Crystallography, Molecular Modeling, and COX-2 Inhibition Studies on Indolizine Derivatives

The cyclooxygenase-2 (COX-2) enzyme is an important target for drug discovery and development of novel anti-inflammatory agents. Selective COX-2 inhibitors have the advantage of reduced side-effects, which result from COX-1 inhibition that is usually observed with nonselective COX inhibitors. In this study, the design and synthesis of a new series of 7-methoxy indolizines as bioisostere indomethacin analogues (5a–e) were carried out and evaluated for COX-2 enzyme inhibition. All the compounds showed activity in micromolar ranges, and the compound diethyl 3-(4-cyanobenzoyl)-7-methoxyindolizine-1,2-dicarboxylate (5a) emerged as a promising COX-2 inhibitor with an IC50 of 5.84 µM, as compared to indomethacin (IC50 = 6.84 µM). The molecular modeling study of indolizines indicated that hydrophobic interactions were the major contribution to COX-2 inhibition. The title compound diethyl 3-(4-bromobenzoyl)-7-methoxyindolizine-1,2-dicarboxylate (5c) was subjected for single-crystal X-ray studies, Hirshfeld surface analysis, and energy framework calculations. The X-ray diffraction analysis showed that the molecule (5c) crystallizes in the monoclinic crystal system with space group P 21/n with a = 12.0497(6)Å, b = 17.8324(10)Å, c = 19.6052(11)Å, α = 90.000°, β = 100.372(1)°, γ = 90.000°, and V = 4143.8(4)Å3. In addition, with the help of Crystal Explorer software program using the B3LYP/6-31G(d, p) basis set, the theoretical calculation of the interaction and graphical representation of energy value was measured in the form of the energy framework in terms of coulombic, dispersion, and total energy.


Chemistry
The design of the target compounds (5a-e) was mainly based on the close chemical structural relationship with the commercially available NSAIDs indomethacin ( Figure 1). The synthesis of the target compounds is depicted in Scheme 1. The intermediates (3a-e) were obtained by stirring a mixture of 4-methoxy pyridine and para-and meta-substituted phenacyl bromides in acetone medium at 5 h and were then further reacted with diethyl but-2-ynedioate in the existence of potassium carbonate in dimethylformamide solvent medium for 30 min. The resulting title compounds were purified by column chromatography using mixture of ethyl acetate and hexane as an eluent, and the purity of the compounds was more than 99% with a satisfactory yield (69% to 77%). The physicochemical property of the target compound ethyl 3-(4-bromobenzoyl)-2-ethyl-7-methoxyindolizine-1-carboxylate (5e) is presented in Table 1. The chemical structure of this compound (5e) was confirmed with spectroscopic techniques such as FT-IR, 1 H-NMR and 13 C-NMR, and LC-MS. The Fourier-transform infrared (FT-IR) spectroscopy revealed benzoyl and ester carbonyl groups at 1699 and 1668 cm −1 , respectively (spectra are available as Electronic Supplementary Materials). The 1 H-NMR spectra revealed the methoxy group at 3.86 ppm, ester peak triplet appearance of 1.34 ppm, and ester group quartet appearance of 4.30 ppm. The 13 C-NMR spectra revealed the appearance of benzoyl and ester carbonyl groups at 186.00 and 166.07 ppm, respectively. The molecular ion peaks of this compound (5e) were in good agreement with its molecular mass. Title compounds diethyl 3-(4-cyanobenzoyl)-7-methoxyindolizine-1,2-dicarboxylate (5a), diethyl 3-(4-fluorobenzoyl)-7-methoxyindolizine-1,2-dicarboxylate (5b), diethyl 3-(4-bromobenzoyl)-7-methoxyindolizine-1,2-

Hirshfeld Surface Analysis
The Hirshfeld surfaces of the crystal structure 5c were investigated to illustrate the nature of intermolecular interactions and visualization of intermolecular close contacts in its crystal structure using Crystal Explorer 17.5 [60], which mapped over de, dnorm, shape index, and curvedness, as shown in Figure 6. The contribution of individual intermolecular interactions on the Hirshfeld surface can be defined by color codes. On the dnorm surface, the red color shows the shorter molecular contacts and the blue color on the dnorm surface area represents the longer molecular contacts. The white color on the dnorm surface indicates the contact around the van der Waals radii. In the dnorm surfaces, the red color shows the hydrogen bonding H···O contacts, whereas the blue surface area represents the H···H contacts (Figure 6a). The de surface features appear as a relatively flat green region where the contact distances are similar (Figure 6b). The adjacent highlighted red and yellow regions on the shape index surface also show the strong hydrogen bonding interactions present in the molecule (Figure 6c), whereas the blue curved and yellow regions on the curvedness surfaces shows the H···H interactions (Figure 6d). The 2D fingerprint plots [61] show the sharp spike, which represents the intermolecular interactions present in the molecule (Figure 7a). Again, it shows that C-H···O interactions were predominant (23.2%) after the H···H contacts, which led to the highest contribution of 35.8% in comparison to other interactions, suggesting that weak C-H···O hydrogen bonding plays an essential role in its crystal packing. The percentage contributions of other intermolecular interactions in this crystal structure were as follows: C···H/H···C (18.4%), H···Br/Br···H (12.2%), C···O/O···C (2.5%), C···C (2.1%), N···H/H···N (1.7%), etc. (Figure 7b).

Hirshfeld Surface Analysis
The Hirshfeld surfaces of the crystal structure 5c were investigated to illustrate the nature of intermolecular interactions and visualization of intermolecular close contacts in its crystal structure using Crystal Explorer 17.5 [60], which mapped over d e , d norm , shape index, and curvedness, as shown in Figure 6. The contribution of individual intermolecular interactions on the Hirshfeld surface can be defined by color codes. On the d norm surface, the red color shows the shorter molecular contacts and the blue color on the d norm surface area represents the longer molecular contacts. The white color on the d norm surface indicates the contact around the van der Waals radii. In the d norm surfaces, the red color shows the hydrogen bonding H···O contacts, whereas the blue surface area represents the H···H contacts ( Figure 6a). The d e surface features appear as a relatively flat green region where the contact distances are similar (Figure 6b). The adjacent highlighted red and yellow regions on the shape index surface also show the strong hydrogen bonding interactions present in the molecule (Figure 6c), whereas the blue curved and yellow regions on the curvedness surfaces shows the H···H interactions (Figure 6d). The 2D fingerprint plots [61] show the sharp spike, which represents the intermolecular interactions present in the molecule (Figure 7a). Again, it shows that C-H···O interactions were predominant (23.2%) after the H···H contacts, which led to the highest contribution of 35.8% in comparison to other interactions, suggesting that weak C-H···O hydrogen bonding plays an essential role in its crystal packing. The percentage contributions of other intermolecular interactions in this crystal structure were as follows: C···H/H···C (18.4%), H···Br/Br···H (12.2%), C···O/O···C (2.5%), C···C (2.1%), N···H/H···N (1.7%), etc. (Figure 7b).

Energy Framework Calculation
Furthermore, the Crystal Explorer 17.5 software was used to evaluate the interaction energies for the crystal structure 5c. Energy frameworks have a strong and remarkable way of imagining the supramolecular existence of molecular crystal structures. The interaction energies between the molecules are obtained using monomer wave functions at the B3LYP/6-31G (d, p) level. [62]. As prescribed, the tube size used in all the energy frameworks was 80 (scale factor), and the cutoff for the energy threshold value was set to zero. In the 3D topological images, the diameter of the tube cylinder reflects the interaction energy in the molecular packing for the corresponding interaction. The molecules present within the 3.8 Å circle within 1 × 1 × 1 unit cell dimensions were selected for this calculation (Figure 8a). Energies between molecular pairs are expressed as cylinders that connect molecular pair centroids with a cylindrical radius proportional to the energy interaction magnitude. The energy framework was outlined as red cylinders for Eelec, green cylinders for Edis, and

Energy Framework Calculation
Furthermore, the Crystal Explorer 17.5 software was used to evaluate the interaction energies for the crystal structure 5c. Energy frameworks have a strong and remarkable way of imagining the supramolecular existence of molecular crystal structures. The interaction energies between the molecules are obtained using monomer wave functions at the B3LYP/6-31G (d, p) level. [62]. As prescribed, the tube size used in all the energy frameworks was 80 (scale factor), and the cutoff for the energy threshold value was set to zero. In the 3D topological images, the diameter of the tube cylinder reflects the interaction energy in the molecular packing for the corresponding interaction. The molecules present within the 3.8 Å circle within 1 × 1 × 1 unit cell dimensions were selected for this calculation (Figure 8a).

Energy Framework Calculation
Furthermore, the Crystal Explorer 17.5 software was used to evaluate the interaction energies for the crystal structure 5c. Energy frameworks have a strong and remarkable way of imagining the supramolecular existence of molecular crystal structures. The interaction energies between the molecules are obtained using monomer wave functions at the B3LYP/6-31G (d, p) level. [62]. As prescribed, the tube size used in all the energy frameworks was 80 (scale factor), and the cutoff for the energy threshold value was set to zero. In the 3D topological images, the diameter of the tube cylinder reflects the interaction energy in the molecular packing for the corresponding interaction. The molecules present within the 3.8 Å circle within 1 × 1 × 1 unit cell dimensions were selected for this calculation (Figure 8a). Energies between molecular pairs are expressed as cylinders that connect molecular pair centroids with a cylindrical radius proportional to the energy interaction magnitude. The energy framework was outlined as red cylinders for Eelec, green cylinders for Edis, and Energies between molecular pairs are expressed as cylinders that connect molecular pair centroids with a cylindrical radius proportional to the energy interaction magnitude.
The energy framework was outlined as red cylinders for E elec , green cylinders for E dis , and blue cylinders for E tot , as shown in (Figure 8b-d), and the relative strength of molecular packing was expressed in various directions by these tubes. The supramolecular nature of the crystal structure was, thus, visualized by energy structures in a special way. The calculated energy values are listed in Table 4 for electrostatic, polarization, dispersion, and total interaction energy, which suggests that 5c crystal structure preferred dispersion energy over others.

Pharmacology
The COX-2-inhibitory activity of the target compounds (5a-5e) is presented in Table 5. As can be seen from Table 5, all the indolizines displayed interesting inhibitory activity against COX-2 similar to the commercially available drug indomethacin. Compound 5a with a 4-cyanobenzoyl group attached to the 3-position of the indolizine scaffold, having two ethyl carboxylate groups attached to the 1-and 2-position of the scaffold, emerged as the most promising compound with the highest COX-2-inhibitory activity (IC 50 = 5.84 µM). Replacement of the electron-withdrawing nitrile group with halogens such as fluorine and bromine atoms at the 4-position of the benzoyl ring exhibited detrimental COX-2 inhibitory activity for compounds 5b and 5c with IC 50 values 6.73 µM and 6.99 µM, respectively. It is remarkable to note that title compound 5e with only one ethyl ester moiety at the first position of the indolizine pharmacophore exhibited a further reduction in activity (IC 50 = 7.38 µM) as compared to the structurally similar compound 5c (IC 50 = 6.99 µM).
The compound 5d, substituted with a methoxy functional group at the 3-position of the benzoyl ring, displayed the least inhibitory activity (IC 50 = 8.49 µM). In general, the availability of the electron-withdrawing functional groups at the para position of the benzoyl ring was found to be favorable for COX-2-inhibitory activity as compared to the presence of the electron-donating functional groups at the meta position. Previously these derivatives demonstrated excellent safety profiles [35]; thus, they could be considered as lead molecules for further improvement of novel potential COX-2 inhibitors.

Compound
Compound Structure IC 50 * (µM) 5a Molecules 2021, 26, x FOR PEER REVIEW 11 of 20      The compound 5d, substituted with a methoxy functional group at the 3-position of the benzoyl ring, displayed the least inhibitory activity (IC50 = 8.49 µM). In general, the availability of the electron-withdrawing functional groups at the para position of the benzoyl ring was found to be favorable for COX-2-inhibitory activity as compared to the presence of the electron-donating functional groups at the meta position. Previously these derivatives demonstrated excellent safety profiles [35]; thus, they could be considered as lead molecules for further improvement of novel potential COX-2 inhibitors.

Computational Studies
Molecular docking studies are considered an invaluable in silico approach to correlate the in vitro structure-activity relationship (SAR) of chemical compounds [63]. To gain insight into the inhibitory activity of indolizines (5a-e), we investigated their key interactions with the COX-2 receptor through a computational approach. The docking study was conducted with Accelrys Discovery Studio Client 4.0 software. The docking interaction energies and the residue interactions of indolizines 5a-e and indomethacin are reported in Table 6. All the compounds demonstrated favorable docking energy ranging from −38.22 to −53.29 kcal/mol, indicating that they have a good binding affinity with the COX-2 receptor, as demonstrated from their biological activities.  50 value is defined as the concentration of test and standard substances required to produce 50% inhibition of human recombinant COX-2 enzyme by means of three determinations using the enzyme-linked immune sorbent assay kit. a-d Title compounds not sharing a letter vary significantly (p < 0.05).

Computational Studies
Molecular docking studies are considered an invaluable in silico approach to correlate the in vitro structure-activity relationship (SAR) of chemical compounds [63]. To gain insight into the inhibitory activity of indolizines (5a-e), we investigated their key interactions with the COX-2 receptor through a computational approach. The docking study was conducted with Accelrys Discovery Studio Client 4.0 software. The docking interaction energies and the residue interactions of indolizines 5a-e and indomethacin are reported in Table 6. All the compounds demonstrated favorable docking energy ranging from −38.22 to −53.29 kcal/mol, indicating that they have a good binding affinity with the COX-2 receptor, as demonstrated from their biological activities.
The predicted docking poses of indolizines 5a-e and indomethacin are depicted in Figure 9. All the compounds adopted a similar conformation to that of indomethacin, where the indolizine ring is taken into a sandwich between the amino-acid residues Ala527, Val523 and Val349, Leu352. The benzoyl ring is oriented toward the residues Tyr385 and Trp387, while the methoxy group is located in the deep region of the receptor. As can be observed from the binding poses, indolizines 5a, 5b, and 5d demonstrated favorable hydrogen bonding interaction between Arg120 with the ester group at the 1-position of the indolizine scaffold, while indomethacin showed hydrogen bonding and ionic bonding interactions between the same residue Arg120 and its carboxylic acid group. This indicated that the ionic interaction with Arg120 is not a requirement for maintaining the potency of the compounds. Furthermore, the indolizines 5c and 5e containing a bromine atom at the para position of the benzoyl ring showed no hydrogen bonding involvement with the residue Arg120. Therefore, the major contribution to the COX-2 activity of our compounds principally involves hydrophobic interactions with the indolizine ring and with the substituents at positions 2 and 3 of indolizine. It can be noted that only the bromine substituent on the benzoyl ring (5c and 5e) demonstrated hydrophobic interactions with residues Leu384 and Met522, while pi-pi interactions were observed for all indolizines with the exception of fluoro indolizine 5b and indomethacin. Therefore, the substituent in the benzoyl ring has a minor contribution to the activity of indolizine. However, it has been demonstrated that the benzoyl ring on indomethacin is important to bioactivity since the replacement of N-benzoyl by N-benzyl led to a reduction in COX-2 inhibition [64]. Table 6. Docking results of indolizines (5a-e) and indomethacin against cyclooxygense-2 (COX-2) receptor (PDB 4COX).
tions with the COX-2 receptor through a computational approach. The docking study was conducted with Accelrys Discovery Studio Client 4.0 software. The docking interaction energies and the residue interactions of indolizines 5a-e and indomethacin are reported in Table 6. All the compounds demonstrated favorable docking energy ranging from −38.22 to −53.29 kcal/mol, indicating that they have a good binding affinity with the COX-2 receptor, as demonstrated from their biological activities. Table 6. Docking results of indolizines (5a-e) and indomethacin against cyclooxygense-2 (COX-2) receptor (PDB 4COX). The predicted docking poses of indolizines 5a-e and indomethacin are depicted in Figure 9. All the compounds adopted a similar conformation to that of indomethacin, where the indolizine ring is taken into a sandwich between the amino-acid residues Ala527, Val523 and Val349, Leu352. The benzoyl ring is oriented toward the residues Tyr385 and Trp387, while the methoxy group is located in the deep region of the receptor. As can be observed from the binding poses, indolizines 5a, 5b, and 5d demonstrated favorable hydrogen bonding interaction between Arg120 with the ester group at the 1-position of the indolizine scaffold, while indomethacin showed hydrogen bonding and ionic bonding interactions between the same residue Arg120 and its carboxylic acid group. This indicated that the ionic interaction with Arg120 is not a requirement for maintaining the potency of the compounds. Furthermore, the indolizines 5c and 5e containing a bromine atom at the para position of the benzoyl ring showed no hydrogen bonding involvement with the residue Arg120. Therefore, the major contribution to the COX-2 activity of our compounds principally involves hydrophobic interactions with the indolizine ring and with the substituents at positions 2 and 3 of indolizine. It can be noted that only the bromine substituent on the benzoyl ring (5c and 5e) demonstrated hydrophobic interactions with residues Leu384 and Met522, while pi-pi interactions were observed for all indolizines with the exception of fluoro indolizine 5b and indomethacin. Therefore, the substituent in the benzoyl ring has a minor contribution to the activity of indolizine. However, it has been demonstrated that the benzoyl ring on indomethacin is important to bioactivity since the replacement of N-benzoyl by N-benzyl led to a reduction in COX-2 inhibition [64].

Indolizine 5a
Indolizine 5b The molecular modeling study provided insight into the structural requirement of indolizines for COX-2-inhibitory activity. Moreover, indolizines 5a-e are more likely to be selective COX-2 inhibitors. It has been validated that COX-1 and COX-2 selectivity is mainly due to the ionic interaction with residue Arg120 since the corresponding ester and amide of indomethacin derivatives presented good selectivity in favor of COX-2 inhibition [64].

Chemistry
All the commercially offered chemicals and solvents were purchased from Sigma-Aldrich Co. (St. Louis, MO, USA). All the chemical reactions were performed in hot-airdried glassware in the presence of a nitrogen atmosphere consuming dry solvents. A Shimadzu FT-IR spectrophotometer (Columbia, MD, USA) was used to record the FT-IR spectra. Furthermore, 1 H-and 13 C-NMR spectra were documented at ambient temperature on Bruker AVANCE III 400 MHz instruments (San Jose, CA, USA) using CDCl3 and DMSO-d6 as solvents. An Agilent 1200 series instrument (Santa Clara, CA, USA) in conjunction with a 6140 single-quadrupole mass spectrometer using positive and negative ESI mode with a mass selective detector (MSD) range of 100-2000, as well as 0.1% aqueous trifluoroacetic acid in an acetonitrile system on the C18-BDS column, was used to record liquid chromatography-mass spectrometry (LC-MS) spectra. Then, an elemental analysis was carried out using the analyzer FLASH EA 1112 CHN (Thermo Finnigan LLC, New The molecular modeling study provided insight into the structural requirement of indolizines for COX-2-inhibitory activity. Moreover, indolizines 5a-e are more likely to be selective COX-2 inhibitors. It has been validated that COX-1 and COX-2 selectivity is mainly due to the ionic interaction with residue Arg120 since the corresponding ester and amide of indomethacin derivatives presented good selectivity in favor of COX-2 inhibition [64].

Chemistry
All the commercially offered chemicals and solvents were purchased from Sigma-Aldrich Co. (St. Louis, MO, USA). All the chemical reactions were performed in hot-airdried glassware in the presence of a nitrogen atmosphere consuming dry solvents. A Shimadzu FT-IR spectrophotometer (Columbia, MD, USA) was used to record the FT-IR spectra. Furthermore, 1 H-and 13 C-NMR spectra were documented at ambient temperature on Bruker AVANCE III 400 MHz instruments (San Jose, CA, USA) using CDCl 3 and DMSO-d 6 as solvents. An Agilent 1200 series instrument (Santa Clara, CA, USA) in conjunction with a 6140 single-quadrupole mass spectrometer using positive and negative ESI mode with a mass selective detector (MSD) range of 100-2000, as well as 0.1% aqueous trifluoroacetic acid in an acetonitrile system on the C18-BDS column, was used to record liquid chromatography-mass spectrometry (LC-MS) spectra. Then, an elemental analysis was carried out using the analyzer FLASH EA 1112 CHN (Thermo Finnigan LLC, New York, NY, USA). A single-crystal X-ray diffraction study was performed using a Bruker KAPPA APEX II DUO diffractometer (Madison, WI, USA) equipped with a charge-coupled device (CCD) detector; monochromated Mo Kα radiation (λ = 0.71073 Å) was used. Data collection was carried out using an Oxford Cryostream cooling system featuring the Bruker Apex II software (Madison, WI, USA) at 173(2) K [16].

Crystallography
Single-crystal X-ray diffraction data of 5c were collected on a Bruker KAPPA APEX II DUO diffractometer using graphite-monochromated Mo-Kα radiation (χ = 0.71073 Å). Data collection was carried out at 173(2) K. Oxford Cryostream was used to control temperature (Oxford Cryostat). The cell refinement and data reduction for 5c were performed using the program SAINT [65], and the absorption correction was performed using SADABS [65].
The crystal structure of 5c was solved by direct methods using SHELXS-18 [66] and refined by the full-matrix least-squares method based on F 2 using SHELXL-2018 [66]. The program WinGx [67] was used to prepare molecular graphic images. All non-hydrogen atoms were refined anisotropically, and all hydrogen atoms were placed in idealized positions and refined in riding models with U iso assigned 1.2 or 1.5 times U eq of the parent atoms [67]. The C-H bond distances was constrained to 0.95 Å for aromatic hydrogen and 1.00 Å for methyl hydrogen. The crystallographic parameters are listed in Table 2.

In Vitro COX-2 Inhibition Assay
The title compounds 5a-e were tested for in vitro human recombinant COX-2 enzyme inhibition activity following our previously reported protocol [68].

Computational Studies
The computational study for test compounds 5a-e was conducted with Accelrys Discovery Studio Client 4.0 (Waltham, MA, USA) using the indomethacin crystal structure PDB 4COX following our previously reported protocol [68].

Conclusions
In this study, a series of diethyl 7-methoxy-3-(3-substituted benzoyl)indolizine-1,2dicarboxylate derivatives (5a-d) and ethyl 3-(4-bromobenzoyl)-2-ethyl-7-methoxyindolizine-1-carboxylate (5e) were synthesized and evaluated for the inhibition of COX-2 enzyme activity. All the compounds were demonstrated to be active inhibitors of COX-2, with the most active compound (5a) having an IC 50 value comparable to that of indomethacin, a marketed COX inhibitor. Computational studies were conducted to analyze the key interactions of these compounds with the amino-acid residues of the COX-2 receptor. Hydrophobic interactions were observed to be mainly responsible for the inhibitory COX-2 activity of indolizines. The compound 5c was crystallized in a monoclinic crystal system with space group P 2 1 /n. The molecule was observed to have both intra-and intermolecular hydrogen bonds and exhibited C-H···π interactions for stability. In order to understand and visualize the contribution of different intermolecular interactions, Hirshfeld surface analysis with 2D fingerprint plots was carried out to provide insight into the stability of the crystal structure. In terms of electrostatic, dispersion, and total energy, the systematic and theoretical energy was calculated using the software program Crystal Explorer, which further provided 3D topological images. Indolizines 5a-e could be considered as lead compounds for developing novel COX-2 inhibitors.