Phytochemical Screening and Isolation of New Ent-Clerodane Diterpenoids from Croton guatemalensis Lotsy

Phytochemical screening of an ethanol–water extract (EWE) from the bark of Croton guatemalensis led to the isolation and identification of eight compounds, among them: five ent-clerodane diterpenoids [junceic acid (1), 6(s)-acetoxy-15,16-diepoxy-ent-cleroda-3,13(16),14-trien-20-oic acid (crotoguatenoic acid A) (2), 6(s)-hydroxyoxy-15,16-diepoxy-ent-cleroda-3,13(16),14-trien-20-oic acid (crotoguatenoic acid B) (3), formosin F (4), bartsiifolic acid (5)], and three flavonoids [rutin (6), epicatechin (7), and quercetin (8)]. Of these, 2 and 3 are reported here for the first time. Structures were established through conventional spectroscopy methods and their absolute configurations were determined by optical rotation and comparison of experimental electronic circular dichroism (ECD) and theoretical calculated ECD spectra. A suitable high performance liquid chromatography (HPLC) method for quantifying rutin (6) was developed and validated according to standard protocols. Affinity-directed fractionation was used to identify possible in vitro active compounds on α-glucosidases from Saccharomyces cerevisiae. HPLC-ESI-MS was used to identify the inhibitors as free ligands after being released from the enzymatic complex by denaturing acidic conditions. The affinity studies led to the identification of ent-clerodane diterpenoids as active compounds. In silico analysis allowed us to determine the best conformational rearrangement for the α-glucosidase inhibitors.


Introduction
About 1300 species of the Croton genus (Euphorbiaceae) were reported, which are distributed in tropical climate regions around the world [1]. Extracts of different parts of the plant (aerial parts, roots, leaves, bark, etc.) are used for the treatment of various ailments, such as stomachache, abscesses, inflammation, cancer, diabetes, and malaria in the Americas, Africa, and South Asia [1][2][3]; for example, in an in vivo assay of the ethanolic extract of the aerial parts of C. zambesicus was performed to determine antiplasmodial activity against chloroquine-sensitive Plasmodium berghei infections in mice [4], crude leaf extracts of Croton cajucara exhibited a significant antinociceptive effect in rats. The cortex bark is one of the most used parts of pharmacological interest and was studied as an analgesic, anti-inflammatory, antiulcerogenic, gastroprotective, antiviral, antibacterial, antitumor, and hypoglycemic agent [2,3,5]. Many compounds were isolated and identified, among the most important are terpenes, and within them the diterpenes with different types of skeletons are the predominant group, with more than 800 registered compounds, including clerodane, tigliane, kaurane, crotofolane, labdane, cembrane, abietane, casbane, halimane, pimarane, cleistanthane, grayanane, atisane, phytane, and laevinane diterpenoids; of which clerodanes are the most abundant [1,2]. Some alkaloids and phenolic compounds were also laevinane diterpenoids; of which clerodanes are the most abundant [1,2]. Some alkaloids and phenolic compounds were also isolated. Recent interest in searching for flavonoids in this genus led to the identification of proanthocyanidins, flavones, glycosylated flavonols, and lignans; among the most common are rutin, quercetin, kaempferol, catechin, and epicatechin [6][7][8][9], which could serve as genus chemical markers. Finally, the essential oils of various Croton species present α-pinene, ß-pinene, camphor, 1,8-cineole, and germacrenes among their most abundant compounds [10][11][12][13][14].
Croton guatemalensis Lotsy (Cg) is a small tree, up to 6 m high, distributed in the tropical and subtropical areas of the Americas, including Mexico, Colombia, Ecuador, and Guatemala; it is also known as "copalchi". Aqueous and methanolic extracts of leaves and cortex were reported as antiplasmodial and cytotoxic [15]; whereas the aqueous extract of the bark is antinociceptive [16]. Our group proved that the aqueous and hydroalcoholic extracts have hypoglycemic activity [17]; so far, no compounds isolated from the plant are reported. In this research, the EWE, such as the one used in the previous work, from the bark of C. guatemalensis was fractionated to isolate the main compounds to provide information about its chemical profile; furthermore, with the aim to understand the ecological role of some of the isolated compounds affinity studies for the identification of new αglucosidase inhibitors were performed.

HPLC Phytochemical Profiling
HPLC profile of the EWE of C. guatemalensis was monitored at different wavelengths and the peak heights were evaluated. Maximum peak heights for the extract were ob-

HPLC Phytochemical Profiling
HPLC profile of the EWE of C. guatemalensis was monitored at different wavelengths and the peak heights were evaluated. Maximum peak heights for the extract were obtained at 205, 240, and 254 nm ( Figure 5) and were selected as optimum wavelengths to analyze the chromatographic profile. The UV spectra of the peaks showed characteristic bands of flavonoids with features of flavans, flavonols, and terpenes, with their maximum absorptions at 200, 266-280 nm (Band II) for flavans [23] in the first 13 min of the profile; then, maximum absorptions at 230-254 (Band II) and 330-370 nm (Band I) for flavonols [23] between 14 and 20 min, and, finally, absorption maxima of 205, 218, and 240 nm for terpenes [24,25]   Diterpenes are characteristic components of the Croton species and clerodane diterpenes skeletons are the most abundant, being part of 27% of the diterpenes found in Croton species [1]. Junceic acid (1) was first isolated from Solidago juncea Ait [18] and previously identified as a major compound in Croton sarcopetalus [26] and Croton arboreus [27], and it was tested as an anti-inflammatory [28] and phytotoxic [20] agent. Formosin F (4) was previously isolated from Excoecaria formosana and analyzed as an antibacterial compound that showed moderate antibacterial activity against two strains of Helicobacter Diterpenes are characteristic components of the Croton species and clerodane diterpenes skeletons are the most abundant, being part of 27% of the diterpenes found in Croton species [1]. Junceic acid (1) was first isolated from Solidago juncea Ait [18] and previ- ously identified as a major compound in Croton sarcopetalus [26] and Croton arboreus [27], and it was tested as an anti-inflammatory [28] and phytotoxic [20] agent. Formosin F (4) was previously isolated from Excoecaria formosana and analyzed as an antibacterial compound that showed moderate antibacterial activity against two strains of Helicobacter pylori [19]. Bartsiifolic acid (5) was previously isolated from Blakiella bartsiifolia [20] and E. formosana [19]. It was studied as phytotoxic and antimicrobial. In terms of its phytotoxic activity, it restrained seed germination at low concentration and hindered elongation of the shoots [20]; its antibacterial activity was not proven. Some flavonoids were isolated from various Croton species [1]; among these are flavans, flavonol aglycones, flavonol glycosides, flavones, etc. Rutin (6) was first described in Croton menthodorus [28], subsequently in Croton caudatus [29], Croton sphaerogynus [30], Croton polycarpus [31], Croton campestris [6], and finally in C. urucurana [9]. This indicates a constant presence of the flavonoid in the genus; therefore, 6 may be useful as a possible phytochemical marker. In addition, this flavonoid meets several requirements to be a chemical marker [32] and its effectiveness as a hypoglycemic agent was demonstrated in several studies [33,34]. For this reason, the rutin (6) quantification method in the EWE of C. guatemalensis was validated. Epicatechin (7) was previously described in Croton lechleri [35] and C. urucurana [36]. This flavanoid was extensively studied as an anti-inflammatory, antioxidant, anti-cancer agent, and as preventing diabetes, cardiovascular diseases, a neuroprotector, and enhancer of muscle performance [37]. Recently, the combination of 7 with rutin (6) (75:25) was tested in the oral administration of alloxan-induced hyperglycemic mice for 28 days, and its chronic hypoglycemic activity yielded similar results to glibenclamide [38]. Quercetin (8) was previously isolated from Croton sylvaticus and proved to be a potent inhibitor of acetylcholinesterase [39]. Quercetin (8) was also identified and, in some cases, quantified in C. sphaerogynus [30], C. polycarpus [31], and C. urucurana [9]. This flavonol was extensively studied as an antioxidant, antimicrobial, anti-Alzheimer's, antiarthritic, anticarcinogenic, and hypoglycemic agent [40]. As far as we know, our current work is the first report of isolation of compounds 1-8 from C. guatemalensis, and 2 and 3 are new for the genus.

Quantification of Rutin (6) in C. guatemalensis Extract
A comprehensive HPLC method was developed and validated for quantifying rutin (6) according to the International Conference on Harmonization guidelines [41]. Rutin (6) was selected as a chemical marker based on its constant presence in the genus, stability, and pharmacological activity, as previously mentioned. Diterpenes 3, 4, and 5 were not included in the validation process due to their instability. Flavonoids 7 and 8 were neither considered in the validation process due to the lack of a standard for compound 7 and the low concentration in the chromatographic profile in the case of compound 8. The calibration curve showed good linearity within the test range (R 2 ≥ 0.9996). The LOD and LOQ values were 0.19 and 0.57 µg/mL, respectively. Intraday and interday precision relative standard deviations (RSDs) were no more than 0.79% (Tables 2 and S1, and Figure S23). No significant degradation of 6 was detected in samples investigated over 72 h at room temperature (20 • C), at 37 • C, and at 4 • C, compared with the initial values. The method was linear, precise, and accurate for the quantitative evaluation of the marker. The content of rutin (6) in three batches of C. guatemalensis from different years (2014, 2015, and 2019) was investigated and the results are summarized in Table 3. Rutin (6) was identified in all batches with amounts between 0.55 and 0.64 mg/g (mg of 6/g of plant). Previous analyses reported a total of 6.02 mg/g of rutin (6) in leaves of a hydroalcoholic extract of C. campestris [6], which suggests a possible higher amount of the flavonoid in the leaves or the use of another solvent, such as methanol, as was shown in other studies [42,43].

Affinity-Directed Fractionation
In 2019, the in vivo hypoglycemic effect of the hydroalcoholic and aqueous extracts of C. guatemalensis was demonstrated, and the in vitro inhibition of α-gucosidases was also tested [17]; this assay did not show inhibition of α-glucosidases from rat intestine, thus, ruling out its hypoglycemic action mechanism as an α-glucosidase inhibitor. However, the extract showed greater activity (IC 50 = 32 µg/mL) than acarbose (IC 50 = 105 µg/mL) against α-glucosidases from Saccharomyces cerevisiae. The affinity-directed fractionation assay was implemented to find the metabolites responsible for this activity. EWE of C. guatemalensis was subjected to a gel permeation chromatography with a spin column packed with polyacrylamide, previously incubated with the α-glucosidases enzymes. The principle of affinity screening is based on the fact that target enzymes incubated with a complex matrix of natural compounds will retain the most tightly non-covalent binding active molecules from a mixture of closely related compounds [44,45]. The HPLC-MS chromatogram obtained from the affinity screening analysis of the EWE allowed the identifying of some of the ent-clerodane diterpenes observed in the previous fractionation procedures. Figure 6 illustrates the HRESI-MS obtained from these affinity screening assays: the HRESI-MS  Table 4). Other signals observed in the HPLC-MS spectrum ( Figures S24 and S25) are related to other high affinity compounds not observed in the previous fractionation procedures. However, the m/z yields molecular weights of structures with the same base skeleton (clerodane diterpenes) with one or more oxidations, for example: at 25.47-25.65 min the m/z is 363.1715, indicating a possible molecular formula, C 21 H 32 O 5 , whereas the peak at 27.84-27.98 min with m/z 347.1770 could be C 20 H 28 O 5 . Further analysis should be performed to confirm these possible structures. New prototypes of modulatory enzymes observed with affinity studies allowed knowing that these diterpenes had a high affinity for the S. cerevisiae α-glucosidase enzyme.
The importance of these experiments could be explained by the hypothesis of Kimura [46] that yeast and mammalian α-glucosidases belonged to two different families that differed in their amino acid sequences and their abilities to act on different substrates. The yeast and insect enzymes belong to family I (GH13) and have greater affinity for heterogeneous substrates, such as sucrose or 4-PNGP, whereas α-glucosidases from mammals belong to family II (GH31) and have greater affinity for homogeneous substrates, such as maltose.
In this sense, according to our findings, the inhibition of the Saccharomyces enzymes by the compounds could be more related to an ecological role that enables the plants to defend themselves against insect herbivory or fungal attacks by inhibiting type 1 enzymes.

Molecular Docking
Compounds 1-5 and acarbose (control) were constructed in 3D models and molec ular docking studies between ligands (acarbose and compounds 1-5) and the amino acid sequence of α-glucosidase from S. cerevisiae (MAL12) and human maltase-glucoamylase (MGAM-C) by AutoDock 4.2 software were performed to improve our understanding o the interaction of the high affinity compounds 1-5 inside the catalytic sites of MAL12 and

Molecular Docking
Compounds 1-5 and acarbose (control) were constructed in 3D models and molecular docking studies between ligands (acarbose and compounds 1-5) and the amino acid sequence of α-glucosidase from S. cerevisiae (MAL12) and human maltase-glucoamylase (MGAM-C) by AutoDock 4.2 software were performed to improve our understanding of the interaction of the high affinity compounds 1-5 inside the catalytic sites of MAL12 and MGAM-C, which were selected as the template for molecular modeling to establish a comparison between the resulting affinity-directed fractionation assay and the theoretical inhibition constant (Ki) obtained from in silico studies. To refine the results, the best conformations observed in the preliminary analysis were docked into a smaller area of the catalytic domain. Data are shown in Table 4. Acarbose fits well in the catalytic pocket of the analyzed enzymes and showed hydrogen-bonding interactions with the amino acid residues HIS279 (2.08 Å), GLN322 (2.00 Å), and ARG312 (2.15 Å) with MAL12, whereas the binding modes inside the catalytic site of MGAM-C corresponded to TYR1251 (1.93 Å), GLN1372 (1.75 Å), ARG1377 (2.11 Å), GLN1561 (2.07 Å), and GLY1588 (2.17 Å). Compounds 1-5 fit well in the catalytic pocket with MAL12 and showed hydrogen-bonding interactions with the amino acid residues HIS279 and ARG312, and preserved catalytic residues around TYR1251 in MGAM, which is involved in the catalytic substrate specificity of this protein [47]. Compounds 4 and 5 have the lowest Ki values of both analyzed enzymes (Table 4). These results plus the results of affinity studies with α-glucosidase indicate that the best conformation for enzyme inhibition is that of compound 5 (Figure 7). MGAM-C, which were selected as the template for molecular modeling comparison between the resulting affinity-directed fractionation assay and inhibition constant (Ki) obtained from in silico studies. To refine the re conformations observed in the preliminary analysis were docked into a the catalytic domain. Data are shown in Table 4. Acarbose fits well in the c of the analyzed enzymes and showed hydrogen-bonding interactions w acid residues HIS279 (2.08 Å), GLN322 (2.00 Å), and ARG312 (2.15 Å) whereas the binding modes inside the catalytic site of MGAM-C co TYR1251 (1.93 Å), GLN1372 (1.75 Å), ARG1377 (2.11 Å), GLN1561 (2.07 Å) (2.17 Å). Compounds 1-5 fit well in the catalytic pocket with MAL12 an drogen-bonding interactions with the amino acid residues HIS279 and preserved catalytic residues around TYR1251 in MGAM, which is involv lytic substrate specificity of this protein [47]. Compounds 4 and 5 have values of both analyzed enzymes (Table 4). These results plus the res studies with α-glucosidase indicate that the best conformation for enzym that of compound 5 (Figure 7). A secondary study was carried out in the catalytic site, using acarbo and the best conformation of each compound 1-5 of the refine study knowing the pharmacophore of compounds 1-5. Figure 8 shows the minim of the α-glucosidase complexed with active compounds 1-5 in the hypoth mode. The furane group at C-13 of all compounds forms a hydrogen bon of the catalytic residue HIS279, inducing a greater steric impediment at th catalytic pocket. A secondary study was carried out in the catalytic site, using acarbose as a control and the best conformation of each compound 1-5 of the refine study. This allowed knowing the pharmacophore of compounds 1-5. Figure 8 shows the minimized structure of the αglucosidase complexed with active compounds 1-5 in the hypothesized binding mode. The furane group at C-13 of all compounds forms a hydrogen bond with the NH of the catalytic residue HIS279, inducing a greater steric impediment at the surface of the catalytic pocket.
Plants 2022, 11, x FOR PEER REVIEW Figure 8. Pharmacophore docking results show the minimized structure of MAL12 compounds 1-5, the furane group forms a hydrogen bond with the NH of the HIS279.

Conclusions
In this study, eight compounds were isolated from de bark of C. gua the absolute configuration of two unreported ent-clerodane diterpenoids established by ECD spectrum. Quantification of the flavonoid rutin (6) was analysis of three different batches indicated very similar amounts of rutin each of them.
The approach for affinity-directed fractionation was applied at vario ing the isolation and purification processes to speed the identific α-glucosidase inhibitors, which could have an impact in the microscale s dereplication of active natural products, as demonstrated here for the clero guatemalensis. The present study provides insights into the phytochemical c the hydroalcoholic extract of C. guatemalensis and reveals new prototyp modulators through affinity studies. As previously mentioned, these find related to an ecological role that enables the plant to defend themselves aga insects or fungal attacks by inhibiting enzymes of family I, according to Because, in the present work, we used similar extracts to those previous some of the compounds isolated herein could be involved in the previo hypoglycemic activity. However, more experiments are needed to confirm

General Experimental Procedure
Analytical and preparative HPLC analyses were performed in an A finity system equipped with a G1311B quaternary pump, G1367E autosam DAD VL+, and controlled by Agilent ChemStation software (Agilent Tech Santa Clara, CA, USA). For analytical and semipreparative HPLC, a Luna C18, 50 × 2.1 mm id., 1.6 μm column (Phenomenex, Inc., Torrance, CA, U

Conclusions
In this study, eight compounds were isolated from de bark of C. guatemalensis, and the absolute configuration of two unreported ent-clerodane diterpenoids (2 and 3) were established by ECD spectrum. Quantification of the flavonoid rutin (6) was validated and analysis of three different batches indicated very similar amounts of rutin (6) content in each of them.
The approach for affinity-directed fractionation was applied at various stages during the isolation and purification processes to speed the identification of new α-glucosidase inhibitors, which could have an impact in the microscale separation and dereplication of active natural products, as demonstrated here for the clerodanes from C. guatemalensis. The present study provides insights into the phytochemical composition of the hydroalcoholic extract of C. guatemalensis and reveals new prototypes of enzyme modulators through affinity studies. As previously mentioned, these findings could be related to an ecological role that enables the plant to defend themselves against herbivory insects or fungal attacks by inhibiting enzymes of family I, according to Kimura [46]. Because, in the present work, we used similar extracts to those previously tested [17], some of the compounds isolated herein could be involved in the previously observed hypoglycemic activity. However, more experiments are needed to confirm this.

Plant Material and Extracts
Croton guatemalensis was collected by Dr. Carola Cruz, based on previous ethnobotanical studies (Cruz, 2011), at the Department of Chimaltenango, Guatemala, in 2019.
Ethanol-water extract (EWE) was made by heating 20 g of the dry plant material with a mixture of ethanol:water (1:1; 500 mL) during 2 h, followed by filtration and concentration under reduced pressure to remove ethanol in a rotary vacuum evaporator (Büchi Labortechnick, AG, Flawil, Switzerland) at 40 • C. Finally, it underwent lyophilization to yield 4.058 g of EWE. The extract was stored at 4 • C for HPLC analysis.

Isolation Compounds
MeOH (200 mL) was added to the WSF to obtain 672 mg soluble in methanol, which was subjected to Sephadex LH-20 using MeOH 100% as eluent. This process led to 17 subfractions (WSF1-WSF17); WSF16 (9.1 mg) was isolated as the pure compound 8 and analyzed by HPLC to be compared with the UV spectrum of a quercetin standard (>94% HPLC; Sigma-Aldrich), which was confirmed.

HPLC Method Validation
The method was validated according to the ICH guidelines for specificity, linearity, accuracy, precision, LOQ, and LOD [41]. Specificity was checked using the extract and a rutin (6) standard. Linearity of the method was evaluated by inspection of a rutin (6) standard solution at a concentration range of 20 to 250 µg/mL. A calibration line was made, and the least square line and correlation coefficient were calculated. Accuracy was evaluated by means of recovery assays carried out by adding known amounts of the standards of 6 to the sample at three different levels of the initial concentration of the sample. Average recoveries were calculated by the Equation (1).
Precision was evaluated by repeatability using six replicates at 100% of the test concentration. Stability was tested by analyzing the sample solution at different time points (0, 24, 48, and 72 h). LOD and LOQ were quantified based on the standard deviation (σ) of the response and the slope (S) calculated by the equations 2 and 3, respectively.

Affinity-Directed Fractionation
Gel permeation chromatography was performed with a spin column (BioRad Laboratories, Hercules, CA, USA) packed with polyacrylamide, 1 cm high, 100 µL swollen). The gel and samples were prepared in a solution of 0.1 M sodium phosphate buffer (pH 6.8) [45]. Aliquots (10 µL; in triplicates) of the extract (200 µg/mL) and acarbose (therapeutic control) were independently incubated for 5 min with 20 µL of the enzyme stock solution (0.9 units/mL of yeast α-glucosidase in 100 µM of buffer solution). Upon loading the test samples at the top of the spin exclusion column, the mixtures were eluted by centrifugation at RCF 42,985 g for 4 min; then, the eluate, corresponding to the solvent front and containing the α-glucosidase-acarbose complex, was collected and a denaturing solution (10 µL) of 3% glacial acetic acid in acetonitrile:water (1:1, v:v) was added and mixed with a vortex mixer. The solution was vacuum-dried and reconstituted with acetonitrile and analyzed by a coupled liquid chromatography system with single quadruple mass spectrometry and time of flight (HPLC-EM-SQ-TOF). Chromatographic profile elaboration was performed using a Phenomenex (Kinetex C 18 , 50 × 2.1 mm id., 2.6 µm) reverse phase column; the same flow gradient conditions mentioned above (item 4.4) were used. ESI mass spectra after the SEC/ESI-MS protocol for the acetonitrile and the enzyme functioned as background signals for the spectrum of the samples of interest.

Molecular Docking
Docking was carried out with the AutoDock 4.2 software (The Scripps Research Institute, La Jolla, CA, USA) using the default parameters. The molecular docking was performed with a model built by homology with Bacillus cereus α-glucosidase (1UOK.PDB) for the amino acid sequence of MAL12 from S. cerevisiae, which was retrieved from the UniProt protein resource data bank (accession code P5334) with preserved catalytic residues His111, Asp205, Glu276, His348, and Asp349 [48]. All files were prepared by adding polar hydrogen atoms and merged non-polar hydrogens to the enzyme structures and computing Gasteiger charges for the molecular model of analyzed compounds (1-5) as previously described for acarbose [48]. The entire system was subjected to a surface scanning and refined docking.

Computational Details
The Spartan'14 software was implemented to calculate the energy-minimized form with geometric optimization for all ligands, utilizing a semiempirical method (PM3). The resulting conformers were filtered and checked for redundancy. All conformers were minimized using a DFT force field at the B3LYP/DGDZVP level of theory employing Gaussian 09 software. The conformers were optimized, and thermochemical properties, IR, and vibrational analyses were obtained at the same level of theory. The TD-SCF with the default solvent model was used to perform the theoretical circular dichroism (TCD) calculations of the major conformers in the MeOH solution, using a B3LYP/DGDZVP force field. The calculated excitation energy (nm) and rotatory strength (R) in dipole velocity (R vel ) form was simulated into a TCD curve using the Harada-Nakanishi equation, as implemented in the SpecDis 1.71 software [49].