the power of computational methods in the design of novel toxins.

Conduction and block of inward rectifier K+ channels: predicted structure of a potent blocker of Kir2.1.

Dysfunction of Kir2.1, thought to be the major component of inward currents, I(K1), in the heart, has been linked to various channelopathies, such as short Q-T syndrome. Unfortunately, currently no known blockers of Kir2.x channels exist. In contrast, Kir1.1b, predominantly expressed in the kidney, is potently blocked by an oxidation-resistant mutant of the honey bee toxin tertiapin (tertiapin-Q). Using various computational tools, we show that both channels are closed by a hydrophobic gating mechanism and inward rectification occurs in the absence of divalent cations and polyamines. We then demonstrate that tertiapin-Q binds to the external vestibule of Kir1.1b and Kir2.1 with K(d) values of 11.6 nM and 131 µM, respectively. We find that a single mutation of tertiapin-Q increases the binding affinity for Kir2.1 by 5 orders of magnitude (K(d) = 0.7 nM). This potent blocker of Kir2.1 may serve as a structural template from which potent compounds for the treatment of various diseases mediated by this channel subfamily, such as cardiac arrhythmia, can be developed.

Complex structures between the N-type calcium channel (CaV2.2) and ω-conotoxin GVIA predicted via molecular dynamics.

The N-type voltage-gated Ca(2+) channel CaV2.2 is one of the important targets for pain management. ω-Conotoxins isolated from venoms of cone snails, which specifically inhibit CaV2.2, are promising scaffolds for novel analgesics. The inhibitory action of ω-conotoxins on CaV2.2 has been examined experimentally, but the modes of binding of the toxins to this and other related subfamilies of Ca(2+) channels are not understood in detail. Here molecular dynamics simulations are used to construct models of ω-conotoxin GVIA in complex with a homology model of the pore domain of CaV2.2. Three different binding modes in which the side chain of Lys2, Arg17, or Lys24 from the toxin protrudes into the selectivity filter of CaV2.2 are considered. In all the modes, the toxin forms a salt bridge with an aspartate residue of subunit II just above the EEEE ring of the selectivity filter. Using the umbrella sampling technique and potential of mean force calculations, the half-maximal inhibitory concentration (IC50) values are calculated to be 1.5 and 0.7 nM for the modes in which Lys2 and Arg17 occlude the ion conduction pathway, respectively. Both IC50 values compare favorably with the values of 0.04-1.0 nM determined experimentally. The similar IC50 values calculated for the different binding modes demonstrate that GVIA can inhibit CaV2.2 with alternative binding modes. Such a multiple-binding mode mechanism may be common for ω-conotoxins.

Binding modes of µ-conotoxin to the bacterial sodium channel (NaVAb).

Polypeptide toxins isolated from the venom of cone snails, known as µ-conotoxins, block voltage-gated sodium channels by physically occluding the ion-conducting pathway. Using molecular dynamics, we show that one subtype of µ-conotoxins, PIIIA, effectively blocks the bacterial voltage-gated sodium channel Na(V)Ab, whose crystal structure has recently been elucidated. The spherically shaped toxin, carrying a net charge of +6 e with six basic residues protruding from its surface, is attracted by the negatively charged residues on the vestibular wall and the selectivity filter of the channel. The side chain of each of these six arginine and lysine residues can wedge into the selectivity filter, whereas the side chains of other basic residues form electrostatic complexes with two acidic residues on the channel. We construct the profile of potential of mean force for the unbinding of PIIIA from the channel, and predict that PIIIA blocks the bacterial sodium channel with subnanomolar affinity.

Engineering a potent and specific blocker of voltage-gated potassium channel Kv1.3, a target for autoimmune diseases.

A polypeptide toxin extracted from scorpion venom, OSK1, is modified such that its potency is drastically enhanced in blocking one class of voltage-gated potassium channels, Kv1.3, which is a pharmacological target for immunosuppressive therapy. The bound complex of Kv1.3 and OSK1 reveals that one lysine residue of the toxin is in the proximity of another lysine residue on the external vestibule of the channel, just outside of the selectivity filter. This unfavorable electrostatic interaction is eliminated by interchanging the positions of two amino acids in the toxin. The potentials of mean force of the wild-type and mutant OSK1 bound to Kv1.1-Kv1.3 channels are constructed using molecular dynamics, and the half-maximal inhibitory concentration (IC(50)) of each toxin-channel complex is computed. We show that the IC(50) values predicted for three toxins and three channels match closely with experiment. Kv1.3 is half-blocked by 0.2 pM mutant OSK1; it is >10000-fold more specific for this channel than for Kv1.1 and Kv1.2.

Binding modes and functional surface of anti-mammalian scorpion α-toxins to sodium channels.

Scorpion α-toxins bind to the voltage-sensing domains of voltage-gated sodium (Na(V)) channels and interfere with the inactivation mechanisms. The functional surface of α-toxins has been shown to contain an NC-domain consisting of the five-residue turn (positions 8-12) and the C-terminus (positions 56-64) and a core-domain centered on the residue 18. The NC- and core-domains are interconnected by the linker-domain (positions 8-18). Here with atomistic molecular dynamics simulations, we examine the binding modes between two α-toxins, the anti-mammalian AahII and the anti-insect LqhαIT, and the voltage-sensing domain of rat Na(V)1.2, a subtype of Na(V) channels expressed in nerve cells. Both toxins are docked to the extracellular side of the voltage-sensing domain of Na(V)1.2 using molecular dynamics simulations, with the linker-domain assumed to wedge into the binding pocket. Several salt bridges and hydrophobic clusters are observed to form between the NC- and core-domains of the toxins and Na(V)1.2 and stabilize the toxin-channel complexes. The binding modes predicted are consistent with available mutagenesis data and can readily explain the relative affinities of AahII and LqhαIT for Na(V)1.2. The dissociation constants for the two toxin-channel complexes are derived, which compare favorably with experiment. Our models demonstrate that the functional surface of anti-mammalian scorpion α-toxins is centered on the linker-domain, similar to that of ß-toxins.

Computational design of a carbon nanotube fluorofullerene biosensor.

Carbon nanotubes offer exciting opportunities for devising highly-sensitive detectors of specific molecules in biology and the environment. Detection limits as low as 10(-11) M have already been achieved using nanotube-based sensors. We propose the design of a biosensor comprised of functionalized carbon nanotube pores embedded in a silicon-nitride or other membrane, fluorofullerene-Fragment antigen-binding (Fab fragment) conjugates, and polymer beads with complementary Fab fragments. We show by using molecular and stochastic dynamics that conduction through the (9, 9) exohydrogenated carbon nanotubes is 20 times larger than through the Ion Channel Switch ICS(TM) biosensor, and fluorofullerenes block the nanotube entrance with a dissociation constant as low as 37 pM. Under normal operating conditions and in the absence of analyte, fluorofullerenes block the nanotube pores and the polymer beads float around in the reservoir. When analyte is injected into the reservoir the Fab fragments attached to the fluorofullerene and polymer bead crosslink to the analyte. The drag of the much larger polymer bead then acts to pull the fluorofullerene from the nanotube entrance, thereby allowing the flow of monovalent cations across the membrane. Assuming a tight seal is formed between the two reservoirs, such a biosensor would be able to detect one channel opening and thus one molecule of analyte making it a highly sensitive detection design.

Permeation and block of the Kv1.2 channel examined using brownian and molecular dynamics.

Using both Brownian and molecular dynamics, we replicate many of the salient features of Kv1.2, including the current-voltage-concentration profiles and the binding affinity and binding mechanisms of charybdotoxin, a scorpion venom. We also elucidate how structural differences in the inner vestibule can give rise to significant differences in its permeation characteristics. Current-voltage-concentration profiles are constructed using Brownian dynamics simulations, based on the crystal structure 2A79. The results are compatible with experimental data, showing similar conductance, rectification, and saturation with current. Unlike KcsA, for example, the inner pore of Kv1.2 is mainly hydrophobic and neutral, and to explore the consequences of this, we investigate the effect of mutating neutral proline residues at the mouth of the inner vestibule to charged aspartate residues. We find an increased conductance, less inward rectification, and quicker saturation of the current-voltage profile. Our simulations use modifications to our Brownian dynamics program that extend the range of channels that can be usefully modeled. Using molecular dynamics, we investigate the binding of the charybdotoxin scorpion venom to the outer vestibule of the channel. A potential of mean force is derived using umbrella sampling, giving a dissociation constant within a factor of ∼2 to experimentally derived constants. The residues involved in the toxin binding are in agreement with experimental mutagenesis studies. We thus show that the experimental observations on the voltage-gated channel, including the toxin-channel interaction, can reliably be replicated by using the two widely used computational tools.

Modeling the binding of three toxins to the voltage-gated potassium channel (Kv1.3).

The conduction properties of the voltage-gated potassium channel Kv1.3 and its modes of interaction with several polypeptide venoms are examined using Brownian dynamics simulations and molecular dynamics calculations. Employing an open-state homology model of Kv1.3, we first determine current-voltage and current-concentration curves and ascertain that simulated results accord with experimental measurements. We then investigate, using a molecular docking method and molecular dynamics simulations, the complexes formed between the Kv1.3 channel and several Kv-specific polypeptide toxins that are known to interfere with the conducting mechanisms of several classes of voltage-gated K(+) channels. The depths of potential of mean force encountered by charybdotoxin, α-KTx3.7 (also known as OSK1) and ShK are, respectively, -19, -27, and -25 kT. The dissociation constants calculated from the profiles of potential of mean force correspond closely to the experimentally determined values. We pinpoint the residues in the toxins and the channel that are critical for the formation of the stable venom-channel complexes.

Synthetic cation-selective nanotube: permeant cations chaperoned by anions.

The ability to design ion-selective, synthetic nanotubes which mimic biological ion channels may have significant implications for the future treatment of bacteria, diseases, and as ultrasensitive biosensors. We present the design of a synthetic nanotube made from carbon atoms that selectively allows monovalent cations to move across and rejects all anions. The cation-selective nanotube mimics some of the salient properties of biological ion channels. Before practical nanodevices are successfully fabricated it is vital that proof-of-concept computational studies are performed. With this in mind we use molecular and stochastic dynamics simulations to characterize the dynamics of ion permeation across a single-walled (10, 10), 36 Å long, carbon nanotube terminated with carboxylic acid with an effective radius of 5.08 Å. Although cations encounter a high energy barrier of 7 kT, its height is drastically reduced by a chloride ion in the nanotube. The presence of a chloride ion near the pore entrance thus enables a cation to enter the pore and, once in the pore, it is chaperoned by the resident counterion across the narrow pore. The moment the chaperoned cation transits the pore, the counterion moves back to the entrance to ferry another ion. The synthetic nanotube has a high sodium conductance of 124 pS and shows linear current-voltage and current-concentration profiles. The cation-anion selectivity ratio ranges from 8 to 25, depending on the ionic concentrations in the reservoirs.

Computational modeling of transport in synthetic nanotubes.

Synthetic nanotubes that have the ability to broadly mimic the function of biological ion channels have extraordinary potential for various applications, from ultrasensitive biosensors to efficient water purification devices. As a result of their immense potential, the design and fabrication of such synthetic nanotubes is rapidly gaining momentum. We briefly review recent theoretical and experimental studies on nanoscale cylindrical hollow tubes constructed from carbon, boron, and nitrogen atoms that are able to selectively transport water molecules, cations (positively charged ions), or anions (negatively charged ions) similar to various biological ion channels. FROM THE CLINICAL EDITOR: This review discusses the current status of synthetic nanotube research, including recent theoretical and experimental studies on nanoscale cylindrical hollow tubes constructed from carbon, boron, and nitrogen atoms that are able to selectively transport water molecules, cations or anions similar to biological ion channels.

Synthetic chloride-selective carbon nanotubes examined by using molecular and stochastic dynamics.

Synthetic channels, such as nanotubes, offer the possibility of ion-selective nanoscale pores which can broadly mimic the functions of various biological ion channels, and may one day be used as antimicrobial agents, or for treatment of cystic fibrosis. We have designed a carbon nanotube that is selectively permeable to anions. The virtual nanotubes are constructed from a hexagonal array of carbon atoms (graphene) rolled up to form a tubular structure, with an effective radius of 4.53 Å and length of 34 Å. The pore ends are terminated with polar carbonyl groups. The nanotube thus formed is embedded in a lipid bilayer and a reservoir containing ionic solutions is added at each end of the pore. The conductance properties of these synthetic channels are then examined with molecular and stochastic dynamics simulations. Profiles of the potential of mean force at 0 mM reveal that a cation moving across the pore encounters an insurmountable free energy barrier of ∼25 kT in height. In contrast, for anions, there are two energy wells of ∼12 kT near each end of the tube, separated by a central free energy barrier of 4 kT. The conductance of the pore, with symmetrical 500 mM solutions in the reservoirs, is 72 pS at 100 mV. The current saturates with an increasing ionic concentration, obeying a Michaelis-Menten relationship. The pore is normally occupied by two ions, and the rate-limiting step in conduction is the time taken for the resident ion near the exit gate to move out of the energy well.

Efficacy of external tetraethylammonium block of the KcsA potassium channel: molecular and Brownian dynamics studies.

Blockade of the KcsA potassium channel by externally applied tetraethylammonium is investigated using molecular dynamics calculations and Brownian dynamics simulations. In KcsA, the aromatic rings of four tyrosine residues located just external to the selectivity filter create an attractive energy well or a binding cage for a tetraethylammonium molecule. We first investigate the effects of re-orienting the four tyrosine residues such that the centers of the aromatic rings face the tetraethylammonium molecule directly. Then, we systematically move the residues inward in both orientations so that the radius of the binding cage formed by them becomes smaller. For each configuration, we construct a one-dimensional free energy profile by bringing in a tetraethylammonium molecule from the external reservoir toward the selectivity filter. The free energy profile is then converted to a one-dimensional potential energy profile, taking the available space between the tyrosine residues and the tetraethylammonium molecule into account. Incorporating this potential energy profile into the Brownian dynamics algorithm, we determine the conductance properties of the channel under various conditions, construct the current-tetraethylammonium-concentration curve and compare it with the experimentally determined inhibitory constant k(i) for externally applied tetraethylammonium. We show that the experimentally determined binding affinity for externally applied tetraethylammonium can be replicated when each of the four tyrosine residues is moved inward by about 0.7 angstroms, irrespective of orientation of their aromatic rings.