Streptozotocin (STZ), S1P and normal rabbit IgG (cat. no. NI01) were purchased from Sigma-Aldrich; Merck KGaA. Mouse anti-Ki-67 monoclonal antibody (cat. no. BD550609) was purchased from BD Biosciences; Becton, Dickinson and Company. Rabbit anti-insulin (cat. no. sc-9168) and anti-S1P3 (cat. no. sc-30024) polyclonal antibodies and normal mouse IgG (cat. no. sc-2025) were purchased from Santa Cruz Biotechnology, Inc. The TUNEL kit (cat. no. 11684817910) was purchased from Roche Diagnostics. Rabbit anti-S1P1 polyclonal antibody (cat. no. ab11424) was purchased from Abcam. Rabbit anti-S1P2 polyclonal antibody (cat. no. BS2594) was purchased from Bioworld Technology, Inc. The biotin-streptavidin horseradish peroxidase detection system include the secondary antibody of 1% biotin-labelled goat anti-mouse/rabbit IgG (cat. no. SP-9000) and diaminobenzidine (DAB) kit (cat. no. ZLI-9017) were purchased from Beijing ZSGB-BIO Technologies, Inc.

A total of 30 male C57BL/6J mice (age, 4 weeks; weight, 16–20 g) were obtained from the Experimental Animal Centre of Xi'an Jiaotong University, Xi'an, China [license no. SCXK (Shan) 2012-003] and were randomly assigned to 2 groups: Normal control (NC) on a normal diet (n=10), and low-dose-STZ-treated diabetic group on an HFD, under standard animal housing conditions (n=20). Food intake and body weight were monitored weekly. Following feeding on an HFD for 4 weeks, a single low dose of STZ (120 mg/kg in 0.1 M citrate buffer; pH 4.5) was administered via intraperitoneal injection after a 14-h fasting to induce diabetes, while the NC group was treated with the same volume of citrate buffer solution. A total of two weeks after the STZ treatment, blood glucose levels were measured to confirm diabetes. Values of >11.1 mmol/l were considered to indicate diabetes in the animals. A total of two mice did not fulfil the conditions to be considered diabetic and one mouse died during the aforementioned process. The remaining 17 diabetic mice were randomly divided into two groups using a random number table: Vehicle diabetic control (DC) group (n=8) and S1P-treated group (S1P group; n=9). S1P storage solution (1 mg S1P dissolved in 1 ml methanol) was preserved in the dark at −20°C and diluted in PBS buffer to make the working concentration. The S1P group animals were administered 20 µg/kg working concentration S1P, whereas the vehicle diabetic control group mice received an equal volume of methanol in PBS by intraperitoneal injection daily for 3 weeks. The DC and S1P groups continued to be fed with an HFD, while the non-diabetic NC group was fed with the normal diet. Fasting blood glucose, body weight, food intake and drinking volume were measured weekly. All mice were housed under conditions of a 12-h light/dark cycle at a temperature of 25°C and a humidity of 55%. All mice were allowed free access to food and water throughout the entire experiment. All animal procedures were approved by the Institutional Animal Care and Use Committee of the Xi'an Jiaotong University of Health Sciences, Xi'an, China.

Glucose tolerance was evaluated using a conventional IPGTT and conducted 21 days after S1P administration. Mice were weighed after fasting for 8 h, before being administered an intraperitoneal glucose injection (2 g/kg). Blood samples (~0.6 µl) were collected from the tail before and at 30, 60 and 120 min after glucose administration. The blood glucose level was determined with an Accu-Check Active digital glucose meter (Roche Diagnostics). The areas under the curves (AUCs) of the IPGTT results were calculated for the blood glucose levels at 0 (BG₀), 30 (BG₃₀), 60 (BG₆₀) and 120 (BG₁₂₀) min, using the following equation: AUC=1/4 (BG₀) + 1/2 (BG₃₀) + 3/4 (BG₆₀) + 1/2 (BG₁₂₀). Fasting insulin was determined by radioimmunoassay. The HOMA-IR index was calculated using Matthew's equation: Fasting insulin level (µIU/ml) × fasting glucose level (mmol/l)/22.5 (29). A HOMA-IR index value of >2.5 was considered to indicate the insulin resistance (29).

Each mouse had one pancreas and in total the mice in 3 groups had a total of 27 pancreases. The tail of the pancreas was taken to make paraffin slides. Pancreatic tissues were carefully excised, fixed in 10% neutral buffered formalin for 24 h at room temperature, dehydrated, embedded in paraffin and cut into 5-µm sections. Overall, ≥7 consecutive sections were cut from the largest cross-sectional area in the middle of each pancreas tissue specimen and were used for hematoxylin and eosin (HE) staining, immunohistochemical staining (insulin, Ki-67, S1PR1, S1PR2 and S1PR3) and TUNEL staining. Each pancreas had one slide for each protein immunohistochemical staining by strictly following the same time and conditions in the process of immunohistochemical staining.

For histological examination, sections were stained with hematoxylin for 5 min and with eosin for 2 min at room temperature. For immunohistochemical staining, the samples were routinely dewaxed, hydrated and incubated with 3% H₂O₂ at room temperature for 10 min. Then, a microwave was used to retrieval the antigen by heating to 95–100°C for 10 min in 0.01 mol/l sodium citrate buffer (pH, 6.0). The slices were then blocked with 0.1% goat serum for 15 min at room temperature. The sections were incubated with anti-insulin (1:1,400), anti-Ki-67 (1:200), anti-S1PR1 (1:400), anti-S1PR2 (1:150) and anti-S1PR3 (1:40) primary antibodies at 4°C overnight, followed by incubation with secondary antibody (1% biotin-labelled goat anti-mouse/rabbit IgG) at 37°C for 15 min. This was followed by incubation with horseradish peroxidase-labelled streptomyces ovalbumin at 37°C for 15 min and staining with DAB at room temperature. Slides were counterstained with hematoxylin for 1 min at room temperature to identify the cell nuclei. As a negative control (Fig. S1A), the insulin, S1PR1, S1PR2 and S1PR3 primary antibodies were replaced with rabbit non-specific IgG, whereas the Ki-67 primary antibody was replaced with mouse non-specific IgG (Ki-67) at the same dilution at 4°C overnight. Myocardial (S1PR1-positive), kidney (S1PR2- and S1PR3-positive) and tumour (Ki-67-positive) mouse tissue was used for positive controls (Fig. S1B). The staining was observed using a DP71 fluorescence microscope (magnification, ×400; Olympus Corporation). The images were analysed using Adobe Photoshop CS4 extended software (Adobe, Inc.). Image pro Plus 6.0 software (Media Cybernetics, Inc.) was used for semi-quantitative analysis of insulin, Ki-67 and S1PR expression in pancreatic tissue. A total of three to five unique fields of view were randomly selected in each slice (magnification, ×400) and averages were calculated for analysis. The integrated optical density (IOD) of positively stained cells and the corresponding islet area in each islet were measured and calculated automatically, and the average values were calculated for analysis. The average optical density (IOD/area) was calculated for the relative amount of protein. After insulin staining, the number of islets in each slide was counted and the area of the maximum islet on each slide was measured in order to assess islet size. Following Ki-67 staining, the proliferating nuclei were stained dark brown. The cells whose nuclei were dyed dark brown were considered to be proliferating islet cells and the proliferation rate in each islet (%) was calculated as the number of proliferating cells/the total number of cells ×100%. Overall, three to five islets were counted for each section and average values were taken.

The formalin-fixed and paraffin-embedded pancreas tissue sections were stained with HE and immunohistochemical staining of insulin. Following the HE and insulin staining, images of at least 5 fields on each slide were captured using a fluorescence microscope (magnification, ×400; Olympus Corporation). Apoptosis was demonstrated using a TUNEL assay kit (Roche Diagnostics), according to the manufacturer's protocols. After TUNEL staining, apoptotic nuclei were stained dark brown and cells with dark brown nuclei were considered to be apoptotic islet cells. The apoptosis rate in each islet (%) was calculated as the number of apoptotic cells/the total number of cells ×100% in each islet. Overall, three to five islets were counted for each section and average values were taken.

Statistical analysis was performed using SPSS version 21.0 software (IBM Corp.). Continuous variables are expressed as the mean ± standard deviation or median (interquartile) from at least three independent experiments. Intervention and control groups were compared with one-way analysis of variance followed by Least Significant Difference post hoc test for parametric analysis and Kruskal-Wallis H test for nonparametric analysis. P<0.05 was considered to indicate statistically significant differences.

The islets of the NC mice were round or oval, with clear boundaries and complete film. In addition, the islets were evenly distributed between the pancreatic acinus; the islet cells were round or oval, of uniform size, strong staining in the cytoplasm and a visible centred nucleus. The number of islets in the DC group was decreased and they were unevenly distributed, with different degrees of atrophy, and vacuoles appeared within them. A further loss of β-cells was noted in the diabetic group within irregularly shaped islets. Ill-defined β-cells were loosely arranged, the cell volume was increased and the nucleus exhibited an abnormal distribution (Fig. 1A). The number of positively stained cells decreased and the insulin staining was decreased compared with in the NC group (Fig. 1B), suggesting that islet β-cells showed marked damage after establishment of the animal model. Due to the limitations of tissue size and the specific parts of sections, the number of the islets was not identical on different slides, and each slide could count ~three to ten islets. A number of the small islets only appear to be small as they were produced by cutting through the edge of an islet, the maximum islet area was used as an indicator in each slide to assess the islet area. Compared with that in the NC group, the islet area in the DC group was significantly smaller (P<0.01; Table I). The S1P administration group exhibited less damage; the number and volume of islet cells was increased, and the islet cells had a clearer structure, and a more regular shape and arrangement compared with the DC group (Fig. 1A). Immunohistochemistry on the pancreas of the NC mice revealed intense insulin staining, whereas weak staining was observed in the core of the islets in the DC group. The islets in the S1P administration group displayed stronger insulin immunostaining intensity compared with the DC group (Fig. 1B). The islet area in the S1P group was significantly decreased compared with that in the NC group (P<0.05) and appeared larger than that in the DC group, but the difference compare with the NC group was not statistically significant (Table I). An increase in positive insulin staining granules suggested that S1P could decrease islet damage and improve β-cell morphology. The average optical density of the insulin immunohistochemistry (IOD/area) and maximum islet area were calculated using Image Pro Plus 6.0 image analysis system (Table I).

The blood glucose level was measured before and 1, 2 and 3 weeks after S1P administration to observe the effect of S1P on the fasting blood glucose levels of T2DM mice. The fasting blood glucose in the DC mice was significantly increased compared with the NC group (P<0.01; Table II); 1 week after S1P administration, the fasting levels in the S1P group were decreased slightly compared with the diabetic model group, but the difference was not significant (P>0.05). The same trend was observed at weeks 2 and 3 after the administration of S1P. No statistically significant difference was observed in the fasting blood glucose between the DC and S1P group (Fig. 2; Table SI).

IPGTT was conducted 3 weeks after S1P administration to observe the effect of S1P on pancreatic β-cell function and metabolic regulation in T2DM mice. Compared with that in the NC group, the blood glucose in the DC mice reached its peak 30 min after glucose injection and maintained a high level throughout the whole experiment. The blood glucose in the S1P mice appeared to decrease 1 and 2 h after glucose injection, compared with in the DC group, but the difference was not significant (P>0.05; Table II).

Fasting serum insulin levels were not significantly different among groups (all P>0.05; Table III) and were consistent with insulin levels in T2DM. The AUC and HOMA-IR index were calculated in all groups. Compared with that measured in the NC group, the HOMA-IR index in the DC mice was significantly increased (P<0.01), whereas that in S1P group decreased slightly compared with that in the DC group, but the difference was not statistically significant (P>0.05; Table III). The HOMA-IR exhibited a trend similar to that observed with the AUC. Compared with that calculated in the NC group, the AUC was significantly increased in the DC and S1P groups (both P<0.01). The AUC calculated for the S1P group appeared to decrease compared with the DC group, but with no statistical significance (P>0.05; Table III).

Next, an investigation into whether exogenous S1P promoted β-cell proliferation was carried out. Ki-67-positive staining appeared as a brownish yellow colour in the nucleus, whereas the cell membrane and cytoplasm displayed no staining (Fig. 3A and C). A total of three to five unique fields of view of the islets were selected randomly and the cell proliferation rate of each islet was calculated separately, then the mean value was taken for analysis. The rates of Ki-67 expression were <3% in all groups of mouse islets (Fig. 3D). Fewer Ki-67-positive cells were observed within the S1P administration group of mouse islets compared with the NC group (Fig. 3A and C), whereas almost no Ki-67-positive cells were noted in the diabetic model control group (Fig. 3B). Ki-67-positive cells increased significantly in the S1P group compared with the DC group (P<0.05; Fig. 3D; Table SII).

Islet β-cell dysfunction and insulin resistance are multifaceted with their interdependence for triggering the pathogenesis of T2DM. TUNEL staining of mouse islet β-cell was used to determine the percentage of β-cells that were undergoing apoptosis (Fig. 4). The apoptotic islet cell nuclei were stained brown, whereas normal nuclei were blue (Fig. 4A-C). A significantly increased number of apoptotic cells were observed in the islets of the DC mice, compared with those counted in the NC mouse islets (P<0.01; Fig. 4D; Table SIII). A significantly increased number of TUNEL-positive cells was observed in the diabetic model group compared with the S1P administration group (P<0.05; Fig. 4D; Table SIII), suggesting that S1P serves a positive role in protecting islet cells against apoptosis.

As most of the S1P-induced effects were mediated via S1PR, it was of interest to investigate the differences of protein expression of the S1PR subtypes in the mouse pancreas. S1P signalling is mediated via the activation of specific S1PRs (S1PR1-5), of which mouse islet β-cell express mainly S1PR1-3 (22). To determine which S1PR subtypes mediate the S1P-induced protective response in islet cells, normal and diabetic mouse islets were incubated with anti-S1PR1-3 antibodies. The positive S1PR1, S1PR2 and S1PR3 protein staining was localised in the cytoplasm and membrane of the islet cells, with almost no expression in the exocrine pancreas, consistent with the insulin staining results (Fig. 5A). Compared with that in the NC group, the positive staining of the S1PR1 protein in the diabetic mice was stronger and exhibited more uneven distribution (Fig. 5A). The Image Pro Plus 6.0 software image analysis system was used to calculate the average density of immunohistochemical S1PR1 protein staining (IOD/area) and significant differences were observed between the two diabetic groups, and the NC group (P<0.01; Fig. 5B; Table SIV). S1PR2 exhibited a similar expression trend to S1PR1 (Fig. 5C; Table SIV). No statistical difference was observed in the expression of S1PR3 protein between the diabetic and NC group (P>0.05; Fig. 5D; Table SIV). These results indicate that S1PR1, S1PR2 and S1PR3 proteins were expressed in the pancreatic β-cells. The protein expression of S1PR1 and S1PR2 in the diabetic mice was significantly increased compared with the NC group (P<0.01 and P<0.05, respectively; Fig. 5B and C), whereas no significant difference was observed in the expression of S1PR3 (Fig. 5D). Overall, the findings of the present study indicate that extracellular S1P induces islet β-cell proliferation and inhibits cell apoptosis, possibly via S1PR1 and S1PR2 activation, leading to enhanced islet β-cell protection.