Multistep Modeling Strategy To Improve the Binding Affinity Prediction

Subscriber access provided by Northern Illinois University

Article

A Multistep Modeling Strategy to Improve the Binding Affinity Prediction of PET Tracers to Amyloid A#42: A Case Study with Styrylbenzoxazole Derivatives. Kanagasabai Balamurugan, Natarajan Arul Murugan, and Hans Ågren ACS Chem. Neurosci., Just Accepted Manuscript • DOI: 10.1021/acschemneuro.6b00216 • Publication Date (Web): 14 Sep 2016 Downloaded from http://pubs.acs.org on September 15, 2016

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

ACS Chemical Neuroscience is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Chemical Neuroscience

A Multistep Modeling Strategy to Improve the Binding Affinity Prediction of PET Tracers to Amyloid Aβ42: A Case Study with Styrylbenzoxazole Derivatives. Kanagasabai Balamurugan,†* Natarajan Arul Murugan,† and Hans Ågren† †

Division of Theoretical Chemistry and Biology, School of Biotechnology, KTH Royal Institute of

Technology, SE-10691 Stockholm, Sweden

ABSTRACT: Positron emission tomography (PET) tracers play an important role in the diagnosis of Alzheimer’s disease, a condition that leads to progressive dementia and memory loss. A high binding affinity and specificity of the PET tracers to amyloid oligomers and fibrils are crucial for their successful application as diagnostic agents. In this sense it is essential to design PET tracers with enhanced binding affinities which can lead to more precise and earlier detection of Alzheimer’s disease conditions. The application of in silico methodology for the design and development of efficient PET tracers

may serve as an important route to improved Alzheimer diagnosis. In this work

the

performance of widely used computational methods is explored for predicting experimental binding affinities of styrylbenzoxazole (SB) derivatives against a common amyloid protofibril. By performing docking, molecular dynamics and quantum chemistry calculations in sequence their combined predictive performance is explored. The present work emphasizes the merits as well as limitations of these simulation strategies in the realm of designing PET tracers for Alzheimer’s diagnosis. Keywords: Alzheimer’s disease, amyloid-β peptide, PET tracers, molecular dynamics, Density functional theory.

*[email protected], [email protected]

1

ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 24

INTRODUCTION Alzheimer’s disease is a chronic neurodegenerative disease which is the most widely occurring cause for dementia in aged populations.1 It accounts for a considerable burden in terms of health care expenditure in the developed nations. It was first described in 1906 by German psychiatrist and pathologist Alois Alzheimer, but still after a century the causative reason for the disease remains inconclusive.2-3 As for any disease the very first step of treatment starts with the diagnosis of the disease condition. The diagnosis of dementia is complicated because of its requirement to differentiate between the normal memory loss against the chronic neurodegenerative condition in the aged patients. Even though contradictions exist regarding the causative reasons of Alzheimer’s condition the amyloid hypothesis is most widely accepted which suggests that amyloid deposits are the hallmarks of the disease.4 Techniques such as magnetic resonance imaging (MRI) and computed tomography (CT) scans were earlier used for the diagnosis of dementia but the problem of these techniques is that they only can reveal structures of the brain which may not be sufficient to confirm the condition of Alzheimer’s.5 With the advent of positron emission tomography (PET) scans, a new avenue for monitoring the brain function and metabolism was opened which is vital for the diagnosis of Alzheimer’s condition.5 A PET scan uses a radioactive substance called a tracer to look for the distribution of the disease related plagues or injury in the brain. The usage of amyloid specific PET tracers and the imaging of the accumulation of tracers in the brain lead to the identification of the amyloid plagues. Thus PET imaging is widely used in the cases where CT and MRI results are inadequate to conclude the condition of Alzheimer’s. A number of PET tracers specific to amyloid fibrils were proposed for diagnosing the Alzheimer’s condition. Particularly popular classes of tracers are

the thioflavin T derived

11

C-

Pittsburgh Compound B (PIB), the benzofurane based 11C-AZD-2184 , the benzoxazole type 11C -BF227, the stilbene compound11C -SB, furthermore the 18

18

F-labeled tracers

18

18

4694, F-florbetaben, F-florbetapir, and the naphthol based F-FDDNP.

18

F-flutemetamol,

6-11 11

18

F-AZD-

18

C and F radioisotopes

are the mainly used tracers for diagnosis and of the two 18F is preferred due to its longer half life time of around two hours, while the lifetime is only around 20 minutes in the case of tracers listed above three are approved by FDA, namley

18

F-florbetapir,

florbetaben (in 2012, 2013 and 2014, respectively). 2


18

11

C. Among the

F-flutemetamol and

18

F-

Page 3 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Molecular modeling and simulations have been widely employed

for an atomistic level

understanding of the mode of interaction of small molecules with biomolecules such as proteins and nucleic acids. Apart from providing understanding of the interaction process, the in silico methodologies are also used in the realm of drug design for the development of molecules with better binding affinity which in turn has a better potency .14 In silico methodologies are used as a preliminary step in the drug discovery process to screen the library of compounds. The promising compounds identified are then carried along for the development of potent drugs.15 A priori, one needs to know about the number of binding sites in a specific biological target and the nature of the microenvironment of these sites in order to optimize the drugs. In the case of amyloids a number of studies have been reported also on the mode of interaction and binding affinity of different optical and PET tracers with amyloid fibrils.16-19 One of the most challenging steps concerning the amyloid targets is polymorphic nature of the amyloid fibril and the

existence of

the

different oligomeric forms and

quarternary structures.20 Molecular modeling methods vary in terms of their ability to account for different effects and the way they describe the subsystem interactions. Docking and other binding free energy calculation approaches, such as molecular mechanics and the Generalized Born (or Poisson Boltzmann) Surface Area approach (MM-GB(PB)SA), are the most widely used methods in the drug discovery parlance. Docking is used for screening libraries of drug candidates and their mode of binding against the target molecules. However, MM-GB(PB)SA approaches rely on other sampling methods like Molecular Dynamics (MD) or Monte Carlo (MC) for the configurations of receptor-ligand complexes which are then used for computing average binding free energies of the complex. Docking is computationally cheap whereas the MD+MM-GB(PB)SA based binding free energy calculations are relatively demanding. Both docking and MM-GB(PB)SA are classical methods based on force fields which account for only the nuclear motions of the molecule without explicit treatment of the electronic degrees of freedom.

The accuracy is limited by the suitability of the governing force-fields in

describing the subsystem interactions. Pitfalls of the above mentioned methods in predicting the free energy of binding of small molecules with biological targets have been discussed in many previous studies.21-23 By employing Quantum Mechanical (QM) methods the subsystem interaction can be described in a better way, as polarization of a part of the system by the rest of the system is accounted for, making the so obtained electrostatic interaction energies much more accurate. This, of course, comes at the cost of high computational resources. 3



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 24

The primary requirement for assessing the accuracy of the modeling strategies is to match the experimental values for a series of compounds from a single experimental study. In this study an attempt is been made to analyze the inherent performance of a combination of theoretical methods to predict experimental binding affinities of styrylbenzoxazole derivatives against amyloid peptides.24 The purpose is to come up with a workflow of modeling methods to improve the binding affinity prediction for the design of new PET tracers. RESULTS AND DISCUSSION Docking The details of the styrylbenzoxazole derivatives, dissociation constant and their binding affinities in terms of kcal/mol are given in Table 1. The ten styrylbenzoxazole derivatives were docked into various binding sites of Aβ42 - the binding free energy results are given in the Table 2. The four docking sites of Aβ42 with the smallest binding affinities are illustrated in Figure 1. The results indicate that site S4 has the smallest binding affinity for the given series of the styrylbenzoxazole derivatives followed by the S3 and S2 sites of Aβ42. The S1 site of Aβ42 possesses the highest binding affinity among the most possible sites for the styrylbenzoxazole derivatives. It is also found from previous studies that this particular site is the most favorable one for various other PET tracers.16,17

So, the complex of

styrylbenzoxazole derivatives at the S1 site is considered for the molecular dynamics simulation and further studies. It is found from the docking results (as reported in Table 1) that the docking approach does not distinguish between SB derivatives with nanomolar and micromolar binding affinities. Moreover, the most effective SB derivative BF-164 was not predicted to have the highest binding affinity whereas BF-124 possesses the highest binding affinity among the SB derivatives.

Molecular dynamics simulations The binding free energies of the SB derivatives with Aβ42 were estimated using the MMGBSA and MMPBSA methodologies - they are presented in the Table 3. The results indicate that the MMGBSA and MMPBSA values are in line with each other whereas there is a considerable increase in the magnitude of the MMPBSA values comparing with MMGBSA. It is also notable that MMPBSA and 4


Page 5 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


MMGBSA do not distinguish between the SB derivatives having nanomolar and micromolar binding affinities. Here also we find that BF-124 (MMGBSA) and BF-168 (MMPBSA) have the highest binding affinity whereas the experimental values indicate that BF-164 possesses the highest binding affinity. The various contributions to the overall binding energies using MMGBSA and MMPBSA are provided in

Table 4 and 5, respectively. It is found from these results that the van der Waals

interactions provide a significant contribution to the overall interaction energy. It is also important to note that the MMGBSA and MMPBSA results may depend on the total time scale of the simulation. In order to verify whether the computed

binding energy values of MMGBSA and MMPBSA are

converged, the binding energies are calculated for the last 2ns of the simulation i.e. 8-10 ns for three SB derivatives (with the highest, lowest and intermediate binding affinities, namely BF-164, BF-208 and BF-168, respectively) and are compared with those computed for the trajectory corresponding to 4-10 ns. The results are presented in Table 6. It is observed that the MMGBSA values deviate up to 1kcal/mol, while

the MMPBSA values deviate up to1.4 kcal/mol. The order is largely maintained,

however, indicating that

in this particular case the convergence of the binding free energy

is

established. Quantum Mechanical Calculations The interaction energy calculated by using the quantum mechanical approach for cluster models of the Aβ42-SB tracer complexes with solvent effects included using the PCM description, is presented in Table 7. The schematic representation of the cluster model derived from the MD trajectories is shown in Figure 2. It is observed from the predicted binding affinity values that the trend in the interaction energy values is independent of whether the calculations are carried out in gas phase or in solvent environment. Although the trend of results remains the same there is a considerable increase in the magnitude of interaction energy in the case of the PCM based model compared to the gas phase calculations. It is notable that the QM based interaction energies are much higher in magnitude compared with

the MD simulation values (300K). The possible reason for this could be the

temperature factor where the QM calculations refer to 0K and the neglect of entropy contributions. A number of previous studies have been reported showing enhanced QM interaction energies in the case of biomolecule ligand interaction, illustrating the common trait of

QM cluster models in

overestimating the interaction energies.25-28 It is also interesting to note that the QM calculations are capable of distinguishing between the micromolar and nanomolar concentration of the SB derivatives. 5



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 24

More importantly, the QM calculations can appropriately identify the experimentally potent SB derivative BF-164 with the lowest binding energy among the SB derivatives. Thus the QM calculations are superior, quantitatively and qualitatively, in terms of predicting the binding energy profiles of SB derivatives , compared to classical force field based docking and the MMGBSA/MMPBSA methods. It is also worthwhile to mention that QM calculations based on a similar cluster model has already been successfully applied for predicting the potency of β-secretase inhibitors.29 The main difference to our work is that β-secretase has a very stable tertiary structure and is an enzyme, where the active site residues along with the respective inhibitor molecules were included in the QM calculations, whereas in our case Aβ42 has a very flexible structure with no well-defined binding sites. Thus in this case we relied upon the most stable Aβ42-SB derivative complex structure as obtained from the MD simulation. In order to make it computationally feasible, only residues within a distance of 4Å from the tracer were included in the cluster model. The importance of the dispersion term in stabilizing the complexes is estimated by calculating the interaction energies with B3LYP and dispersion corrected B3LYP (referred to as B3LYP-D3) functionals – the values of which are provided in the supporting information. Experimental vs Predicted binding affinity In order to assess the performance of different modeling strategies in reproducing experimental data the binding affinities of the SB derivatives were correlated against the predicted values. As the S1 site possesses the highest binding energy among all possible binding sites and is further considered for the MD simulation studies, the S1 binding energy values obtained with docking were plotted against the experimental binding energy values (refer to Figure 3). It is found from the results that the docking energy values correlates poorly with the experimental findings with a pearson’s r value of only 0.19. As can be seen in Figure 4, the MMGBSA method is associated with an r value of 0.39 and MMPBSA correlates with an r value of 0.38. It can be seen from the results that MMGBSA predicts the experimental results slightly better than the MMPBSA method which has already been observed in previous studies.30 In any case, the classical force field methods predict the experimental findings with a meager r value of 0.39 . The QM based interaction energies with and without solvent description using the PCM model are plotted against the experimental binding affinity values in Figure 5. The prediction of the binding affinity values is now considerably improved to an r value of 0.78 in both cases. We note that the two points that significantly deviate from the expected values correspond to the tracers BF-168 and BF-125. A possible reason for this deviation could be that BF-168 possesses the 6


Page 7 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


bulkiest substituent group among all the studied BF tracers and BF-125 possesses, apart from a bulky R2 substituent apart from having H-bond donor, acceptors as its other two substituent groups which may require explicit treatment of the hydrogen bonded solvent molecules. It is interesting that the QM method significantly improves the correlation and that the solvent environment does not have a pronounced effect in terms of improving the predictive ability. It is important to mention that the solvent effect also depends on the conformational flexibility and polarity of the ligand molecule involved. In the case of SB derivatives the core structure of the ligand is composed of aromatic rings connected by a C=C bond where the conformational freedom of the ligand is restricted. Thus, in this particular case the solvent may not have a profound effect in terms of improving the binding energy values. Plausible cause for improvement in prediction accuracy It is observed from the results that the prediction of the binding affinity with respect to the experimental values increases from docking to molecular dynamics which then significantly is further increased using the QM methodology. Many factors are responsible for this observation. The docking studies do not take into account protein flexibility and explicit solvent molecules (that are involved in hydrogen bonding) which is one of the major drawbacks of the method. At the same time, using MD simulations both the abovementioned shortcomings in the docking methodology can be rectified. However, it is often a challenge for the force field methods to distinguish protein-ligand systems that differ in energies by 0.5-1 kcal/mol - the reason being that the approximations involved in the parametrization itself contribute to the errors in that energy range. Above all, a significant factor missing is the electronic effect in the protein ligand interaction. As we are looking towards distinguishing the ligand molecules with similar structural motifs based on their binding affinities with a protein, it is important to account for the electronic effects. As the MD simulations are based on the nuclear motion of the atoms, with the electronic effects completely neglected, the accuracy of the results are purely limited to the accuracy of the force field used. Thus on switching into a QM based cluster model, there is a considerable improvement in the binding energy prediction. It is also important to note that the computational demand also increases while moving from docking to MD and to QM methods. Thus the application of multiple simulation strategies can indeed increase the binding affinity prediction of PET tracers (SB derivatives) with amyloid fibrils. Interestingly, a similar kind of result was reported recently on the potency prediction of the β-secretase inhibitor by Roos and coworkers.29 7



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 24

Hobza et al made valuable contributions in terms of the protein ligand binding energy predictions through the use of the semi-empirical PM6-DH2X approach where the non-covalent contributions are properly accounted for in the description of the interaction energy.28 We believe our approach can be useful when dealing with proteins like amyloid aggregates with highly flexible structures.

CONCLUSION A systematic study has been carried out to evaluate the efficiency of different computational methods in predicting the binding affinity of styrylbenzoxazole derivatives, commonly used as PET tracers, towards the amyloid fibril Aβ42. The results indicate that the docking, MMGBSA and MMPBSA methods poorly predict the experimental results (with r correlation values of 0.19, 0.39, 0.38 for docking MMGBSA,MMPBSA, values, respectively) and are also not able to distinguish between SB derivatives having nano molar and micro molar binding affinities. The above mentioned methods are also not able to identify the compound with lowest binding affinity among the series of SB derivatives studied. The QM method significantly improves the prediction of the experimental binding affinity values (with r=0.78) and distinguishes between the SB derivatives having nano molar and micro molar binding affinities apart from identifying the compound with lowest binding energy. Each of these modeling methods has its own shortcoming which will be overcome to a certain extent if they are combined into a workflow. Such a strategy can quide the design and development of PET tracers with improved binding affinities which in turn can be useful for better imaging of amyloid conditions and early detection. MATERIALS AND METHOD Protein In this study the structure of Aβ42 with the PDB ID 2BEG is taken as the amyloid protofibril structure which is a pentamer.31 It is well established that the amyloid exists predominately in two forms Aβ40 and Aβ42, here we have chosen Aβ42 because the experimental study on the styrylbenzoxazole derivatives were carried out with this fibril.

24

It is known from the previous studies that the Aβ42

protofibril has multiple binding sites for the tracers where they may be classified as surface binding sites and core binding sites. In the case of core sites three possible sites are identified where the two 8


Page 9 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


sites are present in the first conformer structure in the reported NMR models and the site specific to phenylalanine is present in the eighth conformer of the NMR structures.16 As the dodecamer (56-kDa soluble amyloid-β assembly) is found to be the causative agent for memory loss in the Alzheimer’s disease we here also modeled the systems as dodecamer peptides by replicating the penatmer (2BEG) structure appropriately.32

Ligand The 10 different styrylbenzoxazole derivatives were used as the ligand molecule in this particular study. The experimental binding affinity of the ligand molecules with Aβ42 were obtained from the reported previous study.24 The chemical structure of styrylbenzoxazole derivative is shown in Scheme 1. Docking All the possible sites are explored for the interaction with styrylbenzoxazole derivatives using docking. Since the amyloid protofibril does not have well defined binding sites, blind docking has been carried out using the autodock software.33 The grid box dimension has been chosen as 170X100X200 with grid size of 0.375 Å which is to make sure that even the surface binding sites are identified. In particular, the docking has been carried out for the 1st and 8th conformers of amyloid protofibril (as reported in 2BEG) so that all core sites are detected in the docking studies. 500 different low energy configurations for SB tracers have been identified using genetic algorithm and these configurations differ in terms of binding mode and binding pose in different binding sites. The structures of SB tracers with the least binding energy in different binding sites were considered for further analysis. Molecular Dynamics Simulation The Amber 12 package was used for the MD simulation.34 The low energy protein ligand complex structure is taken from docking and is solvated with cubic box of TIP3P water molecules up to 12Å from their edges. Counter ions were added to make the system neutral. FF99SB parameters were used for the proteins and GAFF parameters were used for the ligand molecules. The charges for the ligand molecules were obtained by fitting to electrostatic potential using CHELPG protocol as implemented in Gaussian09.35-38 The level of theory employed was B3LYP/6-31G*.The systems were minimized using the steepest descent algorithm followed by the equilibration using NVT and NPT ensemble with a 9



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 24

timescale of 100ps each. The temperature of the system was raised from 0 to 300K during NVT equilibration phase. The solute structure was constrained with a weak force (10 kcal/mol Å2) in the energy minimization and equilibration steps. Then the production run of 10ns in NPT ensemble was carried out for each system. In this production run, no constraint was applied on fibril and a fully flexible molecular mode was used. The temperature of the system was maintained at 300k and pressure was maintained at 1 atm, respectively. A 12 Å cutoff was set for short-range interactions, while the electrostatic interactions beyond the cutoff were computed with the particle mesh Ewald (PME) method.39 Bonds involving hydrogen atoms were constrained with the SHAKE algorithm. A time step of 2 fs was used, and the trajectory was saved every 2 ps. The trajectories were visualized using the VMD and Pymol.40,41 Binding Free Energy Calculations The python script MMPBSA.py of Amber 12 was used to do Molecular Mechanics Generalized Born Surface Area (MMGBSA) and Molecular Mechanics Poisson Boltzmann Surface Area (MMPBSA) calculations,42

which can be used to calculate the binding free energies of styrylbenzoxazole

derivatives with Aβ42 fibril. The data from the last 6 ns of the MD trajectories were used for MMGBSA calculations in the single trajectory mode; namely, the configurations of ligand, protein, and complex were extracted from the same trajectory (once in 30 frames, totaling to 100 configurations). This mode was efficient and has been used extensively in binding free energy calculations.30 The average values and standard errors were calculated from the results of all the extracted snapshots.

Quantum Mechanical Calculations In order to move forward with the quantum mechanical calculations it is inevitable to reduce the size of the system without much compromising the accuracy of the results. In this case we have analyzed the total energy of the system through simulations ranging from 4 to 10 ns (which is after RMSD equilibration (See Figure SI) and also used for MMGBSA and MMPBSA calculations) and the coordinates of the lowest energy configuration were used for the further QM calculations. As we are concerned about the binding energy of the ligand with Aβ42, the residues of the protofibril which are within 4Å of the respective outermost ligand coordinates are chosen and are appropriately saturated with hydrogen atoms and a similar protocol has been employed previously. 10


43

The coordinates of the

Page 11 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


protein moieties are constrained and the ligand was freely optimized in the protein environment at B3LYP-D3/6-31g* level of theory.44 The protein atoms are fixed in our quantum chemical calculation for the sake of truncation effect on proteins, but it should be considered that the geometry of the protein-ligand atoms are derived from the MD simulation where the protein structure is well equilibrated in the ligand and solvent environment. The calculations were carried out using the Gaussian 09 D.01 suite of programs.37 The interaction energy between the protein and ligand was calculated using the counter poise method as suggested by Boys and Bernandi correcting for basis set super position error.45 The interaction energy of the system was calculated by the equation given below Interaction Energy (I.E) = Energycomplex – ( Energyprotein + Energyligand )

(1)

The solvation effect of the QM system was evaluated using the PCM model of water and interaction energy of the complex systems was calculated as provided in the above given equation. We note that enthalpies were considered as the binding affinities of the Aβ42-PET and that the entropy contributions were not added as the calculations involve the optimization and mimics zero temperature condition. Supporting Information The graphical representation of the root mean square deviation(RMSD) of the Aβ42-SB systems during the course of simulation. The interaction energies calculated with the B97D3 dispersion corrected functional along with their correlation with experimental values are presented. In order to assess the origin of interaction stabilizing the complexes, the interaction energies are also calculated with the B3LYP functional (without dispersion corrections) and are presented. This material is available free of charge via the Internet at http://pubs.acs.org.

Acknowledgments The authors acknowledge support from the Swedish Foundation for Strategic Research (SSF) through the project ``New imaging biomarkers in early diagnosis and treatment of Alzheimer's disease'' and the support from SLL through the project ``Biomolecular profiling for early diagnosis of Alzheimer's disease''. We thank Prof. Agneta Nordberg and Prof. Bengt Långström for valuable discussions on the

11



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 24

project. This work was supported by the grants from the Swedish Infrastructure Committee (SNIC) for the projects “Multiphysics Modeling of Molecular Materials'' (SNIC2015-16-10).

References: (1) Fratiglioni, L., De Ronchi, D., and Aguero-Torres, H. (1999) Worldwide prevalence and incidence of dementia. Drugs aging, 15, 365− 375. (2) Berchtold, N. C., and Cotman, C. W. (1998). Evolution in the conceptualization of dementia and Alzheimer's disease: Greco-Roman period to the 1960s. Neurobiol. Aging 19, 173-189. (3) Tanzi, R.E., and Bertram, L. (2005) Twenty years of the Alzheimer's disease amyloid hypothesis: a genetic perspective. Cell, 120 , 545–555. (4) Hardy, J., and Selkoe, D. J. (2002) The amyloid hypothesis of alzheimer's disease: progress and problems on the road to therapeutics, Science, 297, 353- 356. (5) McGeer, P.L., Kamo, H., Harrop, R. McGeer, E.G., Martin, W.R., Pate, B.D., and Li, D.K. (1986) Comparison of PET, MRI, and CT with pathology in a proven case of Alzheimer's disease, Neurology, 36,1569–1574. (6) Jagust, W. (2009) Mapping brain beta-amyloid. Curr Opin Neurol.22, 356–361. (7) Nordberg, A., Rinne, J.O., Kadir, A., and Langstrom, B. (2010) The use of PET in Alzheimer disease. Nat Rev Neurol, 6,78–87. (8) Villemagne, V. L., Mulligan, R.S., Pejoska, S., Ong, K., Jones, G., O’Keefe, G., et al. (2012) Comparison of 11C-PiB and 18F-florbetaben for Abeta imaging in ageing and Alzheimer’s disease. Eur J Nucl Med Mol Imaging.39,983–9. (9) Joshi, A.D., Pontecorvo, M.J., Clark, C.M., Carpenter, A.P., Jennings, D.L., Sadowsky, C.H., et al. (2012) Performance characteristics of amyloid PET with florbetapir F 18 in patients with Alzheimer’s disease and cognitively normal subjects. J Nucl Med. 53, 378–384. (10) Vandenberghe, R., Van Laere, K., Ivanoiu, A., Salmon, E., Bastin, C., Triau, E., et al. (2010) 18F-flutemetamol amyloid imaging in Alzheimer disease and mild cognitive impairment: a phase 2 trial. Ann Neurol. 68:319–329. (11) Cselenyi, Z., Jonhagen, M.E., Forsberg, A., Halldin, C., Julin, P., Schou, M., et al. (2012) Clinical validation of 18F-AZD4694, an amyloid-beta-specific PET radioligand. J Nucl Med. 53:415–424. (12) Heather, A., Carlson, and McCammon, J.A., (2000) accommodating protein flexibility in computational drug design ,Mol Pharmacol, 57, 213-218. (13) Jung-Hsin, L., Perryman, A. L., Schames, J.R., and McCammon, J.A. (2002) Computational drug design accommodating receptor flexibility: the relaxed complex scheme. J. Am. Chem. Soc. 124, 5632-5633. (14) Terstappena, G. C. and Reggiania, A. (2001) Trends in pharmacological sciences, 22, 23–26. (15) Kitchen, D. B., Decornez, H., Furr, J.R. and Bajorath, J. (2004) Docking and scoring in virtual screening for drug discovery: methods and applications. Nature Reviews Drug Discovery, 3, 935-949.

12


Page 13 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(16) Kuang, G., Murugan, N.A., Tu, Y., Nordberg, A., and Ågren, H. (2015) Investigation of the binding profiles of AZD2184 and thioflavin T with Amyloid-β(1–42) fibril by molecular docking and molecular dynamics methods. J. Phys. Chem. B, 119, 11560–11567. (17) Murugan, N.A., Halldin, C., Nordberg, A., Långström, B., and Ågren, H (2016) The culprit is in the cave: the core sites explain the binding profiles of amyloid-specific tracers. J. Phys. Chem. Lett., 7, 3313-3321. (18) Skeby, K.K., Sørensen, J., and Schiøtt, B. (2013) Identification of a common binding mode for imaging agents to amyloid fibrils from molecular dynamics simulations. J. Am. Chem. Soc.135, 15114−15128. (19) Murugan, N. A., Olsen, J. M. H., Kongsted, J., Rinkevicius, Z., Aidas, K., and Ågren, H. (2013) Amyloid fibril-induced structural and spectral modifications in the thioflavin-T optical probe. J. Phys. Chem. Lett. 4, 70−77. (20) Wetzel, R., Shivaprasad, S. and Williams, A.D. (2007) Plasticity of amyloid fibrils. Biochemistry, 46 ,1–10. (21) Cross, J.B., Thompson, D.C., and Humblet, C. (2009) Comparison of several molecular docking programs: pose prediction and virtual screening accuracy. J. Chem. Inf. Model. 49,1455–1474. (22) Ferrara, P., Gohlke, H., and Brooks, C.L. (2004) Assessing scoring functions for protein-ligand interactions. J. Med. Chem. 47, 3032–3047. (23) Genheden, S., and Ryde, U. (2015) The MM/PBSA and MM/GBSA methods to estimate ligand-binding affinities. Expert Opin Drug Discov. 10, 449–461. (24) Okamura, N., Suemoto, T., Shimadzu, H., Suzuki, M., Shiomitsu, T., Akatsu, H., Yamamoto, T., Staufenbiel, M., Yanai, K., Arai, H., Sasaki, H., Kudo, Y.,and Sawada, T. (2004) Styrylbenzoxazole derivatives for in vivo imaging of amyloid plaques in the brain. J Neurosci 24, 2535–2541. (25) Peräkylä, M.; Pakkanen, T. A. (1994) Quantum mechanical model assembly study on the energetics of binding of arabinose, fucose, and galactose to L-arabinose-binding protein. Proteins: Struct., Funct. Genet., 20, 367−372. (26) Shi, J. Y.; Lu, Z. A.; Zhang, Q. L.; Wang, M. L.; Wong, C. F.; Liu, J. H. (2010) Supplementing the PBSA approach with quantum mechancis to study the binding between CDK2 and N-2substituted O-6-cyclo-hexylmethoxyguanine inhibitors. J. Theor. Comput. Chem., 9, 543−559. (27) Poongavanam, V.; Steinmann, C.; Kongsted, J.(2014) Inhibitor ranking through QM based chelation calculations for virtual screening of HIV-1 RNase H inhibition. PLoS One, 9, e98659. (28) Dobeš, P., Řezáč, J., Fanfrlík, J., Otyepka, M., and Hobza, P.(2011) Semiempirical quantum mechanical method PM6-DH2X describes the geometry and energetics of CK2-inhibitor complexes involving halogen bonds well, while the empirical potential fails. J. Phys. Chem. B, 115, 8581-8589. (29) Roos, K., Viklund, J., Meuller, J., Kaspersson, K., and Svensson, M. Potency prediction of ßsecretase (BACE-1) inhibitors using density functional methods. (2014) J. Chem. Inf. Model. 54, 818−825. (30) Hou, T. J., Wang, J. M., Li, Y. Y., and 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. J. Chem. Inf. Model. 51, 69−82.

13



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 24

(31) Luhrs, T., Ritter, C., Adrian, M., Riek-Loher, D., Bohrmann, B., Dobeli, H., Schubert, D., and Riek, R. (2005) 3D structure of Alzheimer's amyloid-β(1-42) fibrils. Proc.Natl.Acad.Sci.USA. 102, 17342-17347. (32) Lesne,S.,Teng Koh,M., Kotilinek, L., Kayed, R.,Glabe, C.G.,Yang, A., Gallagher, M and Ashe, K.H. (2006) A specific amyloid-β protein assembly in the brain impairs memory. Nature, 440, 352-357. (33) Morris, G. M., Huey, R., Lindstrom, W., Sanner, M. F., Belew, R. K., Goodsell, D. S. and Olson, A. J. (2009) Autodock4 and AutoDockTools4: automated docking with selective receptor flexiblity. J Comput Chem. 16, 2785-2791. (34) Case, D.A. Darden, T.A. Cheatham III, T.E. , Simmerling, C.L., Wang, J., Duke, R.E., Luo, R., Walker, R.C., Zhang, W., Merz, K.M., Roberts, B., Hayik, S., Roitberg, A., Seabra, J. Swails, A.W. Götz, I. Kolossváry, K.F. Wong, F. Paesani, J. Vanicek, R.M. Wolf, G., Liu, J., Wu, X., Brozell, S.R., Steinbrecher, T., Gohlke, H., Cai, Q., Ye, X., Wang, J., Hsieh, M.-J., Cui, G., Roe, D.R., Mathews, D.H., Seetin, M.G., Salomon-Ferrer, R., Sagui, C., Babin, V., Luchko, T., Gusarov, S., Kovalenko, A., and Kollman, P.A. (2012), AMBER 12, University of California, San Francisco. (35) Hornak, V., Abel, R., Okur, A., Strockbine, B., Roitberg, A., and Simmerling, C. (2006) Comparison of multiple Amber force fields and development of improved protein backbone parameters. Proteins. 65,712-725. (36) Wang, J., Wolf, R.M., Caldwell, J.W., Kollman, P.A., and Case, D.A. (2004) Development and testing of a general amber force field. J Comput Chem. 25, 1157-1174. (37) Gaussian 09, Revision D.01, Frisch, M. J., Trucks, G. W., Schlegel, H. B., Scuseria, G. E., Robb, M. A., Cheeseman, J. R., Scalmani, G., Barone, V., Mennucci, B., Petersson, G. A., Nakatsuji, H., Caricato, M., Li, X., Hratchian, H. P., Izmaylov, A. F., Bloino, J., Zheng, G., Sonnenberg, J. L., Hada, M., Ehara, M., Toyota, K., Fukuda, R., Hasegawa, J., Ishida, M., Nakajima, T., Honda, Y., Kitao, O., Nakai, H., Vreven, T., Montgomery, J. A., Jr., Peralta, J. E., Ogliaro, F., Bearpark, M., Heyd, J. J., Brothers, E., Kudin, K. N., Staroverov, V. N., Kobayashi, R., Normand, J., Raghavachari, K., Rendell, A., Burant, J. C., Iyengar, S. S., Tomasi, J., Cossi, M., Rega, N., Millam, J. M., Klene, M., Knox, J. E., Cross, J. B., Bakken, V., Adamo, C., Jaramillo, J., Gomperts, R., Stratmann, R. E., Yazyev, O., Austin, A. J., Cammi, R., Pomelli, C., Ochterski, J. W., Martin, R. L., Morokuma, K., Zakrzewski, V. G., Voth, G. A., Salvador, P., Dannenberg, J. J., Dapprich, S., Daniels, A. D., Farkas, Ö., Foresman, J. B., Ortiz, J. V., Cioslowski, J., and Fox, D. J. (2009) Gaussian, Inc., Wallingford CT. (38) Wang, J., Wang, W., Kollman P. A., and Case, D. A. (2006) Automatic atom type and bond type perception in molecular mechanical calculations. J Mol Graph and Model, 25, 247-260. (39) Essmann, U., Perera, L., Berkowitz, M. L., Darden, T., Lee, H., Pedersen, and L. G. A. (1995) Smooth particle mesh ewald method. J. Chem. Phys. 103, 8577−8593. (40) Humphrey, W., Dalke, A., and Schulten, K. (1996) VMD: Visual Molecular Dynamics. J. Mol. Graphics. 14, 33−38. (41) The PyMOL Molecular Graphics System, Version 1.3, Schrödinger, LLC. (42) Miller, B. R., McGee, T. D., Swails, J. M., Homeyer, N., Gohlke, H., and Roitberg, A. E. (2012) MMPBSA.py: An efficient program for end-state free energy calculations. J. Chem. Theory Comput. 8, 3314−3321.

14


Page 15 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(43) Yilmazer, N.D., and Korth, M. (2013) Comparison of molecular mechanics, semi-empirical quantum mechanical, and density functional theory methods for scoring protein–ligand interactions. J. Phys. Chem. B, 117, 8075-8084. (44) Grimme, S., Antony, J., Ehrlich, S., and Krieg, H. (2010) A consistent and accurate ab initio parametrization of density functional dispersion correction (DFT-D) for the 94 elements H-Pu. J Chem Phys.132,154104. (45) Boys, S. F., and Bernardi, F. (1970) The calculation of small molecular interactions by the differences of separate total energies. Some procedures with reduced errors. Mol. Phys. 19, 553−566.

15



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 24

Table 1. The details of various SB derivatives along with their dissociation constants and binding affinity values.

S.No

Compound

X

R1

R2

Ki(nm)

1 2 3 4 5 6 7 8 9 10

BF-208 BF-191 BF-164 BF-169 BF-165 BF-168 N-282 BF-148 BF-125 BF-124

O O O O O O O O O S

H H H H OH O(CH2)2F H F H H

F Cl NH2 NH(CH3) NH(CH3) NH(CH3) N(CH3)2 N(CH3)2 N(C2H5)2 N(C2H5)2

5000 5000 0.38 7.10 1.80 6.40 4.30 4.20 4.90 10.90

*binding affinity values were calculated using the ∆G=-R*T*lnK

16


Binding Affinity* (kcal/mol) -7.22 -7.22 -12.84 -11.11 -11.91 -11.17 -11.41 -11.42 -11.33 -10.86

Page 17 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 2. The interaction energy of the SB derivatives with various possible binding sites of the Aβ42 calculated using docking. S.No Compound Binding Affinity (kcal/mol) 1 BF-208 -7.22 2 BF-191 -7.22 3 BF-164 -12.84 4 BF-169 -11.11 5 BF-165 -11.91 6 BF-168 -11.17 7 N-282 -11.41 8 BF-148 -11.42 9 BF-125 -11.33 10 BF-124 -10.86

S4 S3 S2 S1 (kcal/mol) (kcal/mol) (kcal/mol) (kcal/mol) -6.43 -6.99 -6.50 -6.58 -6.50 -6.49 -6.95 -6.90 -7.47 -7.67

-8.58 -8.86 -7.85 -8.39 -8.09 -8.62 -8.89 -8.53 -9.19 -9.50

-8.80 -9.54 -8.58 -9.10 -9.40 -9.61 -9.68 -9.65 -10.58 -10.40

-8.97 -9.82 -9.11 -9.45 -9.55 -9.84 -10.11 -10.03 -10.85 -10.99

Table 3. The free energy of binding of the SB derivatives with the S1 binding site of the Aβ42 calculated using the MMGBSA and MMPBSA methods. *all values are entropy corrected and in the units of kcal/mol S.No

Compound

1 2 3 4 5 6 7 8 9 10


Binding Affinity (kcal/mol) -7.22 -7.22 -12.84 -11.11 -11.91 -11.17 -11.41 -11.42 -11.33 -10.86

MMGBSA* (kcal/mol) -21.82 -25.68 -25.45 -26.24 -23.99 -27.83 -25.70 -26.01 -26.26 -29.33

MMPBSA* (kcal/mol) -40.37 -45.99 -48.43 -44.69 -40.46 -53.64 -47.90 -49.73 -48.29 -51.97

17



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 24

Table 4. The various components of the binding energy calculated using MMGBSA method for the SB derivatives with the Aβ42. S.No 1 2 3 4 5 6 7 8 9 10

Compound BF-208 BF-191 BF-164 BF-169 BF-165 BF-168 N-282 BF-148 BF-125 BF-124

∆EvdW ∆Eelec -40.04±0.20 -5.73±0.16 -44.89±0.22 -3.84±0.15 -47.12±0.21 -7.85±0.21 -46.69±0.15 -2.03±0.15 -46.74±0.27 -11.50±0.20 -50.09±0.25 -4.16±0.21 -46.26±0.32 -4.45±0.17 -46.93±0.48 -3.40±0.20 -47.79±0.29 -0.04±0.11 -48.54 ±0.22 -5.03±0.22

∆GGB 15.66±0.12 13.45±0.11 18.32±0.17 13.12±0.15 23.29±0.20 16.58±0.24 14.46±0.24 15.19±0.27 9.59±0.14 13.73±0.21

∆GSASA -4.72±0.00 -4.92±0.00 -4.84±0.01 -5.21±0.01 -5.35±0.01 -5.69±0.01 -5.33±0.01 -4.99±0.03 -5.31±0.02 -5.03±0.02

-T∆S 13.01±1.34 14.51±0.62 16.04±0.64 14.56±0.72 16.31±0.76 15.53±0.61 15.88±0.56 14.12±0.59 17.28±0.79 15.54±0.61

∆Gbind -21.82 -25.68 -25.45 -26.24 -23.99 -27.83 -25.70 -26.01 -26.26 -29.33

Table 5. The various components of the binding energy calculated using MMPBSA method for the SB derivatives with the Aβ42. S.No 1 2 3 4 5 6 7 8 9 10

Compound BF-208 BF-191 BF-164 BF-169 BF-165 BF-168 N-282 BF-148 BF-125 BF-124

∆EvdW -40.04±0.20 -44.89±0.22 -47.12±0.21 -46.69±0.15 -46.74±0.27 -50.09±0.25 -46.26±0.32 -46.93±0.48 -47.79±0.29 -48.54±0.22

∆Eelec -5.73±0.16 -3.84±0.15 -7.85±0.21 -2.03±0.15 -11.50±0.20 -4.16±0.21 -4.45±0.17 -3.40±0.20 -0.04±0.11 -5.03±0.22

∆GPB 19.37±0.18 16.36±0.20 18.33±0.19 18.93±0.25 30.88±0.36 17.74±0.25 17.49±0.21 16.40±0.29 13.48±0.16 16.77±0.20

∆GPol -26.98±0.03 -28.13±0.03 -27.83±0.02 -29.46±0.03 -29.41±0.09 -32.67±0.10 -30.57±0.08 -29.92±0.20 -31.23±0.11 -30.71±0.14

-T∆S 13.01±1.34 14.51±0.62 16.04±0.64 14.56±0.72 16.31±0.76 15.53±0.61 15.88±0.56 14.12±0.59 17.28±0.79 15.54±0.61

∆Gbind -40.37 -45.99 -48.43 -44.69 -40.46 -53.64 -47.90 -49.73 -48.29 -51.97

Table 6. The MMGBSA and MMPBSA free energy values calculated during various time durations of MD simulation (i) 4-10ns and (ii) 8-10ns. S.No

Compound Exp binding 4-10ns affinity MMGBSA* MMPBSA* 1 BF-208 -7.22 -21.82 -40.37 2 BF-168 -11.17 -27.83 -53.64 3 BF-164 -12.84 -25.45 -48.43 *all values are entropy corrected and in the units of kcal/mol 18


8-10ns MMGBSA* -21.58 -27.83 -24.87

MMPBSA* -38.98 -52.63 -48.53

Page 19 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 7. The interaction energy of the SB derivatives with the S1 binding site of the Aβ42 calculated using the B3LYP-D3/6-31g*. S.No

Compound

1 2 3 4 5 6 7 8 9 10


Binding Affinity (kcal/mol) -7.22 -7.22 -12.84 -11.11 -11.91 -11.17 -11.41 -11.42 -11.33 -10.86

QM I.E (kcal/mol) -42.65 -43.26 -61.37 -53.79 -54.86 -57.59 -49.90 -49.01 -45.84 -50.93

QM I.E PCM (kcal/mol) -56.44 -56.03 -74.48 -65.33 -68.58 -73.79 -62.84 -65.27 -59.50 -66.42

Figure 1. A schematic representation of the important binding sites in the Aβ42.

19



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Scheme 1. Chemical structure of the styrylbenzoxazole.

20


Page 20 of 24

Page 21 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 2. Representation of the Aβ42-ligand complex in the MD simulation snapshot and in the quantum mechanical cluster model.

21



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 24

Figure 3. Docking calculated binding affinity vs experimental binding affinity of SB derivatives with Aβ42.

Figure 4. MMGBSA (A) and MMPBSA (B) calculated binding affinity vs experimental binding affinity of SB derivatives with Aβ42.

B

A

22


Page 23 of 24

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 5. QM gas phase (A) and QM PCM (B) calculated binding affinity vs experimental binding affinity of SB derivatives with Aβ42.

B

A

23



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Table of Contents Graphics

24


Page 24 of 24

Multistep Modeling Strategy To Improve the Binding Affinity Prediction

Recommend Documents