Oxazin-5-Ones as a Novel Class of Penicillin Binding Protein Inhibitors: Design, Synthesis and Structure Activity Relationship

Penicillin binding proteins (PBPs) are normal constituents of bacterial which are absent in mammalian cells. The theoretical binding modes of known oxazin-5-ones toward the protein were used as a guide to synthesis new inhibitors. Structural studies of protein-ligand complexes revealed that conformational discrepancies of the derivatives in the protein’s binding site gave rise to the variation in their inhibition constant which ranged from 68.58 μM to 2.04 mM. Biological assay results further confirmed the antibiotic potencies of the studied compounds. Although the outcome of biological screening does not parallel computational predictions, the results obtained from both methods suggest that the oxazin-5-one derivatives are potential PBP inhibitors, hence interesting antibiotic lead agents.

Nowadays, computational methods are routinely employed in drug development processes due to their reliability, time and cost effectiveness [15][16][17]. These methods involve; calculation of pharmacokinetic parameters of chemical compounds using molecular descriptors, pharmacophore screening, docking and binding free energy calculations of a given interaction. Information derived from the binding mode of docked compound has been employed as a guide in structural optimization processes. [18][19].
In the present work, we used results of binding mode predicted from docking calculations of two parent molecules to guide the synthesis of new oxazin-5-ones via palladium catalyzed atom at 6-position with lipophilic and/or an extended hydrophilic moieties could lead to improved potency due to the presence of TRP233 and SER62 residues.

Docking Calculations
To probe the C6 position, compounds 3 to 13 were synthesized (Figs 2 and 3) and docked toward PBP binding site. It was observed that compounds 4, 9, 7 and 12 inhibited the activity of the studied target (K i values ranging from 96.31 to 68.58 μM) ( Table 1) more than compounds 1 and 2, but their relatively poor ligand efficiencies (21-23 kcal/mol per non-H atom) could pose a challenge [23]. Activities were greatly reduced in other derivatives with compound 10 having K i as poor as 2.04 mM.

Assessment of Oral Bioavailability Property
Criteria proposed by Lipinski in his popular "rule of five" (ro5) alongside total polar surface area (TPSA) property were used to assess the oral bioavailability potential of the newly synthesized oxazin-5-ones [24]. Total polar surface area (TPSA) is frequently used in drug design as surrogate property for cell permeability with a rule-of-thumb that a molecule with a TPSA of less than 140 Å 2 would be able to permeate the cell. TPSA has also been used as a surrogate for penetrating the blood-brain-barrier (BBB). Van de et al [25] demonstrated that for a drug molecule to cross the central nervous system, the cut-off for TPSA should be 90 Å 2 . This implies that all compounds can penetrate blood-brain barriers, hence can be used in treating brain cells infections.
According to Lipinski's ro5, derived from 90 th percentile of drug candidates that reached phase II clinical trials, to be drug-like, a drug candidate should have lipophilicity (log P) 5, molecular weight (MW) 500, number of hydrogen bond acceptor (HBA) 10, and number of hydrogen bond donor (HBD) 5. The rule claims that drug candidate which violates more than one property will have bioavailability problem. Table 2 showed that all the compounds are drug-like with respect to ro5. Veber et al [26] observed number of rotatable bond (NRB) experimentally influences bioavailability in rats. Therefore, NRB 10 has been recommended for good oral bioavailability property. Again all the compounds respected NRB criteria for drug-likeness.

Binding Mode Prediction
The docked poses of all the derivatives toward PBP binding site is shown in Part 4a in S1 Fig. It was observed that the compounds adopted varying preferential conformations within the PBP  Their styryl and phenylethynyl moieties were accommodated within the protein groove surrounded by PHE120, VAL302, GLN303, and LEU214 residues. Double bonds are longer than triple bonds. Therefore, compounds 3 and 8 docked deeper into the PBP hydrophobic pocket and hence, made a stronger interactions (546.24 and 893.70 μM respectively) than compounds 5 and 10 (931.82 μM and 2.04 mM respectively). In fact, it appeared the ability to dock deep into PBP binding cavity is a necessary criteria for interaction with the protein because the better inhibitory activities of compounds 6 and 11 than those of compounds 5 and 10 could be attributed to the length of compounds 6 and 11 hexynyl substituent (Part 4d in S1 Fig). In general, the derivatives demonstrated no significant improved affinity for the studied target over the known oxazines (compounds 1 and 2).

Biological Screening
All the derivatives were screened in vitro against selected bacterial following Bauer etal method [27] and results are shown in Table 3. In general, the synthesized derivatives manifested appreciable activity but not in consonant with the docking calculation results. Perhaps PBP was not the drug target inhibited by the derivatives in the whole cell assays and hence the variation in their results. Apart from compounds 3 and 4, the synthesized derivatives exhibited activity against both Gram-positive and Gram-negative bacteria. With exception of compounds 7, 12 and 13 the rest compounds appeared to be generally more active for Gram-positive than Gram-negative bacteria. The compounds (8-13) derived from 6-chloro-5H-naphtho[2,1-b]pyrido[3,2-e] [1,4]oxazin-5-one whose structure contain nitrogen hetro-atom in position-10 of the molecule exhibited enhanced activity compared to those derived from 6-chloro-5H-benzo[a]phenoxazine substrate which has no N-hetero-atom at position 10. Compounds 12 and 13 particularly seem to have broad activity for Gram-positive and Gram-negative bacteria and this was attributed partly to heterocyclic thiophenyl and furanyl moieties contained in the molecules. In addition, the MICs of the compounds were higher than the reference drugs. However, the MIC of compounds 5 and 10 are close to that of tetracycline for B. cereus and S. aureus bacteria respectively.
It can be observed in this study that the results of biological assay and in silico screening do not parallel. This is often the case when comparing the results of in-silico screening, which focuses on a particular enzyme, with a whole organism in vitro testing. The reason could be that the enzyme used in the in silico study might not be in vitro mechanism of the drug candidate action [28].

Experimental Section General Information
All chemicals were purchased from Aldrich Chemical Company UK and were used without further purification. Otherwise stated all compounds were synthesized and characterized in the School of Chemistry of Cardiff University UK. Melting points was determined with a Fischer-Johns apparatus. 1 H and 13 C NMR data were recorded with Brucker DPX 400 MHz spectrometers relative to TMS as internal standard. All and chemical shifts reported in ppm (δ) and coupling constants (J), reported in Hz. Multiplicity is indicated using the following abbreviations: br, for broad; s, for singlet; d, for doublet; t, for triplet; dd, for doublet of doublets and; m, for Table 3

Molecular Modeling
The x-ray crystal structure of PBP (DDTP) with its co-crystallized inhibitor was retrieved from protein data bank (PDB code 1CEF) [20]. Molecular operating environment (MOE) was used to treat the complex dimers as described in our earlier work [29] and to generate the three dimensional structures of the benzophenoxazines. To an oven dried 10 mL RB flask containing 2 mL of CH 3 CN and 1 mL of water was added RX (1 mmol), RB(OH) 2 (1.2 mmol), K 3 PO 4 (588mg, 3 mmol) and the reaction mixture gradually warm to 40°C while stirring under nitrogen atmosphere. Pd(OAc) 2 (8.92mg, 4 mol%), X-Phos (32.5mg, 7 mol%) were added and reaction vessel cork with rubber septum. The entire reaction mixture was heated at 80°C within 5-8 h, and then cooled to room temperature. Solvent evaporated in vacuum and crude product extracted from water with DCM (10 mL x 4). The combined organic extracts were dried with MgSO 4 and concentrated in vacuum. Crude product was purified by flash column chromatography on silica gel.
General Procedure II (Sonogashira Cross-Coupling reactions). Acetonitrile (3 mL) was degassed for 0.5 h before injection into an oven-dried 10 mL RB flask fitted with a rubber septum already charged with Pd(OAc) 2 (8.9 mg, 4 mol%), X-Phos (32.5mg, 7 mol%), RX (1mmol) and K 3 PO 4 (588mg, 3mmol), under an atmosphere of nitrogen. The reaction mixture was stirred and warmed to 50°C during which time 1-alkyne (1.5 mmol) was gradually injected via syringe. The reaction temperature was maintained for 0.5 h before being increased to 80°C. Stirring was continued for 5-8 h then the mixture was cooled to room temperature after reaction completion as monitored by TLC. Water (10 mL) was added mand product extracted with dichloromethane (4 x 10 mL). The combined organic extracts were dried (MgSO 4 ) and concentrated in vacuum. The crude product was separated by flash chromatography on silica gel using petroleum ether-ethyl acetate mixtures.
General Procedure III (Stille Cross-Coupling Reactions). An oven-dried 10 mL RB flask was charged with Pd(OAc) 2 (8.92 mg, 4 mol%) and X-Phos (32.5 mg, 7 mol%) and covered with rubber septum. The vessel was evacuated and back-filled with N 2 thrice before injecting of CH 3 CN (2mL) and H 2 O (1 mL) (both solvents degassed for 30 min) and the reaction mixture warmed to 50°C within 10 min. Rubber septum quickly removed to add chlorophenothiazine (1mmol) and K 3 PO 4 (318 mg, 1.5 mmol), and replaced before injecting tributylthienylstannane or tributylfuranylstannane (1.2 mmol). The temperature was gradually increase to and maintained 80°C. The reaction was terminated in 5 h and the crude product extracted from water (10 mL) four times with DCM. The combined organic extract was dried with MgSO 4 and concentrated in vacuum. The crude product was purified by flash chromatography on silica gel using petroleum ether-ethyl acetate eluent.

General Antimicrobial Sensitivity Testing of Compounds
A pure culture of human pathogenic microbes was obtained from culture collection center, Bishop Shahanan Hospital, Nsukka, Enugu State. The agar cup diffusion method was applied to determine the sensitivity of compounds against bacteria using Muller Hinton Agar. The MHA plates were inoculated with 1 x 10 4 CFU culture of test organism. After which cups were made in each sector after previously dividing the plate into six segments and labeled. Using the sterile pipette, each cup was filled with four drops of compound (0.1 mg/ml). Pre-diffused time of 30 min was allowed before all the plates were incubated at 37°C for 24 h for bacteria. After incubation the inhibition zone diameter (IZD) resulting were measured and result recorded after subtracting the diameter of the cork borer. The cork borer used to make the cup is 8 mm in diameter. The procedure was repeated for tetracycline (standard bacteria) and DMSO (solvent).

Minimum Inhibitory Concentration (MIC) Testing
The method used to determine the MIC was the same as for general sensitivity testing except serial dilution of 0.1 mg/ml DMSO solution of each sample was carried out to have 0.05, 0.025, 0.0125, 0.00625 mg/mL solutions. Fours drops of each dilution were added to the corresponding cup previously cut in the Mueller Hinton Agar (MHA) plate. The plates were incubated at 37°C for 24 h for bacteria and 48 h for fungi. The diameter of zone of inhibition was measured and the value subtracted from the diameter of the borer to give the inhibition zone diameter (IZD). The graph of IZD 2 against the log of concentrations was plotted for each plate containing a specific compound and a microorganism. The anti-log of the intercept on x-axis gives the MIC. The procedure was repeated for tetracycline.

Conclusion
The ease of accessing PBPs from periplasm and its absence in mammalian cells make them target of choice in search for antibiotics. The history of oxazines as having chemotherapeutic potential informed its usage as a parent molecule in the current study. The binding modes of two known oxazin-5-ones were used to guide the synthesis of derivatives. Evaluation of their SAR revealed that the analogues adopted a unique preferential configuration within the binding site cavity of the protein different from that of their parent molecules. This may account for the variation observed in their degree of PBP inhibition. Four of the analogues exhibited improved potencies over the parent molecules and were also drug-like according to Lipinski's ro5. The biological assay results confirmed the antibiotic potencies of the derivatives, but were not in tandem with the computational predictions. Medicinal Chemists could take advantage of the ligand interaction motifs identified in this study in rational optimization by chemical modification of the compounds.