Inhibition of catechol-O-methyltransferase by natural pentacyclic triterpenes: structure–activity relationships and kinetic mechanism

Abstract Inhibitors of COMT are clinically used for the treatment of Parkinson’s disease. Here, we report the first natural pentacyclic triterpenoid-type COMT inhibitors and their structure-activity relationships and inhibition mechanism. The most potent compounds were found to be oleanic acid, betulinic acid and celastrol with IC50 values of 3.89–5.07 μM, that acted as mixed (uncompetitive plus non-competitive) inhibitors of COMT, representing a new skeleton of COMT inhibitor. Molecular docking suggested that they can specifically recognise and bind with the unique hydrophobic residues surrounding the catechol pocket. Furthermore, oleanic acid and betulinic acid proved to be less disruptive of mitochondrial membrane potential (MMP) compared to tolcapone, thus reducing the risk of liver toxicity. These findings could be used to produce an ideal lead compound and to guide synthetic efforts in generating related derivatives for further preclinical testing.


Introduction
Catechol-O-methyltransferase (COMT, C.E. 2.1.6) is a bisubstrate enzyme, that catalyses methyl transfer from S-adenosyl-L-methionine (SAM) to one of the hydroxyls of catecholamine neurotransmitters dopamine, epinephrine and norepinephrine, and catechol oestrogens, resulting in termination of their biological activity 1,2 . The catalytic site consisted of SAM and catechol binding pockets connected by a narrow channel through which methyl transfer occurs, where the SAM site proved to be deeply embedded 1 . COMT followed an ordered reaction mechanism where SAM bound first to the enzyme and the catechol substrate followed by release of the products in the reverse order 3 . COMT is a therapeutic target for the treatment of various peripheral cancers [4][5][6] and central system disorders [7][8][9][10] . In particular, inhibition of peripheral COMT offers a unique advantage, since COMT inhibitors are clinically used as adjunct to levodopa (L-dopa) for the treatment of Parkinson's disease 7 .
Marketed COMT inhibitors are nitrocatechols including tolcapone 11 , entacapone 12 and opicapone 13 . These three drugs are coadministered with L-dopa, a precursor of dopamine, to increase its half-life and improve symptoms resulting from dopamine level fluctuation 14,15 . However, tolcapone was proposed to cause liver toxicity and thus required clinical liver function monitoring 15 .
Although not related to idiosyncratic toxicity, entacapone can cause severe diarrhoea, increased dyskinesia frequency and has low bioavailability 16 . Therefore, the discovery of superior COMT inhibitors with better safety profiles and clinical efficacy is urgently needed.
To improve the complications of nitrocatechol drugs, researchers seek to discovery non-nitrocatechol COMT inhibitors from natural products, which may have the potential to improve toxicity profiles versus tolcapone and entacapone. There have been limited reports of natural COMT inhibitors. Notably, epigallocatechin and epicatechin (the major polyphenol in green tea) 17 , and chlorogenic acid and caffeic acid existent in coffee 18 demonstrated COMT inhibition with the IC 50 values of 6.3-60 lM. Some alkaloids and bioflavonoids (e.g. quercetin and fisetin) also showed inhibitory potency on COMT 19,20 . Unfortunately, the high polarity, acidity or alkalinity of these compounds makes it difficult to optimise them for use as drugs.
Pentacyclic triterpenes including four classes of oleananes, ursanes, lupanes and friedelanes are an excellent reservoir of biologically active compounds 21 . In the present study, we discovered natural COMT inhibitors with characterisation of a chemical scaffold of pentacyclic triterpene, which was distinct from previously reported compounds. An enzymatic activity based fluorescence assay was developed to evaluate COMT inhibitors with 3-BTD or a COMT-specific fluorescent probe as the substrate. That produced three potent inhibitors oleanic acid, betulinic acid and celastrol with IC 50 values of 4.74, 5.07 and 3.89 lM, respectively. It seemed certain that introducing hydroxyl groups into the ring A of oleanic acid was not beneficial to COMT inhibition, while the C-17 site carboxyl of betulinic acid and the conjugated system between ring A and B of celastrol were necessary. These compounds inhibited COMT-catalysed O-methylation of 3-BTD in a mixed (uncompetitive and non-competitive) modality. Furthermore, docking simulation indicated that they can specifically bind with the unique hydrophobic residues (Met-40, Leu-198, Trp-143, Trp-38 and Pro-174) as a "gatekeeper" to guard the catechol site of COMT. Finally, oleanic acid and betulinic acid showed less potent toxicity on the normal human liver-derived cell line LO-2 and on mitochondrial membrane potential (MMP) compared to tolcapone, mitigating the risk of hepatotoxicity. These pentacyclic triterpenes represented a novel type of COMT inhibitors with improved safety profile, which might be further optimised to obtain related derivatives for preclinical testing.

COMT activity inhibition assay
The measurement of COMT activity was carried out as described with modification 22  The tested compound of 20 mM stock in DMSO was used to prepare 10-point 2-fold dilution series. After pre-incubation 37 C for 3 min, the reaction was initiated upon the addition of 10 lL SAM and progressed at 37 C for 6 min. The incubation was terminated with 200 lL ice-cold ACN containing 1% formic acid. After centrifugation at 20,000 Âg for 5 min at 4 C, 200 lL of supernatant was plated into assay wells (Costar V R assay plate, 96 well, black, flat bottom black, polystyrene plate, ref. #3925 from Corning Inc.). The fluorescent intensity were measured using the multi-mode microplate reader (SpectraMaxs iD3 V R , Molecular Devices, Austria). An additional assay include a DMSO treatment which served as a maximum enzyme activity control. The excitation/emission wavelengths set at 390/510 nm. Residual COMT activity can be calculated using the formula: (COMT inhibitor signal/DMSO signal) Â 100%. To evaluate inhibition kinetics, the concentrations of 3-BTD substrate ranged from 0.01 to 5 lM in the presence of oleanic acid, celastrol and betulinic acid with the concentrations of 2.5, 5 and 7.5 lM. The fluorescence signal was detected as detailed above except the rate of Omethylation of 3-BTD which was expressed as nanomoles of methylated product formed per min per milligram of recombinant human S-COMT protein (nmol/min/mg protein).

Cytotoxicity assay
Cytotoxicity assays of oleanic acid, celastrol, betulinic acid, tolcapone and entacapone were performed using standard CCK8 kit. The normal human liver-derived cell line LO-2 were grown in DMEM culture medium supplemented with 10% FBS, penicillin (100 U/mL) and streptomycin (100 mg/mL). After culturing to 80% confluence, cells were digested and exactly 100 mL of cell suspension with a cell concentration of 10 4 cells/well was seeded in 96well plates and incubated in a 5% CO 2 humidified atmosphere at 37 C for 12 h. Then the cells were incubated with 100 lL tested compounds of varying concentrations from 0.25 to 100 lM prepared in FBS-free culture medium for 24 h. Subsequently, CCK-8 was diluted 1:10 with DMEM culture and them was added into the adherent cells and incubated in 37 C for 2 h. The absorbance at 450 nm was measured by using the multi-mode microplate reader. Cell viability can be calculated by A/A 0 Â 100% (A and A 0 are the absorbance of experimental group and control group, respectively).

Mitochondrial membrane potential assay
Toxic effects of oleanic acid, betulinic acid and tolcapone on mitochondrial membrane potential of LO-2 cells were examined. The cells were cultured as described in the Section 2.3. Then the cells were incubated with 100 lL tested compounds of various concentrations (from 0.25 to 25 lM prepared in FBS-free culture medium) for 24 h. To assess the mitochondria membrane potential, JC-10 (5 lg/mL) was incubated with the cells for 30 min, and Hochest (10 lg/mL) was used to stain nuclei. After removal of culture medium and rinsing three times with PBS, cell imaging and quantitative analysis were performed by high-content imaging analysis (Molecular Devices V R ImageXpress Micro 4, the USA).

Molecular docking
Molecular docking was carried out to determine geometrically and energetically stable conformation upon the binding of compounds to COMT. The protein structure of human S-COMT was obtained from a Protein Data Bank (PDB ID: 3BWM). All docking experiments were conducted through AutoDock Vina. To process the receptor file (.pdb) for docking, the following procedures were implemented: water molecules and irrelevant heteroatoms were removed, hydrogen atoms were added and non-polar hydrogens were subsequently merged, ultimately, charges were added using the Kollman method. The ligand compounds depicted in Chem3D were imported into ADT accompanied with the Torsion Tree root detected and were subsequently saved as .pdbqt files. To further verify its inhibition mechanism, the searching grid box was sufficiently large to wrap the whole protein. The conformational search space was generated with a spacing of 0.375 Å and dimensions of (126 Â 122 Â 106) points (.gpf file) with the grid centre XYZ coordinates at À9.82, À5.60 and À9.83, respectively. The highest ranked and lowest energy docking pose were further analysed by Discovery Studio visualiser version.

Data analyses
The kinetic analysis were performed by fitting the initial velocity as a function of concentration to the following equations using the non-linear regression analysis program in GraphPad Prism 5.0 software: The [S] and [I] represent the concentrations of substrate and inhibitor, respectively; V max is the maximal velocity; K m is the Michaelis constant; K i is the inhibition constant suggesting the dissociation of the enzyme-inhibitor complex (EI); aK i is the inhibition constant when the inhibitor binds to an enzyme-substrate complex, suggesting the dissociation constant of the enzyme-substrate-inhibitor complex (ESI); the value of a equals 1, meaning the pure non-competitive inhibition; at a ) 1, meaning competitive inhibition; at a ( 1, meaning uncompetitive inhibition. The reciprocals of velocity and substrate concentrations gave the linear correlation by which K m and V max values can be calculated. Obtained data were presented as mean ± standard error (± SE) of three independent experiments with duplicate determinations for each assay. Statistical difference in IC 50 values were determined using an unpaired two-tailed t-test, which indicated as Ã p < .05, ÃÃ p < .01, ÃÃÃ p < .001.
Of the oleanane-type triterpenes, oleanolic acid (1) displayed inhibitory potency on COMT activity in a concentration-dependent manner [the blue curve in Figure 2(A)]. However, with the introduction of the hydroxyl group into ring A (seeing Figure 1 for ring labeling), the COMT inhibitory potency decreased such as hederagenin (2) and maslinic acid (3). Polygalacic acid (4) with more hydroxyl groups almost did not inhibit COMT activity and the IC 50 value was not calculated due to insufficient concentrationresponse relationship. Thus, the decrease of lipophilicity of ring A did not enhance COMT inhibition. In spite of minor difference in the ring E substitution position of C-29 and C-30 methyl groups, ursolic acid (7, IC 50 ¼ 15.13 ± 1.35 lM) displayed $3-fold lower inhibitory potency than oleanolic acid (1). Introducing hydroxyl into ring A of ursolic acid (7) can weaken COMT inhibition, as it did in the case of asiatic acid (8, IC 50 ¼ 43.11 ± 5.92 lM). Addition of a carbonyl to the C-11 position, such as 11-keto-b-boswellic acid (10), set up a complete loss of inhibitory potency.
Among these lupane-type triterpenes, betulinic acid (11) had strong inhibitory potency with the IC 50 value of 5.07 ± 0.087 lM [the orange curve in Figure 2(C)], nearly equal to oleanolic acid (1, IC 50 ¼4.74 ± 0.29 lM). With addition of one hydroxyl group to ring A, 23-hydroxybetulinic acid (16) showed a reduced COMT-inhibitory effect. Compounds 12-15 also proved to be less potency versus betulinic acid (11), indicating that the converting of (S)-C-17 carboxyl group to methyl (12), hydroxymethyl (13), carbonyl (14) and (R)-C-17 carboxyl group (15) can lead to a slight decrease. Compared to betulinic acid (11), epibetulinic acid (15) showed a 50% reduction in COMT-inhibitory potential, despite having a little difference in the configuration of the C-17 substituent. Of the friedlane-type molecules, celastrol (18, IC 50 ¼3.89 ± 0.15 lM) showed a stronger COMT-inhibitory effect [the green curve in Figure 2(D)]. This might be attributed to the existence of a conjugation system between rings A and B, increasing the lipophilicity of ring A.
Collectively, the primary structure-activity relationship (SAR) is summarised in Figure 3: (1) introducing hydroxyl groups into ring A weakened COMT inhibition, while increasing the conjugated system helped enhance the lipophilicity of ring A, which was beneficial for COMT inhibition; (2) the existence of a carbonyl group at the C-11 position resulted in a complete loss of inhibitory potency; (3) transferring one of the methyl groups at C-20 into C-19 can cause a reduction in COMT-inhibitory potency; (4) the C-17 carboxyl group was necessary, and the COMT inhibitory potential of S-configuration proved to be better than that of R-configuration. It was expected to guide us rational design and synthesis of novel COMT inhibitors of pentacyclic triterpene analogues based on the mentioned results. Figure 4(A) illustrated the inhibition mechanism of pentacyclic triterpenes for the catechol-binding site of COMT. Here, the probe 3-BTD, serving as a typical tool molecule to occupy the catechol pocket, can be O-methylated to 3-BTMD by COMT 22 . Instead of competing with 3-BTD, the inhibitor molecule can bind free COMT by recognising the unique hydrophobic residues surrounding the catechol site, and thus could act as a non-competitive inhibitior. In addition, the COMT-substrate complex bind with the inhibitor molecule which could hamper the release of the O-methylation product of 3-BTD. This was because the hydrophobic residues not only defined the catechol active site, but also served a "gatekeeper" to guard bind and release of molecules 14 . Thus, it was possible for the pentacyclic triterpene to inhibit the catechol site of COMT in a mixed (non-competitive plus uncompetitive) mode.

Inhibition kinetics
We monitored the kinetics in the steady-state as a function of 3-BTD in the presence of compounds 1, 11 and 18 with saturated SAM (Supplementary Figure S3). The O-methylation rates of 3-BTD were plotted as Lineweaver-Burk diagrams [ Figure 4(B-D)]. Obviously, the plot lines of Lineweaver-Burk were intersected in the quadrant III, suggesting that compounds 1, 11 and 18 inhibited the catechol site in a non-competitive and uncompetitive manner. Furthermore, the K m and V max values for 3-BTD were determined to be 0.38 ± 0.029 lM and 13.8 ± 0.30 nmol/min/mg protein, respectively. With the addition of the inhibitors, the corresponding apparent K m and V max values significantly decreased. Thus, compounds 1, 11 and 18 displayed a mixed (non-competitive plus uncompetitive) inhibition pattern with respect to the catechol active site of COMT. The global fit of the inhibition mode to these data yielded the K i values for the three ones in the range of 3.5-5.6 lM. Table 2 showed the kinetic parameters in detail.

Molecular docking
To further explore the binding interactions of compounds 1, 11 and 18, a three-dimensional model of human S-COMT (PDB: 3BWM) was constructed. The entirely hydrophobic residues (Met-40, Leu-98, Trp-143, Trp-38 and Pro-174) as "gatekeepers" surrounding the catalytic pocket of COMT were unique 14,23,24 . This allowed the specific recognition of some small molecules to COMT with hydrophobic interaction force. The three compounds covered the COMT active cave quite well and did not overlap with the catechol-binding pocket (Supplementary Figure S4). For compound 1, the six-membered rings and some small hydrophobic substitutes can initiate alkyl and pi-alkyl interactions with Trp-38, Met-40, Pro-174, Trp-143 and the aliphatic portion of Lys-144 [seeing Obviously, the three molecules can be "trapped" by these "gatekeeper" residues around the active pocket, indicating binding with free COMT or COMT-substrate complex instead of binding the catechol-or SAM-pocket. What is more, the calculated binding energy (DE binding ) values varied from À7.6 to À6.2 (Supplementary  Table S1), revealing the following order of binding affinity to human S-COMT: compounds 18 > 1 > 11. This is partially due to p-p interaction between ring A of compound 18 and the indole ring of Trp-143, which tightened its contact surface and the higher binding affinity with COMT. This was consistent with the biochemical data in which compound 18 had a lower K i value for inhibiting the O-methylation of 3-BTD.

Assessment of mitochondria membrane potential
To investigate the influence of the novel pentacyclic triterpenetype COMT inhibitors on hepatotoxicity, we examined the effects  (Table 1).  (18) was determined using an unpaired two-tailed t-test as indicated by Ã p < 0.05, ÃÃ p < 0.01 and ÃÃÃ p <0 .001. N.E. is the abbreviation of no effect. All the data were the mean ± standard error (± SE) of three independent experiments with duplicate determinations.
Introducing hydroxyl into the A ring can weaken COMT inhibition, while increasing the conjugation of the ring A helped to enhance inhibitory effect.

Introducing carbonyl group to C-11 position can result in a complete loss of COMT-inhibitory effect.
The existence of carboxyl group at C-17 proved to be necessary, and COMT-inhibitory potency of S-configuration was better than that of R-configuration.
Transferring one of the methyl groups at C-20 into C-19 can cause a reduction in COMT-inhibitory potency.     of these compounds on mitochondria membrane potential (MMP) using the normal human liver-derived cell line LO-2, along with tolcapone as a positive control. The dye JC-10 underwent a change in fluorescence emission from red to green when the MMP was disrupted and uncoupled 25 . For tolcapone, strong green signals were observed inside the cells with increasing concentrations [ Figure 6(A), bottom row], indicating a decrease of MMP. In contrast, a strong red fluorescence appeared in the mitochondria with increasing concentrations of compounds 1 or 11 [ Figure 6(A), top and middle rows]. This observation demonstrate that the two compounds tested had the lower mitochondria toxicity than tolcapone. In Figure 6(B), tolcapone at 10 lM led to a 50% reduction in membrane potential, yet compound 1 and 11 even at 50 lM did not display a decrease of MMP. In pilot studies, the toxic effects of compounds 1, 11 and 18 on LO-2 cells were tested in a CCK8 assay. Supplementary Figure S5 demonstrated that 2.4 lM of compound 18 led to a 50% reduction of cell viability, which corresponded to a 12-fold more potent cytotoxicity than tolcapone. However, compounds 1 and 11 up to 100 lM displayed no significant toxicity with 95% cell survival, while entacapone without liver toxicity served as a negative control. The cell viability and mitochondria membrane potential is an indicator of cell viability 26 . Therefore, these data supported the idea that these novel nonnitrocatechol COMT inhibitors had the potential of being developed with a reduced risk of liver toxicity.

Conclusion
In summary, our research led to the discovery of novel COMT inhibitors with a chemical scaffold of pentacyclic triterpene, which is distinct from previously reported ones. Using a combination of the inhibition kinetic assay and molecular docking, we tried and explained the inhibition mechanism and the binding site of the pentacyclic triterpene to COMT. Furthermore, oleanic acid and betulinic acid displayed potent COMT inhibition and significantly less toxicity on the mitochondria membrane potential (MMP) of a human normal liver cell line, serving as an ideal lead compound to develop pentacyclic triterpene-type COMT inhibitors. It was possible that these novel COMT inhibitors can provide a starting point for synthetic efforts to generating related derivatives for further preclinical testing and new drugs used for the treatment of Parkinson's disease, as adjuncts in L-dopa based therapy, or for the treatment of schizophrenia.