Dyhidro-β-agarofurans natural and synthetic as acetylcholinesterase and COX inhibitors: interaction with the peripheral anionic site (AChE-PAS), and anti-inflammatory potentials

Abstract In order to find molecules of natural origin with potential biological activities, we isolate and synthesise compounds with agarofuran skeletons (epoxyeudesmanes). From the seeds of Maytenus disticha and Maytenus magellanica we obtained six dihydro-β-agarofurans, and by means of the Robinson annulation reaction we synthesised five compounds with the same skeleton. The structures were established on the basis of NMR, IR, and MS. The evaluated compounds showed inhibitory activity on the acetylcholinesterase enzyme and on the COX enzymes. Compound 4 emerged as the most potent in the acetylcholinesterase inhibition assay with IC50 17.0 ± 0.016 µM on acetylcholinesterase (AChE). The compounds evaluated were shown to be selective for AChE. The molecular docking, and the propidium displacement assay suggested that the compounds do not bind to the active site of the enzyme AChE, but rather bind to the peripheral anionic site (PAS) of the enzyme, on the other hand, the natural compound 8, showed the best inhibitory activity on the COX-2 enzyme with an IC50 value of 0.04 ± 0.007 µM. The pharmacokinetic profile calculated in silico using the SWISSADME platform shows that these molecules could be considered as potential drugs for the treatment of neurodegenerative diseases such as AD.


Introduction
Alzheimer's disease (AD) is a dementia of unknown aetiology that affects adults, usually over 65 years of age 1 . It is a chronic neurodegenerative disease, which causes a progressive and irreversible decrease in cognitive functions, the onset of the disease being characterised by memory loss. AD is often manifested histologically by synaptic and neuronal loss, intracellular neurofibrillary tangles (NFT), and deposition of b-amyloid (Ab) plaques in human brain 2 . There are several therapeutic targets that can be influenced in order to treat the disease improving the symptoms related to AD or modify its course 3 . The most widely used therapeutic strategy is cholinergic symptomatic therapy, which is based on balance and regulation of neurotransmitters such as acetylcholine.
Acetylcholinesterase (AChE) and butyrylcholinesterase (BChE) are two types of cholinesterase, which catalyse the hydrolysis of choline-based esters to terminate cholinergic signal transmission. These two enzymes differ significantly in substrate specificity, enzyme kinetics, expression and activity in different brain regions and complexity of gene regulation 4 . Both AChE and BChE are therapeutic targets to improve cholinergic dysfunction associated with cognitive and behavioural abnormalities during the development of AD. In the early stages of this disease, the level of AChE increases at a much higher rate than that of BChE in the brain 5 , hence the importance of the use of AChE inhibitors in patients who shown early clinical features of this disease. Currently, inhibitors of AChE are the main strategy for the development of drugs against AD, since by inhibiting this enzyme, the levels of acetylcholine in cholinergic neurons may increase and cholinergic function could improve.
Several decades have passed since the detection of inflammatory mediators in sections obtained from post-mortem brains tissues of AD patients. This findings led to the proposal that neuroinflammatory events could promote AD progression 6 . Recently, in vivo studies using single photon emission (PET) and positron emission computed tomography (SPECT) of AD transgenic mice and AD patients further confirmed that neuroinflammation precedes the outbreak of AD. Therefore, the presence of innate immune cells and the synthesis of inflammatory biomarkers are linked to the progression and severity of the disease 7 . It is suggested that microglia can be activated by Ab peptides, releasing cytokines and neurotoxins, as if they were responding to a pathogen, which cause neuronal damage, thus mediating neurodegeneration. COX enzymes converts arachidonic acid to inflammatory prostaglandins and thromboxane. These reactions are driven by two isoforms of this enzyme involved in prostaglandin biosynthesis, COX-1 and COX-2. COX-1 is known as a household enzyme and is constitutively expressed in all tissues, while COX-2 is constitutively expressed in the kidneys, brain, and ovaries. COX-2 is expressed during inflammatory conditions by pro-inflammatory molecules such as IL-1, TNF-a, LPS and is practically absent in healthy individuals 8 . In this sense, early epidemiological data suggest that long-term treatment with nonsteroidal anti-inflammatory drugs (NSAIDs) reduces the risk of AD due to the inhibition of the cyclooxygenase 2 (COX-2) up-regulation promoted by Ab in glial cells 9 . The stimulation of COX-2 in the microglia promote the synthesis of IL-1b, which once secreted stimulate COX-2 in neuron cells leading to the activation PI3-K/AKT and PKA/CREB pathways. In consequence, this cycle is complete when both kinases phosphorylate NF-kB enabling the transactivation of BACE-1, responsible for APP cleavage to produce more Ab. These events connect the altered expression of COX-2 in the hippocampus with the severity of dementia.
COX-2 inhibitors are widely prescribed anti-inflammatory agents, while COX-1 inhibitors have been associated with several unwanted side effects 10 . Therefore, the development of compounds that exclusively inhibit COX-2 is an important goal to reduce adverse side effects during non-steroidal anti-inflammatory treatment, thus enhancing therapeutic benefits.
Both isoenzymes have similar structures and catalytic activities, the amino acid sequences for the catalytic and binding sites of the substrate are almost identical. So, the main difference between these enzymes lies in that COX-2 has valine instead of isoleucine at positions 434 and 523 11 Valine is smaller than isoleucine because of a methyl group. These substitutions result in a larger and more flexible substrate channel and a secondary internal pocket outside the COX-2 inhibitor binding site, which is not seen in COX-1.
The Celastraceae family is widespread in tropical and subtropical regions of the world, including North Africa, Central and South America, and Central and East Asia 12 . The plants of this family have a long tradition of use in folk medicine and agriculture. The crude extracts of plants of this family have shown an extraordinary variety of pharmacological properties, among which we can mention their use in digestive disorders, fever, arthritis, bacterial infections and cancer. In addition, these plants are used as stimulants, appetite suppressants and insect repellants 13 . The bulk of bioactive constituents of Celastraceae representative are terpenes, being the sesquiterpene dihydro-b-agarofuran the most characteristic 14 , and the quinone methide triterpenoids (called celastroids). Both class of molecules are recognised as chemotaxonomic markers in this family. Dihydro-b-agarofuran sequiterpenoids show a high degree of oxidation and can form polyesters and pyridine-sesquiterpenes alkaloids. These biosynthetic derivatives have a wide spectrum of biological activities such as immunosuppressants, chemopreventives, anti-HIV, anti-insect agents and to circumvent multidrug resistance (MDR) in cancer cells 15 .
Considering the above mentioned information, in the present work we have evaluated the inhibitory activity of synthetic and natural dihydro-b-agarofurans focussed in acetylcholinesterase and COX enzymes. Our findings suggested that dihydro-b-agarofuran sequiterpenes has significant inhibitory activity, which would allow further design or modification of these molecules in order to improve their potential use in the treatment of AD.

Synthetic dihydro-b-agarofurans
The synthesis of five compounds with dihydro-b-agarofuran skeleton by Robinsons annulation reaction from (±)-dihydrocarvone and ethyl vinyl ketone was development as outlined in Figure 1 and the structures were identified by their melting point, FTIR, 1 H NMR, 13 C NMR, and HRMS spectral data. Stereochemistry was unambiguously confirmed with the help of cross-peak intensities observed in nuclear overhauser effect (NOE) and 2 D NOESY (nuclear overhauser effect spectroscopy) spectrum.
Following a published procedure 16 , an epimeric mixture of 10epiedusmane-4-ene-3,11-diol 4 was prepared in four steps starting from (þ)-dihydrocarvone. Treatment with H 2 SO 4 , afforded a-agaro- Compound 6 was obtained by esterification of 4 with furoyl chloride. The 1 H NMR spectrum of compound 6 show the presence of three singlet signals in high field that were assigned to the methyl groups in C-12, C-13, and C-15. In addition, it is observed in this compound an AB system between at 4.01-3.98 (J¼ 6.5 Hz) and 3.65-3.63 (J¼ 8.6 Hz), that is not present in the precursor. Further comparison of the 1 H and 13 C NMR spectroscopic data of compound 4 to those of compound 6 reveals that compound 6 contains one less methyl group and one more methylene bearing a hydroxyl group. This is further confirmed by the HMBC correlations of H-14 (d H 3.99) with C-4 (d C 47.57 ppm), C-3 (d C 76.71 ppm) and C-5 (d C 89.91 ppm). Similarly compound 7 was prepared by esterification of 4 with benzoyl chloride. The 1 H NMR spectrum of compound 7 show the presence of three singlet signals in high field that were assigned to the methyl groups in C-12, C-13, and C-15. In addition, it is observed in this compound an AB system between at d 3.91-3.88 (J¼ 8.0 Hz) and 3.55 À 3.52 (J¼ 10.6 Hz).
These signals are possibly due to an intramolecular rearrangement that facilitated the migration of the hydroxyl group that was present at the C-3 position to the C-14 position. This migration would be facilitated by the use of 4-dimethylamino pyridine (DMAP) as catalyst.

Natural dihydro-b-agarofurans
The chemical study of the fractions obtained from the methanolic extract of the Maytenus disticha and M. magellanica seeds by CC and preparative TLC resulted in the isolation of six compounds ( Figure 3). Compounds 8, 10 and 13 were isolated from M. disticha and compounds 11, 12 and 9 were isolated from M. magellanica. Their NMR spectra showed signals characteristic of dihydrob-agarofurans 17-20 . The 1 H NMR spectrum of compound 8 showed the typical signals of a bsubstituted furan ring (d 6.72; 7.40; 8.01 ppm) in C-9 and the change of field down to d 5.53 of H-6 indicates the relative position of a second furan group ester in C-6. The double doublet at d 5.27 (1H, dd, J¼ 4.0, 12 Hz) was assigned to H-1 because in this class of compound H-1 had generally axial stereochemistry. Furthermore, typical signals were present for eudesmane with an angular methyl group (H-15) at 1.34 ppm (s), a secondary methyl group (H-14) at 1.40 ppm (s), and singlets at 1.51 ppm (H-12) and 1.53 ppm (H-13), the signals at 2.08 ppm (s) corresponded to protons of acetate group. Together with the 13 C NMR spectrum, the presence of the eudesmane system with oxygen functions at C-1, C-4, C-5, C-6, and C-9 Methyl groups for acetates for compounds 11 and 13 appeared as singlets between d 2.10 and 2.24 ppm depending chemical environment of the molecule.

Inhibition of the activity of the enzyme acetylcholinesterase and COX 1/2 and kinetic study of the inhibition of AChE
All compounds synthesised were evaluated in vitro as dual AChE and BChE inhibitors according to the modified Ellman's method 21 . Galantamine and carvacrol, both inhibitors of AChE, were used as the positive controls. The concentration of the compound required for 50% enzyme inhibition (IC 50 ) was calculated by means of regression analysis. The results of this assay are listed in Table 1. IC 50 values are expressed in mM as means ± SEM and were compared by using ANOVA analysis. A p values of less than 0.05 was considered significant.
As it can be observed in Table 1, all the evaluated compounds show inhibitory activity of acetylcholinesterase enzyme. The IC 50 values were from 17.0 ± 0.016 to 740.0 ± 0.045 mM. The results obtained are satisfactory in relation to the reference compounds  used (galantamine and carvacrol), and when compared with other terpenoids of natural origin 22 . None of the compounds evaluated showed significant inhibitory effects on the BChE enzyme (IC 50 >500 mm, data not show). Therefore, these compounds can serve as selective inhibitors for AChE, thus avoiding unexpected effects on peripheral tissues due to BChE inhibition.
Compound 4 was the most potent anti-AChE agent with an IC 50 value of 17.0 ± 0.016 mM to AChE, which makes it significantly better than natural agarofurans isolated from previously reported 16,23 . Inhibitory activity is probably favoured by free hydroxyl groups at positions C-3 and C-4, which favour the formation of hydrogen bonds with the residues of the enzyme.
Based on the obtained results in the cholinesterase inhibition assay, and with the aim of assessing the kinetic mode of AChE inhibition of the target compounds, the most active compound 4 was evaluated. The rate of enzyme activity was measured at eight different concentrations of substrate acetylthiocholine iodide (ATC). After plotting the reciprocal of enzyme velocity (l/v) versus the reciprocal of substrate concentration (l/S), the established Lineweaver-Burk reciprocal plots ( Figure 4) showed that the Vmax decreases in the presence of the inhibitor, while the intercepts remain unchanged, the value of Km ¼ 0.11 mM being at increasing concentrations of the inhibitor for Tc-AChE. These kinetic parameters would indicate that 4 is a non-competitive inhibitor.
Recognising the action mode of enzyme inhibitors is very important; however, much of the drug discovery process only focuses instead upon stablishing whether one inhibitor is more powerful than another. Reversible inhibition comprises of the ensuing subtypes: competitive, uncompetitive, non-competitive or mixed-type. Such inhibitions can be determined using Lineweaver-Burk plots. When a molecule binds to the catalytic active site (CAS), the type of inhibition of the enzyme is classified as competitive; by contrast, when a molecule interacts with the peripheral anionic site (PAS), the enzymatic inhibition is called non-competitive 24 . Furthermore, once the active molecule acts on both CAS and PAS, the enzymatic inhibition is designated as mixed inhibition. Considering the above mentioned, our kinetics study for the compound 4 suggest an interaction with the PAS reflexed by a lower Vmax value and unchanged Km.
In a previous work carried out by Alarc on et al., 25 sesquiterpene esters with a dihydro-b-agarofuran skeleton were isolated from Chilean Celastraceae and it was found that most of the compounds had inhibitory activity on AChE. The IC 50 values for said compounds were found in the range of 0.046 to 0.738 mM. When comparing the results, it can be seen that the synthesised dihydro-b -agarofurans have a better AChE inhibitory activity than the compounds isolated from plants. These results show that there would be a possible relationship between the size of the substituents surrounding decalin and its inhibitory activity. Synthetic compounds with less bulky groups would have better AChE enzyme inhibitory activity.
The synthesised compounds were evaluated as COX-1 and COX-2 inhibitors. The IC 50 values of the compounds against COX-1 ranged from 8.69 ± 0.03 to 14.21 ± 0.2 mM, while the IC 50 values obtained for COX-2 ranged from 0.04 ± 0.007 to 0.84 ± 0.012 mM as shown in Table 1. The compounds showed prominent inhibition constants towards both COX isoforms comparable to those of the standard drugs used, diclofenac sodium and celecoxib. It can be seen that dihydro-b-agarofurans show more potent and selective inhibitory effects on the COX-2 isoform. Interestingly, the best activities come from the naturally occurring compounds 8-13,  It is worthy to note that, structural variations appeared to affect both the potency and the selectivity of these derivatives against the enzymes tested. This is a specially sought trait for the development of selective anti-inflammatory agents to avoid undesirable side effects.
Regarding the results of the COX-2 enzyme inhibition assay, compound 8 was the most potent, showing lower IC 50 value 0.04 ± 0.007 mM and SI 264.75, compared to the known selective COX-2 inhibitor Celecoxib (IC 50 ; 0.049 mM and SI; 308. 16).
The increase in the initial expression of COX-2 and in the enzymatic activity promotes the appearance of neuronal lesions in various models of neurological disease. It is an interesting approach to alter the microglial inflammatory response through COX-2 inhibition 26 . Pharmacological inhibition of COX-2 is neuroprotective in rodent models of stroke 27 . Inhibition of COX-2 enzyme activity with non-steroidal anti-inflammatory drugs has been shown to reduce inflammation in murine transgenic models of AD, as well as the use of COX-2 inhibitors to decrease the risk of developing AD in populations. healthy ageing 28 . COX-mediated neuronal injury is presumed to be due to downstream effects of one or more prostaglandin products including PGE2, PGD2, PGF2a, PGI2 and TXA2 that affect cellular changes through activation of specific prostaglandin receptor subtypes. The suppression of these proteins has been linked to improved memory in transgenic AD mouse models 29 .

Docking Studies on AChE and COX-1/2
To gain insight into the binding interactions of compounds with enzymes, molecular coupling studies were carried out. The data in Table 2 show the free binding energies (À9.1 to À4.2 kcal/mol) of each compound to the enzyme AChE. The most active compound 4 formed three H bonds with AChE. The hydroxyl group at C-3 forms two hydrogen bonds with the residue Arg 289 (1.9 and 1.73 Å) and the hydroxyl group at C-4 forms a hydrogen bond with the residue of Ser 286 (2.14 Å). In addition, the methyl groups at the C- The rest of the compounds evaluated interact in a similar way with some of the amino acids that are part of the PAS. In addition, all of them, with the exception of compound 11, interact through p bonds with amino acids located in the gorge centre of the enzyme such as Phe 330, Trp 279, Phe 290 and Phe 331. These interactions are important because these amino acid residues connect the peripheral site and the anionic subsite in the active site of the enzyme. When inhibitor entering the enzym es active site, possibilities that substrate can reach the bottom of the gorge are limited 30 . This way, dihydro-b-agarofurans would obstruct the entrance of the substrate to the active site. The PAS is responsible for non-cholinergic functions of AChE and plays a role in the deposition of amyloid plaque. Therefore, molecules that bind to PAS site could, plays an important role slowing down the progression of AD 31 .
AChE has been recognised as a target for a number of toxins and promising drugs such as huperzine 32 and derivates, propidium iodide 33 , gallamine triethiodide 34 among others. In order to maintain a stable binding of these inhibitors with the enzyme, many cooperative interactions are established such as hydrogen bonds, hydrophobic contacts, p-p (aromatic) interactions, and hydrophilic-hydrophobic interactions 35 . The chemical structure of the evaluated compounds allows many of these interactions to be formed. For example, the free hydroxyl groups, the epoxide group and the carbonyl groups can form hydrogen bonds; the methyl groups generate nonpolar interactions and it is also feasible to form p-type interactions with the aromatic amino acids that are part of the PAS portion of the enzyme 36 .
In our study, the best inhibition of the COX-1 enzyme was given by compound 9. Molecular docking for this ligand-enzyme complex shows that one non-conventional hydrogen bond is formed between the benzoyl group at position C-9 with the Ser 89 residue, p-cation and anion interactions with Arg 467 and Glu 520 residues and several nonpolar interactions of the p and p alkyl type were generated with the residues of Pro 514 and Lys 511 ( Figure 5).
The data in Table 2 show the free binding energies (À7.9 to À3.3 kcal/mol) of the compounds to the enzyme COX-2. The most active compound 8 formed one conventional hydrogen bond between Asn 43 residue and the furoyl group of C-6 position, three non-conventional hydrogen bonds between the furoyl groups at position C-6 and C-9 and the residues of Gln 42, Lys 468 and Leu 472 and several non-polar interactions of the type p and p alkyl with residues Lys 473, Leu 80, Pro 86 and Lys 83.

Displacement of propidium iodide from the peripheral anionic site
The fluorescent assay used to evaluate competitive propidium iodide displacement from the PAS of AChE is commonly used as primary screening method of AChE pro-aggregation activity inhibitors. Propidium iodide is a selective ligand for the PAS of AChE responsible for Ab binding. It exhibits a fluorescence increase upon binding to AChE. A decrease in propidium iodide fluorescence in the presence of test compounds suggests that they are able to displace propidium, consequently bind to the PAS of AChE 37 .
Compounds 3, 4 y 11 were examined for their ability to bind to the PAS of AChE and displace propidium iodide. The three compounds evaluated decreased fluorescence intensity by 12.1% and 13.9% at 290 mM, and by 13.4 and 17.6% at 340 mM (Table 2). Our data suggest that the conjugates are able to bind to the PAS of AChE and therefore might be able to inhibit the aggregation of amyloid peptides induced by AChE. The above is supported by studies demonstrating that ligands binding selectively to the PAS, such as propidium iodide, are capable of blocking Ab aggregation. Therefore, the development of drugs blocking site PAS in AChE and prevent interactions with b-amyloid peptide (and thus decreasing its AChE-induced aggregation) is a promising approach for anti-amyloid AD treatment 38 .

Prediction of absorption, distribution, metabolism, excretion and toxicity (ADMET) and related properties
The pharmacokinetic profile, that is, the absorption, distribution, metabolism and excretion (ADME) of the compounds was predicted with the help of SWISSADME (http://www.swissadme.ch/index.php).
The application was used for in silico prediction of drug similarity of all compounds based on various molecular descriptors and the results are presented in Table 3. Druglikeness plays an effective role in drug design for a new drug (molecules studied) and is generally based on five basic rules (RO5) defined by Lipinski 39 . Lipinski's RO5 is a general rule of thumb for determining drug similarity and determining whether an inhibitor with specific biological and pharmacological characteristics can be an orally active drug in the human body. Furthermore, from the SwissADME website it was possible to obtain properties such as log S (Ali: aqueous solubility), molar refractivity values and inhibitory (þ) and non-inhibitory (-) properties) of CYP450 enzymes such as CYP1A2, CYP2C19, CYP2C9, CYP2D6 and CYP3A4.
Negative BBB values were obtained for natural compounds 8-13 and positive values for synthetic compounds 3-7, which coincides with the best AChE inhibition obtained experimentally. The values of human intestinal absorption that is, the use of molecules in the human body (a crucial barrier in the formulation of new drugs), were high in most of the compounds, both synthetic and natural and only the compounds 8, 11 and 12 showed low absorption. Pan-Assay interference compound (PAINS) index was negative for all the compounds and, as can be seen in Table 3, most of the compounds comply with the five Lipinski rules.

General experimental procedures
NMR spectra, including COSY, NOESY, HMBC, and HSQC experiments, were recorded on a Bruker spectrometer operating at 300 and 400 MHz ( 1 H) and 100 MHz ( 13 C), respectively, using CDCl 3 and DMSO-d 6 as solvents. Tetramethylsilane (TMS) was used as an internal standard. IR spectra (KBr pellets, 500-4000 cm À 1 ) were recorded on a NEXUS 670 FT-IR spectrophotometer. The compounds were analysed by to high-resolution mass spectrometry (HPLC-HRMS/MS) on a maxi G3 quadrupole time-of-flight mass spectrometer (Bruker Daltonics, Bremen, Germany) equipped with an ESI source. The MS was connected to a Shimadzu CBM-20A system with communication module CBM-20A, diode-array detector (DAD) SPD-20A and a Phenomenex Luna C-18 reversed-phase column (250 mm Â 4.6 mm, 5 mm) (Torrance, CA). The 1 H RMS analysis was carried out in positive mode (ESIþ) with a data acquisition range of 2 scans/s at m/z 50-1500. The eluent flow of 200 mL/min was carried to the ion source. Dry nitrogen (N 2 ) was used as nebuliser gas, with a pressure of 2 bar, corresponding to a gas volume flow of 8 L/min. The other ESI conditions: capillary tension of 4500 V, end plate offset of 500 V, drying temperature of 200 C, collision energy of 6.0 eV. Analytical and preparative TLC was performed on Silica Gel 60 Merck plates, and the spots visualised by spraying with a 10% solution of H 2 SO 4 , followed by heating at 110 C. All reagents were purchased from either Merck or Sigma Aldrich (St. Louis, MO, USA) and used without further purification.

Synthesis of the dihydro-b-agarofurans 3 and 4
Agarofuran 3 and 4 (60 and 90% yield respectively) were both prepared in five steps from (þ)-Dihydrocarvone as described by Alarc on et al 16 . Compounds 5 to 7 are reported for the first time both their structure and their activity.

Esterification of 4 to the preparation of compounds 5-7
Compound 4 (1.49 mmol) was dissolved in 250 mL of anhydrous pyridine were heated at 70 C and under stirring with 1.49 mmol of CH 3 COCl and 0.75 mmol of 4-dimethylamino pyridine (DMAP) for 1 h. The resulting mixture was diluted with water, allowed to stand for 30 min, extracted with EtOAc and dried with anhydrous sodium sulphate. Removal of the solvent gave a residue which crystallised from n-hexane in the form of a colourless fine needle of 5 (yield 99%). Compounds 6 and 7 were prepared in the same manner using furoyl chloride (C 5 H 3 ClO 2 ) and benzoyl chloride (C 7 H 5 ClO) respectively (95% and 97% yield respectively). The final purification of all the products for analysis was carried out by recrystallization.

In vitro AChE/BChE inhibitory activity assay
The evaluation of enzyme inhibition was carried out using an adapted version of Ellman's method 21 in the 96-well plates. AChE (from Electrophorus electricus) and BChE (from equine serum), 5,5 0dithio-bis-(2-nitrobenzoic acid (DTNB), acetylthiocholine and butyrylthiocholine iodides were purchased from Sigma-Aldrich. The stock solutions of the test compounds were prepared in 100 mL of DMSO and 900 mL of phosphate buffer (8 mmol/L K 2 HPO 4 , 2.3 mmol/L NaH 2 PO 4 , 150 mmol/L NaCl, and 0.05% Tween 20 at pH 7.6). Enzyme solutions were prepared with buffer to give 0.25 U/mL and 50 mL was added to the plate. After 30 min of incubation, the substrate solution consisting of Na 2 HPO 4 (40 mmol/L), acetylthiocholine/butyrylthiocholine (0.24 mmol/L) and DTNB (0.2 mmol/L) was added. Absorbance of the yellow anion product -due to the spontaneous substrate hydrolysis -was measured at 405 nm for 5 min on a Microliter plate reader (Epoch, Biotek). The enzyme activity was calculated as a percentage compared with a control using only the buffer and enzyme solution. The compounds were assayed in a dilution interval of 500-1.3 mg/mL. Galantamine and carvacrol was used as a positive control.

Kinetic characterisation of AChE inhibition
The kinetic characterisation of AChE was performed using a previously reported method 25 . The enzyme solution was preincubated with different substrate concentrations ranging from 9.38 Â 10 À4 to 0.48 mM. Compound 4, used like as inhibitor was added to the test solution at five different concentrations and the experiment was performed in triplicate. Data analysis was carried out with SigmaPlot v10.0.

COX inhibitory activity
Synthesised compounds and the positive control drugs, Celecoxib, Diclofenac sodium and Indomethacin, were tested for their ability to inhibit COX-1 and COX-2 using COX Inhibitor Screening Assay kit (ACE TM EIA kits, catalog no. 560131). The assay includes both ovine COX-1 and human recombinant COX-2 enzymes. Various concentrations for each of the tested compounds' stock solutions (0.5, 1, 2.5, 5 and 10 mM) were used to obtain the concentration inhibition response curve and calculate the concentration of the test compound causing 50% inhibition (IC 50 , mM).

Ligand Construction and molecular docking studies of compounds with AChE and COX-1/2
All structures of ligands were constructed using Spartan'10 1.1.0 2011 and their geometries were optimised using Density Functional B3LYP 6-31 ÃÃ G ab initio methods in a vacuum. These studies were performed using X-ray crystal structure of T. californica (PDB 1EA5; 1, 8 Å resolution) 40 acetylcholinesterase, the crystal structure of COX-1 enzyme with 2.6 Å resolution complex (PDB code: 3N8Y) and COX-2 enzyme with 1.73 Å resolution complex (PDB code: 3NT1) was obtained from the PDB database https:// www.rcsb.org/. Docking molecular of ligands 3-13 were carried out with the AUTODOCK 4 package using the Lamarckian genetic algorithm 41 . After the preparation of the autodock format of protein and selected compounds, the obtained enzyme structure was used as an input for the AUTOGRID program.
AUTOGRID for each atom type in the ligand performed a precalculated atomic affinity grid map, plus an electrostatics map and a separate desolation map presented in the ligand. The centre of the grid box to AChE was placed in coordinates x ¼ 3.436, y ¼ 63.459, z ¼ 66.849. The dimensions of the active site box were set at 112 Â 126 Â 122 points. All maps were calculated with 0.236 Å spacing between the grid points. The parameters were the following: a grid box was prepared individually for COX-2 to cover the pocket with the main residues for enzyme binding site by maintaining the grid size at 82 Â 70 Â 62 Å in the x-, y-, and zaxis, respectively, and a grid size of x ¼ 80, y ¼ 70, z ¼ 58 Å for the COX-1 enzyme, centred at coordinate x, y, z À33.802, À45.341, À24.413 (for COX-2) and 36.007, À52.132, À1.553 (for COX-1) respectively, with 1 Å spacing between the grid points 42 . Searching parameters were to 50 runs and with 25.000.000 maximum number of evaluations for each ligand. All experiment was making a physiological pH. Root-mean-square deviation (RMSD) between the best redocked conformation and the original conformation of the ligand was then calculated using the PyMOL software. The best pose of each ligand was selected for analysing the interactions between enzyme and the inhibitor. The results were visualised using Discovery Studio 4.0 Client.

Propidium displacement studies
The ability of compounds to competitively displace propidium iodide was evaluated by a fluorescence method 43 . To determine the degree of displacement (% displacement) of propidium iodide from the PAS of AChE, EeAChE (fnal concentration 7 mM) was incubated with the test compound at a concentration of 290 and 340 mM in 1 mM Tris-HCl bufer pH 8.0, 25 C for 15 min. Then, propidium iodide solution (final concentration 8 mM) was added, the samples were incubated for 15 min and the fluorescence spectrum 530 nm (excitation) and 600 nm (emission) was taken. Donepezil was used as reference compound. The blank contained propidium iodide of the same concentration in 1 mM Tris-HCl buffer pH 8.0. The measurements were carried out in triplicate on a microplate reader Perkin Elmer VictorX2 (Perkin Elmer, Singapur).

Conclusion
In conclusion, we either isolate or synthesise a group of molecules with dihydro-b-agarofuran skeleton (epoxieudesmanes) and evaluated their potential as inhibitors of the enzyme acetylcholinesterase, butyrylcholinesterase, COX-1 and COX-2. The structures of the compounds were characterised in the basis of NMR, IR, and MS. Both, synthetic and natural compounds showed inhibitory activity against the acetylcholinesterase enzyme of Torpedo californica with IC 50 values ranging from 17.0 ± 0.016 À 740.0 ± 0.045 mM. Compound 4 has the best inhibitory activity of the enzyme AChE. The compounds evaluated showed selectivity for AChE, with inhibitory activity on the BChE enzyme observed only at high concentrations.
Regarding the COX-1 and COX-2 enzymes, we can say that the compounds act as inhibitors of both isoforms. Compound 9 (IC 50 8.69 mM) showed the best inhibitory activity on COX-1, while compound 8 (IC 50 0.04 mM) showed the best inhibitory activity on COX-2. Natural compounds are better inhibitors than synthetic compounds, in addition the compounds exhibited a preference for the COX-2 isoform.
The study of molecular docking and enzymatic kinetics showed that the sesquiterpenes with dihydro-b-agarofuran skeleton inhibit the AChE enzyme, through a non-competitive mechanism. Moreover, our data suggest that the inhibition is due to different interactions with the peripheral anionic site (PAS). Among the forces involved in these interactions it could be mention Van der Waals attraction and Pi-sigma, Pi-alkyl and hydrogen bonds formation. Binding of the compounds to the AChE PAS together with the type of AChE inhibition suggests their potential to block the aggregation of b-amyloid induced by AChE. In fact, the compounds studied effectively displaced the propidium of AChE PAS.
Molecular docking studies of COX isoforms show that compounds interact with these enzymes through conventional and unconventional hydrogen bonds, stabilising the compounds with the protein, transforming them into a potential source of new anti-inflammatory compounds that could be explored in the future as a cure for the many neurodegenerative diseases mediated by neuroinflammation.
Finally, based on druglikeness studies, ADMET and related properties, it was determined that the evaluated molecules had good pharmacokinetic properties. When the results were evaluated in terms of in silico and in vitro analysis, good agreement was achieved between experimental and theoretical results. In summary, this study provided new small molecules of AChE and COX inhibitors with good therapeutic potential for neurological disorders such as Alzheimer's disease and other causes of dementia.