In silico modeling and in vitro evaluation of antidiabetic potential of exopolysaccharides from Brassicaceae root endophytic bacteria Authors: Kusmiati Kusmiati, Arif Nurkanto, Asrul Fanani, Jendri Mamangkey, Dian Alfian Nurcahyanto, Khatarina Meldawati Pasaribu, Indriati Ramadhani, Ade Lia Putri, Fifi Afiati, Ni Wayan S Agustini, Herman Irawan, Dody Priadi, Yelin Adalina, Muhammad Taufik, Silvia Mandasari, Hasnadhiazahra Rohadi Published: 2026 ABSTRACT AbstractEndophytic bacteria associated with Brassicaceae roots produce extracellular polysaccharides (EPS) that may inhibit carbohydrate-hydrolyzing enzymes related to antidiabetic properties. This study evaluated EPS produced by Brassicaceae root endophytes for inhibitory activity against carbohydrate-digesting enzymes and examined putative ligand–enzyme interactions using molecular docking. Endophytic bacteria were isolated from roots of cabbage, broccoli, Chinese broccoli, and pakcoy, phenotypically characterized, and identified by 16S rRNA gene sequencing. Crude EPS were extracted from bacterial cultures and quantified as dry weight, with total carbohydrate (glucose equivalents) and protein contents determined for compositional comparison. α-Glucosidase inhibitory activity was assessed in vitro using p-nitrophenyl-α-D-glucopyranoside as the substrate, and IC50 values were calculated from dose–response curves with acarbose as a positive control. In silico docking was performed against α-glucosidase (PDB: 2QMJ) using EPS-related monosaccharides, and pose reliability was evaluated by RMSD. Ten isolates produced EPS with yields ranging from 76.7 to 309.7 mg L−1. EPS-associated glucose equivalents ranged from 22.7 to 68.8 mg L−1 and protein from 22.8 to 46.7 mg L−1, indicating isolate/strain-dependent compositional variation. Two crude EPS samples showed the strongest α-glucosidase inhibition, with IC50 values of 121.16 ppm (Pseudomonas plecoglossicida AK8A4) and 168.72 ppm (Neobacillus drentensis AK4B2), whereas the remaining samples showed IC50 values >200 ppm; acarbose yielded an IC50 of 79.79 ppm. Docking produced stable poses (RMSD <2 Å) and predicted stronger binding for acarbose than for individual monosaccharides; among the latter, D-mannose showed the most favorable energy against α-glucosidase. Brassicaceae root bacterial endophytes produced crude EPS with potential α-glucosidase inhibitory activity, supporting further purification and structure–activity analysis to define active fractions and inhibitory mechanisms.Keywords: Exopolysaccharides, Endophytic bacteria, Brassicaceae, α-Glucosidase inhibition, Molecular docking Introduction Diabetes mellitus is a chronic metabolic disorder characterized by hyperglycemia caused by impaired insulin secretion, insulin action, or both. In 2024, approximately 589 million adults aged 20–79 years were living with diabetes worldwide, and this number is projected to reach 853 million by 2050.1 Persistent hyperglycemia increases the risk of cardiovascular, renal, neurological, and other chronic complications.2 One approach to controlling postprandial hyperglycemia is to inhibit carbohydrate-hydrolyzing enzymes. Inhibition of intestinal α-glucosidase delays the conversion of oligosaccharides and disaccharides into absorbable monosaccharides, thereby reducing the postprandial rise in blood glucose.3 Microbial exopolysaccharides (EPS) are extracellular carbohydrate polymers that vary widely in monosaccharide composition, glycosidic linkage, branching, molecular mass, and functional groups.4 Bacterial EPS are generally high-molecular-mass polymers, although their molecular masses vary among strains and recovery methods. In lactic acid bacteria (LAB), reported values ranged from less than 10 kDa to more than 6000 kDa.5 The structural differences may affect their physicochemical properties and biological activity. Some bacterial EPS have shown α-glucosidase inhibitory activity in vitro. For example, purified EPS1–1 from Lacticaseibacillus rhamnosus LB1lac10 inhibited α-glucosidase, and its activity was associated with its monosaccharide composition and glycosidic structure.6 However, the activity reported for certain EPS cannot be extended to other polymers because EPS composition and function are strongly strain-dependent. Endophytic bacteria are a useful source of EPS because microbial fermentation permits controlled and reproducible production, while the specialized plant-associated niche may support the formation of structurally diverse polymers.7, 8 Endophytic bacteria produce EPS that contribute to surface attachment, stress tolerance, and colonization of plant tissues. For example, EPS produced by the endophytic bacterium Paraburkholderia phytofirmans PsJN was required for optimal stress tolerance and plant colonization.9 Bacterial EPS also contributed to salt-stress alleviation as in the case of the rice endophyte Pantoea alhagi NX-11.10 A recent study further showed that a Burkholderia sp. isolated from Grifola frondosa produced a mannose-rich EPS at high yield under controlled fermentation and that the purified polymer had antibacterial and prebiotic properties.11 These findings support further investigation of endophytic bacteria as sources of EPS with distinct α-glucosidase inhibitory activity. Members of the Brassicaceae are widely recognized as sources of health-promoting phytochemicals with potential antidiabetic activity. Sulforaphane and related organosulfur compounds derived from broccoli and other cruciferous vegetables have been reported to improve glycemic control, reduce hepatic glucose production, and alleviate insulin resistance in patients with type 2 diabetes.12, 13, 14, 15 Several Brassicaceae microgreens have also demonstrated inhibitory activity against α-amylase and α-glucosidase, further supporting the importance of this plant family as a source of compounds capable of regulating postprandial carbohydrate metabolism.16 In parallel, endophytic microorganisms associated with Brassicaceae species have increasingly been investigated as sources or modulators of biologically active metabolites. For example, inoculation with an endophytic Jeotgalicoccus strain isolated from Eruca sativa enhanced the plant's α-amylase- and α-glucosidase-inhibitory activities,17 whereas an endophytic Enterobacter hormaechei isolated from Alliaria petiolata produced secondary metabolites previously associated with antidiabetic activity.18 However, studies have largely focused on plant phytochemicals, microbial secondary metabolites, and endophyte-mediated enhancement of host bioactivity. EPS from Brassicaceae-associated bacterial endophytes remain poorly characterized, particularly as α-glucosidase inhibitors, highlighting their potential as novel antidiabetic carbohydrate-based metabolites. Computational approaches can support the preliminary evaluation of EPS bioactivity by predicting the interaction of representative structural units with α-glucosidase. Molecular docking has been increasingly applied to natural products to estimate binding affinity, identify key interacting residues, and prioritize candidates for subsequent experimental validation.19, 20, 21 Combining molecular docking with biochemical assays can provide preliminary information on enzyme inhibition and predicted ligand–enzyme interactions, which could lead to further phytocompound optimization or development.22, 23, 24, 25 This study then aimed to evaluate the α-glucosidase inhibitory activity of crude EPS fractions produced by endophytic bacteria isolated from Brassicaceae roots as the preliminary investigation and to examine the interactions of representative carbohydrates with α-glucosidase through molecular docking. Materials and methods Materials Media: nutrient broth (Merck), agar (Merck), and Mueller-Hinton medium (Merck). Chemicals and reagents: sodium hydroxide; potassium sodium tartrate; copper sulfate; sodium chloride; Folin–Ciocalteu reagent; sodium hypochlorite (5% stock); ethanol (90%); distilled water; acetone; NaOH (5 N); hydrogen peroxide (2%); potassium bromide (KBr); Gram-staining reagents (crystal violet, iodine, safranin, and 70% alcohol); immersion oil; and ethyl acetate. Isolation and characterization of endophytic bacteria Healthy Brassicaceae roots (cabbage, broccoli, kailan, and pak choy) were collected local agricultural sites in Bogor, West Java, Indonesia. The roots were washed under running tap water to remove adhering soil. Surface sterilization was performed by immersion in 70% ethanol for 5 min followed by 3% (v/v) sodium hypochlorite for 5 min. The roots were rinsed several times with sterile distilled water, cut into 1–2 cm segments, and macerated in phosphate-buffered saline until homogeneous. The suspension was serially diluted to 10−6, and aliquots were streaked onto nutrient agar plates. Plates were incubated at 28 °C for 24–48 h. Morphologically distinct colonies were selected and purified by repeated streaking. Isolates were characterized by Gram staining and a catalase test. DNA extraction, PCR, sequencing, and phylogenetic analysis Crude DNA templates were prepared from bacterial biomass grown on agar using a heat-lysis method. Briefly, bacterial cells were suspended in sterile water, heated at 98 °C for 10 min, and cooled at 4 °C for 10 min. Cellular debris was removed by centrifugation, and the resulting supernatant was used as the PCR template. The nearly full-length 16S rRNA gene was amplified using the universal bacterial primers 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5′-GGTTACCTTGTTACGACTT-3′). PCR products were verified by agarose gel electrophoresis, purified, and sequenced in both directions using an ABI 3730XL DNA Analyzer (Applied Biosystems). Forward and reverse chromatograms were quality-checked, trimmed to remove low-quality regions, and assembled into consensus sequences using BioEdit v7.2.5.25, 26 Taxonomic identification was performed by comparing the consensus sequences with reference sequences in the NCBI nucleotide database using BLASTn. Species assignments were based on the highest sequence identity and query coverage among validly identified reference strains. EPS production and extraction Each endophytic isolate was cultured in nutrient broth at 30 °C for 48 h with shaking. Cells were removed by centrifugation (3600 rpm, 30 min). The supernatant was treated with activated carbon at 40 °C for 30 min to reduce pigmentation and was then concentrated by rotary evaporation. Proteins were removed by extraction with chloroform–n-buthanol (4,1, v/v). EPS was precipitated by addition of cold ethanol followed by overnight incubation. The precipitate was collected for subsequent analyses.27 Total carbohydrate was quantified using the phenol–sulfuric acid method,28 and total protein was quantified using the Lowry method.29, 30, 31 Total carbohydrate determination in crude EPS Total carbohydrate content was determined using the phenol–sulfuric acid method. A glucose calibration curve was prepared using standard solutions at concentrations of 10, 20, 40, 60, 80, and 100 ppm. Briefly, 100 μL of each glucose standard or appropriately diluted EPS sample was mixed with 50 μL of 5% (w/v) phenol, followed by the rapid addition of 250 μL of 96% sulfuric acid. The reaction mixture was allowed to stand for 10 min, heated in a boiling water bath for 15 min, and subsequently cooled to room temperature. Absorbance was measured at 490 nm using a Varioskan LUX multimode microplate reader (Thermo Fisher Scientific, Waltham, MA, USA). A reagent blank containing distilled water instead of the standard or sample was prepared in parallel. Total carbohydrate concentration was calculated from the linear regression equation of the glucose calibration curve, (y = a + bx), and expressed as glucose equivalents. Protein determination in crude EPS Protein content was determined using the Lowry method with bovine serum albumin (BSA) as the reference standard. A calibration curve was prepared from a 1000 ppm BSA stock solution using working standards ranging from 10 to 80 ppm. Briefly, 200 μL of each BSA standard or appropriately diluted EPS sample was mixed with 200 μL of 1 N NaOH, heated for 20 min, and cooled to room temperature. The alkaline copper reagent was freshly prepared by mixing 5% (w/v) sodium carbonate, 1% (w/v) copper sulfate, and 2% (w/v) potassium sodium tartrate at a volumetric ratio of 25:1:1. Subsequently, 200 μL of the alkaline copper reagent was added to each reaction mixture, followed by incubation for 10 min at room temperature. Then, 200 μL of Folin–Ciocalteu reagent was added, and the mixture was incubated for 30 min before absorbance was measured at 750 nm using a Varioskan LUX multimode microplate reader (Thermo Fisher Scientific, Waltham, MA, USA). A reagent blank containing distilled water instead of the standard or sample was prepared in parallel. Protein concentration was calculated from the linear regression equation of the BSA calibration curve, (y = a + bx), and expressed as protein equivalents. In vitro α-glucosidase inhibition assay α-Glucosidase inhibitory activity was evaluated as described by Salehi et al.,32 with minor modifications. For initial screening, crude EPS fractions and acarbose were tested at final concentrations of 500 and 100 μg/mL (ppm), respectively. For IC50 determination, each EPS fraction (100–1000 ppm) and acarbose (25–400 ppm) were tested over their respective concentration ranges. Acarbose (100 ppm) was used as a positive control. In a 96-well plate, 50 μL of EPS solution was mixed with 20 μL of phosphate buffer (pH 6.8) and 15 μL of p-nitrophenyl-α-D-glucopyranoside (pNPG; 5 mM). The mixture was pre-incubated at 37 °C for 5 min. The reaction was initiated by addition of 15 μL of α-glucosidase and was incubated at 37 °C for 15 min. The reaction was terminated by addition of sodium carbonate (Na2CO3), and absorbance was recorded at 405 nm. Percentage inhibition was calculated as: One unit of α-glucosidase activity was defined as the amount of enzyme that releases 1 μmol p-nitrophenol per minute from pNPG under the assay conditions. Measurements at each concentration were performed in triplicate and are presented as mean ± SD. IC50 values were estimated from nonlinear regression of the mean percentage inhibition against concentration. These triplicate measurements represented technical replicates and did not provide independent IC50 estimates; therefore, statistical comparisons among IC₅₀ values were not performed. Molecular docking Molecular docking was performed using AutoDockTools v1.5.6. Because the monosaccharide composition and structural characteristics of the crude EPS fractions were not experimentally determined, monosaccharides commonly reported as constituents of microbial EPS were selected as representative ligands for a preliminary monomer-level assessment of potential ligand–enzyme interactions. Eight ligands were included in the molecular docking analysis: acarbose, α-D-galacturonic acid, D-glucosamine, d-glucose, D-glucuronic acid, D-mannose, maltose, and raffinose. Acarbose was used as the reference α-glucosidase inhibitor, whereas the remaining ligands represented neutral sugars, an amino sugar, uronic acids, and small oligosaccharides selected for comparative evaluation of carbohydrate–enzyme interactions.20, 33, 34 The three-dimensional structures of all ligands were retrieved from PubChem in SDF format and geometry-optimized using Avogadro before molecular docking.35, 36 Crystal structure of α-glucosidase (PDB: 2QMJ) were retrieved from the RCSB Protein Data Bank.37 Protein preparation included removal of crystallographic water molecules and non-essential ligands, followed by addition of polar hydrogens and assignment of Gasteiger charges. Docking performance was validated by redocking the native ligands into the corresponding binding sites. Test ligands were docked under the same parameters, and predicted binding was compared with that of acarbose as the reference inhibitor. Protein–ligand interactions were visualized in 2D and 3D using LigPlot+, Discovery Studio Visualizer v16.1.0.15350, and PyMOL v2.5.5.38, 39, 40 Results and discussion Phenotypic and 16S rRNA-based identification of endophytic bacteria from Brassicaceae roots Endophytic bacteria isolated from Brassicaceae roots were characterized by colony morphology, Gram staining, catalase activity, and 16S rRNA gene sequencing (Table 1). A total of 10 isolates were obtained from cabbage (1), broccoli (4), Chinese broccoli (1), and pakcoy (4) roots. All isolates were catalase-positive, whereas Gram reactions were split between three Gram-positive isolates (3/10) and eight Gram-negative isolates (8/10). The Gram-positive fraction comprised Bacillus/Neobacillus members, specifically Bacillus toyonensis (AK3A1; cabbage) and Neobacillus drentensis (AK4B2; broccoli), together with an additional Bacillus isolate from broccoli identified as Bacillus altitudinis (AK5B1). In contrast, the Gram-negative isolates were dominated by Proteobacteria and Acinetobacter spp. Broccoli yielded two Gram-negative taxa, identified as Pseudomonas granadensis (AK5A2) and Enterobacter asburiae (AK5A4), while Chinese broccoli yielded Acinetobacter calcoaceticus (AK6A1). Pakcoy showed the greatest diversity among the sampled hosts, producing four Gram-negative isolates assigned to Pseudomonas fluorescens (AK7A2), Enterobacter huaxiensis (AK7A4), Acinetobacter haemolyticus (AK8A3), and Pseudomonas plecoglossicida (AK8A4). Sequence-based identification showed very high similarity to reference strains, with nine isolates showing 100% 16S rRNA sequence resemblance and one isolate (AK7A4) showing 99.93% resemblance to Enterobacter huaxiensis. The GenBank accession numbers are provided in Table 1, supporting their similarity-based taxonomic identification. Table 1Phenotypic characterization and 16S rRNA gene-based identification of endophytic bacteria isolated from roots of Brassicaceae plants.IsolateSample sourceGram/CatalaseClosest BLASTn matchSequence identityGenBank accession numberAK3A1Cabbage roots+/+Bacillus toyonensis strain T7100%MW786726.1AK4B2Broccoli roots+/+Neobacillus drentensis strain 23,325100%OR429348.1AK5A2Broccoli roots−/+Pseudomonas granadensis strain FJT3100%MG571665.1AK5A4Broccoli roots−/+Enterobacter asburiae strain TPY8100%KT766080.1AK5B1Broccoli roots+/ +Bacillus altitudinis strain DES103G100%OQ092432.1AK6A1Kailan roots−/ +Acinetobacter calcoaceticus strain LZH-E31100%PP563811.1AK7A2Pak coy roots−/ +Pseudomonas fluorescens strain psf3100%MN256389.1AK7A4Pak coy roots−/ +Enterobacter huaxiensis strain 09000899.93%CP043342.1AK8A3Pak coy roots−/ +Acinetobacter haemolyticus strain ZH1100%KX817287.1AK8A4Pak coy roots−/ +Pseudomonas plecoglossicida strain ZXDC-18100%GQ398824.1 Phenotypic characterization and 16S rRNA gene-based identification of endophytic bacteria isolated from roots of Brassicaceae plants. The culturable endophytic bacteria isolated from Brassicaceae roots comprised taxa frequently reported from vegetable endospheres. Gram-negative isolates predominated (8/10), and catalase activity was detected in all isolates. This composition is consistent with culture-dependent surveys of Brassica roots in which Proteobacteria and Bacillus-related lineages constitute major recoverable fractions of the endosphere.41, 42 The concurrent detection of Bacillus/Neobacillus with Pseudomonas, Enterobacter, and Acinetobacter further agrees with the widespread occurrence of these genera as plant-associated endophytes and with their frequent association with plant-beneficial traits.43 Several isolates were assigned to genera with established roles in plant performance. In particular, Pseudomonas spp. have been associated with growth promotion and disease suppression via multiple mechanisms, including iron sequestration through siderophore production and induction of host defense responses; P. fluorescens strains have also been reported to prime systemic resistance in planta.44 The uniform catalase-positive phenotype observed across isolates is compatible with selection for oxidative stress tolerance during endophytic colonization, given that reactive oxygen species are generated in roots during basal defense and stress signaling.45 EPS production and yields Extracellular polysaccharide production differed significantly among the tested isolates based on an ANOVA (F9,20 = 443.52; p ≤ 0.001) as depicted in Fig. 1. The dry weight of EPS from Brassicaceae-associated root bacterial endophytes ranged from 76.7 to 309.7 mg L−1. The highest yields were produced by Enterobacter asburiae AK5A4 (309.7 mg L−1) and Pseudomonas fluorescens AK7A2 (307.0 mg L−1), whereas lower yields were detected for Bacillus altitudinis AK5B1 (76.7 mg L−1). In plant root systems, bacterial EPS are recognized as major constituents of the biofilm matrix and are associated with initial surface attachment, microcolony development, and persistence during plant-associated colonization.46 The predominance of Gram-negative isolates with high EPS-associated activity is consistent with reports that root endophytic communities are frequently dominated by Proteobacteria and that these taxa are well adapted for root-associated lifestyles.47 Because EPS biosynthesis is regulated by defined genetic determinants, the inherent capacity for matrix production after isolation in laboratory culture can be interpreted as evidence of retained biosynthetic potential, although expression levels are expected to remain responsive to growth conditions.48 EPS production by endophytic bacterial isolates. Bars represent the dry weight of EPS (mg L−1) quantified for each isolate under the assay conditions. Data are presented as the mean of three independent replicates (n = 3) with error bars indicating standard deviation. Statistical differences among isolates were evaluated by one-way ANOVA, followed by Tukey's HSD multiple-comparisons test; isolates not sharing a common letter (a–f) are significantly different at p < 0.05. EPS characterization The soluble carbohydrate (expressed as glucose equivalents, F9,20 = 95.81; p ≤ 0.001) and protein (F9,20 = 5.51; p ≤ 0.001) contents associated with the EPS fraction differed significantly among isolates (Table 2). Glucose equivalents ranged from 22.7 to 68.8 mg L−1, with the highest values detected for AK8A3 (68.8 mg L−1) and AK4B2 (67.1 mg L−1), whereas the lowest value was measured for AK3A1 (22.7 ± 0.54 mg L−1). Protein concentrations ranged from 22.8 to 46.7 mg L−1 and were highest for AK7A4 (46.7 mg L−1), followed by AK8A3 (43.3 mg L−1). When expressed as the carbohydrate-to-protein ratio, values ranged from 0.91 to 1.99, indicating isolate-specific differences in EPS-associated composition; the highest ratio was observed for Neobacillus drentensis AK4B2 (∼1.99), whereas the lowest ratio was observed for P. fluorescens AK7A2 (∼0.91). EPS composition varied among isolates, as expected for extracellular matrices that contain polysaccharides alongside measurable amounts of proteins and other macromolecules, with the final proportions shaped by strain-specific traits and culture conditions. In Table 2, carbohydrate and protein did not show a parallel trend across isolates: AK4B2 and AK8A3 showed significantly high carbohydrate contents, whereas the highest protein concentration was detected in AK7A4. This pattern supports the interpretation that EPS should not be treated as a carbohydrate-only product. Rather, proteins can comprise a major fraction of the matrix and may contribute to structural integrity, cell-surface interactions, and the localization or retention of extracellular enzymes, depending on the organism and the growth environment.49, 50, 51 Table 2Soluble carbohydrate (glucose equivalents) and protein contents of EPS produced by endophytic bacterial isolates. Values are presented as mean ± SD. Different superscript letters within a column indicate significant differences among isolates (p < 0.05) by Tukey's HSD multiple-comparisons test.IsolatesGlucose equivalents (mg L−1)Protein (mg L−1)Ratio (Carbohydrate-to-protein)AK3A122.7 ± 0.54d22.8 ± 2.51c0.999AK4B267.1 ± 1.74a33.8 ± 12.66abc1.983AK5A239.9 ± 3.04c32.5 ± 3.68abc1.226AK5A456.1 ± 0.95b37.1 ± 0.51abc1.513AK5B130.9 ± 1.77d25.7 ± 0.76c1.202AK6A144.9 ± 6.04c35.0 ± 1.94abc1.283AK7A227.2 ± 1.59d29.9 ± 8.85bc0.907AK7A454.2 ± 1.76b46.7 ± 4.33a1.160AK8A368.8 ± 1.75a43.3 ± 0.84ab1.589AK8A440.1 ± 4.35c34.3 ± 1.41abc1.167 Soluble carbohydrate (glucose equivalents) and protein contents of EPS produced by endophytic bacterial isolates. Values are presented as mean ± SD. Different superscript letters within a column indicate significant differences among isolates (p < 0.05) by Tukey's HSD multiple-comparisons test. In vitro inhibition of α-glucosidase by crude EPS Crude EPS fractions produced by endophytic bacteria isolated from Brassicaceae roots showed specific α-glucosidase inhibition, a standard target for reducing postprandial glucose release from dietary carbohydrates. In vitro α-glucosidase assays are commonly conducted using a spectrophotometric method with the synthetic substrate p-nitrophenyl-α-D-glucopyranoside (pNPG). In this system, inhibition of α-glucosidase decreases the release of p-nitrophenol, which is quantified at 405 nm. Crude EPS from ten isolates produced varying α-glucosidase inhibition, with two isolates showing moderate inhibitory potency (IC50 within 100–200 ppm). The EPS from AK4B2 and AK8A4 produced IC50 values of 168.72 ppm and 121.16 ppm, respectively, whereas the remaining eight EPS samples showed IC50 values >200 ppm. Under the same assay conditions, acarbose (positive control) produced an IC50 of 79.79 ppm which is the highest among tested extracts. The results are summarized in Table 3. Studies on microbial endophytes have reported inhibition of carbohydrate-hydrolyzing enzymes by EPS-containing fractions and other microbial extracts. To our knowledge, no standardized IC50 thresholds have been established for classifying the potency of microbial EPS as α-glucosidase inhibitors. α-Glucosidase inhibitory activities have also been reported for EPS from other microorganisms. Shao et al.52 found that EPS from Penicillium janthinellum N29 produced 70.52% inhibition at 5 mg/mL, although the activity was slightly lower than that of acarbose at the same concentration. From marine environment, Alharbi et al.53 reported the α-glucosidase inhibitor by the bacterial EPS (EPSR9) with an IC50 value of 26.73 μg/mL from Bacillus rugosus SYG20. Ghareeb et al.54 reported IC50 value of 6.48 μg/mL against α-glucosidase for EPS produced by a marine actinobacterium Streptomyces vinaceusdrappus AMG31. Direct comparisons among these findings are limited by differences in microbial source, EPS purity and structure, enzyme source, substrate, concentration range, and assay conditions. Residual phenolic compounds may also contribute to the activity of partially purified EPS preparations,55 although phenolic content was not measured in the present study. The hydroxyl groups of polysaccharides may support hydrogen bonding with amino acid residues, whereas carboxyl and sulfate groups in acidic or sulfated polysaccharides may contribute to charge-dependent interactions with the enzyme.56, 57, 58 Further purification, structural characterization, enzyme-kinetic analysis, and safety testing are needed to determine whether microbial EPS have practical value as inhibitors of carbohydrate-hydrolyzing enzymes. Table 3In vitro α-glucosidase inhibitory activity of crude EPS produced by Brassicaceae-root-endophytic bacterial isolates. IC50 values were calculated from concentration-response curves. Percentage inhibition values are presented as mean ± SD. a) Acarbose or positive control was tested at 100 ppm.Bacterial EPS extract(s)α-Glucosidase inhibition% Inhibition at 500 ppma)IC50 (ppm)AK3A150.82 ± 0.58500.68AK4B260.70 ± 1.42168.72AK5A262.45 ± 0.45316.18AK5A456.03 ± 0.38347.60AK5B164.78 ± 0.71214.36AK6A147.51 ± 0.43583.12AK7A247.08 ± 0.45587.31AK7A449.41 ± 1.16525.68AK8A349.22 ± 1.03522.35AK8A462.25 ± 0.19121.16Acarbosea)59.72 ± 0.2579.79 In vitro α-glucosidase inhibitory activity of crude EPS produced by Brassicaceae-root-endophytic bacterial isolates. IC50 values were calculated from concentration-response curves. Percentage inhibition values are presented as mean ± SD. a) Acarbose or positive control was tested at 100 ppm. Molecular docking analysis Ribbon representations of the protein receptor used for in silico EPS-protein interaction analyses are shown in Fig. 2A. The protein structure (PDB: 2QMJ) was prepared as docking targets to evaluate the predicted binding of representative carbohydrate ligands. The protein showed mixed secondary-structure architecture, with α-helices (cyan) and β-sheets (red) connected by loop regions (magenta), providing multiple surface-exposed regions that may contribute to ligand accommodation and interaction during inhibition screening. Before the test ligands were docked, the protocol was validated by independently redocking the native ligand, acarbose, 10 times into the active site of α-glucosidase (PDB ID: 2QMJ). Validation of acarbose redocking against α-glucosidase (PDB ID: 2QMJ): (A) protein structure and (B) superimposed crystallographic and redocked poses. Docking was performed using AutoDock Vina with a grid box centered at x = −21.727, y = −6.323, and z = −5.281 Å and dimensions of 15.0 × 15.0 × 16.5 Å. Exhaustiveness and the number of output poses were set to 8 and 20, respectively. The 10 runs produced a mean RMSD of 1.76 Å, below the accepted threshold of 2.0 Å, indicating that the protocol reproducibly recovered the crystallographic binding pose. The superimposed crystallographic and representative redocked poses are shown in Fig. 2B. Following protocol validation, seven test ligands were docked into the α-glucosidase binding site, with acarbose used as the reference ligand. The predicted binding energies are summarized in Table 4. Acarbose showed the most favorable value (−7.8 kcal/mol), followed by raffinose (−7.0 kcal/mol), maltose (−6.6 kcal/mol), D-mannose (−6.0 kcal/mol), d-glucose and D-glucuronic acid (−5.6 kcal/mol), D-glucosamine (−5.5 kcal/mol), and α-galacturonic acid (−5.2 kcal/mol). The monosaccharides showed predicted binding energies ranging from −5.2 to −6.0 kcal/mol, with D-mannose producing the most favorable value. Maltose and raffinose showed more favorable docking scores than the monosaccharides, possibly because their larger structures provide greater contact surfaces and more opportunities for hydrogen bonding and other non-covalent interactions within the binding site. Acarbose, a larger pseudo-oligosaccharide, showed the strongest predicted binding. However, these differences cannot be attributed solely to saccharide chain length because the ligands also differ in monosaccharide composition, glycosidic linkages, conformation, and functional-group arrangement. Docking scores may also favor larger ligands that form more protein contacts. Therefore, maltose and raffinose should be regarded as simplified oligosaccharide models rather than direct representations of the isolated EPS. The results indicate differences in predicted α-glucosidase interactions among the tested carbohydrates but do not establish the binding affinity or inhibitory activity of intact EPS. Table 4Predicted binding energies of ligands docked to α-glucosidase (PDB: 2QMJ).Ligandα-Glucosidase (kcal/mol)Acarbose−7.8α-Galacturonic acid−5.2D-Glucosamine−5.5d-Glucose−5.6D-Glucuronic acid−5.6D-Mannose−6.0Maltose−6.6Raffinose−7.0 Predicted binding energies of ligands docked to α-glucosidase (PDB: 2QMJ). Comparative analysis of ligand interactions with α-glucosidase showed that all tested ligands formed hydrogen bonds with residues within the binding site, including Asp327, Asp443, Asp542, Arg526, and His600, although the number and distance of these interactions varied among ligands (Table 5, Fig. 3, Fig. 4). Acarbose formed three hydrogen bonds with Asn207, His600, and Arg526 at distances of 2.80–3.34 Å. Its binding pose was further supported by nonpolar contacts with Trp406, Trp441, Met444, Trp539, and Phe575. Among the monosaccharides, D-mannose and D-glucuronic acid each formed seven hydrogen bonds, while d-glucose, D-glucosamine, and α-galacturonic acid formed six, six, and five hydrogen bonds, respectively. These ligands also formed nonpolar contacts with residues such as Tyr299, Trp406, Trp441, and Phe575. Ligand-specific contacts were observed between D-glucosamine and Asp203 and between D-mannose and Ile328, indicating differences in ligand orientation and functional-group positioning within the binding site. Table 5Comparative analysis of ligand interactions with 2QMJ protein.Ligand(s)Number of H-bondsH-bond distance range (Å)Key interacting residuesKey hydrophobic interactionsAcarbose32.80–3.34Asn207, His600, Arg526Trp441, Trp406, Met444, Trp539, Phe575α-Galacturonic acid52.67–3.23Asp327, Asp542, Arg526, Asp443, His600Trp441, Tyr299, Trp406, Phe575, Met444D-Glucuronic acid72.70–3.21Asp327, Asp542, Arg526, Asp443Trp441, Tyr299, Trp406, Phe575, Met444, Trp539D-Glucosamine62.84–3.25Asp542, Arg526, Asp443, His600, Asp203Trp441, Tyr299, Trp406, Phe575, Met444, Trp539d-Glucose62.81–3.22Asp327, Asp542, Arg526, Asp443, His600Trp441, Tyr299, Trp406, Phe575, Met444, Trp539D-Mannose72.70–3.23Asp327, Asp542, Arg526, Asp443, His600, Ile328Trp441, Tyr299, Trp406, Phe575, Met444Maltose62.62–3.33Asp203, Asp327, Met444, Arg526, Asp542, His600Tyr299, Ile328, Trp406, Trp441, Asp443, Trp539, Phe575Raffinose72.57–3.30Asp203, Asp327, Ser448, Asn449, Arg526, Asp542, His600Tyr299, Ile328, Ile364, Trp406, Trp441, Asp443, Met444, Phe450, Lys480, Phe575 Comparative analysis of ligand interactions with 2QMJ protein. Visualization using a Liplot regarding the complex interaction of EPS monomers and protein 2QMJ. (a) α-Galacturonic-acid (b) D-Glucosamine (c) d-Glucose (d) D-Glucuronic acid (e) D-Mannose (f) Maltose and (g) Raffinose. 3D Visualization of binding Ligand and protein 2QMJ. (a) α-Galacturonic-acid (b) D-Glucosamine (c) d-Glucose (d) D-Glucuronic acid (e) D-Mannose (f) Maltose and (g) Raffinose. The monosaccharides formed more hydrogen bonds than acarbose, yet their docking scores were still less favorable. The Vina score accounts for steric fit, hydrophobic contacts, hydrogen bonding, and ligand flexibility rather than hydrogen-bond number alone.59 The contribution of each hydrogen bond also depends on its geometry and the desolvation of the interacting polar groups, which may reduce the net energetic benefit of bond formation.60 Acarbose occupies a larger region of the binding pocket and contacts more aromatic and nonpolar residues, which may explain its more favorable score despite having fewer hydrogen bonds. Because docking scores may favor larger ligands with greater contact surfaces,61 hydrogen-bond counts and Vina scores describe different features of the predicted binding pose and should not be regarded as direct measures of experimental affinity. These in vitro and in silico findings support further investigation of microbial EPS as inhibitors of carbohydrate-hydrolyzing enzymes. Purification, structural characterization, enzyme-kinetic analysis, and safety testing are required to determine their biological relevance. Conclusions Brassicaceae root endophytes produced crude EPS fractions with isolate-dependent variation in yield and macromolecular composition. Among the fractions tested, AK8A4 and AK4B2 showed the greatest α-glucosidase inhibition, although both were less potent than acarbose. Docking indicated that the tested carbohydrates could be accommodated within the binding sites of α-glucosidase (2QMJ), with acarbose showing the most favorable predicted binding energy. However, the monosaccharides and representative oligosaccharides used for docking do not fully represent the structural complexity and multivalent interactions of intact EPS; therefore, these results are still preliminary and may not be directly linked to the activity of the crude fractions. Further studies should include EPS purification and structural characterization, assessment of co-extracted components, enzyme kinetics, and molecular dynamics simulations of experimentally defined EPS fragments. Cytotoxicity and other safety assessments are also required before these fractions can be considered for functional or therapeutic applications. CRediT authorship contribution statement Kusmiati Kusmiati: Writing – review & editing, Writing – original draft, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Arif Nurkanto: Methodology, Investigation, Formal analysis, Data curation. Asrul Fanani: Validation, Software, Methodology, Formal analysis, Data curation. Jendri Mamangkey: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation, Data curation. Dian Alfian Nurcahyanto: Supervision, Resources, Methodology, Investigation, Data curation. Khatarina Meldawati Pasaribu: Supervision, Methodology, Investigation, Data curation. Indriati Ramadhani: Resources, Investigation, Formal analysis, Data curation. Ade Lia Putri: Resources, Methodology, Investigation, Data curation. Fifi Afiati: Visualization, Methodology, Investigation, Data curation. Ni Wayan S. Agustini: Supervision, Resources, Methodology, Investigation, Conceptualization. Herman Irawan: Visualization, Validation, Supervision, Methodology, Investigation. Dody Priadi: Validation, Methodology, Formal analysis, Data curation. Yelin Adalina: Supervision, Resources, Investigation, Formal analysis. Muhammad Taufik: Validation, Supervision, Software, Investigation. Silvia Mandasari: Resources, Methodology, Investigation, Data curation. Hasnadhiazahra Rohadi: Software, Investigation, Formal analysis, Data curation. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments The authors sincerely thank the National Research and Innovation Agency (BRIN) for providing research facilities and the Indonesia Endowment Fund for Education (LPDP) for financial support through the Riset dan Inovasi untuk Indonesia Maju (RIIM) Program under Contract Nos. 7501/II.7.5/FR.06.00/11/2024 and B-7766/III.5/FR.06.00/11/2024, and through the RIIM Invitasi Strategis Platform Kolaborasi Biologi Struktur Biomolekul Keanekaragaman Hayati, Indonesia, under Grant No. 55/III.5/HK/2025. Contributor Information Kusmiati Kusmiati, Email: kusm001@brin.go.id. Arif Nurkanto, Email: arif028@brin.go.id. Asrul Fanani, Email: asrul0403fanani@apps.ipb.ac.id. Jendri Mamangkey, Email: jendri.mamangkey@unm.ac.id. Dian Alfian Nurcahyanto, Email: dian036@brin.go.id. Khatarina Meldawati Pasaribu, Email: khat004@brin.go.id. Indriati Ramadhani, Email: indr046@brin.go.id. Ade Lia Putri, Email: adel004@brin.go.id. Fifi Afiati, Email: fifi001@brin.go.id. Ni Wayan S. Agustini, Email: niwa003@brin.go.id. Herman Irawan, Email: herm015@brin.go.id. Dody Priadi, Email: dody002@brin.go.id. Yelin Adalina, Email: yeli001@brin.go.id. Muhammad Taufik, Email: mtaufik.chem@gmail.com. Silvia Mandasari, Email: silviamandasari899@gmail.com. Hasnadhiazahra Rohadi, Email: Hasnadhiazahrarohadi@mail.ugm.ac.id. References