Future Medicinal ChemistryVol. 2, No. 5 EditorialFree AccessUnderstanding ion channels using computational approachesMarcel J de GrootMarcel J de GrootPfizer Global Research & Development, World-Wide Medicinal Chemistry, Ramsgate Road, Sandwich, CT13 9NJ, Kent, UK. Search for more papers by this authorEmail the corresponding author at marcel.degroot@pfizer.comPublished Online:12 May 2010https://doi.org/10.4155/fmc.10.177AboutSectionsPDF/EPUB ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareShare onFacebookTwitterLinkedInReddit Ion channels have been investigated for years in academia. However, with the rise of (pre)clinical data and various genetic links between ion channels and diseases, ion channels have been drawing the interest of pharmaceutical industry. One example is Nav1.7, in which there is a strong genetic linkage between the ion channel and pain in humans [1]. Screening has become more high throughput throughout the years, as summarized recently [2], opening up possibilities for the pharmaceutical industry to screen (parts of) their compound collection.Computational methods have been applied to a large number of targets over the years, both ligand and structure based. Until the elucidation of the first crystal structure of an ion channel in 1998 [3], a variety of computational techniques have been used to refine the understanding of the structure and function of ion channels [4,5].With the appearance of structural data for ion channels, the question arises of how these data can be used to gain additional insights into the activity and binding of specific ion channels and how information can be transferred across the ion channel families.Structure-based computational approachesThe groundbreaking structural determination of the KcsA channel from Streptomyces lividans[3,6,7] has had a major impact in the fields of crystallography and computational chemistry. Since these initial structures, a variety of ion channels have been crystallized in recent years, not only showing the fourfold symmetry as displayed by the KcsA channel – which was also shown to exist in KvAP [8], Kv1.2 [9], KirBac3.1 [10], MthK [11], NaK [12], GluA2 [13] and TRPM7 [14], for example – but also trimeric channel layouts as shown in ASIC [15] and P2X4 [16], and even pentameric arrangements as seen in ELIC [17], GLIC [18] and the nACh receptor pore [19]. Crystal structures of open-state structures and closed-state structures are available for the pore domain. The structure of Kv1.2 is currently the only structure that contains both the pore domain and the voltage sensor domains [9].The occurrence of all these structures has enabled different computational methodologies to have an impact on the field of ion channels. Comparative modeling (also known as homology modeling) arises from the observation that proteins with similar amino acid sequences have a tendency to adopt similar 3D structures [20]. Therefore, it is possible to predict the 3D structure of a protein based solely on knowledge of its amino acid sequence and the 3D structures of proteins with similar sequences. Although these models will be inherently less accurate than those derived experimentally, they are invaluable as they provide testable hypotheses in the absence of experimental data.Since the emergence of the various ion channel crystal structures, a variety of comparative models has been published. This started even before the KcsA structures became available and the results obtained with this early model show remarkable similarities with the crystal structures obtained almost a decade later [21]. A variety of potassium channels have been modeled based on the various (tetrameric) potassium channel crystal structures. Three very recent examples of potassium channel models are a model for KCNQ3 based on Kv1.2 and used to refine the binding site [22], a hERG model based on KcsA looking at the impact of the model on guiding a medicinal chemistry program [23] and a Kv1.5 model based on Kv1.2 used to select compounds by docking and scoring techniques [24]. Similarly, a variety of sodium channels have been based on these tetrameric templates. Due to the nonsymmetrical nature of the sodium channels, alignment of the four domains brings additional assumptions and difficulties. Two of the latest models aimed to explain the structure–function relationships in sodium channels [25] and determine the drug-binding determinants in Nav1.8 [26]. Several models for transient receptor potential channels (TRPs) have also been constructed, including a model for TRPM8 looking at analyzing binding features within the channel[27], and a TRPV1 model focused on closed and desensitized states [28], both based on the Kv1.2 crystal structure. Models for twin-pore domain potassium channels have recently been based on KcsA, such as the TREK1 and TREK2 models used to investigate interactions of a specific residue [29], or on Kv1.2, like the model used to analyze binding site residues for K2P[30]. Although the majority of the models are generated for tetrameric channels, this is not exclusive. Trimeric models have been constructed and used for docking and molecular dynamics studies for human ASIC1a [31,32], while pentameric nACh was used as a template for a model of the transmembrane domain of 5-HT3[33].Current status of using comparative models of ion channels in drug discoveryNot unexpectedly, the emergence of atomic level structural information has rekindled interest in generating comparative (homology) models. As mentioned, this creates various assumptions that complicate the drawing of conclusions from these models. Therefore, a valid question would be: ‘do the benefits these models bring outweigh the difficulties?’ The similarities between the various ion channel families can be used to transfer ‘indirect’ evidence from one channel to another, as exemplified in a recent in-house example in which the design team was facing exactly this question. However, there was an abundant amount of structural and nonstructural data available: ▪ A large number of structure–activity relationships (SAR);▪ Small-molecule crystal structures;▪ NMR structures of molecules binding to the target;▪ Crystal structures of compounds bound in anti-targets/selectivity targets;▪ Site-directed mutagenesis (SDM) data on residues changing or eradicating compound activity;▪ Published crystal structures as described above.The initial step consisted of generating a comparative model using various available crystal structures and the SDM data. The main difficulty (often underestimated in the generation of comparative models) was the generation of a reliable multi-alignment of the various crystal structures with the target channel. Although there are programs available to kick-start the alignments, manual tweaking to improve the amino acid-based alignment is important. Once this step is accomplished, an initial model can be generated. Evaluation of the model may lead to changes in alignment with the templates and one or more iterations optimizing the comparative model. The NMR data, SDM data and the data on observed interactions between our compounds and residues in other crystal structures could then be used to come up with hypotheses of possible binding modes. Compounds can be docked flexibly or by utilizing small‑molecule crystal structure geometries to generate an initial docking pose (conformation). At this point, the various docked hypotheses were used to guide compound design. Due to the large number of assumptions that were used to generate the model and the docked orientations, the aim was not to make small changes to the molecules, but rather to guide the syntheses by identifying broad areas to build into or to stay clear off, starting from the docked-scaffold molecule. Numerous compounds were designed that would improve interactions with the modeled channel, and some compounds were added that should not be accommodated. After screening, the results from this iteration were fed back into the model. After several iterations of proposing compounds and updating the model based on the readouts, the project ended up with highly active compounds that would not have been picked up easily by simple small‑molecule overlays.In the literature, numerous applications of comparative models have also been described. In some cases, such as hERG [23], the modeling approaches were not yet precise enough to give sufficient predictive power to guide a late-stage drug-development program. On the other hand, modeling of the Kv1.5 binding modes resulted in the construction of a pharmacophore model that could be used in a rational drug-design approach [24].In the case of blockers for the eukaryotic shaker K+ potassium channel, this was taken one step further and a virtual library of compounds was automatically docked in the binding pore of a comparative model based on the KcsA crystal structure. Molecular dynamics were then performed on the best compounds (inhibitors) in order to cut the number of compounds down and only select the best ones for synthesis and testing, resulting in six active compounds (30% hit rate) [34].The examples above show the impact comparative models can have on drug-discovery projects. These kinds of computational approaches open up avenues for exploring broad changes in order to introduce compounds or select the best compounds from a large compound collection. However, this methodology currently lacks the predictive power to guide a late-stage drug-discovery project where only small tweaks are made to the molecules. In order to use these approaches successfully, additional data (structural or otherwise) will help greatly. SDM on the channel of interest is a very valuable tool to pinpoint the binding site within the channel as well as to guide docking methodologies.Future perspectiveIf the current trend continues, more and more crystal structures will become available for more and more ion channels. With greater numbers of similar structures, comparative models will get more reliable as multiple crystal structures can be used to build the models. Furthermore, different states of these channels may get crystallized, leading to a better understanding of the structural implications the state changes of the ion channels will have.In a perfect world, high-resolution crystal structures would be obtained for the ion channels in which the pharmaceutical industry is really interested. This would not remove the need for computational chemistry approaches, but merely change the kind of calculations. For example, automated docking will become more common with the emergence of high-resolution crystal structures. This will still need critical evaluation as the scoring functions used in docking methodologies are known to present shortcomings and may not work for the compounds that are being docked.A more realistic perspective will be the emergence of more ion channel crystal structures, but not necessarily of prime pharmaceutical targets. Together with more SDM data, this will facilitate the alignments and improve the quality of comparative models. Some regions will be modeled with relative high confidence (e.g., transmembrane helical bundles), while other regions will still be of lower confidence (e.g., intra- and extra-cellular loops of different length and composition as those found in the crystal structures). The literature contains a variety of examples where the comparative model of proteins with low amino acid similarity with the crystal structures on which they were based gave reliable models that were used by the pharmaceutical industry in the design of compounds, often years before a crystal structure validated the comparative model (e.g., CYP2D6, where the crystal structure [35] was obtained several years after the models [36,37]).Regardless of which ion channel crystal structures will be obtained, additional structural data will still be a key determinant in pinpointing locations of binding sites (e.g., SDM), predicting possible binding orientations (e.g., molecules of interest cocrystallized in other proteins, showing interactions with similar amino acids) and possible compound conformations (NMR or small-molecule crystal structures). Most of these additional data have caveats (i.e., inactivity caused by SDM does not have to reflect disruption of a direct interaction of that amino acid with the compounds); conformations seen in other proteins only show options and small molecule crystal structures are governed by crystal packing effects. However, when used carefully and appropriately, these additional structural data will be able to guide some of the inherent assumptions in the comparative models.In addition, when screening data are becoming available for large numbers of compounds, small-molecule computational approaches can be used, either as a replacement for comparative models or ideally to support the comparative model by highlighting areas in molecules responsible for activity or inactivity (e.g., pharmacophores or comparative molecular field analysis [38]) or highlighting chemical motifs (fingerprints) to include or avoid in the compounds (e.g., fingerprint approaches, decision trees and Bayesian models). At all stages, it is also important to include at least some compounds in the iterative design cycles that will challenge the models (i.e., purposely synthesize and test a couple of compounds predicted to be less active).Financial & competing interests disclosureThe author is an employeee of Pfizer Ltd. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.No writing assistance was utilized in the production of this manuscript.Bibliography1 Cox JJ, Reimann F, Nicholas AK et al. An SCN9A channelopathy causes congenital inability to experience pain. Nature444,894–898 (2006).Crossref, Medline, CAS, Google Scholar2 Clare JJ. Targeting voltage-gated sodium channels for pain therapy. Expert Opin. Investig. Drugs19,45–62 (2010).Crossref, Medline, CAS, Google Scholar3 Doyle DA, Morais Cabral J. Pfuetzner RA et al. The structure of the potassium channel: molecular basis of K+ conduction and selectivity. Science280,69–77 (1998).Crossref, Medline, CAS, Google Scholar4 Roux B. Theoretical and computational models of ion channels. Curr. Opin. Struct. Biol.12,182–189 (2002).Crossref, Medline, CAS, Google Scholar5 Jakobsson E, Mashl RJ, Tseng T-T. Investigating ion channels using computational methods. Curr. Topics Membr.52,255–273 (2002).Crossref, CAS, Google Scholar6 Zhou Y, Morais-Cabral JH, Kaufman A, Mackinnon R. Chemistry of ion coordination and hydration revealed by a K+ channel–Fab complex at 2.0Å resolution. Nature414,43–48 (2001).Crossref, Medline, CAS, Google Scholar7 Zhou M, Morais-Cabral JH, Mann S, Mackinnon R. Potassium channel receptor site for the inactivation gate and quaternary amine inhibitors. Nature411,657–661 (2001).Crossref, Medline, CAS, Google Scholar8 Jiang Y, Lee A, Chen J et al. X-ray structure of a voltage-dependent K+ channel. Nature423,33–41 (2003).Crossref, Medline, CAS, Google Scholar9 Long SB, Tao X, Campbell EB, MacKinnon R. Atomic structure of a voltage-dependent K+ channel in a lipid membrane-like environment. Nat. Rev. Drug Discov..450,376–382 (2007).CAS, Google Scholar10 Kuo A, Domene C, Johnson LN, Doyle DA, Vénien-Bryan C. Two different conformational states of the KirBac3.1 potassium channel revealed by electron crystallography. Structure13,1463–1472 (2005).Crossref, Medline, CAS, Google Scholar11 Jiang Y, Lee A, Chen J, Cadene M, Chalt BT, MacKinnon R. Crystal structure and mechanism of a calcium-gated potassium channel. Nature417,515–522 (2002).Crossref, Medline, CAS, Google Scholar12 Alam A, Shi N, Jiang Y. Structural insights into Ca2+ specificity in tetrameric cation channels. Proceed. Natl Soc. Sci. USA104,15334–15339 (2007).Crossref, Medline, CAS, Google Scholar13 Sobolevsky AI, Rosconi MP, Gouaux E. X-ray structure, symmetry and mechanism of an AMPA-subtype glutamate receptor. Nature462,745–758 (2009).Crossref, Medline, CAS, Google Scholar14 Fujiwara Y, Minor DL Jr. X-ray Crystal structure of a TRPM assembly domain reveals an antiparallel four-stranded coiled-coil. J. Mol. Biol.383,854–870 (2008).Crossref, Medline, CAS, Google Scholar15 Jasti J, Furukawa H, Gonzalez EB, Gouaux E. Structure of acid-sensing ion channel 1 at 1.9Å resolution and low pH. Nature449,316–324 (2007).Crossref, Medline, CAS, Google Scholar16 Kawate T, Carlisle Michel J, Birdsong WT, Gouaux E. Crystal structure of the ATP-gated P2X4 ion channel in the closed state. Nature460,592–598 (2009).Crossref, Medline, CAS, Google Scholar17 Hilf RJC, Dutzler R. X-ray structure of a prokaryotic pentameric ligand-gated ion channel. Nature452,375–379 (2008).Crossref, Medline, CAS, Google Scholar18 Bocquet N, Nury H, Beaaden M et al. X-ray structure of a pentameric ligand-gated ion channel in an apparently open conformation. Nature457,111–114 (2008).Crossref, Medline, Google Scholar19 Miyazawa A, Fujiyoshi Y, Unwin N. Structure and gating mechanism of the acetylcholine receptor pore. Nature423,949–955 (2003).Crossref, Medline, CAS, Google Scholar20 Chothia C, Lesk AM. The relation between the divergence of sequence and structure in proteins. EMBO J.5,823–826 (1986).Crossref, Medline, CAS, Google Scholar21 Durell SR, Guy HR. Atomic scale structure and functional models of voltage-gated potassium channels. Biophys. J.62,238–250 (1992).Crossref, Medline, CAS, Google Scholar22 Lange W, Giebendörfer J, Schenzer A et al. Refinement of the binding site and mode of action of the anticonvulsant retigabine on KCNQ K+ channels. Mol. Pharmacol.75,272–280 (2009).Crossref, Medline, CAS, Google Scholar23 Bell IM, Bilodeau MT. The impact of IKr blockade on medicinal chemistry programs. Curr. Topics Med. Chem.8,1128–1139 (2008).Crossref, Medline, CAS, Google Scholar24 Yang Q, Du L, Wang X, Li M, You Q. Modeling the binding modes of Kv1.5 potassium channel and blockers. J. Mol. Graph. Model.27,178–187 (2008).Crossref, Medline, CAS, Google Scholar25 Duclohier H. Structure–function studies on the voltage-gated sodium channel. Biochim. Biophys. Acta.1788,2374–2379 (2009).Crossref, Medline, CAS, Google Scholar26 Browne LE, Blaney FE, Yusaf SP, Clare JJ, Wray D. Structural determinants of drugs acting on the Nav1.8 channel. J. Biol. Chem.284,10523–10536 (2009).Crossref, Medline, CAS, Google Scholar27 Pedretti A, Marconi C, Bettinelli I, Vistoli G. Comparative modeling of the quarternary structure for the human TRPM8 channel and analysis of its binding feature. Biochim. Biophys. Acta.1788,973–982 (2009).Crossref, Medline, CAS, Google Scholar28 Fernández-Ballester G, Ferrer-Montiel A. Molecular modeling of the full-length human TRPV1 channel in closed and desensitized states. J.Membrane Biol.223,161–172 (2008).Crossref, Medline, CAS, Google Scholar29 Sandoz G, Douguet D, Chatelain F, Lazdunski M, Lesage F. Extracellular acidification exerts opposite actions on TREK1 and TREK2 potassium channels via a single conserved histidine residue. Proceed. Natl Soc. Sci. USA106,14628–14633 (2009).Crossref, Medline, CAS, Google Scholar30 Kollewe A, Lau AY, Sullivan A, Roux B, Goldstein AAN. A structural model for K2P potassium channels based on 23 pairs of interacting sites and continuum electrostatics. J. Gen. Physiol.134,53–68 (2009).Crossref, Medline, CAS, Google Scholar31 Pietra F. Docking and MD simulations of the interaction of the tarantula peptide psalmotoxin-1 with ASIC1a channels using a homology model. J. Chem. Inf. Model.49,972–977 (2009).Crossref, Medline, CAS, Google Scholar32 Qadri YJ, Berdiev BK, Song Y, Lippton HL, Fuller CM, Benos DJ. Psalmotoxin-1 docking to human acid-sensing ion channel-1. J. Biol. Chem.284,17625–17633 (2009).Crossref, Medline, CAS, Google Scholar33 Lummis SCR, Beene DL, Lee LW, Lester HA, Broadhurst RW, Dougherty DA. Cis–trans isomerization at a proline opens the pore of a neurotransmitter-gated ion channel. Nature438,248–252 (2005).Crossref, Medline, CAS, Google Scholar34 Liu H, Gao A-B, Yao Z et al. Discovering potassium channel blockers from synthetic compound database by using structure-based virtual screening in conjucntion with electrophysological assay. J. Med. Chem.50,83–93 (2007).Crossref, Medline, CAS, Google Scholar35 Rowland P, Blaney FE, Smyth MG et al. Crystal structure of human cytochrome P450 2D6. J. Biol. Chem.281,7614–7622 (2006).Crossref, Medline, CAS, Google Scholar36 de Groot MJ, Ackland MJ, Horne VA, Alex AA, Jones BC. Novel approach to predicting P450 mediated drug metabolism. The development of a combined protein and pharmacophore model for CYP2D6. J. Med. Chem.42,1515–1524 (1999).Crossref, Medline, CAS, Google Scholar37 Lewis DFV. Homology modelling of human CYP2 family enzymes based on the CYP2C5 crystal structure. Xenobiotica32,305–323 (2002).Crossref, Medline, CAS, Google Scholar38 Cramer RD, Patterson DE, Bunce JD. Comparative molecular field analysis (CoMFA). 1. Effect of shape on binding of steroids to carrier proteins. J. Am. Chem. Soc.110,5959–5967 (1988).Crossref, Medline, CAS, Google ScholarFiguresReferencesRelatedDetailsCited ByIon channel drug discovery: challenges and future directionsA Wickenden, B Priest & G Erdemli29 March 2012 | Future Medicinal Chemistry, Vol. 4, No. 5 Vol. 2, No. 5 Follow us on social media for the latest updates Metrics History Published online 12 May 2010 Published in print May 2010 Information© Future Science LtdFinancial & competing interests disclosureThe author is an employeee of Pfizer Ltd. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.No writing assistance was utilized in the production of this manuscript.PDF download