Flavonoids as Human Intestinal α-Glucosidase Inhibitors

Certain flavonoids can influence glucose metabolism by inhibiting enzymes involved in carbohydrate digestion and suppressing intestinal glucose absorption. In this study, four structurally-related flavonols (quercetin, kaempferol, quercetagetin and galangin) were evaluated individually for their ability to inhibit human α-glucosidases (sucrase, maltase and isomaltase), and were compared with the antidiabetic drug acarbose and the flavan-3-ol(−)-epigallocatechin-3-gallate (EGCG). Cell-free extracts from human intestinal Caco-2/TC7 cells were used as the enzyme source and products were quantified chromatographically with high accuracy, precision and sensitivity. Acarbose inhibited sucrase, maltase and isomaltase with IC50 values of 1.65, 13.9 and 39.1 µM, respectively. A similar inhibition pattern, but with comparatively higher values, was observed with EGCG. Of the flavonols, quercetagetin was the strongest inhibitor of α-glucosidases, with inhibition constants approaching those of acarbose, followed by galangin and kaempferol, while the weakest were quercetin and EGCG. The varied inhibitory effects of flavonols against human α-glucosidases depend on their structures, the enzyme source and substrates employed. The flavonols were more effective than EGCG, but less so than acarbose, and so may be useful in regulating sugar digestion and postprandial glycaemia without the side effects associated with acarbose treatment.


Introduction
One of the earliest signs of type 2 diabetes (T2D) is elevated and erratic postprandial glycaemia that promotes oxidative stress at various sites within the body [1]. Controlling postprandial glycaemia is an important strategy in the management of T2D. One way is by slowing down carbohydrate digestion and glucose absorption in the intestine via the inhibition of salivary/pancreatic α-amylases and membrane-bound brush-border α-glucosidases.
Flavonoids are found ubiquitously in plants and represent~60% of all dietary (poly) phenolic compounds [19,20]. Flavonols, a sub-class of flavonoids, are present in onions, kale, apples, berries, leeks and broccoli [19]. Some flavonols excreted in urine can be used as biomarkers of flavonol intake and are significantly associated with a lower T2D risk [21]. Many flavonoids extracted from plants inhibit α-amylase and α-glucosidases activities in vitro and improved postprandial glycaemia in diabetic animal models and limited human studies [22,23]. Very few studies have reported on the inhibition of isomaltase, however. The disaccharide isomaltose is rarely present in nature but is commonly added as low-caloric food sweeteners in industrial-scale production [24,25], or produced from amylopectin hydrolysis to α-limit dextrins. Studies assessing the isomaltase inhibitory potential by flavonoids and acarbose are therefore of interest.
Unfortunately, many enzyme inhibition studies have been conducted using α-glucosidases from yeast or bacteria, with fewer studies using human intestinal enzymes. The inhibition of yeast and human α-glucosidases is very different, specific to the type of substrate, as reported for maltose [2]. Here we used Caco-2 cells, originating from human colon cancer cells, which form monolayers that differentiate to produce apical microvilli with high expression of maltase and sucrase. The Caco-2/TC7 clone specifically expresses high SI levels at 19-25 days post-confluence [24,26]. Using an enzyme preparation from these cells, we have evaluated sucrase, maltase and isomaltase inhibition by several flavonols and compared them to acarbose and (−)-epigallocatechin-3-gallate (EGCG), a flavan-3-ol known for its inhibitory activity on sucrase and maltase of various sources [27]. These natural compounds may provide promising alternatives for diabetes management with no undesirable side effects.

Reagents and Instruments
Buffer components, sugar substrates and standards, and most inhibitors (acarbose, galangin, kaempferol and EGCG) were purchased from Sigma-Aldrich Corp., Merck (St. Louis, MO, USA), with purity >98%. Quercetagetin was purchased from EMD Millipore, Merck (Burlington, MA, USA) and quercetin was purchased from Extrasynthese (Genay, France). Maltose monohydrate, sucrose and isomaltose were used as substrates for the enzyme assay, and together with fructose and glucose, were used as sugar standards for the chromatographic analyses. All other chemicals were of analytical grade and also purchased from Sigma-Aldrich, unless specified otherwise.
The Dionex™ Integrion™ HPIC™ (High Performance Ion Chromatography) system was used for High-Performance Anion-Exchange chromatography with Pulsed Amperometric Detection (HPAE-PAD) (Thermo Fisher Scientific Inc., Waltham, MA, USA) for the separation and analysis of sugars. A PHERAstar FS plate reader (BMG Labtech, Ortenberg, Germany) was used for measuring absorbance in the total protein assay. High-purity (18.2 mΩ/cm) H 2 O supplied by a MilliQ system (Millipore) was used throughout.

HPAE-PAD Instrumentation and Chromatographic Conditions
Disaccharides and monosaccharides were analyzed by HPAE-PAD on the Dionex™ Integrion™ HPIC™ system (Thermo Fisher Scientific). Separation of the carbohydrates was achieved on a CarboPac PA210 column (2 × 150 mm), preceded by a CarboPac PA210 Guard column (2 × 30 mm), with the column and compartment temperatures maintained at 30 • C and 20 • C, respectively. Eluent was generated using a Dionex™ EGC 500 KOH eluent generator cartridge with Dionex™ continuously regenerated-anion trap column 600, with eluent concentration following a multistep gradient: 12 mM for 12 min, 100 mM for 8 min and 12 mM for 12 min, at a flow rate of 0.2 mL/min and with a sample injection volume of 2.5 µL. Detection was performed using a gold working electrode and AgCl reference electrode at pH~12.0, with a collection rate of 2.00 Hz using the "Gold, Carbo, Quad" waveform. The total run time per sample was 32 min, with 3 min allowed between analyses for sample injection by autosampler. A wash injection of only H 2 O while following the same multistep eluent gradient was performed at the end of every batch (12-14 samples). All injections were performed in duplicate and peak identification was achieved by comparing retention times to the standards. Dionex™ Chromeleon™ 7 Chromatography Data System, version 7.2.9 (Thermo Fisher Scientific, Waltham, MA, USA), was used to process the chromatograms, ensuring peaks were suitably integrated before recording peak area. Sugars were quantified from peak areas using standard curves, with the standards prepared in the same buffer as the samples.

Substrate, Inhibitor/Flavonoid and Sample Preparation
Stock solutions (250 mM) of sucrose, maltose and isomaltose (substrates), plus glucose and fructose (standards only), were prepared in sodium phosphate buffer (SPB, 10 mM, pH 7.0). Stock solutions of acarbose (1 mM) and all flavonoids (10 mM) were prepared by dissolving in their respective solvents, stored at −20 • C and used within 2 weeks. Acarbose and EGCG were prepared in H 2 O, quercetin and quercetagetin were prepared in DMSO, while kaempferol and galangin were dissolved in absolute ethanol. The maximum working concentrations for each compound was pre-determined before enzymatic reaction assay to ensure no precipitation occurred in the system. Compared to acarbose and EGCG, lower solubility of tested flavonols was expected due to their structural differences, where compounds with less than 100 µg/mL solubility were considered poorly soluble, as reported previously [28,29]. Inhibitors were tested at various concentrations: acarbose (0.1-100 µM), EGCG (5-1500 µM), quercetin (20-200 µM), quercetagetin (1-50 µM), kaempferol (5-40 µM) and galangin (1-25 µM). The flavonols were tested up to their maximum soluble concentrations. Working solutions were prepared fresh at various concentrations in SPB buffer immediately before assaying. The maximum concentrations (v/v) of DMSO were ≤2% and ≤0.5% for quercetin and quercetagetin, respectively, and ethanol was ≤0.5% for kaempferol and galangin. The solvents did not affect enzyme activity, as demonstrated by vehicle controls. Cell-free extracts (CFE) were prepared as described in Section 2.6 and used as the enzyme source.
All prepared standards and assay samples, for both method validation and post-assay quantification, underwent the same treatment prior to injection on the HPAE-PAD system. All were deproteinated by mixing with an equal volume of acetonitrile, vortexed for 30 s and centrifuged at 17,000× g, 15 min at 4 • C. The resulting supernatants were then diluted at least 10× in H 2 O (maximum final acetonitrile concentration of 5% (v/v)). Additionally, all standards and samples containing enzyme and substrates were filtered through 0.2 µM polyether sulfone (PES) filters (Sartorius, Göttingen, Germany). All standards, samples and blanks were kept at 4-8 • C until analysis by HPAE-PAD, as described in Section 2.2, was complete.

Validation Parameters for Quantification
The HPAE-PAD method for the quantification of glucose, sucrose, fructose, isomaltose and maltose was set up based on our previously published method [30], but with improved sensitivity and run time. The method was validated for specificity, linearity, sensitivity, precision, and accuracy as percent extraction recovery, according to the guidelines issued by the U.S. Food and Drug Administration (FDA) 2018 [31] and International Conference on Harmonisation (ICH) 2005 [32].

Specificity and Matrix Effect
Specificity was determined by evaluating any endogenous interferences from the CFE. A comparison study was conducted on chromatograms of a blank incubated CFE matrix sample (CFE/enzyme only), CFE incubation sample with inhibitors (without substrate), CFE incubation sample with a substrate (without inhibitor), individual inhibitors only and blank assay solvent (DMSO and ethanol (v/v <2%)). Blank samples spiked with a known amount of maltose, sucrose and isomaltose served as reference. The matrix effect was evaluated by comparing the analytical response of sugar spikes in SPB to those in H 2 O, to ensure accurate calibration plots were constructed.

Linearity
The linearity of the HPAE-PAD method was evaluated by a calibration curve constructed by plotting concentrations of standards against their peak areas within the determined limits of detection and quantification (LOD and LOQ, respectively; see below). Six different concentrations of maltose, sucrose and isomaltose (0.1, 0.5, 1.0, 2.5, 5.0 and 10.0 µg/mL) were prepared in assay incubation buffer or distilled water and measured in triplicate on four different days, giving a total of twelve replicates to construct the curve. Linearity was evaluated by calculating a regression line by the least-squares method, deter-mining a linear equation (Equation (1)), where y = peak area, x = concentration, a = slope, b = intercept, and R 2 for each standard [33].
The sensitivity of the method was evaluated by determining the LOD and LOQ using data generated from the calibration curve. LOD and LOQ were measured using the SD of the y-intercept and the slope of the calibration curve, as shown in Equations (2) and (3) below, where SD y−int is the standard deviation of the y-intercept and S is the calibration curve slope. Both LOD and LOQ were expressed as analyte concentration (µM):

Precision
Precision (repeatability and reproducibility) was determined through the analysis of intra-and inter-day assay using standards in SPB quantification buffer. Intra-assay precision was assessed by measuring six concentrations of each standard measured in triplicate on the same day, in one laboratory by one person. Inter-assay precision was carried out by measuring the same concentrations of standards measured (in triplicate) over four different days by two analysts in the same laboratory. Precision was expressed as percent coefficient of variance (%CV), according to Equation (4), calculated as:

Accuracy as Extraction Recovery
As standard reference material was not used, accuracy was determined using the extraction recovery calculated by comparing the analytical response of two different concentrations of substrates spiked pre-assay to the values recovered post-assay in triplicates. The accuracy was calculated following Equation (5) as below: where C measured = measured concentration calculated from the calibration curve (µM); C standard = real (prepared) concentration of the standard solution (µM), which was used in the calculation of the percent relative error (%RE), as shown in Equation (6) below:

Enzyme Activity Assay
To mimic intestinal digestion, an in vitro assay using Caco-2/TC7 cell extracts containing sucrase, maltase and isomaltase was conducted. Frozen cells were thawed, 1 mL ice-cold SPB added and then passed through a 21-G needle 15-20 times. The lysate was centrifuged at 14,000× g, 10 min at 4 • C and the supernatant containing cell-free extract (CFE) collected. Total protein concentration in the CFEs was determined by Bradford assay [34,35], using the Pierce Coomassie Bradford reagent and BSA standards (Thermo Fisher Scientific, Waltham, MA, USA).
Assay mixtures, total volume 250 µL, containing CFE (final protein concentration at 0.1-0.35 mg/mL), with or without various concentrations of inhibitors/flavonoids, were prepared and kept on ice. The enzyme reaction was initiated by the addition of ≥20 mM sucrose, maltose or isomaltose and immediately incubating in a 37 • C water bath for 10 min (or various time points during method setup and validation). Following incubation, the enzyme activity was terminated by incubating in a 96 • C water bath for a further 10 min. A positive control without any added inhibitor/flavonoid was simultaneously tested in each batch, and negative controls without enzymes or substrates were also assayed to evaluate the stability of the inhibitors/compounds. Samples were prepared for HPAE-PAD analysis as described in Section 2.3. Specific enzyme activities were determined (U/mg CFE protein) and expressed as a percentage of control enzyme activities accordingly.

Optimization of Assay Conditions and Enzyme Kinetics
Preliminary assays to optimize the substrate and enzyme concentrations and incubation time were performed to ensure enzyme kinetic experiments were carried out under initial linear velocity conditions (substrate depletion <10%). CFE protein concentration was tested, and specific activities found to be linear, at 0.10-0.35 mg/mL (Table 1), with substrates tested at 10-80 mM for maltase and 5-50 mM for sucrase and isomaltase, while incubation times were tested for 10-60 min. Michaelis-Menten and Lineweaver-Burk plots were used to obtain the kinetic parameters of the digestive enzymes, using GraphPad Prism 8.0 software (GraphPad Software, San Diego, CA, USA) ( Table 2). The specific activities of sucrase and maltase were similar to those we reported previously [27]. Lower CFE concentrations (0.25 mg/mL for sucrase and 0.1 mg/mL for maltase and isomaltase) were used to determine the substrate concentration required to achieve maximal catalytic efficiency or velocity of reaction (V max ) ( Table 2). Sucrase, maltase and isomaltase exhibited a linear production of glucose up to 40 min using 20 mM maltose, sucrose and isomaltose, respectively. A 10 min incubation time was used in the inhibition assays to ensure reactions were in the initial linear velocity phase. Assays were performed using CFEs from biological triplicates.

Inhibition by Acarbose and Flavonoids
Using the optimal assay conditions, various concentrations of acarbose and flavonoids were tested. Controls (CFE and substrate only) were prepared by replacing the volume of inhibitor with SPB. Activities of sucrase, maltase and isomaltase were considered as 100% (or 0% inhibition) in the absence of an inhibitor. Compounds that exhibited enzyme inhibition of at least 15%, 25% and 50% were subjected to IC 15 , IC 25 and IC 50 value determination, respectively, or the maximum percentage of inhibition expressed. Estimation of inhibition values were determined in GraphPad Prism and the percentage of inhibition of the sample was calculated following Equation (7) below, where SA = specific activity:

Data Analysis
Enzyme assays were performed at least once for CFEs from biological triplicates, with duplicate injections of each analyzed by HPAE-PAD. MS Excel (Microsoft, Redmond, WA, USA) was used for data processing and analysis. Final inhibition values were expressed as the percentage of control activity (%). The IC 15 , IC 25 and IC 50 values were calculated based on the plots created using GraphPad, using the dose-response inhibition (log (inhibitor) vs. normalized response-variable slope) model. The same software was used to determine the apparent K m and V max values under the enzyme-kinetic inhibition model, and for statistical analyses using non-parametric multiple comparisons tests. A difference was considered significant at p < 0.05 for all comparisons. All data are expressed as the mean ± SD or SEM, specified accordingly.

Method Validation
The potential inhibition of key human intestinal α-glucosidases by flavonoids was evaluated. Initially, the analytical method was optimized and validated. Chromatograms for mixed sugar standards in H 2 O or in sodium phosphate assay buffer (SPB) were almost identical, with the same retention times and peak areas, demonstrating the absence of matrix effects (Figure 2a). Specificity was confirmed by comparing the retention times and peak areas of the sugars when run individually and as a mixture. The peak areas from mixed sugar standards prepared in SPB were used to plot standard curves (0-10 µg/mL), with excellent linearity for all sugars in this range (Figure 2b). The intercepts were not significantly different from zero (p = 0.225). LOD and LOQ for all five sugars were determined as a signal to noise ratio of 3:1 and 10:1, respectively. All sugar standards showed low LOD (ranging from 0.106 µM for maltose to 0.619 µM for fructose) and LOQ (ranging from 0.320 µM for maltose to 1.876 µM for fructose) ( Table 3).   Intra-and inter-run precision was determined by analyzing the standards in triplicate in a single run on the same day and repeated on four different days within four months. Samples were kept at 4 • C at all times to avoid repeated freeze/thaw cycles or deterioration at RT. The mean peak areas and coefficients of variation (%CV) were calculated to determine the precision, as per ICH guidelines, [32] as presented in Table 4. Intra-run data collected from triplicate injections in a single run on 1 day; inter-run assays performed on four separate days in triplicate. The %CV of peak areas were <15%, indicating excellent precision and recoveries.
The intra-day precision range was calculated to be 0.63-2.14% for glucose, 0.94-5.52% for sucrose, 1.03-6.52% for fructose, 1.05-3.47% for isomaltose and 1.08-10.39% for maltose, while inter-day evaluations were 5.01-12.44%, 3.73-7.03%, 8.13-10.22%, 4.39-5.14% and 3.98-12.78% for glucose, sucrose, fructose, isomaltose and maltose, respectively. Precision results were considered excellent, as they fall below 15% of ICH guidelines, even at the lowest sugar concentration of 0.1 µg/mL. Extraction efficiency was between 95.9% and 108.9% for sucrose, maltose and isomaltose at two different concentrations, where the %RE accuracy and the %CV precision were <10% for all concentrations (Table 5). All values are mean ± SEM (n = 3). Pre-assay concentrations of maltose, sucrose and isomaltose spikes were 10 mM (C1) and 20 mM (C2). The %RE accuracy and %CV precision values are <10%, demonstrating excellent recoveries. Figure 2 displays the chromatogram of the sugar standards with efficient separation in a single 32-min elution, with good resolution. To analyze the efficiency of the enzyme reaction and sample extraction, substrates with or without cell-free extract (CFE) were digested, extracted and analyzed accordingly. The representative chromatograms in Figure 3 show the breakdown of sucrose into glucose and fructose (Figure 3a,b), and the breakdown of maltose (Figure 3c,d) and isomaltose (Figure 3d,e) into glucose was detectable with good resolution only when the relevant enzymes were present.   Table 6 indicates the peak areas of maltose, sucrose and isomaltose in the presence of (potential) inhibitors were not significantly different to when the compounds were not added (ANOVA: F < F critical one tail, p > 0.05), and without the presence of any additional glucose or fructose peaks in the chromatograms. High precision values were obtained for all tested sugars with inhibitors (%CV precision < 15%), indicating no interference between sugars and tested substances. No peaks were observed in blank/control samples, where (potential) inhibitors, or assay buffer, or water, acetonitrile and eluent alone were tested.

Inhibition of α-Glucosidase Activities
Based on the evidence that some flavonols consumed in the diet may reach concentrations as high as 50 µM or more in the intestinal lumen [36], a concentration up to 200 µM, or maximal solubility, was used in this study. Where half-maximal inhibitory potential (IC 50 ) could not be determined, IC 25 and IC 15 were calculated instead. All flavonols, acarbose and EGCG inhibited α-glucosidase activity to some extent (Figures 4-6 and Table 7).

Maltase
Based on the IC 25 values, the decreasing order of the maltase inhibitory activity of the studied inhibitors was acarbose ≥ quercetagetin ≥ galangin ≥ kaempferol ≥ EGCG > quercetin, where the significantly weakest inhibition of maltase was exhibited by quercetin (p < 0.05) ( Table 7). Only acarbose and EGCG showed more than 50% maltase inhibition at their maximal tested concentrations ( Figure 5).

Isomaltase
On the basis of the IC 15 values, the decreasing order of the isomaltase inhibitory activity of the studied inhibitors was concluded to be quercetagetin ≥ acarbose ≥ galangin ≥ kaempferol > quercetin ≥ EGCG. This pattern is almost identical to sucrase inhibition. The inhibition potentials of acarbose and EGCG were generally lower towards isomaltase than sucrase and maltase, where a few fold higher concentrations were required to exhibit half-maximal inhibition ( Table 7). The inhibition shown by acarbose and EGCG was 96% and 71%, respectively, while all flavonols exhibited <30% inhibition of isomaltase ( Figure 6).

Discussion
Flavonoids and other (poly)phenols potently inhibit α-amylase and α-glucosidase activities [37] without associated adverse gastrointestinal effects [38] and so may be useful in the management of T2D. However, previously, most inhibitory activities have been tested using α-glucosidase from yeast (Saccharomyces cerevisiae) with limited reports on enzymes of mammalian or human origin [37]. Multiple α-glucosidases are widely distributed in microorganisms, plants and animal tissues with variations in >20 amino acid sequences between species [39,40]. We explored disaccharide digestion in the human intestine by determining the inhibitory potential of flavonoids on sucrase, maltase and isomaltase in a specific mature Caco-2/TC7 clone, with high expression of SI [24]. Four flavonols were compared to a commercial α-glucosidase inhibitor, acarbose and a flavan-3-ol, EGCG. All compounds inhibited human sucrase, maltase and isomaltase in a dose-dependent manner.

Quercetagetin
Quercetagetin was first identified as part of spinacetin (quercetagetin-3',6-dimethyl ether) in spinach [41], and a few recent reports identified a possible function in glucose metabolism [42][43][44][45]. Compared to quercetin, it has an additional C6-OH in the A ring, and exhibits various biological activities [46][47][48]. The additional C6-OH confers a strong affinity to proteins, speculated to weaken the binding of substrates to the active sites of enzymes and reduce or inhibit their activities [49]. Our study revealed quercetagetin as a strong human sucrase inhibitor, similar to acarbose and more potent than quercetin and EGCG. This was a greater inhibition than that seen previously against yeast α-glucosidase [42], demonstrating the varied activities between species and substrates used. Quercetagetin could be a promising α-glucosidase inhibitor, provided its high susceptibility to degradation [21] is considered.

Galangin
Galangin has an unsubstituted B ring and is rich in many root plants, and possesses antiviral and anti-inflammatory properties with no toxic effects observed even at high doses in rats [61][62][63]. This compound regulates glucose homeostasis and enzymes responsible for glycolysis and gluconeogenesis in rats [64]. Reports have shown strong inhibition of α-glucosidase in yeast, better than acarbose [65] but similar to kaempferol [66]. In contrast, both galangin and kaempferol were poorer maltase inhibitors than acarbose when tested using enzymes from rat intestine [54]. Like kaempferol, the poor aqueous solubility of the aglycone is a drawback for practical use.

Quercetin
Quercetin is a widely distributed flavonoid and therefore most researched in human and animal models. With a half-life of~24 h [67], a few fold increase of this compound in plasma after several weeks of ingestion was noted (reviewed in [68]). Quercetin is a potent inhibitor of intestinal GLUT2, substantially reducing glucose absorption [69]. Quercetin has repeatedly been reported to inhibit yeast α-glucosidases more so than acarbose [42,50,[70][71][72]. In contrast, for rat maltase and sucrase, it was shown to be weaker than acarbose (IC 50 = 281.2 µM for maltase, IC 50 > 400 µM for sucrase) [52], similar to the data reported here.

EGCG
Among all tested flavonoids, while EGCG was most soluble with the highest inhibition reached for human sucrase (100%), maltase (68%) and isomaltase (71%) at 1500 µM, this is a supra-physiological concentration and EGCG generally exhibited the weakest inhibitory potential when compared by concentration alone, as indicated previously [30]. Conversely, much stronger α-glucosidase inhibitory effects were demonstrated against yeast or recombinant enzymes [73][74][75]. It has been suggested that EGCG (and quercetin) may exert much slower but more effective inhibition of disaccharide digestion in the intestine [75]. EGCG potently reduced glycaemic response in a diabetic animal model by binding to the active site of α-amylase and α-glucosidase [76] and decreased glucose uptake and GLUT2 expression in vitro [77]. However, a recent systematic review and meta-analysis from fourteen eligible articles demonstrated that the regular intake of EGCG-rich green tea had no significant effects on fasting blood glucose and insulin, HbA1c or HOMA-IR in T2D patients [78], which may be partially explained by the weak inhibition towards all three intestinal α-glucosidases shown in this study.

Structure-Function Relationships
The most active flavonol was quercetagetin, with IC 50 values closest to acarbose. This suggests that stronger enzyme inhibition is observed with increasing hydroxyls on the A ring since quercetagetin is a stronger inhibitor than quercetin. Increasing hydroxylation of the B ring (from galangin to kaempferol to quercetin) improves the solubility of compounds but lowers inhibition. The lower aqueous solubility of flavonols, observed in quercetagetin, kaempferol and galangin, is a shortcoming of this study, and is the reason why some IC 50 values could not be determined.
At the molecular level, the binding between hydroxyls in ring A, B or C of flavonoids to the active sites of α-glucosidases leads to structural changes in the enzyme evidenced by several docking studies with yeast α-glucosidase [20,55,72,79]. The inhibitory activity of flavonoids was concluded to be in the decreasing order of anthocyanidin ≥ isoflavone ≥ flavonol ≥ flavone ≥ flavonone ≥ flavan-3-ol [50], indicating the crucial role of A ring hydroxylation for potent α-glucosidases inhibition [80]. The A ring hydroxylation at C5 (fisetin converted to quercetin) or C6 (quercetin converted to quercetagetin) increased αglucosidase and α-amylase inhibition in yeast and rat [50,81], and in human α-glucosidases as shown here.
The hydroxylation patterns, particularly 3-OH at a B ring catechol moiety, are among the major determinants of various biological effects of flavonoids [82]. The hydrophilicity of compounds is enhanced with increasing hydroxyls in the B ring, which also affects α-glucosidase inhibition, varying between species and substrates used [20,83]. The αglucosidase inhibitory activity of flavonols increased with increasing hydroxyls on the B ring in rat and yeast (myricetin > quercetin > kaempferol) [50,51], in contrast to the results shown in this study (quercetagetin > galangin > kaempferol > quercetin > EGCG), and again emphasizing the importance of using human enzymes. Further, the hydrogenation of the C2 = C3 double bond in flavan-3-ols on the C ring weakened their enzyme inhibition activities [50,80], despite higher binding affinities [84]. Saturated C2-C3 bonds in flavan-3-ols are speculated to allow more twisting of the B ring and, together with additional hydroxyls on the gallate group in the C ring, increase solubility [85].

Comparing Flavonoids to Acarbose
Although mild α-amylase inhibition is beneficial for blunting glucose spikes, excessive inhibition may induce starch indigestion and abnormal bacterial fermentation, causing abdominal pain, bloating or cramping [22]. Acarbose can induce these undesirable effects due to its potent inhibition of human and mammalian pancreatic α-amylase [11]. Medicinal plant extracts containing quercetin and kaempferol consistently exhibited favorable inhibition against yeast and mammalian α-glucosidase over pancreatic α-amylase [53,[86][87][88][89]. Many flavonoids have a higher inhibition of α-glucosidases, leading to a slow-release effect, than of α-amylase [75], which may be favoured over acarbose to decrease postprandial glucose spikes without the unpleasant side effects.
We have elucidated the inhibitory effects of flavonoids against human α-glucosidases compared with acarbose, influenced by structure, enzyme origin and substrates. A higher concentration of acarbose is required to inhibit maltase than sucrase, while all flavonoids showed similar inhibition of sucrase and maltase, in agreement with our previous findings using olive leaf extracts [30]. Isomaltose is known to be hydrolyzed slowly by the SI complex, reflected by the accumulation of isomaltose in the intestine [90]. We have shown that quercetagetin (at higher concentrations) inhibits starch digestion through direct αamylase inhibition and starch complexation [91], making it a promising compound for regulating postprandial glycaemia.
Enzyme inhibition by plant extracts was consistently superior to acarbose when tested using yeast α-glucosidase [72,73], in contrast to data on human or mammalian sucrase and maltase. Previously, an IC 50 of 20 µM for EGCG was determined for human maltase expressed in yeast [74], which is 8-10-fold lower than reported here, and by us previously [30], using human intestinal Caco-2/TC7 as the enzyme origin. This emphasizes the importance of using a relevant substrate and enzyme source when screening for inhibitory potentials of compounds.

Conclusions
Our study highlights the potential of selected flavonoids to inhibit human intestinal αglucosidases, hence slowing carbohydrate digestion and reducing postprandial glycaemia. A sensitive and accurate method to determine sugar hydrolysis by sucrase, maltase and isomaltase has been successfully developed and validated. The use of HPAE-PAD to detect subtle changes in the concentrations of five sugars simultaneously, with minimal sample preparation and high precision within 32 min, has been central to this study. Acarbose and flavonoids exhibit different inhibition of human enzymes to those reported for yeast or mammalian α-glucosidases, emphasizing the need for a more pragmatic screening approach on individual human enzymes to elucidate their actual inhibitory potentials in vivo. Flavonoids from various sources are more effective against α-glucosidase than α-amylase [37]. The low solubility of some flavonoids limits the experimental concentration which can be employed, preventing the determination of IC 50 values and necessitating the use of IC 25 or IC 15 values instead.
Quercetagetin, similar to acarbose, followed by kaempferol and galangin, exhibited greater inhibitory action against sucrase, maltase and isomaltase than EGCG and quercetin, although the latter compounds were more soluble in aqueous buffer. Two key structural elements of flavonoids for enhanced α-glucosidase inhibition in humans are the C6-OH A ring hydroxylation and reduced B ring hydroxylation. Improving understanding of how flavonoids bind to human α-glucosidases should provide a rational basis for exploiting antidiabetic compounds from dietary sources.