- Email: [email protected]
Atomic determinants of state-dependent block of sodium channels by charged local anesthetics and benzocaine
FEBS Letters 580 (2006) 6027–6032
Hypothesis
Atomic determinants of state-dependent block of sodium channels by charged local anesthetics and benzocaine Denis B. Tikhonov, Iva Bruhova, Boris S. Zhorov* Department of Biochemistry and Biomedical Sciences, McMaster University, 1200 Main Street West, Hamilton, Ont., Canada L8N 3Z5 Received 4 September 2006; revised 11 October 2006; accepted 12 October 2006 Available online 24 October 2006 Edited by Maurice Montal
Abstract Molecular modeling predicts that a local anesthetic (LA) lidocaine binds to the resting and open Nav1.5 in different modes, interacting with LA-sensing residues known from experiments. Besides the major pathway via the open activation gate, LAs can reach the inner pore via a ‘‘sidewalk’’ between D3S6, D4S6, and D3P. The ammonium group of a cationic LA binds in the focus of the pore-helices macrodipoles, which also stabilize a Na+ ion chelated by two benzocaine molecules. The LA’s cationic group and a Na+ ion in the selectivity filter repel each other suggesting that the Na+ depletion upon slow inactivation would stabilize a LA, while a LA would stabilize slow-inactivated states. Ó 2006 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. Keywords: Homology modeling; Lidocaine; QX-314; Monte Carlo-minimization; Ion permeation; Channel block; Slow inactivation
1. Introduction Local anesthetics (LAs) are classical blockers of voltagegated sodium channels. LAs produce three main effects on the channels: resting block, use-dependent block, and the shift of inactivation [1,2]. To explain these effects, the ‘‘modulatedreceptor’’ hypothesis was elaborated [3]. According to this hypothesis, LAs slowly enter the closed channel through a hydrophobic pathway. When the channel is activated by membrane depolarization, the fast hydrophilic pathway becomes open. The binding of LAs blocks ion permeation. LAs have the highest affinity to the inactivated state and further stabilize the inactivated state. Mutational experiments revealed that LAs bind inside the inner pore. LA-sensing residues were found in segments D1S6, D3S6, and D4S6 [4,5]. It should be noted that effects
* Corresponding author. Fax: +1 905 522 9033. E-mail address: [email protected] (B.S. Zhorov).
Abbreviations: LA, local anesthetic; MC, Monte Carlo; DEKA, the circular locus of residues Asp, Glu, Lys, and Ala from P-loops in domains D1–D4, respectively, which form the Na+ channel selectivity filter; D1S6–D4S6, the inner helices in domains D1–D4, respectively; D3P, the pore helix in domain 3
of mutations on the resting and use-dependent block are unequal [6]. Except for benzocaine and its analogs, LAs are protonated or permanently charged molecules. Intriguingly, neutral benzocaine produces generally the same effects on the channel as charged LAs. Benzocaine shares a common binding site with other LAs, but some mutations have different influence on the effects of benzocaine and etidocaine [7]. Benzocaine does not demonstrate a use-dependent block, but this peculiarity is likely due to its fast kinetics [8]. The fundamental difference between benzocaine and other LAs is the Hill coefficient, which in the case of benzocaine is not equal to 1 [9]. In this work, we use homology modeling and ligand docking to address the following questions. What are energetically optimal binding modes of LAs in the open and resting channels? Whether a relatively large molecule like QX-314 could pass to the inner pore of the resting channel between D3S6 and D4S6, some residues in which had been shown to affect the hydrophobic pathway? Why do so different molecules as charged LAs and neutral benzocaine act similarly on sodium channels? How slow inactivation could stabilize the binding of both charged LAs and neutral benzocaine? Why do LAs stabilize slow-inactivated states?
2. Model building and ligand docking In the absence of X-ray structures of Na+ channels, the atomic determinants of the LA receptor can only emerge from molecular models based on the X-ray structures of potassium channels. We have built homology models of Nav1.5 (rH1) in the open and closed states using, respectively, KvAP [10] and KcsA [11] structures. The models were composed of P-loops and S6 segments, whose sequences were aligned with K+ channels as proposed earlier [12]. For the selectivity-filter region, which is significantly different between Na+ and K+ channels, we used our earlier model of the P-loop domain, which was shaped around tetrodotoxin and saxitoxin [13]. The models were optimized by the Monte Carlo-minimization protocol as described elsewhere [13]. To prevent large deviations of the models from the templates, ‘‘pin’’ constraints were imposed on Ca atoms of a-helices. The optimal binding modes of LAs were searched by a two-stage random-docking approach. In the first stage, 30 000 binding modes of a ligand were randomly generated and MC-minimized in short trajectories of 20 energy minimizations. A thousand of the energetically best structures found at this stage were further
0014-5793/$32.00 Ó 2006 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2006.10.035
6028
optimized in long MCM trajectories, each of which terminated when the last 1000 energy minimizations did not improve the apparent global minimum. Both ligand and protein were kept flexible during energy minimizations. Henceforth, we use the term ‘‘docking’’ to refer to the above procedure of searching energetically optimal binding modes of the ligand. Thus, all models discussed in this work represent stable low-energy conformations. Although results of in silico experiments with homology models are always questionable, the obtained models integrate seemingly controversial experimental data on the action of LAs in Na+ channels and provide answers to the questions formulated in the Introduction.
3. Ion dependence of LA binding We have built four models of Nav1.5 in which the activation gate was either open or closed, and the circular locus of residues Asp, Glu, Lys, and Ala from P-loops in domains D1– D4, respectively, which form the Na+ channel selectivity filter (DEKA) ring was loaded with either a Na+ ion (Na+-DEKA) or a water molecule (H2O-DEKA). A water molecule was placed in the model to stabilize side-chain conformations of the DEKA ring. Docking lidocaine, QX-314, and other cationic LAs in the closed and open H2O-DEKA models predicted several binding modes in which the ammonium group occurs at the focus of the pore-helix macrodipoles (Fig. 1A). In this respect, Nav1.5-bound LAs resemble KcsA-bound tetrabutylammonium, whose nitrogen is close to position 5 of K+ in the ligand-free channel [14]. In the open and closed Na+-DEKA models, there is a significant electrostatic repulsion between the ammonium group of the ligand and the Na+ ion. The repulsion weakens the LA-channel interaction energy and moves the ligand’s ammonium group away from the focus of helical macrodipoles (Fig. 1B). These results suggest that LAs should have higher affinity to channels that lack
D.B. Tikhonov et al. / FEBS Letters 580 (2006) 6027–6032
Na+ in the DEKA locus and that the binding of LAs should prevent Na+ to occupy the DEKA locus. Based on this result, we performed other computational experiments on the H2ODEKA models. The docking yielded two significantly different binding modes of LAs: planar and axial (Fig. 2). In the planar mode, the aromatic ring of the ligand protrudes between D3S6 and D4S6. In the axial mode, the aromatic ring faces intracellularly. Importantly, in both modes, F1762 and Y1769 in D4S6 (F1764 and Y1771 in rat brain IIA channel; F1579 and Y1586 in rat skeletal muscle isoform), which were identified in mutational studies as critical determinants of the LA receptor, are significant contributors to the ligand binding energy. Segments D1S6 and D3S6 provide modest contributions, whereas the interactions of LAs with D2S6 are very weak. In both binding modes, the ammonium group of the ligand is located at the focus of helical macrodipoles. The axial binding mode coincides with the proposal based on mutational and ligand-binding studies [15] and with a recent model of usedependent block [16]. A possibility of an alternative planar binding mode, impact of helical dipoles, and the role of ions in the selectivity filter are new proposals.
4. LA access and binding in the resting and open channels In the open channel, the obvious access of LAs to the inner pore is through the widely open activation gates. In the closed state, this pathway is impermeable for LAs. Some residues in S6s are known to affect the resting-channel block by the external quaternary LAs [17,18]. Mapping these residues in the model Nav1.4, visualizes a ‘‘sidewalk’’ to the inner pore between D3S6 and D4S6 along the pore helix in domain 3 (D3P) [13]. A Cys mutant of Nav1.5, whose position is immediately C-terminal to Asp of the DEKA locus, affects egress of the protonated lidocaine from the pore suggesting that a hydrophilic pathway via the outer pore may coexist with the
Fig. 1. The side views of the lowest-energy complexes of lidocaine in H2O-DEKA (A) and Na+-DEKA (B) models of the open Nav1.5. The inner helices and pore helices are shown as violet and green ribbons, respectively. Domain 2 is removed for clarity. The P-loop segments lining the outer pore are shown by Ca tracings. Lidocaine is shown by thick sticks. Side chains of the DEKA locus are shown by thin sticks. Side chains of Phe and Tyr residues in D4S6, which are critical determinants of the LA receptor, are colored magenta. The water molecule and Na+ ion at the DEKA locus are shown by red and yellow spheres, respectively. In the H2O-DEKA model (A), which represents a slow-inactivated channel, the ammonium group of the ligand is stabilized at the focus of macrodipoles of the pore helices. In the Na+-DEKA model (B), the repulsion between the Na+ ion and the ligand’s ammonium group cause the latter to face the cytoplasmic entrance to the open pore.
D.B. Tikhonov et al. / FEBS Letters 580 (2006) 6027–6032
6029
Fig. 2. The cytoplasmic (A,C) and side (B,D) views of lidocaine in the H2O-DEKA models of the open (A,B) and closed (C,D) Nav1.5. In both the open and closed channels, the ligand interacts with Phe and Tyr residues of D4S6 (magenta sticks). In the open channel, the ligand is in the axial binding mode, which is complementary to the open inner pore. In the closed channel, the ligand is in the planar binding mode. In both open and closed channels, the ligand’s ammonium group binds at the focus of the macrodipoles of the pore helices. For clarity the side views have a transparent D2S6.
hydrophobic pathway [19]. In this work, we explored only the hydrophobic pathway, which is lined by the protein segments whose mutual disposition is defined in the X-ray structures of the K+ channel templates. We have pulled QX-314 along the hydrophobic pathway ˚ and MC-minimizing the energy at each with the step of 1 A step. The superposition of the MC-minimized structures with the ligand at different positions along the hydrophobic pathway is shown in Fig. 3. No high-energy barriers were predicted in these computations suggesting that the hydrophobic pathway is passable by QX-314 and hence by smaller LAs. Exact molecular mechanisms of isoform specificity of QX-314 access and effects of mutations need systematic study. A possible difference in the access of the protonated and deprotonated forms of LAs also should be addressed in future works. For example, since the hydrophobic pathway is close to the selectivity filter, presence of an ion in the DEKA locus could affect access of charged molecules.
Our results suggest the following mechanism of LA action. A deprotonated LA molecule dissolved in the membrane would reach the resting-channel inner pore through the pathway between D3S6, D4S6, and D3P. Inside the pore, the LA molecule would accept a proton and become a hydrophobic cation enjoying hydrophobic interactions with hydrophobic residues in the inner helices and electrostatic interactions with the pore-helices macrodipoles. In the cardiac channel, quaternary compounds also can enter the channel via this pathway. Importantly, the lowest-energy structures obtained at the inner-pore end of the hydrophobic pathway (Fig. 3) are very similar to the planar binding mode of LAs found by random docking (Fig. 2). Since the closed inner pore does not provide enough room for a LA molecule to maneuver, the molecule that enters the pore via the hydrophobic pathway would remain in the planar binding mode. In the open channel, LAs reach the binding site rapidly via the open intracellular gate and, most likely, would bind in the axial mode, which is
6030
D.B. Tikhonov et al. / FEBS Letters 580 (2006) 6027–6032
Fig. 3. Moving QX-314 to the closed Nav1.5 via the hydrophobic pathway between D3S6, D4S6, and D3P. Panels A and B show orthogonal side views at the superposition of MC-minimized structures with the ligand constrained at different distances from the pore axis. Residues in the extracellular half of D4S6, which affect the Na+ channel block by the extracellularly applied LAs with the quaternary ammonium group [17,18], are space-filled and magenta-colored.
complementary to the open pore. Thus, the binding mode of LAs in the open and closed channels may depend on the access pathway. This model explains the observation that mutations of F1762 and Y1769 in D4S6 cause qualitatively different effect on the use-dependent and resting block [6].
5. Relations between slow inactivation and binding of cationic LAs Our models provide an explanation to the notion that cationic LAs bind preferably to slow-inactivated channels [2]. The presence of a Na+ ion in the DEKA locus would destabilize the binding of a cationic ligand. Slow-inactivated channels do not permeate ions and therefore may lack Na+ in the DEKA locus. X-ray structures suggest that the extracellular gate closure in K+ channels may be associated with the ion deficiency in the selectivity filter [14,20]. The binding of cationic ligands in the inner pore of K+ channels enhances slow inactivation [21,22]. The cause of this phenomenon may be the depletion of a K+ ion from the selectivity filter by a cationic ligand [23]. Berneche and Roux [24] suggested that the rearrangements of the selectivity filter in KcsA caused by ion deficiency might be associated with slow inactivation. By analogy with K+ channels, we propose that the H2ODEKA and Na+-DEKA models correspond, respectively, to the slow inactivated and conducting states of the Na+ channel. Docking LAs in the H2O-DEKA and Na+-DEKA models explains the state-dependent action of cationic ligands by repulsion between the ligand’s ammonium group and an ion in the DEKA locus. Cationic LAs bind stronger to the slow-inactivated (H2O-DEKA) channels and stabilize the slow-inactivated state by antagonizing occupation of the DEKA ring by an ion. Certainly, slow inactivation is a complex process that probably involves yet unknown structural rearrangements of the channel protein. Our simple model of the slow-inactivated state just suggests that the DEKA locus occupancy by
either a cation or a water molecule may be of critical importance for the state-dependent ligand binding.
6. The benzocaine paradox Cationic ligands block the ion permeation by occupying the focus of macrodipoles of the pore helices. From this point of view, neutral benzocaine seems like a puzzle because the binding of a small amphiphilic molecule is not expected to block ion permeation. The shift of steady-state inactivation by benzocaine [25] is also a puzzle. Our models explain the channel block and effect of inactivation by the antagonism between Na+ in the DEKA locus and the ammonium group in the focus of the macrodipoles, but benzocaine lacks a charged group. A key to resolve the paradox is the assumption that the ester group of benzocaine binds a Na+ ion. Earlier we proposed a model, in which the sodium channel agonist batrachotoxin occupies the inner pore close to the selectivity filter and a permeant Na+ ion directly binds to oxygen atoms of the ligand [26]. A touchstone prediction of the model was that Phe preceding Lys of the DEKA locus contributes to the batrachotoxin receptor. Recent experiments confirmed the prediction [27]. Since the Hill coefficient for benzocaine binding is greater than 1, we further suggest that two benzocaine molecules interact with a Na+ ion inside the pore. To explore this possibility computationally, we generated a large number of starting points with two benzocaine molecules and a Na+ ion randomly seeded in a large sphere volume in the inner pore of the closed channel. The two-stage random docking protocol (see 2) yielded lowest-energy structures with two benzocaine molecules jointly chelating the Na+ ion by their ester groups. One of such structures (Fig. 4) shows that the Na+ ion bound to benzocaine can mimic the ammonium group of a charged LA. One benzocaine molecule occurs in the planar, and another in the axial orientation in the channel, while aro-
D.B. Tikhonov et al. / FEBS Letters 580 (2006) 6027–6032
6031
between the slow inactivation and the channel block. Considerable efforts are needed to experimentally verify the hypothesis, which integrates data from decades of experimental studies. If correct, the hypothesis could be of interest beyond the sodium channels field. Further analysis of existent and new experimental data in view of our models is necessary. Coordinates of the models are available upon request from BSZ ([email protected]). Acknowledgements: This work was supported by the grant from the Canadian Institutes of Health Research. Computations were performed, in part, using the facilities of the Shared Hierarchical Academic Research Computing Network (SHARCNET: www. sharcnet.ca).
References
Fig. 4. The side view of the H2O-DEKA model of the closed channel blocked by the ternary complex involving a Na+ ion and two benzocaine molecules. Four oxygen atoms of the benzocaine molecules coordinate the Na+ ion (yellow sphere), which occurs at the focus of macrodipoles of the pore helices. Phe and Tyr in D4S6 (magenta sticks), which are critical determinants of the LA receptor, interact with benzocaine molecules in the planar and axial orientations, respectively.
matic rings interact with critical F1762 and Y1769 residues in D4S6. The concentration dependence of benzocaine block is complex: the Hill coefficient increases with the benzocaine concentration [9]. Until now, this non-typical concentration dependence was not explained. In the view of our model, at low concentrations one molecule of benzocaine can bind to the Na+ ion in the focus of macrodipoles, stabilize it, and block ion permeation. At high concentrations, two benzocaine molecules would jointly chelate the Na+ ion. Benzocaine binding is not a unique example of a concentration-dependent Hill coefficient. Similar concentration dependence was observed with S-nitrosodithiothreitol action on Kv1 channels [28]. The authors of the study suggested that two S-nitrosodithiothreitol molecules bind to the channel, but even a single molecule can block it. A notable analogy with benzocaine is that S-nitrosodithiothreitol is a small, uncharged molecule, which has nothing in common with a typical cationic blocker of K+ channels, tetraethylammonium. Like benzocaine, S-nitrosodithiothreitol has polar groups that could bind to a permeant cation inside the pore and stabilize it. Indeed, the voltage dependence of S-nitrosodithiothreitol action suggests that K+ forms a charged complex with the ligand(s), and that this complex blocks the channel.
7. Concluding remarks Results of molecular modeling of the cardiac Na+ channel and its complexes with LAs provide new insights to the problem of state-dependent ligand action. Our study integrates various experimental data and suggests a hypothesis on the common mechanism of the channel block by essentially different ligands and on molecular interactions underlying relations
[1] Nau, C. and Wang, G.K. (2004) Interactions of local anesthetics with voltage-gated Na+ channels. J. Membr. Biol. 201, 1–8. [2] Fozzard, H.A., Lee, P.J. and Lipkind, G.M. (2005) Mechanism of local anesthetic drug action on voltage-gated sodium channels. Curr. Pharm. Des. 11, 2671–2686. [3] Hille, B. (1977) Local anesthetics: hydrophilic and hydrophobic pathways for the drug-receptor reaction. J. Gen. Physiol. 69, 497– 515. [4] Wang, S.Y., Barile, M. and Wang, G.K. (2001) Disparate role of Na(+) channel D2-S6 residues in batrachotoxin and local anesthetic action. Mol. Pharmacol 59, 1100–1107. [5] Yarov-Yarovoy, V., McPhee, J.C., Idsvoog, D., Pate, C., Scheuer, T. and Catterall, W.A. (2002) Role of amino acid residues in transmembrane segments IS6 and IIS6 of the Na+ channel alpha subunit in voltage-dependent gating and drug block. J. Biol. Chem. 277, 35393–35401. [6] Li, H.L., Galue, A., Meadows, L. and Ragsdale, D.S. (1999) A molecular basis for the different local anesthetic affinities of resting versus open and inactivated states of the sodium channel. Mol. Pharmacol. 55, 134–141. [7] Wang, G.K., Quan, C. and Wang, S. (1998) A common local anesthetic receptor for benzocaine and etidocaine in voltage-gated mu1 Na+ channels. Pflugers Arch. 435, 293–302. [8] Quan, C., Mok, W.M. and Wang, G.K. (1996) Use-dependent inhibition of Na+ currents by benzocaine homologs. Biophys. J. 70, 94–201. [9] Meeder, T. and Ulbricht, W. (1987) Action of benzocaine on sodium channels of frog nodes of Ranvier treated with chloramine-T. Pflugers Arch. 409, 265–273. [10] Jiang, Y., Lee, A., Chen, J., Ruta, V., Cadene, M., Chait, B.T. and MacKinnon, R. (2003) X-ray structure of a voltage-dependent K+ channel. Nature 423, 33–41. [11] Doyle, D.A., Morais Cabral, J., Pfuetzner, R.A., Kuo, A., Gulbis, J.M., Cohen, S.L., Chait, B.T. and MacKinnon, R. (1998) The structure of the potassium channel: molecular basis of K+ conduction and selectivity. Science 280, 69–77. [12] Zhorov, B.S. and Tikhonov, D.B. (2004) Potassium, sodium, calcium and glutamate-gated channels: pore architecture and ligand action. J. Neurochem. 88, 782–799. [13] Tikhonov, D.B. and Zhorov, B.S. (2005) Modeling P-loops domain of sodium channel: homology with potassium channels and interaction with ligands. Biophys. J. 88, 184–197. [14] Zhou, Y., Morais-Cabral, J.H., Kaufman, A. and MacKinnon, R. (2001) Chemistry of ion coordination and hydration revealed by a K+ channel-Fab complex at 2.0 A resolution. Nature 414, 43–48. [15] Ragsdale, D.S., McPhee, J.C., Scheuer, T. and Catterall, W.A. (1994) Molecular determinants of state-dependent block of Na+ channels by local anesthetics. Science 265, 1724–1728. [16] Lipkind, G.M. and Fozzard, H.A. (2005) Molecular modeling of local anesthetic drug binding by voltage-gated sodium channels. Mol. Pharmacol. 68, 1611–1622. [17] Qu, Y., Rogers, J., Tanada, T., Scheuer, T. and Catterall, W.A. (1995) Molecular determinants of drug access to the receptor site for antiarrhythmic drugs in the cardiac Na+ channel. Proc. Natl. Acad. Sci. USA 92, 11839–11843.
6032 [18] Sunami, A., Glaaser, I.W. and Fozzard, H.A. (2001) Structural and gating changes of the sodium channel induced by mutation of a residue in the upper third of IVS6, creating an external access path for local anesthetics. Mol. Pharmacol. 59, 684–691. [19] Lee, P.J., Sunami, A. and Fozzard, H.A. (2001) Cardiac-specific external paths for lidocaine, defined by isoform-specific residues, accelerate recovery from use-dependent block. Circ. Res. 89, 1014–1021. [20] Zhou, Y. and MacKinnon, R. (2003) The occupancy of ions in the K+ selectivity filter: charge balance and coupling of ion binding to a protein conformational change underlie high conduction rates. J. Mol. Biol. 333, 965–975. [21] Grissmer, S. and Cahalan, M. (1989) TEA prevents inactivation while blocking open K+ channels in human T lymphocytes. Biophys. J. 55, 203–206. [22] Choi, K.L., Aldrich, R.W. and Yellen, G. (1991) Tetraethylammonium blockade distinguishes two inactivation mechanisms in voltage-activated K+ channels. Proc. Natl. Acad. Sci. USA 88, 5092–5095.
D.B. Tikhonov et al. / FEBS Letters 580 (2006) 6027–6032 [23] Lenaeus, M.J., Vamvouka, M., Focia, P.J. and Gross, A. (2005) Structural basis of TEA blockade in a model potassium channel. Nat. Struct. Mol. Biol. 12, 454–459. [24] Berneche, S. and Roux, B. (2005) A gate in the selectivity filter of potassium channels. Structure 13, 591–600. [25] Elliott, J.R., Haydon, D.A. and Hendry, B.M. (1987) The mechanisms of sodium current inhibition by benzocaine in the squid giant axon. Pflugers Arch. 409, 596–600. [26] Tikhonov, D.B. and Zhorov, B.S. (2005) Sodium channel activators: model of binding inside the pore and a possible mechanism of action. FEBS Lett. 579, 4207–4212. [27] Wang, S.Y., Mitchell, J., Tikhonov, D.B., Zhorov, B.S. and Wang, G.K. (2006) How batrachotoxin modifies the sodium channel permeation pathway: computer modeling and sitedirected mutagenesis. Mol. Pharmacol. 69, 788–795. [28] Brock, M.W., Mathes, C. and Gilly, W.F. (2001) Selective openchannel block of Shaker (Kv1) potassium channels by s-nitrosodithiothreitol (SNDTT). J. Gen. Physiol. 118, 113–134.
Hypothesis
Atomic determinants of state-dependent block of sodium channels by charged local anesthetics and benzocaine Denis B. Tikhonov, Iva Bruhova, Boris S. Zhorov* Department of Biochemistry and Biomedical Sciences, McMaster University, 1200 Main Street West, Hamilton, Ont., Canada L8N 3Z5 Received 4 September 2006; revised 11 October 2006; accepted 12 October 2006 Available online 24 October 2006 Edited by Maurice Montal
Abstract Molecular modeling predicts that a local anesthetic (LA) lidocaine binds to the resting and open Nav1.5 in different modes, interacting with LA-sensing residues known from experiments. Besides the major pathway via the open activation gate, LAs can reach the inner pore via a ‘‘sidewalk’’ between D3S6, D4S6, and D3P. The ammonium group of a cationic LA binds in the focus of the pore-helices macrodipoles, which also stabilize a Na+ ion chelated by two benzocaine molecules. The LA’s cationic group and a Na+ ion in the selectivity filter repel each other suggesting that the Na+ depletion upon slow inactivation would stabilize a LA, while a LA would stabilize slow-inactivated states. Ó 2006 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. Keywords: Homology modeling; Lidocaine; QX-314; Monte Carlo-minimization; Ion permeation; Channel block; Slow inactivation
1. Introduction Local anesthetics (LAs) are classical blockers of voltagegated sodium channels. LAs produce three main effects on the channels: resting block, use-dependent block, and the shift of inactivation [1,2]. To explain these effects, the ‘‘modulatedreceptor’’ hypothesis was elaborated [3]. According to this hypothesis, LAs slowly enter the closed channel through a hydrophobic pathway. When the channel is activated by membrane depolarization, the fast hydrophilic pathway becomes open. The binding of LAs blocks ion permeation. LAs have the highest affinity to the inactivated state and further stabilize the inactivated state. Mutational experiments revealed that LAs bind inside the inner pore. LA-sensing residues were found in segments D1S6, D3S6, and D4S6 [4,5]. It should be noted that effects
* Corresponding author. Fax: +1 905 522 9033. E-mail address: [email protected] (B.S. Zhorov).
Abbreviations: LA, local anesthetic; MC, Monte Carlo; DEKA, the circular locus of residues Asp, Glu, Lys, and Ala from P-loops in domains D1–D4, respectively, which form the Na+ channel selectivity filter; D1S6–D4S6, the inner helices in domains D1–D4, respectively; D3P, the pore helix in domain 3
of mutations on the resting and use-dependent block are unequal [6]. Except for benzocaine and its analogs, LAs are protonated or permanently charged molecules. Intriguingly, neutral benzocaine produces generally the same effects on the channel as charged LAs. Benzocaine shares a common binding site with other LAs, but some mutations have different influence on the effects of benzocaine and etidocaine [7]. Benzocaine does not demonstrate a use-dependent block, but this peculiarity is likely due to its fast kinetics [8]. The fundamental difference between benzocaine and other LAs is the Hill coefficient, which in the case of benzocaine is not equal to 1 [9]. In this work, we use homology modeling and ligand docking to address the following questions. What are energetically optimal binding modes of LAs in the open and resting channels? Whether a relatively large molecule like QX-314 could pass to the inner pore of the resting channel between D3S6 and D4S6, some residues in which had been shown to affect the hydrophobic pathway? Why do so different molecules as charged LAs and neutral benzocaine act similarly on sodium channels? How slow inactivation could stabilize the binding of both charged LAs and neutral benzocaine? Why do LAs stabilize slow-inactivated states?
2. Model building and ligand docking In the absence of X-ray structures of Na+ channels, the atomic determinants of the LA receptor can only emerge from molecular models based on the X-ray structures of potassium channels. We have built homology models of Nav1.5 (rH1) in the open and closed states using, respectively, KvAP [10] and KcsA [11] structures. The models were composed of P-loops and S6 segments, whose sequences were aligned with K+ channels as proposed earlier [12]. For the selectivity-filter region, which is significantly different between Na+ and K+ channels, we used our earlier model of the P-loop domain, which was shaped around tetrodotoxin and saxitoxin [13]. The models were optimized by the Monte Carlo-minimization protocol as described elsewhere [13]. To prevent large deviations of the models from the templates, ‘‘pin’’ constraints were imposed on Ca atoms of a-helices. The optimal binding modes of LAs were searched by a two-stage random-docking approach. In the first stage, 30 000 binding modes of a ligand were randomly generated and MC-minimized in short trajectories of 20 energy minimizations. A thousand of the energetically best structures found at this stage were further
0014-5793/$32.00 Ó 2006 Federation of European Biochemical Societies. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.febslet.2006.10.035
6028
optimized in long MCM trajectories, each of which terminated when the last 1000 energy minimizations did not improve the apparent global minimum. Both ligand and protein were kept flexible during energy minimizations. Henceforth, we use the term ‘‘docking’’ to refer to the above procedure of searching energetically optimal binding modes of the ligand. Thus, all models discussed in this work represent stable low-energy conformations. Although results of in silico experiments with homology models are always questionable, the obtained models integrate seemingly controversial experimental data on the action of LAs in Na+ channels and provide answers to the questions formulated in the Introduction.
3. Ion dependence of LA binding We have built four models of Nav1.5 in which the activation gate was either open or closed, and the circular locus of residues Asp, Glu, Lys, and Ala from P-loops in domains D1– D4, respectively, which form the Na+ channel selectivity filter (DEKA) ring was loaded with either a Na+ ion (Na+-DEKA) or a water molecule (H2O-DEKA). A water molecule was placed in the model to stabilize side-chain conformations of the DEKA ring. Docking lidocaine, QX-314, and other cationic LAs in the closed and open H2O-DEKA models predicted several binding modes in which the ammonium group occurs at the focus of the pore-helix macrodipoles (Fig. 1A). In this respect, Nav1.5-bound LAs resemble KcsA-bound tetrabutylammonium, whose nitrogen is close to position 5 of K+ in the ligand-free channel [14]. In the open and closed Na+-DEKA models, there is a significant electrostatic repulsion between the ammonium group of the ligand and the Na+ ion. The repulsion weakens the LA-channel interaction energy and moves the ligand’s ammonium group away from the focus of helical macrodipoles (Fig. 1B). These results suggest that LAs should have higher affinity to channels that lack
D.B. Tikhonov et al. / FEBS Letters 580 (2006) 6027–6032
Na+ in the DEKA locus and that the binding of LAs should prevent Na+ to occupy the DEKA locus. Based on this result, we performed other computational experiments on the H2ODEKA models. The docking yielded two significantly different binding modes of LAs: planar and axial (Fig. 2). In the planar mode, the aromatic ring of the ligand protrudes between D3S6 and D4S6. In the axial mode, the aromatic ring faces intracellularly. Importantly, in both modes, F1762 and Y1769 in D4S6 (F1764 and Y1771 in rat brain IIA channel; F1579 and Y1586 in rat skeletal muscle isoform), which were identified in mutational studies as critical determinants of the LA receptor, are significant contributors to the ligand binding energy. Segments D1S6 and D3S6 provide modest contributions, whereas the interactions of LAs with D2S6 are very weak. In both binding modes, the ammonium group of the ligand is located at the focus of helical macrodipoles. The axial binding mode coincides with the proposal based on mutational and ligand-binding studies [15] and with a recent model of usedependent block [16]. A possibility of an alternative planar binding mode, impact of helical dipoles, and the role of ions in the selectivity filter are new proposals.
4. LA access and binding in the resting and open channels In the open channel, the obvious access of LAs to the inner pore is through the widely open activation gates. In the closed state, this pathway is impermeable for LAs. Some residues in S6s are known to affect the resting-channel block by the external quaternary LAs [17,18]. Mapping these residues in the model Nav1.4, visualizes a ‘‘sidewalk’’ to the inner pore between D3S6 and D4S6 along the pore helix in domain 3 (D3P) [13]. A Cys mutant of Nav1.5, whose position is immediately C-terminal to Asp of the DEKA locus, affects egress of the protonated lidocaine from the pore suggesting that a hydrophilic pathway via the outer pore may coexist with the
Fig. 1. The side views of the lowest-energy complexes of lidocaine in H2O-DEKA (A) and Na+-DEKA (B) models of the open Nav1.5. The inner helices and pore helices are shown as violet and green ribbons, respectively. Domain 2 is removed for clarity. The P-loop segments lining the outer pore are shown by Ca tracings. Lidocaine is shown by thick sticks. Side chains of the DEKA locus are shown by thin sticks. Side chains of Phe and Tyr residues in D4S6, which are critical determinants of the LA receptor, are colored magenta. The water molecule and Na+ ion at the DEKA locus are shown by red and yellow spheres, respectively. In the H2O-DEKA model (A), which represents a slow-inactivated channel, the ammonium group of the ligand is stabilized at the focus of macrodipoles of the pore helices. In the Na+-DEKA model (B), the repulsion between the Na+ ion and the ligand’s ammonium group cause the latter to face the cytoplasmic entrance to the open pore.
D.B. Tikhonov et al. / FEBS Letters 580 (2006) 6027–6032
6029
Fig. 2. The cytoplasmic (A,C) and side (B,D) views of lidocaine in the H2O-DEKA models of the open (A,B) and closed (C,D) Nav1.5. In both the open and closed channels, the ligand interacts with Phe and Tyr residues of D4S6 (magenta sticks). In the open channel, the ligand is in the axial binding mode, which is complementary to the open inner pore. In the closed channel, the ligand is in the planar binding mode. In both open and closed channels, the ligand’s ammonium group binds at the focus of the macrodipoles of the pore helices. For clarity the side views have a transparent D2S6.
hydrophobic pathway [19]. In this work, we explored only the hydrophobic pathway, which is lined by the protein segments whose mutual disposition is defined in the X-ray structures of the K+ channel templates. We have pulled QX-314 along the hydrophobic pathway ˚ and MC-minimizing the energy at each with the step of 1 A step. The superposition of the MC-minimized structures with the ligand at different positions along the hydrophobic pathway is shown in Fig. 3. No high-energy barriers were predicted in these computations suggesting that the hydrophobic pathway is passable by QX-314 and hence by smaller LAs. Exact molecular mechanisms of isoform specificity of QX-314 access and effects of mutations need systematic study. A possible difference in the access of the protonated and deprotonated forms of LAs also should be addressed in future works. For example, since the hydrophobic pathway is close to the selectivity filter, presence of an ion in the DEKA locus could affect access of charged molecules.
Our results suggest the following mechanism of LA action. A deprotonated LA molecule dissolved in the membrane would reach the resting-channel inner pore through the pathway between D3S6, D4S6, and D3P. Inside the pore, the LA molecule would accept a proton and become a hydrophobic cation enjoying hydrophobic interactions with hydrophobic residues in the inner helices and electrostatic interactions with the pore-helices macrodipoles. In the cardiac channel, quaternary compounds also can enter the channel via this pathway. Importantly, the lowest-energy structures obtained at the inner-pore end of the hydrophobic pathway (Fig. 3) are very similar to the planar binding mode of LAs found by random docking (Fig. 2). Since the closed inner pore does not provide enough room for a LA molecule to maneuver, the molecule that enters the pore via the hydrophobic pathway would remain in the planar binding mode. In the open channel, LAs reach the binding site rapidly via the open intracellular gate and, most likely, would bind in the axial mode, which is
6030
D.B. Tikhonov et al. / FEBS Letters 580 (2006) 6027–6032
Fig. 3. Moving QX-314 to the closed Nav1.5 via the hydrophobic pathway between D3S6, D4S6, and D3P. Panels A and B show orthogonal side views at the superposition of MC-minimized structures with the ligand constrained at different distances from the pore axis. Residues in the extracellular half of D4S6, which affect the Na+ channel block by the extracellularly applied LAs with the quaternary ammonium group [17,18], are space-filled and magenta-colored.
complementary to the open pore. Thus, the binding mode of LAs in the open and closed channels may depend on the access pathway. This model explains the observation that mutations of F1762 and Y1769 in D4S6 cause qualitatively different effect on the use-dependent and resting block [6].
5. Relations between slow inactivation and binding of cationic LAs Our models provide an explanation to the notion that cationic LAs bind preferably to slow-inactivated channels [2]. The presence of a Na+ ion in the DEKA locus would destabilize the binding of a cationic ligand. Slow-inactivated channels do not permeate ions and therefore may lack Na+ in the DEKA locus. X-ray structures suggest that the extracellular gate closure in K+ channels may be associated with the ion deficiency in the selectivity filter [14,20]. The binding of cationic ligands in the inner pore of K+ channels enhances slow inactivation [21,22]. The cause of this phenomenon may be the depletion of a K+ ion from the selectivity filter by a cationic ligand [23]. Berneche and Roux [24] suggested that the rearrangements of the selectivity filter in KcsA caused by ion deficiency might be associated with slow inactivation. By analogy with K+ channels, we propose that the H2ODEKA and Na+-DEKA models correspond, respectively, to the slow inactivated and conducting states of the Na+ channel. Docking LAs in the H2O-DEKA and Na+-DEKA models explains the state-dependent action of cationic ligands by repulsion between the ligand’s ammonium group and an ion in the DEKA locus. Cationic LAs bind stronger to the slow-inactivated (H2O-DEKA) channels and stabilize the slow-inactivated state by antagonizing occupation of the DEKA ring by an ion. Certainly, slow inactivation is a complex process that probably involves yet unknown structural rearrangements of the channel protein. Our simple model of the slow-inactivated state just suggests that the DEKA locus occupancy by
either a cation or a water molecule may be of critical importance for the state-dependent ligand binding.
6. The benzocaine paradox Cationic ligands block the ion permeation by occupying the focus of macrodipoles of the pore helices. From this point of view, neutral benzocaine seems like a puzzle because the binding of a small amphiphilic molecule is not expected to block ion permeation. The shift of steady-state inactivation by benzocaine [25] is also a puzzle. Our models explain the channel block and effect of inactivation by the antagonism between Na+ in the DEKA locus and the ammonium group in the focus of the macrodipoles, but benzocaine lacks a charged group. A key to resolve the paradox is the assumption that the ester group of benzocaine binds a Na+ ion. Earlier we proposed a model, in which the sodium channel agonist batrachotoxin occupies the inner pore close to the selectivity filter and a permeant Na+ ion directly binds to oxygen atoms of the ligand [26]. A touchstone prediction of the model was that Phe preceding Lys of the DEKA locus contributes to the batrachotoxin receptor. Recent experiments confirmed the prediction [27]. Since the Hill coefficient for benzocaine binding is greater than 1, we further suggest that two benzocaine molecules interact with a Na+ ion inside the pore. To explore this possibility computationally, we generated a large number of starting points with two benzocaine molecules and a Na+ ion randomly seeded in a large sphere volume in the inner pore of the closed channel. The two-stage random docking protocol (see 2) yielded lowest-energy structures with two benzocaine molecules jointly chelating the Na+ ion by their ester groups. One of such structures (Fig. 4) shows that the Na+ ion bound to benzocaine can mimic the ammonium group of a charged LA. One benzocaine molecule occurs in the planar, and another in the axial orientation in the channel, while aro-
D.B. Tikhonov et al. / FEBS Letters 580 (2006) 6027–6032
6031
between the slow inactivation and the channel block. Considerable efforts are needed to experimentally verify the hypothesis, which integrates data from decades of experimental studies. If correct, the hypothesis could be of interest beyond the sodium channels field. Further analysis of existent and new experimental data in view of our models is necessary. Coordinates of the models are available upon request from BSZ ([email protected]). Acknowledgements: This work was supported by the grant from the Canadian Institutes of Health Research. Computations were performed, in part, using the facilities of the Shared Hierarchical Academic Research Computing Network (SHARCNET: www. sharcnet.ca).
References
Fig. 4. The side view of the H2O-DEKA model of the closed channel blocked by the ternary complex involving a Na+ ion and two benzocaine molecules. Four oxygen atoms of the benzocaine molecules coordinate the Na+ ion (yellow sphere), which occurs at the focus of macrodipoles of the pore helices. Phe and Tyr in D4S6 (magenta sticks), which are critical determinants of the LA receptor, interact with benzocaine molecules in the planar and axial orientations, respectively.
matic rings interact with critical F1762 and Y1769 residues in D4S6. The concentration dependence of benzocaine block is complex: the Hill coefficient increases with the benzocaine concentration [9]. Until now, this non-typical concentration dependence was not explained. In the view of our model, at low concentrations one molecule of benzocaine can bind to the Na+ ion in the focus of macrodipoles, stabilize it, and block ion permeation. At high concentrations, two benzocaine molecules would jointly chelate the Na+ ion. Benzocaine binding is not a unique example of a concentration-dependent Hill coefficient. Similar concentration dependence was observed with S-nitrosodithiothreitol action on Kv1 channels [28]. The authors of the study suggested that two S-nitrosodithiothreitol molecules bind to the channel, but even a single molecule can block it. A notable analogy with benzocaine is that S-nitrosodithiothreitol is a small, uncharged molecule, which has nothing in common with a typical cationic blocker of K+ channels, tetraethylammonium. Like benzocaine, S-nitrosodithiothreitol has polar groups that could bind to a permeant cation inside the pore and stabilize it. Indeed, the voltage dependence of S-nitrosodithiothreitol action suggests that K+ forms a charged complex with the ligand(s), and that this complex blocks the channel.
7. Concluding remarks Results of molecular modeling of the cardiac Na+ channel and its complexes with LAs provide new insights to the problem of state-dependent ligand action. Our study integrates various experimental data and suggests a hypothesis on the common mechanism of the channel block by essentially different ligands and on molecular interactions underlying relations
[1] Nau, C. and Wang, G.K. (2004) Interactions of local anesthetics with voltage-gated Na+ channels. J. Membr. Biol. 201, 1–8. [2] Fozzard, H.A., Lee, P.J. and Lipkind, G.M. (2005) Mechanism of local anesthetic drug action on voltage-gated sodium channels. Curr. Pharm. Des. 11, 2671–2686. [3] Hille, B. (1977) Local anesthetics: hydrophilic and hydrophobic pathways for the drug-receptor reaction. J. Gen. Physiol. 69, 497– 515. [4] Wang, S.Y., Barile, M. and Wang, G.K. (2001) Disparate role of Na(+) channel D2-S6 residues in batrachotoxin and local anesthetic action. Mol. Pharmacol 59, 1100–1107. [5] Yarov-Yarovoy, V., McPhee, J.C., Idsvoog, D., Pate, C., Scheuer, T. and Catterall, W.A. (2002) Role of amino acid residues in transmembrane segments IS6 and IIS6 of the Na+ channel alpha subunit in voltage-dependent gating and drug block. J. Biol. Chem. 277, 35393–35401. [6] Li, H.L., Galue, A., Meadows, L. and Ragsdale, D.S. (1999) A molecular basis for the different local anesthetic affinities of resting versus open and inactivated states of the sodium channel. Mol. Pharmacol. 55, 134–141. [7] Wang, G.K., Quan, C. and Wang, S. (1998) A common local anesthetic receptor for benzocaine and etidocaine in voltage-gated mu1 Na+ channels. Pflugers Arch. 435, 293–302. [8] Quan, C., Mok, W.M. and Wang, G.K. (1996) Use-dependent inhibition of Na+ currents by benzocaine homologs. Biophys. J. 70, 94–201. [9] Meeder, T. and Ulbricht, W. (1987) Action of benzocaine on sodium channels of frog nodes of Ranvier treated with chloramine-T. Pflugers Arch. 409, 265–273. [10] Jiang, Y., Lee, A., Chen, J., Ruta, V., Cadene, M., Chait, B.T. and MacKinnon, R. (2003) X-ray structure of a voltage-dependent K+ channel. Nature 423, 33–41. [11] Doyle, D.A., Morais Cabral, J., Pfuetzner, R.A., Kuo, A., Gulbis, J.M., Cohen, S.L., Chait, B.T. and MacKinnon, R. (1998) The structure of the potassium channel: molecular basis of K+ conduction and selectivity. Science 280, 69–77. [12] Zhorov, B.S. and Tikhonov, D.B. (2004) Potassium, sodium, calcium and glutamate-gated channels: pore architecture and ligand action. J. Neurochem. 88, 782–799. [13] Tikhonov, D.B. and Zhorov, B.S. (2005) Modeling P-loops domain of sodium channel: homology with potassium channels and interaction with ligands. Biophys. J. 88, 184–197. [14] Zhou, Y., Morais-Cabral, J.H., Kaufman, A. and MacKinnon, R. (2001) Chemistry of ion coordination and hydration revealed by a K+ channel-Fab complex at 2.0 A resolution. Nature 414, 43–48. [15] Ragsdale, D.S., McPhee, J.C., Scheuer, T. and Catterall, W.A. (1994) Molecular determinants of state-dependent block of Na+ channels by local anesthetics. Science 265, 1724–1728. [16] Lipkind, G.M. and Fozzard, H.A. (2005) Molecular modeling of local anesthetic drug binding by voltage-gated sodium channels. Mol. Pharmacol. 68, 1611–1622. [17] Qu, Y., Rogers, J., Tanada, T., Scheuer, T. and Catterall, W.A. (1995) Molecular determinants of drug access to the receptor site for antiarrhythmic drugs in the cardiac Na+ channel. Proc. Natl. Acad. Sci. USA 92, 11839–11843.
6032 [18] Sunami, A., Glaaser, I.W. and Fozzard, H.A. (2001) Structural and gating changes of the sodium channel induced by mutation of a residue in the upper third of IVS6, creating an external access path for local anesthetics. Mol. Pharmacol. 59, 684–691. [19] Lee, P.J., Sunami, A. and Fozzard, H.A. (2001) Cardiac-specific external paths for lidocaine, defined by isoform-specific residues, accelerate recovery from use-dependent block. Circ. Res. 89, 1014–1021. [20] Zhou, Y. and MacKinnon, R. (2003) The occupancy of ions in the K+ selectivity filter: charge balance and coupling of ion binding to a protein conformational change underlie high conduction rates. J. Mol. Biol. 333, 965–975. [21] Grissmer, S. and Cahalan, M. (1989) TEA prevents inactivation while blocking open K+ channels in human T lymphocytes. Biophys. J. 55, 203–206. [22] Choi, K.L., Aldrich, R.W. and Yellen, G. (1991) Tetraethylammonium blockade distinguishes two inactivation mechanisms in voltage-activated K+ channels. Proc. Natl. Acad. Sci. USA 88, 5092–5095.
D.B. Tikhonov et al. / FEBS Letters 580 (2006) 6027–6032 [23] Lenaeus, M.J., Vamvouka, M., Focia, P.J. and Gross, A. (2005) Structural basis of TEA blockade in a model potassium channel. Nat. Struct. Mol. Biol. 12, 454–459. [24] Berneche, S. and Roux, B. (2005) A gate in the selectivity filter of potassium channels. Structure 13, 591–600. [25] Elliott, J.R., Haydon, D.A. and Hendry, B.M. (1987) The mechanisms of sodium current inhibition by benzocaine in the squid giant axon. Pflugers Arch. 409, 596–600. [26] Tikhonov, D.B. and Zhorov, B.S. (2005) Sodium channel activators: model of binding inside the pore and a possible mechanism of action. FEBS Lett. 579, 4207–4212. [27] Wang, S.Y., Mitchell, J., Tikhonov, D.B., Zhorov, B.S. and Wang, G.K. (2006) How batrachotoxin modifies the sodium channel permeation pathway: computer modeling and sitedirected mutagenesis. Mol. Pharmacol. 69, 788–795. [28] Brock, M.W., Mathes, C. and Gilly, W.F. (2001) Selective openchannel block of Shaker (Kv1) potassium channels by s-nitrosodithiothreitol (SNDTT). J. Gen. Physiol. 118, 113–134.