Inhibitors of Human Dihydrofolate Reductase : A Computational Design and Docking Studies Using Glide

Abstract: Dihydrofolate reductase (DHFR) plays a vital role in the DNA synthesis by reducing dihydrofolic acid to tetrahydrofolic acid which is an essential component. Synthetic ligands like methotrexate (MTX), aminopterin (AMP) and their analogues act as potential anti metabolites by mimicking the coenzyme dihydrofolic acid (DHFA) they inhibit the activity of DHFR antagonistically. Several ligands which are similar to MTX analogues and 6, 8 substituted 2 – naphthyls (NAP) which can mimic the pteridyl group of DHFA have been computationally designed. These ligands were proposed to hinder the formation N, N methylene tetrahydrofolic acid (THFA) coenzyme, which is essential for the DNA synthesis. The docking studies were done using grid, generated with the 0.9 Vander Waals scaling for non polar bonds in the active site of the receptor. These newly designed ligands such as 14 -21 ,23 and 28 have shown good docking scores and predicted activities when compared to already existing ligands MTX and its analogues.


Introduction
The therapeutic effect of the clinical drug will be effective if the biochemical pathway of the enzyme can be exploited.DHFR enzyme plays a vital role by catalyzing the biosynthesis of nucleotide (DNA).DNA is a very important genetic component which influences the occurrence and prognosis of cancer.The reduction of dihydrofolate to tetra hydrofolate is catalyzed by DHFR via transfer of hydride from NADPH.THFA thus formed picks up one carbon unit from l-serine and is converted to coenzyme N 5 , N 10 methylene tetrahydrofolic acid (THFA) where CH 2 group bridges N 5 and N 10 tetra hydrofolate 1 .This coenzyme is utilized by thymidylate synthase for the synthesis of DNA.In this process the hetero cyclic ring, the methylene group at the 9 th position and N 10 methylated and demethylated aminopteridine play a vital role while undergoing reduction or formation of N 5 , N 10 methylene tetrahydrofolic acid coenzyme.
Taking these points into account several ligands similar to MTX analogues having different substituents were designed computationally.The designed molecules had changes on methylene group at 9 th position, substituting different substituents at N 10 and also totally replacing the pteridyl group by naphthyl group.So that these ligands can be used for inhibiting the formation of N 5 , N 10 methylene tetrahydrofolic acid coenzyme.

Materials and methods
The ligands were built using Maestro build panel and prepared by Lig Pep 2.0 application which uses MMFF 94s 2 force field and gave the corresponding low energy 3D conformers of the ligands.These ligands were then neutralized and checked for their ADME 3 properties using Qikprop.Qikprop helps in analyzing the pharmacokinetics and pharmacodynamics of the ligand by accessing the drug like properties.Before docking the screened ligands into the protein active site, the protein was prepared by deleting the substrate cofactor as well as the crystallographically observed water molecules and then the active site of the protein was defined for generating the grid.Receptor Vander Waals scaling for the non polar atoms was set to 0.9 4 which makes the protein site "roomier" by moving back the surface of non Polar regions of the protein and ligand.This kind of adjustments emulate to some extent the effect of breathing motion to the protein site, it is a kind of giving breathing to the receptor, this approach softens the active site region of the receptor making it flexible. 5he screened ligands were docked into the prepared grid, for which "standard precision mode" was selected.The ligands with known human activity values were taken for the calculation of predicted activity.
Regression analysis for these ligands were calculated by using activity values of the existing ligands 6 gave correlation coefficient (R) as 0.930334 and standard error of estimate (SEE) as 0.71762.Activity = (-1.6869* Dock score) -6.76188 (1) Using this equation the predicted activity for the rest of the ligands has been calculated by incorporating the dock score into the equation obtained by regression.Among 19 best docked ligands 10 had good predicted activity than the existing ligand.COP was not incorporated due to lack of interaction of the core with the active site.Activities of known molecules, Predicted activity and Dock Score values are included in the following Table-1 and corresponding graphs in Figure -1.

Results and Discussion
Computational strategies for structure based drug discovery offer a valuable alternative to the costly and time consuming process of random screening 7 .Synthetic programs directed towards the design of new DHFR inhibitors have most often followed an empirial medicinal approach involving replacement of ring nitrogens by carbon 8,9 .Although a large number of MTX analogues featuring diamino pyrimidine and diamino quinazoline motif have been synthesized, number of them have been quite potent and in some cases fairly selective.
An important quantification has to be made whenever docking simulations of small molecules to enzymes or other complex protein structures are performed, namely consideration of a protein as a rigid structure which may lead to an oversimplified view of the system.Thus a drawback of such a static model is that it does not take into account the possibility that the protein can undergo a ligand induced conformational change of its 3D structure.To vigorously address the issue, a modified Vander Waals scaling lesser than 1.0 (rigid) i.e. 0.9 has been adopted to avoid serious steric clashes with the rigid site 10 .
Our computationally designed molecules showed greater affinities towards the hDHFR receptor compared to the already existing molecules.Total 400 screened ligands having pteridine, quinazoline and naphthalene cores were docked into the generated grid.Out of which 130 were computationally designed molecules.It was found that 14 newly designed molecules (14-21 and 23-28) gave good docking scores compared to the existing ligands.Their structures, docking scores were included in Figure 2 and Table 1 respectively.Substituting hydrogen's of the methylene group at the 9 th position by amine, phenyl amine and replacing phenyl ring of p-amino benzoic acid (PABA) with the pyridine ring have given good docking scores with top ranking.When the pteridyl group was totally replaced by naphthalene (NAP) then also gave good ranking compared to the existing highly potent ligands.These molecules had much higher docking scores compared to the MTX (-9.11),COP (-8.75), and the coenzyme DHFA (-8.42).A comparison of the simulated contact distances between the receptor and ligand several key amino acid residues in the active site of hDHFR were carried out for all high ranking designed molecules such as 14 -16,17,18 and 19.Models of the binding compounds 14, 16-18 and 28 to the active site of hDHFR were generated using the molecular graphics program Glide 11 to replace the tight binding inhibitor COP in the corresponding enzyme complexes of known structure A stereoscopic image for pteridine core is displayed in Figure 3 where as the other naphthalene core is given in Figure 4.As required the 2, 4-diamino groups of pteridyl ring of MTX analogues 12 (14-21) the 2 nd amino group interacts in an unusual manner with gamma carbonyl group of Glu-30 and 4 th amino group interacting with Val-115 and Ile-7.When the simulations were checked for the naphthyl group the same interactions were seen as that of pteridyl ring for the ligands 23, 24, and 25 of greatest interest.When the side chain glutamic acid interactions were compared compounds of pteridyl ring and naphthyl ring showed that the gamma carboxylic acid group of glutamic acid side chain interacts with Asn-64 and alpha carboxylic acid group interacts with Gln-35 and Arg-70.In compounds with pteridyl ring, when hydrogen of the methylene group at the 9 th position was substituted by amine, gave a top score of (-12.21K.cal/mol).Replacing the phenyl ring of PABA with pyridine ring and substitution of the methylene hydrogen with ethyl and amino methyl, chloro, amine, and phenyl amine gave docking scores of -11.91, -11.68, and -11.69 respectively.When an ester containing bridge was used to connect the 2, 4-diamino naphthalene and aromatic ring gave a docking score of -11.74 standing at 4 th position in the ranking list.Other naphthalene molecules were ranked next to these molecules as 7, 8, 9 and 10.The existing molecules were ranked at 13 th and 14 th position with G-scores of -11.19 and -11.10 respectively.Whereas the MTX, coenzyme DHFA and COP ranked at 62, 90 and 70 with docking scores of -9.11, -8.42 and -8.75 respectively.

Conclusion
The designed molecules having substitutions on methylene group at 9 th position, replacing the phenyl ring of PABA with pyridine ring, molecules with ester linkage at bridge region connecting the naphthalene and phenyl ring and replacement of pteridine with naphthalene core have given hit molecules of greatest interest.These molecules have greater affinity towards the hDHFR compared to the existing molecules.Their QSAR and synthetic studies are in progress.
and predicted activity (b) Figure 1.(a) Scatter plot of experimental activity (-log Ki) verses Docking Score.(b) Comparison plot of experimental activity (in blue) verses Predicted activity (in pink).

Figure 2 .
Figure 2. Top ranking ligand structures 1 -13 and 22 are already existing MTX analogues and 29 is the co crystallized ligand in the hDHFR (1OHJ) protein, and others are computationally designed molecules.

Figure 3 .Figure 4 .
Figure 3. DHFR protein with ligand of pteridyl group showing 6 hydrogen bonds with the active site

Table 1 .
Predicted activity from regression analysis Plot of docking scores Vs Activity (a)