PS1 alone and in combination with metformin can reduce diabetes and increase survival and life span in db/db mice

PS1 was rationally developed as a Pdia4 small-molecule inhibitor using a combination of a molecular docking strategy and total chemical synthesis (Figure S1A). Furthermore, PS1 had a half maximal inhibitory concentration (IC50) value of 4 μM for Pdia4. These data prompted us to assess the anti-diabetic effect of PS1 in db/db mice in new-onset diabetes (Figure S1B). First, we assessed the glucose-reduction caused by PS1 in diabetic db/db mice. As anticipated, db/db mice had manifested diabetes by 8 weeks and this disease lasted for their life time (CTR, Fig. 1A). In contrast, STG is a commercial DPP4 inhibitor for diabetes. Sixteen-week treatment with STG, a positive control of monotherapy, and 16-week treatment with STG plus metformin, a positive control of combinational therapy, modestly lowered FBG and PBG in diabetic db/db mice (STG@15 mg/kg versus STG@15 mg/kg + Met@60 mg/kg, Fig. 1A). Of note, PS1 dose-dependently reduced diabetes in diabetic db/db mice as evidenced by FBG (PS1, left, Fig. 1A) and PBG (PS1, right, Fig. 1A). A combination of PS1 and metformin had slightly better glucose-lowering effects than PS1 alone (FBG and PBG, PS1@2.5 + Met@60 mg/kg, Fig. 1A).

Fig. 1

Beneficial effects of PS1 on diabetes development in db/db mice. A Female db/db mice that had new-onset diabetes were grouped and treated with PBS (CTR), STG (15 mg/kg), a combination of STG (15 mg/kg) and metformin (Met, 60 mg/kg), PS1 (0.8 mg/kg, 2.5 mg/kg, and 7.5 mg/kg), and a combination of PS1 (2.5 mg/kg) and metformin (60 mg/kg) from 8 to 56 weeks of age. Their fasting blood glucose (FBG) and postprandial blood glucose (PBG) levels were found using a glucometer at the indicated ages. B HbA1c of the mice A was monitored. C GTT of the mice A was monitored by the aged of 8, 16 and 24 weeks. Mouse number (n) is shown in parentheses. P (*) < 0.05, P (**) < 0.01 and P (***) < 0.001 were considered statistically significant

HbA1c is known as a reliable marker of chronic glycemia. Therefore, we next checked the effects of PS1 on HbA1c in diabetic db/db mice. Each group of db/db mice, aged 8 weeks, had an HbA1c value of 4% (8 weeks, Fig. 1B). This value increased to 10% in control db/db mice aged 16 weeks and beyond (CTR, 16–56 weeks, Fig. 1B). In contrast, db/db mice treated with STG per se and in combination with metformin reduced the HbA1c value to 8.3–7.7% in age-matched db/db mice (STG@15 mg/kg versus STG@15 mg/kg + Met@60 mg/kg, Fig. 1B). Furthermore, PS1 dose-dependently reduced the HbA1c value to 7.9–5.9% in age-matched db/db mice (PS1, Fig. 1B). The db/db mice given a combination of PS1 and metformin had an HbA1c value of 5.3%, which was slightly lower than the HbA1c value in the age-matched mice treated with PS1 alone (PS1@2.5 + Met@60 mg/kg versus PS1, Fig. 1B).

Next, GTT was measured in each group of db/db mice aged 8, 16 and 24 weeks. No difference was observed in the GTT of each group of mice at the age of 8 weeks (left, Fig. 1C). However, STG per se and in combination with metformin significantly improved GTT compared to PBS vehicle in db/db mice aged 16 weeks (STG@15 mg/kg versus STG@15 mg/kg + Met@60 mg/kg, middle, Fig. 1C). PS1 alone and in combination with metformin further improved GTT in 16-week-old db/db mice in a dose-dependent fashion (PS1 at 0.8, 2.5 and 7.5 mg/kg versus PS1@2.5 mg/kg + Met@60 mg/kg, middle, Fig. 1C). A combination of PS1 and metformin improved GTT more than PS1 alone in 16-week-old db/db mice (PS1 at 0.8, 2.5 and 7.5 mg/kg versus PS1@2.5 mg/kg + Met@60 mg/kg, middle, Fig. 1C). By 24 weeks of age, PS1 alone and in combination with metformin for 16 weeks improved GTT in db/db mice more than 8-week treatment with the same drugs (right, Fig. 1C).

As far as diabetic incidence is concerned, 100% of new-onset diabetic db/db mice developed severe diabetes over time (CTR, Fig. 2A). Although STG per se and in combination with metformin lowered the blood glucose level of the db/db mice (Fig. 1A), such treatments failed to reduce diabetic incidence (STG@15 mg/kg versus STG@15 mg/kg + Met@60 mg/kg, Fig. 2A). However, PS1 at 2.5, and 7.5 mg/kg reduced diabetic incidence in db/db mice by 0%, 13–29%, and 43–86%, respectively (Fig. 2A). In sharp contrast, a combination of PS1 and metformin fully reversed diabetes in db/db mice (PS1@2.5 mg/kg + Met@60 mg/kg, Fig. 2A). We also examined survival rate and life span in different mouse groups. Sixteen-week treatment with PS1 significantly improved survival (PS1, Fig. 2B) and life span (PS1, Fig. 2C) in db/db mice according to dose. Furthermore, db/db mice treated with PS1 and metformin had better survival and life span (PS1@2.5 mg/kg + Met@60 mg/kg, Fig. 2B, C). Overall, the data indicated that the Pdia4 inhibitor, PS1, alone and in combination, could treat and reverse diabetes.

Fig. 2

Promotion of diabetes reversal, survival rate and longevity in db/db mice by PS1. Diabetic incidence (A), survival rate (B), and life span (C) of the mice from Fig. 1A from birth to death were measured. Diabetic incidence of the mice was calculated based on their PBG over 126 mg/dL. Mouse number (n) is shown in parentheses. P (*) < 0.05, P (**) < 0.01 and P (***) < 0.001 were considered statistically significant

PS1 increases serum insulin and serum c-peptide and improves HOMA indices in db/db mice

In parallel, we tested how PS1 affected the function of pancreatic islets in db/db mice. No difference in levels of postprandial serum insulin was observed in any of the groups of mice aged 7 weeks (left, Fig. 3A). Levels of postmeal serum insulin in control db/db mice gradually decreased over time (CTR, 7–58 weeks, Fig. 3A). However, mice given STG alone and in combination with metformin had slightly more postmeal serum insulin than control mice (STG@15 mg/kg versus STG@15 mg/kg + Met@60 mg/kg, 18–58 weeks, Fig. 3A). In contrast, PS1 dose-dependently elevated the levels of postmeal serum insulin in db/db mice (PS1 at 0.8, 2.5 and 7.5 mg/kg, 18–58 weeks, Fig. 3A). Since c-peptide is a predictor of β-cell function, next, the effect of PS1 on the levels of c-peptide in db/db mice was examined. There was no difference in the levels of postprandial serum c-peptide in any of the groups of mice aged 7 weeks (7 weeks, Fig. 3B). Levels of serum c-peptide in control db/db mice gradually decreased over time (CTR, 7–58 weeks, Fig. 3B). However, mice given STG per se and in combination with metformin slightly increased the levels of serum c-peptide in db/db mice over time (STG@15 mg/kg versus STG@15 mg/kg + Met@60 mg/kg, 18–58 weeks, Fig. 3B). In contrast, PS1 per se and in combination with metformin dose-dependently augmented the levels of serum c-peptide in db/db mice (PS1 at 0.8, 2.5 and 7.5 mg/kg versus PS1@2.5 mg/kg + Met@60 mg/kg, 18–58 weeks, Fig. 3B). Since HOMA indices are useful markers of β-cell function and insulin resistance, next, we checked the effect of PS1 on the HOMA-β and HOMA-IR indices in diabetic db/db mice. There was no difference in the HOMA-β index in any group of db/db mice at 8 weeks (Fig. 3C). However, the HOMA-β index in db/db mice dramatically declined over time (CTR, Fig. 3C). STG per se and in combination with metformin modestly reduced this decline in db/db mice (STG@15 mg/kg versus STG@15 mg/kg + Met@60 mg/kg, 16–56 weeks, Fig. 3C). In remarkable contrast, PS1 per se and in combination with metformin significantly increased the HOMA-β index in db/db mice according to dose (PS1 at 0.8, 2.5 and 7.5 mg/kg versus PS1@2.5 mg/kg + Met@60 mg/kg, 16—56 weeks, Fig. 3C). In agreement, no difference in the HOMA-IR index was perceived in any group of 8-week-old db/db mice (Fig. 3D). However, the HOMA-IR index in db/db mice gradually increased over time (CTR, Fig. 3D). STG per se and in combination with metformin moderately reduced this increase in db/db mice (STG@15 mg/kg versus STG@15 mg/kg + Met@60 mg/kg, 16–56 weeks, Fig. 3D). PS1 per se and in combination with metformin significantly reduced the HOMA-IR index in db/db mice in a dose-dependent fashion (PS1@0.8, 2.5 and 7.5 mg/kg versus PS1@2.5 + Met@60 mg/kg, 16—56 weeks, Fig. 3D). As a result, the HOMA-IR index in each group of mice was inversely proportional to the HOMA-β index (Fig. 3C, D). Collectively, the data showed that PS1 ameliorated the function of the pancreatic islets.

Fig. 3

Improvement of serum insulin, c-peptide and HOMA indices in db/db mice by PS1. A The insulin concentration of the mice (Fig. 1A) was measured at the age of 7, 18, 30 and 58 weeks. B C-peptide of the mice (Fig. 1A) was determined at the indicated ages. HOMA-β (C) and (D) HOMA-IR of the mice (Fig. 1A) were measured at the age of 8, 16, 24 and 56 weeks. Mouse number (n) is shown in parentheses. P (*) < 0.05, P (**) < 0.01 and P (***) < 0.001 were considered statistically significant

PS1 reduces islet atrophy, islet ROS, and serum ROS in db/db mice

We also looked at the effect of PS1 on the architecture of pancreatic islets in db/db mice. Control db/db mice, aged 18 weeks, had smaller islet size than those treated with STG alone or in combination with metformin (STG@15 mg/kg versus STG@15 mg/kg + Met@60 mg/kg, Ins, Fig. 4A, B). Of note, db/db mice treated with PS1 per se and in combination with metformin further increased the islet size (PS1 at 0.8, 2.5 and 7.5 mg/kg versus PS1@2.5 + Met@60 mg/kg, Ins, Fig. 4A, B). This increase seemed to be dose-dependent. The data suggested that PS1 reduced the islet atrophy in a dose-dependent fashion. Consistently, ROS content in the islets of db/db mice treated with PS1 and a combination of PS1 plus metformin was inversely proportional to their islet size as evidenced by the signals of DHE staining (DHE, Fig. 4A, C). The data on the ROS content of the mouse islets were in good agreement with the data on serum ROS of db/db mice (Fig. 4D).

Fig. 4

Reduction of islet atrophy, islet ROS, and serum ROS in db/db mice by PS1. A The pancreata of the mice (Fig. 1A) at the age of 18 weeks were fixed and stained with anti-insulin antibody (top) and DHE (bottom). Representative images were photographed. Scale bar: 50 µm. Islet area (μm2) (B) and relative fluorescence intensity (RFI) (C) of the pancreatic islets of the mice A were quantified and re-plotted into histograms. Scale bar = 100 μm. The dashed circles show islet regions. D Serum ROS of the mice (Fig. 1A) was measured at the age of 18 weeks. The number of mice (n) is indicated in parentheses. P (*) < 0.05, P (**) < 0.01 and P (***) < 0.001 were considered statistically significant

The overall data demonstrated that PS1 reduced islet atrophy and elevated the ROS level of the islets and sera in db/db mice.

PS1 reduces cell death rather than cell proliferation in the islets of db/db mice

To probe the mechanism by which PS1 protected against β-cell failure, we first studied the action of PS1 on β-cell proliferation and demise using Ki67 staining and TUNEL assays, respectively. There was no statistical difference in the proliferation of cells of the islets among control db/db and the mice treated with PS1 and other drugs based on the IHC staining with the antibody against Ki67, a cell proliferation marker (Fig. 5A). However, control db/db mice had a comparable number of TUNEL-positive islet cells, primarily dead islet cells, as the age-matched mice treated with STG alone and in combination with metformin (STG@15 mg/kg versus STG@15 mg/kg + Met@60 mg/kg, Fig. 5B). In remarkable contrast, treatment with PS1 per se or in combination with metformin significantly diminished the number of TUNEL-positive islet cells in db/db mice (PS1 at 0.8, 2.5 and 7.5 mg/kg versus PS1@2.5 mg/kg + Met@60 mg/kg, Fig. 5B). Furthermore, the diminishment of TUNEL-positive islet cells in the mouse pancreata by PS1 seemed to be dose-dependent. We also explored the effect of PS1 on the expression of Aldh1a3, an indicator of β-cell dedifferentiation. PS1 dose-dependently decreased the expression level of Aldh1a3 in mouse islets (Fig. 5C). To sum up, both types of assays suggested that PS1 was able to reduce cell death but not cell proliferation in pancreatic islets of db/db mice.

Fig. 5

Decrease in cell death and Aldh1a3 expression in the pancreatic islets of db/db mice by PS1. A, B The pancreata of the mice (Fig. 1A) were fixed and stained with anti-Ki67 and TUNEL reagent. Ki67-positive cells per islet area (0.05 mm2) A were visualized in the islets of the mice (left). After quantification, the signals of each group were quantified and re-plotted into histograms (right). TUNEL-positive cells per islet area (0.05 mm2) B were visualized in the islets of the mice (left). After quantification, the signals of each group were quantified and re-plotted into histograms (right). C The pancreata of the mice (Fig. 1A) were fixed and stained with anti-Aldh1a3 (red). Representative images of Aldh1a3 in the islets were acquired using a confocal microscope. After quantification, the mean fluorescence intensity (MFI) of each group was quantified and re-plotted into histograms (right). The dashed circles show islet regions and the black arrowheads point out Ki67+ or TUNEL+ cells. Mouse number (n) is shown in parentheses. P (*) < 0.05, P (**) < 0.01 and P (***) < 0.001 are considered statistically significant

PS1 decreases cell death and increases insulin secretion in β-cells

To probe the action of PS1 in β-cells, we investigated the impact of PS1 on cell demise and function in Min6 cells, a murine β-cell line, in response to high glucose. Min6 cells were pre-incubated with cell medium, and the medium was supplemented with STG (10 μg/mL) and PS1 (0.2, 1, and 5 μg/mL) in the presence of 5 mM or 30 mM glucose. Flow cytometric data showed that control Min6 cells grown in the medium containing 5 mM glucose, had a basal level of cell death, 7.4% of PI-positive cells (CTR, Fig. 6A). Glucose at 30 mM increased cell death to 18.6% (HG, Fig. 6). STG failed to lower cell death (17%) (STG, Fig. 6A). In contrast, PS1 at 0.2 μg/mL, 1 μg/mL, and 5 μg/mL reduced cell death to 12.8%, 8.8%, and 4%, respectively (PS1, Fig. 6A). Remarkably, PS1 at 5 μg/mL even reduced this death in Min6 cells by 3.4% (from 7.4 to 4%). The data demonstrated that PS1 dose-dependently rescued Min6 cells from cell death as a result of high glucose. However, STG failed to rescue Min6 cells from cell death under the same conditions.

Fig. 6

Down-regulation of cell death and up-regulation of insulin secretion in Min6 cells by PS1. A Min6 cells were treated with 5 mM (CTR) or 30 mM glucose with PBS (HG), STG (10 µg/mL) and PS1 (0.2 µg/mL, 1 µg/mL, and 5 µg/mL). The cells were stained with propidium iodide (PI) and analyzed with flow cytometry. The percentage of PI-positive cells from 3 experiments was analyzed and is expressed as mean ± SD. B Min6 cells were stimulated with DMEM medium or the medium containing KCl (30 mM), STG (10 µg/mL) and PS1 (0.2 µg/mL, 1 µg/mL, and 5 µg/mL). The insulin level of their supernatants was quantified and replotted into histograms. The data from 3 experiments are expressed as mean ± SD. C Min6 cells were pre-treated with DMEM medium containing 30 mM. After washing, the cells were treated with high glucose (16.7 mM), STG (10 µg/mL) and PS1 (0.2 µg/mL, 1 µg/mL, and 5 µg/mL). The insulin level of their supernatants was quantified and replotted into histograms. The data from 3 experiments are expressed as mean ± SD. P (*) < 0.05, P (**) < 0.01 and P (***) < 0.001 were considered statistically significant

We also examined the effect of PS1 on insulin release in Min6 cells. First, we tested the insulinotropic effect of PS1 in Min6 cells. As expected, there was a basal level of insulin in the supernatant of Min6 cells (CTR, Fig. 6B). Potassium chloride, a positive control, elevated the level of insulin in the supernatant of Min6 cells (KCL, Fig. 6B). STG did not alter the level of insulin in the supernatant of Min6 cells (STG, Fig. 6B). However, PS1 dose-dependently up-regulated the level of insulin in the supernatant of Min6 cells (PS1, Fig. 6B). In parallel, we also treated Min6 cells with 30 mM glucose with or without PS1. As expected, high glucose compromised the glucose stimulated insulin secretion (GSIS) in Min6 cells (CTR, Fig. 6C), However, STG failed to improve the GSIS in Min6 cells (STG, Fig. 6C). In remarkable contrast, PS1 dose-dependently improved the GSIS in Min6 cells (PS1, Fig. 6C). The data suggest that PS1 protected against β-cell dysfunction and death.

PS1 down-regulates ROS production via the Pdia4/Ndufs3 and Pdia4/p22 axes in β cells

To investigate whether PS1 exerted its action through ROS pathways, we first analyzed the amount of ROS in the mitochondria and cytosol of Min6 cells. Min6 cells had a basal level of mitochondrial ROS (CTR, Fig. 7A). High glucose augmented mitochondrial ROS in Min6 cells (HG, Fig. 7A). STG failed to diminish mitochondrial ROS in Min6 cells (STG, Fig. 7A). In sharp contrast, PS1 dose-dependently reduced mitochondrial ROS in Min6 cells (PS1, Fig. 7A). Similarly, we found that Min6 cells had a basal level of cytosolic ROS (CTR, Fig. 7B). High glucose increased cytosolic ROS in Min6 cells (HG, Fig. 7B). STG failed to diminish cytosolic ROS in Min6 cells (STG, Fig. 7B). In sharp contrast, PS1 dose-dependently reduced cytosolic ROS in Min6 cells (PS1, Fig. 7B). The data showed that PS1 reduced ROS in β-cells.

Fig. 7

Inhibition of the Pdia4/ETC C1 and Pdia4/Nox pathways in Min6 cells by PS1. (A-B) Min6 cells were treated with 5 mM glucose (CTR) or 30 mM glucose with PBS (HG), STG (15 μg/mL), and PS1 (0.2 µg/mL, 1 µg/mL, and 5 µg/mL). The cells were then incubated with MitoGreen plus MitoSOX (A) and Hoechst 33,342 (Ho) plus CellROX (B) in the presence of glucose at 5 mM (CTR) and 30 mM (HG). The cells were visualized (left) and quantified (right). Scale bar: 5 µm. C The ETC C1 (left) activity of mitochondria isolated from control and PS1-treated Min6 cells A was measured using MitoCheck assays. D The Nox activity of the membrane fraction of control and PS1-treated Min6 cells A was measured using luminometric assays. Min6 cells that expressed Flag-Pdia4 plus Myc/Flag-tagged Ndufs3 (E) or p22 (F) were treated with PS1 (5 μg/mL) for 30 min. The cells were lysed and precipitated using anti-Pdia4 antibodies and protein G beads. Their total lysates (TL) and immunoprecipitates (IP) were analyzed with immunoblots using anti-Flag antibody. P (*) < 0.05, P (**) < 0.01 and P (***) < 0.001 were considered statistically significant. G Schema outlining the regulation of the Pdia4/Ndufs3 and Pdia4/p22 pathways. The intermolecular association of Pdia4 with Ndufs3 and p22phox increases the activity of ETC C1 and Nox. As a result, the excess ROS induces β-cell pathology and diabetes (top). On the other hand, Pdia4 inhibition by PS1 can disrupt the interplay of Pdia4 and Ndufs3 or p22, resulting in declined ROS production, β-cell failure and diabetes (bottom)

To pinpoint the pathways that PS1 targeted, we checked the effect of PS1 on Pdia4 activity and the intermolecular interaction between Pdia4 and Ndufs3 or p22. The data showed that PS1 could inhibit the enzymatic activity of Pdia4 in Min6 cells (Figure S1) and, in turn, decreased the activity of ETC C1, Fig. 7C) and Nox (Fig. 7D). Further, immunoblotting data showed that PS1 intervened with the intermolecular interaction between Pdia4 and Ndufs3 (Fig. 7E) and that between Pdia4 and p22 (Fig. 7F). Taken together, the data suggested that PS1 down-regulated production of mitochondrial and cytosolic ROS via the Pdia4/Ndufs3 and Pdia4/p22 pathways in β-cells.

A schematic model describing the pharmacological function and mechanism of PS1 is shown in Fig. 7G. Under hyperglycemia, nutrient overload up-modulated Pdia4 expression. This up-modulation increased the activity of Ndufs3 and p22 through intermolecular interplay and thus escalated the generation of ROS in β-cells. Eventually, the excessive ROS caused β-cell dysfunction and death and, subsequently, diabetes (Fig. 7G). On the other hand, PS1 reversed diabetes through reduced ROS production and β-cell pathology in diabetic animals (Fig. 7G). The data also unveiled the pharmacological action and mechanism of PS1 in β-cell pathology and diabetes.