Meroterpene-Like α-Glucosidase Inhibitors Based on Biomimetic Reactions Starting from β-Caryophyllene

Background: Natural meroterpenes derived from phloroglucinols and β-caryophyllene have shown high inhibitory activity against α-glucosidase or cancer cells, however, the chemical diversity of this type of skeletons in Nature is limited. Methods: To expand the chemical space and explore their inhibitory activities against α-glucosidase (EC 3.2.1.20), we employed β-caryophyllene and some natural moieties (4-hydroxycoumarins, lawsone or syncarpic acid) to synthesize new types of meroterpene-like skeletons. All the products (including side products) were isolated and characterized by NMR, HR-MS, and ECD. Results: In total, 17 products (representing seven scaffolds) were generated through a one-pot procedure. Most products (12 compounds) showed more potential activity (IC50 < 25 μM) than the positive controls (acarbose and genistein, IC50 58.19, and 54.74 μM, respectively). Compound 7 exhibited the most potent inhibition of α-glucosidase (IC50 3.56 μM) in a mixed-type manner. The CD analysis indicated that compound 7 could bind to α-glucosidase and influence the enzyme’s secondary structure. Conclusions: Compound 7 could serve as a new type of template compound to develop α-glucosidase inhibitors. Full investigation of a biomimic reaction can be used as a concise strategy to explore diverse natural-like skeletons and search for novel lead compounds.


Introduction
Natural products (NPs) are a diverse family of organic molecules, most of which exhibit significant bioactivities to treat human diseases [1][2][3][4][5][6][7][8]. Thus NPs are an essential source for discovering novel drugs or lead compounds [6]. However, the range of scaffolds readily accessible in Nature is limited [9]. Furthermore, interest in generating novel synthetic scaffolds has recently declined in pharmaceutical research due to the elaborate isolation procedures required or lengthy total synthesis pathways [9]. Recently, there has been increased interest in NP-inspired or NP-like products to improve the chemical diversity of NPs through different routes: (1) diversity-oriented synthesis (DOS) starting from isolated NPs [10,11], (2) expansion of NP space catalysed by uncommon P450 reactions [12], (3) construction of new lead compounds inspired by bioactive NPs [13], (4) recombined NP moieties generated via coupling reactions [14], (5) diversity-enhanced extraction directly from plants [15]. In particular easily

Synthesis
The synthesis was carried out by a one-pot procedure from coumarin, paraformaldehyde, and β-caryophyllene. In total, 15 compounds corresponding to five different skeletons (Scheme 2) were acquired. Due to the electron donation of O-19a in the coumarin building blocks, the major products were obtained from cycloaddition at the 16 and 19a positions (Scheme 2). The product diversity can be rationalized by the reaction mechanism shown in Scheme 3. The reaction starting from the unsubstituted 4-hydroxycoumarin has already been reported [33,34]. However, the reported procedure did not separate the isomer pairs of 1a/2a and 3a/4a since their chromatographic retention times are very close. We optimized the separation to purify these isomers. In the presence of Cl-or Br-substituents on the 4-hydroxycoumarin moiety, the total conversion increased to 81% or 78%, respectively. Generally, the stable conformers of caryophyllene favored generating (4R,5S)configuration products. However, in the reported total synthesis [35], it was still tricky to obtain enantioselective products. We did not optimize the reaction to promote geneartion of a single target

Synthesis
The synthesis was carried out by a one-pot procedure from coumarin, paraformaldehyde, and β-caryophyllene. In total, 15 compounds corresponding to five different skeletons (Scheme 2) were acquired. Due to the electron donation of O-19a in the coumarin building blocks, the major products were obtained from cycloaddition at the 16 and 19a positions (Scheme 2). The product diversity can be rationalized by the reaction mechanism shown in Scheme 3. The reaction starting from the unsubstituted 4-hydroxycoumarin has already been reported [33,34]. However, the reported procedure did not separate the isomer pairs of 1a/2a and 3a/4a since their chromatographic retention times are very close. We optimized the separation to purify these isomers. In the presence of Cl-or Br-substituents on the 4-hydroxycoumarin moiety, the total conversion increased to 81% or 78%, respectively. Generally, the stable conformers of caryophyllene favored generating (4R,5S)-configuration products. However, in the reported total synthesis [35], it was still tricky to obtain enantioselective products. We did not optimize the reaction to promote geneartion of a single target product since the purpose of the reaction Molecules 2020, 25, 260 3 of 16 in the current work was to acquire diverse skeletons and screen for bioactive products. Thus we explored all the major and side products to get diverse compounds to seek higher bioactivities.
Molecules 2020, 25, x FOR PEER REVIEW 3 of 16 product since the purpose of the reaction in the current work was to acquire diverse skeletons and screen for bioactive products. Thus we explored all the major and side products to get diverse compounds to seek higher bioactivities.

Scheme 3.
A speculated mechanism for the formation of the skeletons 1-5.
To assign the absolute configurations (ACs) of those newly formed chiral carbons (C-4 and C-5) unambiguously, we carried out ECD calculations [36,37] and compared the calculated curves with the experimental ones. The skeletons 1 and 2 possess the same planar structures but inverse chiralities at C-4 and C-5. Their experimental ECD spectra showed an almost mirror curve from 212 to 320 nm. Comparing their experimental ECD spectra with the calculated curves of the (4S,5R) and (4R,5S)isomers allowed us to establish the absolute configurations of 1a and 2a (Figure 1a). The ACs of 3a and 4a were also determined by ECD associated with the calculated spectra ( Figure 1b). The halogen (Cl or Br)-substituted products b and c in the series 1-4 had the same Cotton effect as the Scheme 2. One-pot synthesis of meroterpene-like compounds starting from caryophyllene and 4-hydroxycoumarins. Reagents and conditions: (CH 2 O) n , anhydrous dioxane, reflux, 24 h. product since the purpose of the reaction in the current work was to acquire diverse skeletons and screen for bioactive products. Thus we explored all the major and side products to get diverse compounds to seek higher bioactivities.

Scheme 3.
A speculated mechanism for the formation of the skeletons 1-5.
To assign the absolute configurations (ACs) of those newly formed chiral carbons (C-4 and C-5) unambiguously, we carried out ECD calculations [36,37] and compared the calculated curves with the experimental ones. The skeletons 1 and 2 possess the same planar structures but inverse chiralities at C-4 and C-5. Their experimental ECD spectra showed an almost mirror curve from 212 to 320 nm. Comparing their experimental ECD spectra with the calculated curves of the (4S,5R) and (4R,5S)isomers allowed us to establish the absolute configurations of 1a and 2a (Figure 1a). The ACs of 3a and 4a were also determined by ECD associated with the calculated spectra ( Figure 1b). The halogen (Cl or Br)-substituted products b and c in the series 1-4 had the same Cotton effect as the Scheme 3. A speculated mechanism for the formation of the skeletons 1-5.
To assign the absolute configurations (ACs) of those newly formed chiral carbons (C-4 and C-5) unambiguously, we carried out ECD calculations [36,37] and compared the calculated curves with the experimental ones. The skeletons 1 and 2 possess the same planar structures but inverse chiralities at C-4 and C-5. Their experimental ECD spectra showed an almost mirror curve from 212 to 320 nm. Comparing their experimental ECD spectra with the calculated curves of the (4S,5R) and (4R,5S)-isomers allowed us to establish the absolute configurations of 1a and 2a (Figure 1a). The ACs of 3a and 4a were also determined by ECD associated with the calculated spectra ( Figure 1b   The planar structure of 5a was confirmed by the HMBC correlations, as shown in Figure 3. The chiral C-5 is connected to the chromophore via a flexible CH2 group, thus its configuration cannot be determined from the ECD spectra. However, the chirality of C-1 and C-9 is inherited from βcaryophyllene and remains unchanged during the synthesis procedure. Thus, the NOESY correlation from H-9S to H-5 (as shown in Figure 2) indicated the chiral C-5 was in an (R) configuration. The    The planar structure of 5a was confirmed by the HMBC correlations, as shown in Figure 3. The chiral C-5 is connected to the chromophore via a flexible CH2 group, thus its configuration cannot be determined from the ECD spectra. However, the chirality of C-1 and C-9 is inherited from βcaryophyllene and remains unchanged during the synthesis procedure. Thus, the NOESY correlation from H-9S to H-5 (as shown in Figure 2) indicated the chiral C-5 was in an (R) configuration. The The planar structure of 5a was confirmed by the HMBC correlations, as shown in Figure 3. The chiral C-5 is connected to the chromophore via a flexible CH 2 group, thus its configuration cannot be determined from the ECD spectra. However, the chirality of C-1 and C-9 is inherited from β-caryophyllene and remains unchanged during the synthesis procedure. Thus, the NOESY correlation from H-9S to H-5 (as shown in Figure 2) indicated the chiral C-5 was in an (R) configuration. The corresponding 1 H-and 13 C-NMR data of 5b and 5c when compared with those of 5a indicated that the compounds 5b and 5c possess the same scaffolds and configuration as 5a.
Molecules 2020, 25, x FOR PEER REVIEW 5 of 16 corresponding 1 H-and 13 C-NMR data of 5b and 5c when compared with those of 5a indicated that the compounds 5b and 5c possess the same scaffolds and configuration as 5a. Inspired by the high inhibition against α-glucosidase exhibited by compound 5a (Table 1), we considered that a flexible connection between coumarins and caryophyllene would be beneficial for the bioactivities. To verify our speculation, we attempted to connect 4-hydroxycoumarin and the caryophyllene skeleton via an ether bond. However, according to the GC-EIMS and LC-QTOF-MS/MS analysis of the crude product, none of the expected MS signals were detected (see Figure S1 in Supplementary Materials). Then we used another two similar building blocks (syncarpic acid and lawsone) to replace the 4-hydroxycoumarin to acquire products 6 and 7 (Scheme 4). These two building blocks can also provide hydrogen binding sites. In the presence of ZnI2, the skeleton of caryophyllene rearranged via a transannular reaction to yield a relatively stable carbocation, which was then attached to the hydroxyl to afford compounds 6 and 7.  Inspired by the high inhibition against α-glucosidase exhibited by compound 5a (Table 1), we considered that a flexible connection between coumarins and caryophyllene would be beneficial for the bioactivities. To verify our speculation, we attempted to connect 4-hydroxycoumarin and the caryophyllene skeleton via an ether bond. However, according to the GC-EIMS and LC-QTOF-MS/MS analysis of the crude product, none of the expected MS signals were detected (see Figure S1 in Supplementary Materials). Then we used another two similar building blocks (syncarpic acid and lawsone) to replace the 4-hydroxycoumarin to acquire products 6 and 7 (Scheme 4). These two building blocks can also provide hydrogen binding sites. In the presence of ZnI 2 , the skeleton of caryophyllene rearranged via a transannular reaction to yield a relatively stable carbocation, which was then attached to the hydroxyl to afford compounds 6 and 7.
Molecules 2020, 25, x FOR PEER REVIEW 5 of 16 corresponding 1 H-and 13 C-NMR data of 5b and 5c when compared with those of 5a indicated that the compounds 5b and 5c possess the same scaffolds and configuration as 5a. Inspired by the high inhibition against α-glucosidase exhibited by compound 5a (Table 1), we considered that a flexible connection between coumarins and caryophyllene would be beneficial for the bioactivities. To verify our speculation, we attempted to connect 4-hydroxycoumarin and the caryophyllene skeleton via an ether bond. However, according to the GC-EIMS and LC-QTOF-MS/MS analysis of the crude product, none of the expected MS signals were detected (see Figure S1 in Supplementary Materials). Then we used another two similar building blocks (syncarpic acid and lawsone) to replace the 4-hydroxycoumarin to acquire products 6 and 7 (Scheme 4). These two building blocks can also provide hydrogen binding sites. In the presence of ZnI2, the skeleton of caryophyllene rearranged via a transannular reaction to yield a relatively stable carbocation, which was then attached to the hydroxyl to afford compounds 6 and 7.
The planar structures of 6 and 7 were elucidated according to the 2D NMR correlations, as shown in Figure 3. Since the chirality of C-1(R) and C-9(S) is already known, the newly formed chiral centers (C-4 and C-8) in compounds 6 and 7 were deduced according to the NOESY correlation H-1/H-13 as shown in Figure 3.

α-Glucosidase Inhibition
α-Glucosidase is a key enzyme for hydrolyzing amylum into glucose, which will raise the glucose levels in the blood [38]. Thus inhibiting this enzyme will postpone glucose absorption, thus lowering postprandial blood glucose levels [39]. To screen for lead compounds, we primarily bioassayed all the compounds and building blocks on α-glucosidase at the 25 and 50 µM level. Those compounds (as listed in Table 1) that showed high inhibition (>50% at 25 µM level) were then subjected to further IC 50 evaluation. Compounds 5a, 6, and 7 showed promising inhibition (IC 50 < 10 µM). Interestingly, none of the building blocks, including the coumarins and caryophyllene, showed any inhibitory effects against α-glucosidase. Only when the coumarins were combined with caryophyllene to form meroterpene-like compounds, inhibition was observed. The compound set 1 (1a, 1b, and 1c) all showed weak activities against the bio-target. The compound sets 2-4 showed moderate inhibition against the target protein. The halogen atoms might provide an additional site for hydrogen bonding since compounds 2a, 3a and 4a showed limited inhibition at 25 µM while the compounds 2b and 3b (Cl substituted) and 2c and 3c (Br substituted) all showed moderate activities (IC 50 12.0-23.6 µM).
The compound set 5 showed better inhibition than compounds 2-4. These results indicated that a flexible connection between β-caryophyllene and 4-hydroxycoumarin is beneficial to improve the inhibition. Furthermore, the hydroxyl group could be necessary for binding to the target protein.
When the rearranged caryophyllene was linked with aromatic ketones through an oxygen bridge (compounds 6 and 7), the product showed better inhibition than the positive control. Among our synthetic products, compound 7 showed the highest inhibition, probably due to the presence of the para-substituted carbonyls and the hydrophobic caryophyllene moiety.
To explore the interaction mechanism of the best inhibitor 7, we carried out the enzyme kinetic study using Lineweaver-Burk plots analysis [40,41]. α-Glucosidase was treated with pNPG and compound 7 at various concentrations. As shown in Figure 4, compound 7 showed a non-competitive type of inhibition against α-glucosidase. Replotting the slope and Y-intercept values taken from each line in the primary Lineweaver-Burk plot allowed extrapolation of the inhibition constants K i,free (the affinity to the free enzyme) and K i,bound (the affinity to the complex enzyme-substrate). The K i,free and K i,bound values of 7 were 2.03 and 5.57 µM, respectively. CD spectroscopy is considered as a reliable and sensitive method for monitoring the secondary structural changes of macromolecules interacting with small molecules [42]. The α-glucosidase yield two negative bands at 209 and 219 nm on CD the spectrum which originated from the n π* and π ⟶ CD spectroscopy is considered as a reliable and sensitive method for monitoring the secondary structural changes of macromolecules interacting with small molecules [42]. The α-glucosidase yield two negative bands at 209 and 219 nm on CD the spectrum which originated from the n π* and π → π* electron transfer of the α-helix motifs [43,44]. With increasing levels of compound 7 to the enzyme, a regular increasing intensity of both negative bands occurred (as shown in Figure 5). These CD changes indicated that the secondary structure of the target protein was influenced by the small molecule 7. CD spectroscopy is considered as a reliable and sensitive method for monitoring the secondary structural changes of macromolecules interacting with small molecules [42]. The α-glucosidase yield two negative bands at 209 and 219 nm on CD the spectrum which originated from the n π* and π ⟶ π* electron transfer of the α-helix motifs [43,44]. With increasing levels of compound 7 to the enzyme, a regular increasing intensity of both negative bands occurred (as shown in Figure 5). These CD changes indicated that the secondary structure of the target protein was influenced by the small molecule 7.

Products Starting from 4-Hydroxycoumarin
To a 50 mL reaction tube was added 4-hydroxycoumarin (170 mg, 1.049 mmol) and paraformaldehyde (252 mg). The reaction tube was then flushed with argon and evacuated three times. Anhydrous 1,4-dioxane (5 mL) was added by syringe, and the mixture was stirred under reflux for 1 h. Then anhydrous 1,4-dioxane (2 mL) and β-caryophyllene (630 µL, 3.150 mmol) were injected into the reaction tube. The resulting solution was stirred for 24 h at reflux under an argon atmosphere, after which complete conversion of the starting material was confirmed by TLC inspection. After cooling to rt, the reaction mixture was washed with 10 mL of saturated Na 2 CO 3 . The organic layer was collected and removed the solvent under vacuum to yield crude product (869 mg). The crude products were separated on silica gel column (EtOAc in petroleum ether gradient, from 25:1 to 5:1, v/v) to 5 fractions (Fr. A~E). These fractions were further purified by HPLC using the Dr. Maisch GmbH semi-preparative column, eluting with 100% MeCN, to yield compounds 1a (110 mg, 28%), 2a (48 mg, 12%), 3a (31 mg, 8%), 4a (10 mg, 3%) and 5a (7 mg, 2%).

Products Starting from 6-Bromo-4-Hydroxycoumarin
The procedure was the same as the above reaction.
After stirring for 18 h at 110 • C, the mixture was concentrated. To the residue was added 5 mL water and the mixture was then extracted with EtOAc (10 mL) three times. The organic layers were combined, dried over Na 2 SO 4 and the solvent removed under reduced pressure to yield the crude product, which was then separated on a silica gel column (petroleum ether-EtOAc 50:1) to afford oily compound 6 (38 mg, 36%).

ECD Calculation Method
The calculation was performed as our previously reported procedure [45,46]. All the conformers of every calculated compound were searched by Conflex using the MMFF94s force field [47,48]. Further optimization was performed at B3LYP/6-31 + G(d,p) level in Gaussian 09 package [49]. The theoretical CD spectra were calculated by cam-B3LYP/TZVP and added in SpecDis [50] according to their Boltzmann-calculated distributions.

α-Glucosidase Inhibitory Assay
The α-glucosidase inhibitory assay of the synthesized compounds was performed as previously reported procedure [45]. α-Glucosidase (EC 3.2.1.20) isolated from Saccharomyces cerevisiae was purchased from Sigma (St. Louis, MO, USA). The enzyme was dissolved in 200 µL of 10 mM phosphate buffer (pH 6.80) and incubated with 12 µL of the test compound in DMSO at 37 • C for 5 min. Then the enzymatic reaction was started by the addition of 36 µL of 4-nitrophenyl α-D-glucopyranoside (pNPG) and kept under 37 • C for 40 min. The amount of released 4-nitrophenol was determined according to the absorbance at 400 nm. The primary screening was carried out at two concentrations (25 and 50 µM). While the IC 50 assay was performed with five different concentrations around the IC 50 values. In each set of experiments, the assay was performed in triplicate. The percentage inhibition of α-glucosidase activity was calculated via the following formula: Inhibition ratio (%) = 100 × (A control − A sample )/A control . The IC 50 values were calculated in Prism 7 using a nonlinear regression method with the normalized response and variable slope.

Kinetic Analysis of α-Glucosidase Inhibition
The kinetic parameters of α-glucosidase inhibition by compound were evaluated by the Lineweaver-Burk plots and its secondary plots. The double-reciprocal plots were constructed with enzyme reaction initial velocity (V) versus substrate (S) concentration (1/v vs. 1/[S]) in the absence (control) or presence of compound 9 at different levels (0-8 µM). The initial rate was measured by stopping the reaction after 2 min. The type of inhibition, K m, and V max values were determined from the plots. Slopes and Y-intercepts of these reciprocal plots were also replotted against the inhibitor concentration, respectively. Data analysis was performed by the Prism software.

Circular Dichroism Measurement of Inhibitor-Enzyme Complex
CD spectra (190-250 nm) of α-glucosidase with and without compound 7 were recorded on a Chirascan CD spectrometer at room temperature. All the CD spectra were corrected with the buffer signal under constant nitrogen flush. The concentration of α-glucosidase was 2.0 µM, and the molar ratios of compound 7 to α-glucosidase were set to 0:1, 1:1, and 2:1.

Conclusions
In summary, we use a biomimetic hetero-Diels-Alder reaction as a tool for generating NP-like skeletons. Although the yields and stereoselectivities of these reactions were very limited, the product diversity provided a chance to explore new NP-like scaffolds. These reactions starting from 4-hydroxycoumarins and β-caryophyllene could provide a facile way to acquire meroterpene-like skeletons. All 17 products, representing seven different scaffolds, provided candidates to identify potentially new chemotypes. Due to the limited stereoselectivity of the construction reaction, all the ACs of products were determined unambiguously by ECD calculations. Inspired by the inhibition of compound 5a, we furtherly linked β-caryophyllene and lawsone (or syncarpic acid) via an ether connection to yield compound 6 and 7. Compounds 5a, 6, and 7 showed interesting potential activities (IC 50 < 10 µM). They can be considered novel scaffolds for the design of α-glucosidase inhibitors. Furthermore, side products from biomimetic reactions can provide diverse NP-like scaffolds, and will favor searching for lead candidates for drug discovery. Since a hydrogen bonding site is essential for inhibition, we will further explore the chemical space in the future based on natural polyphenol moieties. Also, the inhibition mechanism should be further investigated based on a more promising molecule.