1-Deoxynojirimycin (DNJ) and fagomine (FAG) were purchased from MedChem Express (HY-14860 and HY-13005, USA). 1,4-Dideoxy-1,4-imino-D-arabinitol (DAB) was purchased from Sigma (D1542, USA). Morus alba L. (Sangzhi) alkaloid (SZ-A) powder (Lot number: 201708008; the total polyhydroxy alkaloids content in SZ-A powder is approximately 63%, which includes 39% DNJ, 10.5% FAG, and 7% DAB, as shown in Supplementary Figure S1) was synthesized and provided by Beijing Wehand-bio Pharmaceutical Co. Ltd. (Beijing, China).

C57BL/6J mice (male, 14 weeks) were fed a normal diet and comprised the normal group (Nor group). KKA^y mice (male, 14 weeks) were fed high-fat diets (45% of kcal from fat; D12451; Research Diets, USA). After 4 weeks, the KKA^y mice were randomly divided into three groups (Con group and two doses of SZ-A-treated groups, 10 mice per group) according to the levels of fasting blood glucose, triglycerides, total cholesterol, body weight and the percentage of increasing blood glucose level at 30 min after glucose (2.0 g/kg) overloading (Diagram of animal experiment was shown as Figure 1A). The SZ-A groups were orally administered SZ-A (100 mg/kg, SZ-A_100; 200 mg/kg, SZ-A_200) once per day. The Nor group and Con group were treated with vehicle. All animals were housed in communal cages at 23 ± 1°C on a 12-h light-dark cycle with free access to food and water. Blood glucose was tested in the 2nd week. The oral glucose tolerance test (OGTT), insulin tolerance test (ITT) and intraperitoneal glucose-stimulated insulin secretion test (IPGSIST) were carried out consecutively on the 5th, 6th, and 8th weeks. The hyperglycemia clamp test was conducted in the 11th week. Then, the mice were sacrificed via cervical dislocation, and the pancreas was separated and subsequently fixed with formalin. The experiments were conducted following the “3R” principles and the standards for laboratory animals (GB14925-2001 and MOST 2006a) established by the People’s Republic of China. The animal protocol was approved by the Institutional Animal Care and Use Committee of the Institute of Materia Medica, Chinese Academy of Medical Sciences and Peking Union Medical College (Beijing, China).

The mice were fasted for 6 h, and blood was obtained from the tail for the fasting blood glucose test. For the OGTT, blood samples were acquired at 0, 30, 60 and 120 min after oral glucose overload (2.0 g/kg) for the blood glucose test. For the ITT, mice were hypodermically injected with insulin (0.4 IU/kg), and subsequently, blood was taken from the tail at 40 and 90 min. For the IPGSIST, blood samples were acquired at 0 and 30 min after intraperitoneal injection of glucose (2.0 g/kg) for blood glucose and insulin tests.

The mice were anesthetized with amobarbital sodium after fasting overnight, and then 5% (w/v) glucose was perfused into the jugular vein through a peristaltic pump. During the first 1 min, a glucose bolus (100 mg/kg) was given. Blood was obtained from the tail at 0, 2, 5, 10, and 15 min after glucose loading for the phase I insulin secretion test. Subsequently, glucose was microperfused persistently until the blood glucose level remained steady at 14.0 ± 0.5 mM. The whole testing process took approximately 2.0–2.5 h, during which blood was obtained at 60, 100, and 120 min for the phase II insulin secretion test.

The fixed pancreases of KKA^y mice were embedded in paraffin. Tissues were sliced to 5 μm for morphology analyses. Some sections were stained with hematoxylin-eosin to observe the morphology of islets under a microscope. Others were immunostained with rabbit anti-glucagon antibody (1:200, ab92517, Abcam, USA), rabbit anti-ALDH1A3 antibody (1:100, NBP2-15339, NOVUS, USA), and rat anti-insulin antibody (1:200, MAB1417, R&D Systems, USA). The secondary antibodies were Alexa Fluor 488 donkey anti-rabbit IgG (1:200, A-21206, Invitrogen, USA) and Alexa Fluor 594 donkey anti-rat IgG (1:200, A-11007, Invitrogen, USA). Then, the tissues were mounted utilizing mounting medium with DAPI (ZLI-9557, ZSGB-BIO, China). Finally, the tissues were visualized by confocal laser scanning microscopy (CLSM) (Leica Microsystems, Germany). ImageJ software was used to analyze the picture. The area percentage of β-cell/islet and α-cell/islet were calculated via the insulin-positive or glucagon-positive area in one islet divided by the islet area. The ratio of ALDH1A3-positive β cells/β cells were calculated via the both ALDH1A3 and insulin-positive area divided by insulin-positive area in the islet.

The mouse β-cell line MIN6 was a gift from Professor Xiao Han at Nanjing Medical University, and the cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM, 25 mM glucose) supplemented with 15% fetal bovine serum (FBS) (Chen et al., 2016). Mouse islets were isolated from C57BL/6J mice (male, 24 ± 2 g) and diabetic KKA^y mouse model (male, 38 ± 2 g) using the collagenase V (C9263, Sigma, Germany) perfusion method and were cultured in RPMI-1640 medium supplemented with 10% FBS as reported previously (Lei et al., 2015).

MIN6 cells and mouse islets were prepared to test insulin secretion in response to low and high concentrations of glucose. MIN6 cells were plated in 96-well plates (5 × 10⁴/well) and starved for 1 h in Krebs buffer (2.8 mM glucose) with or without different concentrations of SZ-A, DNJ, FAG or DAB. Then, the cells were cultured for another 1 h in new Krebs buffer containing 2.8 mM or 16.8 mM glucose combined with SZ-A, DNJ, FAG, DAB or vehicle. Ten islets were transferred to new Krebs buffer containing 2.8 or 16.8 mM glucose and were cultured for 1 h after fasting for 1 h in Krebs buffer with 2.8 mM glucose, which was combined with SZ-A, DNJ, FAG, DAB or vehicle. The supernatant was collected for the insulin test using the Mouse Ultrasensitive Insulin ELISA Kit (80-INSMSU-E10, ALPCO, USA). MIN6 cells and mouse islets were used for quantitative protein determination with BCA reagents (P1511, APPLYGEN, China).

MIN6 cells in 96-well plates were cultured with 25, 50, and 100 μg/ml SZ-A; 10 μg/ml FAG; 10 μg/ml DAB; or vehicle for 1 h. The cells were then loaded with the Ca²⁺-sensing fluorescent probe Fluo-4 AM (F14201, Invitrogen, USA) for 30 min at 37°C. The internal calcium concentration ([Ca²⁺]_i) was measured using a Multimode Microplate Reader (Synergy 2, BIO-TEK, USA) through time-series recording of the fluorescent signal. A 5-min basal recording was acquired when the cells were in Krebs buffer containing 2.8 mM glucose. For stimulation, the cells were perfused with Krebs buffer containing elevated glucose concentration (16.8 mM) combined with SZ-A or vehicle. The changes in the MIN6 cell [Ca²⁺]_i were recorded for 20 min after stimulation.

Intracellular cAMP ([cAMP]_i)was measured with a LANCE cAMP Detection Kit (AD0262, PerkinElmer, USA) through time-resolved fluorescence resonance energy transfer (TR-FRET) technology. MIN6 cells were preincubated for 1 h in Krebs buffer containing 2.8 mM glucose and different concentrations of SZ-A or vehicle. Then, both Alexa Fluor 647-cAMP antibody solution and stimulation solution containing 25 μM isobutyl methylxanthine (312460, J&K, China) and different concentrations of SZ-A or vehicle were added to each well and cultured for 30 min. The europium chelate of the Eu-SA/b-cAMP tracer was subsequently added to test the cAMP production of MIN6 cells through a multilabel reader (Envision 2104, PerkinElmer, USA).

For high glucose (Glc) and palmitic acid (PA) treatment, MIN6 cells were incubated in DMEM with 15% FBS, 33 mM glucose and 0.25 mM palmitic acid. The blank control cells were incubated in DMEM with 15% FBS, 25 mM glucose and 8 mM mannitol (63559, Sigma, USA).

MIN6 cells were seeded at 3 × 10⁴/100 μl per well in 96-well plates and incubated under normal conditions. The next day, they were treated with 25, 50, or 100 μg/ml SZ-A or vehicle for 24 h. At the same time, high glucose- and palmitic acid-treated MIN6 cells were co-incubated with different concentrations of SZ-A, DNJ, FAG, DAB or vehicle for 24 h. Cell proliferation was detected by EdU labeling according to the recommended protocol of the assay kit (C0071, Beyotime, China). The cells were fixed with 4% paraformaldehyde, and the nuclei were labeled with Hoechst 33342. The fluorescence signal was detected at 530/25 nm (EdU signal) and 460/40 nm (Hoechst 33342 signal) with a multifunctional microplate reader (Synergy 2, BioTek, USA) and calculated as the ratio of Edu to Hoechst 33342.

High glucose- and palmitic acid-treated MIN6 cells were coincubated with different concentrations of SZ-A, DNJ, FAG, DAB or vehicle for 72 h and fixed with 4% paraformaldehyde. The TUNEL assay was performed according to the recommended procedure of the TransDetect ^® In Situ Fluorescein TUNEL Cell Apoptosis Detection Kit (FA201-01, TransGen Biotech, China). Finally, the cells were mounted utilizing mounting medium with DAPI (ZLI-9557, ZSGB-BIO, China). The images were visualized by inverted fluorescence microscopy (Olympus, Japan). The data are presented as the ratio of TUNEL/DAPI.

High glucose- and palmitic acid-treated MIN6 cells were coincubated with different concentrations of SZ-A, DNJ or vehicle for 24 h. Then, the activity of caspase 3 was tested using Caspase-Glo®3/7 Assays (G8981, Promega, USA). The signal was detected with a multifunctional microplate reader (Synergy 2, BioTek, USA).

The islets of KKA^y mice were incubated in RPMI-1640 medium containing vehicle or 100 μg/ml SZ-A for 24 h. MIN6 cells were fasted for 1 h in Krebs buffer (2.8 mM glucose) with or without different concentrations of SZ-A, FAG, or DAB. Then, they were cultured for another 1 h in new Krebs buffer containing 2.8 mM or 16.8 mM glucose combined with SZ-A, FAG, DAB or vehicle. High glucose- and palmitic acid-treated MIN6 cells were coincubated with different concentrations of SZ-A, DNJ or vehicle for 24 h.

Islets or MIN6 cells were homogenized in RIPA lysis buffer (C1053; APPLYGEN, China) containing protease inhibitors (P1265; APPLYGEN, China) and phosphatase inhibitors (P1260; APPLYGEN, China). Protein (10 μg) from each sample was resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis and transferred to polyvinylidene fluoride membranes for immunoblotting. Rabbit anti-phospho-AMPKα (1:1000, 2535s), rabbit anti-AMPKα (1:1000, 2532s), rabbit-phospho-Erk (1:1000, 9101s), rabbit anti-Erk (1:1000, 9102s), and rabbit anti-HSP90 (1:1000, 4877s) were purchased from Cell Signaling Technology (USA). Goat anti-rabbit IgG/HRP (1:5000, ZDR-5306) was purchased from ZSGB-BIO (China).

The islets of KKA^y mice were incubated in RPMI-1640 medium containing vehicle or 100 μg/ml SZ-A for 24 h. RNA was isolated using TRIzol Reagent (15596018, Life Technology, USA), and cDNA was synthesized using TransScript ^® One-Step gDNA Removal and cDNA Synthesis SuperMix (AT311, TransGen Biotech, China) for gene expression analysis. Quantitative PCR (qPCR) was performed with TransStart Tip Green qPCR SuperMix (AQ141–04, TransGen Biotech, China) using the primers shown in Table 1.

GraphPad Prism 8 was used for all statistical analyses. Two experimental groups were analyzed by Student’s t test. Multiple groups were compared by one-way analysis of variance (ANOVA) followed by Dunnett’s t test or two-way ANOVA with Tukey’s test depending on the experiments. Differences were considered statistically significant at p < 0.05.

The type 2 diabetic KKA^y mice have a similar metabolic syndrome phenotype as humans and exhibit hyperglycemia, hyperinsulinemia, hyperlipidemia and obesity. We have reported that SZ-A ameliorates glucose and lipid metabolism in KKA^y mice (Liu et al., 2021). In this study, we again identified that non-fasting and fasting blood glucose was reduced significantly with SZ-A treatment (Figures 1C,D), which may be ascribed to the increased insulin secretion and sensitivity. In order to identify these further, OGTT, ITT and IPGSIST were carried out sequentially. The OGTT and ITT showed that SZ-A alleviated hyperglycemia after oral glucose load (Figures 1E,F) and IP injection of insulin (Figures 1G,H) in a dose-dependent manner. Notably, 200 mg/kg SZ-A could ameliorate glucose tolerance significantly. At the same time, body weight in the two SZ-A treatment groups decreased significantly (Figure 1B).

IPGSIST was carried out in KKA^y mice. We observed that at either 0 min or 30 min after IP injection of glucose (2.0 g/kg), SZ-A significantly decreased blood glucose (Figures 2A,B). In addition, blood insulin at 0 min was significantly reduced by SZ-A, but insulin secretion after glucose overload was promoted significantly in the SZ-A group (Figures 2A,B). Subsequently, the hyperglycemic clamp test showed that SZ-A could raise the first-phase insulin secretory response to glucose by 86.17% (AUC_0–15 min) compared with that of the Con group (Supplementary Figures S2B,C).

The morphology of islets was depicted by staining glucagon (green) in α cells and insulin (red) in β cells. The islet architecture of the Con group was damaged, as evidenced by the slightly diminished β cells as well as by the augmented and diffusely distributed α cells. Despite all of these changes, in the SZ-A treated islets, more β cells were interspersed among the islets, and α cells tended to be located toward the edge of the islets (Figure 2C), which were associated with significantly increased β-cell area and decreased α-cell area (Figures 2D,E). At the same time, the ALDH1A3⁺ cells were significantly diminished in the SZ-A group (Figures 2F,G). Moreover, some cells in the central area of the islets displayed vacuolar degeneration and necrosis in mice of the Con group, but the islet morphology in the SZ-A group was normal (Supplementary Figure S2A).

To investigate the possible role of SZ-A in the β-cells insulin secretion, both the mouse β-cell line MIN6 and mouse primary islets were treated with SZ-A. In both MIN6 cells (Figure 3A) and KKA^y mouse islets (Figure 3C), SZ-A at 100 μg/ml promoted insulin secretion regardless of the glucose concentration (2.8 mM or 16.8 mM). In C57BL/6J mouse islets, SZ-A mainly enhanced insulin secretion at 16.8 mM glucose (Figure 3B). In addition, three major effective fractions, namely, DNJ, FAG and DAB, showed different efficacies on insulin secretion in β cells. Among them, DAB and FAG promoted insulin secretion in MIN6 cells, but DNJ did not (Supplementary Figure S3A). In addition, DAB and FAG facilitated insulin secretion in C57BL/6J mouse islets and KKA^y mouse islets, respectively (Supplementary Figures S3B,C). However, SZ-A has a greater ability to stimulate insulin secretion in β cells than any of its constituents.

Glucose facilitates insulin secretion through changes in the free [Ca²⁺]_i and [cAMP]_i, so we examined the actions of SZ-A on [Ca²⁺]_i and [cAMP]_i in MIN6 cells. In MIN6 cells, elevating the glucose concentration from 2.8 to 16.8 mM initiated a marked rise in [Ca²⁺]_i and [cAMP]_i. However, in the presence of 2.8 mM glucose, SZ-A had no effect on [Ca²⁺]_i. At a glucose concentration of 16.8 mM, changes in [Ca²⁺]_i were observed after SZ-A treatment (Figure 3D). At the same time, 100 μg/ml SZ-A significantly decreased AMPK phosphorylation levels (Figures 3E,F) at 2.8 mM glucose. However, in the presence of 2.8 mM or 16.8 mM glucose, SZ-A did not cause an increase in [cAMP]_i (Supplementary Figure S4A). In addition, two of the major effective fractions, that is, FAG and DAB, modulated the glucose-dependent increases in [Ca²⁺]_i (Supplementary Figure S4B) and inhibited AMPK phosphorylation at 2.8 mM glucose (Supplementary Figure S4C).

Islet β-cell dedifferentiation has been suggested to participate to the β-cell failure (Zhang and Liu, 2020), and Aldh1a3 was reported as a marker of β-cell dedifferentiation (Kim-Muller et al., 2016). In KKA^y mice, the Aldh1a3 expression level was significantly increased in islet β cells compared with normal mice (Supplementary Figure S5). The expression level of Aldh1a3 was inhibited significantly in KKA^y mouse islets treated with SZ-A (Figures 4A–C). Moreover, in islet cells, SZ-A increased β-cell identifying genes such as Ins1, Ins2, Nkx2.2 and Pax4 and decreased β-cell excluding genes such as Gcg and MafB (Figure 4D).

In addition to dedifferentiation, apoptosis was another cause to β-cell mass reduction. High glucose and palmitic acid treatment (24 h) inhibited proliferation, and if treatment was prolonged to 72 h, apoptosis were recognized by increased TUNEL-positive cells. SZ-A did not influence cell proliferation in basic culture medium (25 mM glucose) but significantly reversed the growth inhibition of MIN6 cells induced by high glucose and palmitic acid (Figure 4E). Furthermore, SZ-A at concentrations of 50 and 100 μg/ml significantly decreased the number of TUNEL-positive cells after 72 h of high glucose and palmitic acid treatment (Figures 4F,G). Interestingly, among the three major fractions of SZ-A, DNJ but not FAG and DAB showed a similar effect (Supplementary Figure S6).

Furthermore, SZ-A significantly attenuated the elevated Erk1/2 phosphorylation level and caspase 3 activity under high glucose- and palmitic acid-induced stress conditions (Figures 5A–C). Similarly, DNJ, one of the major fractions of SZ-A, decreased both Erk1/2 phosphorylation and caspase 3 activity (Supplementary Figure S7).