Phytochemical constituents of Hydrangea macrophylla var. acuminata leaves and their inhibitory activity against PTP1B and α-glucosidase

Abstract

This study presents the first comprehensive phytochemical analysis of Hydrangea macrophylla var. acuminata leaves, resulting in the isolation of 41 secondary metabolites, including four new compounds: one phenolic (3) and three bis-iridoid glycosides (30–32). The inhibitory activities of these compounds against PTP1B and α-glucosidase were evaluated. Among them, compound 12 exhibited the most potent dual inhibition, with IC₅₀ values of 8.0 ± 1.1 µM for PTP1B and 3.4 ± 0.2 µM for α-glucosidase. Compounds 8 and 9 showed notable α-glucosidase inhibitory activity, with IC₅₀ values of 21.9 ± 0.4 µM and 43.8 ± 2.1 µM, respectively. Enzyme kinetics and molecular docking studies revealed their inhibition mechanisms and binding interactions. This study is the first detailed phytochemical investigation of H. macrophylla var. acuminata and highlights its potential as a natural source of PTP1B and α-glucosidase inhibitors. These findings underscore the plant’s promise for developing antidiabetic agents targeting PTP1B and α-glucosidase.

GRAPHIC ABSTRACT

Introduction

The genus Hydrangea, comprising approximately 80 species, is widely cultivated worldwide for ornamental purposes, with Hydrangea macrophylla being one of the most prominent representatives.¹ The leaves of H. macrophylla have traditionally been consumed as herbal tea, primarily due to the presence of phyllodulcin, a naturally occurring sweetener.² This traditional use has stimulated scientific interest in the chemical constituents and biological activities of this plant. Recent studies have reported that H. macrophylla leaves exhibit diverse pharmacological properties, including antimalarial, antidiabetic, and immunostimulatory effects, as well as anti-wrinkle, anti-photoaging, and skeletal muscle-enhancing activities.^3–9 These biological activities are attributed to various phytochemical constituents, primarily coumarins, isocoumarins, and phenolics.^10–13 However, most studies have focused on H. macrophylla var. serrata, var. thunbergii, and var. otaksa,^5,6,14 whereas studies on H. macrophylla var. acuminata remain limited.

Diabetes mellitus (DM) is a complex metabolic disorder categorised into type 1 (T1D) and type 2 (T2D), both of which constitute major global health concerns due to their association with chronic complications, including obesity, cardiovascular disease, and kidney failure.^15,16 While T1D management requires lifelong insulin administration, T2D typically necessitates a combination of pharmacological agents targeting various aspects of glucose metabolism. Among the therapeutic targets for T2D, protein tyrosine phosphatase 1B (PTP1B) and α-glucosidase have gained attention. PTP1B negatively regulates insulin signalling by dephosphorylating key downstream molecules, whereas α-glucosidase facilitates glucose absorption by catalysing the breakdown of complex carbohydrates in the small intestine.^17,18 Inhibitors of PTP1B enhance insulin sensitivity and modulate leptin signalling, contributing to improved diabetic and obesity-related conditions.¹⁹ Conversely, α-glucosidase inhibitors delay carbohydrate digestion, thereby lowering postprandial blood glucose and glycosylated haemoglobin levels.¹⁹ Despite the available therapeutic strategies, many T2D patients experience clinical inertia, alongside the financial burden of long-term medication use.²⁰ Consequently, natural products with antidiabetic potential are increasingly explored as complementary interventions. H. macrophylla leaves, traditionally consumed as herbal tea, contain bioactive isocoumarins that have demonstrated efficacy in diabetes management.^5,14 However, phytochemical studies on H. macrophylla var. acuminata remain limited. In this context, the present study aimed to isolate and characterise secondary metabolites from the leaves of H. macrophylla var. acuminata and evaluate their inhibitory activities against PTP1B and α-glucosidase, thereby providing deeper insight into the phytochemistry and pharmacological potential of this subspecies.

To our knowledge, this study is the first comprehensive investigation into the phytochemical profile and dual enzyme inhibitory potential of this botanical variety against PTP1B and α-glucosidase.

Materials and methods

General experimental procedures

The general experimental procedures were conducted following a previous study with a slight modification.²¹ The compounds were isolated using various column chromatographic methods with different stationary materials, including silica gel 60 (40–63 µm), reversed-phase RP-18 gel (40–63 µm) (LiChroprep), Diaion HP-20 resin (250–850 µm), Sephadex LH-20 (25–100 µm), and MCI gel (75–150 µm) (Merck, Darmstadt, Germany). High-performance liquid chromatography (HPLC) was employed using a Waters HPLC system (MA, USA), equipped with a YMC-Pack ODS-A column (20 × 250 mm, 5 µm; YMC, Kyoto, Japan). Thin-layer chromatography (TLC) was conducted on pre-coated glass plates containing either silica gel 60 F₂₅₄ or RP-18 F_254S (Merck). Visualisation was achieved under UV light at 254 and 365 nm, followed by spraying with 10% sulphuric acid and heating at 120 °C for 1 min. High-resolution fast atom bombardment mass spectra (HR-FAB-MS) were acquired with a JMS-700 instrument (JEOL, Tokyo, Japan). Nuclear magnetic resonance (NMR) spectra were measured on a Bruker Avance Digital 500 MHz spectrometer (Bruker, Karlsruhe, Germany). Optical rotation was determined using a JASCO P-2000 polarimeter (JASCO, Tokyo, Japan).

Plant materials

The dried leaves of H. macrophylla var. acuminata were purchased from a traditional market in Jeonju, Korea, in October of 2018. Authentication was performed by Professor Byung Sun Min at Daegu Catholic University. The plant specimen (22 A-HM) was preserved in the the Pharmacognosy Lab, College of Pharmacy, Kyungpook National University, Korea.

Isolation and extraction

The dried leaves of H. macrophylla var. acuminata were extracted with methanol (MeOH, 15 L × 3, 4 h each). The MeOH extract (1.3 kg) was suspended in distilled water (1.5 L) and partitioned with n-hexane and ethyl acetate (EtOAc) to yield three fractions: n-hexane, EtOAc, and water (Figure S36).

The EtOAc-soluble fraction (411.5 g) was fractionated into eight subfractions (1A–1H) using vacuum liquid chromatography (VLC) with a gradient of dichloromethane (CH₂Cl₂)/MeOH (50:1 to 0:1, v/v). Fraction 1C was subjected to silica gel column chromatography (CC) with n-hexane/EtOAc (10:1–1:0, v/v), to yield seven subfractions (2A–2G). Fraction 2B was identified as compound 9, whereas compounds 1 and 10 were crystallised from MeOH in fractions 2E and 2F, respectively. Compounds 2 and 4 were purified from fraction 2 G using LH-20 Sephadex CC with MeOH/H₂O (1:1, v/v). Fraction 1D was chromatographed on MCI gel CC (MeOH/H₂O; 1:5–1:0) to yield six subfractions (3A–3F). Fraction 3D was identified as compound 29, and fraction 3E was crystallised in MeOH to obtain compound 11. Fraction 3 F was subjected to LH-20 Sephadex CC with MeOH/H₂O (1:1, v/v) to produce three subfraction (4A–4C). Fraction 4C was identified as compound 12, whereas fraction 4A was subjected to HPLC with MeOH/H₂O (3:5, v/v) to isolate compound 28. Compound 36 was purified from fraction 4B using HPLC (MeOH/H₂O (3:5, v/v). Fraction 3C was loaded into MCI gel CC with MeOH/H₂O (3:10, v/v) and separated into seven subfractions (5A–5G). Fractions 5D and 5F were identified as compounds 6 and 5, respectively, whereas fraction 5C was subjected to HPLC with MeOH/H₂O (2:5, v/v) to isolate compounds 37 and 43. Compounds 7, 8, 35, and 42 were isolated from fraction 5G using HPLC with MeOH/H₂O (9:20–11:20, v/v). Fraction 5B was separated into four subfractions (6A–6D) using LH-20 Sephadex CC with MeOH/H₂O (1:1, v/v). Fraction 6C was identified as compound 24, whereas fraction 6D was subjected to HPLC (MeOH/H₂O, 1:1, v/v) to isolate compounds 3 and 22. Fraction 1F was chromatographed on MCI gel CC using MeOH/H₂O (1:5–1:0, v/v) to obtain 11 subfractions (7A–7K). Fraction 7D was identified as compound 20, whereas fraction 7F was separated into eight subfractions (8A–8H) through MCI gel CC using MeOH/H₂O (3:10, v/v). Fraction 8B was identified as compound 19, whereas fractions 8E, 8F, 8G, and 8H were further purified by HPLC with MeOH/H₂O (1:1, v/v) to yield compounds 25 and 39 from 8E, compound 38 from 8F, compound 23 from 8G, and compounds 26 and 27 from 8H. From fractions 8C and 8D, compounds 18 and 41 were isolated using LH-20 Sephadex CC with MeOH/H₂O (1:1, v/v), respectively. Compounds 32, 33, and 34 were isolated from fraction 7G by MCI gel CC with MeOH/H₂O (2:5–4:5, v/v), whereas from fraction 7G subjected to the same conditions, yielded seven subfractions (9A–9G). Fraction 9F was identified as compound 30, whereas fraction 9G was subjected to RP-18 silica gel CC with MeOH/H₂O (2:1, v/v) to obtain compound 31. Fraction 7J was purified by HPLC with MeOH/H₂O (2:5–1:0, v/v) to yield compound 16 and subfraction 10A, which was further subjected to silica gel CC with EtOAc/MeOH/H₂O (12:1:0.1) to isolate compound 17.

The water-soluble fraction (488.3 g) was separated into four fractions (11A–11D) using HP-20 Diaion CC (MeOH/H₂O, 0:1–1:0, v/v). Fraction 11A was further separated into eight subfractions (12A–12H) using silica gel CC (CH₂Cl₂: MeOH; 20:1–0:1, v/v). Fractions 12A and 12B were combined and subjected to silica gel CC with EtOAc/MeOH/H₂O (8:1:0.1) to yield compound 40, whereas fraction 12C was chromatographed on silica gel CC using CH₂Cl₂/MeOH/H₂O (4:1:0.1) to isolate compounds 13, 14, and 15.

PTP1B inhibition assay

The PTP1B inhibitory assay was performed following a previously described protocol, with slight modifications.²² The assay was conducted in a 96-well plate. Test compounds were prepared at concentrations ranging from 1 to 100 µM by dilution in a reaction solution composed of 50 mM citrate buffer (pH 6.0), 0.1 M NaCl, 1 mM EDTA (Sigma-Aldrich), and 1 mM dithiothreitol (DTT; Bio-Rad Laboratories, CA, USA). An aliquot of 10 µL of each sample was dispensed into individual wells, followed by the addition of 30–40 µL of the prepared buffer solution. Subsequently, 10 µL of recombinant human PTP1B enzyme (5 µg/mL, PTB2001, NKMAX, Korea) was added. The reaction mixture was incubated at 37 °C for 10 min, after which 50 µL of 2 mM p-nitrophenyl phosphate (p-NPP, Sigma-Aldrich) was introduced as the substrate. The mixture was then incubated for an additional 20 min at 37 °C. The reaction was terminated by adding 10 µL of 10 M NaOH, resulting in the formation of the yellow-colored p-nitrophenolate anion. Absorbance was measured at 405 nm using a microplate reader (Tecan, Männedorf, Zurich, Switzerland). The percentage of inhibition was calculated using Microsoft Excel 365 according to the following formula:

$$\text{\%\hspace{0.33em}}\mathit{\text{Inhibition}}=\left[\left(\Delta \mathit{\text{control}}\hspace{0.17em}-\Delta \mathit{\text{inhibitor}}\right)/\Delta \text{\hspace{0.33em}control}\right]\times 100\text{\%}$$

α-Glucosidase inhibition assay

The α‑glucosidase inhibition assay was conducted in a 96‑well plate following a protocol previously described in a study.²³ Test compounds (1–100 µM) and the substrate p‑nitrophenyl-α-D-glucopyranoside (p‑NPG; 48756; Sigma‑Aldrich) at 2.5 mM were each diluted in 100 mM phosphate buffer (pH 6.8). The α‑glucosidase enzyme (0.1 U/mL), sourced from Saccharomyces cerevisiae (G5003, Sigma‑Aldrich), was prepared in 10 mM phosphate buffer (pH 6.8). To initiate the reaction in each well, 20 µL of the test compound, 20–40 µL of phosphate buffer, 20 µL of enzyme solution, and 20 µL of substrate were sequentially added. The mixture was incubated at 37 °C for 10 min. The reaction was terminated by adding 0.2 M sodium carbonate (Na₂CO₃), which facilitates the formation of the yellow p-nitrophenol anion. Finally, absorbance at 405 nm was measured using a Tecan microplate reader (Männedorf, Zurich, Switzerland) to assess α‑glucosidase inhibition. The percentage of inhibition was calculated using Microsoft Excel 365 according to the following formula:

$$\text{\%\hspace{0.33em}}\mathit{\text{Inhibition}}=\left[\left(\Delta \mathit{\text{control}}\hspace{0.17em}-\Delta \mathit{\text{inhibitor}}\right)/\Delta \text{\hspace{0.33em}control}\right]\times 100\text{\%}$$

Enzyme kinetic analysis

Enzyme kinetic studies were performed using standard kinetic models, including Lineweaver–Burk and secondary plots. Based on the PTP1B inhibitory assay results, kinetic experiments were designed to elucidate the inhibition mechanism of compound 12 against PTP1B, while compounds 8, 9, and 12 were evaluated for their effects on α‑glucosidase. For the PTP1B assay, compound 12 was tested at concentrations of 2, 4, 10, and 12 µM in combination with substrate concentrations of 0.5, 1, and 2 mM. For the α‑glucosidase assay, substrate concentrations of 0.625, 1.25, and 2.5 mM were used with varying concentrations of the test compounds: compound 8 (7, 12, 20, and 40 µM), compound 9 (20, 30, 60, and 80 µM), and compound 12 (2, 5, 10, and 12 µM). Variations in enzyme activity observed at different inhibitor and substrate concentrations provide insights into the mode of inhibition exerted by the compounds on their respective target enzymes.

Molecular docking

Docking simulations were performed using AutoDock Vina 1.1.2 to predict the binding interaction of active compounds with PTP1B and α-glucosidase. The structure of PTP1B and its allosteric reference inhibitor, [3–(3,5-dibromo-4-hydroxy-benzoyl)-2-ethyl-benzofuran-6-sulfonic acid (4-sulfamoyl-phenyl)-amide, compound A, PDB ID: 1T49] was retrieved from the RCSB Protein Database. Given that the structure of S. cerevisiae α-glucosidase has not been reported, a three-dimensional (3D) structure was constructed following the approach described by Phong et al. using SWISS-MODEL.^23,24 The amino acid sequence of S. cerevisiae isomaltase (PDB ID: 3AJ7), which shares 72.37% sequence identity with S. cerevisiae α-glucosidase, was used as the template. The target sequence was retrieved from UNIPROT (accession ID: E1AFY8). The quality of the predicted α-glucosidase model was assessed using the SAVES v6.1 webserver and ProSA-web²⁵ (Figure S39–S42). The 3D structures of the α-glucosidase allosteric site inhibitor (gentisein) and active compounds 8, 9, and 12 were constructed using ChemDraw Professional 22.2 (PerkinElmer, Shelton, CT, USA). Ligand structures were optimised using Avogadro 2 (Kitware, Inc., NY, USA). Non-residues of the proteins were removed using ChimeraX 1.9 (RBVI, SF, USA). Protein preparation for docking was carried out in AutoDock Tools 1.5.6 by removing water molecules, adding hydrogen atoms, and defining rotatable bonds. Both ligands and proteins were saved in PDBQT format. Blind docking simulations were performed using grid boxes sized 126 × 126 × 126 for PTP1B and 118 × 126 × 126 for α-glucosidase. The grid centres were set at X = 46.093, Y = 17.633, Z = 20.397 for PTP1B, and X = 21.515, Y = −0.615, Z = 18.824 for α-glucosidase. Binding conformations were analysed using PyMOL version, while two-dimensional molecular interactions were visualised with Discovery Studio Visualiser 21.1 (Accelrys, Inc., CA, USA).

Acid hydrolysis and sugar analysis of new compounds

Compounds 3, 30, 31, and 32 (15.0 mg each) were dissolved in 10 mL of 10% HCl and subjected to acid hydrolysis at 70 °C for 2 h. After cooling to room temperature, the reaction mixtures were extracted with ethyl acetate (EtOAc), resulting in the separation of the EtOAc and aqueous layers. The aqueous layer was neutralised with 1 M NaOH and subsequently purified by column chromatography on Sephadex LH-20 using MeOH: H₂O (1:1, v/v) to isolate the monosaccharide fractions (3a, 30a, 31a, and 32a).

Sugar analysis was performed following a previously reported procedure with minor modifications.²⁶ Briefly, 800 µL of a 0.0139 M aqueous solution of each monosaccharide was mixed with 800 µL of 0.5 M PMP (1-phenyl-3-methyl-5-pyrazolone) in methanol and 800 µL of 0.5 M NaOH. The reaction was conducted at 70 °C for 2 h with gentle agitation every 30 min. After derivatization, each solution was neutralised at 4 °C with 0.5 M HCl. The reaction by-products were removed by extraction with 2 ml of dichloromethane, repeated twice more (three extractions in total). The remaining aqueous phase containing the PMP-monosaccharide derivatives was collected for analysis.

The PMP-monosaccharide derivatives were further purified by preparative HPLC using MeOH: H₂O (1:1, v/v) as the mobile phase. Standard D-glucose and L-glucose were derivatized and analysed under identical conditions. HPLC analysis was conducted under the conditions summarised in Table S1, and the results are presented in Figure S37.

Results and discussion

Structural elucidation of compounds from the leaves of H. macrophylla var. acuminata

From the leaves of H. macrophylla var. acuminata, a total of 41 compounds were isolated, including one new phenolic (3), three new iridoid glycosides (30–32), and 37 known compounds (1,2, 4–29, 33–41) (Figure 1). The structures of the isolated compounds were elucidated using various spectroscopic techniques and subsequently confirmed by comparison with published data. In the case of the megastigmanes (34–41), structural characterisation and stereochemical verification were performed by comparing the obtained data with the comprehensive review reported by Reham et al.²⁷ (Supplemental material).

Figure 1.

Chemical structures of isolated compounds (1–41) from the leaves of H. macrophylla var. acuminata.

Compound 3 was isolated as a yellow powder. Its molecular formula was determined as C₂₀H₂₆O₇ by HR-FAB-MS, with a sodium adduct molecular ion peak at m/z 401.1572 [M + Na]⁺ (calcd. for C₂₀H₂₆O₇Na, 401.1576) (Figure S7). The ¹H and 1³C NMR data of compound 3 are presented in Table 1. The doublet signals at δ_H 7.15 (H-2′, H-6′) and 7.04 (H-3′, H-5′) indicate the presence of a 1,4-disubstituted phenolic ring, which is further supported by HMBC correlations of H-2′ with C-1′ and C-4′, confirming the carbon chemical shifts associated with the 1,4-disubstituted aromatic system (Figure 2A). The singlet signal at δ_H 5.82 (H-2) suggests the presence of an olefinic proton, indicative of a double bond in the structure. Additionally, five methylene groups were observed, with resonances in the δ_H range of 1.99–2.83. The signal of the anomeric proton at δ_H 4.89 indicates the presence of a sugar moiety in the structure. The ¹³C NMR spectrum also shows a carbonyl carbon at δ_C 202.7, characteristic of a ketone functional group. HMBC correlations from H-2 (1H, δ_H 5.82) to C-1 (δ_C 169.6), C-3 (δ_C 202.7), C-4 (δ_C 30.7), and C-6 (δ_C 38.1), and from H-7 (2H, δ_H 2.57) to C-1, C-2 (δ_C 126.5), and C-6, along with COSY correlations from H-8 (2H, δ_H 2.83) to H-7 (2H, δ_H 2.57) (Figure 2A), support the presence of an ethylcyclohexenone moiety.²⁸ The position of this moiety was further confirmed by HMBC correlations from H-8 (2H, δ_H 2.83) to the signals δ_C 136.1 (C-1′), and δ_C 130.3 (C-2′, and C-6′).

Table 1.

¹H and ¹³C NMR data for compound 3 in CD₃OD.

Position	δH (J in Hz)	δ C
1	−	169.6
2	5.82, s	126.5
3	−	202.7
4	2.39, t (5.92)	30.7
5	1.99, m	23.8
6	2.35, m	38.1
7	2.57, t (7.65)	41.0
8	2.83, t (7.76)	33.6
1′	−	136.1
2′	7.15, d (8.47)	130.3
3′	7.04, d (8.05)	117.8
4′	−	157.6
5′	7.04, d (8.05)	117.8
6′	7.15, d (8.47)	130.3
1′′	4.89, overlap	102.5
2′′	3.42, overlap	75.0
3′′	3.47, overlap	78.1
4′′	3.43, overlap	71.4
5′′	3.47, overlap	78.0
6′′	3.91, m 3.71, m	62.6

Figure 2.

Key ¹H–¹H COSY (bold) and HMBC (→) correlations of compounds 3 (A), 30 (B), 31 (C), and 32 (D).

The ¹³C NMR spectra of this compound resemble those of prelunularin, except for the absence of a hydroxy group at the C-5 position and the addition of a sugar moiety at the C-4′ position.²⁹ The carbon chemical shifts at δ_C 102.5 (C-1′′), 75.0 (C-2′′), 78.1 (C-3′′), 71.4 (C-4′′), 78.0 (C-5′′), and 62.6 (C-6′′) are characteristic of a glucose unit, which was further confirmed through sugar analysis (Figure S37). The HMBC correlation between H-1′′ (δ_H 4.89) and C-4′ (δ_C 157.6) confirmed the attachment of the glu moiety at the C-4′ position (Figure 2A). Therefore, the structure of compound 3 was elucidated and named as hydrageneone.

Compound 30 was isolated as a yellow powder. Its molecular formula was identified as C₃₄H₅₀O₁₉ by HR-FAB-MS, with a sodium adduct molecular ion peak at m/z 785.2849 [M + Na]⁺ (calcd. for C₃₄H₅₀O₁₉Na, 785.2844) (Figure S16). The ¹H and ¹³C NMR data for compound 30 are summarised in Table 2. The ¹³C NMR spectrum revealed 34 distinct carbon signals, corresponding to 34 carbon atoms. Analysis of both the ¹H and ¹³C NMR spectra confirmed the presence of two methoxy groups (δ_C 51.7 and 53.2; δ_H 3.67 and 3.29), one methyl group (δ_C 20.9; δ_H 1.06), a double bond (δ_C 120.1; δ_H 5.27), and two carboxy carbon (δ_C 169–171). Additional signals corresponding to a conjugated cyclic double bond (δ_C 111–113 and 152–153) and a glucose unit (δ_C 62–100) support the identification of compound 30 as a novel bis-iridoid glucoside, comprising two parts: iridoid A (IA) and iridoid B (IB).

Table 2.

¹H (500 MHz) and ¹³C NMR (125 MHz) data for compounds 30, 31, and 32.

Position	30		31		32
	δH (J in Hz)a	δ C b	δH (J in Hz)a	δ C b	δH (J in Hz)a	δ C b
1a	5.13, d, (5.61)	97.8	5.38, d, (3.20)	97.4	5.40, d, (4.32)	97.7
3a	7.37, s	152.2	7.39, d, (1.28)	153.1	7.40, d, (1.45)	153.0
4a	−	113.6	−	110.3	−	111.6
5a	2.87, m	35.3	3.35, m	28.1	2.97, m	28.31
6a	2.16, m 1.36, m	33.3	2.98, m 2.15, dd, (9.75, 16.51)	35.1	1.53, m 2.20, m	32.6
7a	1.84, m 1.28, m	34.1	−	176.0	4.47, dd, (4.25, 7.37)	104.3
8a	1.95, m	36.5	5.59, m	134.4	5.67, m	135.9
9a	1.72, m	49.3	2.83, m	45.1	2.70, m	45.0
10a	1.06, d, (6.72)	20.9	5.22, m	120.7	5.28, m	120.3
1a′	4.63, overlap	100.5	4.63, d, (7.65)	100.0	4.64, d, (6.21)	100.1
2a′	2.27, overlap	74.3	3.34, overlap	74.2	3.39, m	74.1
3a′	3.27, m	78.3	3.34, overlap	78.1	3.35, overlap	78.1
4a′	3.47, m	71.2	3.43, m	70.8	3.56, m	70.7
5a′	3.36, overlap	77.2	3.34, overlap	77.8	3.35, overlap	77.8
6a′	3.80, d, (2.38)	65.1	3.81, 2H, m	65.2	3.82, dd, (1.56, 12.08) 3.89, dd, (2.91,11.63)	65.5
1b	5.44, d, (4.25)	97.7	5.48, d, (3.48)	97.7	5.46, d, (3.08)	97.5
3b	7.40, s	153.1	7.43, d, (1.91)	153.3	7.41, d, (1.01)	152.9
4b	−	111.5	−	111.4	−	111.5
5b	2.99, m	28.3	3.00, m	27.9	3.04, overlap	27.8
6b	2.25, m 1.54, m	33.3	2.35, m 1.44, m	32.7	1.46, m 2.38, m	32.7
7b	4.69, dd, (4.29, 7.48)	103.6	4.69, dd, (3.62, 8.09)	103.7	4.72, dd, (3.76, 8.13)	103.8
8b	5.69, m	135.7	5.67, m	135.6	5.72, m	135.6
9b	2.83, m	44.6	2.96, m	44.0	3.04, overlap	43.8
10b	5.30, m	120.1	5.28, m	120.3	5.33, m	120.1
1b′	4.64, overlap	99.9	4.60, d, (7.95)	100.0	4.66, d, (6.05)	100.1
2b′	3.33, overlap	74.2	3.50, m	73.9	3.53, m	73.8
3b′	3.33, overlap	78.1	3.24, m	78.1	3.26, m	78.1
4b′	3.36, overlap	71.1	3.53, m	70.7	3.47, m	70.6
5b′	3.36, overlap	77.2	3.34, overlap	77.5	3.35, overlap	77.5
6b′	3.86, dd, (1.90, 11.85)	62.3	3.87, dd, (2.09, 11.89) 3.72, dd, (4.50, 11.93)	61.9	3.91, dd, (1.96, 12.08) 3.76, dd, (4.26, 11.94)	61.8
4a-COOH	−	171.5	−	−	−
4a-COOCH3	−	−		169.0		169.2
4a-COOCH3	−		3.66, s	51.7	3.70	51.7
4b-COOCH3	−	169.1	−	169.1		169.1
4b-COOCH3	3.67, s	51.7	3.66, s	51.7	3.69	51.7
7b-OCH3	3.29, s	53.2	3.27, s	53.0	3.29	52.9
7a-COOCH3	−	−	−	−	3.28	52.8
7a-COOCH3	−	−	−	−	3.28	53.9

Recorded in CD₃OD; ^a500 MHz; ^b125 MHz

According to the HMBC spectra (Figure 2B), the observed correlations from H-3a (1H, δ_H 7.37) to C-4a (δ_C 113.6), C-5a, and C-1a; from H-1a (1H, δ_H 5.13) to C-5a, C-9a, and C-8a; from H-3b (1H, δ_H 7.40) to C-4b (δ_C 111.5), C-5b, and C-1b; and from H-1b (1H, δ_H 5.44) to C-5b, C-9b, and C-8b (δ_C 135.7), support the presence of two iridoid skeletons, each bearing a cyclic double bonds. HMBC correlations from H-10a (3H, δ_H 1.06) to C-8a, C-7a, and C-9a, and from H-5a (1H, δ_H 2.87) to C-6a, C-7a, C-9a, and C-1a, suggest the presence of a methyl-substituted cycloheptane ring in the IA structure. Additionally, HMBC correlations from H-3a (1H, δ_H 7.37) and H-5a (1H, δ_H 2.87) to the carbon signal at δ_C 171.5 confirm the location of the carboxyl group at the C-4a position. NOESY correlations of H-8a with H-1a, H-5a with H-9a, and H-10a with H-5a, with the absence of correlation between H-1a and H-5a or between H-5a and H-8a (Figure 3A), indicate that H-1a and H-8a were α-oriented, whereas H-5a, H-9a, and H-10a were β-oriented. The absolute configuration of C-8a was determined by comparing its ¹³C chemical shifts with those of deoxyloganic acid isomers. Specifically, the 8aR configuration exhibits a ¹³C resonance at approximately 32.5 ppm, whereas the 8aS configuration shows a resonance at approximately 36.1 ppm.³⁰ Thus, the recorded ¹³C chemical shift at δ_C 36.5 (C-8a) indicated an S configuration for C-8a, and the structure of the IA was further confirmed by comparison with the NMR data of 7-deoxyloganic acid.³¹

Figure 3.

Key NOESY correlations of compounds 30 (A), 31 (B), and 32 (C).

Similarly, HMBC correlations from H-3b (1H, δ_H 7.40) to C-1b, C-4b, and C-5b, and from H-8b (1H, δ_H 5.69) to C-1b, C-5b, C-9b, and C-10b, along with COSY correlations between H-8b (1H, δ_H 5.69) and H-10b (2H δ_H 5.30), and between H-8b and H-9b (1H, δ_H 2.83) (Figure 2B), confirm the position of the terminal double bond at C-9b. Moreover, HMBC correlations from H-3b (1H, δ_H 7.40) and the methoxy proton signal at δ_H 3.67 to the carbon resonance at δ_C 169.2 confirmed the presence of a methyl ester functional group at C-4b. The methoxy proton signal observed at δ_H 3.29 exhibited a clear HMBC correlation to the carbon resonance at δ_C 103.6 (C-7b), thereby confirming its attachment at C-7b. Consequently, the recorded ¹³C chemical shifts of IB closely resembled those of secologanin dimethylacetal, with the notable exception of the absence of a methoxy group at C-7b position.³²

Signals corresponding to 12 carbons (δ_C 62–100), including two anomeric carbons [δ_C 100.5 (C-1a′) and 99.9 (C-1b′)], indicate the presence of two sugar moieties, both identified as glucose (glu) units. Based on COSY correlations (Figure 2B), the carbon signals of each sugar moiety were assigned as follows: δ_C 100.5 (C-1a′), 74.3 (C-2a′), 78.3 (C-3a′), 71.2 (C-4a′), 77.2 (C-5a′), and 65.1 (C-6a′) for the first glucose moiety, and δ_C 99.9 (C-1b′), 74.2 (C-2b′), 78.1 (C-3b′), 71.1 (C-4b′), 77.2 (C-5b′), and 62.3 (C-6b′) for the second. The sugar analysis further confirmed D-glucose in the structure (Figure S37). The position of each glu moiety was established through HMBC correlations of their anomeric protons. Specifically, the HMBC correlation between H-1a′ (δ_H 4.63) and C-1a (δ_C 97.8) indicated the attachment site of the first glu moiety, whereas the correlation between H-1b′ (δ_H 4.64) and C-1b (δ_C 97.7) indicated the position of the second glucose unit. Building on these findings, the structure of compound 30 comprises two iridoid skeletons, connected through a glycosidic bond between C-6a′ and C-7b, as further supported by the HMBC correlation between H-7b (δ_H 4.69) and C-6a′ (δ_C 65.1). Therefore, the structure of compound 30 was identified as a bis-iridoid glucoside consisting of 7-deoxyloganic acid and secologanin dimethylacetal moieties.

Overall, the NMR spectroscopic data of compound 30 closely matched those of chrysathain, a constituent isolated from Lonicera chrysatha, with the primary difference being the presence of a methyl-substituted cycloheptane ring in the IA moiety of compound 30, compared with the terminal double bond present in chrysathain.³³ A notable difference was observed in the ¹H NMR chemical shift of the H-7b proton between chrysathain and compound 30. In chrysathain, H-7b appeared at δ_H 4.64 (1H, t, J = 7.8 Hz), whereas in compound 30, it was detected at δ_H 4.69 (1H, dd, J = 4.29, 7.48 Hz). This variation is likely attributable to the difference in stereochemical configuration (R/S) at C-7b.33,34

The intense NOESY correlations observed between H-7b and H-5b, H-9b, and H-10b, along with the weak correlation between 7b-OCH₃ and H-1b (Figure 3B), closely resemble the NOESY interactions reported for (7 R)-secologanin n-butyl methyl acetal.³⁴ Furthermore, the ¹H NMR chemical shift of H-7 in (7 R)-secologanin n-butyl methyl acetal was recorded at δ_H 4.54 (1H, dd, J = 4.8, 7.8 Hz),³⁵ showing coupling constants closely resembling those of H-7b in compound 30 (δ_H 4.69, 1H, dd, J = 4.29, 7.48 Hz). This similarity in coupling patterns supports the assignment of the R absolute configuration at C-7b in compound 30. Therefore, the structure of compound 30 was elucidated and named as hydrasecologanin A.

Compound 31 was isolated as a yellow powder. Its molecular formula was identified as C₃₅H₅₀O₂₁ by HR-FAB-MS, with a sodium adduct molecular ion peak at m/z 829.2745 [M + Na]⁺ (calcd. for C₃₅H₅₀O₂₁Na, 829.2742) (Figure S25). The ¹³C NMR spectrum of compound 31 revealed 35 distinct carbon signals, corresponding to 35 carbon atoms (Figure S19). Analysis of both the ¹H and ¹³C NMR spectra confirmed the presence of three methoxy groups (δ_C 51.7 − 53.0; δ_H 3.27 − 3.66), two double bonds (δ_C 120.7, δ_H 5.22 − 5.28), and three carboxy carbon (δC 169.0 − 176.0). Additional signals corresponding to a conjugated cyclic double bond (δ_C 110.3–111.4 and 153.1–153.3) and two glucose units (δ_C 62–100) support the identification of compound 31 as a novel bis-iridoid glucoside, comprising two components: IA and IB.

The ¹³C chemical shifts of IA in compound 31 were similar to those of compound 30, except for the absence of the signals corresponding to a methyl-substituted cycloheptane ring and the addition of a carboxyl group at C-7a, which was confirmed through the HMBC correlation of H-5a (δ_H 3.00) with C-7a (δ_C 176.0). Therefore, IA was identified as structurally similar to secoxyloganin, a compound isolated from Lonicera periclymenum.³⁶ The ¹³C chemical shifts of IB closely resembled those of compound 30, notably exhibiting a terminal double bond signal at C-9b. This structural feature was confirmed by HMBC correlations from H-10b and H-8b to C-9b, as well as COSY correlations between H-10b and H-8b, and between H-8b and H-9b (Figure 2C).

Signals in the δ_C 62–100 range indicate the presence of two sugar moieties in the structure, both identified as glucose units, confirmed by HMBC and COSY correlations (Figure 2C). The first glu moiety exhibited δ_C 100.0 (C-1a′), 74.2 (C-2a′), 78.1 (C-3a′), 70.8 (C-4a′), 77.8 (C-5a′), 65.2 (C-6a′), and the second exhibited δ_C 100.0 (C-1b′), 73.9 (C-2b′), 78.1 (C-3b′), 70.7 (C-4b′), 77.5 (C-5b′), and 61.9 (C-6b′). These glu moieties were further analysed using HPLC analysis of PMP-glucose derivatives (Figure S37). The attachment positions of these glucose moieties were determined through HMBC correlations of their respective anomeric protons. Specifically, the correlation between H-1a′ (δ_H 4.63) and C-1a (δ_C 97.4) confirmed the attachment site of the first glu unit, whereas the correlation between H-1b′ (δ_H 4.60) and C-1b (δ_C 97.7) identified the position of the second glu moiety.

The NOESY correlations observed between H-5a and H-9a; H-1a and H-10a; H-5b and H-9b; H-1b and H-10b, combined with the absence of correlations between H-1a and H-5a; H-5a and H-10a; H-1b and H-5b; H-5b and H-10b (Figure 3B), indicate that H-5a, H-9a, H-5b, and H-9b are β-oriented, whereas H-1a and H-1b are α-oriented. This NOESY pattern closely resembled that of compound 30, except for correlations associated with the methyl group in compound 30. Notably, the ¹H chemical shift of H-7b in compound 31 was recorded at δ_H 4.69 (1H, dd, J = 3.62, 8.09 Hz), showing a coupling constant pattern similar to that of compound 30 and (7 R)-secologanin n-butyl methyl acetal.³⁴ Based on this evidence, the absolute configuration at C-7b was assigned as the R-enantiomer. Compound 31 was characterised as a conjugate of secoxyloganin and secologanin dimethylacetal, linked via a glycosidic bond between C-6a′ and C-7b, as supported by HMBC correlation from H-7b (δ_H 4.69) to C-6a′ (δ_C 65.2). Consequently, the structure of compound 31 was elucidated and designated as hydrasecologanin B.

Compound 32 was isolated as a yellow powder. Its molecular formula was identified as C₃₇H₅₆O₂₁ by HR-FAB-MS, with a sodium adduct molecular ion peak at m/z 859.3217 [M + Na]⁺ (calcd. For C₃₇H₅₆O₂₁ Na, 859.3212) (Figure S34). The ¹H and ¹³C NMR data for compound 32 are summarised in Table 2. The ¹³C NMR spectrum revealed 37 distinct carbon signals, corresponding to 37 carbon atoms (Figure S28). Analysis of both the ¹H and ¹³C NMR spectra confirms the presence of five methoxy groups (δ_C 51.7–53.9; δ_H 3.28–3.70), two double bonds (δ_C 120.1–120.3; δ_H 5.28–5.33), and two carboxy carbon (δ_C 169.1–169.2). Additional signals corresponding to an oxygen-conjugated cyclic double bond (δ_C 111.5–111.6 and 152.9–153.0) and two glucose units (δ_C 62–100) support the identification of compound 32 as a novel bis-iridoid glucoside, comprising two parts: IA and IB.

The IA moiety exhibited features similar to those of compound 31, differing primarily by the absence of a carboxy group at the C-7a position and the presence of two methoxy groups. HMBC correlations between the two methoxy groups (δ_H 3.28) and C-7a, as well as between H-7a (δ_H 4.47) and C-5a/C-6a, and between H-5a (δ_H 2.97) and C-1a, C-3a, C-4a, and C-9a (Figure 2D), confirmed the presence of a dimethyl acetal moiety at C-6a. Additionally, the presence of a terminal double bond was confirmed by HMBC correlations from H-8a (δ_H 5.67) to C-1a, C-5a, and C-9a, along with COSY correlations between H-10a (δ_H 5.28) and H-8a, as well as between H-8a (δ_H 5.67) and H-9a (δ_H 2.70) (Figure 2D), indicating attachment at the C-9a position. These findings elucidated the IA skeleton, which closely resembled secologanin dimethyl acetal.³²

The observed ¹³C chemical shifts for the IB moiety in compound 32 were similar to those of IA, except for the absence of the dimethyl acetal group, rather than the addition of a methyl hemiacetal group at C-6b. The structure of IB was confirmed by HMBC correlations of H-5b (δ_H 3.04) with C-1b, C-3b, C-4b, C-9b, C-6b, and C-7b, and of H-10b (δ_H 5.33) with C-8b and C-9b. Thus, the structure of IB in compound 32 was elucidated, closely resembling that of compound 31. Notably, the ¹H NMR signal at H-7b (4.72, dd, J = 3.76, 8.13 Hz) was identical to that of compound 31, and the NOESY correlations recorded in compound 32 closely resembled those of compound 31 (Figure 3C). Additionally, compound 32 shared structural similarity with chrysathain, differing mainly in the ¹H chemical shift at H-7b. In chrysathain, H-7b in chrysathain appeared at δ_H 4.64 (t, J = 7.8 Hz),³³ whereas in compound 32, it was observed at δ_H 4.72 (dd, J = 3.76, 8.13 Hz). This variation in chemical shift and coupling pattern reflects a difference in the absolute configuration at C-7b, as previously discussed. Consequently, the absolute configuration at C-7b was assigned as the R-enantiomer. Based on these spectroscopic analyses, compound 32 was determined to comprise two secologanin dimethyl acetal units linked via a glycosidic bond between C-6a′ and C-7b, as supported by an HMBC correlation from H-7b to C-6a′. Therefore, the structure of compound 32 was elucidated and named as hydrasecologanin C.

PTP1B inhibitory activity of isolated compounds

The PTP1B inhibitory activities of various extracts derived from H. macrophylla var. acuminata along with isolated compounds were evaluated using ursolic acid as the positive control, and the results are summarised in Table 3. The EtOAc extract demonstrated the strongest inhibition, with an IC₅₀ value of 3.1 ± 1.0 µg/mL, indicating high potency against PTP1B. Both the n-hexane and MeOH extracts also exhibited notable activity, with IC₅₀ values of 14.2 ± 3.4 µg/mL and 78.2 ± 2.2 µg/mL, respectively. In contrast, the aqueous fraction showed the weakest inhibitory effect, requiring a much higher concentration (IC₅₀ = 149.8 ± 2.0 µg/mL) to achieve 50% enzyme inhibition. These results suggest that medium-polarity metabolites, particularly those obtained in the EtOAc fraction, may be primarily responsible for the observed PTP1B inhibitory activity.

Table 3.

PTP1B and α-glucosidase inhibitory activity of isolated compounds (1–41).

Compound	PTP1B			α-glucosidase
	IC50 (µM)	Inhibition type	Ki value (µM)	IC50 (µM)	Inhibition type	Ki value (µM)
MeOH extract	78.2 ± 2.2 (µg/ml)	−	−	735.1 ± 1.5 (µg/ml)	−	−
n-hexane extract	14.2 ± 3.4 (µg/ml)	−	−	83.2 ± 0.4 (µg/ml)	−	−
EtOAc extract	3.1 ± 1.0 (µg/ml)	−	−	32.2 ± 1.1 (µg/ml)	−	−
Water layer	149.8 ± 2.0 (µg/ml)	−	−	3128.1 ± 3.3 (µg/ml)	−	−
1	>100	−	−	>100	−	−
2	>100	−	−	>100	−	−
3	>100	−	−	>100	−	−
4	>100	−	−	>100	−	−
5	>100	−	−	>100	−	−
6	>100	−	−	>100	−	−
7	>100	−	−	>100	−	−
8	>100	−	−	21.9 ± 0.4	Uncompetitive	Kiu = 14.3
9	>100	−	−	43.8 ± 2.1	Mixed	Kic = 41.4 Kiu = 28.6
10	>100	−	−	> 100	−	−
11	>100	−	−	> 100	−	−
12	8.0 ± 1.1	Mixed	Kic = 12.2 Kiu = 5.5	3.4 ± 0.2	Uncompetitive	Kiu = 0.7
13	>100	−	−	>100	−	−
14	>100	−	−	>100	−	−
15	>100	−	−	>100	−	−
16	>100	−	−	>100	−	−
17	>100	−	−	>100	−	−
18	>100	−	−	>100	−	−
19	>100	−	−	>100	−	−
20	>100	−	−	>100	−	−
21	>100	−	−	>100	−	−
22	>100	−	−	>100	−	−
23	70.2 ± 2.0	−	−	>100	−	−
24	>100	−	−	>100	−	−
25	>100	−	−	>100	−	−
26	87.5 ± 4.4	−	−	>100	−	−
27	>100	−	−	>100	−	−
28	>100	−	−	>100	−	−
29	>100	−	−	>100	−	−
30	>100	−	−	>100	−	−
31	>100	−	−	>100	−	−
32	>100	−	−	>100	−	−
33	>100	−	−	>100	−	−
34	>100	−	−	>100	−	−
35	>100	−	−	>100	−	−
36	>100	−	−	>100	−	−
37	>100	−	−	>100	−	−
38	>100	−	−	>100	−	−
39	>100	−	−	>100	−	−
40	>100	−	−	>100	−	−
41	>100	−	−	>100	−	−
Ursolic acida	11.6 ± 4.7	−	−	−	−	−
Acarboseb	−	−	−	251.3 ± 3.1	−	−

^a Positive control for PTP1B

^b Positive control for α-glucosidase

(−) No test

As shown in Table 3, among the tested compounds, compound 12, which was isolated from the EtOAc extract, exhibited the strongest PTP1B inhibitory activity, with an IC₅₀ value of 8.0 ± 1.1 µM, surpassing the potency of ursolic acid (IC₅₀ = 11.6 ± 4.7 µM).

None of the isolated phenolics (1–6), isocoumarins (7–11), coumarins (13–15), flavonoids (16–17), iridoids (18–33), or megastigmanes (34–41) exhibited inhibitory activity against PTP1B. Notably, among the isocoumarins, only compound 12 showed significant PTP1B inhibition, with an IC₅₀ value of 8.0 ± 1.1 µM, whereas compounds 7–11 were inactive (IC₅₀ > 100 µM). The key structural distinction of compound 12 is the presence of a double bond in the δ-lactone ring, which appears to be crucial for its potent inhibitory effect on PTP1B (Figure S38). Lactones are widely distributed in natural compounds, and α,β-unsaturated δ-lactone rings, in particular, are known to exhibit a broad spectrum of biological activities, including cytotoxic,³⁷ anticancer,³⁸ antifungal and antiviral,³⁹ and antibacterial effects.⁴⁰ These activities are generally attributed to their function as Michael acceptors, enabling covalent interaction with target enzymes via nucleophilic amino acid residues.⁴¹

Moreover, although hydrangenol (9), a bioactive compound isolated from the leaves of H. macrophylla var. thunbergii has been reported to possess antidiabetic potential by upregulating adiponectin, PPARγ2, and GLUT4 mRNA, while downregulating IL-6 mRNA expression.⁵ However, it showed no inhibitory activity against PTP1B in this study. This finding further underscores the critical role of the double bond in the δ-lactone ring of compound 12 as a key structural feature for potent PTP1B inhibition, highlighting arylidene isocoumarins as promising scaffolds for the development of PTP1B inhibitors.

α-Glucosidase inhibitory activity of isolated compounds

The α-glucosidase inhibitory potential of the same extract series from H. macrophylla var. acuminata was also evaluated (Table 3). The ethyl EtOAc extract again emerged as the most active fraction, displaying a potent IC₅₀ value of 32.2 ± 1.1 µg/mL. The n-hexane extract showed moderate activity (IC₅₀ = 83.2 ± 0.4 µg/mL), whereas the crude MeOH extract exhibited weaker inhibition (IC₅₀ = 735.1 ± 1.5 µg/mL). The aqueous layer was significantly less effective, with an IC₅₀ of 3128.1 ± 3.3 µg/mL, over 97-fold higher than that of the EtOAc fraction. In agreement with the PTP1B findings, the strong α-glucosidase inhibition by the EtOAc extract reflects its high content of active secondary metabolites.

Among the tested compounds, α-glucosidase inhibition was exclusively observed within the isocoumarin group (7–12), whose activity varied markedly with structural modifications and substitution patterns. All other compound classes, including phenolics (1–6), coumarins (13–15), flavonoids (16–17), iridoids (18–33), and megastigmanes (34–41), showed no inhibitory effects.

A detailed comparison of the isocoumarins containing a γ-lactone ring (compounds 7 and 8) revealed that hydramacrophyllol B (8) exhibited potent α-glucosidase inhibitory activity, with an IC₅₀ value of 21.9 ± 0.4 µM, significantly stronger than acarbose (IC₅₀ = 251.3 ± 3.1 µM), whereas hydramacrophyllol A (7) was inactive (IC₅₀ > 100 µM). The primary structural difference between the two compounds is the absolute configuration at C-8. Notably, the 8 R configuration in compound 8 correlates with enhanced inhibitory potency, whereas the 8S configuration in compound 7 is associated with a lack of activity (Figure S38). This stereochemistry-dependent activity highlights the importance of enantiomer selection in drug development, as seen in optimised pharmaceutical ingredients such as esomeprazole, the purified S-enantiomer of omeprazole.⁴² Therefore, the 8 R stereoisomer of isocoumarins containing a γ-lactone ring and a hydroxy group at C-8 is critical for significantly improving α-glucosidase inhibitory activity.

While hydrangenol (9) demonstrated no inhibitory effect on PTP1B, it exhibited potent α-glucosidase inhibition (IC₅₀ value = 43.8 ± 2.1 µM), significantly more potent than acarbose (IC₅₀ = 251.3 ± 3.1 µM). Notably, despite their structural similarity to hydrangenol (9), compounds 10 and 11 exhibited no α-glucosidase inhibitory activity. This finding suggests that, in isocoumarins bearing a δ-lactone ring, the addition of a hydroxy group at C-6 (compound 10) or a glucose moiety at C-8 (compound 11) significantly diminishes inhibitory potency (Figure S38). Combined with previously reported antidiabetic effects of hydrangenol,⁵ these findings identify α-glucosidase as a novel molecular target, underscoring its therapeutic potential in diabetes management through multiple mechanisms.

Thunberginol A (12), beyond its potential as a PTP1B inhibitor, also exhibited strong α-glucosidase inhibitory activity, with an IC₅₀ of 3.4 ± 0.2 µM, over 70 times more potent than acarbose. Notably, most reported α-glucosidase inhibitors exhibit IC₅₀ values in the range of approximately 30 to 100 µM, with a few showing activity below 10 µM.⁴³ Thus, thunberginol A emerges as one of the most potent α-glucosidase inhibitors identified to date. Moreover, its dual inhibitory activity against both PTP1B and α-glucosidase highlights thunberginol A (12) as a promising lead compound for type 2 diabetes treatment via a dual-targeted mechanism.

It is noted that yeast α-glucosidase (S. cerevisiae) was used for this study, owing to the broad availability of purified enzyme preparations.⁴⁴ In general, α-glucosidases constitute a broadly distributed enzyme family present in yeasts, fungi, bacteria, plants, archaea, and animals. Despite their shared catalytic role, these proteins exhibit pronounced heterogeneity in their primary sequence, structure, and substrate specificity across species. As a result, both inhibitor potency and mechanism of inhibition are expected to vary across α-glucosidase isoforms.⁴⁵ Importantly, mammalian intestinal α-glucosidases (e.g. in rats) belong to Family II, whereas yeast enzymes fall into Family I.⁴⁶ Furthermore, intestinal α-glucosidase activity in rats is mediated by four distinct enzymes: maltase, isomaltase, sucrase, and glucoamylase, whereas yeast-derived α-glucosidase preparations predominantly display maltase activity. Consequently, assays employing yeast versus rat α-glucosidase preparations may yield substantially different IC₅₀ values and kinetic parameters for identical compounds.⁴⁷ Yonekura et al. found that the IC₅₀ of an olive extract against human α-glucosidase was roughly one-tenth of that measured for the yeast enzyme, indicating substantially greater potency towards the human isoform.⁴⁴ Likewise, although acarbose potently inhibits human α-glucosidase, its activity is markedly reduced against yeast and rat intestinal enzymes. Barber’s review revealed that quercetin more potently inhibits yeast α-glucosidases than acarbose, whereas its activity against rat maltase (IC₅₀ = 281.2 µM) and sucrase (IC₅₀ > 400 µM) is markedly weaker.⁴⁴ These observations underscore the critical role of selecting both an appropriate substrate and enzyme source to accurately evaluate a compound’s inhibitory potency. Hence, further in vitro investigations using human α-glucosidase isoforms are essential to thoroughly evaluate the inhibitory potency of the compounds identified in this study. These data will enable direct comparison with our current findings and clarify the compounds’ potential as α-glucosidase inhibitors for the treatment of T2D.

PTP1B enzyme kinetic analysis

Based on the observed PTP1B inhibitory activity, enzyme kinetic analysis was performed to elucidate the inhibition mode and determine the inhibition constant (K_ic and K_iu) of compound 12. The Lineweaver–Burk plot was employed to characterise the type of inhibition. The inhibition mode is inferred by examining the intersection point of the lines: intersection at the y-axis indicates competitive inhibition; at the x-axis, non-competitive inhibition; between the axes, mixed-type inhibition; and parallel lines indicate uncompetitive inhibition.⁴⁸ As shown in Figure 4(A), the lines intersect in the x-y region, indicating that compound 12 inhibits PTP1B via a mixed-type mechanism. Therefore, secondary plot analysis was performed to determine whether compound 12 preferentially inhibits the free enzyme or the enzyme-substrate complex.⁴⁹ As illustrated in Figure 5(A,B), K_ic > K_iu suggests that the inhibitor preferentially binds to the enzyme-substrate complex, providing a rationale for subsequent molecular docking studies.

Figure 4.

Lineweaver–Burk plots of compound 12 with PTP1B (A), and compounds 8, 9, and 12 with α-glucosidase (B–D), respectively.

Figure 5.

Secondary plots of compound 12 with PTP1B (A, B), and compounds 8 (C), 9 (D, E), and 12 (F) with α-glucosidase.

α-Glucosidase enzyme kinetic analysis

The enzyme kinetics of α-glucosidase were evaluated using a methodology analogous to that of PTP1B, employing Lineweaver–Burk plots to determine the mode of inhibition and secondary plots to calculate the inhibition constants (K_ic and K_iu) of the most potent inhibitors. Based on the α-glucosidase inhibitory activity results, compounds 8, 9, and 12 demonstrated notable inhibitory effects. Consequently, enzyme kinetics were conducted on these three compounds. As shown in Figure 4(B), the Lineweaver–Burk plot for compound 8 reveals parallel lines, indicating that compound 8 inhibits α-glucosidase via an uncompetitive mechanism. A similar trend is observed for compound 12 (Figure 4D), suggesting it also functions as an uncompetitive inhibitor. In contrast, the plot for compound 9 revealed intersecting lines in the x-y axis region, indicating that it inhibits α-glucosidase via a mixed-type mechanism (Figure 4C).

Based on these analyses, secondary plots were employed to determine the inhibition constants for the compounds. Specifically, the K_iu values were obtained for compounds 8 and 12, whereas both K_iu and K_ic values were determined for compound 9. As shown in Figures 5(C,F), compounds 8 and 12 exhibit K_iu values of 14.3 µM and 0.7 µM, respectively. Compound 9 was characterised as a mixed-type inhibitor, with a K_ic value of 41.4 µM and a K_iu value of 28.6 µM (Figure 5D,E). Notably, the observation that K_ic is higher than K_iu for compound 9 indicates a preferential binding to the enzyme-substrate complex. Overall, the low inhibition constant values of compounds 8, 9, and 12 underscore their potential as effective α-glucosidase inhibitors.

Molecular docking study

Based on the enzyme kinetic analysis, compound 12 was identified as a mixed-type inhibitor of PTP1B, exhibiting a stronger affinity for the enzyme-substrate complex. To further elucidate its interaction mechanism, molecular docking simulations were performed comparing compound 12 with compound A, a known allosteric inhibitor (Figure 6A). As shown in Table 4, compound 12 demonstrates a notably low binding affinity of −8.4 kcal/mol, which is lower than that of compound A (–7.8 kcal/mol). This lower binding energy suggests a stronger and more stable interaction between compound 12 and PTP1B. Interestingly, molecular docking analysis (Figure 6B) suggests that while the α,β-unsaturated δ-lactone moiety contributes to the overall binding affinity and correct positioning within the binding site, the specific, high-affinity interactions responsible for stabilising the complex involve other functional groups. Specifically, compound 12 interacts with PTP1B primarily via van der Waals interactions involving residues Ala189, Asn193, Glu276, and Gly277. Hydrogen bonds are formed between the hydroxyl group at C-4′ and Glu200, and between the hydroxyl group at C-3′ and Lys197. An amide-π stacking interaction is observed between the aryl moiety at the C-3 position of compound 12 and Phe196, accompanied by a π-alkyl interaction between the aryl portion of the coumarin core and Leu192. Furthermore, a π-π stacking interaction between Phe280 and the coumarin ring further stabilises the binding of compound 12 to PTP1B. This evidence indicates that the α,β-unsaturated δ-lactone is a necessary structural scaffold, but the hydroxyl groups act as key pharmacophores mediating direct contact with the target enzyme.

Figure 6.

Molecular docking illustrating PTP1B inhibition by compound A (blue stick) and compound 12 (magenta stick) (A), and 2D diagram of PTP1B inhibition by compound 12 (B).

Table 4.

Interactions analysis and binding affinities of compound 12 and compound A (an allosteric site inhibitor) with PTP1B.

Compound	Binding affinity (kcal/mol)	Van der Waals interactions	Hydrogen bonds	Other interactions
				Amide-π stacked	π-alkyl	Others
12	−8.4	Ala189, Asn193, Glu276, Gly277	Lys197, Glu200	Phe196	Leu192	Phe280 (π-π stacked)
A	−7.8	Ser187, Glu276	Asn193	Leu192	Phe196, Phe280, Ile281	Ile281 (alkyl), Phe280 ((π-π stacked), Ala189 (π-sigma), Pro188 (carbon-hydrogen bond)

Enzyme kinetic analysis of α-glucosidase indicates that compounds 8 and 12 act as uncompetitive inhibitors, binding preferentially to the enzyme-substrate complex. In contrast, compound 9 functions as a mixed-type inhibitor, also demonstrating a higher affinity for the enzyme-substrate complex. To date, six allosteric sites of α-glucosidase have been identified.⁵⁰ Therefore, blind docking simulations were performed to determine the allosteric binding sites for these inhibitors. The results reveal that compounds 8, 9, and 12 exhibit the strongest binding affinity at allosteric site 1, compared to genistein, a known allosteric inhibitor of α-glucosidase (Figure 7A).⁵¹

Figure 7.

Molecular docking illustrating α-glucosidase inhibition by gentisein (reference inhibitor, magenta stick), compound 8 (yellow), compound 9 (orange), and compound 12 (green). The active site is docked with p-nitrophenyl α-glucopyranoside (blue) (A). 2D diagrams of α-glucosidase inhibition by compound 8 (B), compound 9 (C), and compound 12 (D). Colour representations: van der Waals (light green), hydrogen bonds (dark green), carbon-hydrogen bond and π-donor hydrogen bonds (pale green), alkyl and π-alkyl interactions (light pink), π-π T-shaped (dark pink), and π-anion/π-cation interactions (orange).

As presented in Table 5, compound 8 exhibits a binding energy of −7.3 kcal/mol with α-glucosidase, equivalent to that of genistein, a known allosteric inhibitor. In contrast, compounds 9 and 12 demonstrate lower binding energies of −7.6 kcal/mol and −8.3 kcal/mol, respectively, indicating stronger interactions with the enzyme. More negative binding energies generally correspond to enhanced stability of the ligand-enzyme complex. These results suggest that compounds 8, 9, and 12 may possess enhanced allosteric inhibitory potential compared to genistein, highlighting them as promising candidates for further development.

Table 5.

Interactions analysis and binding affinities of compounds 8, 9, 12, and gentisein (an allosteric site inhibitor) with α-glucosidase.

Compound	Binding energy (kcal/mol)	Van der Waals interactions	Hydrogen bonds	Other interactions
				π-alkyl	π-π T-shaped	Others
8	−7.3	Trp14, His258, Met261, Arg269, Glu270, Tyr292, Val294, Ser295	−	Lys262, Val265, Ile271, Ala289	−	−
9	−7.6	Lys12, Met261, Lys262, Val265, Arg269, Val294, Ser295	−	Ala289	Trp14	His258 (carbon-hydrogen bond), Ile271(π-donor hydrogen bond), Glu270 (π-anion)
12	−8.3	Lys12, Met261, Lys262, Val265, Glu270, Ser295	Arg269, Val294	Ala289	Trp14	Ile271 (π-donor hydrogen bond)
Gentisein	−7.3	Gly268, Glu270, Met272, Val293, Val294	Arg269, Ser295	Lys262, Val265, Ile271	−	Ile271 (π-donor hydrogen bond), His258 (π-cation), Met261 (π-sulfur)

As illustrated in Figure 7(B), compound 8 interacts with α-glucosidase primarily through van der Waals forces and π-alkyl interactions. Specifically, van der Waals interactions are established between the aryl moiety and residues Trp14, Glu270, Val294, and Ser295, as well as between the coumarin scaffold and residues His258, Met261, Arg269, and Tyr292. Additionally, π-alkyl interactions occur between the aryl portion of the coumarin ring and residues Lys262, Val265, and Ile271, along with interactions between the aryl moiety and Ala289. Collectively, these interactions contribute to the stable binding conformation of compound 8 within the allosteric site of α-glucosidase.

Compound 9 interacts with α-glucosidase primarily through van der Waals forces and carbon-hydrogen bonds (Figure 7C). Van der Waals interactions occur between the aryl moiety and residues Met261, Lys262, Val265, and Arg269, as well as between the coumarin scaffold and residues Lys12, Val294, and Ser295. Carbon–hydrogen bonding is observed between the aryl moiety and Ile271, along with an interaction between the oxygen atom in the δ-lactone ring and His258. Additionally, the aryl portion of the coumarin ring engages in a π-π T-shaped interaction with Trp14, a π-alkyl interaction with Ala289, and a π-anion interaction with Glu270. Compared with compound 8, compound 9 forms a broader range of interactions, which likely contributes to its lower binding energy and stronger overall affinity for α-glucosidase.

Notably, compound 12 exhibited the strongest binding affinity with α-glucosidase among the tested compounds, with a binding energy of −8.3 kcal/mol, notably lower than that of genistein (–7.3 kcal/mol). This enhanced affinity can be attributed to a range of interactions, as illustrated in Figure 7(D). Specifically, van der Waals forces were established between the aryl moiety at C-3 and residues Met261, Lys262, and Val265, as well as between the coumarin part and residues Lys12, Glu270, and Ser295. In contrast to compounds 8 and 9, compound 12 formed hydrogen bonds with Arg269 and Val294 via hydroxyl groups at C-4′ and C-8, respectively. Additionally, the aryl portion of the coumarin ring engaged in a π-alkyl interaction with Ala289 and a π-π T-shaped interaction with Trp14. The aryl moiety at C-3 contributed a π-donor hydrogen bond with Ile271, whereas the δ-lactone ring formed a π-cation interaction with His258.

Molecular docking analysis further underscores the critical role of the hydroxyl groups at C-3′ and C-4′ in the biological activity of compound 12. In PTP1B, these hydroxyl groups form hydrogen bonds with Lys197 and Glu270, respectively, whereas in α-glucosidase, the hydroxyl group at C-4′ establishes a hydrogen bond with Arg269. These interactions indicate that the hydroxy groups play a critical role in the binding mechanism and inhibitory potential of compound 12 towards both PTP1B and α-glucosidase.

Conclusion

In this study, we comprehensively investigated the phytochemical constituents of H. macrophylla var. acuminata leaves, resulting in the successful isolation of 41 secondary metabolites. These included six phenolics (1–6), six isocoumarins (7–12), three coumarins (13–15), two flavonoids (16–17), sixteen iridoids (18–33), and eight megastigmanes (34–41). Among these, four compounds were identified as new molecules, including one phenolic (3) and three bis-iridoid glycosides (30–32). To the best of our knowledge, this represents the first report on the phytochemical profile of this subspecies. The biological activities of the isolated compounds were subsequently evaluated against PTP1B and α-glucosidase using in vitro assays, with ursolic acid and acarbose as positive controls. Among the tested compounds, compound 12 exhibited the most potent inhibitory activity against both enzymes, with IC₅₀ values of 8.0 ± 1.1 µM for PTP1B and 3.4 ± 0.2 µM for α-glucosidase. Additionally, compounds 8 and 9 demonstrated notable inhibitory effects against α-glucosidase, with IC₅₀ values of 21.9 ± 0.4 µM and 43.8 ± 2.1 µM, respectively. Enzyme kinetic analyses revealed that compound 12 functions as a mixed-type inhibitor of PTP1B and an uncompetitive inhibitor of α-glucosidase. Compound 8 exhibited uncompetitive inhibition of α-glucosidase, whereas compound 9 acted as a mixed-type inhibitor of the same enzyme. Molecular docking simulations further elucidated the binding interactions between these active compounds and their target enzymes. Overall, this study provides the first comprehensive phytochemical profile of H. macrophylla var. acuminata leaves and identifies promising dual inhibitors of PTP1B and α-glucosidase. These findings establish a foundation for future development of natural therapeutic agents targeting both PTP1B and α-glucosidase for potential applications in diabetes management.

Funding Statement

This research was supported by the National Research Foundation of Korea (NRF) funded by the Korea government (MSIT) (No. NRF-2020R1A5A2017323).

Author contributions statement

CRediT: Thi Ly Pham: Data curation, Formal analysis, Investigation, Methodology, Writing – original draft; Viet Phong Nguyen: Data curation, Formal analysis, Investigation, Methodology, Writing – review & editing; Thi Thuy An Nguyen: Formal analysis, Investigation, Methodology; Byung Sun Min: Resources, Supervision, Writing – review & editing; Jeong Ah Kim: Conceptualization, Funding acquisition, Supervision, Writing – review & editing.

Disclosure statement

The authors declare no competing financial interests.

Data availability statement

The datasets generated and/or analysed during the current study are available from the corresponding author upon reasonable request.