Design and Synthesis of New Modified Flexible Purine Bases as Potential Inhibitors of Human PNP

The great interest in studying the structure of human purine nucleoside phosphorylase (hPNP) and the continued search for effective inhibitors is due to the importance of the enzyme as a target in the therapy of T-cell proliferative diseases. In addition, hPNP inhibitors are used in organ transplant surgeries to provide immunodeficiency during and after the procedure. Previously, we showed that members of the well-known fleximer class of nucleosides are substrates of E. coli PNP. Fleximers have great promise as they have exhibited significant biological activity against a number of viruses of pandemic concern. Herein, we describe the synthesis and inhibition studies of a series of new fleximer compounds against hPNP and discuss their possible binding mode with the enzyme. At a concentration of 2 mM for the flex-7-deazapurines 1–4, a decrease in enzymatic activity by more than 50% was observed. 4-Amino-5-(1H-pyrrol-3-yl)pyridine 2 was the best inhibitor, with a Ki = 0.70 mM. Docking experiments have shown that ligand 2 is localized in the selected binding pocket Glu201, Asn243 and Phe200. The ability of the pyridine and pyrrole fragments to undergo rotation around the C–C bond allows for multiple binding modes in the active site of hPNP, which could provide several plausible bioactive conformations.


Introduction
Human purine nucleoside phosphorylase (hPNP) belongs to a family of enzymes involved in the purine salvage pathway of nucleoside biosynthesis, which promotes the utilization of purine bases [1]. Its primary role is to catalyze the cleavage of inosine, 2'-deoxyinosine, guanosine and 2'-deoxyguanosine (dG) to the corresponding base and sugar-1-phosphate via reversible phosphorolysis. PNP deficiency in humans leads to a decrease in the number of T cells and immunodeficiencies. Therefore, hPNP is considered an attractive target in the therapy of T-cell proliferative diseases, primarily T-cell leukemias and lymphomas, as well as psoriasis, multiple sclerosis, rheumatoid arthritis, etc. [2,3]. In addition, hPNP inhibitors are used in organ transplant surgeries to provide an immunodeficiency status during and after the procedure. In some cases, hPNP inhibitors are also able to enhance the activity of antiviral and antitumor nucleoside drugs [1,4] by inhibiting the premature metabolism of the nucleoside drug. All of these reasons explain the great interest in studying the structure of human purine nucleoside phosphorylase and the continued search for more effective inhibitors.
Most inhibitors of purine nucleoside phosphorylases are structural analogues of nucleoside substrates, modified in the nucleobase and/or the carbohydrate moiety. Using rational design, Schramm et. al. [5] discovered analogues of C-nucleosides-known as immucillinswhich proved to be inhibitors for the transition state of PNP [6]. Several immucillins have

Inhibition Studies
Compounds 1-4 and 13-16 were then studied as potential inhibitors of hPNP (Figure 1). At a concentration of 2 mM for flexible analogues of 7-deazapurines 1-4, a decrease in enzymatic activity by more than 50% was observed. Flex-acyclovir analogues 13-16 did not appear to be inhibitors, although at a concentration of 0.5 mM, acyclovir slightly reduced the rate of inosine phosphorolysis, as reported by Rabuffetti et al. [1].
The inhibition constants of compounds 1-4 are shown in Table 1, which reveal that the most pronounced inhibition occurred in the presence of compound 2. The flexhypoxanthine analogue 1 showed better binding to the enzyme active site than the flexadenine analogue 4. The inhibition constant for 7-deazaguanine (7DG) is 0.2 mM, which is lower than the constants for the investigated compounds [18].
Non-competitive inhibition was observed in all cases ( Figure S1, see Supplementary Materials). This may be caused by the formation of a dead-end complex upon the binding of the heterocyclic base and an inorganic phosphate at the active site of the human enzyme. This was seen for various heterocyclic bases, including 7-deazahypoxanthine, in the case of E. coli PNP [10]. not appear to be inhibitors, although at a concentration of 0.5 mM, acyclovir slightly reduced the rate of inosine phosphorolysis, as reported by Rabuffetti et al. [1]. The inhibition constants of compounds 1-4 are shown in Table 1, which reveal that the most pronounced inhibition occurred in the presence of compound 2. The flex-hypoxanthine analogue 1 showed better binding to the enzyme active site than the flex-adenine analogue 4. The inhibition constant for 7-deazaguanine (7DG) is 0.2 mM, which is lower than the constants for the investigated compounds [18]. Non-competitive inhibition was observed in all cases ( Figure S1, see Supplementary Materials). This may be caused by the formation of a dead-end complex upon the binding of the heterocyclic base and an inorganic phosphate at the active site of the human enzyme. This was seen for various heterocyclic bases, including 7-deazahypoxanthine, in the case of E. coli PNP [10].

Molecular Modeling
Compound 2 was chosen for computational studies. Before docking, four crystallographic structures of human purine nucleoside phosphorylases from the Protein Database were analyzed. This was a necessary step to determine the most frequent atomic interactions (hydrogen bonds, π-π stacking, etc.) of the PNP active site with various ligands [19]. The following complexes were chosen: with the natural substrate inosine [20] and with inhibitors, including Immucillin H, 7-deazaguanine (7DG) [18] and acyclovir [16] ( Figure S2, see Supplementary Materials). Using the ProteinsPlus web service

Molecular Modeling
Compound 2 was chosen for computational studies. Before docking, four crystallographic structures of human purine nucleoside phosphorylases from the Protein Database were analyzed. This was a necessary step to determine the most frequent atomic interactions (hydrogen bonds, π-π stacking, etc.) of the PNP active site with various ligands [19]. The following complexes were chosen: with the natural substrate inosine [20] and with inhibitors, including Immucillin H, 7-deazaguanine (7DG) [18] and acyclovir [16] ( Figure S2, see Supplementary Materials). Using the ProteinsPlus web service (https://proteins.plus accessed on 15 December 2022) and a special PoseView tool, two-dimensional diagrams of the selected complexes were created [21,22].
Note that, in all cases, the same patterns were responsible for the binding of the purine heterocyclic base and its orientation at the site of enzyme binding. The following amino acid residues of the protein side chains were identified: hydrogen bonding with Glu201 and Asn243, and π-π stacking with Phe200. It is, therefore, assumed that the inhibition of hPNP by flex-base 2 occurs due to interaction with these amino acid residues ( Figure 2).
Structures of flex-base 2 were generated and structural optimizations were performed by using HyperChem software [23]. Quantum chemical analysis of 2 shows that the pyrrole and pyridine fragments are located in different planes (are non-coplanar).
Docking of 2 was then performed in a limited area in a volume of 12 Å × 10 Å × 10 Å, which is sufficient to accommodate the ligand inside the receptor. After clustering, the resulting ligand-protein complexes were ranked according to the established binding energies. The states with the lowest energy and specific conservative ligand-protein interactions were selected as potentially possible. Interactions of protein-ligands were measured using the BIOVIA Discovery Studio Visualizer 2021. rine heterocyclic base and its orientation at the site of enzyme binding. The follow amino acid residues of the protein side chains were identified: hydrogen bonding Glu201 and Asn243, and π-π stacking with Phe200. It is, therefore, assumed that th hibition of hPNP by flex-base 2 occurs due to interaction with these amino acid resi ( Figure 2).   Table 2). Docking experiments showed that 2 is localized in the selected binding pocket. The free energy of binding (∆G) is listed in Table 2.
Three models were chosen from various positions of the fleximer base 2 in the hPNP active site. In the selected models, a similar interaction pattern is observed in the binding site: the base is positioned in the same plane as 7DG, one of the aromatic rings is oriented perpendicular to the Phe200 side residue and the amino group (at the pyridine ring) and the proton at nitrogen (pyrrole ring) participate in the formation of hydrogen bonds with Glu201 and Asn243.
Model A is characterized by the minimum energy ∆G. Model B is similar to the arrangement of 7DG in crystal structure of the hPNP [18]. However, model B is less energetically favorable than models A and C. It is somewhat difficult to assess the exact position of 2 in the active site, which leads to the inhibitory activity. In our opinion, due to the flexibility of flex-base 2, various binding options are possible in the hPNP active site and inhibition of the enzyme is presumably realized in one of the three variants presented.

Molecular Modelling and Docking Studies
Computational docking studies were conducted on protein-ligand docking server SwissDock, based on EADock DSS [24,25]. Structures of fleximer molecules were generated and structural optimizations were performed by using HyperChem software [23]. The ab initio amber FF 6-31G** method; RMS values 0.3 kcal/mol. Analysis of molecular structures and conformational searching from related docking data were performed using the UCSF Chimera program [26].
A crystallographic model of the human purine nucleoside phosphorylase was published and posted on the Protein Data Bank website (PDB ID code 3INY). It is the cryo-EM structure of the hPNP complex bound to its inhibitor 7-deazaguanine [18,27].
To confirm the docking result, a 'pose selection' method was selected to re-dock 7deazaguanine with the active site PNP [28]. Values of the root-mean-square deviation of atomic positions (RMSD), docking pose, accuracy and coverage of contacts were compared with the co-crystallized structure. RMSD of atomic coordinates between two molecules was calculated using the PyMOL Molecular Graphics System, Version 2.5, Schrodinger, LLC. The obtained value has acceptable range of RMSD docking < 1.5 Å [13]. RMSD = 0.112 Å (279 to 279 atoms) (see Supplementary Materials ( Figure S3)).

Conclusions
We synthesized two new groups of compounds, namely fleximer analogues of 7deazapurines and acyclic derivatives of pyrazole-containing bases. The compounds were studied as potential inhibitors of hPNP. At a concentration of 2 mM for the flex-7-deazapurines 1-4, a decrease in enzymatic activity by more than 50% was observed. 4-Amino-5-(1H-pyrrol-3-yl)pyridine (2) was the best inhibitor, with a Ki = 0.70 mM. Flexacyclic analogues 13-16 did not appear to be inhibitors. In order to understand the possible binding mode of the most active compound 2 and the enzyme, computer modeling was performed, which could also aid in additional investigations of hPNP inhibitors. Structures of the flex-base 2 were generated and structural optimizations were performed by using HyperChem soft. Quantum chemical analysis of 2 showed that the pyrrole and pyridine fragments are located in different planes (are non-coplanar). Docking experiments revealed that 2 is localized in the selected binding pocket interacting with Glu201, Asn243 and Phe200. The ability of the pyridine and pyrrole fragments to undergo rotation around the C-C bond allows for multiple binding options in the active site of hPNP, which could provide several plausible bioactive conformations.