Bio‐guided fractionation of (9Z )‐5,8,11‐trihydroxyoctadec‐9‐enoic acid from Abelmoschus sagittifolius root as a potent α‐glucosidase inhibitor: An integrated in vitro and in silico study
Authors: Nguyen Phuong Mai, Nguyen Van Quan, Nguyen Dang Duc, Tran Dang Xuan
Published: 2026

ABSTRACT
AbstractBACKGROUND Abelmoschus sagittifolius stands out as a valuable medicinal source for developing functional food targeting type 2 diabetes. This study investigated the α‐glucosidase inhibitory potential of an active fraction obtained from A. sagittifolius roots (ASR), enriched in (9Z)‐5,8,11‐trihydroxyoctadec‐9‐enoic acid (OA), using both enzyme kinetics and molecular docking analyses. The effect of simulated digestion and in silico acute oral toxicity was also evaluated.RESULTSThe bioactive fraction, with OA as major component, exhibited strong α‐glucosidase inhibitory activity [half‐inhibitory concentration (IC50) = 15.96 μg mL−1], which was markedly higher than acarbose (IC50 = 2109.51 μg mL−1) and palmitic acid (IC50 = 618.93 μg mL−1). Kinetic analysis indicated that OA acts as a mixed inhibitor, with K i < K i′, suggesting stronger affinity for the free enzyme than for the enzyme–substrate complex. Molecular docking supported these findings by revealing stable binding of OA within the catalytic pocket of α‐glucosidase through multiple hydrogen bonds and hydrophobic interactions. The reduction in bioaccessibility of OA through gastrointestinal tract indicates it is less accessible in the digestive system. Qualitative and quantitative in silico acute oral toxicity analysis suggested OA to be a non‐toxic substance.CONCLUSIONThis study provides the first investigation into the potential of an OA enriched fraction and its inhibitory mechanism, highlighting OA as a promising bioactive compound for the development of functional foods or nutraceuticals aimed at managing type 2 diabetes. However, further studies, including enzyme‐based validation, toxicity evaluation, bioavailability assessment and clinical investigation, are required to comprehensively confirm the efficacy and safety of OA as a potent α‐glucosidase inhibitor. © 2026 The Author(s). Journal of the Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

INTRODUCTION

Diabetes mellitus was recognized as the ninth most prevalent cause of mortality by the World Health Organization (WHO), being responsible for over two million deaths each year (WHO, 2024) 1 . One of the effective diabetes management strategies is suppressing ⍺‐glucosidase, an important enzyme involved in carbohydrate metabolism, which can reduce postprandial hyperglycemia. 2 As ⍺‐glucosidase inhibitors, acarbose, voglibose and miglitol are the most commercially available for consumption as synthetic drugs; however, their application is limited by various adverse side effects. 2 Therefore, it is critically important to identify α‐glucosidase inhibitors that are secure, efficiently utilized, natural and have fewer adverse effects.

Abelmoschus sagittifolius has been used in Ayurvedic and Traditional Chinese Medicine since the 19th century for the treatment of cough, tuberculosis, backache, stomachache, constipation, neurasthenia, dizziness, abscesses and lumbosacral pain. 3 Among the A. sagittifolius plant parts, A. sagittifolius roots are the most nutritious, which have been predominantly utilized as a source of mucilage, starch, phytosterols, coumarin, fatty acids, organic acids and reducing sugar. 4 These phytochemical compounds have been confirmed to bring significant benefits in cytotoxic activity, 5 antidiabetic activity 6 and antioxidant activity. 3 These findings suggest that A. sagittifolius holds promise as a functional food, although further studies are imperative.

Among 31 compounds isolated from A. sagittifolius, only one fatty acid has been isolated and showed potential in cytotoxic activities; the antidiabetic potential of the fatty acids isolated from A. sagittifolius has never been examined. This is noteworthy because previous studies on other plant species have demonstrated that specific fatty acids, such as palmitic, pentadecanoic and heptadecanoic acids, exhibit inhibitory effects against α‐glucosidase activity. 7 Compared with acarbose and other synthetic drugs, fatty acids are naturally occurring compounds that may present fewer adverse effects, commercial value and potential applications as additives and supplements for diabetic people in the management of type 2 diabetes. 8 Structurally, based on the length of the hydrocarbon chain, fatty acids are characterized into: short chain fatty acids (2–5 C atoms); medium chain fatty acids (6–12 C atoms) and long chain fatty acids (more than 12 C atoms). Of these, (9Z)‐5,8,11‐trihydroxy octadec‐9‐enoic acid (OA) is a monounsaturated fatty acid derived from long‐chain fatty acids with three hydroxy groups in its structure. Currently, only a limited number of hydroxy fatty acids have been investigated for their biological activities.9, 10 Specifically, none of the comprehensive studies concern the α‐glucosidase inhibition activity and bioaccessibility of OA from A. sagittifolius roots. Therefore, for the first time, the present study investigated the potential and mechanism of OA derived from Abelmoschus sagittifolius root in inhibiting α‐glucosidase enzyme activities using a kinetic study and molecular docking, as well as the bioaccessibility of OA through simulated gastrointestinal digestion and toxicity assessment. The results could provide significant insights into the digestive stability and functional mechanism of α‐glucosidase inhibition from OA in A. sagittifolius extracts, supporting its further exploration for use in nutraceutical and therapeutic applications.

MATERIALS AND METHODS

Chemicals and reagents

All organic extraction solvents were purchased from Junsei Chemical Co., Ltd (Tokyo, Japan). Saccharomyces cerevisiae‐derived α‐glucosidase (CAS number: 9001‐42‐7), p‐nitrophenyl‐α‐d‐glucopyranoside (pNPG) (CAS number: 3767‐28‐0), p‐nitrophenol (PNP) (CAS number: 100‐02‐7), silica gel (60 Å, 60–100 mesh and 200–400 mesh) and all buffer solutions were purchased from Sigma‐Aldrich (St Louis, MO, USA). All of the standards, including acarbose (CAS number: 56180‐94‐0) quercetin (CAS number: 117‐39‐5) and palmitic acid (CAS number: 57‐10‐3), were acquired from Wako Pure Chemical Co., Inc. (Tokyo, Japan).

Plant sample

Abelmoschus sagittifolius was collected in Son Tay district, Ha Noi, Vietnam, in 2023. The samples were identified based on phenotypic characteristics and confirmed by all of the investigators. The specimen voucher ASR_HU2023 was deposited at the Laboratory of Plant Physiology and Biochemistry, Graduate School of Advanced Science and Engineering, Hiroshima University, Japan. The AS root was prewashed with tap water several times before drying in the sunlight for 7 days.

Bio‐guided fractionation of bioactive extract via column chromatography

Dried samples (6.5 kg) were extracted three times with methanol (15 L) for 3 days at room temperature. After filtration, the methanolic extract was evaporated by a rotary evaporator to dryness under reduced pressure. The extraction and fractionation processes are shown in Fig. 1. The bioactive fractions were run through three column chromatographies based on α‐glucosidase inhibitory activity assays. In addition, the C2‐F2 and C2‐F3 fraction was dissolved in MeOH before running through the third column.

Bio‐guided fractionation of bioactive compounds from Abelmoschus sagittifolius (AS) root.

Identification and quantification of phytocompounds in AS root

OA was detected and quantified using liquid chromatography coupled with electrospray ionization tandem mass spectrometry (LC–ESI‐MS/MS). The system included a Tribrid mass spectrometer Orbitrap Eclipse (Thermo Fisher Scientific, Waltham, MA, USA) and an ESI source, which were optimized as follows: ion source voltage = 4.5 kV, capillary voltage = −50 V, tube lens offset = −80 V and capillary temperature = 350 °C. Nitrogen was used as both the sheath and auxiliary gases, with flow rates of 50 and 10 arb (arbitrary unit), respectively. For the analysis, 1.0 μL of a sample was injected into the XBridge®Shield RP18 column (130 Å, 5 μm, 2.1 × 100 mm; Waters Corp., Milford, MA, USA). The column temperature was maintained at 30 °C, and the flow rate was 400 μL min−1. The gradient mode consisted of two solvents: solvent A (0.1% FA) and solvent B (ACN‐0.1% FA). The gradient program used was: 80% A and 20% B, 0–2 min; increasing up to 100% B, 2–5 min; and maintaining at 100% B, 5–8 min. At 8–9 min, the mobile phase was adjusted to the initial condition of 80% A and 20%, and finally equilibrated at 9–15 min. The total running time was 15 min.

α‐Glucosidase inhibition assays

α‐Glucosidase inhibitory activity of the sample was evaluated using the method described previously 11 with modifications. At the beginning, 20 μL of the sample in 40 μL of 0.1 m potassium phosphate buffer (pH 7) was pre‐mixed with 20 μL of 0.5 U of α‐glucosidase enzyme. A 20‐μL aliquot of 5 mm pNPG substrate (in the buffer) was added after 5 min of incubation at 25 °C. Subsequently, the prepared mixture was incubated for 10 min at 25 °C. Finally, the reaction was suspended by adding 100 μL of 0.1 m Na2CO3. The resulting mixture was measured at 405 nm using a spectrophotometer. Acarbose was used as a positive control. The inhibition percentage of the sample against α‐glucosidase was calculated using: %Inhibition=Ac−Ab−As−AbAc−Ab×100 where Ac is the absorbance of the reaction with dimethyl sulfoxide as a negative control, As is the absorbance of the sample and Ab is the absorbance of the blank sample. The linear equation obtained from the dose‐dependent curve was used to calculate the half‐inhibitory concentration (IC50). A lower IC50 value presents a stronger inhibitory effect.

Kinetic study of α‐glucosidase inhibitory effect

The kinetic analysis was performed following the procedure described for the α‐glucosidase inhibition assay, with minor modifications. A constant amount of α‐glucosidase was incubated with increasing concentrations of pNPG (1.25, 2.5, 5 and 10 mm) in the presence of various sample concentrations (0, 25, 50 and 100 μg mL−1). The reaction rates obtained under each condition were converted to their double reciprocal values (1/V and 1/[S]) to construct Lineweaver–Burk plots. The maximum initial reaction rate (V max) was obtained from the intercept of the Lineweaver–Burk plot, and the Michaelis constant (K m) was calculated from the slope of Eqn (1): (1) 1V=1Vmax+KmVmax1S

Homology model and validation of the homology model

The 3D structure of the protein was built based on the Swiss Modeling server (http://swissmodel.expasy.org) using homology modeling based on the amino acid sequence (UniProt ID: P53341), which has 72.68% sequence identity with the target protein. The homology model was evaluated for its stereochemical quality check using the SAVES server (https://saves.mbi.ucla.edu) through PROCHECK, VERIFY 3D and ERRAT. This model was accepted as the best model for molecular docking studies.

Molecular docking

The protein was hydrogenated by Autodock Tools 1.5.6 to remove water and set as a receptor. The 3D structure of OA and acarbose was obtained through the Pubchem database (https://pubchem.ncbi.nlm.nih.gov), and their structure were optimized by energy minimization and set as ligand. Molecular docking was carried out using Autodock Vina v1.2.0 (https://vina.scripps.edu) with an exhaustiveness value set as 20, and generating 50 poses at maximum for each run. The homology model was enclosed in a grid having 0.375 Å spacing, and the inhibitor was allowed to move within the whole region to achieve the lowest energy conformation. The best docking results were selected for visual analysis using ChimeraX (https://www.rbvi.ucsf.edu/chimerax) and BIOVIA Discovery Studio (https://www.3ds.com/products/biovia/discovery-studio) software.

In vitro simulated digestion and bioaccessibility evaluation

The in vitro digestion model was conducted based on a previously reported method 11 with minor modifications. The model was designed into three phases, including the mouth, stomach and small intestine. The preparation of chemicals for in vitro digestion is shown in the Supporting information (Data S1). The process is carried out by imitating the gastrointestinal physiology, consisting of chemical components, stomach pH, temperature (37 °C) and transition time to determine the effect of in vitro digestion on the biological activities of A. sagittifolius plant extracts. Particularly, at the beginning of the process, 1 mL of sample (5 mg mL−1) was added to 4 mL of artificial saliva for 5 min in a shaking water bath at 37 °C. Then, 4 mL of gastric juice was added to the sample tube, and the mixture was kept for 1 h at 37 °C. Finally, 4 mL of small intestinal juice was added to the sample tube and incubated for 1 h at 37 °C. Ethyl acetate was added to the sample tube (1:1, v/v) at the end of each digestive phase to reconstitute bioactive compounds. After vigorous mixing, the ethyl acetate phase was collected, evaporated and prepared for biological assays.

The bioaccessibility of OA was calculated using Eqn (2): (2) Bioaccessibility%=DOAIOA×100 where DOA is the quantity of OA in the digested extract and IOA is the initial quantity of OA in the undigested extract.

In silico acute oral toxicity analysis

The evaluation of acute oral toxicity of the OA compound was performed, following the method described previously 12 with minor modifications. In detail, seven different in silico prediction methods were employed for both qualitative analysis (STopTox, admetSAR3.0 and ADMETlab3.0) and quantitative analysis (TEST 5.1.2; ProTox 3.0; QSAR Toolbox 4.8 and VEGA QSAR 1.2.6). Details of each model and validation method are provided in the Supporting information (Data S2).

Statistical analysis

The experiment was conducted with three replications. Statistical analysis of the raw data was performed through one‐way analysis of variance and a general linear model using Minitab 16.2.3 (Minitab Inc., State College, PA, USA). The results are expressed as the mean ± SD. Significant differences were determined using Tukey's test at a 95% confidence level, based on the groups.

RESULTS

Bio‐guided fractionation and identification of OA

Bioactive guided fractionation was performed on the roots of A. sagittifolius based on α‐glucosidase inhibition. All extracts and fractions were screened to identify the bioactive fractions, ultimately selecting the fraction C3‐F1 as the most active (Fig. 1) for further analysis. A preliminary screening of the phytochemical constituents in fractions C3‐F1 was conducted using LC–ESI‐MS/MS for characterization and identification of the major compounds. The retention of the principal peak, which accounts for 79.94% of the total peak area of C3‐F1, is recorded at 3.85 min (Fig. 2(A)), in which a precursor ion [M‐H]− of m/z 329.23 was characterized as OA (Fig. 2(B)) based on NIST 2.4 (National Institute of Standards and Technology, Gaithersburg, MD, USA) databases and the previous literature13, 14, 15 The major mass fragmentation of OA (Fig. 2(B)) is: 229.10 (100); 211.13 (66); 171.03 (62); 311.21 (49); 293.22 (47); 183.09 (7); 1139.03 (4); 93.16 (3); 127.07 (3); 249.20 (2). The OA‐enriched fraction C3‐F1 will continue to be tested in the α‐glucosidase inhibitory activity assays.

Liquid chromatography coupled with electrospray ionization (LC‐ESI) chromatogram (A) and MS/MS product ions for m/z 329.23 (B) of C3‐F1 fractions.

Inhibitory effect of OA‐enriched fraction on α‐glucosidase enzyme activities

The α‐glucosidase inhibitory activities of the AS crude extract, potential fractions and standards are summarized in Table 1. It could be seen that C3‐F1 effectively inhibited α‐glucosidase enzyme activities in a dose‐dependent manner with the IC50 values of 15.96 μg mL−1. This value is much higher than that of crude extract (IC50 = 429.26 μg mL−1), ethyl acetate extract (IC50 = 143.44 μg mL−1), standard quercetin (IC50 = 19.48 μg mL−1), standard acarbose (IC50 = 2109.51 μg mL−1) and standard palmitic (IC50 = 618.93 μg mL−1).

Table 1 α‐Glucosidase inhibition activities IC50 (μg mL−1)SampleDescriptionIC50 (μg mL−1)R1Methanol crude extract429.26 ± 4.00 aREEthyl acetate extract143.44 ± 1.37 bC1‐F6Highly α‐GI fraction obtained from first column49.76 ± 7.82 cC2‐F2Highly α‐GI fraction obtained from second column46.16 ± 1.07 cC2‐F3Highly α‐GI fraction obtained from second column25.47 ± 0.35 dM1MeOH dissolved part of C2‐F2 and C2‐F352.01 ± 0.41 cC3‐F1Highly α‐GI fraction obtained from third column15.96 ± 0.12 eQuercetinStandards19.48 ± 0.40Acarbose2109.51 ± 59.23Palmitic acid618.93 ± 91.83Note: Data are expressed as the mean ± SD. Significant differences at P < 0.05 by Tukey's test are shown by different lowercase letters in the table within the column; α‐GI: α‐glucosidase inhibitory.

α‐Glucosidase inhibition activities IC50 (μg mL−1)

Note: Data are expressed as the mean ± SD. Significant differences at P < 0.05 by Tukey's test are shown by different lowercase letters in the table within the column; α‐GI: α‐glucosidase inhibitory.

Kinetic analysis

To analyze the type of α‐glucosidase inhibition by OA, the Lineweaver–Burk plots were constructed. These plots were generated by plotting 1/[S] (substrate concentration) against 1/[V] (reaction rate), with varying concentrations of OA (0, 6.25, 12.5 and 25 μg mL−1). The results are shown in Fig. 3. The intersection points on the Lineweaver–Burk plot indicated that OA is a mixed‐type inhibitor on α‐glucosidase. In addition, the presence of OA led to reductions in V max but an increase in K m value (Table 2). The inhibition constants K i (toward free α‐glucosidase) and K i′ (toward α‐glucosidase substrate complex) values were calculated to be 8.76 and 37.52 μm, respectively, from the secondary mapping of slope and intercept versus inhibitor concentrations. The smaller K i value than K i′ value suggested a stronger affinity of OA to free enzyme than to enzyme substrate complex.16, 17 Moreover, the linear secondary mapping (Fig. 3(B),(C)) also indicated that OA inhibited α‐glucosidase activity through one site. 18

(A) Inhibition of identified compound by Lineweaver–Burk plot. (B, C) K i and K i′ by the secondary mapping of slope and intercept versus inhibitor.

Table 2 K m and V max values of OAConcentration (μg mL−1) K m V max 00.092.006.250.341.1712.50.420.83250.480.61

K m and V max values of OA

Model assessment

The homology model was assessed by using PROCHECK, VERIFY 3D and ERRAT through the SAVES server. The results from Ramachandran plot (Fig. 4) indicated 89.3% of the residues within the most favorable regions, 10.1% within the additional allowed regions, 0.2% in the generally allowed region and 0.4% in the disallowed region. Furthermore, the ERRAT quality factor and VERIFY 3D score were observed as 93.88 and 98.62%, respectively. Furthermore, the QMEANDisCo (https://swissmodel.expasy.org) results of 0.91 ± 0.05 were also confirmed. Overall, these results indicated that the generated model is reliable and of good quality (Fig. 4 and Table 3).

Homology modeling assessment. (A) Homology model. (B) QMEANDisCo check. (C) ERRAT check. (D) VERIFY 3D check. (E) Ramachandran plot of the homology model obtained by PROCHECK.

Table 3Evaluation plot statistics results of model and template by PROCHECK, VERIFY3D and ERRATEvaluationPROCHECKVerify 3DERRATMost favored regionsAdditional allowed regionsGenerally allowed regionsDisallowed regions3D‐ID scoreQuality factorHomology model89.3%10.1%0.2%0.4%98.62%93.8811

Evaluation plot statistics results of model and template by PROCHECK, VERIFY3D and ERRAT

Molecular docking

The results from the above experiments confirmed the potential of OA in the inhibition of α‐glucosidase. However, the binding interaction between them has not been clearly defined. For this reason, the binding characteristics between OA and α‐glucosidase were further explored using molecular docking. Molecular docking is a valuable and essential method for drug discovery. It can reveal the theoretical mode of action between ligand and receptor, and predict the potential binding site of the ligand.

Previous kinetic results suggested OA exhibited significant mixed inhibition of α‐glucosidase through an in vitro assay. Therefore, OA was docked into different binding sites, including the active site and several identified hotspots as the allosteric site of the S. cerevisiae α‐glucosidase (Table 4 and Fig. 5). The best docking conformation, which possessed the lowest energy, is described in Fig. 6. The estimated binding energy between the α‐glucosidase and OA was −6.821 kcal mol−1 (Table 5). As shown in Fig. 6, the residue ASP 349, which belongs to the active site, formed a key hydrogen bond (bond length: 2.01 Å) interaction with OA, which was the main binding mode between OA and α‐glucosidase (Fig. 6). Furthermore, OA can form hydrogen bonds with ARG 212, HIS 348, ASP 408, ASP 214 and ARG 439. Additional van der Waals bonds were also seen with the residues ASN 412, TYR 313, PHE 158, GLU 276, ARG 443, TYR 71, THR 215, PHE 177 and PHE 300. In addition, these results were consistent with inhibition kinetics results, confirming that OA was a mixed‐type inhibitor and preferred binding to the active site. In this molecular docking simulation, we considered standard inhibitor acarbose (the first α‐glucosidase inhibitor approved for type 2 diabetes treatment) as the standard ligand for validating the Autodock Vina results, and the molecular docking results showed that the binding sites between α‐glucosidase and OA were the same as the binding site for acarbose, a known competitive inhibitor.

Table 4Predicted allosteric sites of S. cerevisiae α‐glucosidase enzymeSiteResiduesReferenceAS1Lys12, Trp14, His258, Lys262, Val265, Glu270, Ile271, Thr287, Ala289, Tyr292, Val294, Thr295, Ser339 19, 20 AS2Gln66, Gln67, Met69, Ser179, Arg180, Glu405, Val407, Lys410, Asn411, Trp465 20 AS3Thr287, Val297, Ser299, His302, Ile334, Trp340, Ala341, Thr342, Tyr344 21 AS4Thr9, Pro11, Lys15, Ile334, Asn335, Ser339, Trp340, Thr380, Tyr508, Tyr510, Tyr529 19 AS5Tyr142, Ile149, Lys147, Pro150, Asp227, Asp232, Ile236 19 AS6Lys155, Phe157, Leu176, Leu237, Gln238, Gly243, Ser244, Phe311, Arg312 19 AS7Ser157, Thr245, Ser240, Asp242, Ser241, Tyr158, Lys156, Gln239 21 AS8Glu271, Trp15, Thr274, Ser298, Trp343, Lys16 22 AS9Glu405, Arg413, Asp68, Met70, Glu408, Gln67, Trp468, Trp36, Tyr470, Lys406, Tyr407 23 Active siteAsp349, Glu276, Asp214

Predicted allosteric sites of S. cerevisiae α‐glucosidase enzyme

Structural topology of S. cerevisiae α‐glucosidase with active sites and predicted allosteric sites. The generated cavities are displayed in boxes. The predicted allosteric sites and active site are shown in a sphere model, and each site is labeled.

The best docking conformation and interaction pattern of α‐glucosidase with OA. (A) Overall view of small molecule and protein binding. (B) Local view with OA in grey ball stick, amino acid in cyan cartoon as binding site, and yellow dashed lines indicating hydrogen bonding interactions. (C) 2D diagram of OA interaction with α‐glucosidase.

Table 5Molecular docking at different sitesDocking siteBinding energyInteractionBinding siteAS1−5.84Hydro bondVal294AS2−5.14Hydro bondGln67; Lys410; Glu414AS3−5.19Hydro bondAsn411AS4−4.23––AS5−4.69Hydro bondSer161AS6−6.03Hydro bondArg312, Asp408AS7−5.72–Glu304AS8–––AS9−5.19Hydro bondGlu405, Gln67, Asn411Blind−5.97Hydro bondHis348Active site−6.82Hydro bondArg212, His348, Asp349Acarbose/active site−7.78Hydro bondHis348, Asn241, Arg439, Glu276

Molecular docking at different sites

In vitro digestion

The preliminary screening revealed that the most potent α‐glucosidase activity was determined by fractions C3‐F1; therefore, these fractions were selected to evaluate their bioaccessibility, bioactive components and bioactivity via in vitro digestion. For the first time, the present study indicates that simulated digestion significantly affected the α‐glucosidase activities of bioactive fractions from AS roots (Table 6).

Table 6Effect of in vitro digestion on α‐glucosidase activity of OASampleQuantification (μg mL−1)IC50 (μg mL−1)Bio‐accessibility (%)Un‐digest999.76 ± 41.90 a15.96 ± 0.12 c100.00Oral phase159.38 ± 5.64 b51.04 ± 1.96 b15.94Gastric phase79.19 ± 3.63 c153.94 ± 4.00 a7.92Intestinal phase93.73 ± 17.10 c154.13 ± 5.40 a9.37Note: Data are expressed as the mean ± SD. Significant differences at P < 0.05 by Tukey's test are shown by different lowercase letters in the table within the column.

Effect of in vitro digestion on α‐glucosidase activity of OA

Note: Data are expressed as the mean ± SD. Significant differences at P < 0.05 by Tukey's test are shown by different lowercase letters in the table within the column.

In general, the α‐glucosidase inhibition activities of fractions C3‐F1 decreased over the digestive stages. Particularly, the α‐glucosidase capacity of C3‐F1 was reduced at the oral phase, gastric phase and intestinal phase by 3.4‐, 9.58‐ and 9.65‐fold, respectively.

The change in OA content and its bioaccessibility over the in vitro digestion results are presented in Table 6. The results indicated that the OA contents in C3‐F1 significantly declined over the digestive stages, by 6.27‐, 12.62‐ and 10.66‐fold, respectively, throughout the in vitro digestion. Significantly, the reduction of bioaccessibility of OA suggests that this natural α‐glucosidase inhibitor is less accessible in the digestive system (Table 6). In addition, the OA content exhibited a strong correlation (0.986, p 〈 0.001) with α‐glucosidase inhibition activities in bioactive fractions that contained it from the A. sagittifolius root during in vitro digestion. The data indicate that the variation in OA content was in line with that of the α‐glucosidase inhibition activity of the bioactive extracts of A. sagittifolius root during digestion. This also suggests that the accessible content of OA may be the major contributor to the α‐glucosidase inhibition activity of bioactive fractions during stages of digestion.

In silico acute oral toxicity

Qualitative in silico evaluation of OA acute oral toxicity is presented in Table 7. The results from STopTox prediction model indicated OA as non‐toxic, with a confidence level of 78%. This model also defined the hydroxy groups in OA as potential toxicophores. Similarly, ADMETlab and admetSAR predicted OA to be non‐toxic in terms of acute oral toxicity, with predicted toxicity probability of 1.1 and 25.3%, respectively.

Table 7Qualitative predictions of acute toxicity for oral dosage of OA in ratsSTopToxADMETlab 3.0ADMET sarPredictionConfidence (%)Applicability domainPredicted toxicophore(s)Probability (%)Probability (%)Non‐toxic78% 1.1%25.3%

Qualitative predictions of acute toxicity for oral dosage of OA in rats

Quantitative in silico assessment of acute oral toxicity was further performed using multiple prediction tools, targeting oral administrations in rat (Table 8). The predicted toxicity was expressed as the median lethal doses, LD50. The results showed considerable variation among the different software platforms and prediction methods. TEST 5.1.2 (consensus method) indicated an LD50 value of 4596.83 mg kg−1 bw, whereas TEST 5.1.2 (hierarchical clustering) and TEST 5.1.2 (nearest neighbor) showed an LD50 of 15 633.34 and 1351.65 mg kg−1 bw, respectively. Furthermore, ProTox 3.0 software exhibited 11 800 mg kg−1 bw with a prediction accuracy of 69.26%. The QSAR Toolbox 4.8 and VEGA QSAR 1.2.6 were also employed and indicated an LD50 value of 6340 and 199.34 mg kg−1 bw, respectively.

Table 8Quantitative predictions of acute toxicity (LD50) for oral dosage of OA in ratsSoftwareRat oral LD50 (mg kg−1 bw)Similarity coefficientTEST 5.1.2. (consensus method)4596.83SC ≥ 0.5TEST 5.1.2 (hierarchical clustering)15 633.34SC ≥ 0.5TEST 5.1.2 (nearest neighbor)1351.65SC ≥ 0.5ProTox 3.011 800SC = 69.26%QSAR Toolbox 4.86340NAVEGA QSAR 1.2.6199.34NA

Quantitative predictions of acute toxicity (LD50) for oral dosage of OA in rats

DISCUSSION

Bioassay‐guided fractionation of the A. sagittifolius root showed the high potency of the OA enriched fraction with respect to inhibiting α‐glucosidase. The presence of OA has been reported earlier, but this is the first time that it has been isolated and identified as the predominant fatty acid in C3‐F1 fractions by bioassay‐guided fractionation from the A. sagittifolius root. Because of the complex structure of OA, difficulty was encountered during the purification and identification process of OA. Various purification methods, including Sep‐Pak cartridge chromatography, preparative thin‐layer chromatography, preliminary purification via MeOH dissolution and silica gel column chromatography, were evaluated. In addition, different LC modes and MS detectors were also used for compound identification. After careful screening and selection, the reported repeated column chromatography combined with OA identification based on relative abundance was selected as the most suitable purification and identification method. Numerous studies have successfully applied bioactivity‐guided fractionation to isolate both novel and known phytoconstituents with significant biological activities.24, 25, 26, 27 These investigations commonly demonstrate a high quantitative contribution of the active compound within enriched fractions. For example, PC1 isolated from Phyllanthus muellarianus stem bark, representing 68.7% of the total peak area, was identified as a selective inhibitor of human butyrylcholinesterase. 24 Similarly, a triterpenoid acid enrichment fraction enriched from Hippophae rhamnoides L pomace extraction fraction (75.23% of the total peak area) exhibited a strong inhibitory effect on α‐glucosidase. 25 An enzymatically fructosylated of 4PHF‐enriched fraction (74.5% of the total peak area) showed potent inhibition on α‐amylase, 26 whereas a polyphenol‐enriched fraction (72.3% of the total peak area) of Tagetes erecta L. flower showed inhibition of α‐melanocyte‐stimulating hormone‐induced melanogenesis. 27 In the present study, the quantitative contribution of OA to the observed biological activity accounted for 79.94% of the total peak area based on the relative abundance. Therefore, these values could be considered sufficient to support OA as the principal compounds responsible for the α‐glucosidase inhibitory activity and further mechanistic investigations of the fraction.

The structure of OA revealed its significant role in the α‐glucosidase inhibitory activity. The three hydroxyl groups located in C5, C8 and C11 and the free hydroxyl in the carboxyl group indicated a significant role in the α‐glucosidase inhibitory activity (IC50 = 15.96 μg mL−1). The higher activity of known α‐glucosidase inhibitors is generally a result of the presence of numerous hydroxyl groups. 7 Hydroxyl groups are known to give a fatty acid special properties, especially in their reactivity compared to other non‐hydroxylated fatty acids. 28 Furthermore, the inhibition of α‐glucosidase may be significantly influenced by the presence of a double bond, 29 indicating that unsaturated fatty acids presented more potent inhibitory activity against α‐glucosidase than that of saturated fatty acids. 29 Additionally, they also suggested that fatty acids with odd double bonds inhibited α‐glucosidase more strongly than those with even double bonds. 29 Various studies indicated higher potential of fatty acids than standard acarbose in their α‐glucosidase inhibitory activities, such as (9Z,12Z)‐octadecadienoic acid (linoleic acid) (IC50 = 14.8 μm), 30 10‐hydroxy‐8(E)‐octadecenoic acid (IC50 = 0.07 mm), 28 palmitic (IC50 = 0.74 mm) and linolenic acid (IC50 = 0.17 mm). 31 Although these studies have suggested that long‐chain unsaturated fatty acids are strong inhibitors of α‐glucosidase, 32 their mechanisms of action remain unclear.

To gain deeper insight into the inhibitory mechanism, we adopted an experimental and computational chemistry‐based strategy to systematically evaluate the binding affinities using kinetic analysis and a molecular docking study. Because the 3D structure of α‐glucosidase used in our biological assays from Saccharomyces cerevisiae is not available yet in the protein data bank, we built a 3D structure of the protein based on the Swiss Modeling server using homology modeling based on the amino acid sequence (UniProt ID: P53341). Many studies verified that α‐glucosidase possessed allosteric sites.19, 20 Notably, the allosteric inhibitors generally function as mixed or non‐competitive inhibitors. Therefore, in the present study, a total of 10 allosteric sites and active site docking were run to confirm the OA mechanism of inhibition. Among them, the conformation docking within the active site with the lowest energy was the most optimal binding gesture, which included hydrogen bonding and hydrophobic interactions. The hydrogen bonding between the hydroxyl groups of OA and ASP349 and ASP214, the catalytic triad in the active site of S. cerevisiae α‐glucosidase, can effectively decrease enzyme activity. This suppression could be effective because of the maintenance of equilibrium between strong binding affinity, high interaction energy and effective hydrogen bond forming capacity by the multipoint interaction between ligands and the enzyme catalytic region.33, 34 The reported results on the binding of the mixed inhibitor to the active site were consistent with findings reported in multiple previous studies33, 34 indicating that mixed inhibitors predominantly bind to the active site of enzymes. 35 However, because the OA dynamic changes in conformation and microenvironmental impacts of biological systems are not fully reflected in in silico studies, the mechanism analysis offered in the present study by computational docking approaches mainly provides conceptual guidance rather than conclusive structural evidence. To fully understand the OA inhibitor mechanism, additional investigations, including molecular dynamics, multispectral techniques such as fluorescence spectroscopy, circular dichroism spectroscopy, etc, need to be performed. 34

Understanding fatty acid behavior during and after digestion would provide further insight into their potential application in pharmaceutical industry. Even although human nutritional research continues to be considered as the benchmark approach for dealing with food‐related issues, in vitro studies are frequently employed since the advantages of speed, cost‐effectiveness, labor and freedom from ethical constraints. 36 During the digestion process, numerous variables, such as time, temperature and pH, could affect the absorption efficiency of fatty acids. 37 Additionally, variations in chain length, geographical distribution and fatty acid saturation significantly impact bioaccessibility and bioavailability. Bioaccessibility is defined as the percentage of a digested bioactive compound that is released after digestion and available for absorption, whereas bioavailability is the amount absorbed by the body. 38 In some cases, especially for hydrophobic substances, bioaccessibility is measured as the amount of bioactive substances solubilized within the mixed micelle phase compared to the original amount ingested. 39 Mixed micelles originate from triglyceride digestion products, including bile salts, phospholipids, monoacylglycerols and free fatty acids, and are employed for the hydrophobic substances on the solubilization, transportation, absorption and chemical reactions that occur within the gastrointestinal tract.39, 40 Considering the hydrophobic characteristics of OA, which include poor water solubility, natural vulnerability to oxidation and quality degradation during digestion, the formation of OA loaded in mixed micelles is mandatory. 40 However, this formation might result in a reduction in bioaccessibility and bioavailability of OA after digestion as a result of modifications in its molecular and physiochemical characteristics to fit within the mixed micelles. 39 Additionally, bioaccessibility might be underestimated because of the loss of bioactive substances in solubility vesicles and lipid droplets from mixed micelles during the centrifugation step. 39 To prevent this loss, in the present study, ethyl acetate was employed at the end of each digestive phase to terminate the enzymatic reaction and reconstitute OA from the mixed micelles. 41 To enhance OA bioaccessibility, further studies should investigate how OA moves into mixed micelles, factors that impact this process, and the dimensions of the hydrophobic domains formed within the mixed micelles. 39 Additional studies should also be conducted to preserve these lipids from oxidation, such as encapsulation or emulsification. 42

On the other hand, acute oral toxicity stands out as a pivotal parameter in toxicological assessment. In the present study, the results from qualitative analysis indicated the non‐toxicity of OA regarding acute oral toxicity. Additionally, variation in the LD50 values in quantitative analysis from different prediction models could be the result of a limited number of experimental toxicity data on the compound related to OA, and none on OA have been reported. Because toxicity evaluation of bioactive compounds is a critical and indispensable step, further comprehensive toxicity assessments of OA are required to fully characterize its toxicity profile. These assessments should include both acute and chronic studies, along with the evaluation of various indicators such as drug interactions, reactive metabolites, non‐specific targets, cardiotoxicity, mutagenicity, cytotoxicity and teratogenicity 43 . 44 Moreover, investigations at multiple levels, including cellular studies, in vivo experiments and preclinical assessments, are also necessary.

In addition, the relatively high IC50 value observed for acarbose in the present study is not without precedent. Our study followed a conventional and reliable colorimetric spectrophotometric method, 45 which has already been established, standardized and applied in our laboratory in previous studies for the evaluation of natural products and active fractions. 11 Published reports show that acarbose exhibits substantial assay dependent variability in α‐glucosidase inhibition, 46 especially when studies utilize the same S. cerevisiae‐derived α‐glucosidase and/or method based on pNPG as substrate (IC50 range from 1.03 to 4.67 mg mL−1).47, 48, 49 This variability has been attributed to differences in enzyme source, substrate identity, reaction conditions and analytical method. 45 In addition, in the present study, to further indicate the potential as an α‐glucosidase inhibitor, OA was also interpreted in relation to palmitic acid, a known natural α‐glucosidase inhibitor, rather than only against acarbose. Therefore, the stronger inhibitory activity of OA relative to acarbose observed in the present study should be interpreted as a valid assay specific finding, whereas broader cross study comparisons should be made with caution.

Finally, although OA exhibited promising α‐glucosidase inhibitory activity in the present study, the findings were obtained using yeast α‐glucosidase and an in vitro experimental system only. Therefore, the biological relevance and translational potential of OA should be interpreted with caution. Further validation is needed using more physiologically relevant enzyme systems, particularly mammalian or human intestinal α‐glucosidase related enzymes such as sucrase, maltase and isomaltase. In addition, cell‐based studies using intestinal epithelial models, such as Caco‐2 or HT 29 cells, would help clarify its functional effects and cellular safety. Evaluation in normal cell lines, including intestinal, hepatic or fibroblast derived cells, would also be important to assess cytotoxicity. Furthermore, in vivo studies in diabetic or insulin resistant animal models, such as streptozotocin‐induced rodents, high‐fat diet induced models or db/db mice, together with acute and subchronic toxicity studies, are required before any therapeutic or nutraceutical application can be more confidently proposed.

CONCLUSIONS

To the best of our knowledge, the present study study is the first to report the α‐glucosidase inhibitory activity of (9Z)‐5,8,11‐trihydroxyoctadec‐9‐enoic acid (OA) enriched fraction. The OA‐enriched fraction exhibited potent inhibitory effects at substantially lower concentrations than standard acarbose and other reference inhibitors. Molecular docking and enzyme kinetic analyses consistently confirmed that OA functions as a mixed‐type inhibitor binding to the active site, and exhibiting a stronger binding affinity for the free enzyme than for the enzyme–substrate complex. However, the observed reduction in bioaccessibility and bioavailability of OA after simulated digestion indicates limited accessibility in the gastrointestinal tract. Further studies should focus on evaluating the in vivo bioaccessibility, stability, toxicity and bioavailability of the OA enriched fraction to validate its potential as an α‐glucosidase inhibitor.

AUTHOR CONTRIBUTORS

NPM was responsible for validation, data curation, formal analysis, writing and reviewing. NVQ and TDX were responsible for methodology. NVQ and TDX were responsible for conceptualization. NVQ was responsible for visualization. NVQ and NDD were responsible for writing and editing. NVQ and TDX were responsible for supervision. NDD was responsible for reviewing. TDX was responsible for writing. All authors have read and approved the final version of the manuscript submitted for publication.

FUNDING INFORMATION

No funding was received for this study.

Supporting information

Data S1. Preparation of chemicals for in vitro digestion.

Data S2. Toxicity assessment and validation of method.