Synthesis , Antibacterial Evaluation , and QSAR of Caffeic Acid Derivatives

0e present study evaluates the antibacterial effects of a set of 16 synthesized caffeic acid ester derivatives against strains of Staphylococcus aureus and Escherichia coli, as well as discusses their structure-activity relationship (SAR). 0e antibacterial assays were performed using microdilution techniques in 96-well microplates to determine minimal inhibitory concentration (MIC). 0e results revealed that five of the compounds present strong to optimum antibacterial effect. Of the sixteen ester derivatives evaluated, the products with alkyl side chains, as propyl caffeate (3), butyl caffeate (6), and pentyl caffeate (7), presented the best antibacterial activity withMIC values of around 0.20 μMagainst Escherichia coli and only butyl caffeate (6) showing the sameMIC against Staphylococcus aureus. For products with aryl substituents, the best MIC results against the tested strain of Escherichia coli were 0.23 μM for (di-(4-chlorobenzyl)) caffeate (13) and 0.29 μM for diphenylmethyl caffeate (10) and all were less active against the Staphylococcus aureus strain. Preliminary quantitative structure-activity relationship (QSAR) analyses confirmed that certain structural characteristics, such as a median linear carbon chain and the presence of electron withdrawal substituents at the para position of the aromatic ring, help potentiate antibacterial activity.


Introduction
Bacterial diseases are a global health problem and are prevalent in developing and low-income countries [1,2], causing 13 to 17 million deaths annually, equivalent to 25% of global deaths and 45% for developing countries [3].Due to the bacterial resistance of Staphylococcus aureus to methicillin and vancomycin [4] and Escherichia coli to β-lactams [5], there is an urgent need for discovery and development of new classes of antimicrobial compounds to treat bacterial infections.Methicillin-resistant Staphylococcus aureus (MRSA) is responsible for high hospital infection rates [6,7].ese infections range from simple acnes, boils, and abscesses to more severe forms such as pneumonia, meningitis, endocarditis, and septicemia [8,9].β-Lactams are not always fully effective against strains of MRSA [10].Vancomycin is an alternative; however, there are already case reports of intermediate [11] and full resistance to the drug [12].

Chemicals
All of the chemical products used during synthesis were from Sigma-Aldrich. 1 H-NMR (200 and 50 MHz) and 13 C-NMR (500 and 125 MHz) spectra were, respectively, recorded on Varian Mercury and Varian-RMN-System spectrometers.Chemical shifts (δ) are expressed in parts per million (ppm) using TMS as an internal standard.Spin multiplicities are given as s (singlet), brs (broad singlet), d (doublet), dd (doublet of doublets), t (triplet), q (quartet), quint (quintet), sext (sextet), sept (septet), and m (multiplet).High-resolution mass spectrometry was carried out using an Ultraflex II TOF/ TOF spectrometer with a high performance solid state laser (λ � 355 nm) and a reflector using the MALDI (matrixassisted laser desorption ionization) technique.Column adsorption chromatography (CC) was performed on silica gel (Merck 60, 230-400 mesh); analytical TLC was performed on precoated silica gel plates (Merck 60 F 254 ).FTIR spectra were recorded on a Bruker FTIR spectrometer, Vertex 70 model, using KBr pellets.

Synthesis of Compounds 1-8 for Fischer's Esterification.
A mixture of caffeic acid (0.25 g, 1.39 mmol) and alcohol (50 ml) was heated under reflux in the presence of sulfuric acid (0.4 ml) until completion of the reaction (5-21 hours) and verified by a single spot in TLC. e alcohol was then removed under reduced pressure, and the solution was diluted with 20 ml of water.e product was extracted with ethyl acetate (15 ml).
e organic phase was neutralized successively with 5% sodium bicarbonate and water, dried over anhydrous sodium sulfate, and filtered.After evaporation under reduced pressure, this phase yielded the ester derivatives [44].

Reaction with Alkyl and Aryl
Halides for Compounds 9-14 3.2.1.Synthesis of Compounds 9 and 10.Caffeic acid (0.2 g, 1.11 mmol) in acetone (13.5 ml) was heated under reflux in the presence of triethylamine (0.6 ml) and halide (1.14 mmol) until complete reaction (44-120 hours), which was verified by a single spot in TLC. e solvent was then removed under reduced pressure, and the solution was diluted with 20 ml of water.e product was extracted with ethyl acetate (20 ml).e organic phase was treated with water, dried over anhydrous sodium sulfate, filtered, and concentrated under reduced pressure.
e mixture was stirred at room temperature, until complete reaction (27-50 hours), which was verified by a single spot in TLC. e solvent was then removed under reduced pressure, and the solution was diluted with 15 ml of water.e product was extracted with ethyl acetate (10 ml).e organic phase was neutralized successively with 5% sodium bicarbonate (10 ml), washed with brine (10 ml), dried over anhydrous sodium sulfate, filtered, and concentrated under reduced pressure.
e residue was purified by using a chromatographic column with hexane and ethyl acetate eluents in an increasing polarity to isolate the ester derivatives [47].

Synthesis of Compounds 15 and 16 for the Mitsunobu
Reaction.A mixture of caffeic acid (0.1 g, 0.55 mmol) and alcohol (0.55 mmol) was dissolved in 1.85 ml tetrahydrofuran.
e reaction mixture was stirred under magnetic stirring at 0 °C for about 30 minutes.Afterwards, diisopropyl azodicarboxylate (0.11 ml, 0.55 mmol) and triphenylphosphine (0.15 g, 0.55 mmol) were added as esterification agents, with continuous stirring at room temperature for about 46-50 hours, which was verified by a single spot in TLC. e solvent was then removed under reduced pressure, and the solution was diluted with 10 ml of water.
e product was then extracted with ethyl acetate (10 ml).e organic phase was neutralized successively with saturated sodium bicarbonate, washed with brine, dried over anhydrous sodium sulfate, filtered, and evaporated under reduced pressure.e residue was then purified on a silica gel column (eluent: hexane-ethyl acetate, 7 : 3) [48].
Culture Medium.e antifungal activity assays were performed in brain heart infusion (BHI) broth liquid medium (Difco Laboratories/France/USA), which was prepared and used according to the manufacturer's instructions.

Determination of the Minimum Inhibitory Concentration (MIC).
e MIC values were determined by the microdilution method [54] using 96-well "U"-shaped microtiter plates in duplicate.In each plate well, 100 μl of twice concentrated BHI liquid medium was added, except the first line which received 160 μl.en, 40 μl of product solution (also doubly concentrated) was placed in the first row of the plate wells.By serial dilution (ratio of two), concentrations from 200-12.5 μg/ml were obtained, such that in the first line of the plate was the highest concentration and in the latter lower concentration.Finally, another 100 μl of the BHI medium was added to all wells, and 10 μl of inoculum was added to the wells in each plate column that referred specifically to a strain.e same was also done in the culture medium with the bacterial drug chloramphenicol (100 mg/ ml).e plates were incubated at 37 °C for 24 h for further reading; this was performed with the addition of 10 μl of a 0.01% (w/v) solution of resazurin (Sigma), a colorimetric metabolic activity indicator.For each strain, the MIC was defined as the lowest concentration capable of inhibiting bacterial growth in the wells, as visually observed compared with the control.All of the tests were performed in duplicate, and the results were expressed as a geometric mean of the MIC values obtained from both tests [56].

3D-QSAR Modeling
Otherwise noted, default parameters were employed for all software.One initial 3D conformer per compound was generated with OpenEye Omega [57].Each compound's initial conformation was subjected to 200 cycles (100 ps each) of quenched molecular dynamics simulations with Open3DAlign [58] using an implicit solvent model.Molecules alignment was also performed with Open3DAlign by means of its mixed algorithm considering multiple conformers for the template and candidate compounds.
Bioactivities were converted to pIC 50 values (pIC 50 � −log 10 IC 50 ), and QSAR models were separately trained for Escherichia coli and Staphylococcus aureus using Open3-DQSAR [59].e box step for the grid was set to 0.5 Å, and van der Waals (VDW) and electrostatic molecular interaction fields (MIFs) were computed.e dielectric constant was set to 4. Grid points providing VDW values larger than 10 4 kcal/mol were removed from both MIFs.Values of the grid exceeding 30 kcal/mol or below −30 kcal/mol were set to 30 kcal/mol and −30 kcal/mol, respectively, for both MIFs.In addition, all grid points with an absolute value lower than 0.05 were set to zero, and variables with standard deviation lower than 0.1 were excluded from the calculations.Variables spanning up to four levels were also excluded.After these variables filtering steps, those remaining in the dataset were scaled according to the BUW algorithm as implemented in Open3DQSAR.
We considered up to 10 principal components (PCs) for models training.e optimal number of PCs considered in the final model was selected according to the r 2 value of the leave-one-out (LOO) cross validation of the model (q 2 LOO ).With this optimal number of PCs, variables were further grouped using the Smart Region Definition procedure implemented in Open3DQSAR.e grouped variables were subjected to a selection procedure according to the fractional factorial design using leave-many-out (LMO) cross validation with 20 runs and 5 groups.Only selected variables in the previous step remained on the dataset.Finally, given the limited number of available samples, the PLS model was recomputed and its quality was evaluated in 100 cycles of LMO cross validation.
e results obtained in this study revealed that five compounds presented strong to optimum antibacterial activity; the compounds were found to be more active against the Escherichia coli strain.Of the sixteen ester derivatives evaluated, propyl caffeate (3), butyl caffeate (6), and pentyl caffeate (7) were found to exhibit the best antibacterial activity of the products with alkyl side chains, with an MIC value of 0.2 μM against the Escherichia coli strain and only butyl caffeate (6) showed the same MIC for the Staphylococcus aureus strain, whereas of the products with aryl side chains, the best MIC results against the Escherichia coli strain were 0.23 µM for (di-(4-chlorobenzyl)) caffeate ( 13) and 0.29 µM for diphenylmethyl caffeate (10), both being less active against Staphylococcus aureus.Table 1 summarizes the in vitro susceptibilities of the two Escherichia coli and Staphylococcus aureus strains against all of the test compounds.
Of the screened ester derivatives, certain structural features were observed that may increase antibacterial activity, giving information on functional groups that might be important to antibacterial effect, such as alkyl ester side chain, especially a median linear carbon chain, benzyl radicals with two aromatic rings without substituents, and Data analyses revealed that introduction of alkyl substituents as short chains and the presence of a heteroatom and long (10 carbon atoms) or branching chains resulted in decreased antibacterial effect.For example, butyl caffeate (6: butyl group) was found to be the most bioactive compound for both strains (MIC � 0.20 μM), whereas methyl caffeate (1; methyl group), ethyl caffeate (2; ethyl group), isopropyl caffeate (4; isopropyl group), methoxyethyl caffeate (5; methoxyethyl group), isopentyl caffeate (8; isopentyl group), and decyl caffeate (9; decyl group) presented increased MICs (decreased activities).is reduction in antibacterial activity probably occurs due to steric hindrance caused by the presence of bulky groups and also poorly lipophilic shortchain groups.Activity achieved a maximum for the esters with a median linear carbon chain conferring lipophilicity to the molecules.e compounds presenting aryl radicals resulted in decreased antibacterial effect, such as aromatic ring substituents with electron donor groups in the para position such as methyl and methoxy or with electron withdrawal groups such as chloro (except compound (13) which was active against E. coli).
ose without substituents also resulted in decreased antibacterial effect, except compound (10) which was active against E. coli.e presence of an electron-withdrawing group such as di-(4-chlorobenzyl) caffeate (13; di-(4-chlorobenzyl) group) provided more activity against the Escherichia coli strain, which probably occurred due to the para chloro substituent in the caffeic ring, which was the only difference compared to 4-chlorobenzyl caffeate (11; chlorobenzyl group), and presented similarity to the reference drug having two chloro substituents.Compound (10) presented good results against Escherichia coli due to the presence of bulky groups without substituents with two aromatic rings, such as with diphenylmethyl caffeate (10; diphenylmethyl group); this was compared to benzyl caffeate (15; benzyl group), which presented a better bond to its target.
e antimicrobial activity of caffeic acid derivatives has been reported in other studies [43,[60][61][62][63][64][65].For example, Merkl et al. [66] investigated the antibacterial activity of a series of caffeic acid esters (methyl, ethyl, propyl, and butyl caffeate) against E. coli DMF 7503.All of the esters screened showed potent antibacterial activity against the test strain, principally propyl and butyl caffeate, with MIC values of 5.00 mM. e MIC values of the other caffeic acid esters tested were roughly 10.00 mM [67].In the study of Meyuhas et al. [42], the phenethyl caffeate exhibited antibacterial activity against S. aureus (48 mM) and E. coli (254 mM). is compound has structural similarity to ester 15, which showed strong antibacterial action (Table 1).e ester derivatives of caffeic acid were prepared using short linear carbon chain radicals, such as methyl and ethyl caffeate, and from them, compounds synthesized with (4-chloro-benzyl) radicals on the two hydroxyls of the meta and para positions of the caffeic ring.e antibacterial activities of the compounds were evaluated against E. coli and S. aureus; the esters, methyl and ethyl caffeates, showed activity for both strains, while their by-products were only active for the E. coli strain.Derivatives thus prepared are promising candidates for treating bacterial pathologies, and their test results corroborate those already published in the literature.
e statistical parameters of the 3D-QSAR models obtained for the E. coli and S. aureus inhibitory activities are summarized in Table 2.As can be seen, both models have good statistical parameters.Furthermore, for both E. coli and S. aureus, bioactivity is mainly sterically driven.Also, given the lower number of PCs included in the S. aureus model and its statistical parameters, it can be considered statistically more robust.6.2.1.Scaffold.Bioactivity is mostly steric-guided.Models show 77% and 73% contributions of the steric factor to the Escherichia coli and Staphylococcus aureus activities, respectively (Figure 1).

Electrostatic Molecular Interaction Field.
e electrostatic contribution to bioactivity shows that certain substitutions at R 2 can improve bioactivity.In general, the electronegative molecular interaction field next to the O-R 2 bond indicates a positive contribution to bioactivity of substitutions at this position.
is positive contribution cannot be exploited when R 2 � H. is later observation holds for both Escherichia coli and Staphylococcus aureus.Figures 2 and 3 summarize the electrostatic contribution to bioactivity in both strains.
e comparisons of compounds 11 and 13 and of compounds 12 and 14 reveal some interesting details.In the case of Escherichia coli, there is an electronegative molecular interaction field close to the 4-chlorobenzyl substitution at Journal of Chemistry R 2 , indicating that benzyl groups with electronegative parasubstitutions can have an additional positive contribution to bioactivity against this strain.In contrast, substitutions at R 2 such as 4-methoxybenzyl present in compound 14 are unable to increase the bioactivity of R 2 -substituted compounds.is can be clearly exemplified for compounds 12 and 14. e lack of any additional electrostatic molecular interaction field surrounding R 2 in the Staphylococcus aureus model indicates that no increase in the anti-S.aureus bioactivity could be obtained with specific substitutions at R 2 .

Steric Molecular Interaction Field.
e analysis of the steric molecular interaction field shows that there is a steric favorable region around R 1 in both E. coli and S. aureus models.Compounds with small aliphatic substituents at R 1 are favored, while too bulky or too long substituents such as long aliphatic chains and aromatic substitutions are disfavored at this position.is effect is more pronounced in the Staphylococcus aureus model where there is a large unfavorable steric region close to R 1 .For example, the R 1 substituent of compound 6 completely falls into the steric favorable region defined by both strains models.On the other hand, the 4-methylbenzyl substituent of compound 16 falls beyond the steric favorable region in Escherichia coli and in the steric unfavorable region defined by the Sthaphylococcus aureus model.In addition, bulky substitutions at R 2 (such as those present in compound 14) decrease anti-E.coli bioactivity.
e influence of the steric factor in bioactivity is summarized in Figures 4 and 5.
Considering all these observations, for dual Escherichia coli and Staphylococcus aureus bioactivity, small aliphatic substitutions are essential at R 1 .Furthermore, basing on the available data, substitutions at R 2 could be beneficial for dual bioactivity.
is last hypothesis should be confirmed by testing more compounds substituted at R 2 with small groups.
Our QSAR studies confirmed the biological activity of the tested compounds, connecting with chemical structures for (di-(4-chlorobenzyl)) caffeate ( 13), through the electronegative electrostatic molecular interaction field at R 2 as against Escherichia coli and for butyl caffeate (6), due to the small aliphatic radical completely falling into the sterically favorable region as defined by both strain models.In contrast to these, 4-methylbenzyl caffeate (16) exceeded the sterically favorable regions of these models and presented the worst bacterial activity against the tested strains.

Conclusion
e present study investigated the antibacterial activity of sixteen caffeic acid ester derivatives against a Grampositive Staphylococcus aureus strain and a Gram-negative
Contribution of VDW and electrostatic factors to bioactivity, respectively.

Figure 2 :
Figure 2: Electrostatic molecular interaction field for compounds 11 and 13 in Escherichia coli.

Figure 3 :
Figure 3: Electrostatic molecular interaction field for compounds 11 and 13 in Staphylococcus aureus.
(ATCC-2536).e strains belonged to the Bacterioteca of the Laboratory of Toxicological Tests, Institute of Research in Drugs and Medications, Paraíba Federal University.e strains were maintained in an appropriate culture medium, BHI (Difco Laboratories/France/USA), and stored at 4 4. Antibacterial Assaye microbiological testing used single strains of Staphylococcus aureus (ATCC-25619) and Escherichia coli °C and 35 °C.

Table 2 :
Statistical parameters of the 3D-QSAR models for the antibacterial activity against E. coli and S. aureus.
a Number of PCs in the final model.b Coefficient of determination.c LOO cross-validation performance.d LMO cross-validation performance.