2.1. Cyclooxygenase Inhibition Studies
The effect of CDXs previously obtained in our group [
40,
41] as inhibitors of membrane located enzymes that might be involved in inflammatory processes, namely COX-1 and COX-2, was evaluated. Three enantiomeric pairs of CDXs were chosen to study the inhibitory effect of both enantiomers of each pair face to biological targets in order to evaluate potency and enantioselectivity. The effect on COXs activity was studied by spectrofluorimetry using a commercial kit measuring the peroxidation activity of COXs. The kit includes isoenzyme-specific inhibitors for distinguishing COX-1 from COX-2 activities.
The results, given as percentage (%) of inhibition and expressed as mean ± standard deviation of two independent experiments, are summarized in
Table 1. The overall results indicate that all the CDXs evaluated exhibited COX-1 and COX-2 inhibition potential, although about 20 times less active than indomethacin (positive control). Paired
t-test was also performed to compare inhibitory effects within each enantiomeric pair to verify the occurrence of enzyme-type and enantioselectivity. Among all the compounds, (
R)-(+)-CDX2 was the only compound presenting statistically significant enzyme-type selectivity (
p < 0.05). This enantiomer was more active at inhibiting COX-2 than COX-1. All pairs demonstrated enantioselectivity for COX-1 and |
t|
calculated values were 3.613, 7.249 and 2.891 for CDX1, CDX2 and CDX3 pairs, respectively (
ttab(p = 0.05; d.f. = 10) = 2.228). (
S)-(−)-CDX1, (
S)-(−)-CDX2 and (
S)-(+)-CDX3 were more active than their antipode.
Concerning enantioselectivity for COX-2, the results obtained for the pairs of CDX2 and CDX3 should be highlighted. For instance, the % of COX-2 inhibition for (
S)-(+)-CDX3 was 93.4 ± 11.4, however the inhibitory effect of (
R)-(−)-CDX3 was statistically significantly lower. |
t|
calculated value was 2.891 for COX-2 (
ttab(p = 0.05; d.f. = 10) = 2.228). Similarly, (
S)-(−)-CDX2 was more active than (
R)-(+)-CDX2 at inhibiting COX-2 (|
t|
calculated = 6.777;
ttab(p = 0.05; d.f. = 10) = 2.228). Accordingly, for these two pairs, weak enantioselectivity was observed. It is important to stress that even though enantioselectivity was observed, its extent was not as evident as that obtained for ketoprofen for instance [
10].
Molecular docking studies were also performed in order to predict the potential anti-inflammatory activity and to postulate a hypothetical binding model of the tested compounds. The binding affinity between the target and the small molecule was evaluated by the binding free energy approximation (ΔGb, kcal/mol) using AutoDock Vina. The best scored conformation of each compound predicted by AutoDock Vina was selected and further evaluated. The docking score was used to predict the strength of the non-covalent interactions between two molecules after they have been docked (also referred to as binding energy). The docking score is a mathematical approximation of the binding free energy between the ligand and its target.
Diclofenac, indomethacin, naproxen, and piroxicam [
46] were used as positive controls and showed negative binding energy values (
Table 2). Ligands obtained from the database established more stable complexes with COX-1, with an average binding free energy of −7.8 kcal/mol. Moreover, the docking scores predicted for decoys into the COX-1 was surprisingly low (−7.3 kcal/mol). From the tested compounds, only (
S)- and (
R)-CDX3 and (
S)-CDX2 presented docking scores more negative than the known COX-1 inhibitors indomethacin and piroxicam. However, as only a very small difference was observed between known ligands and decoys scores, this model cannot be used to predict COX-1 inhibition. Hence, more detailed analysis of docking poses and binding mechanisms was performed for the other studied COX isoform: COX-2. Diclofenac, indomethacin, celecoxib, and valdecoxib [
46,
47] were used as positive controls for COX-2 inhibition. The average binding energy predicted for decoys and known ligands into the COX-2 was −7.6 and −9.3 kcal/mol, respectively (
Table 2). Both diclofenac and indomethacin presented −7.9 kcal/mol, whereas celecoxib and valdecoxib exhibited −11.5 and −9.5 kcal/mol, respectively. Among the tested CDXs, (
R)- and (
S)-CDX1 exhibited the highest binding affinities and, therefore, lower binding free energies than negative controls.
Both positive controls and CDXs could dock into the active site of COX-2 successfully (
Figure 2A). The binding mode of celecoxib, valdecoxib, indomethacin, and diclofenac (
Supplementary Data, Figure S1) are in accordance to the previously reported binding modes. This is of particular importance considering docking accuracy. (
S)-CDX1 presented the highest binding energy (−8.0 kcal/mol) into the COX-2 model, similar to the known ligands value. Docking energies of −7.8, −7.5, and −7.0 kcal/mol were obtained to the (
R)-CDX1, (
S)-CDX3, and (
S)-CDX2, respectively. CDX3 enantiomers presented very different poses within the binding site of COX-2, whereas CDX1 enantiomers showed a minor difference (
Figure 2).
CDX1 enantiomers interact through hydrogen bonds with His90, Leu352, Ser353, Tyr355, and Ala527, also involved in the binding of known anti-inflammatory compounds to COX-2 [
48,
49,
50]. CDX1 enantiomers bind similarly to COX-2 binding pocket, with the xanthone scaffold aligned approximately in the same special position, with a slightly different orientation of the aromatic ring and OH group of the chiral moiety (
Figure 2B). (
R)-CDX1 shows an additional hydrogen interaction between OH and Gln-192, similarly to celecoxib and valdecoxib. On the other hand, CDX3 enantiomers bind in very different poses in COX-2 binding pocket, almost perpendicular to each other (
Figure 2C). In fact, (
S)-CDX3 (
Figure 2C, dark blue sticks) binds in a pose similar to CDX1, establishing hydrogen interactions with residues Gln192, His90, Ser353, Leu352, and Tyr355, documented as being important for COX-2 inhibition [
50,
51,
52]. Concerning (
R)-CDX3, the aromatic backbone projects deep COX active site from the hydrophobic channel, with the C3 chain establishing hydrogen interactions with Arg513, Pro86, and Arg120, and the C6-methoxy group establishing hydrogen interactions with Tyr385 (
Figure 2C, light blue sticks), which is important in the catalysis or inactivation of the enzyme [
53]. (
S)-CDX-2 establishes hydrogen interactions with Gln192, Leu352, Ser353, His90, Tyr355, and Arg513; and (
R)-CDX-2 establishes hydrogen interactions with Tyr385, Tyr355, Arg513, Pro86, and Arg-120 (not shown).
2.2. Human Serum Albumin Affinity Studies
HSA-CDX binding parameters are compiled in
Table 3. The binding process has reached a completion state as indicated by the
Ymax that reached about 100%. For all the compounds, HSA binding occurred spontaneously (ΔG values < 0) in a single binding site (
n = 1). All compounds presented dissociation constants (K
d) below 100 µM, which accounts for high affinity binding to HSA [
54]. Paired
t-test was performed to compare K
d obtained for (
S)- and (
R)-enantiomers of each CDX pair. For all the pairs, there was a statistically significant difference between K
d values. Hence, weak enantioselectivity concerning albumin binding was observed. Among all pairs, the highest difference in binding affinity was observed for CDX1 (
ca. 2.6 fold) as demonstrated by the |
t|
calculated value which was 10.103 (
ttab(p = 0.05; d.f. = 4) = 4.303). The (
S)-enantiomer presented higher binding affinity compared to the (
R)-enantiomer. For CDX3 and CDX2, the difference in HSA binding affinity obtained for (
S)- and (
R)-enantiomers was less pronounced compared to CDX1. For CDX2, the (
S)-enantiomer presented slightly higher affinity than the (
R) one; |
t|
calculated value was 5.484 (
ttab(p = 0.05; d.f. = 4) = 2.776). On the contrary, (
R)-(−)-CDX3 has shown slightly higher affinity than (
S)-(+)-CDX3. In this case, |
t|
calculated value was 3.713 (
ttab(p = 0.05; d.f. = 4) = 2.776).
Regarding computational studies, drugs that are described as having high affinity to HSA lead to docking scores between −4.4 kcal/mol (Iophenoxid acid) and −8.5 kcal/mol (warfarin) (
Table 3). CDXs presented scores from −7.0 to −7.3 kcal/mol, and therefore it is hypothesized that they will have high affinity to albumin target. There is an offset between the ∆G binding and the docking scores. This relates to the ability of the docking algorithm to predict the strength of ligand binding to the protein target, and therefore, other scoring functions will be used in the future to increase the accuracy.
(
S)-Ibuprofen binds albumin through hydrogen interactions with Arg-140, Tyr-411, and Lys-414 (
Figure 3A), residues described as being involved in binding of substrates to HSA [
55]. The present study indicates that CDXs fit within the hydrophobic pocket of subdomain IIIA, presenting low negative docking scores. This groove was selected for the docking studies as it was described as being the binding pocket for (
S)-ibuprofen and most ligands [
56]. CDXs bind in a similar position in the binding groove, with the central xanthone ring aligned with ibuprofen ring. The binding of CDX enantiomers in HSA subdomain IIIA present differences concerning the number of hydrogen interactions. For example, the complex (
R)-CDX1-HSA is stabilized by three hydrogen-bond interactions with residues Leu-430, Ser-489, and Asn-391, already described as being involved in the binding of drugs to HSA [
57,
58,
59], whereas (S)-CDX1 lacks those interactions (
Figure 3B). Therefore, there is concordance between in silico and in vitro studies, as enantiosselectivity can be found on the binding of CDX1 to HSA.