The model proposed by Grant & Pickup [J. Phys. Chem. 99, 3503-3510 (1995)] is a well-established method to estimate accurately volume and surface areas of molecules. The method is based on the inclusion-exclusion formula of statistical physics (also known as the Poincaré formula). It states that the volume of an object composed of multiple overlapping bodies is given by the sum of the volumes of the bodies, minus the sum of the overlap volumes between pairs of bodies, plus the sum of the overlap volumes between triplets of bodies, and so on:
Atomic volumes are represented by Gaussian densities and overlap volumes are computed using standard Gaussian overlap integrals. The model is fully analytic. For example, the solvent-accessible surface area of an atom is obtained as the derivative of the molecular volume with respect to the atomic radius of an atom.
The model leads to a tree-based algorithm in which overlap volumes are recursively evaluated.
We have used this model extensively to implement the non-polar component and the self-adjusting pair-wise descreening method for the Generalized Born solvation electrostatic component of the of the AGBNP solvation model. Our CPU implementation is based on a dept-first traversal of the tree.
Lately we have been working on a GPU implementation of the Gaussian volumetric model. As one can imagine, tree-based recursive algorithms do not easily lend themselves to GPUs. However, we finally found an efficient solution based on a flat arrays representation of the tree and breadth-first traversal (see paper in JCC).
According to Newman et al., in all D3 receptor antagonists a salt bridge between the protonated amine in the primary pharmcophore and Asp 110 is observed. Previous antagonists candidates having a stepholidine ring as a primary pharmacophore have been synthetized in our lab. The protonated nitrogen of the pharmacophore makes the same interaction with Asp 110 observed in Newman et al.10 Some of the molecules used in this study like 216F (figure 4 left) have the same feature. However, molecules like 217F don't do the same interaction (figure 4 right). In order to find out the reason why those molecules interact differently 216F and 217F were superimposed (figure 5). From the superimposition it looks that there is electronic repulsion between a cyano group in para of the aromatic ring and the carbonyl of Val 189. This repulsion would cause the molecule to rearrange in another way in order to fit in the receptor. This rearrangement would prevent the protonated amine of 217F to make the salt bridge with Asp 110 in the primary binding site (OBS). This phenomenon occurs in all molecules with a substitution in para.
Figure 4. left, SG-216F makes the salt bridge with Asp 110. Righ, SG-217F does not make the salt bridge interaction with Asp 110.
Figure 5. SG-216F in blue and SG-217F in pink. SG-217F cannot make the salt bridge interaction because of the clash between the cyano group and valine 189.
Prediction of Binding Energy affinities of Cucurbituril clip(host) with various guests as a part of SAMPL5 Challenge
By: Divya K.Matta
Ph.D student in Chemistry
The binding energies of Cucurbituril clip with various guests have been calculated using Binding energy distribution Analysis Method(BEDAM). The method employs AGBNP2(Analytic Generalized Born Plus NonPolar2) as an implicit(continuum) solvation model. Molecular Dynamic(MD) simulations were carried out to predict the binding free energies of these host-guest systems. Cucurbituril clip is deprotonated and has a charge of -4 due to the presence of four sulfonate groups.The values of Binding energies of these host guests system depends on the number of factors such as the strength of interactions between them, for example, depending upon whether they have hydrogen bonding, electrostatic or hydrophobic interactions, the binding energies vary accordingly. Also, it depends on the charge of guests molecules whether they are neutral or ionised.With the guidance of Prof.Emilio, I learnt the BEDAM methodology and its applications on host-guest systems.
Image :Interaction of Cucurbituril clip(host) with the guest
Our lab participated in the SAMPL5 host-guest blind challenge During September 2015 to February 2016. The challenge aimed at the use of computational methodologies to study and rank the binding of a set of guest molecules with three specific hosts. In this work, Hydration Site Analysis(Young et al., 2007; Lazaridis, T. 1998) was used to score the water sites within the binding pocket to favor the binding of the guests and the binding free energy was calculated using BEDAM(Gallicchio et al. 2010) with the incorporation of salt effects.
A summary of the final result was presented as a talk and also as a poster at the SAMPL5 workshop.
by Godfrey Rollins
HSP90 is a heat shock protein that plays an important role in our bodies by causing them to function properly. This role includes assisting in correct protein folding as well as stabilizing them when they are exposed to high temperatures. However, cancerous cells rely on the protein for growth and survival. In addition to that, these cells produce more of this protein than their healthy counterparts. Inhibition of this protein makes it easier for the cancer cells to be destroyed. Several HSP90 inhibitors were discovered and synthesized and are tested against this protein. As there are few crystal structures of the inhibitors present, more structures need to be predicted in order to have a better on how these inhibitors alter their structure and bind to HSP90. In this project, we use computational docking methods in order to predict what these structures of the complexes the inhibitors form when they bind to HSP90.
We first start off by obtaining a structure of an HSP90 inhibitor. We check it to see that the atoms are correct and that the formal charge of the inhibitor is 0. In this example, we use BIIB021.
Secondly, we prepare possible structures of the inhibitor using LigPrep. This results in various protonation states and possible tautomers. Here, we see a protonated N atom on the left aromatic ring on BIIB021.
Thirdly, we create a receptor grid of a known crystal structure we obtained. In this example we will dock BIIB021 onto a receptor grid obtained from the crystal structure with PDB ID code: 4B7P This code consists of another inhibitor, NMS-E973 bound to HSP90. The receptor grid showing the binding site is seen below.
Finally, we dock our inhibitor onto the receptor grid.
From what we see here, BIIB021 in its protonated state forms a hydrogen bond with Aspartic Acid and a halogen bond with Lysine. This is one of the possible structures of BIIB021 when it binds to HSP90. As several of these inhibitors have different structures, there are different possible complex structures that can be predicted.
2D Structures of the inhibitors used in this study are available in the PDF file below. FP IC50 values are expressed in µM.
by Sajeev Saluja
In a recent work to appear in Protein Science (see this post) we have worked with Dr. Daniele di Marino and collaborators in Italy and Belgium to predict the elusive structure of the complex between the Cytoplasmic Fmrp Interacting Protein 1 (CYFIP1) and the eukaryotic translation initiation factor 4E (eIF4E). This interaction is known to be disrupted in people affected by Fragile X syndrome, one of the major causes of autism in children. Researchers have not been able to solve the structure of this complex by experimental means. In this work, illustrated below by means of movies of molecular dynamics trajectories, we have obtained what we believe is a very good guess of how these two proteins might be interacting.
This work has been made possible by the WEB Computing Grid at Brooklyn College. Thanks!
The structure of the CYFIP1p-eIF4E complex is predicted by parallel BEDAM conformational search. CYFIP1p is found to bind in the same region as other known peptide inhibitors, but in a unique orientation.
The predicted structure showed remarkable stability during 50 ns of explicit solvent molecular dynamics with the Desmond program (DE Shaw Research, New York, NY). In contrast, the CYFIP1 peptide unfolded when prepared in other poses.
Simulation of Monovalent and Bivalent Salts : Does a Computer Model Distinguish Soluble and Non-Soluble Salts?
By Shivam Suleria
Undergraduate student at Brooklyn College of CUNY
The aim of the research was to find out that whether a computer model, the DESMOND MD software in the Maestro program (Schrodinger, inc.), is able to distinguish between small samples of crystals of soluble and non-soluble salts. According to hypothesis, the software should distinguish between various monovalent and bivalent salts. Small crystals of Sodium Chloride (cubic crystal structure, monovalent) and Calcium Carbonate (trigonal rhombohedral, bivalent) were selected. Proper structures were constructed using data from crystallographic databases.
First, a typical 64 atom crystal of NaCl was taken in a cubic box of dimension 22 Angstroms. The salt was then hydrated. The molarity obtained after adding the water was 9.979M. The expectations were that the salt will dissolve quickly. Several simulations of 1.2ns, 3.0ns and 6.0ns were done. The results showed that the NaCl crystal dissolves quickly in water (in around 5.0ns). Hence the software correctly predicts dissolution of the crystal into solvated ions (Na+ and Cl-). This is what was expected.
Later on, a small 54 atoms crystal of Calcium carbonate was taken in a cubic box with 20.5 Angstrom sides. Water was then added to it to obtain a molarity of 10.407M. The expected result was that small crystal will not dissolve at all. Molecular simulation for 50.0ns was ran. Results showed that even after 50.0ns the atoms present in the sample remained together. Hence the assumption was correct. The salt did not get dissolved. Hence it is an evidence that the software, Schrodinger Maestro, is able to distinguish between various salts. The results supported the hypothesis.
Sodium chloride dissolves rapidly in water (6 ns)
Calcium carbonate does not dissolve within 50 ns of simulation.
This was part of an undergraduate honor project for the General chemistry II class of Spring 2015 at Brooklyn College.
Over the past couple of weeks further optimizations were applied to the AGBNP3 algorithm. These mainly focused on the pairwise-descreening calculation for Born radii. Prior to this, the descreening kernel employed an interpolation procedure which utilizes two C-spline calculations. The new algorithm utilizes only one. The benefits are substantial, especially for the larger systems, as illustrated below (see May results vs. April results):
These results complete the main phase of the AGBNP3 serial performance optimization project. We nearly tripled the speed of the code relative to AGBNP2 without non-bonded cutoffs and doubled it relative to AGBNP2 with cutoffs (for system sizes illustrated above). Future efforts will focus on parallelization on multi-core and GPU architectures.
Last weekend I presented my work at the Sigma Xi Northeastern Regional Research Conference at Western Connecticut State University. The conference aims to showcase undergraduate research, and it was certainly exciting to share my research with others and get to see all of the interesting research being done by other undergraduates.