In Vitro and In Silico Studies of the Antimicrobial Activity of Prenylated Phenylpropanoids of Green Propolis and Their Derivatives against Oral Bacteria

Artepillin C, drupanin, and plicatin B are prenylated phenylpropanoids that naturally occur in Brazilian green propolis. In this study, these compounds and eleven of their derivatives were synthesized and evaluated for their in vitro antimicrobial activity against a representative panel of oral bacteria in terms of their minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC) values. Plicatin B (2) and its hydrogenated derivative 8 (2′,3′,7,8-tetrahydro-plicatin B) were the most active compounds. Plicatin B (2) displayed strong activity against all the bacteria tested, with an MIC of 31.2 μg/mL against Streptococcus mutans, S. sanguinis, and S. mitis. On the other hand, compound 8 displayed strong activity against S. mutans, S. salivarius, S. sobrinus, Lactobacillus paracasei (MIC = 62.5 μg/mL), and S. mitis (MIC = 31.2 μg/mL), as well as moderate activity against Enterococcus faecalis and S. sanguinis (MIC = 125 μg/mL). Compounds 2 and 8 displayed bactericidal effects (MBC: MIC ≤ 4) against all the tested bacteria. In silico studies showed that the complexes formed by compounds 2 and 8 with the S. mitis, S. sanguinis, and S. mutans targets (3LE0, 4N82, and 3AIC, respectively) had energy score values similar to those of the native S. mitis, S. sanguinis, and S. mutans ligands due to the formation of strong hydrogen bonds. Moreover, all the estimated physicochemical parameters satisfied the drug-likeness criteria without violating the Lipinski, Veber, and Egan rules, so these compounds are not expected to cause problems with oral bioavailability and pharmacokinetics. Compounds 2 and 8 also had suitable ADMET parameters, as the online server pkCSM calculates. These results make compounds 2 and 8 good candidates as antibacterial agents against oral bacteria.


Introduction
Oral pathogens comprise a group of microorganisms that adhere to and colonize the oral cavity, causing periodontal inflammatory diseases, caries, or infective endocarditis [1].Oral bacteria are responsible for dental caries, an infectious disease that affects people worldwide, especially in low-income regions of developing countries [2].These bacteria adhere to the mature acquired salivary pellicle [3,4], which provides specific receptors for bacteria to attach and initiate colonization of the tooth surface [5].The initial colonizers that adhere to acquired salivary pellicles are usually Streptococcus species and Actinomyces species [6,7].Streptococcus mutans, one of the primary bacteria associated with dental caries, can establish a strongly acidic microenvironment (pH below 5.0) due to acidic by-products of the bacterial fermentation of different types of carbohydrates from the diet [8], thereby forming biofilms on the tooth surface (i.e., the dental plaque), which culminates in teeth cavitation (i.e., caries) and the onset of tooth decay [9].However, oral diseases generally have a polymicrobial origin, and together with S. mutans, other acid-resistant bacteria can cause cavities, including S. sobrinus and S. sanguinis [10].
The most effective method of preventing dental biofilm is through mechanical removal by brushing and flossing [11].However, in most cases, this mechanical method is performed improperly, especially in hard-to-reach places, such as sub-gums and fissures, which end up favoring biofilm accumulation.Thus, the use of chemical methods, such as mouthwashes with antimicrobial action, becomes necessary to promote the reduction of the adhesion of bacteria to the dental surface and inhibit cariogenic bacteria growth and proliferation [12,13].Chlorhexidine is the gold standard anti-cariogenic agent used in mouthwash formulations to reduce plaque [13,14].However, despite its effectiveness, chlorhexidine has shown adverse effects, especially in terms of its daily use, in addition to the emergence of resistant strains [15,16].In this scenario, the interest in new compounds with antimicrobial action as effective as chlorhexidine in controlling biofilm but with fewer adverse effects has grown in the last few years [17,18].
Chemically, Brazilian green propolis is rich in prenylated compounds [37,38], mainly phenylpropanoids like drupanin (I), and artepillin C (II, Figure 1).These phenolic acids are commonly detected and identified in Brazilian green propolis [32,[39][40][41].In the literature, many studies have reported the biological and pharmacological activities of artepillin C [32,39,42,43] and drupanin [40,41,44].However, data on the antibacterial activity of these prenylated compounds against oral bacteria are still scarce [45].In this study, we investigated the antibacterial activity of a series of synthetic prenylated phenylpropanoids, including the naturally occurring compounds artepillin C and drupanin and their derivatives, against a representative panel of oral bacteria.In silico studies with the most active compounds aiming to understand the target-compound interactions and to estimate the physicochemical, drug-likeness, and ADMET/pharmacokinetic properties were also performed.

In Vitro Antibacterial Activity of Compounds 1-14
Artepillin C (5), drupanin (6), and compounds 7-14 were obtained by hydrolysis, reduction with LiAlH4, and/or catalytic hydrogenation of prenylation products of methyl pcoumarate 1-4, as shown in Scheme 1. Table 1 summarizes the results of the antimicrobial assays with the fourteen compounds against a representative panel of oral bacteria regarding their minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC).The lowest MIC values (31.2 µg/mL) were observed for compound 2 (plicatin B) against Streptococcus mutans (S. mutans), Streptococcus mitis (S. mitis), and Streptococcus sanguinis (S. sanguinis) and for compound 8 against S. mitis.In this study, we investigated the antibacterial activity of a series of synthetic prenylated phenylpropanoids, including the naturally occurring compounds artepillin C and drupanin and their derivatives, against a representative panel of oral bacteria.In silico studies with the most active compounds aiming to understand the target-compound interactions and to estimate the physicochemical, drug-likeness, and ADMET/pharmacokinetic properties were also performed.

In Vitro Antibacterial Activity of Compounds 1-14
Artepillin C (5), drupanin (6), and compounds 7-14 were obtained by hydrolysis, reduction with LiAlH 4 , and/or catalytic hydrogenation of prenylation products of methyl p-coumarate 1-4, as shown in Scheme 1. Table 1 summarizes the results of the antimicrobial assays with the fourteen compounds against a representative panel of oral bacteria regarding their minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC).The lowest MIC values (31.2 µg/mL) were observed for compound 2 (plicatin B) against Streptococcus mutans (S. mutans), Streptococcus mitis (S. mitis), and Streptococcus sanguinis (S. sanguinis) and for compound 8 against S. mitis.In this study, we investigated the antibacterial activity of a series of synthetic pren ated phenylpropanoids, including the naturally occurring compounds artepillin C a drupanin and their derivatives, against a representative panel of oral bacteria.In sili studies with the most active compounds aiming to understand the target-compound teractions and to estimate the physicochemical, drug-likeness, and ADMET/pharmaco netic properties were also performed.

In Vitro Antibacterial Activity of Compounds 1-14
Artepillin C (5), drupanin (6), and compounds 7-14 were obtained by hydrolysis, duction with LiAlH4, and/or catalytic hydrogenation of prenylation products of methyl coumarate 1-4, as shown in Scheme 1. Table 1 summarizes the results of the antimicrob assays with the fourteen compounds against a representative panel of oral bacteria garding their minimum inhibitory concentration (MIC) and minimum bactericidal co centration (MBC).The lowest MIC values (31.2 µg/mL) were observed for compound (plicatin B) against Streptococcus mutans (S. mutans), Streptococcus mitis (S. mitis), and Str tococcus sanguinis (S. sanguinis) and for compound 8 against S. mitis.Some information related to the crystal structures of the S. mitis (PDB ID: 3LE0) [46], S. sanguinis (PDB ID: 4N82) [47], and S. mutans (PDB ID: 3AIC) [48] targets are given in Table 2.The docking simulation results for compounds 2 and 8, the most active compounds against the selected oral bacteria, along with both X-ray crystals of the studied targets, are listed in Tables 3, S1 and S2.The complexes formed by compounds 2 and 8 with the active site residue of S. mitis (3LE0) have low energy values (−4.228 and −4.476 kcal/mol, respectively, Table 3) compared to the native ligand glycerol (GOL, −3.655 kcal/mol).Four strong hydrogen bonds between compound 8 and the active site residue of the S. mitis (3LE0) target were observed: one conventional H-bond type (H/HIS85(A)-NE2/bond distance = 2.23 Å) and three other carbon H-bonds (H/ASP114(A)-OD1/bond distance = 2.87 Å, H/ASP77(A)-OD2/bond distance = 2.90 Å), and H/ASP114(A)-OD1/bond distance = 2.95 Å).One electrostatic interaction between compound 8 and ARG112(A) was also formed.On the other hand, this compound formed one hydrophobic interaction with VAL117(A) (Table 3 and Figure 2b).In addition, the native ligand (GOL) formed three strong hydrogen bonds, two conventional hydrogen bonds with CYS241 and one carbon hydrogen bond with HIS85(A) (Figure 2c).Noteworthily, the residues ARG120(A), ARG112(A), and HIS85(A) were found in the binding site of the S. mitis target and they were involved in different interactions in the compounds 2, 8, and native ligand (GOL) (Figure 2a-c).

Molecular Dynamics Simulation
Figure 5 presents the 3D diagrams of the best pose for compounds 2 and 8 with the active binding sites 3LE0, 4N82, and 3AIC, as determined by molecular docking (green) and molecular dynamics (yellow) studies.As shown in Figure 6, a fluctuation in all the curves was initially observed during the first 100 picoseconds.A slight variation in the potential energy was observed between 100 and 800 ps in curve (a), and between 100 and 400 ps in curve (e), and between 100 and 900 ps in curve (c).Finally, the 3LE0-compound 8, 4N82-compound 2, and 3AIC-compound 2 complexes retained their stabilities in the last intervals (between 800 and 1500 ps, between 400 and 1500 ps, and between 900 and 1500 ps, respectively) (Figure 6a-c).

Molecular Dynamics Simulation
Figure 5 presents the 3D diagrams of the best pose for compounds 2 and 8 with the active binding sites 3LE0, 4N82, and 3AIC, as determined by molecular docking (green) and molecular dynamics (yellow) studies.As shown in Figure 6, a fluctuation in all the curves was initially observed during the first 100 picoseconds.A slight variation in the potential energy was observed between 100 and 800 ps in curve (a), and between 100 and 400 ps in curve (e), and between 100 and 900 ps in curve (c).Finally, the 3LE0-compound 8, 4N82-compound 2, and 3AIC-compound 2 complexes retained their stabilities in the last intervals (between 800 and 1500 ps, between 400 and 1500 ps, and between 900 and 1500 ps, respectively) (Figure 6a-c).

Drug-Likeness Prediction and ADMET Properties Drug-Likeness Evaluation
Different physicochemical properties were calculated for compounds 2 and 8, aiming to verify the drug-likeness rules using the SwissADME online server.As shown in Table 4, it is apparent that compounds 2 and 8 have hydrogen bond donors < 5 (n-HD: 0~7) and hydrogen bond acceptors < 10 (n-HA: 0~10).In addition, the molecular weight values of these compounds belong to the interval 100~500 g/mol, and their MLogP and WLogP values are <5.Also, the nROTB values are <11, which denotes the flexibility of these compounds.Moreover, the TPSA values obtained for both compounds are less than 140 Å.

ADMET Properties
The absorption, distribution, metabolism, excretion, and toxicity (ADMET) were calculated for compounds 2 and 8 using the online server pkCSM (Table 5).The Caco-2 values obtained for 2 and 8 are 1.178 and 1.417, respectively, whereas the human intestinal absorption (HIA) values are 99.207% and 97.682%, respectively.In terms of the distribution, the logPS values of 2 and 8 are 1.800 and −1.692, respectively, while their logBB values are 0.311 and 0.030, respectively (Table 5).Compounds 2 and 8 are CYP1A2 inhibitors, but not CYP2C19 and CYP2D6 inhibitors.In addition, these compounds are not CYP2D6 and CYP3A4 substrates and are likely not OCT2 substrates.These compounds have a total clearance of 0.744 and 0.780, respectively.Furthermore, compounds 2 and 8 are neither hERG I nor hERG II inhibitors.

In Vitro Antibacterial Activity of Compounds 1-14
The antimicrobial activity of compounds 1-14 was classified based on the MIC (minimum inhibitory concentration) values as follows: MIC values lower than 10 µg/mL, between 11 and 100 µg/mL, between 101 and 500 µg/mL, and between 501 and 1000 µg/mL correspond to very strong, strong, moderate, and weak activities, respectively, while MIC values higher than 1000 µg/mL denote inactivity [1,[52][53][54].Based on these criteria, compounds 2 and 8 displayed strong or moderate activity against all the tested bacteria (Table 1).However, the lowest MIC value was observed for 2 (plicatin B), which showed strong activity against S. mutans, S. mitis, and S. sanguinis (MIC = 31.2µg/mL).This compound also showed strong activity against S. salivarius and S. sobrinus (MIC = 62.5 µg/mL).The strong activity of compound 2 against S. mutans is an auspicious result because it is one of the leading agents causing dental caries [1].Plicatin B (2) has been isolated from Brazilian propolis samples, even though its occurrence is less commonly reported [55].However, data on the antimicrobial activity of plicatin B are still scarce in the literature [56].On the other hand, artepillin C (5) and drupanin (6), which are among the most common prenylated compounds found in Brazilian green propolis, showed moderate or weak activity against most of the tested bacteria (MIC between 125 and 1000 µg/mL), with drupanin displaying MIC values slightly lower than artepillin C. Compounds 1, 10, and 14 had no activity against any of the oral bacteria tested, with MIC values higher than 1000 µg/mL.
According to the literature, compounds with an MBC:MIC ratio ≤ 4, and 4 < MBC:MIC ≤ 32 display bactericidal and bacteriostatic effects, respectively, whereas MBC:MIC values > 32 denote a bacteria-resistant effect [57].Based on these criteria, compounds 2 and 8 are considered to display bactericidal activity against all the oral bacteria tested in this study.
We compared our data on the antibacterial activity of compounds 1-14 with the relevant literature to derive insights into the potential structure-activity relationships.Based on the comparison of the antimicrobial activities of artepillin C (5), drupanin (6), and 4-hydroxycinnamic acid, Aga and coworkers considered that the antimicrobial activity of 4-hydroxycinnamic acid derivatives may be boosted by an increasing number of prenyl groups in the structure [58].However, in this study, the mono-prenylated compound 2 (plicatin B) and its corresponding carboxylic acid 6 (drupanin) were more active against the tested oral bacteria than the di-prenylated compounds 1 and 5 (artepillin C), respectively.These results indicated clearly that other structural features than the prenyl groups also play a key role in the antibacterial activity of compounds 1-14.For example, the phenolic hydroxyl is a crucial structural feature in relation to the antibacterial activity, as evidenced by the comparison between the MIC values of the isomers 2 (a C-prenylated compound) and 4 (an O-prenylated compound).The role played by phenolic hydroxyl is likely due to its capability to form a hydrogen bond to the active binding sites of the bacteria enzymes, thus inhibiting microbial enzymes and simultaneously increasing the affinity with cytoplasmic membranes [59].This is reinforced by the higher activity of 2 (a mono-C-prenylated ester) compared to compound 3 (a tri-prenylated ester that contains a prenyl group at the oxygen at C4), despite the higher lipophilicity of 3. Indeed, docking studies confirmed the formation of a hydrogen bond between the phenolic hydroxyl of compounds 2 and 8 with the residues HIS 85(A), and ASN 104(A) in S. mitis, and S sanguinis, respectively.
The nature of the oxygenated function at C9 and the type of covalent bond between C7 and C8 were also found to affect the antibacterial activity of compounds 1-14; however, the effects of these structural features vary according to the number and position of the prenyl groups.For example, the mono-C-prenylated methyl ester 2 (plicatin B) and its hydrogenated derivative 8 are more active than their corresponding carboxylic acids 7 (drupanin) and 13, indicating that the presence of an ester function at C9 may potentialize the antibacterial activity of mono-C-prenylated compounds.However, in the case of compounds di-C-prenylated 5 (artepillin C) and O-prenylated 7, the carboxyl group at C9 increases the antibacterial activity as compared to the corresponding methyl esters 1 and 4. Similarly, the presence of a single bond between C7 and C8 in the structure of mono-C-prenylated compounds 8 (a methyl ester) and 13 (a carboxylic acid) decreases the antibacterial activity as compared to compounds 2 and 6, which display a double bond between C7 and C8.On the other hand, for compounds di-and tri-prenylated 12 and 9, a single bond between C7 and C8 increases the antibacterial activity as compared to 5 and 3, respectively.
The important role played by the number and position of the prenyl groups in terms of the antibacterial activity of compounds 1-14 can be better understood based on the studies reported by Kubo and coworkers.According to those authors, the mechanism by which phenolic acids enter the molecular structure of the bacteria membrane involves the orientation of the phenol hydroxyl into the aqueous phase by hydrogen bonding, and the non-polar carbon chain alignment into the lipid phase by dispersion forces.The activity tends to disappear when the hydrophilic force exceeds the hydrophobic one [60].Therefore, based on the literature [60], the antibacterial activity of compounds 1-14 can be understood in terms of the balance between hydrophilicity/lipophilicity.In this sense, compounds 2 and 8 can be inferred to achieve the best lipophilicity/hydrophilicity balance among the tested compounds.Hydrolysis of plicatin B (2) produces drupanin (6), which has a carboxyl group at C9.Although this group can anchor the compound to the lipid bilayer [61], it also increases the hydrophilicity of drupanin ( 6) as compared to plicatin B (2) and consequently decreases its antibacterial activity.On the other hand, the presence of two C-prenyl groups in the structure of artepillin C (5) increases lipophilicity as compared to 2 and 6, thus decreasing its antibacterial activity.Finally, the presence of an α,β-unsaturated carbonyl conjugated with the aromatic ring of phenylpropanoids 1-7 confers conformational and electronic characteristics that are strongly influenced by the phenol-OH group in the para position [62].In principle, because of the rotation around the single bond between C7 and C8, compounds 8, 9, 12, 13, and 14 can assume a wider variety of conformations as compared to 1-7, 10, and 11, which could improve their intermolecular interactions with a potential target [63].However, data from this study suggest a combined effect of the nature of the bond between C7 and C8 and the number and position of the prenyl groups on the molecular shape of compounds 1-14 and, consequently, on their antibacterial activity.

Molecular Docking and Molecular Dynamics
The antibacterial activity of compounds 2 and 8 against S. mitis, S. sanguinis, and S. mutans was further investigated using molecular docking.Molecular docking is a computational technique that predicts the precise positioning of a ligand molecule over a receptor protein molecule to create a stable complex [64].This orientation is essential for predicting the binding affinity and strength of the ligand-protein connection through the use of a scoring function [65].The interaction between a drug and its receptor reduces the overall free energy of the system, providing insights into the molecule's affinity and activity, which is crucial for drug design and discovery [66].Docking studies can both describe the protein and the ligand as complementary surfaces [67,68] and provide the ligand-protein pairwise interaction energies [69].
The protein targets in S. mitis (3LE0) [46], S. sanguinis (4N82) [47], and S. mutans (3AIC) [48] were selected on the basis of their relevance to the microorganism virulence and the availability of their resolved crystallographic structures in the Protein Data Bank (PDB) [70].Based on the binding score energy values, compounds 2 (−4.228 kcal/mol) and 8 (−4.476 kcal/mol) were found to inhibit S. mitis (3LE0) more effectively than the native ligand glycerol (GOL) (−3.655 kcal/mol).However, compound 8 was predicted to be a stronger S. mitis (3LE0) target binder compared to compound 2 due to the formation of a more stable complex, as indicated by its negative score energy of −4.476 kcal/mol.These results are in full agreement with the lower MIC value of compound 8 (0.12 mM) compared to compound 2 (0.13 mM) (Table 2).Furthermore, these results suggest that compound 8 has a high affinity with the S. mitis (3LE0) pocket, which is confirmed by establishing four strong hydrogen bonds (a conventional H-bond type and three carbon H-bonds) [71,72].Of particular interest is the electrostatic interaction with ARG112(A).Recent studies have revealed that residues ARG112(A) and HIS85(A) play an essential role in the inhibition of the S. mitis (3LE0) target [73,74].
Conversely, compound 2 is more active against S. sanguinis (MIC = 31.2µg/mL, 0.13 mM) than compound 8 (MIC = 62.5 µg/mL, 0.25 mM) (Table 1).Indeed, molecular docking with these compounds revealed the higher stability of the complex formed by compound 2 (score energy of −6.156 kcal/mol) with the protein target in S. sanguinis (4N82) compared to that of the complex formed by compound 4 (score energy of −5.575 kcal/mol).This stability is similar to that of the flavin mononucleotide (FMN), the native ligand (score energy of −6.671 kcal/mol, Table 3).This higher stability can be due to the two strong carbon H-bonds (O/ASN104(A)-H and O/GLY103(A)-HA3), one Pi-lone pair interaction (with TYR63(A)), and four hydrophobic interactions between compound 2 and residues PHE107(A), MET132(A), PRO62(A), and TYR64(A) of the S. sanguinis protein target (Table 3 and Figure 3A).Indeed, the role played by these residues in the stability of the formed complex has been reported [47,73].
The score energies of the complexes formed between the S. mutans target (3AIC) and compounds 2 (−5.049 kcal/mol) and 8 (−5.042 kcal/mol) are similar to those of acarbose (−6.674 kcal/mol, Table 3), the native ligand.Furthermore, compound 2 fits nicely into the S. mutans (3AIC) pocket due to the formation of three strong hydrogen bonds [71,72] with active site residue of the S. mutans (3AIC) target: one conventional H-bond type and two other carbon H-bonds.An electrostatic interaction with GLU515(A) was also observed.In contrast, this compound formed five hydrophobic interactions with the following active site residues: LEU382(A), HIS587(A), TYR610(A), TYR916(A), and LEU433(A) (Table 3, Figure 4A).This result was supported by several recent studies [49][50][51].
MD simulations were also performed to investigate the stability of the best complexes obtained through molecular-docking calculations for compounds 2 and 8 with the protein targets.Based on a generic model of the physics driving interatomic interactions, MD simulations predict the movements of individual atoms inside proteins and other molecular systems across time [75].These simulations display the locations of all the atoms with femtosecond temporal precision, and they can capture a wide range of significant biomolecular processes, such as conformational change, ligand binding, and protein folding [76].Crucially, these simulations can also forecast, down to the atomic level, how biomolecules will react to various perturbations like mutation, phosphorylation, protonation, or ligand addition or removal [77].In this study, the stability of the best complexes obtained through molecular-docking calculations for compound 2 (4N82-compound 2 and 3AIC-compound 2) and 8 (3LE0-compound 8) with the protein targets was investigated by MD simulations (Figure 5).The energy potential fluctuation observed in all the curves in the first 100 picoseconds (ps) (Figure 5) can be justified based on the variation in the potential energy of three complexes: 3LE0-compound 8 (Figure 5a); 4N82-compound 2 (Figure 5b); and 3AIC-compound 2 (Figure 5c).However, all the studied complexes exhibit high stability, as confirmed by the MD simulations, because they maintained almost the same types of interactions compared to the molecular-docking studies.This was also confirmed by the stability of the potential energy as a function of time (Figure 5a-c), as reported in previous papers [78,79].

Drug-Likeness and ADMET Properties
We also estimated the drug-likeness and ADMET properties of compounds 2 and 8.The drug-likeness considers how simple physicochemical properties impact molecular behavior in vivo [80].Lipinski's "rule-of-five" predicts the drug-likeness of an oral therapeutic agent based on its physicochemical properties [81].This depicts the connection between physicochemical and pharmacokinetics indices [82].According to this rule, an orally active drug-like compound should not have more than one violation of the following criteria: hydrogen bond donors not greater than 5, hydrogen bond acceptors not greater than 10, molecular weight not greater than 500 Da, and an octanol-water partition coef-ficient (log P) not greater than 5 [81,82].As shown in Table 4, all the physicochemical properties of compounds 2 and 8 follow Lipinsky's rule.
The ADMET (absorption, distribution, metabolism, excretion, and toxicity) properties play a critical role in sorting ligands in drug discovery programs, as they account for the failure of 60% of drug candidates during the drug development process [83].In this study, the Caco-2 values obtained for 2 and 8 are higher than −5.15 (>−5.15cm/s), which confirms that these compounds have good human intestinal permeability.Moreover, both compounds have HIA values higher than 30%, which means that compounds 2 and 8, when administered orally, can be absorbed from the gastrointestinal system into the human body's bloodstream.The logPS values (−3< logPS <−2) indicate that compounds 2 and 8 cannot penetrate the CNS.Additionally, the logBB values of compounds 2 (0.311) and 8 (0.030) indicate that compound 2 is expected to readily cross the blood-brain barrier, whereas compound 8 is poorly distributed in the brain.Compounds 2 and 8 are CYP1A2 inhibitors, but not CYP2C19 and CYP2D6 inhibitors.In addition, these compounds are not CYP2D6 and CYP3A4 substrates and are not likely OCT2 substrates.These compounds are also expected to have a low clearance (<5 mL/min/kg).Furthermore, compounds 2 and 8 are neither hERG I nor hERG II inhibitors, showing no hepatotoxicity risk.Therefore, compounds 2 and 8 satisfy the drug-likeness criteria without violating the Lipinski [81], Veber [84], and Egan rules.Furthermore, based on these results, these compounds are not expected to cause problems with regard to oral bioavailability and pharmacokinetic parameters.

Synthesis of Compounds 1-14
Compounds 1-4 were synthesized by alkylation of methyl p-coumarate according to the methodology employed by Patra and coworkers [85] with modifications (Scheme 1).In this procedure, methyl p-coumarate (0.5 mmol) was added to a 25 mL flask equipped with a magnetic stirrer bar.Next, 15 mL toluene was added.The mixture was cooled to 0 • C, and 1.5 mmol NaH was added in portions.After 15 min, 1.5 mmol (153 µL) of prenyl bromide was added dropwise.The reaction progress was monitored through TLC using a Hex:EtOAc 8:2 (v/v) solution as the eluent.After 24 h, the solvent was removed under reduced pressure on a rotary evaporator, and the reaction mixture was then extracted with EtOAc (3 × 15 mL).The organic phase was washed with saturated NaCl solution (15 mL), dried over MgSO 4, and filtered off.The solvent was removed by evaporation at reduced pressure on a rotaevaporator.The compounds were isolated by column chromatography with gradient elution, starting with 100% hexane and changing to Hex:EtOAc 9.8:0.2(v/v) after the separation of the first compounds.The resulting solids were dried under a vacuum to yield compounds 1 (32% yield), 2 (18% yield), 3 (15% yield), and 4 (35% yield) (Scheme 1).
Compounds 5, 6, and 7 were obtained from a hydrolysis reaction according to the methodology described by Uto and co-workers [86].In this procedure, a solution of KOH (15 mL of a 10% aqueous) was added to a solution of the methyl esters 1, 2, and 4 in MeOH (15 mL) (Scheme 1).The mixture was heated under reflux for 1 h, cooled to 0-5 • C, and acidified with 1 mol/L HCl.After removing MeOH under reduced pressure, the aqueous residue was extracted with EtOAc (3 × 15 mL).The organic phase was washed with saturated NH 4 Cl and brine, dried over MgSO 4 , and evaporated under reduced pressure.The resulting white solids were dried under a vacuum to afford compounds 5, 6, and 7 in a 100% yield.
Compounds 10 and 11 were synthesized according to the methodology reported by Kantee and co-workers [87] with some modifications (Scheme 1).In this procedure, LiAlH 4 (0.68 mmol) was quickly added to a solution of compounds 3 and 4 (0.34 mmol) in THF (5.5 mL) at 0 • C. The reaction mixture was stirred under an N 2 atmosphere at 0 • C for 1 h.After one hour, the mixture was stirred at room temperature for 9-10 h.The reaction mixture was then augmented with H 2 O, conc.HCl and extracted with EtOAc (x3).The combined organic layers were washed with brine, dried over MgSO 4 , and concentrated under a vacuum.Purification of the crude residue by column chromatography with isocratic elution with Hex/EtOAc 9.8:0.2(v/v) gave the corresponding alcohol derivative.The resulting compounds were dried under a vacuum to yield compounds 10 (65% yield) and 11 (89% yield).

Antibacterial Assays
The in vitro antimicrobial action of artepillin C (5), drupanin (6) and derivatives 1-4, and 7-14 was evaluated in terms of their minimum inhibitory concentration (MIC) values [88], which were interpreted as the lowest concentrations that inhibited bacterial growth.To this end, Streptococcus mutans (ATCC 25175), Streptococcus mitis (ATCC 49456), Streptococcus salivarius (ATCC 25975), Streptococcus sanguinis (ATCC 10556), Streptococcus sobrinus (ATCC 33478), Enterococcus faecalis (ATCC 4082), and Lactobacillus paracasei (ATCC 11578) were assayed by the broth microdilution method in 96-well microplates.The bacterial colonies were cultured at 37 • C for 24 h in blood agar (Difco Labs, Detroit, MI, USA).Further standardization of the inoculum quantity was accomplished on a Femto spectrophotometer (São Paulo, Brazil) operating at a wavelength of 625 nm to match 0.5 on the McFarland scale (1.5 × 10 8 CFU/mL).The microorganism suspensions were diluted to a final concentration of 5 × 10 5 CFU/mL.Samples of compounds 1-14 were dissolved in DMSO (Merck, Darmstadt, Germany) and tryptic soy broth (TSB, Difco) to obtain final concentrations varying from 0.98 to 2000 µg/mL.Inoculated microplate wells containing DMSO (1%) and TSB (1:5 v/v and 100%) were employed as the negative control.A non-inoculated well was also added to ensure the medium sterility.Chlorhexidine (Sigma-Aldrich, St. Louis, MO, USA) was used as the positive control at concentrations ranging from 0.115 to 59 µg/mL in TSB (Difco).The microplates were sealed with plastic film and incubated at 37 • C for 24 h.Next, 30 µL of revealing 0.02% resazurin (Sigma-Aldrich, St. Louis, MO, USA) was added to each microplate well to indicate the microbial viability.Before resazurin was added and to determine the MBC, a 10 µL aliquot of the inoculum was aseptically removed from each well and plated onto blood agar (Difco).The plates were incubated as described previously.The minimum bactericidal concentration (MBC) was determined as the lowest compound concentration that killed > 99.9% of the initial bacteria population, at which point no visible bacterial growth occurred [89].The MIC values were assessed by analysis of the color change of the resazurin solution from blue (without metabolic activity) to pink (with metabolic activity) [90].The MIC and MBC were determined in triplicate for each microorganism, and the results are presented as the mean of three replicates.

Computational Methodology 4.3.1. Ligands and Targets Preparation
The 3D structures of the most active compounds 2 and 8 (Scheme 1) were optimized using the semi-empirical method AM1 [91], which was implemented in Hyperchem 8.0.8 software (Hypercube Inc., Gainesville, FL, USA).Next, the 3D structures were converted into .mdbformat for use as input in the MOE-docking.

Docking Method Protocol and Validation
The molecular-docking studies were carried out to identify the binding interactions of the most active compounds 2 and 8 within the binding site residue of the targets by using MOE software [92], and the docking protocol steps were followed and detailed in our previous research [93,94] by using the following default parameters: placement: Triangle Matcher; rescoring 1: London dG scoring function.The visualization of all the possible interactions that have been formed between the compounds and receptor active site residues was generated by the BIOVIA DS visualize package (Dassault Systèmes BIOVIA, Discovery Studio Modeling Environment, 2020).
The re-docking of all the native ligands to their targets was conducted using "Dock Option" implemented in the MOE software [92] to validate the used method.The RSMD values of the obtained complexes (targets-crystallized ligands) were less than 2.50 Å [95], meaning that the docking method is accurate and successful.

MD Simulations
MD simulations were carried out using the "Compute/Simulations/Dynamics" options in the MOE 2014.0901software [92] to study the stability of the two best-formed complexes.The following default parameters were used: the Nose-Poincaré-Andersen (NPA) algorithm to search the interactions of different residues in each system, and the MMFF94x force field for the energy minimization step of these complexes [96,97].The MD protocol was settled for 1500 ps (an equilibrium period of 100 ps, followed by a production period of 1400 ps, at a constant temperature of 310 K).Finally, the plots of the potential energies U (kcal/mol) variations as a function of time t (ps) were produced using OriginPro 9.1.software[98].

Conclusions
Plicatin B (2) was identified as a compound with promising activity against oral bacteria, including S. mutans, one of the main causative agents of bacteria caries.Moreover, this study revealed plicatin B to be more active than artepillin C, which has been highlighted as the main factor responsible for the antimicrobial activity of the Brazilian green propolis.However, among all the tested compounds, the most remarkable activity against S. mutans was found for compound 8, which was obtained by hydrogenation of plicatin B.
The molecular-docking/dynamics simulation results proved that compounds 2 and 8 have high binding affinities against three targets, S. sanguinis (4N82) and S. mitis (3LE0), as confirmed by the low energy values and various interactions with the active site residues of these targets.The ADMET properties were predicted for all the candidates to validate the pharmacodynamics and pharmacokinetics profiles, and these compounds verified the three rules, Lipinski, Veber, and Egan.In summary, the results of the present research revealed that the most active compounds 2 and 8 should be further investigated to explore their anti-cariogenic properties, including evaluation of their cytotoxicity, shelf life, anti-

Table 2 .
Some information related to the studied targets.

Table 3 .
Docking results of compounds 2 and 8 when docked into the S. mitis target (3LE0).

-Score (kcal/mol) RMSD (Å) Bonds between Atoms of Compounds and Active Site Residues Atom of Compound Involved Receptor Atoms Involved Receptor Residues Category Type of Interaction
HB: hydrogen bond.GOL: glycerol.

Table 4 .
Physicochemical properties and drug-likeness predictions of compounds 2 and 8. : topological polar surface area; n-ROT: number of rotatable; MW: molecular weight; Log P: logarithm of partition coefficient of compound between n-octanol and water; n-HA: number of hydrogen bond acceptors; n-HD: number of hydrogen bonds donors. TPSA