Insights into molecular interactions between CaM and its inhibitors from molecular dynamics simulations and experimental data.

This article was downloaded by: [Massachusetts PRIM Board] On: 05 April 2015, At: 07:55 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Biomolecular Structure and Dynamics Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/tbsd20

Insights into molecular interactions between CaM and its inhibitors from molecular dynamics simulations and experimental data ac

b

b

Martin González-Andrade , Rogelio Rodríguez-Sotres , Abraham Madariaga-Mazón , José b

b

a

c

Rivera-Chávez , Rachel Mata , Alejandro Sosa-Peinado , Luis del Pozo-Yauner & Imilla I. d

Arias-Olguín a

Facultad de Medicina, Universidad Nacional Autónoma de México (UNAM), México Distrito Federal, CP 04510, México b

Click for updates

Facultad de Química, Universidad Nacional Autónoma de México (UNAM), México Distrito Federal, C.P 04510, México c

Unidad de Vinculación de la Facultad de Medicina, UNAM en el INMEGEN, Secretaría de Salud, México Distrito Federal, C.P. 14610, México d

Escuela de Medicina Ignacio Santos, Ciencias de la Vida, Instituto Tecnológico de Estudios Superiores de Monterrey (ITESM) Campus, Ciudad de México, México Distrito Federal, C.P 14380, México Accepted author version posted online: 23 Feb 2015.

To cite this article: Martin González-Andrade, Rogelio Rodríguez-Sotres, Abraham Madariaga-Mazón, José Rivera-Chávez, Rachel Mata, Alejandro Sosa-Peinado, Luis del Pozo-Yauner & Imilla I. Arias-Olguín (2015): Insights into molecular interactions between CaM and its inhibitors from molecular dynamics simulations and experimental data, Journal of Biomolecular Structure and Dynamics, DOI: 10.1080/07391102.2015.1022225 To link to this article: http://dx.doi.org/10.1080/07391102.2015.1022225

Disclaimer: This is a version of an unedited manuscript that has been accepted for publication. As a service to authors and researchers we are providing this version of the accepted manuscript (AM). Copyediting, typesetting, and review of the resulting proof will be undertaken on this manuscript before final publication of the Version of Record (VoR). During production and pre-press, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal relate to this version also.

PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any

Downloaded by [Massachusetts PRIM Board] at 07:55 05 April 2015

form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions

Publisher: Taylor & Francis Journal: Journal of Biomolecular Structure and Dynamics DOI: http://dx.doi.org/10.1080/07391102.2015.1022225


Insights into molecular interactions between CaM and its inhibitors from molecular dynamics simulations and experimental data

Martin González-Andrade*,1,3, Rogelio Rodríguez-Sotres2, Abraham Madariaga-Mazón2, José Rivera-Chávez2, Rachel Mata2, Alejandro Sosa-Peinado1, Luis del Pozo-Yauner3, and Imilla I. Arias-Olguín4

1

Facultad de Medicina, Universidad Nacional Autónoma de México (UNAM), México Distrito Federal, CP 04510, México. 2 Facultad de Química, Universidad Nacional Autónoma de México (UNAM), México Distrito Federal, C.P 04510, México. 3 Unidad de Vinculación de la Facultad de Medicina, UNAM en el INMEGEN, Secretaría de Salud, México Distrito Federal, C.P. 14610, México. 4 Escuela de Medicina Ignacio Santos, Ciencias de la Vida, Instituto Tecnológico de Estudios Superiores de Monterrey (ITESM), Campus Ciudad de México, México Distrito Federal, C.P 14380, México.

*Corresponding author. Dr. Martin González-Andrade, Facultad de Medicina, Universidad Nacional Autónoma de México (UNAM), México Distrito Federal, C.P 04510, México. E-mail address: [email protected]

ABSTRACT In order to contribute to the structural basis for rational design of calmodulin (CaM) inhibitors, we analyzed the interaction of CaM with 14 classic antagonists and two compounds


that do not affect CaM, using docking and molecular dynamics (MD) simulations, and the data were compared to available experimental data. The Ca2+-CaM-ligands complexes were simulated 20 ns, with CaM starting in the “open” and “closed” conformations. The analysis of the MD simulations provided insight intro the conformational changes undergone by CaM during its interaction with these ligands. Theses simulations were used to predict the binding free energies (G) from their H and S contributions, giving useful information about CaM-ligand binding thermodynamics. The G predicted for the CaM’s inhibitors correlated well with available experimental data as the r2 obtained was 0.76 and 0.82 for the group of xanthones. Additionally, valuable information is presented here: I) CaM has two preferred ligand binding sites in the open conformation known as site 1 and 4, II) CaM can bind ligands of diverse structural nature, III) the flexibility of CaM is reduced by the union of its ligands, leading to a reduction in the Ca2+-CaM entropy, IV) enthalpy dominates the molecular recognition process in the system Ca2+-CaM-ligand, and V) the ligands making more extensive contact with the protein have higher affinity for Ca2+-CaM. Despite their limitations, docking and MD simulations in combination with experimental data, continue to be excellent tools for research in pharmacology, towards a rational design of new drugs.

Keywords: calmodulin; protein-ligand interactions; protein-peptide interactions; MMPBSA; molecular recognition, docking, molecular dynamics; conformational flexibility.

1. Introduction


Molecular recognition and protein-ligand interaction in solution are dynamic processes involving transitions in conformational ensembles of receptor, ligand, and hydration states (Baron & McCammon, 2013). The conformational flexibility of proteins plays a key role in molecular recognition and calmodulin (CaM) is considered as a paradigm of protein flexibility (Junker & Rief, 2009). CaM is an ubiquitous calcium binding protein, and an important molecular target for drug design, because it regulates multiple pathways, critical to several biological functions (Sorensen & Shea, 1998); such as gene regulation, protein synthesis, fast axonal transport, smooth muscle contraction, secretion, growth, organelle tubulation, ion channel function, cell motility, and chemotaxis to mention a few. CaM is a protein of 148 residues formed by two domains containing two Ca2+-binding loops each, known as EF hands. These domains are separated by a long central helix giving a dumbbell-shaped structure to this protein. This protein has no enzymatic activity but acts in response to Ca2+ concentration to modulate a wide variety of target proteins. Upon Ca2+ binding, CaM exposes hydrophobic patches, thus making it possible its interaction with more than 300 target proteins, and regulates a large number of biochemical processes (Junker & Rief, 2009). This protein also participates in pathological processes, such as cancer, inflammation, and viral penetration. As CaM plays a key role in these previously described vital processes, it has become an important protein target for the development of new drugs, including antitumorals,

antipsychotics, antidepressants, muscle relaxants and local anaesthetics (Bouche, Yellin, Snedden, & Fromm, 2005; Chin & Means, 2000; Du, Szabo, Gray, & Manji, 2004; Seales, Micoli, & McDonald, 2006).Therefore, CaM should be considered as a multi-functional protein, and some classic drugs such as chlorpromazine (CPZ) or trifluoroperazine (TFP) may cause sideeffects, many of which are likely to arise due to their effects on CaM, as well as on other proteins, which are related directly or indirectly. With this in mind, it is clear that it should be


necessary to avoid possible side effects and potential toxicity of CaM inhibitors. In this context, studies of CaM-ligand interactions may help provide information towards a rational drug design of CaM inhibitors. These inhibitors may provided an insight into the role of CaM in many biological processes, Therefore, it may be questionable use these inhibitors as drugs of first election. Today, many new drugs are developed that are more specific towards CaM (e.g., N-(6aminohexyl)-5-chloro-1-naphthalenesulfonamide W7)..So, the study of CaM-ligand interactions at a molecular level has gained relevance, as reflected by more than 20,000 studies involving CaM, published in the last years. In addition, around one fifth of these publications are directly related to the development of drugs (data obtained in SciFinder® https://scifinder.cas.org). As part of the research for new anti-CaM compounds, several crystallographic and NMR studies have focused on how CaM interacts with different ligands including chemical inhibitors, peptides, and binding proteins. Given the high flexibility of CaM, the molecular recognition during its binding to ligands is expected to be a very dynamic process. For example, when a small molecule like a ligand approaches its target (for example, CaM), it encounters not a single, static structure, but rather a macromolecule in constant motion. Classically, the ligand may bind and stabilize a subgroup of the many conformations presented by its dynamic target, this event should lead to a reduction of the receptor’s entropy (Baron & McCammon, 2013). However, the

ligands can also induce conformations not typically sampled, when this one ligand is absent. In the case of CaM, the important conformational flexibility during binding of its ligands is illustrated in Figure 1 (Laine, Yoneda, Blondel, & Malliavin, 2008). The molecular recognition is an essential process in the receptor-ligand interaction; that can be better understood from the enthalpy-entropy reflecting the compensation, conformational flexibility, and the solvent distribution (Baron & McCammon, 2013; Dolenc, Baron, Missimer, Steinmetz, & van


Gunsteren, 2008). Within this context, molecular dynamics (MD) studies have become a fundamental tool for understanding protein-ligands and protein-protein interactions (Miao et al., 2012; Norberto de Souza & Ornstein, 1999; Thomas, Mao, & Elcock, 2013). Herein, we have reported thirty-two MD studies of different CaM-ligands complexes. The

following

ligands

were

used:

N-(3,

3-diphenylpropyl)-N'-[1-R-(3,

4-bis-

butoxyphenyl)ethyl]-propylene-diamine (AAA) (Harmat et al., 2000), chlorpromazine (CPZ) (Martin Gonzalez-Andrade, Figueroa, Rodriguez-Sotres, Mata, & Sosa-Peinado, 2009), Ca2+CaM-peptide 1 derived from the human II–spectrin (αII–spec) (Shifman & Mayo, 2002), trifluoroperazine (TFP) (M. Gonzalez-Andrade et al., 2011), N-(6-aminohexyl)-5-chloro-1naphthalenesulfonamide (W7) (Osawa et al., 1998), malbrancheamide (MBC), malbrancheamide B (MBC-B), isomalbrancheamide B (MBC-I), premalbrancheamide (MBC-P) (Figueroa et al., 2011),

14-methoxytajixanthone

(XAN-14),

15-acetyl

tajixanthone

hydrate

(XAN-15),

variecoxanthone A acetate (XAN-A), emericellin (EM), tajixanthone hydrate (TJX-H), tajixanthone (TJX) (M. Gonzalez-Andrade et al., 2011), and citrinin (CIT) (M. GonzalezAndrade et al., 2013) (Figure 2); in the conformations “open” and “closed” of the CaM (open 1CLL.pdb and closed 1A29.pdb). These MD studies were performed using the AMBER 99SB and GAFF force-fields (AMBER 12; sander as integrator) (Case et al., 2005). The predicted

binding energies (Gcal) were compared to the experimental available data for the complexes, showing correlations higher than 0.76. These thermodynamic estimations also provided valuable information about CaM´s conformational assemblies. Therefore, in this study we proved that MD studies can provide insights into the processes that govern molecular recognition of CaM and different types of ligands, which in combination with experimental data has allowed to infer binding properties, and thus establish the molecular basis for a rational design of inhibitors with


greater selectivity and specificity to reduce side effects of the new anti-CaM drugs.

2. Results and discussion MD is a powerful tool for modeling macromolecular systems such as protein-ligand complexes (R. Chen & Chung, 2013; Zhou & Madura, 2004). Scheme 1S illustrates a basic information flow of how MD was performed (Supporting Information). Most of the MD studies of CaM focused on the flexibility between the two globular domains, on their conformational plasticity and on the CaM’s mechanisms mediating its molecular recognition of proteins, peptides, drugs, and ions (Hou, Wang, Li, & Wang, 2011; Laine et al., 2008; Yang, Jas, & Kuczera, 2001). It can be concluded that the conformational changes adopted by CaM upon binding to different ligands are essential to achieve its regulatory effects. CaM protein is an excellent model for performing MD studies since it allows comparing or complementing theoretical studies with a large amount of available experimental data, thus facilitating a better understanding of the mechanisms involved in the molecular recognition of the protein with its different ligands. In this context, a recent review written by Baron and McCammon (Baron & McCammon, 2013) suggests that computational methods should include the estimation of free

energy and entropic effects for a better understanding of the molecular recognition processes undergone in solution by protein-ligand complexes. This task can be done using the following equation G=H-TS, where G indicates the degree of stability of a system which results from the balance of two different contributions H and S. H refers to the strength of interaction forces in the system, while S indicates the changes in the degree of disorder of the system. Therefore, it is important to calculate both enthalpy and entropy, in order to relate molecular Downloaded by [Massachusetts PRIM Board] at 07:55 05 April 2015

recognition with the affinity of ligands to their targets. Herein, we aimed to study the mechanism of interaction between CaM and a set of antagonists previously reported, in order to contribute to the design of more specific CaM inhibitors. To achieve our goal, we applied docking and MD simulations in order to compare experimental data for 16 ligands of CaM with the dynamics of binding, as well as to calculate the thermodynamic contributions of CaM-ligand complexes. Ligands used are considered to be classical antagonists of CaM (Figure 2), with dissociation constants (Kd) within the range of pM (αII–spec), nM (AAA, XAN-14, XAN-15, XAN-A, EM, TJX-H and TJX), and M (CPZ, TFP, W7, MBC, MBC-B, and MBC-I). In addition two compounds (MBC-P and CIT) which do not bind to CaM were included as negative controls.

2.1. Molecular docking studies Docking studies were performed for all CaM-ligands. For those complexes lacking structural experimental data support, the possible site of interaction of the ligands with CaM was predicted, and a theoretical inhibition constant (Ki) was estimated using the program AutoDock4. This program calculates a Ki from docking energy using the following equation: Ki = exp(G*1000)/RT), where G is the docking energy; R is the universal constant of ideal gas (

1.98719 cal K-1 mol-1), T is the temperature (here 298.15 K), and Ki is the estimated inhibition constant. The docking energy values of AutoDock4 comprise intermolecular energy (constituted by van der Waals energy, hydrogen bonding energy, solvation energy, and electrostatic energy). The predicted docked complexes selected were those conformations showing the lowest binding free energy. Table 1S presents the Ki and the estimated free energy of binding (EFEB) for all the complexes considered in this work. Four binding sites for CaM ligands had been identified, two


in each lobe (Vandonselaar, Hickie, Quail, & Delbaere, 1994). Docking rounds were performed for both open and closed CaM conformations. In the closed conformation all the ligands tested had a preferred binding pocket at the so called site 1 (Figure 3, blue surface), which has been identified as the binding pocket in most of the ligand-CaM complexes resolved to date by experimental methods. This site is formed by the amino acids Phe92, Ile100, Leu105, Met124, Ile125, Glu127, Ala128, Val136, Phe141, and Met144, on the C-terminal domain. In the open conformation the ligands docked preferentially at site 1, as before, or site 4 (Figure 3, green surface). Site 4 is formed by amino acids Phe19, Ile27, Leu32, Met51, Ile52, Glu54, Val55, Ile63, Phe68, and Met71 on the N-terminal domain. It must be noted that sites 1 and 4 are well separated in the open CaM conformation and get close to each other in the closed conformation. Both sites are hydrophobic, as only Glu127 in site 1 and Glu54 in site 4 form bind polar side chains. Each complex structure was then subjected to MD simulations with AMBER 12 force field, as described in the methodology. 2.2. Molecular dynamics simulations Figure 1S shows the changes in system energy along the simulation process carried out for one Ca2+-CaM-ligand complex as an example. The length of the MD simulations for each process was: 50 ps for heating to 298 K, 50 ps for density relaxation (constant volume), 500 ps

for system equilibration, and 20 ns for production. The 20 ns production period is a compromise between the computational cost of all simulations required for the study, and the time required to sample the conformational fluctuations in a highly flexible protein. The average computational time required for heating, density, equilibration, and production were: 2.32, 2.24, 46.32, and 14233.68 CPU-hours, respectively for the closed conformation, and 6.04, 8.10, 97.57, and 22111.13 CPU-hours, respectively, for the open conformation. MD simulations of the open


conformation require larger unit cells with a ~2.3 larger solvent to protein ratio. Figure 4 shows the RMSD (root mean square deviation) of the backbone atoms (C, Cα, N and O), relative to the initial conformation, for some representative complexes, starting in the closed conformation of CaM (Figure 4A), and the comparison between the closed and open complexes with some representative ligands (Figure 4B-E). The starting RMSD at the production period was roughly 1.5 Å for the closed complexes and 2 Å for the open complexes. For the closed complexes the RMSD did not exceed 5 Å, and reached a plateau after 5 ns (Figure 4A), except for some cases like the Ca2+-CaM-TXJ complex, where the plateau was reached after 15 ns. For the open Ca2+CaM-ligand complexes, the RMSD begins around 2 Å and large variations were observed in the RMSD reaching up to 14 Å in some complexes (Figure 4B-E) The closed complexes with larger changes in RMSD along the MD trajectory [(Max-Min)] were 1A29 (3.428 Å), 1A29-CPZ (3.294 Å) and 1A29-TJX (3.804 Å). In contrast, 1A29-αII-spec and 1A29-XAN15 complexes suffered small changes (1.321 and 1.596 Å, respectively). Direct comparison of the RMSD between open and closed conformation for the ligand-free CaM reveals important fluctuations along the simulation for both conformations (Figure 4B). Instead, for Ca2+-CaM-αII-spec complex in the closed conformation displayed only minor changes in the RMSD and remained stable, while in the open conformation the change was large and did not stabilize within the 20 ns

of simulation (Figure 4C). A similar trend is observed for Ca2+-CaM-W7 complex, despite being W7 a relatively small molecule, in comparison to the αII-spec peptide (Figure 4D). Finally, for the case of Ca2+-CaM-MBC-P complex where MBC-P was used as the negative control (no detectable affinity for the CaM) (Figueroa et al., 2011) the RMSD of both conformations showed the same kind of stochastic fluctuations observed for the free proteins, along of the 20 ns MD


trajectory (Figure 4E). Most of the crystallographic structures of Ca2+-CaM bound to ligands have RMSD values around 2 Å, changes larger than this value are usually related to conformational changes and therefore also associated thermodynamically with an important entropic component (Baron & McCammon, 2013; Xu, Yu, Wan, & Huang, 2011). Figure 5 shows a descriptive statistical analysis of the RMSD along the MD trajectory using a boxplot. For the closed conformation, the highest median reached was 3.5 Å, with a range from 1 to 5 Å. In contrast, for the open conformation the smaller median was 4.5 Å and the range was from 1 to 10 Å (Figure 5). The fluctuations observed for CaM in the closed conformation were smaller in magniyude and in their variance, indicating a more stable conformation, which was further enhanced by some ligands, particularly the αII-spec peptide, TFP, W7, MBC-B, and XAN-15, but not by CIT which does not bind to CaM in vitro. On the other hand, the higher medians and larger range in RMSD along the CaM MD starting in the open conformation reflect larger conformational changes, and the presence of the ligands had only a modest effect on theses parameters (Figure 5B). Figure 6, shows the fluctuation in the root mean square deviation per residue (RMSF), this is a measure of the average mobility of every amino acid in the protein along the MD. The RMSF value gives an indication of the flexibility/mobility of each segment of the proteins. As

expected, regions with regular secondary structure exhibit smaller RMSF values than loops, and the N- and C- terminal loops are particularly mobile. In the case of CaM in the closed conformation (Figure 6A) the three systems exemplified here (Ca2+-CaM, Ca2+-CaM-II-spec, and Ca2+-CaM-TFP) showed less mobility than the simulations in the open conformation (Figure 6B). It is worth noting that the simulations for Ca2+-CaM without ligands showed lower values of RMSF, in comparison with the complexes (Figure 6B), maintaining the trend of lower mobility


in regions corresponding to -helices, except for the central helix (Phe65-Phe92). These data are consistent with the known properties of CaM because the central helix suffers a large distortion when going from the open to the closed conformation. Another important parameter in the study of MD is the radius of gyration (RG). The analysis of the changes in the RG provided an additional measure of the global changes in the protein structure. The radius of gyration through the simulation gives information on the changes in the overall shape of the protein. A larger RG means an expanded structure, and in our system there is consistency, as the open conformation has higher RG values, which also fluctuate more than those of the closed conformation. This means that the ligands increase the structural stability to the protein in the closed conformation. Figure 2S illustrates RG vs time along the MD simulations of complex CaM, and shows that at 20 ns this parameter has reached a stable value. In order to show that the MD simulations of the complexes achieved a reasonable sampling, we took independent structures of Ca2+-CaM-II-spec complex at 15, 17 and 19 ns and started new independent simulations for these complexes for 5, 3 and 1 ns respectively to complete the 20 ns. Figure 3S shows the RMSD along the trajectories of these independent MD simulations, which followed a very similar trend. Though it was not the main interest of this

work, a 100 ns simulation of the transition from the open to the close form of Ca2+-CaM-IIspec was performed taking the final conformation after the first 20 ns. The total simulation time for this complex would be 0.12 µs, which is still well below the expected time to observe a full transition. In order to make the exploration more effective, we used a simulated annealing strategy, the system was heated for 1 ns to 423 K, cooled down to 298 K in two steps, and simulated to complete 20 ns, and the whole cycle was repeated. Figure 4S shows how the 1CLL-


II-spec complex starts to close, as the distance from the N- to the C-terminal domains is reduced from 4.3 to 3.9 nm, and the angle between two arbitrary planes, selected in the closed conformation to be nearly parallel, reduced from ~130 to 70 degrees. Both parameters indicate that this complex was about 25 % of its way between the open and the closed conformation. This simulation was not further extended, due to the computational cost involved, and because it was not the aim of the present work. This last result allowed us to conclude that simulations of 20 ns are considered suitable for our purposes because the binding free energies predicted by MM/PBSA will reflect the properties of either the open or the closed CaM conformations, longer MD simulations would be unlikely to give us better predictions (Hou et al., 2011).

2.3. Comparison between theoretical and experimental values of the binding free energy In this study we calculated binding free energies of all complexes using the MM/PBSA methods and normal mode analysis, which provides estimates of the enthalpy and entropy contributions, respectively. The strategy used to estimate the G, allowed a separate calculation of the energetic contributions (enthalpy-entropy compensations) associated with molecular recognition in the binding of ligands to proteins. This kind of estimation is informative, because

determining the thermodynamic signature of binding (including free energy, enthalpy, and entropy) is significantly more insightful than information based only on the change in the binding free energy, Therefore, the thermodynamic signature that underlies the receptor-ligand association provides more detailed information about the driving forces involved during proteinligand binding. The calculation of thermodynamic signatures of binding is also useful for the enhancement of the affinity of ligands towards their targets on a rational basis. Figure 7 and


Table 1 show the results and comparisons between available experimental data and the theoretically predicted values. The data were grouped according to the chemical nature of the ligand, because the group of xanthones clearly followed a different trend, hence, the grouping was required in order to observe the correlation between theory and experiment. For the different inhibitors (Figure 7A) a correlation of 0.76 was obtained and for the group of xanthones (Figure 7B) this parameter was 0.82. Clearly the chemical nature of the ligand is important, because the correlation was better for the homogenous group of the xanthones. On the other hand, calculated thermodynamic parameters indicate that the enthalpy is the component that contributes more to G in all cases, except for the negative control (Ca2+-CaM-CIT). In this last system the degree of disorder is the one that governs the process of molecular recognition between the CaM protein and its ligands. In particular, the Ca2+-CaM-II-spec complex showed the lowest binding free energy experimentally determined (-15 kcal/mol) and a theoretically calculated value of -20 kcal/mol; where both thermodynamic terms (H and TS) are large compared to the rest of the ligands, indicating a remarkable affinity of the peptide for Ca2+-CaM. In the case of Ca2+-CaMMBC-B and Ca2+-CaM-MBC-I complexes; experimentally MBC-B and MBC-I showed the same affinity for the protein (Kd of 4.8x10-6 M), and the predicted values of G from MD simulations of the closed CaM differ by 10 kcal mol-1, attributed to the enthalpy component. On

the other hand, the Ca2+-CaM-MBC-P complex gave a predicted Gcal, but experimentally, the binding of MBC-P to the protein was not detected. Nevertheless, the predicted G is higher compared to its three analogs that bind to calmodulin (MBC, MBC-B and MBC-I), and it must be taken into consideration that the complex was set up in silico at the start of the MD simulation, but as shown below, the complex was unstable, which reflects the lack of specificity and transient nature of this interaction. In the case of CIT our negative control, which does not


bind to the CaM, a predicted G could be calculated, but its theoretical binding energy was the most unfavorable (positive G). These results reinforce the value of the theoretical tools used in this study to estimate the binding energy of Ca2+-CaM-ligand complexes. 2.4. Process of ligand-protein interaction along the molecular dynamics During the MD runs, multiple interactions occur within the global system (protein, ligands and solvent), and it should be possible to correlate some amino acid-ligand interactions with the degree of affinities of the anti-CaM compounds. The ligands can both induce a conformational change in the macromolecule and/or stabilize a subset of conformations, corresponding to a reduction of the receptor entropy. Therefore, we decide to monitor ligandresidue contacts along the MD simulations for the ligands tested here. We identified generic contacts by calculating if any atom of each amino-acid residue in the protein was at less than 4 Å from any atom of the ligand, at every collected frame of the simulation. More elaborate search criteria can be used, if one wishes to focus on a specific subset of interactions, such as hydrogen bonds or salt-bridges, but this was not done here. Figure 8 shows graphs and structural models of interactions along the dynamics of some of the complexes. The graphs show the state (in contact or detached) of protein residues at 4 Å or

less from the corresponding ligand, as well as selected snapshots of the structure along the 20 ns MD simulations. In Figure 8A, the MD simulation of the Ca2+-CaM-MBC complex in the open conformation is shown. At the beginning of the simulation (~5000 ps) only residues 90 and 120 participate in the contact, and subsequently many residues at the C-terminal domain participate (residues 81 to 148). In the case of the Ca2+-CaM-MBC-P complex in the open conformation (Figure 8B), the set of initially interacting residues lies down in the C-terminal domain like with


MBC, however when the complex MBC-P reaches nearly half of the MD simulation (~ 9 ns) this ligand loses all interaction with the protein (Movie S1). Therefore the binding energy shown in the previous section refers to the first 9 ns of the simulation, and is clearly not low enough to make the interaction stable and eventually promote the formation of the closed complex. This result is consistent with experimental data, because MBC-P is the only analogue of the malbrancheamides unable to binds to CaM, as monitored using the biosensor hCaM-M124CmBBr (Figueroa et al., 2011). In panel C and D of Figure 8, MD data is presented for the open conformation of Ca2+-CaM-CPZ and Ca2+-CaM-II-spec complexes, respectively. During the full MD trajectory of these last complexes, the residues corresponding to the C-terminus stayed in contact with the ligands, in contrast to the Ca2+-CaM-MBC-P complex. As expected in its closed conformation the protein interacts with the ligands with participation of residues from both N and C-terminal domains, but there are differences between each ligand. For example, for the Ca2+-CaM-CPZ complex the C-terminal residues contact the ligand with higher frequency that the N-terminal ones (Figure 8E). Furthermore, in the case of CaM-II-spec complex, residues from both lobes of the protein (Figure 8D and Movie S2) do have an extensive participation in the contact in agreement with the high affinity of the peptide (Kd = 2.5 pM) for this protein. From all these results, some general principles can be derived: I) the proximity of

the ligands to specific areas of the macromolecule is crucial, II) an interaction with a larger number of residues usually correlates with a higher the degree of affinity (a reduced entropy component), and III) interactions with longer residence time between the protein and the ligand


are indicative of complex stability (with highly favorable enthalpy component).

3. Conclusions In this work, we presented data based on MD simulations of Ca2+-CaM complex with sixteen ligands, these simulations gave reasonable predictions for the theoretical binding energy, from the enthalpy and entropy contributions. In general terms, the results showed an acceptable correlation with experimental data (r2 of 0.76 and 0.82), as long as the nature of the compound is taken into account. In addition, this study also provides valuable information about the mechanisms of Ca2+-CaM-ligand interactions: i) CaM has two preferred sites for ligand binding in the open conformation site 1 and 4, ii) CaM can bind different types of ligands having a well defined hydrophobic and a polar region, iii) the flexibility of CaM can be reduced by the binding of its ligands, reflected as a reduction of Ca2+-CaM entropy, and iv) enthalpy contribution dominates the molecular recognition process in this system (Ca2+-CaM-ligand). Further studies are needed in order to address additional questions left unexplored in the present work, such as Ca2+-CaM-target protein-ligand interactions, complexes of higher stoichiometry, molecular dynamics with ligands combinations, and some other. In conclusion, docking studies and MD simulations in combination with experimental data are excellent tools for identifying binding sites and improve our understanding of the processes involved in the molecular recognition

between ligands and their molecular targets. Therefore, the use of these computational tools represent nowadays an exciting area in pharmacological research, providing valuable information for the rational design of new anti-CaM drugs in particular, but likely, the strategy


may be of more general use.

4. Methods and computational details 4.1. Preparation of initial coordinate files The coordinates corresponding to the structure of CaM were obtained from the Protein Data Bank (PDB, http://www.rcsb.org). For the open conformation of CaM, the X-ray structure with calcium refined at a resolution of 1.7 Å (PDB entry 1CLL) was selected and for the closed conformation of CaM refined at 2.7 Å in complex with TFP (PDB entry 1A29) was chosen (Chattopadhyaya, Meador, Means, & Quiocho, 1992; Vertessy et al., 1998). The structure of ligands was obtained from the PDB co-crystallized structure when available, or constructed using HyperChem 8 software. All structures of the ligands were minimized using Gaussian 09, revision A.02 (Gaussian Inc., Wallingford, CT) at DTF B3LYP/3-21G level of theory. Partial charges and force field parameters of the inhibitors were generated using the antechamber program in AMBER 12 (Case et al., 2005; S. F. Chen, Cao, Han, & Chen, 2013; Wang, Wang, Kollman, & Case, 2006). 4.2. Docking Protocol

Docking was conducted using the PDB X-ray structure of CaM (1CLL and 1A29) and PDBs files of the 15 ligands. The structures of CaM and ligands were further prepared using the utilities implemented by AutoDockTools 1.5.4 (http://mgltools.scripps.edu/). All hydrogen atoms as well the Kollman united-atom partial charges were added to the protein structures, while Gasteiger-Marsili charges and rotatable groups were assigned automatically to the structures of the ligands, at the active torsions were added to the structures of the ligands. Blind docking was


carried out with AutoDock4 version 4.2 (http://autodock.scripps.edu/) (Huey, Morris, Olson, & Goodsell, 2007; Morris et al., 1998) as indicated: default parameters for the Lamarkian genetic algorithm with local search, number of individuals in population (150), maximum number of energy evaluations (2.5 million), maximum number of generations (27000), rate of gene mutation (0.02), rate of crossover (0.8), and 1000 runs for docking. Electrostatic grid maps were generated for each atom type using the auxiliary program AutoGrid4 that is part of the software AutoDock4. The initial grid box size was 60Å  60Å  60Å in the x, y, and z dimensions. Afterwards, a refined docking analysis was performed with a smaller grid box of 30Å  30Å  30Å, centered at the previously identified ligand´s binding site. Docking was analyzed with AutoDockTools using cluster analysis, LigPlot (Wallace, Laskowski, & Thornton, 1995), and PyMOL (DeLano, 2004) software.

4.3. Molecular dynamics simulation The coordinates of the ligands, resulting from the docking study, were processed with antechamber (a set of auxiliary programs for molecular mechanic studies) in order to generate suitable topologies for the LEaP module from AMBER 12 (Case et al., 2005; D.A. Case, 2012).

Each structure and complex were subjected to the following protocol: hydrogen’s and other missing atoms were added using the LEaP module with the parm99 parameter set, Na+ counterions were added to neutralize the system, the complexes were then solvated in an octahedral box of explicit TIP3P model water molecules localizing the box limits at a distance of 8 Å from the protein surface. The total number of atoms in each simulated system ranged from 14500 to 17700, including solvent molecules. MD simulations were performed at 1 atm and 298


K, maintained with the Berendsen barostat and thermostat, using periodic boundary conditions and particle mesh Ewald sums (grid spacing of 1 Å) for treating long-range electrostatic interactions with a 10 Å cutoff for computing direct interactions. The SHAKE algorithm was used to satisfy bond constraints, allowing employment of a 2 fs time step for the integration of Newton’s equations as recommended in the Amber package (D.A. Case, 2012; Walker, Crowley, & Case, 2008). Amber f99SB force field (Case et al., 2005; Lindorff-Larsen et al., 2010; Thomas et al., 2013) parameters were used for all residues and Gaff force field (Wang et al., 2006; Wang, Wolf, Caldwell, Kollman, & Case, 2004) parameters were used for the ligands. The protocol consisted in performing an optimization of the initial structure, followed by 50 ps heating step at 298 K, 50 ps for equilibration at constant volume and 500 ps for equilibration at constant pressure. Several independents 20 ns MD simulations were performed. Frames were saved at 100 ps intervals for subsequent analysis 4.4. Binding free energies calculated by molecular mechanics/poisson boltzmann surface area (MM/PBSA) This method involves a combination of molecular mechanics energy with implicit solvation models to calculate binding free energies. In MM/PBSA (Treesuwan & Hannongbua,

2009; Zhou & Madura, 2004), binding free energy (ΔGbind) between a ligand (L) and a target (T)


to form a complex is calculated as:

where ΔEMM, ΔGSol and −TΔS are the changes of the gas phase molecular mechanics energy, the solvation free energy and the conformational entropy upon binding, respectively. ΔEMM comprises ΔEInternal (bond, angle and dihedral energies), ΔEElectrostatic (electrostatic energies), and ΔEVdw (van der Waals energies). ΔGSolv is the sum of electrostatic solvation energy (polar contribution) -ΔGPB- and non-electrostatic solvation component (non-polar contribution) -ΔGSA-. The polar contribution is calculated using the Poisson-Boltzmann surface area model, while the non-polar energy is estimated from the solvent accessible surface area (SASA). The conformational entropy change (-TS) was computed by normal mode analysis from a set of conformational snapshots taken from the MD simulations (Hou et al., 2011; Kollman et al., 2000; Treesuwan & Hannongbua, 2009).

Acknowledgment This work was supported by grants from DGAPA-UNAM (IN218110) and CONACyT 99395. The authors are very grateful to Dr. A. Olson and his colleagues at the Scripps Research Institute for providing AutoDock. We are indebted to Dirección General de Cómputo y de Tecnologías de Información y Comunicación, UNAM, for providing the resources to carry out

computational calculations through Nuevo Equipo de Supercomputo (NES) System and the project number SC14-1-I-36 Miztli supercomputer.

Supplementary material Supplementary data associated with this article plots, figures, and movies can be found, in the


online version.

References

Baron, R., & McCammon, J. A. (2013). Molecular recognition and ligand association. Annual review of physical chemistry, 64, 151-175. Bouche, N., Yellin, A., Snedden, W. A., & Fromm, H. (2005). Plant-specific calmodulin-binding proteins. Annual Review of Plant Biology, 56, 435-466. Case, D. A., Cheatham, T. E., 3rd, Darden, T., Gohlke, H., Luo, R., Merz, K. M., Jr., . . . Woods, R. J. (2005). The Amber biomolecular simulation programs. Journal of Computational Chemistry, 26, 1668-1688. Chattopadhyaya, R., Meador, W. E., Means, A. R., & Quiocho, F. A. (1992). Calmodulin structure refined at 1.7 A resolution. Journal of molecular biology, 228, 1177-1192. Chen, R., & Chung, S. H. (2013). Molecular dynamics simulations of scorpion toxin recognition by the Ca(2+)-activated potassium channel KCa3.1. Biophysical journal, 105, 18291837.

Chen, S. F., Cao, Y., Han, S., & Chen, J. Z. (2013). Insight into the structural mechanism for PKBalpha allosteric inhibition by molecular dynamics simulations and free energy calculations. Journal of molecular graphics & modelling, 48C, 36-46. Chin, D., & Means, A. R. (2000). Calmodulin: a prototypical calcium sensor. Trends in Cell Biology, 10, 322-328. DeLano, W. L. (2004). Use of PYMOL as a communications tool for molecular science.


Abstracts of Papers of the American Chemical Society, 228, U313-U314. Dolenc, J., Baron, R., Missimer, J. H., Steinmetz, M. O., & van Gunsteren, W. F. (2008). Exploring the conserved water site and hydration of a coiled-coil trimerisation motif: a MD simulation study. Chembiochem : a European journal of chemical biology, 9, 17491756. Du, J., Szabo, S. T., Gray, N. A., & Manji, H. K. (2004). Focus on CaMKII: a molecular switch in the pathophysiology and treatment of mood and anxiety disorders. International Journal of Neuropsychopharmacology, 7, 243-248. Figueroa, M., Gonzalez-Andrade, M., Sosa-Peinado, A., Madariaga-Mazon, A., Del Rio-Portilla, F., Gonzalez Mdel, C., & Mata, R. (2011). Fluorescence, circular dichroism, NMR, and docking studies of the interaction of the alkaloid malbrancheamide with calmodulin. Journal of enzyme inhibition and medicinal chemistry, 26, 378-385. Gonzalez-Andrade, M., Del Valle, P., Macias-Rubalcava, M. L., Sosa-Peinado, A., Del Carmen Gonzalez, M., & Mata, R. (2013). Calmodulin inhibitors from Aspergillus stromatoides. Chemistry & biodiversity, 10, 328-337.

Gonzalez-Andrade, M., Figueroa, M., Rodriguez-Sotres, R., Mata, R., & Sosa-Peinado, A. (2009). An alternative assay to discover potential calmodulin inhibitors using a human fluorophore-labeled CaM protein. Analytical Biochemistry, 387, 64-70. Gonzalez-Andrade, M., Rivera-Chavez, J., Sosa-Peinado, A., Figueroa, M., Rodriguez-Sotres, R., & Mata, R. (2011). Development of the fluorescent biosensor hCalmodulin (hCaM)L39C-monobromobimane(mBBr)/V91C-mBBr, a novel tool for discovering new


calmodulin inhibitors and detecting calcium. Journal of medicinal chemistry, 54, 38753884. Harmat, V., Bocskei, Z., Naray-Szabo, G., Bata, I., Csutor, A. S., Hermecz, I., . . . Ovadi, J. (2000).

A new

potent

calmodulin

antagonist

with

arylalkylamine

structure:

crystallographic, spectroscopic and functional studies. Journal of Molecular Biology, 297, 747-755. Hou, T., Wang, J., Li, Y., & Wang, W. (2011). Assessing the performance of the MM/PBSA and MM/GBSA methods. 1. The accuracy of binding free energy calculations based on molecular dynamics simulations. Journal of chemical information and modeling, 51, 6982. Huey, R., Morris, G. M., Olson, A. J., & Goodsell, D. S. (2007). A semiempirical free energy force field with charge-based desolvation. Journal of Computational Chemistry, 28, 1145-1152. Junker, J. P., & Rief, M. (2009). Single-molecule force spectroscopy distinguishes target binding modes of calmodulin. Proceedings of the National Academy of Sciences U S A, 106, 14361-14366.

Kollman, P. A., Massova, I., Reyes, C., Kuhn, B., Huo, S., Chong, L., . . . Cheatham, T. E., 3rd. (2000). Calculating structures and free energies of complex molecules: combining molecular mechanics and continuum models. Accounts of chemical research, 33, 889897. Laine, E., Yoneda, J. D., Blondel, A., & Malliavin, T. E. (2008). The conformational plasticity of calmodulin upon calcium complexation gives a model of its interaction with the oedema


factor of Bacillus anthracis. Proteins, 71, 1813-1829. Lindorff-Larsen, K., Piana, S., Palmo, K., Maragakis, P., Klepeis, J. L., Dror, R. O., & Shaw, D. E. (2010). Improved side-chain torsion potentials for the Amber ff99SB protein force field. Proteins, 78, 1950-1958. Miao, Y., Yi, Z., Cantrell, C., Glass, D. C., Baudry, J., Jain, N., & Smith, J. C. (2012). Coupled flexibility change in cytochrome P450cam substrate binding determined by neutron scattering, NMR, and molecular dynamics simulation. Biophysical journal, 103, 21672176. Morris, G. M., Goodsell, D. S., Halliday, R. S., Huey, R., Hart, W. E., Belew, R. K., & Olson, A. J. (1998). Automated docking using a Lamarckian genetic algorithm and an empirical binding free energy function. Journal of Computational Chemistry, 19, 1639-1662. Norberto de Souza, O., & Ornstein, R. L. (1999). Molecular dynamics simulations of a proteinprotein dimer: particle-mesh Ewald electrostatic model yields far superior results to standard cutoff model. Journal of biomolecular structure & dynamics, 16, 1205-1218. Osawa, M., Swindells, M. B., Tanikawa, J., Tanaka, T., Mase, T., Furuya, T., & Ikura, M. (1998). Solution structure of calmodulin-W-7 complex: the basis of diversity in molecular recognition. Journal of molecular biology, 276, 165-176.

Seales, E. C., Micoli, K. J., & McDonald, J. M. (2006). Calmodulin is a critical regulator of osteoclastic differentiation, function, and survival. Journal of Cellular Biochemistry, 97, 45-55. Shifman, J. M., & Mayo, S. L. (2002). Modulating calmodulin binding specificity through computational protein design. Journal of Molecular Biology, 323, 417-423. Sorensen, B. R., & Shea, M. A. (1998). Interactions between domains of apo calmodulin alter


calcium binding and stability. Biochemistry, 37, 4244-4253. Thomas, A. S., Mao, S., & Elcock, A. H. (2013). Flexibility of the bacterial chaperone trigger factor in microsecond-timescale molecular dynamics simulations. Biophysical journal, 105, 732-744. Treesuwan, W., & Hannongbua, S. (2009). Bridge water mediates nevirapine binding to wild type and Y181C HIV-1 reverse transcriptase--evidence from molecular dynamics simulations and MM-PBSA calculations. Journal of molecular graphics & modelling, 27, 921-929. Vandonselaar, M., Hickie, R. A., Quail, J. W., & Delbaere, L. T. (1994). Trifluoperazine-induced conformational change in Ca(2+)-calmodulin. Nature structural biology, 1, 795-801. Vertessy, B. G., Harmat, V., Bocskei, Z., Naray-Szabo, G., Orosz, F., & Ovadi, J. (1998). Simultaneous binding of drugs with different chemical structures to Ca2+-calmodulin: crystallographic and spectroscopic studies. Biochemistry, 37, 15300-15310. Walker, R. C., Crowley, M. F., & Case, D. A. (2008). The implementation of a fast and accurate QM/MM potential method in Amber. Journal of Computational Chemistry, 29, 10191031.

Wallace, A. C., Laskowski, R. A., & Thornton, J. M. (1995). LIGPLOT: a program to generate schematic diagrams of protein-ligand interactions. Protein engineering, 8, 127-134. Wang, J., Wang, W., Kollman, P. A., & Case, D. A. (2006). Automatic atom type and bond type perception in molecular mechanical calculations. Journal of molecular graphics & modelling, 25, 247-260. Wang, J., Wolf, R. M., Caldwell, J. W., Kollman, P. A., & Case, D. A. (2004). Development and


testing of a general amber force field. Journal of Computational Chemistry, 25, 11571174. Xu, M., Yu, L., Wan, B., & Huang, Q. (2011). Predicting inactive conformations of protein kinases using active structures: conformational selection of type-II inhibitors. PLoS One, 6, e22644. Yang, C., Jas, G. S., & Kuczera, K. (2001). Structure and dynamics of calcium-activated calmodulin in solution. Journal of biomolecular structure & dynamics, 19, 247-271. Zhou, Z., & Madura, J. D. (2004). Relative free energy of binding and binding mode calculations of HIV-1 RT inhibitors based on dock-MM-PB/GS. Proteins, 57, 493-503.


Table 1. Experimental and theoretical binding properties of Ca2+-CaM-Ligands complexes. Figure 1. Three-dimensional structures of CaM in its different conformations; in green cartoon (N-Terminus; residues 1-80), blue cartoon (C-Terminus; residues 81-148), and white spheres the Ca2+ ions. (A) calcium free (Protein Data Bank [PDB] code: 1CFD); (B) with calcium (PDB code: 1CLL); (C) with TFP and II–spec (PDB code: 1A29 and 2FOT, respectively). The structures were drawn using the PyMOL program. Figure 2. Chemical structures of CaM inhibitors AAA (N-(3, 3-diphenylpropyl)-N'-[1-R-(3, 4bis-butoxyphenyl)ethyl]-propylene-diamine), CPZ (chlorpromazine), TFP (trifluoroperazine), W7 (N-(6-aminohexyl)-5-chloro-1-naphthalenesulfonamide), MBC (malbrancheamide), MBCB (malbrancheamide B), MBC-I (isomalbrancheamide B), MBC-P (premalbrancheamide), XAN-14 (14-methoxytajixanthone), XAN-15 (15-acetyl tajixanthone hydrate), XAN-A

(variecoxanthone A acetate), EM (emericellin), TJX-H (tajixanthone hydrate), TJX (tajixanthone), and CIT (citrinin). Figure 3. Structural model of Ca2+-CaM-Ligands, in the open and closed conformation. Shown the complexes in cartoon the (Ca2+-CaM), red sticks (AAA), blue sticks (TFP), yellow sticks (W7), magenta sticks (MBC), and cyan sticks (XAN-14). Also shown the pocket of binding


corresponding to the site 1 and 4, are shown in surface blue and green, respectively. The amino acid corresponding at site 1 are: Phe92, Ile100, Leu105, Met124, Ile125, Glu127, Ala128, Val136, Phe141, and Met144 and site 4 are: Phe19, Ile27, Leu32, Met51, Ile52, Gku54, Val55, Ile63, Phe68, and Met71. Figure 4. The RMSD & Time plot for 20 ns MD simulation. (A) shows the differences between closed complex; 1A29 (), 1A29-CPZ (), 1A29-TXJ (), 1A29- αII–spec (), and 1A29XAN15 (). Down are shows the differences between open () and closed () conformations of different complexes; 1CLL (B), 1CLL-αII–spec (C), 1CLL-W7 (D), and 1CLL-MBC-P (E).

Figure 5. Box and whiskers plot of the RMSD along the MD trajectory for the CaM backbone atoms and using the starting structure as reference. (A) The open form of CaM (1a29) in complex with the ligand indicated in the x-axis was simulated for 20 ns. (B) The closed form of CaM (1cll) in complex with the ligand indicated in the x-axis was simulated for 20 ns. Figure 6. Fluctuations per residue considering 20 ns of MD. Side-chains fluctuations of Ca2+CaM (), Ca2+-CaM-II-spec (), and Ca2+-CaM-TFP () are shown for the closed (A) and open (B) conformation with their respective secondary structure at the bottom of each graph.

Figure 7. Correlation between the binding free energies calculated by MM/PBSA and the experimental values for the Ca2+-CaM complexes. In the panel A shown the different inhibitors (AAA, II-spec, TFP, W7, XAN-15, MBC) and the panel B shown the compounds corresponding at xhantones. Figure 8. Ca2+-CaM-ligands interactions along 20 ns of molecular dynamic. (A)1CLL-MBC,


(B) 1CLL-MBC-P, (C) 1CLL-CPZ, (D) 1CLL-II-spec, (E) 1A29-CPZ and (F) 1A29-IIspec.


Table 1. Experimental and theoretical binding properties of Ca2+-CaM-Ligands complexes. MD studies Complex Evdw Eele GPB GSA H S Gcal (kcal mol-1) (kcal mol-1) (kcal mol-1) (kcal mol-1) (kcal mol-1) (kcal mol-1) (kcal mol-1) 1A29-AAA -47.80±3.2 1A29-CPZ -26.58±2.0 1A29-αII–spec -121.87±6.0 1A29-TFP -33.76±3.1 1A29-W7 -29.39±3.7 1A29-MBC -34.53±5.4 1A29-MBC-B -36.71±3.3 1A29-MBC-I -38.62±2.5 1A29-MBC-P -36.02±2.5 1A29-XAN-14 -42.97±2.8 1A29-XAN-15 -41.93±2.6 1A29-XAN-A -39.43±2.5 1A29-EM -42.47±5.6 1A29-TJX-H -44.42±2.6 1A29-TJX -37.37±2.3 1A29-CIT -24.32±4.0 1 Apparent dissociation constants

-742.84±26.6 -369.61±17.8 -1214.59±75.7 -876.50±15.6 -388.64±20.8 -369.25±30.4 -398.15±24.1 -425.73±11.2 -380.80±11.7 7.78±4.2 1.89±2.1 -12.77±2.8 -4.30±4.1 -0.26±0.1 -2.60±2.3 -13.07±8.8

753.15±29.0 376.70±17.5 1282.57±70.0 872.16±13.5 395.58±19.6 374.36±34.4 401.42±21.6 430.94±9.7 393.69±12.3 14.04±5.1 23.05±11.3 31.90±4.1 23.33±3.6 24.79±8.3 17.46±4.4 27.77±10.3

-5.24±0.2 -2.78±0.3 -15.27±0.4 -3.79±0.1 -3.02±0.3 -2.60±0.6 -3.30±0.1 -3.32±0.1 -3.07±0.1 -4.44±0.2 -4.55±0.2 -3.99±0.1 -4.45±0.2 -4.33±0.1 -4.02±0.2 -2.23±0.1

-42.75±6.2 -22.27±2.5 -69.16±12.8 -41.90±3.6 -25.47±3.5 -32.02±3.0 -36.74±4.3 -36.74±3.1 -26.21±2.8 -25.58±3.9 -21.54±3.3 -24.28±3.0 -27.59±3.7 -24.22±3.2 -26.53±3.2 -11.85±3.5

-29.35±4.8 -17.75±7.7 -48.46±9.64 -16.57±2.7 -24.09±5.2 -24.73±3.81 -17.72±6.2 -7.62±3.2 -12.45±3.95 -17.23±5.7 -14.56±4.2 -16.72±3.7 -18.9±5.7 -15.99±3.8 -18.55±4.6 -13.60±5.2

-13.4±6.0 -4.52±1.8 -20.69±6.1 -25.32±3.2 -1.38±0.3 -7.29±2.8 -19.02±6.3 -29.11±8.5 -3.76±4.8 -8.35±3.1 -6.98±2.4 -7.56±2.6 -8.69±3.2 -8.23±2.7 -7.98±2.2 1.74±0.8

Gexp (kcal mol-1) -11.93 -7.97 -15.82 -8.55 -9.49 -8.12 -7.25 -7.25 -10.42 -8.59 -9.42 -11.14 -9.59 -10.13 -

Experimental Kd1 References (M) 18.0x10-9 1.43x10-6 2.50x10-12 0.53x10-6 11.0x10-6 1.10x10-6 4.80x10-6 4.80x10-6 23.1x10-9 498.4x10-9 124.2x10-9 6.80x10-9 93.0x10-9 3.7x10-9 -

Harmat et al., 2000 Figueroa et al., 2011 Shifman & Mayo, 2002 M. Gonzalez-Andrade et al., 2011 Osawa et al., 1998 Figueroa et al., 2011 Figueroa et al., 2011 Figueroa et al., 2011 Figueroa et al., 2011 M. Gonzalez-Andrade et al., 2011 M. Gonzalez-Andrade et al., 2011 M. Gonzalez-Andrade et al., 2011 M. Gonzalez-Andrade et al., 2011 M. Gonzalez-Andrade et al., 2011 M. Gonzalez-Andrade et al., 2011 M. Gonzalez-Andrade et al., 2013


Figure 1


Figure 2


Figure 3

Figure 4 5.0 4.5

A

1A29 1A29-CPZ 1A29-TJX 1A29-II-spec 1A29-XAN15

RMSD (Å)

4.0 3.5 3.0 2.5

1.5 1.0 0

5000

10000

15000

20000

Time (ps) 8

B

14

1A29 1CLL

6

12

RMSD (Å)

RMSD (Å)

7 5 4 3 2

0

5000

10000

15000

10 8 6 4

20000

0

D

10

1A29-W7 1CLL-W7

8 6 4 2 0

5000

10000 Time (ps)

15000

20000

RMSD (Å)

12

1A29-II-spec 1CLL-II-spec

C

2

1

RMSD (Å)


2.0

9 8 7 6 5 4 3 2 1

5000

10000

15000

1A29-MBC-P 1CLL-MBC-P

E

0

20000

5000

10000 Time (ps)

15000

20000

W7

CIT

TJX

TJXH

EM

XANA

XAN15

XAN14

MBCP

MBCI

MBCB

0 MBC

1

TFP

3

CPZ

14

aIIsprec

A

AAA

4

RMSD (Å)

6

none

CIT

TJX

TJXH

EM

XANA

XAN15

XAN14

MBCP

MBCI

MBCB

MBC

W7

TFP

aIIsprec

CPZ

AAA

RMSD (Å) 5

none


Figure 5

16

B

12

10 8

2

6

4

2

0

Figure 6

0.7

0.9

1A29 1A29-II-spec 1A29-TFP

RMSF (Å)

0.5 0.4 0.3 0.2

0.7 0.6 0.5 0.4 0.3 0.2

0.1


1CLL 1CLL-aII-spec 1CLL-TFP

B

0.8

RMSF (Å)

A

0.6

0.1

0.0

0.0

-0.1

-0.1 0

20

40

60

80

Residue Number

100

120

140

0

20

40

60

80

Residue Number

100

120

140

Figure 7

-6.8

0

W7


XAN-15

MBC

-10

AAA -15

-20

R2=0.76

II-spec -16

-14

-12

-10

-8 -1

Experimental Gbind (kcal mol )

-1

CPZ

-5

Predicted Gbind (kcal mol )

-7.0

-1

Predicted Gbind (kcal mol )

A

B

XAN-15

-7.2 -7.4

XAN-A

-7.6 -7.8

TJX

-8.0 -8.2

TJX-H

XAN-14 -8.4 -8.6

R2=0.82

EM -8.8 -11.5

-11.0

-10.5

-10.0

-9.5

-9.0 -1

Experimental Gbind (kcal mol )

-8.5

Figure 8 140

A

B

B

D

E

C

Residue Number

120

100

80

60

20

0

140

Residue Number

120

100

80

60

40

20

0 140

120

Residues Number


40

100

80

60

40

20

0 0

5000

10000

Time (ps)

15000

20000

0

5000

10000

Time (ps)

15000

20000

Graphical abstract

-1 Predicted Gbind (kcal mol )

8

RMSD (Å)


0

7 6 5

W7 CPZ

-5

XAN-15 -10

AAA -15

-20

4

MBC

R2=0.76

II-spec

3 2

-16 0

2

4

6

8

10

Time (ns)

12

14

16

18

20

-14

-12

-10

-8

-1 Experimental Gbind (kcal mol )

Interactions of Multimodal Ligands with Proteins: Insights into Selectivity Using Molecular Dynamics Simulations.

P-113 peptide: New experimental evidences on its biological activity and conformational insights from molecular dynamics simulations.

CH4 Hydrate Formation between Silica and Graphite Surfaces: Insights from Microsecond Molecular Dynamics Simulations.

Dual-inhibitors of STAT5 and STAT3: studies from molecular docking and molecular dynamics simulations.

Insights into ligand binding to PreQ1 Riboswitch Aptamer from molecular dynamics simulations.

Molecular dynamics simulations: insight into molecular phenomena at interfaces.

Insights into the role of Asp79(2.50) in β2 adrenergic receptor activation from molecular dynamics simulations.

Insights into the molecular interactions between aminopeptidase and amyloid beta peptide using molecular modeling techniques.

Molecular docking and molecular dynamics simulation studies of Trypanosoma cruzi triosephosphate isomerase inhibitors. Insights into the inhibition mechanism and selectivity.

Factors affecting the interactions between beta-lactoglobulin and fatty acids as revealed in molecular dynamics simulations.

Insights into the unfolding pathway and identification of thermally sensitive regions of phytase from Aspergillus niger by molecular dynamics simulations.

Insights into the key interactions between human protein phosphatase 5 and cantharidin using molecular dynamics and site-directed mutagenesis bioassays.

Nonadiabatic molecular dynamics simulations: synergies between theory and experiments.

Insights into the Tunnel Mechanism of Cholesteryl Ester Transfer Protein through All-atom Molecular Dynamics Simulations.

Understanding Miltefosine-Membrane Interactions Using Molecular Dynamics Simulations.

Modeling of annexin A2-Membrane interactions by molecular dynamics simulations.

Detailed molecular dynamics simulations of human transferrin provide insights into iron release dynamics at serum and endosomal pH.

Understanding the thermostability and activity of Bacillus subtilis lipase mutants: insights from molecular dynamics simulations.

Molecular dynamics simulations of anchored viral peptide interactions.

Characterizing a partially ordered miniprotein through folding molecular dynamics simulations: Comparison with the experimental data.

Molecular dynamics simulations of membrane proteins and their interactions: from nanoscale to mesoscale.

Insights into the interactions between maleimide derivates and GSK3β combining molecular docking and QSAR.

Insights into carbon nanotube and graphene formation mechanisms from molecular simulations: a review.

Interactions of the EGFR juxtamembrane domain with PIP2-containing lipid bilayers: Insights from multiscale molecular dynamics simulations.