harnessing human n-type ca2+ channel receptor by identifying the atomic hotspot regions for its...

DOI: 10.1002/minf.201200025

Harnessing Human N-type Ca2 + Channel Receptor byIdentifying the Atomic Hotspot Regions for Its Structure-Based Blocker DesignAshish Pandey,[a] Jigneshkumar P,[a] Satyaprakash Tripathi,[a] and C. Gopi Mohan*[a]

1 Introduction

Calcium (Ca2 +), potassium (K+) and sodium (Na+) ion chan-nels assemble in the membranes to form functional tetram-ers. K+ channels are formed by four a-subunit monomers,while for Na+ and Ca2 + channels, a single a-subunit poly-peptide with four internal hydrophobic repeats folds toform a functional tetrameric structure. Each repeat containssix TMhs (S1-S6), constituting two functional domainswhich include the voltage-sensing domain (S1-S4) and thepore-forming domain (S5-P-S6), respectively. S5-P-S6 TMhsconfers pore properties including channel selectivity, block-er specificity, and conductance in voltage gated Ca2 + chan-nels (VGCC).[1–4]

Ion channel therapeutics for many pathophysiologicalconditions exists, which include: affective and allergic disor-ders, autoimmune diseases, epilepsy, hypertension, insom-nia, pain, anesthesia, anxiety and stroke. Voltage gated N-type Ca2 + channel (NCC) plays dominant roles in neuro-pathic pain and cerebral ischemia. Experimentally, it waswell established that the NCC inhibitory activity is essentialfor the treatment of chronic neuropathic pain andstroke.[1,2] A major obstacle with this membrane proteinwas that the experimental three dimensional (3D) structuresolved by either X-ray or NMR technique was not yet avail-able, in order to understand the mode of small molecule(antagonist or agonist) binding at its active sites. Blockadeof NCC receptor has recently been shown in the treatmentof chronic pain associated with cancer, AIDS and neuropa-

thy. NCC will stabilize/destabilize its closed/open stategating mechanism by its blocker (antagonist) binding atthe ligand sensing residues and thus play an important rolein ion channelopathies.[1–3]

Many potent NCC blockers are in different stages of thedrug discovery program, such as

(i) Ziconotide (also known as SNX111 or Prialt) a US Foodand Drug Administration (FDA) approved drug, whichis a synthetic version of w-Conotoxin MVIIA derivedfrom marine snail Conus magus.

(ii) Morphine (also known as Avinza or Roxanol) isa potent narcotic pain reliever used to treat moderateto severe pain.

Abstract : The voltage dependent N-type Ca2 + channel(NCC) receptor was identified to have therapeutic potentialfor the treatment of neuropathic pain and stroke disease.The Ca2 + ion transport through the transmembrane influxis mainly dependent on the closing, opening, or intermedi-ate state gating mechanism of NCC. Harnessing this dy-namic gating mechanism at the structural level is an impor-tant and challenging physiological phenomenon. The threedimensional (3D) structure of this membrane receptor isnot yet experimentally determined to understand its mech-anism of action. Based on these observations, we have de-veloped for the first time the structure of the closed state

of the NCC receptor at the pore forming domains whichmainly involve three transmembrane helices (TMhs) S5, Pand S6. Hot-spot binding site residues of this receptormodel were identified by molecular docking techniqueusing amlodipine, cilnidipine and nifedipine compoundsknown to be potent Ca2 + channel antagonists. Further, theCa2 + ion permeability and the hydrophobic gating mecha-nism provided better structural and functional insights onthe NCC receptor. These results are in consonance withother Ca2 + channel receptors and would provide guidancefor further biochemical investigations.

Keywords: Ca2 + channel receptors · Homology model · Molecular docking · Amlodipine · Pain

[a] A. Pandey, J. P, S. Tripathi, C. Gopi MohanDepartment of Pharmacoinformatics, National Institute ofPharmaceutical Education and Research (NIPER), S.A.S. Nagar,Punjab 160 062, Indiaphone: + 91-172-2214682, fax: 0091-172-2214692Present address : Amrita Centre for Nanosciences and MolecularMedicine, Amrita Institute of Medical Sciences and ResearchCentre, Ponekkara, Kochi-682 041, Kerala State, Indiaphone: + 91-484-4008769, fax: + 91-484-2802120*e-mail : [email protected]

[email protected]

Supporting Information for this article is available on the WWWunder http://dx.doi.org/10.1002/minf.201200025

Mol. Inf. 0000, 00, 1 – 15 � 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim &1&

These are not the final page numbers! ��

(iii) Gabapentin (also known as Neurontin) drug was devel-oped for the treatment of epilepsy, but is now widelyused to relieve neuropathic pain.

(iv) NMED160 is a lead compound discovered by the Neu-romed company for future pain therapy, and

(v) ZC-88 a novel NCC blocker is known for the treatmentof acute and neuropathic pain. The dihydropyridine(DHP) class of compounds is also known as potentand selective Ca2 + channel blockers.[3,4]

Over the past two decades, technological advancementsin theoretical and experimental methods such as homologymodeling, NMR and X-ray crystallography has allowedstructure-based inhibitor design to attain a vital position inthe preclinical drug discovery program. Owing to the dra-matic increase in the availability of 3D structure of proteintargets and rapid advancement in computational tech-niques, structure- or ligand-based inhibitor design hasbecome an integral strategy for both lead generation andlead optimization.[5–12] Recently, we have developeda robust 2D-quantitative-structure-activity relationship(QSAR) model for structurally diverse non-peptidyl deriva-tives acting as potent NCC blockers.[13]

Many hotspots residues important for the binding ofVGCC antagonists especially for L-type Ca2 + channel (LCC)was identified, and substantial progress has been made incharacterizing its channel pore structure.[14,15] If availablecomputer models of S5-P-S6 TMhs are accurate for otherCa2 + channels, including KcsA and Na+ channel, then thepredicted hot-spot blocker sensing residues can be experi-mentally tested. Based on these observations, we carriedout homology modeling of human NCC receptor at S5-P-S6 TMhs (pore forming domain) in the closed state usingbacterial KcsA crystal structure as a template.

NCC three-dimensional (3D) structure provided for thefirst time the detail of its structure-based blocker sensingresidues and forms the main objective of the present study.Strategies for direct inhibition of NCC by peptide toxinsand small organic blockers have been realized only recently.Moreover, potential ways to target NCC specifically and se-lectively were not yet fully exploited. Further, experimentalstructure of LCC, RCC, PCC and other Ca2 + channels are notavailable which hinders the discovery of channel selective/specific blockers. Thus, the present structure-based NCCblocker study was just an initial step in this direction andfuture studies will potentially address these questions toidentify more effective channel specific small moleculedrugs.

The second objective of the present study was to identi-fy the antagonist hotspot binding site residues of theclosed state of human NCC using its 3D structural model.Recently, amlodipine a potent DHP Ca2 + channel antago-nist was experimentally tested for NCC inhibitory activityand key residues at different TMhs were identified.[14] Furu-kawa et al. studied inhibition of the dihydropyridine (DHP)derivatives towards different Ca2 + channels, and among

them amlodipine was known to be the most potent NCCblocker.[15] 3D structure-based blocker binding of amlodi-pine to NCC was computed by us using molecular dockingtechnique. Amlodipine, was docked at the ligand sensingregion in the closed state of the NCC receptor to under-stand its mode of binding and mechanism of NCC function.This antagonist was known to inhibit the transmembraneinflux of Ca2 + ions at the vascular smooth muscle and car-diac muscle for the treatment of hypertension and coro-nary artery disease. The contractile processes of these mus-cles are dependent on the movement of extracellular Ca2 +

ions into these cells through different channels.[15] Further,two other Ca2 + channel DHP blockers cilnidipine and nife-dipine were also docked to obtain better insight on themode of its binding with NCC ligand sensing residues.

NCC conductance and gating mechanisms were alsostudied by us using different computational techniques.The consistency of NCC-DHP’s predicted binding poseswith the structure-activity relationships and mutagenesisdata on KcsA and LCC receptor supports the confidence ofour predicted results. Altogether, the structure of NCCmodel provided for the first time an in depth detail of thehot-spot binding site residues of NCC in complex with dif-ferent DHPs which include amlodipine, cilnidipine and nife-dipine.

2 Materials and Methods

All computational experiments were carried out usingMODELLER8v2, SYBYL7.1, GOLD, APBS, HOLE and PyMOLmolecular modeling packages on a Sun workstation withRed Hat Enterprise Linux 3 and Silicon Graphics Fuel Work-station with IRIX 6.5 operating system.

2.1 Homology Modeling of Human NCC Receptor

The homology model of the NCC receptor at the S5-P-S6TMhs was constructed using MODELLER software.[6] Con-struction of protein models by homology modeling tech-nique involves well-defined steps:

(i) sequence alignment between the target (NCC at S5-P-S6) and the template (KcsA at M1-P-M2),

(ii) building an initial NCC crude model,(iii) refining the NCC model,(iv) and evaluating the quality of the final NCC homology

model using different structure validation tools.[16]

The target sequence of the a1B TMhs (S5-P-S6) of thehuman NCC receptor was retrieved as CAC1B_Human inthe SWISS-PROT database (accession number Q00975) andthe template sequence of the TMhs (M1-P-M2) of bacterialpotassium channel (KcsA) was retrieved from Swiss-Prot da-tabase (accession number P0A334). Homology modeling ofNCC receptor was performed using the crystal structure of

&2& www.molinf.com � 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Mol. Inf. 0000, 00, 1 – 15

�� These are not the final page numbers!

Full Paper Ashish Pandey et al.

www.molinf.com

KcsA protein solved at 3.2 �.[17] We have chosen this crystalstructure based on the anisotropic thermal refinement lead-ing to its better structural quality, reported by Chen et al.[17]

This technique was successfully employed in differentmembrane proteins.[18] Also, KcsA was used as a templatefor the present study, because it is well accepted that theKcsA architecture might describe the pore of K+ , Ca2 + , Na+

and other ion channels, thus confirming an evolutionarypredecessor of 6 TMhs ion channels.[19]

2.2 NCC Model Structure Refinement and Validation

Initially, 100 NCC models of S5-P-S6 TMhs were generatedusing MODELLER, and model 77 having the best DOPEscore and molpdf score was selected by us for structure re-finement and validation. The Ramachandran plot of the77th NCC model showed some steric hindrance with 0.9 %residues in its disallowed region as explained below. Inorder to further improve the structure quality, this NCCmodel was subjected to MOE refinement using Energyrefine module in MOE program. Employing coarse modelrefinement, different models were constructed by the Boltz-mann-weighted randomized modeling procedure. Each ofthese intermediate models was then subjected to a suffi-cient degree of minimization to relieve any serious stericclashes. AMBER99 force field was used for this purpose andNCC model quality was further evaluated by Ramachandranplot in the PROCHECK validation package. The main pur-pose of energy minimization was to eliminate undesirablesteric interactions (bad contact) of the NCC structure andthereby eliminating any residues in the disallowed regionof the Ramachandran plot. The MOE report of the NCC re-ceptor model having best packing quality was generatedshowing its stereochemical significance which includes pla-narity, chirality, phi/psi preferences, chi angles, non-bondedcontact distances, unsatisfied donors and acceptors. In thisreport, the value of the final NCC model was comparedwith the high resolution structures from PDB and its statis-tical significance is presented. Finally, the NCC geometricquality of the backbone conformation, the residue interac-tion, the residue contact and the energy profile of thestructure was shown to be within the restrictions estab-lished for reliable structures.

2.3 Molecular Docking Study

The molecular docking technique is helpful in understand-ing the NCC receptor ability to bind with different DHP li-gands (blockers), which is an important phenomenon inmany biological processes. We performed docking ofpotent antagonists-amlodipine, cilnidipine and nifedipinewhich are known to be effective NCC/LCC blockers. Experi-mentally, the binding site of the ligand sensing domain ofthe NCC receptor was reported by Miyashita et al.[14] Fur-thermore, identical/similar ligand sensing residues of the

LCC receptor with the corresponding NCC receptor resi-dues were identified by sequence alignment techniques.[19]

Ligand docking was performed at the active site of theNCC receptor model using GOLD 5.0.1 (Genetic Optimiza-tion for Ligand Docking) program.[20] GOLD uses a geneticalgorithm (GA) for docking ligands into the protein bindingsites in a flexible manner, the so called induced fit dockingprocedure. The binding affinity of the ligands with the pro-tein is represented in terms of the GoldScore and Chem-Score fitness function. The total GoldScore, which is repre-sented as “Fitness,” was calculated from the contribution ofhydrogen bonds and van der Waals interactions betweenprotein and the ligand, as well as the contribution of theintra-molecular hydrogen bonds and strains experienced bythe ligand. The annealing parameters of van der Waals andhydrogen bonding interactions were considered within6.0 � and 3.0 � respectively. All other parameters were keptas default values in GOLD.

We extended our docking analysis using Glide ver5.5software[21] as well for judging the quality of the antago-nists-NCC mode of interactions. The results obtained usingboth these docking techniques were then used to identifythe hotspots binding site residues in the NCC receptormodel. Using Glide software, 20 � cubic boxes were usedas the search region within which ligands must fit in orderto be scored. We also defined the binding site regions inthe NCC as reported by Miyashita et al.[14] OPLS_2005 (opti-mized potential for liquid simulation) all atom force fieldwas used, and other parameters were kept at default forthe Glide docking. Antagonists (Amlodipine, cilnidipine andnifedipine) with the best SP Glide score were then dockeda second time using the Extra Precision (XP) mode of Glide,which was more rigorous, as it takes into account polarityand hydrophobicity of the environment during dockingprocess.

2.4 Ca2 + Ion Permeability Study

Ion permeation across the channel is affected by the differ-ence in the potential at the channel interface and the aque-ous environment.[22,23] Pore analysis was performed usingthe HOLE program to determine the channel attributes,and which allows the determination of the pore dimen-sions using its structural coordinate file.[24] It sets the vander Waals radius of each atom record and then proceedsalong the NCC from the initial point in the direction of thechannel vector, such that the largest sphere can be accom-modated without overlap with the van der Waals surface ofany atom. The algorithm uses a Monte Carlo simulated an-nealing procedure to find the best route for a sphere withthe variable radius to squeeze through the channel (flexiblesphere positions “squeezing” through the ion channel indiscrete steps), to analyze the anisotropy within the NCCpore in the direction of the channel vector. This process isvisualized using visual molecular dynamics (VMD 1.8.6)graphics package.

Mol. Inf. 0000, 00, 1 – 15 � 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.molinf.com &3&


Harnessing Human N-type Ca2 + Channel Receptor

www.molinf.com

POREWALKER program was used to find the NCC poreresidues and coordinates along the NCC pore. An ideaabout the NCC gating and selectivity filter regions wascomputed using Adaptive Poisson-Boltzmann Solver (APBS)software (Ver-1.3). It uses finite difference Poisson�Boltz-mann method to calculate the free energy of solvation forany ion to pass through the channel pore.[23,24] The NCCmodel was centered at the origin and aligned along the Z-axis. Using the PDB2PQR-1.7 software, protein charge andatom radius (PQR) file was generated from the structure co-ordinate file for which the values were determined for thechannel protein. The temperature around the molecule wasset at 300 K and dielectric constant of 78.5 respectively.Since the protein of our interest is an ion selective, the em-ulating ion used was Ca2 + with a radius of 1.68 �. The elec-trostatic profile was evaluated and its contribution to thefree energy of transfer was calculated as,

DEPB ¼ Ecomplex�Eion�Eprotein

Calculations were performed by moving Ca2 + ions alongthe pore axis in discrete steps. Mobile ions move to mini-mize the free energy of the NCC system, so that theirneighboring positions can be determined by finding pointswhere the total energy of the system is minimized to attainits stable state.

3 Results and Discussion

3.1 Construction of NCC Transmembrane Helices

In VGCCs, it is a well established fact that the flow of Ca2 + ,K+ and Na+ ions are affected by the changes in the mem-brane action potentials, which include ion channel conduc-tion, selection and gating functions. But what do the chan-nels look like? How do they conduct ions? How do themetabolic cues regulate ion channel gating? These ques-tions could not be answered until the knowledge of the 3Dstructure of the ion channels was revealed. The first crystalstructure of potassium channel KcsA from the bacteriumStrptomyces lividans was reported two decades ago.[25] Be-cause both prokaryotic and eukaryotic pores were closelyrelated, the channel’s architectural design was essentiallythe same and evolutionary preserved for higher organismsas well. On the basis of this, it was rational to use the crys-tal structure of KcsA as a template to model the humanNCC receptor, and to discuss various structural insightsthus obtained.[26]

In order to understand the conserved motifs present inthe pore-forming domain of different human VGCCs, multi-ple sequence alignment of different channels such as NCC(denoted as a1B, Cav2.2) with L-type Ca2 + channel (LCC, de-noted as a1C, Cav1.2), P/Q-type Ca2 + channel (PCC, denoted

Figure 1. Multiple sequence alignment of four TM regions of N-type, L-type, P/Q-type, and R-type Ca2 + channels along with KcsA and Na+

-channel. Ligand (DHP’s) sensing residues are marked as boxes. Solid black boxes represent conserved residues, dashed black boxes repre-sent active site residues and clouded red boxes represent Ca2 + ion selectivity filter. General sequence numberings of the four TMhs arealso shown.




www.molinf.com

as a1A, Cav2.1), KcsA and Na+ channel were performed, andis depicted in Figure 1. Analysis of Ca2 + , Na+ and K+ chan-nel sequences at the pore-forming domain has enabled thedesign of different biochemical experiments in order to ex-plore the relationship between the sequence motifs andthe various aspects of its physiological functions.[19,26]

It was observed that there is a high sequence identity/similarity in TMhs at S5-P-S6 pore forming domains inthese four channels and was pictorially depicted inFigure 1. The percentage sequence similarity (identity) be-tween NCC and other three VGCCs (at the S5-P-S6 TMhs)i.e. LCC was 88.86 % (67.97 %), PCC was 98.05 % (94.71 %)and RCC was 96.65 % (91.36 %) respectively. The amino acidsequence length used for this estimation was: 359 residuesfor NCC, LCC and RCC; 358 residues for P/QCC; 360 resi-dues for KcsA and 359 residues for Na+ channel respective-ly, and is depicted in Figure 1. The higher degree of NCCsequence similarity with respect to LCC, RCC and P/QCCchannels is consistent with the significant conservation ofboth its biophysical and pharmacological properties.[19,27]

The percentage sequence similarity (identity) between NCCand KcsA was 44.8 % (17.49 %), and those between NCCand Na+ channel was 58.76 % (30.91 %) respectively. Thesignificance of the sequence alignment between NCC andKcsA was calculated using the BLAST program withBlosum45 matrix and the expectation value (E-value) ob-tained was 9.00 E-04, as expected. Since, there is low se-quence identity/similarity between NCC and KcsA.

Site directed mutagenesis experiments on LCC receptorwas extensively carried out and these studies provided thebasis of the present investigations in the identification of

hotspots binding site residues in NCC receptor. Dihydropyr-idine (DHP) derivatives such as amlodipine and cilnidipineare reported to bind at this pore forming region (S5-P-S6),and thereby act as potent NCC/LCC blockers.[14,28]

Figure 2 shows the sequence alignment of NCC (S5-P-S6)TMhs with the template KcsA (M1-P-M2) TMhs at the pore-forming as proposed by Lipkind and Fozzard.[12] For clarity,all the four TMhs of KcsA and NCC were marked in Figure 2and Table 1 respectively. Sequence alignment of this TMhsis also represented in different form in Figure 1s (Support-ing Information). The most important part of any homologymodeling study is the proper alignment of the sequence ofthe test protein (NCC) with that of the template protein(KcsA). The atomic coordinate of the KcsA crystal structurein the closed state was downloaded from the protein data-bank (PDB code: 2QTO). The central pore of NCC is formedby the S6 segment of each subunit and by the extracellularregion between S5 and S6 segments (P-loop) that deepensinto the pore, forming the extracellular mouth of the chan-nel. Sequence alignment of human NCC at S5-P-S6 TMhsand bacteria KcsA (template) at M1-P-M2 TMhs wereshown in Figure 2. The alignment shows that the TMhs ofNCC shares 32 % sequence similarity with the TMhs ofKcsA. It is important to mention at this juncture that gener-al rule w.r.to sequence alignment is not feasible in the pres-ent study due to low sequence identity/similarity betweenNCC and KcsA. So based on the existing knowledge, wemanually aligned KcsA template sequence (359 aminoacids) with NCC sequence (360 amino acids). The alignmentshowed the only single gap at the IIIS6 TM helix, and is pre-sented in Figure 1 and Figure 1s (Supporting Information)

Figure 2. Sequence alignment of NCC at (S5-P-S6) with the template KcsA at (M1-P-M2) pore-forming TMhs. Residue numbering is depict-ed in two ways: (i) genuine number of first and last residues of each TMhs as shown in the Uniprot databank; (ii) relative numbers of resi-dues (see text). Black boxes correspond to key mutational residues causing disease conditions and underlined correspond to selectivityfilter residues of KcsA.




www.molinf.com

respectively. Broadly, the similar sequence alignment ap-proach was earlier adopted by other research groups forstudying the structure–function relationship of different ionchannels.[19,26,28] Further, modelling extracellular loop re-gions between different TMhs was not quite feasible dueto the non-availability of the experimental structure atthese regions.

Model construction at the central pore (S5-P-S6 TMhs) ofthe NCC was challenging due to the low sequence identitybetween KcsA and NCC. We aligned the NCC sequence ofS5-P-S6 TMhs (359 residues) with that of the correspondingM1-P-M2 TMhs (360 residues) of KcsA, and which is in ac-cordance with that of the LCC receptor model publishedearlier.[14,28] Recent experiments on the construction of theNCC chimeric channels and mutagenesis experiments onNCC/LCC demonstrated that the IIIS5, IIIS6, and IVS6 TMhsinteraction with the known NCC/LCC antagonists.[14,29]

Since modeling of NCC may be relevant to understandthe structures of various voltage gated Ca2 + and Na+ chan-nels, a general nomenclature is desirable to designate resi-dues in the S5-P-S6 TMhs, as first proposed by Zhorovet al.[28] We assign a relative number of 1 to the first residuein each TMhs of KcsA. Positions in the TMhs of NCC alignedwith KcsA (Figure 2) are designated by using the numberof repeats and TMhs along with the relative number of thehomologous positions in KcsA.[28] Thus, Thr1653 of NCCfrom the human is designated ThrIVP.48 since it occupies rela-tive position 48 in the segment IVP (Figure 2). Repeatnumber is omitted in the designation of residues of K+

channel that have four identical subunits. Similar designa-tions are used for KcsA that has only two TMhs. Since theP-loops of voltage gated ion channels are vastly different inlength, counting relative positions from either the N- or C-

terminus of the P-loop segments is not feasible. The selec-tivity filter glutamate (E) was assigned 50 by us; the mostconserved residues in the P-loops of the VGCCs, and usethis as a marker for counting the residues in the aligned se-quences (Figure 2), which is in accordance with the earlierreport on LCC28. For example, Glu1655 and Glu1659 in therepeat IV of the NCC P-loop are designated as GluIVP.50 andGluIVP.54 respectively. It is well known that an interface be-tween domains III and IV is the prime binding site for Ca2 +

channel antagonists, shown in Figure 3.Most precisely of NCC mutagenesis studies, it emerges

that in IVS6 segment: IleIVS6.11, PheIVS6.12, LeuIVS6.19 andAsnIVS6.20 are critical for the DHPs binding. For IIIS6 segment-TyrIIIS6.10, PheIIIS6.11, PheIIIS6.14, PheIIIS6.16, PheIIIS6.17 and ValIIIS6.19

residues are important, while in the case of IIIS5 segment-TyrIIIS5.14 and MetIIIS5.18 residues will interact with DHPs.[14]

However, mutational data an important source of informa-tion for identifying the hot-spot binding sites need to betreated carefully. Mutated protein has 3D structural changeat the site of mutation and nearby sites, which in turn willinfluence its structure-based ligand design studies. Further,it can hinder movement distal to the binding site, therebyaltering the ligand binding or the downstream signalingmechanism, without necessarily changing direct interac-tions with the ligand. Nevertheless, various mutationalstudies on LCC receptor revealed that its IVS6 segment:Tyr1508 (TyrIVS6.11), Met1509 (MetIVS6.12) and Ile1516 (IleIVS6.19)(shown as black boxes in Figure 1) residues are thought todirectly participate in the DHP binding.[13] For IIIS6 segment,five LCC residues- Tyr1169 (TyrIIIS6.10), Ile1170 (IleIIIS6.11),Ile1173 (IleIIIS6.14), Met1177 (MetIIIS6.18) and Met1178 (MetIIIS6.19)(shown as black boxes in Figure 1) influence the binding ofDHPs. While in IIIS5 segment- Thr1066 (ThrIIIS5.14) and

Table 1. Molecular docking analysis of amlodipine with the NCC receptor in NCC-amlodipine (R) complex; residue numbering in parenthe-ses obtained from Uniprot sequence database. Residues were identified using 2D-interaction map from Discovery studio module. The con-served residues of NCC with other ion channels (L, P/Q, R, KcsA and Na+) are also given in parentheses.

Hydrogen-bonding interactions probability be-tween NCC-amlodipine

Van der Waals interactions probability be-tween NCC-amlodipine

Residues showing probable solvent ac-cessible surface

ThrIVP.48 (1653) (N, L, P/Q, R, KcsA) MetIS6.18 (347) (N, P/Q) ThrIVP.48 (1653)GlyIVP.49 (1654) (N, L, P/Q, R) MetIVS6.18 (1699) (N, P/Q, R, Na+) PheIIIS6.22 (1411)GluIVP.50 (1655) (N, L, P/Q, R) LeuIS6.19 (348) (N, L, P/Q, R) AsnIIS6.15 (697)GluIIP.51 (663) (N, L, P/Q, R) PheIVS6.22 (1703) (N, L, P/Q, R) LeuIIS6.18 (700)GlyIIP.50 (662) (N, L, P/Q, R) PheIIIS6.22 (1411) (N, L, P/Q, R, Na+) MetIP.49 (313)ThrIIP.48 (661) (N, L, P/Q, R) ValIIIS6.19 (1408) (N, L, P/Q, R) GlyIIP.50 (662)MetIP.49 (313) (N, L, P/Q, R) LeuIIS6.18 (700) (N, L, P/Q, R) SerIVS6.15 (1696)AsnIIS6.15 (697) (N, L, P/Q, R) LeuIIP.47 (660) (N, L, P/Q, R) ThrIIIP.48 (1363)Ca2 + ion PheIIIS6.11 (1400) (N, P/Q, R)

GlyIIS6.14 (696) (N, L, P/Q, R, KcsA, Na+)ThrIIIP.48 (1363) (N, L, P/Q, R, KcsA)ThrIIS6.11 (693) (N, P/Q, R)GluIP.50 (314) (N, L, P/Q, R)ThrIP.48 (312) (N, L, P/Q, R, KcsA, Na+)SerIS6.15 (344) (N, L, P/Q, R, Na+)SerIVS6.15 (1696) (N, P/Q, R)ProIIIS6.15 (1404) (N, P/Q, R)GlyIIIP.49 (1364) (N, P/Q, R)




www.molinf.com

Gln1070 (GlnIIIS5.18) residues of LCC receptor are importantfor DHP binding.[14,28] Recently, the importance of domainIII on NCC receptor as a primary controller of Ca2 + channeldeactivation was studied by Yarotskyy et al. using kineticand pharmacological techniques.[29]

Thus, all the above mentioned residues and mutationalstudy performed by Miyashita et al.[14] directly indicate thatDHP’s interactions with NCC/LCC by binding between IIIS6and IVS6 TMhs respectively. Further, it is important to men-tion that most of the above mentioned key LCC residue isalso conserved in NCC obtained by sequence alignmentwith different VGCCs, as depicted in Figure 1. All the abovementioned observations were taken into consideration indeveloping the homology model of NCC receptor and itsstructure-based antagonist (blocker) binding study.

The M1-P-M2 TMhs of KcsA has extra-cellular regionswhich deepen into the pore forming a narrow region of~12 �, and is lined by main chain carbonyl oxygens of theselectivity filter sequence.[25] The selectivity filter sequenceTXGYG in KcsA allows only the passage of K+ ions. On theother hand, Ca2 + selectivity filter of NCC is made by theside chains of four glutamates (EEEE locus), and these resi-dues are broadly conserved among other VGCCs, which in-clude LCC, PCC and RCC receptors respectively (Figure 1).The selectivity-filter of NCC contains negatively chargedresidues and three Ca2 + ions were added to the different P-loop regions (IP, IIP, IIIP and IVP) to interact with theseacidic residue pairs, which include GluIP50…Ca2 +…GluIVP50,GluIIP50…Ca2 +…GluIIIP50 and AspIP54…Ca2 +…AspIIP51 respec-tively. In the third/fourth repeat (MetIIIP54/GluIVP54) of the P-loop region in NCC, MetIIIP54 is present, and this residue was

Figure 3. Homology model of NCC at (S5-P-S6) TMhs. Active site residues are marked with sticks and the Ca2 + ion selectivity filter regionis highlighted. Important residues of the template KcsA crystal structure (PDB code: 2QTO) involved in mutations[19] (shown as light greensticks) along with the corresponding residues of the NCC model (shown as white sticks). For clarity only some of the TMhs are marked.




www.molinf.com

not favorable for binding with the Ca2 + ion. Thus, we haveincluded only three Ca2 + ions in our NCC model for furtheranalysis.

The homology model structure of the NCC was obtainedfrom the MODELLER software, as explained earlier. 100 dif-ferent NCC models were generated and among this 77thmodel were chosen as the best model on the basis ofDOPE score and molpdf score respectively. NCC model re-finement was performed on the 77th NCC model usingenergy refine option implemented in MOE software.[30] Em-ploying coarse model refinement, different NCC modelswas constructed by the Boltzmann-weighted randomizedmodeling procedure. Each of these intermediate modelswas then subjected to a sufficient degree of minimizationto relieve any serious steric clashes. After structural refine-ment using MOE, the NCC geometric quality of the back-bone conformations, the residue interactions, the residuecontacts and the energy profile of the structure was wellwithin the restrictions established for reliable structures.The quality of the NCC homology model before and afterMOE refinement in the closed state was assessed by Rama-chandran plot in the PROCHECK validation package,[5] andis depicted in Figure 2s (Supporting Information). We ob-served an improvement in the stereochemical structuralquality of the NCC model in the Ramachandran plot afterMOE refinement, i.e. 91.6 % residues were in the allowedregion, 6.7 % were in the additionally allowed region, 1.7 %were in the generously allowed region and no residues inthe disallowed region. But, before MOE refinement 90.9 %residues were in the allowed region, 7.4 % were in the addi-tionally allowed region and 0.9 % were in the generously al-lowed as well as in the disallowed regions respectively. Thisindicated that after MOE refinement the backbone dihedralangles f and y of the final NCC homology model have im-proved showing no residues in the disallowed region (Sup-porting Information Figure 2s). We also computed the G-factor for NCC model, which provide a measure of hownormal or unusual a stereochemical property is (dihedrals,bond lengths and angles), with values below �0.5 beingunusual, and values below �1.0 being highly unusual. Ourfinal NCC model had an overall G-factor of 0.11 showing itsstereochemical significance. Further, different structural pa-rameters generated using MOE software of the NCC modeland its expected values were shown in Table 1s (SupportingInformation), depicting the good stereochemical signifi-cance and model structure quality. In order to further testthe structure and stability of our NCC model (after MOE re-finement) we performed 10 ns MD simulation study usingGROMACS4.5 software.[31] During simulation the root meansquare deviation of the backbone atoms and the structuralcomparison w.r. to its simulation time showed stable fold-ing conformation for NCC model (data not shown).

Figure 3 shows the final homology model structure ofthe closed state of the NCC receptor at (S5-P-S6) TMhs.Active site residues are marked with sticks and the Ca2 +

ion selectivity filter region was also highlighted. It also de-

picts important residues of the template KcsA crystal struc-ture (PDB code: 2QTO) involved in mutations[19] (shown aslight green sticks) alongwith the corresponding residues ofNCC model (shown as white sticks). The top view of thefinal NCC model showing three Ca2 + ions is depicted inFigure 3s (Supporting Information).

The resulting NCC model was then used to dock threedifferent potent antagonists- amlodipine, cilnidipine and ni-fedipine respectively to shed light on its binding mode.Thus, we developed NCC homology model and experimen-tal data available for KcsA, NCC and LCC receptor gave usthe confidence to propose our 3D-model as a valuable plat-form for structure-based blocker (antagonist) design studiesfor the treatment of chronic pain and neurological disor-ders.

3.2 DHPs Docking and Interaction Analysis

Knowledge of the structure of the inner channel pore is im-portant for understanding the mechanisms of ion permea-tion, drug binding, and gating mechanism. The inner poreof VGCCs, is formed by four homologous S6 segments withnon-identical repeats. Large sized organic compounds suchas phenylalkylamines, benzothiazepines, and DHPs are ob-served to bind within or close to the ion channel conduc-tion pathway. DHPs are well characterized by differentsecond and third generation agents and the VGCC bindingdomains of these drugs were extensively probed by radio-labeled ligands.[32,33] Also, VGCCs has the ability of dynamicmovement from a closed state to an open state, and it wasobserved that tetramethyl ammonium compound havinga diameter of 6 � can pass through the channel pore andthereby revealing the pore size.[34]

DHPs bind to a single site at which agonists increaseCa2 + channel activity and antagonists abate it.[35] Therefore,DHP antagonists are believed to block the pore indirectlyby stabilizing the channel closed state with a single Ca2 +

ion bound to its pore. In fact, site directed mutagenesis ex-periments confirmed that the binding of Ca2 + ion to theselectivity filter stabilizes the VGCC receptors in its high af-finity closed state.[36] Interestingly, our NCC model in com-plex with the antagonists also showed single Ca2 + ionbinding at its pore region, and is explained below in thedocking section.

To the best of our knowledge, two theoretical models ofLCC had been developed for structure-based inhibitordesign studies. Pioneering studies by Lipkind and Foz-zard,[11] Cosconati et al.[12] and Zhorov et al. ,[28] analyzedthoroughly the binding pose of few DHPs with the LCCmodel. To understand the mode of binding mechanism ofthe LCC model with DHPs, a manual docking procedurewas adopted by Lipkind and Fozzard. While other researchgroups performed studies on LCC model interactions withsmall compounds by Monte Carlo minimization methodand Autodock software.[12,13,28]




www.molinf.com

Goldmann et al. suggested different regions in DHP’s tounderstand its SARs i.e. the preferred conformation of theDHP ring will be specified as a flattened boat with C4 asthe bow, the axial aryl ring as the bowsprit, and the N1

atom as the stern.[36] Accordingly, the two sides of the DHPring will be designated as the port side (left) and the star-board side (right), and is shown in Figure 4 for amlodipinecompound. Similar nomenclature was adopted for Cilnidi-pine and Nifedipine structure as well (Figure 4).

The structural mechanism of NCC-blocker interactionsand the energetics of its molecular complexes, which isa key factor in understanding the biological function, wererevealed in our molecular docking study. We adopted twodifferent docking algorithms (GOLD and Glide) in order totest the binding mode of NCC with three different antago-nists: amlodipine-R, cilnidipine and nifedipine, in a morecomprehensive manner. Both these algorithms gave similarmode of binding of these three antagonists with the NCCreceptor. Thus, we have chosen in our future discussions

the docking results obtained for these three antagonistsusing GOLD program.

3.2.1 Amlodipine Binding Analysis with NCC

Amlodipine compound has five oxygen atoms that mayaccept H-bonds and/or coordinate metal ions with the re-ceptor (Figure 4a). A highly electronegative chlorine atomat the bow part of amlodipine forms a strong negativeelectrostatic potential, which can be neutralized by theCa2 + ion of the channel receptor.[37] Coordination of Ca2 +

by negatively charged residues glutamate (GluP50) in thepore regions of each TMhs is known to affect the DHPbinding in LCC receptor.[38,39]

Molecular docking derived binding conformation of am-lodipine showed that the plane of DHP ring is parallel tothe NCC pore axis, the starboard side of its heterocyclicring points upward, and the plane of its 4-aryl substituentis perpendicular to the pore axis (Figure 5). This orientationallows amlodipine to establish different favorable contactswith the NCC receptor and is presented in Figure 5 andTable 1 respectively. Amlodipine made strong hydrogenbonding interactions with ThrIVP.48 and AsnIIS6.15 residues ofthe NCC receptor model. These residues belong to the cat-alytic/selectivity region and the blocking of which may leadto impaired NCC receptor activity. Other closely hydrogenbond interacting residues of NCC with the amlodipine wasmainly from the pore regions of IP, IIP and IVP TMhs whichinclude MetIP.49, ThrIIP.48, GlyIIP.49, GluIIP.50, GlyIVP.49 and GluIVP.50

residues respectively (Table 1). These probable hydrogenbonding residue interactions and other weak interactionsin NCC-Amlodipine complex was identified from 2D-interac-tion map module of the Discovery studio program,[40] andis presented in Figure 4s (Supporting Information).

Other weak forces which are important in biological phe-nomenon was van der Waals (VDW) based interactions.VDW radii were used to describe atoms that form favorablepolar interactions and these radii overlap with each otheramong neighboring residues. We identified folowing NCCresidues: SerIS6.15, MetIS6.18, LeuIS6.19, ThrIP.48, GluIP.50, ThrIIS6.11,GlyIIS6.14, LeuIIS6.18, LeuIIP.47, PheIIIS6.11, ProIIIS6.15, PheIIIS6.22, ValIIIS6.19,ThrIIIP.48, GlyIIIP.49, SerIVS6.15, MetIVS6.18 and PheIVS6.22 making fa-vorable VDW interactions with amlodipine (Figure 5 andTable 1), and is also depicted in Figure 4s and Figure 5s(Supporting Information). It is important to note that mostof the hydrogen bonding and VDW based interacting resi-dues of NCC-amlodipine(R) complex are conserved amongdifferent VGCC receptors, suggesting an evolutionary rela-tionship among them (Figure 1). Thus, our docking analysisbroadly suggests the mode of binding of amlodipine withthe NCC receptor.

The conserved residues Thr and Trp at the pore (IP, IIP,IIIP and IVP) of the TMhs of VGCC receptors in each repeatwith the corresponding residues in Na+ channel receptorshas important functions in pharmaceutical applications, asthese channels have local anesthetic receptors.[19] SARs and

Figure 4. Chemical structure of amlodipine (a), cilnidipine (b), andnifedipine (c) showing different sides (portside, bowspirit, star-board and stern) adopted as general nomenclature of DHP drugsfrom the literature.[36]




www.molinf.com

mutagenesis experiments confirmed the importance ofthese residues in VGCCs. SARs studies indicated that the N1

hydrogen atom has a key role in the binding of the DHPsto NCC. Mutational analysis clearly demonstrated that Thrfrom the pore region of each TMhs repeat contributes inthe binding of DHPs through polar interactions.[19] Ourdocking result also showed MetIP.49 of NCC made hydrogenbonding interaction with amlodipine and ThrIP.48 made vander Walls interactions with amlodipine, as shown in Table 1and Figure 4s (Supporting Information). The NH group ofamlodipine forms hydrogen bond with ThrIVP.48 (Table 1 andFigure 5) of NCC. The portside COOMe group establishes fa-vorable van der Waals contacts with MetIVS6.18, ValIIIS6.19 and

PheIIIS6.11 residues, and is depicted in Figure 4s and Figure 5srespectively (Supporting Information).

As depicted in Figure 5, ester oxygen on the starboardside of the DHP ring forms hydrogen bond with theAsnIIS6.15 side chain. This is in accordance with the SAR dataon LCC receptor indicating the involvement of this groupin hydrogen bonding interactions with the channel.[41] SARdata indicate that only small-sized ester groups are tolerat-ed on the starboard side of the DHP ring.[41] The 4-aryl sub-stituent of the docked amlodipine is in close contact withthe ligand sensing residue PheIIIS6.11 by making favorablevan der walls interactions (Table 1 and Figure 5). Also, inthis case the involvement of PheIIIS6.11 in the binding of the

Figure 5. Amlodipine (R)-NCC docking based interactions. Amlodipine blocker and binding site (hot-spot) residues are shown in stickmodel (green color). Hydrogen bonding interactions are shown as blue dotted lines and the Ca2 + ion is marked as blue sphere. LCC resi-dues (blue color) involved in mutation[28] are also shown along with the corresponding NCC residues. Brown colored residues belong toNCC and blue colored residues belong to LCC.




www.molinf.com

NCC DHP antagonists was experimentally proven by muta-genesis experiments.[19]

Ester group on the port side of the DHP ring adopts a cisconformation with the double bond of the heterocyclicring. The trans-conformation of DHP does not appear to befeasible due to the unfavorable steric clashes that the largeport side esters would make with the IVS6 segment.Indeed, synthesis of DHP derivatives with an immobilizedester group demonstrated the preference of cis conforma-tion on the port side ester. The large lipophilic substituentat the port side ester of amlodipine in complex with NCCshowed favorable van der Waals interactions with MetIVS6.18

(Table 1 and Figure 5).[42]

When amlodipine stern is anchored to ThrIVP.48 andIleIVS6.11, the bow approaches TyrIIIS5.14, ValIIIS6.19 and AsnIIS6.15

residues respectively (Table 1 and Figure 5). Simple hydra-tion of stern-anchored amlodipine is unlikely because DHPsare lipophilic compounds.[43] Extensive SAR studies have un-ambiguously demonstrated that electron-withdrawing sub-stituent in the 4-phenyl ring enhances its activity in theortho and meta positions, while any substituent in the paraposition diminishes its activity.[36–41] Amlodipine compoundhaving Cl atom attached at the meta position of 4-phenylring (Bowspirit) was closure for second hydrophobic brace-let of NCC in NCC-Amlodipine complex (Figure 5 and Fig-ure 6a), and which is in good agreement with the SAR stud-ies. Further, it was speculated that Cl atom of amlodipinehas the power to neutralize the Ca2 + ion passing throughthe NCC channel interface due to strong electrostatic inter-actions between them, and this phenomenon needs to befurther confirmed. An important hydrogen bonding interac-

tion between the Ca2 + ion of NCC with the ester oxygenatom of amlodipine compound was observed in NCC-Amlo-dipine complex (Figure 5), and this would clearly restriction movements in NCC leading to its blocking action. Thisphenomenon could be biologically significant and restric-tion of Ca2 + ion movement in the closed state of NCC willbe useful for understanding pain and other neurologicaldisorders. Similar observation of charge neutralization atthe LCC pore region interface was also experimentally con-firmed earlier, in support of our docking study.[37–39]

3.2.2 Cilnidipine Binding Analysis with NCC

Cilnidipine compound has seven oxygen atoms that mayaccept H-bonds and/or coordinate metal ions with the NCCreceptor (Figure 4b). NO2 group of the bow part of cilnidi-pine form strong negative electrostatic potential, whichcan be neutralized by the Ca2 + ion of the NCC receptor.

Molecular docking derived binding conformation of cilni-dipine showed strong hydrogen bonding interactions withAlaIVP.47 and AsnIIS6.15 residues of NCC receptor model (Fig-ure 6s – Supporting Information). These residues are closeto the catalytic/selectivity region and the blocking of whichmay lead to impaired NCC receptor activity. Other closelyhydrogen bond interacting residues of NCC with the cilnidi-pine was mainly from the pore region TMhs which include,MetIP.49, ThrIP.48, ThrIIIP.48, GlyIIP.49, GluIP.50, SerIVS6.15, CysIVS6.14,GlyIIIP.49, ThrIVP.48 and AlaIVP.47 respectively (Table 2s SupportingInformation). Further, we identified- SerIS6.15, MetIS6.18,LeuIS6.19, ValIS6.22, ThrIIS6.11, TyrIIS6.16, LeuIIS6.18, PheIIIS6.11, PheIIS6.22,PheIIIS6.14, ProIIIS6.15, PheIIIS6.18, PheIIIS6.22, ValIIIS6.19, SerIIIP.47, GlyIVP.49,

Figure 6. Hydrophobic gating regions (a) and selectivity filter regions (b) of N-type Ca2 + channel.




www.molinf.com

GluIVP.50, SerIVP.46, IleIVS6.11, MetIVS6.18 and PheIVS6.22 residuesmaking VDW interactions with cilnidipine, and is shown inTable 2s and Figure 6s (Supporting Information) respective-ly. These results are broadly in agreement with the site di-rected mutagenesis experiments and NCC/LCC-blocker in-teractions reported earlier.[14,31] These probable hydrogenbonds and other weak interactions in NCC-Cilnidipine com-plex was identified from the 2D-interaction map module ofthe Discovery studio program,[40] and is presented in Fig-ure 6s (Supporting Information).

Thus, our docking analysis broadly suggests the mode ofbinding of cilnidipine with the NCC receptor in closedstate. It is important to note that most of the hydrogenbonding and VDW based interacting residues of NCC-cilni-dipine complex are conserved among different VGCC re-ceptors (Figure 1), suggesting an evolutionary relationshipbetween them.

MetIP.49 and ThrIP.48 of NCC made hydrogen bonding inter-action with cilnidipine in NCC-cilnidipine complex. Thisresult is in agreement with the mutational study showingthe importance of Thr residue in the pore region, and ispresented in Table 2s and Figure 6s respectively (Support-ing Information). NH group of cilnidipine forms hydrogenbond with AlaIVP.47 and this residue is close to ThrIVP.48 ofNCC. The portside COOCH2OH2Me group establishes favora-ble van der Waals contacts with MetIVS6.18, ValIIIS6.19 andPheIIIS6.11 residues respectively, shown in Figure 6s (Support-ing Information).

Also, methyl group on the starboard side of the DHPring forms hydrogen bond with the AsnIIS6.15 side chain andis shown in Figure 6s (Supporting Information). SAR dataon LCC receptor indicated that only small-sized estergroups are tolerated on the starboard side of the DHPring.[41] The 4-aryl substituent of the docked cilnidipine is inclose contact (van der walls interactions) with the ligandsensing residue PheIIIS6.11 of NCC (Table 2s and Figure 6s).

Ester group on the port side of the DHP ring adopts a cisconformation to the double bond of the heterocyclic ring.The trans-conformation of DHP does not appear to be fea-sible due to the unfavorable steric clashes that the largeport side esters would make with the IVS6 segment.Indeed, synthesis of DHP derivatives with an immobilizedester group demonstrated the preference for a cis confor-mation on the port side ester.

Large lipophilic ester substituent on the port side of cilni-dipine establishes hydrogen bonding interactions with GlyIIP.49 and ThrIIIP.48 residues (Table 2s and Figure 6s), and itadopts a cis conformation to the double bond of the heter-ocyclic ring to prevent unfavorable steric clashes that thelarge port side esters would make with the NCC IVS6 seg-ment.

When cilnidipine stern is anchored to ThrIVP.48 and IleIVS6.11,the bow approaches MetIP.49, ThrIP.48 and AsnIIS6.15 residues re-spectively (Table 2s and Figure 6s). The role of the highlyelectronegative NO2 group at the bowspirit was significantand it was located nearer to the second hydrophobic bra-

celet. Further, we speculate that the oxygen atom of NO2

group of cilnidipine has the power to neutralize the Ca2 +

ion passing through the NCC channel interface due tostrong electrostatic interactions between them and thisphenomenon needs to be further confirmed. An importanthydrogen bonding interaction between the Ca2 + ion ofNCC with the NO2 group in cilnidipine compound was ob-served (Figure 6s), and this interaction will further restrictthe ion movements (or ion competition) in NCC leading toits blocking action. Another key observation was NO2

group at the ortho position of the 4-phenyl ring in cilnidi-pine is broadly supported by SAR study, suggesting anelectron withdrawing group at this position is crucial forbetter activity. These phenomena will be biologically signifi-cant for understanding different neurological disorders.[36–41]

3.2.3 Nifedipine Binding Analysis with NCC

Nifedipine compound has six oxygen atoms that mayaccept H-bonds and/or coordinate metal ions with the re-ceptor (Figure 4c). NO2 group of the bow part of nifedipineform strong negative electrostatic potential, which can beneutralized by the Ca2 + ion of NCC receptor.

NCC-nifedipine complex showed strong hydrogen bond-ing interactions with AsnIIS6.15 residue and Ca2 + ion, shownin Figure 7s (Supporting Information). Other hydrogenbond interacting residues of NCC-nifedipine complex iden-tified from 2D-interaction map module was mainly fromthe pore region TMhs which include: MetIP.49, GluIP.50 ,ThrIP.48,GluIIP.50, ThrIIIP.48, GlyIIIP.49, GlyIVP.49 and GluIVP.50 residues respec-tively (Table 2s). Further, we identified VDW interactions ofNCC residues with nifedipine: SerIS6.15, LeuIIS6.18, ThrIIP.48,GlyIIP.49, AlaIVP.47, PheIIIS6.11, PheIIIS6.14, ProIIIS6.15, PheIIIS6.18,PheIIIS6.22, ValIIIS6.19, ThrIVP.48, SerIVS6.15 and MetIVS6.18 respectivelyshown in Table 2s and Figure 7s (Supporting Information).Further, most of these residues are also conserved amongdifferent VGCC receptors, suggesting an evolutionary rela-tionship between them.

NCC-nifedipine complex showed MetIP.49 and ThrIP.48 resi-dues making hydrogen bonding interactions as shown inTable 2s and Figure 7s (Supporting Information). The NHgroup of nifedipine form hydrogen bond with ThrIP.48 ofNCC. The portside COOMe group establishes favorable vander Waals contacts with MetIVS6.18, ValIIIS6.19 and PheIIIS6.11 resi-dues respectively, and is depicted in Figure 7s (SupportingInformation).

Also, starboard side methyl group in nifedipine forms hy-drogen bond with ThrIP.48 side chain, shown in Figure 7s.Ester group on the starboard side of the DHP ring adoptsa cis conformation to the double bond of the heterocyclicring. The trans-conformation of DHP does not appear to befeasible due to the probability of the unfavorable stericclashes made by the ester group at the starboard side withthe NO2 group of nifedipine, and also with the NCC IVS6segment. The bowsprit region (4-aryl substituent) of the




www.molinf.com

docked nifedipine is in close contact (van der walls interac-tion) with the ligand sensing residue PheIIIS6.11 (Table 2s).

When nifedipine stern is anchored to AsnIIS6.15, the bowapproaches IleIVS6.11, ThrIVP.48 and MetIVS6.18 residues respec-tively (Table 2s and Figure 7s – Supporting Information). Animportant hydrogen bonding interaction between the Ca2 +

ion of NCC with the ester oxygen atom and the NO2 groupof nifedipine compound was observed (Figure 7s), and thiswould clearly restrict the ion movements in NCC leading toits blocking action. This result is in accordance with theNCC-Cilnidipine complexes described earlier.

Docking experiments described above for amlodipine,cilnidipine and nifedipine blockers led us to make a compa-rative SAR analysis of its binding with different ligand sens-ing residues of NCC receptor and is tabulated in Table 2s(Supporting Information). Further, the binding affinity ofthese blockers towards NCC was computed using Gold andGlide software in the form of docking scores. The dockingscore showed amlodipine having better affinity than othertwo blockers in the order of amlodipine>cilnidipine>nife-dipine. Further, an excellent correlation coefficient of 0.85between the docking scores computed using Gold andGlide software was obtained, and is presented in Table 3s(Supporting Information).

3.3 Sequence-Structure-Function Relationship of KcsA, LCCand NCC

Sequence analysis and mutagenesis study on KcsA and cal-cium channel receptors (LCC and NCC) at the pore formingdomains (M1-P-M2/ S5-P-S6) revealed several key residuesof pharmacological interest.[19] Mutation of SerM1.22Trp at M1domain of KcsA causes loss of channel function and benignfamilial neonatal convulsions (BFNC).[19] For simplicity, rela-tive residue numbering is adopted as shown in Figures 2and 3. Mutational residues at the pore (P) domain wereknown to cause various diseases i.e. residues essential forK+ selectivity- ThrP.49, ValP.50 and mutation of other two resi-dues GlyP.51Ser and TyrP.52Ser leads to long QT syndrome,BFNC, Jervell and Lange-Nielsen, Romano-Ward (RW) etc.Further, mutation of LeuM2.20Phe in the M2 domain of KcsAcaused RW and JLN inherited cardiac arrhythmias[44–46] (Fig-ures 2 and 3). The motif Thr-Thr-X-Gly-Tyr-Gly is well con-served in potassium channels and was important for K+ ionselectivity and its structure stability[25] (Figure 2). Here X de-notes any residue insertion in the motif at that particularposition.

The LCC mutagenesis study revealed few key residues atthe S5-P-S6 TMhs which are known for the channel func-tion and conductance. The pore (P) domain residues ThrIIIP.48

and TrpIIIP.52 are involved in key pharmaceutical functions, asthis channel had local anesthetic receptors.[47] Mutation ofThrIIIP.48Met decreased both the density of the functionalchannels and their unitary conductance.[48] The calcium se-lectivity pore (Glu-Glu-Glu-Glu) from IP, IIP, IIIP and IVPrepeat at GluP50 was affected by mutating GluIVP50Asp or

GluIVP50AlaIVP50 residue. Further, S6 domains are associatedwith the channel gating and other pharmacological pro-cesses. Point mutation at IIS6 i.e. ValIIS6.27Ala decreases thechannel function, and IleIIS6.25Thr cause retinal disorder bychanging the channel activation[49] . Mutational effect onDHP binding to LCC revealed many residues from IIIS5, IIIS6and IVS6 domains. These mutated residues includeThrIIIS5.14Tyr, GlnIIIS5.18Met, TyrIIIS6.10Ala, IleIIIS6.11Ala, IleIIIS6.14Ala,MetIIIS6.18Ala, MetIIIS6.19Ala, TyrIVS6.11Ala, MetIVS6.12Ala,IleIVS6.18Met, IleIVS6.19Ala and AsnIVS6.20Ala respectively[28]

(Figure 5).By sequence analysis and docking study we identified

few key residues in NCC which were important for thechannel stability, function and conductance. The highlyconserved sequence at IIP among different VGCCs involvePhe-[GlnArg]-x(2)-Thr-x-Glu-x-Trp-Asn-x-ValMetTrp consen-sus motif. As mentioned earlier, the Ca2 + ion selectivityfilter region GluP50 (EEEE locus) and ThrP48 is conservedamong different VGCCs. Mutation of ThrP48 to Met andGluP50 to Asp or Ala affect the channel selectivity and con-ductance[19] . Further, by sequence analysis key residues Ileand Val in IIS6 domain are found to be conserved amongNCC and LCC receptors (Figure 2). So we speculate that thebiological significance of mutation of these residues werecrucial for NCC function, which is in agreement with that ofthe LCC receptor mentioned earlier (Figure 5).

A comparative analysis on NCC in complex with thethree different antagonists (Figures 4s, 6s and 7s – Support-ing Information) revealed significant insight on the ligandsensing residues for the channel function. Most of theseresidues are from IIIS, IIIP and IVS domains i.e. (i) MetIVS6.18 inNCC made van der Waals interaction with our studied an-tagonists and the corresponding residue in KcsA wasPheM2.23 and at LCC was IleIVS6.18 respectively. These residuesare important for the channel activity (Figures 1 and 5). (ii)SerIVS6.15 made polar hydrogen bonding interaction with cil-nidipine and van der Waals interaction with amlodipineand nifedipine. Corresponding residue in KcsA was IleM2.15

and at LCC was AlaIVS6.15 respectively (Figures 1 and 2) be-longing to the active site region.[33] (iii) IleIVS6.11 in NCC madevan der Waals interaction with cilnidipine. The correspond-ing residue in KcsA was MetM2.11 (pore facing residue) andLCC was TyrIVS6.11. This residue on mutation with Ile decreas-es the affinity of LCC[28] (Figures 1 and 5). (iv) ValIIIS6.19 inNCC made van der Waals interaction with amlodipine, cilni-dipine and nifedipine DHP’s. The corresponding residue inKcsA was GlyM2.19 (pore facing residue) and LCC was Me-tIIIS6.19 which on mutation with Ala decreases the affinity ofLCC[28] (Figures 1 and 5). (v) SerIIIP.47 in NCC made van derWaals interaction with cilnidipine and the correspondingconserved residue in LCC is also its key ligand sensing resi-due.[31] (vi) Similarly we noted that PheIIIS6.11 and PheIIIS6.14 inNCC and the corresponding residues in KcsA and LCC wereimportant for its ligand binding[14,28] (Figure 1).




www.molinf.com

3.4 Ca2 + Ion Permeability Analysis of Closed State of NCC

Pore dimension of any ion channels can provide a quantita-tive analysis of ion channel gating, ion selectivity and con-formational state of the channel, by the knowledge of thePoisson-Boltzmann electrostatic profile (ESP), which in turnpotentially guide the nature of its ion permeability.[50] APBS,POREWALKER and HOLE program was used by us to ana-lyse these dynamic processes in the closed state of theNCC receptor.[24,51,52] POREWALKER was used to identify thechannel lining residues and the coordinates along the NCCpore, whereas HOLE can determine the nature of the NCCpore area for the Ca2 + ion passage.

The positive and negative ESPs inside the closed state ofthe NCC was measured, which in turn is related to theradius of the pore dimensions through which Ca2 + ion per-meation takes place. The ESPs experienced by an emulatingCa2 + ion was interesting and it showed positive ESP andnegative ESP along the NCC interface. We also noticed thatat the two hydrophobic bracelets (gating), there is a strongtendency of repulsion of Ca2 + ion due to the high positiveESP between (i) Ca2 + ion and Trp residues at different NCCTMhs near the selectivity filter region and (ii) Ca2 + ion andhydrophobic residues at the bottom of the constricted areaof NCC. HOLE program identified this constricted area con-firming the closed state of our developed NCC model. Thisarea is lined with a ring of four hydrophobic residuesLeuIS6.19, LeuIIS6.19, ValIIIS6.19 and LeuIVS6.19 respectively, leadingto its high positive potential, also termed as a first hydro-phobic bracelet in NCC (Figure 6a). This makes the unfavor-able situation for Ca2 + ion movement further inside thechannel. Thus, it is speculated that these hydrophobic resi-dues lining the pore would impede Ca2 + ion permeationby a process called “hydrophobic gating”. In NCC model,the hydrophobic residues LeuIS6.19, LeuIIS6.19, ValIIIS6.19, andLeuIVS6.19 are conserved in Ca2 + and Na+ channels formingthe pore-facing bracelet (Figure 1). The bracelet may playa crucial role in the mechanism of gating modulation byDHPs. Mutations in the bracelet affect the binding of DHPsto LCC, as well as the binding of antagonists and steroidalagonists to Na+ channels[28] . However, a negative ESPmean force is also experienced by the Ca2 + ions in NCCdue to the four conserved glutamate residues (EEEE) pres-ent at the selectivity filter (Figure 2). This region strongly at-tracts the Ca2 + ions towards the channel pore due tohighly negative electrostatic forces.

HOLE program provided different colored regions withrespect to NCC receptor as shown in Figure 6b. The differ-ent colored regions inside the channel pore include: (i)blue region- correspond to the wide channel area havingenough passage for the ions, (ii) green region – correspondto the restricted channel area having a limited passage ofthe ions and (iii) red region- correspond to the hydrophobicgating where the channel radii is not enough for the pas-sage of the ions (Figure 6b). Thus, these observations fur-

ther support the NCC channel gating, selectivity filter andclosed conformational state of the NCC receptor.

4 Conclusions

The homology model of the closed state of the NCC recep-tor at S5-P-S6 TMhs was built for the first time in order tocarry out structure-based blocker (antagonist) design. Theaccuracy of the model was validated using different tech-niques. Identification of the NCC receptor putative hot-spotbinding site residues which mainly include- ThrIVP.48, GlyIVP.49,GluIVP.50, AlaIVP.51, TrpIVP.52, AsnIVS6.20, AsnIIS6.15, and MetIIIS5.18

using sequence and structure analysis with the template(KcsA) and LCC was established. Further, molecular dockingof amlodipine, cilnidipine and nifedipine antagonists re-vealed other putative key residues of NCC receptor in-volved in channel blocking. Thus, the present work formsthe basis for further molecular modeling and biochemicalstudies on targeting NCC receptor for pain and stroke ther-apy.

Altogether, our study provided for the first time a detaileddescription of the main interaction of closed state of theNCC receptor with amlodipine, cilnidipine and nifedipine(potent DHP antagonists) respectively. The consistency ofthe predicted binding poses with SARs and site-directedmutagenesis on some key residues in KcsA, LCC and NCCsupports the feasibility of our computed results. The Ca2 +

ion conduction and permeability study also provided usefulinsight for the first time about the nature of electrostaticpotential behavior along the NCC passage. The presentstudy can thus be carried forward for the possible dynamicnature of NCC at the closed state, intermediate state andan open state for understanding its function in pain andstroke therapy.

Although computer-predicted protein structures are stillnot as accurate as X-ray structures, the 3D modeled struc-ture of NCC presented here can be used to understand andtest hypotheses about its biological function. Particularly,the present homology model provides useful insights forunderstanding the mechanism of the closed state of NCC-blocker interactions at the structural level. The model canalso serve as a structural frame for conducting site-directedmutagenesis experiments and docking studies or as a foot-ing for encouraging novel strategies in developing newdrugs to treat ion channel disorders.

References

[1] J. G. McGivern, Drug Discov. Today 2006, 11, 245 – 253.[2] T. Yamamoto, A. Takahara, Curr. Top. Med. Chem. 2009, 9, 377 –

395.[3] C. G. Mohan, A. Pandey, J. Mungalpara, in Advances in the

Studies on Ion Channels and Their Blockers (Ed: S. P. Gupta),Springer, Heidelberg, 2011, 289 – 305.




www.molinf.com

[4] A. Takahara, S. Fujita, K. Moki, Y. Ono, H. Koganei, Hypertens.Res. 2003, 26, 743 – 747.

[5] a) A. Sali, M. A. Marti-Renom, A. C. Stuart, A. Fiser, R. Sanchez,F. Melo, Annu. Rev. Biophys. Biomol. Struct. 2000, 29, 291 – 325;b) K. S. Brian, K. K. Brian, Tr. Pharmacol. Sci. 2012, 33, 268 – 272.

[6] J. Vaibhav, P. Saravanan, A. Akanksha, C. G. Mohan, Chem. Biol.Drug Des. 2011, 77, 373 – 387.

[7] W. Steffen, B. Marcus, H. Udo, S. J�rgen, G. Klaus, FEBS Lett.2008, 582, 3335 – 3342.

[8] S. Horrick, C. Xiaolin, K. B. John, J. Chem. Inf. Model. 2012, 52,515 – 544.

[9] F. Pietra, J. Chem. Inf. Model. 2009, 49, 972 – 977.[10] J. Carlsson, R. G. Coleman, V. Setola, J. J. Irwin, H. Fan, A.

Schiessinger, A. Sali, B. L. Roth, B. K. Shoichet, Nat. Chem. Biol.2011, 18, 769 – 778.

[11] G. M. Lipkind, H. A. Fozzard, Mol. Pharmacol. 2003, 63, 499.[12] S. Cosconati, L. Marinelli, A. Lavecchia, E. Novellino, J. Med.

Chem. 2007, 50, 1504 – 1513.[13] J. Mungalpara, A. Pandey, V. Jain, C. G. Mohan, J. Mol. Model.

2010, 16, 629 – 644.[14] Y. Miyashita, T. Furukawa, E. Kamegaya, M. Yoshii, Eur. J. Phar-

macol. 2010, 632, 14 – 22.[15] T. Furukawa, T. Yamakawa, T. Midera, T. Sagawa, Y. Mori, T.

Nukada, J. Pharmacol. Exp. Ther. 1999, 291, 464 – 473.[16] L. R. Forrest, C. L. Tang, B. Honig, Biophys. J. 2006, 91, 508 –

517.[17] X. Chen, B. K. Poon, A. Dousis, Q. Wang, J. Ma, Structure 2007,

15, 955 – 962.[18] A. J. Mason, S. L. Grage, S. K. Straus, C. Glaubitz, A. Watts, Bio-

phys. J. 2004, 86, 1610 – 1617.[19] P. Guda, P. E. Bourne, C. Guda, Biochem. Biophys. Res. Commun.

2007, 352, 292 – 298.[20] M. L. Verdonk, J. C. Cole, M. J. Hartshorn, C. W. Murray, R. D.

Taylor, Proteins 2003, 52, 609 – 623.[21] R. A. Friesner, J. L. Banks, R. B. Murphy, T. A. Halgren, J. J. Klicic,

T. Daniel, M. P. Repasky, E. H. Knoll, M. Shelley, J. K. Perry, J.Med. Chem. 2004, 47, 1739 – 1749.

[22] R. Oliva, G. Calamita, J. M. Thornton, M. Pellegrini-Calace, Proc.Natl. Acad. Sci. USA 2010, 107, 4135.

[23] O. Beckstein, K. Tai, M. S. P. Sansom, J. Am. Chem. Soc. 2004,126, 14694 – 14695.

[24] M. Pellegrini-Calace, T. Maiwald, J. M. Thornton, PLOS Comp.Biol. 2009, 5, 1 – 16.

[25] D. A. Doyle, J. M. Cabral, R. A. Pfuetzner, A. Kuo, J. M. Gulbis,S. L. Cohen, B. T. Chait, R. MacKinnon, Science 1998, 280, 69 –77.

[26] a) B. S. Zhorov, D. B. Tikhonov, J. Neurochem. 2004, 88, 782 –799; b) G. M. Lipkind, H. A. Fozzard, Biochemistry 2000, 39,8161 – 8170; c) D. B. Tikhonov, I. Bruhova, B. S. Zhorov, FEBSLett. 2006, 13, 6027 – 6032; d) A. Rossokhin, G. Teodorescu, S.Grissmer, B. S. Zhorov, Mol. Pharm. 2006, 69, 1356 – 1365;e) S. Y. Wang, J. Mitchell, D. B. Tikhonov, B. S. Zhorov, G. K.Wang, Mol. Pharm. 2006, 69, 788 – 795; f) A. Giorgetti, P. Carlo-ni, P. Mistrik, V. Torre, Biophys. J. 2005, 89, 932 – 944.

[27] T. P. Snutch, NeuroRx 2005, 2, 662 – 670.[28] B. S. Zhorov, E. V. Folkman, V. S. Ananthanarayanan, Arch. Bio-

chem. Biophys. 2001, 393, 22 – 41.

[29] V. Yarotskyy, G. Gao, B. Z. Peterson, K. S. Elmslie, J. Neurophy-siol. 2012, 107, 1942 – 1951.

[30] Molecular Operating Environment (MOE 2004.03), ChemicalComputing Group, 1255 University St. Suite 1600, Montreal,Quebec, Canada H3B 3X3, 2004.

[31] H. J. C. Berendsen, D. van der Spoel, D. R. Van, Comput. Phys.Commun. 1995, 91, 43 – 56.

[32] a) S. Yamaguchi, B. S. Zhorov, K. Yoshioka, T. Nagao, H. Ichijo,S. Adachi-Akahane, Mol. Pharmacol. 2003, 64, 235 – 248; b) H.Glossmann, D. R. Ferry, A. Goll, M. Rombusch, J. Cardiovasc.Pharmacol. 1984, 6, S608 – S621.

[33] G. H. Hockerman, B. Z. Peterson, B. D. Johnson, W. A. Catterall,Annu. Rev. Pharmacol. Toxicol. 1997, 37, 361 – 396.

[34] S. Tang, G. Mikala, A. Bahinski, A. Yatani, G. Varadi, A. Schwartz,J. Biol. Chem. 1993, 268, 13026 – 13029.

[35] P. Hess, J. B. Lansman, R. W. Tsien, Nature 1984, 311, 538 – 544.[36] S. Goldmann, J. Stoltefuss, Angew. Chem. Int. Ed. Engl. 1991,

30, 1559 – 1578.[37] V. A. Govyrin, B. S. Zhorov, in Ligand-Receptor Interactions in

Molecular Physiology, St. Petersburg, Russia 1994, in Russian.[38] B. Z. Peterson, W. A. Catterall, J. Biol. Chem. 1995, 270, 18201 –

18204.[39] J. Mitterdorfer, M. J. Sinnegger, M. Grabner, J. Striessnig, H.

Glossmann, Biochemistry 1995, 34, 9350 – 9355.[40] Discovery studio 2.5 Guide, Accelrys, San Diego, 2009.[41] a) D. J. Triggle, D. Rampe, Tr. Pharmacol. Sci. 1989, 10, 507 –

511; b) D. J. Triggle, Curr. Pharm. Des. 2006, 12, 443 – 457.[42] B. Z. Peterson, B. D. Johnson, G. H. Hockerman, M. Acheson, T.

Scheuer, W. A. Catterall, J. Biol. Chem. 1997, 272, 18752 –18758.

[43] L. G. Herbette, Y. M. H. Vant Erve, D. G. Rhodes, J. Mol.Cell Car-diol. 1989, 21, 187 – 201.

[44] W. Zareba, A. J. Moss, G. Sheu, E. S. Kaufman, S. Priori, G. M.Vincent, J. A. Towbin, J. Benhorin, P. J. Schwartz, C. Napolitano,W. J. Hall, M. T. Keating, M. Qi, J. L. Robinson, M. L. Andrews, J.Cardiovasc. Electrophysiol. 2003, 14, 1149 – 1153.

[45] K. Dedek, L. Fusco, N. Teloy, O. K. Steinlein, Epilepsy Res. 2003,54, 21 – 27.

[46] C. Chouabe, N. Neyroud, P. Guicheney, M. Lazdunski, G.Romey, J. Barhanin, EMBO J. 1997, 16, 5472 – 5479.

[47] S. Y. Tsang, R. G. Tsushima, G. F. Tomaselli, R. A. Li, P. H. Backx,Mol. Pharmacol. 2005, 67, 424 – 434.

[48] M. Hans, S. Luvisetto, M. E. Williams, M. Spagnolo, A. Urrutia,A. Tottene, P. F. Brust, E. C. Johnson, M. M. Harpold, K. A. Stau-derman, D. Pietrobon, J. Neurosci. 1999, 19, 1610 – 1619.

[49] A. Hohaus, S. Beyl, M. Kudrnac, S. Berjukow, E. N. Timin, R.Marksteiner, M. A. Maw, S. Hering, J. Biol. Chem. 2005, 280,38471 – 38477.

[50] F. Fogolari, A. Brigo, H. Molinari, J. Mol. Recognition 2002, 15,377 – 392.

[51] O. S. Smart, J. G. Neduvelil, X. Wang, B. A. Wallace, M. S. P.Sansom, J. Mol. Graph. Model. 1996, 14, 354 – 360.

[52] T. J. Dolinsky, J. E. Nielsen, J. A. McCammon, N. A. Baker, Nucle-ic Acids Res. 2004, 32, W665 – W667.

Received: May 12, 2012Accepted: July 18, 2012

Published online: && &&, 0000




www.molinf.com

harnessing human n-type ca2+ channel receptor by identifying the atomic hotspot regions for its...

Documents