The use of human tissue was compliant with the Declaration of Helsinki and the study protocol was approved by the Ethics Committee of the West China Hospital of Sichuan University. A total of sixteen (eight males and eight females) patients with T2D and eighteen (nine males and nine females) age-matched healthy controls, 12 male patients with type 1 diabetes (T1D) and eight male age-matched normal voluteers were recruited in this study. All participants signed an informed consent form for the use of serum for research purposes. Blood samples from each participant were collected after an overnight fast. The plasma was separated by centrifugation at room temperature.

All animal studies were approved by the Laboratory Animal Care and Use Committee of West China Hospital of Sichuan University. Seven-week-old male C57BL/6J mice were purchased from Huafukang Bioscience Technology Co. (Beijing, China). Mice were housed in a standard laboratory animal facility under a 12-h light-dark cycle with free access to standard chow food and water. After 1 week of acclimatization, mice were randomly divided into four groups (n = 10 in each group): the control group with an injection of saline vehicle DMSO (5% v/v), CS-treated group with an administration of CS (25 mg/kg body weight/dose, two doses daily) (Bide Pharmatech Ltd., Shanghai, China). CS dosage was selected from previous studies on its beneficial effect on glucose and lipid metabolism (Shi et al., 2014), and this regiment of CS was not toxic to the mice (Bi et al., 2018). STZ-treated group with a 7-day injection of saline vehicle followed by a 4-day administration of STZ (40 mg/kg of body weight, once daily) (Meilun Bio, Dalian, China), and CS + STZ-treated group with a 7-day CS injection followed by a STZ administration for four continuous days. In the control and STZ-treated groups, saline vehicle was intraperitoneally injected daily for 42 days, while the CS- and CS + STZ-treated groups were given the same volume of CS twice daily instead of saline (Supplementary Figure S1A). Mice were weighed weekly, and blood glucose levels were measured daily during the experimental treatment period. In addition, obesity was induced in ten male mice by feeding them a high-fat diet (HFD) (60% fat, TROPHIC Animal Feed High-Tech Co., Ltd., Jiangsu, China) for 8 weeks.

Anesthetized mice were laid in a supine position in an imaging cell, and three-dimensional (3D) X-ray images were acquired using a SPECT-μCT system (General Electric, Waukesha, United States). Adipose tissue was recognized by the reduced attenuation of X-rays in comparison with other structures. Total adipose tissue volume was determined between the proximal end of lumbar vertebra 1 (L1) and the distal end of caudal vertebra 4 (C4).

The plasma levels of insulin, glucagon, triglyceride and cholesterol were measured as previously described (Zhang Yanjie et al., 2018). The intraperitoneal glucose tolerance test (IPGTT), glucose-stimulated insulin secretion (GSIS) and intraperitoneal insulin tolerance test (IPITT) were conducted in mice with overnight fasting, as previously described (Liu et al., 2017a).

CS levels in mouse and human plasma were determined using liquid chromatography tandem-mass spectrometry (LC-MS/MS) as previously described (Bandaru and Haughey 2014). Briefly, 30 μL of plasma was mixed with 120 μL of honokiol internal control solution (honokiol dissolved in methanol, 200 ng/ml). After centrifugation at 10,000 g for 15 min, 0.2 μL of the supernatant was injected into a Shimadzu LC-20AD HPLC. Chromatography was conducted using an Agilent Extend 5 μM C18 column (Agilent Technologies, Inc., Santa Clara CA, United States) with solvent A (pure methanol) and solvent B (0.01% ammonia) under the gradient conditions of 60% A at 0–1 min and gradient to 95% A at 1–2.2 min at a flow rate of 0.5 ml/min. Quantification of CS was performed on an AB SCIEX Triple Quad 5500 LC/MS/MS system (AB SCIEX, United States) by multiple reaction monitoring (MRM) using Analyst 1.6.2 software (Applied Biosystems). CS stock solution with a concentration range of 0.5 ∼3.0 μg/ml was used as an internal standard, and calibration curves were plotted using the peak area ratios of CS to honokiol (internal control).

Mice were given BrdU intraperitoneally at a dose of 50 mg/kg body weight once daily for 3 days. Two hours after the final injection, mice were euthanized and pancreata processed for histology.

Pancreatic islets were isolated from 10-week-old C57BL/6J mice using the intraductal collagenase digestion technique (Li et al., 2009). Isolated islets were plated on 12-well plates overnight and then incubated with DMSO or 10 μM CS for 12 h. Thereafter, islets were incubated in Krebs-Ringer bicarbonate (KRB) buffer with low glucose (2.8 mM) or high glucose (16.7 mM) for 1 h at 37°C. Insulin in islets was extracted by overnight incubation with ethanol/HCl buffer at 4°C. The insulin secreted into the buffer was measured using an ELISA kit (Crystal Chem, IL).

Dissociated islets (10 islets/well) or INS-1 β-cells (1×10⁴ cells/well) seeded in 96-well plates were incubated with DMSO or 10 µM CS for 12 h followed by a 30-min incubation of 100 μg/ml 2-NBDG (Cayman Chemical, MI, United States) in KRB buffer. The fluorescence was measured using a fluorescence plate reader with an excitation wavelength of 475 nm and an emission wavelength of 550 nm.

As previously described (Liu et al., 2017a; Chen et al., 2019), hematoxylin and eosin (HE) staining and immunohistochemistry (IHC) were conducted on paraffin sections, and immunofluorescence staining was performed on frozen sections of mouse pancreas with antibodies presented in Supplementary Table S2.

As previously described (Liu et al., 2021), isolated mouse islets were plated on 12-well plates overnight and then incubated with or without 10 µM CS for 12 h. Afterwards, islets were incubated in 5 µM Fluo-3 AM working solution (Fluo-3 AM dissolved in anhydrous DMSO was diluted with KRB buffer) at 37°C for 60 min. Subsequently, islets were perfused with KRB buffer-based solutions containing glucose (16.7 mM) or KCl (30 mM) at 37°C at a flow rate of 2 ml/min. Fluorescence imaging was conducted in an N-STORM and A1 confocal scanning microscope (Nikon, Japan).

INS-1 β-cells plated on black 96-well plates were treated with or without 10 µM CS for 12 h following a 20-min incubation with glucose at a concentration of 2.8 mM or 16.7 mM. ATP levels in cell lysates were measured using a cellular ATP determination kit (Beyotime Biotech, Shanghai, China) in accordance with the manufacturer’s protocol.

Mice pancreas sections were postfixed in 1% OsO₄, washed in 0.1 M phosphate buffer, and dehydrated via a graded ethanol series. Samples were embedded in propylene oxide/Polybed 812 epoxyresin overnight. Ultrathin sections (70 nm) stained with 2% uranyl acetate and 1% lead citrate were examined under a Philips CM120 scanning TEM.

INS-1 β-cells were seeded in 96-well plates (1×10⁴ cells/well) overnight and then incubated with or without varying concentrations of CS for 6 h followed by a coincubation of CS and STZ (2.5 mM), or H₂O₂ (200 µM) or palmitic acid (PA, 400 µM) for 12 h or high concentration glucose (HG, 33.3 mM) for 72 h. Cell viability was assessed by a Cell Counting Kit-8 (CCK-8) (4A Biotech, Beijing) assay according to the manufacturer’s instructions.

Primary islet apoptosis was analyzed by means of terminal deoxynucleotidyl mediated dUTP nick end labeling (TUNEL) following the manufacturer’s instructions (Yeasen Biotech, Shanghai, China). The samples were stained with DAPI to visualize total cells. The ratios of TUNEL-positive nuclei to the total islet nuclei were calculated from the digitized images. INS-1 cell apoptosis was assessed by flow cytometric analysis (Thermo Scientific, United States) after cells were stained with Annexin V (AV)-fluorescein isothiocyanate (FITC)/propidium iodide (PI) (4A Biotech, Beijing, China). Data were analyzed using Flow Jo v.10.0.7 software (Tree Star, United States), and the number of apoptotic cells was expressed as the percentage of the total number of cells.

INS-1 β-cells plated on 6-well plates (1×10⁵ cells/cm²) were incubated with DMSO or 10 µM CS for 12 h and then stained with EdU and Hoechst 33,342 dye sequentially. Images were captured by a laser scanning confocal microscope.

Intracellular ROS was estimated using a 2′,7′-dichlorofluorescin diacetate (DCFDA) Cellular ROS Assay Kit (Beyotime, Shanghai, China) following the manufacturer’s protocol. Briefly, INS-1 β-cells (1×10⁴ cells/well) or mouse islets (10 islets/well) were seeded in black 96-well plates. After a 6-h preincubation of CS (10 µM) or DMSO, cells were coincubated with CS (10 µM) and STZ (2.5 mM) or H₂O₂ (200 µM) or palmitic acid (PA, 400 µM) for an additional 12 h. Afterwards, the cells were incubated for 30 min at 37°C with 20 µM DCFHDA. The fluorescence was examined using an AxioVert fluorescence microscope (Carl Zeiss) with a 448-nm excitation filter. The fluorescence intensity was quantified by ImageJ software and normalized to the number of cells in the field. Cells treated only with DMSO or STZ were used as a negative or positive control, respectively.

MMP was determined using a 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzi-midazolylcarbocyanine iodide (JC-1) fluorescent probe (Beyotime Biotech, Shanghai, China). After the indicated treatments as described above, mouse islets or INS-1 β-cells were incubated with JC-1 staining solution for 30 min at 37°C and then analyzed using an Axiovert fluorescence microscope (Carl Zeiss) with 488-nm and 561-nm emission filters.

Total RNA was extracted from mouse islets or INS-1 cells with TRIzol reagent (Invitrogen, Carlsbad, CA) and reversely transcribed into cDNA using a reverse-transcription kit (Invitrogen). qPCR analysis with the indicated primers (Supplementary Table S1) was carried out as described elsewhere (Liu et al., 2017b).

Protein extraction from mouse islets or INS-1 cells and Western blotting with the indicated primary antibodies (Supplementary Table S2) were performed as previously described (Kuang et al., 2014).

The data were presented as the mean ± standard deviation (SD) from at least three independent experiments. The data were analyzed using SPSS 22.0 software (SPSS, United States). Statistical significance with p < 0.05 was determined by using a two-tailed Student’s t test (unpaired) for groups of two and one-way ANOVA (followed by Newman–Keuls multiple comparison test) for groups of three or more.

Since CS is a major sterol sulfate in plasma, we wanted to determine the potential changes in plasma CS content under various pathologic conditions linked to diabetes. Clinically, multiple low doses of STZ induce a mild impairment of insulin secretion that is more similar to the later stages of T2DM (Srinivasan et al., 2005). Compared to the control mice, STZ-treated mice showed markedly increased fasting blood glucose (Supplementary Figure S1B) and diminished serum insulin concentration (Supplementary Figure S1C), paralleled by significantly lower body weight (Supplementary Figure S1D). Additionally, other characteristic diabetic symptoms, such as polyphagia and polydipsia, were also observed in these mice (Supplementary Figures S1E,F). With respect to plasma CS level, this parameter was significantly higher in the STZ group than in the control group (2.52 ± 0.13 μg/ml vs. 1.98 ± 0.10 μg/ml, p < 0.01) (Figure 1A). The HFD-induced obese mouse is generally accepted as a T2DM model (Heydemann 2016) displaying hyperglycemia and a significantly higher body weight (Supplementary Figures S1G,H), and the plasma CS level was detected to be remarkably higher than that in the lean mice (3.15 ± 0.37 μg/ml vs. 1.19 ± 0.10 μg/ml, p < 0.001) (Figure 1B).

Subsequently, we wanted to know whether the level of CS is changed in diabetic patients. As expected, fasting blood glucose level was significantly higher in both type 1 and 2 diabetic patients than in the corresponding sex- and age-matched normoglycemic controls (Supplementary Figures S1I,J). Quantitative analysis of plasma CS revealed a significantly higher CS level in men than in women (1.44 ± 0.07 μg/ml vs. 0.89 ± 0.06 μg/ml, p < 0.01), and male T2D patients exhibited a markedly elevated CS level than control subjects (1.99 ± 0.17 μg/ml vs. 1.44 ± 0.07 μg/ml, p < 0.01) (Figure 1C). Although there was no significant difference in women (1.38 ± 0.31 μg/ml vs. 0.89 ± 0.06 μg/ml, p > 0.05), the plasma CS level did show an increasing tendency in diabetic patients (Figure 1C). Furthermore, the plasma CS level was found to be markedly increased in T1D patients compared to the controls (2.29 ± 0.16 μg/ml vs. 1.34 ± 0.07 μg/ml, p < 0.001) (Figure 1D). Collectively, these data suggest a correlation between increased plasma CS level and hyperglycemia.

To determine whether CS mediates glucose homeostasis, we examined its effects in mice treated with multiple low-dose STZ-induced diabetes. From the second week onwards, the body weight of mice in the STZ and CS + STZ groups was significantly lower than that of the control group (Figure 2A). However, the mice treated with CS had a significantly increased body weight compared to that of the mice in the STZ group (Figure 2A). Further evidence demonstrated that CS can relieve STZ-induced adipose tissue loss in mice at the end of the experiment (Figures 2B,C, Supplementary Figure S2). The blood glucose levels of mice in the CS + STZ group and CS group were not different during the first 30 days, but they were significantly higher than those of control mice. On the 31st day, the mice in the CS + STZ group exhibited significantly lower fasting blood glucose levels than those in the STZ group (Figure 2D). The GTT results showed that control mice were glucose tolerant, whereas STZ mice had an impaired tolerance to glucose (Figure 2E), the area under the blood glucose curve (AUC) was increased by 54.5 ± 1.8% in the STZ group compared to that in the control group (Figure 2E). CS-treated mice exhibited a remarkably improved glucose tolerance, and the AUC in the GTT was reduced by approximately 29.2 ± 2.3% in the STZ + CS group compared with that in the STZ group (Figure 2E). However, the effectiveness of insulin in lowering blood glucose, as shown in the ITT, was comparable between mice treated with STZ + CS and STZ alone (Figure 2F). Together, these results indicate that CS could alleviate the STZ-induced impairment of glucose homeostasis but had no effect on systemic insulin sensitivity.

CS reduced STZ-induced diabetic hyperglycemia without an alternation of the insulin sensitivity, indicating preservation of islet β-cell mass and function in these mice. We next examined the morphology of the pancreatic islets. STZ-treated mice exhibited a markedly decreased insulin-positive area and a severe reduction in the β-cell proportion in the islets, with an expansion of α-cells infiltrating into the center of the islets (Figures 3A–D). Daily CS injections initiated 7 days before STZ treatment partially prevented β-cell loss, resulting in a preservation of the β-cell mass and a reduction of the α-cell infiltration by 15.9 ± 2.2% (Figures 3C,D). However, the α-cell arrangement did not recover to the normal peripheral positioning pattern (Figure 3C).

The protection of CS against STZ-induced β-cell loss could arise from a change in β-cell viability or proliferation. To determine whether CS influences β-cell survival under stressful conditions, cell viability was investigated in STZ-exposed INS-1 β-cells. STZ at a concentration of 2.5 mM markedly decreased INS-1 cell viability by 51.4 ± 6.5% (p < 0.001) (Supplementary Figure S3A). However, a preincubation with CS alleviated STZ-induced cell death in a dose-dependent manner (Figure 3E), and 10 μM CS significantly ameliorated STZ-induced cell death by 22 ± 5.9% (p < 0.05). In addition, CS exposure also protected β-cells against oxidation (H₂O₂)_- or lipitoxity (palmitic acid, PA)- or glucotoxity (high concentration glucose, HG)-induced cell death at concentrations of 2.5 μM or 5 μM, respectively (Supplementary Figures S3B–F). These results thus unequivocally demonstrated that CS improves β-cell viability under stress.

Subsequently, we determined whether CS could protect β-cells against apoptosis in primary islets and INS-1 β-cells. STZ significantly induced apoptosis tending to a 67.3 ± 5.4% increase in primary islets (Figures 3F,G). This proapoptotic effect was significantly suppressed by CS, and 10 μM CS inhibited apoptosis by 32.3 ± 7.2% (p < 0.01) (Figures 3F,G). Furthermore, CS exposure at a concentration of 10 μM also significantly attenuated STZ- or H₂O₂- or PA- or HG-induced apoptosis by 8.5 ± 1.4% (p < 0.01), 11.8 ± 1.2% (p < 0.01), 7.7 ± 1.4% and 9.6 ± 1.3% (p < 0.01) of INS-1 β-cells, respectively (Figure 3H, Supplementary Figures S3G–I). Moreover, we determined the impact of CS on pancreatic β-cell proliferation as defined by BrdU incorporation into insulin positive cells. As shown in Figures 3I,J, STZ-treated mice showed a markedly reduced β-cell proliferation (p < 0.001), daily CS injections initiated 7 days before STZ treatment led to a significant increase of β-cell proliferation (p < 0.05). We further performed an EdU incorporation assay in the INS-1 cell line. We found that 10 μM CS significantly increased the number of EdU⁺ cells (Figure 3K). Simultaneously, the protein levels of PCNA and cyclin D1 implicated in β-cell proliferation were significantly upregulated in these cells (Figure 3L). Collectively, these observations suggest that CS promotes both β-cell replication and survival.

Since the AKT signaling pathway is critical in the regulation of human and rodent pancreatic β-cell survival and islet mass (Yuan et al., 2018; Tuttle et al., 2001), we tested the impact of CS on the activation of the AKT pathway. Western blotting showed that CS markedly promoted AKT activation in INS-1 β-cells (Figures 4A,B). cAMP-response element–binding protein (CREB), a crucial transcription factor for β-cell gene expression and function (Dalle et al., 2011; Blanchet et al., 2015), is one of the crucial downstream effectors of AKT (Manning and Toker 2017). We next determined the levels of CREB protein and its serine 133 phosphorylation, which is required for CREB-mediated transcription. We found that CREB was remarkably phosphorylated in CS-treated cells (Figures 4A,B). In particular, this was simultaneously associated with an increased abundance of proteins and transcripts of insulin (Ins1 and Ins2), along with β-cell markers of Pdx1 and Mafa (Figures 4A–C). However, our data also showed that pharmacological blockade of Akt activation did not attenuate CS-induced CREB phosphorylation (Figure 4D, Supplementary Figure S4A). Consistent with these observations, CS protected INS-1 cells against H₂O₂- or PA-induced cell death even when the cells were exposure to the AKT blockade (Figure 4E, Supplementary Figure S4B). These results suggested that CS stimulates β-cell signaling through a mechanism involving the independent activation of the AKT and CREB pathways.

We next wished to delineate whether the protective effect of CS on STZ-induced pancreatic β-cell injury is concomitant with β-cell functional restoration. To this end, we first performed a glucose-stimulated insulin secretion (GSIS) assay in vivo. As shown, CS partially but significantly restored the STZ-induced declination of insulin secretion in response to glucose stimulation (Figure 5A), and the area under the insulin secretion curve (AUC) was increased by 65.2 ± 4.6% in CS-exposed mice compared to that in STZ-treated mice (Figure 5B). We next examined insulin secretion from isolated islets. Despite a comparable basal insulin release between CS-exposed islets and DMSO-treated control islets, the insulin secretion was significantly elevated in islets exposed to CS in comparison with the control islets after being challenged with high-concentration glucose (16.7 mM) (Figure 5C). Together, these results suggest that CS preserves the ability of GSIS in STZ-injured islet β-cells.

Since glucose stimulates insulin secretion by inducing depolarization and Ca²⁺ influx (Yang et al., 2020), we measured intracellular Ca²⁺ levels in perfused islets exposed to 16.7 mM glucose. Glucose-evoked Ca²⁺ influx was significantly increased in islets with CS exposure compared with control islets (Figures 5D,E). β-Cell Ca²⁺ entry is induced by the closure of ATP-sensitive potassium channels (K_ATP channels) with consequent depolarization; therefore, we examined the potential effect of CS exposure on the activity of K_ATP channels in β-cells. Ca²⁺ influx induced by 30 mM KCl in the presence of basal 2.8 mM glucose was equivalent between CS-exposed and control islets (Figures 5F,G). Consistent with these observations, insulin secretion was comparable between CS-treated and control islets challenged with KCl (Figure 5H). These findings indicated that the enhanced GSIS in CS-exposed islets was independent of K_ATP channels, and signaling events upstream of membrane depolarization appear to be responsible for CS-modulated insulin secretion.

In an attempt to identify the upstream event by which CS regulates GSIS, we evaluated whether CS can increase glucose uptake into β-cells, which was monitored by the uptake of 2-NBDG. As shown, glucose uptake was significantly elevated in CS-exposed islets compared to DMSO-treated control islets (Figure 5I). This was further confirmed in INS-1 β-cells exposed to CS (Figure 5J). These findings strongly suggested that CS-enhanced Ca²⁺ entry and GSIS result from increased glucose uptake. Given that blood glucose is taken up by β-cells mainly through GLUT2 (Thorens 2015), we postulated that CS might increase GLUT2. Indeed, both the mRNA expression (Figure 5K) and protein levels (Figure 5L) of GLUT2 were markedly upregulated in cells exposed to CS. Thus, we concluded that CS-induced insulin secretion was at least partially mediated by increases in GLUT2 levels and GLUT2-mediated glucose uptake.

Increased oxidative metabolism of glucose in the mitochondria and the resultant elevated cellular ATP synthesis are key steps in the regulation of glucose-induced insulin release in β-cells (Las et al., 2020; Liu et al., 2021). Although the intracellular ATP content was significantly increased in both CS- and DMSO- treated islets under high glucose (16.7 mM) conditions compared to low glucose (2.8 mM) conditions, CS exposure markedly increased ATP content relative to DMSO treatment (Figure 6A). Given that mitochondrial ATP synthesis in pancreatic β-cells is driven by a change in mitochondrial membrane potential (MMP) in response to glucose stimulation, we then detected the effect of CS on the MMP of β-cells by using JC-1. The degree of depolarization of mitochondria can be measured by the ratio of red/green fluorescence intensity. As shown, CS markedly attenuated the STZ-induced reduction in MMP in both isolated mouse islets (Figures 6B,C) and INS-1 β-cells (Figures 6D,E), indicating that CS can preserve MMP under stressful conditions. This was further confirmed in islets or INS-1 β-cells challenged with either H₂O₂ or PA (Supplementary Figure S5). Additionally, the expression of mRNAs encoding mitochondrial respiratory complex components, such as As9, CytB, Cox2 and atpase6, showed an increased trendency in CS-exposed islets relative to control islets (Figure 6F), suggesting an improved mitochondrial function in β-cells after CS exposure. Further ultrastructural analysis by TEM revealed that STZ-exposed β cells exhibited serious mitochondrial structural abnormalities characterized by marked intracristal swelling and reduced electron density, which were significantly alleviated by CS exposure (Figure 6G). Overall, these data indicate that CS can maitain mitochondrial integrity in islet β-cells.

To further gain insight into the possible mechanisms underlying the effect of CS on mitochondrial homeostasis, we examined its effect on mitochondrial fission and fusion, which are important processes for functional mitochondria after cell damage (Dorn and Kitsis 2015). In INS-1 β-cells, exposure to STZ or PA stimulated a partial mitochondrial fission process as evidenced by the alternation of mitochondrial shape from wirelike to shorter and granular, and CS could not prevent this process (Supplementary Figure S6). As mitochondrial ROS generation is a hallmark of STZ- or PA-induced islet toxicity, we also determined whether CS impacts ROS release. In isolated mouse islets, both STZ and H₂O₂ markedly increased ROS release (p < 0.001), while CS showed a significantly effective anti-ROS effect (Figures 6H,I, Supplementary Figure S7A). This was further confirmed in INS-1 β-cells. When INS-1 cells were exposed to STZ or H₂O₂, mitochondrial ROS increased and was significantly suppressed by CS (Figures 6J,K, Supplementary Figure S7B). Furthermore, the expression of the antioxidant protein SOD2 was significantly increased in CS-treated INS-1 cells (Figure 6L).