Dihydrofolate reductase inhibitors among pteridine and furo[3,2- g ] pteridine derivatives

Aim. To the purposeful search for the DHFR-inhibitors among substituted pteridine-2,4,7-triones and 7-aryl-(hetaryl-)furo[3,2- g ]pteridine-2,4(1 H ,3 H )-diones for further biological re-search. Methods. In vitro methods, molecular docking, SAR-analysis, statistical methods. Results. The DHFR-inhibitory activity of substituted 1-methylpteridine-2,4,7-triones ( 2 , 3 , 4 ) and 7-aryl-(hetaryl-)furo[3,2- g ]pteridine-2,4(1 H ,3 H )-diones ( 5 , 6 ) was studied. It was estab-lished that 6-(2-hydroxy-2-aryl-(hetaryl-)ethyl)-1-methylpteridine-2,4,7(1 H ,3 H ,8 H )-triones ( 3 ) and butyl 2-(7-aryl- (hetaryl-)-1-methyl-2,4-dioxo-1,4-dihydrofuro[3,2- g ]pteridine-3(2 H )-yl)acetates ( 6 ) inhibited DHFR by 14.59–52.11 %, and were less active comparing to methotrexate. It was found that the introduction of aryl moiety with electron-accepting group, naphthyl substituent or electron-accepting heterocycle (furan, thiophene and benzofuran) caused an increase in the DHFR-inhibitory activity. Additionally, it was shown, that annulation of the furan cycle to the pteridine system was reasonable in the scope of new DHFR-inhibitors synthesis. Thereby it may be concluded that the calculated values of affinity are not reliable predictors for the DHFR-inhibiting activity of studied compound. However, the molecular docking study may be used for evaluation of the interactions between the studied inhibitor and active center of DHFR. Conclusions. The conducted primary in vitro screening revealed low or moderate DHFR-inhibiting activity of the synthesized compounds. The visualization of molecular docking showed that despite the structural similarity to methotrexate, the obtained compounds form different ligand-enzyme interactions. The calculated values of affinity cannot be used as predictors of DHFR-inhibiting activity because of the absence of correlation between the abovementioned indicators. The obtained compounds may be of interest for further studies aimed at the search for anti-inflammatory, anti-viral, hypoglycemic, hypotensive, anti-ische mic agents due to the expected low-toxicity associated with the slight DHFR-inhibiting activity.

Despite the existence of drugs with high selectivity and affinity to DHFR, they do not always meet the requirements of modern medicine in terms of toxicity and the possibility of resistance appearance [15,16]. The search for new drugs, the inhibitors of the main enzymatic pathways of folate formation, which consists in structural modification («bioisosteric» substitutions) of the pteridine cycle and modification or replacement of p-aminobenzoyl-glutamate fragment by other functional groups remains an urgent problem [10][11][12][13][17][18][19]. Moreover, many drugs ("non-classical" antifolates) do not have a p-aminobenzoyl-glutamate fragment in the molecule and show high activity [10][11][12][13].

Reagents
Dihydrofolate Reductase Assay Kit (Sigma-Aldrich, Catalog Number CS0340, Batch Number 067M4065V) was used for evaluation of the DHFR-inhibitory activity of synthesized compounds. The protein content in supplied dihydrofolate reductase was 0.032 mg/ml and the activity of enzyme was 3.75 U/mgP.

The procedure of estimation of the DHFRinhibitory activity of studied compounds
To the microcentrifuge tube (volume 2 ml) 966 µl of diluted 1:10 assay buffer were ad ded. Then sequentially 13 µl of dihydrofolate reductase and 10 µl of 100 µM solution of studied compound in DMSO were added. The tube was sealed, intensively shacked and the formed mixture was transferred to the 1.4 ml quartz cuvette. To the formed mixture 6 µl of 10 mM solution of the NADPH were added, cuvette was sealed by parafilm and shacked. 5 µl of 10 mM of dihydrofolic acid solution were added the mixture, the cuvette was sealed by parafilm, shacked, and immediately transferred to spectrophotometer ULab 108 UV. The absorption of sample at 340 nm was measured each 15 seconds during 150 seconds.
The activity of enzyme was calculated according to the following formula: where: ∆OD/min blank = ∆OD/min for the blank, from the spectrophotometer readings; ∆OD/min sample = ∆OD/min. for the reaction, from the spectrophotometer readings; 12.3 = extinction coefficient (ε, mM -1 *cm -1 ) for the DHFR reaction at 340 nm; 0.013 = enzyme volume in ml (the volume of enzyme used in the assay); 0.032 = enzyme concentration of the original sample. The value of DHFR-inhibitory activity in % was calculated according to the formula: Methotrexate was used as a reference compound.

Molecular docking
The кesearch was conducted by flexible molecular docking, as an approach of finding the molecules with affinity to a specific biological target. Macromolecule from Protein Data Bank (PDB) was used as biological targets, namely DHFR (PDB ID -1RG7) [Protein Data Bank. http://www.rcsb.org/pdb/home/home.-do. Accessed September 6, 2020]. The choice of biological targets was based on the literature data about novel antifolate drugs [10][11][12][13].
Ligand preparation. Substances were drawn using MarvinSketch 20.20.0 and saved in mol format [MarvinSketch version 20.20.0, Chem-Axon http://www.chemaxon.com]. After that they were optimized by program Chem3D, using molecular mechanical MM2 algorithm and saved as pdb files. Molecular mechanics was used to produce more realistic geometry values for most of organic molecules, owing to the fact of being highly parameterized. Using AutoDockTools-1.5.6, the pdb files were converted into PDBQT, the number of active torsions was set as default [23].
Protein preparation. The PDB files were downloaded from the protein data bank. Discovery

Molecular docking
Molecular docking was used to elucidate the main types of interactions of substituted pteridine-2,4,7-triones (2-4) and 7-arylfuro[3,2-g] pteridine-2,4(1H,3H)-diones (5, 6) with an active center of DHFR as well as a tool for possible future prediction of enzyme-inhibitory activity of these compounds. The results of molecular docking showed, that in most of cases the tested compounds revealed higher calculated affinity comparing to methotrexate ( Table 1).
Visualization of the X-ray diffraction study by Discovery Studio showed [Protein Data Bank. http://www.rcsb.org/pdb/home/home.  36 Å). Additionally, the methotrexate molecule formed hydrogen bonds with water in the active center of the enzyme (Fig. 2).
Visualization of the interaction of compound 3.8, 5.1 and 6.6 with DHFR, showed that these structures were characterized by other types of interactions with amino acids residues in the active site (Fig. 2). Thus, 6-(2-hydroxy-2-(4-chlorophenyl)ethyl)-1-methylpteridine-2,4,7(1H,3H,8H)-trione (3.8) had two conventional hydrogen bonds between Oxygen atoms of 2 nd position of the pteridine ring with A:ALA7 (3.40Å), A:TYR100 (2.79Å) and carbon hydrogen bond with A:ILE94 (3.24Å) and A:ALA6 (3.28Å). Whereas, 1-methyl-7-phenylfuro[3,2-g]pteridine-2,4(1H,3H)-dione (5.1) unlike compound 3.8 had a methotrxate-like location in the active center of the enzyme and provided three conventional hydrogen bonds of the Oxygen Note: -* Methotrexate atom at the 2 nd and 4 th positions of the pteridine ring with A:ALA7 (3.15Å), A:TYR100 (2.70Å) and A:ASN18 (3.09Å) and carbon hydrogen bond with A:ALA6 (3.18Å) (Fig. 2, B). Visualization of the interaction between compound 6.6 and DHFR active center (Fig. 2, C) revealed that this structure, like the previous ones (3.8 and 5.1), did not have a significant number of interactions similar to methotrexate -enzyme interactions. Visualization was characterized by conventional hydrogen bonds between the Oxygen atom of the butoxycarbonylmethyl group at the 3 rd position of the pteridine ring with A:ASN18 (3.13Å) and the carbon hydrogen bond between the Oxygen atoms of the 2 nd position of the pteridine cycle with A:SER49 (3.53Å) and A:ASN23 (3.58Å). Other enzyme binding sites of compounds 3.8, 5.1 and 6.6 were rather weak and were associated with the presence of Van der Waals, Pi-sigma, Pi-Pistaked, Pi-alkyl and alkyl interactions. The analysis of correlation between DHFRinhibiting activity of synthesized compounds and their calculated affinity to DHFR showed the absence of direct dependency of abovementioned values. Thus, among compounds 2.5, 5.1, 5.2 and 5.3 that according to the docking results have affinity values ≥ 10.0 the compounds 2.5 and 5.2 reveal low DHFR-inhibiting activity. Noteworthy, despite the highest enzyme-inhibiting activity, the calculated affinity of reference compound methotrexate to DHFR has value 8.7 kcal/mol, which comparable or lower than the affinity of most of studied compounds. Thereby it may be concluded that for the studied compound the calculated values of affinity are not reliable predictors for DHFR-inhibiting activity. However, the molecular docking study may be used for evaluation of interactions between studied inhibitor and active center of DHFR. Abovementioned information is valuable for the search for promising DHFR-inhibitors among various pteridine derivatives.

Conclusion
The conducted primary in vitro screening revealed low or moderate DHFR-inhibiting ac-tivity of synthesized compounds. The visualization of molecular docking showed that despite the structural similarity to methotrexate, the obtained compounds form different ligandenzyme interactions. The calculated values of affinity cannot be used as the predictors of DHFR-inhibiting activity because of the absence of correlation between abovementioned indicators. The obtained compounds may be of interest for further search for anti-inflammatory, anti-viral, hypoglycemic, hypotensive, anti-ischemic agents due to the expected lowtoxicity associated with the slight DHFRinhibiting activity.