Male C57BL/6J mice (4-week-old, approximately 17–18 g) purchased from Jackson Laboratories (Bar Harbor, ME, United States) were housed in a specific pathogen-free animal vivarium at St. Michael’s Hospital, and maintained on a 12-h light-dark cycle, with free access to standard rodent chow and water. For low-dose STZ induced diabetic mouse model, 20 mice were randomly assigned into four groups after 1-week of adaptive housing: non-treatment diabetic control group (Water), GABA treatment group (GABA), sitagliptin treatment group (Sita), and GABA plus sitagliptin group (GABA+Sita) (Figure 1A). For high-dose STZ induced diabetic mouse model, 20 mice were also randomly assigned into four groups: non-treatment diabetic control group (HD-STZ+Water), GABA treatment group (HD-STZ+GABA), sitagliptin treatment group (HD-STZ+Sita), and GABA plus sitagliptin group (HD-STZ+GABA+Sita) (Supplementary Figure S1A). GABA (Sigma–Aldrich, St. Louis, MO, United States, 6 mg/mL in drinking water), or sitagliptin (Merck Inc., Montreal, ON, Canada, 0.4 mg/L in drinking water), or GABA and sitagliptin were orally administrated at the age of 6 weeks and maintained during the treatment course of 6 weeks. At the age of 7 weeks, all mice received STZ (Sigma–Aldrich, 40 mg/kg for 4 consecutive days or 125 mg/kg for 2 consecutive days) via injection, as we have previously described (Soltani et al., 2011). Body weights and blood glucose levels were measured twice a week; metabolic cages were used before the termination of the experiments (Soltani et al., 2011). Blood samples were collected before treatment and sacrifice. During blood sample collection, diprotin A (Sigma–Aldrich) and aprotinin (Bioshop, Burlington, ON, Canada) were added to inhibit the degradation of GLP-1 and glucagon (Farngren et al., 2014). All animal experiments were conducted in accordance with the guidelines put forth by the Canadian Council on Animal Care and were approved by the University of Toronto Animal Care Committee.

Non-fasting blood glucose levels were measured with mouse tail blood using a One Touch Basic glucometer (LifeScan Inc., Burnaby, BC, Canada) or glucose assay kit (Abcam, Cambridge, United Kingdom). Glucose tolerance and insulin sensitivity were evaluated by IPGTT and IPITT at 5–6 weeks old (as baseline) and at 11–12 weeks old (after drug intervention). Mice were fasted for 15 h for IPGTT and 6 h for IPITT, as we described previously (Ip et al., 2015).

Pancreases were paraffin embedded and prepared for histological analysis (Wang and Brubaker, 2002). Briefly, freshly isolated pancreas was cut into 8–10 segments followed by formaldehyde fixation. All pancreatic pieces were embedded in paraffin after being dehydrated in ethanol and cleaned with xylene. Insulin and glucagon dual staining were performed on tissue sections (5 μm) by using guinea pig anti-insulin and rabbit anti-glucagon antibodies (1:1000; DAKO, Burlington, ON, Canada); then detected with fluorescent (Cy3- and FITC- conjugated IgG) or biotinylated secondary antibodies (1:200; Abcam, Cambridge, United Kingdom). For IHC staining, samples were incubated with avidin-biotin-peroxidase complex (Vector Laboratories, Burlington, ON, Canada) before staining with DAB (Vector Laboratories) or Fuchsin red (DAKO) and subsequent hematoxylin counterstaining. Entire pancreatic images were scanned and viewed with NanoZoomer (Hamamatsu, Hamamatsu-shi, Shizuoka-ken, Japan) and analyzed by using the ImageScope program (Aperio Technologies, Vista, CA, United States) (Purwana et al., 2014). Total alpha and beta cell mass were determined by the product of cross-sectional alpha and beta cell area over total tissue area and the weight of pancreatic tissue before fixation.

Proliferative beta cells were detected in the pancreatic sections by double immunofluorescence staining with guinea pig anti-insulin (1:1000; DAKO), rabbit anti-Ki67 (1:200; Thermo Fisher, Burlington, ON, Canada) antibodies, and relevant secondary antibodies (1:1000; Abcam). Regenerative beta cells were detected by double immunofluorescence staining with guinea pig anti-insulin (1:1000; DAKO), rabbit anti-PDX-1 (1:400; Cell Signaling Technology, Danvers, MA, United States) antibodies, and relevant secondary antibodies (1:200; Jackson ImmunoResearch Laboratories, West Grove, PA, United States). Apoptotic beta cells were also identified in pancreatic sections with insulin and terminal deoxynucleotidyl transferase dUTP nick end labeling (Tunel) labeling (TMR red, Roche, Mississauga, ON, Canada) (Wang and Brubaker, 2002; Robertson et al., 2008). Results are expressed as the percentage of Ki67⁺, PDX-1⁺, or Tunel⁺ beta cells. All immunofluorescent images were captured by an Olympus upright BX50 fluorescence microscope (Olympus, Richmond Hill, ON, Canada) at ×40 magnification.

Plasma insulin, glucagon, and active GLP-1 concentrations were measured using an ultra-sensitive mouse insulin ELISA kit (Crystal Chemical Inc., Wakefield, MA, United States), a glucagon RIA kit (Millipore, Etobicoke, ON, Canada), and a high sensitivity GLP-1 active chemiluminescent ELISA Kit (Millipore), according to the manufacturer’s instructions (Chiang et al., 2014).

All data were expressed as mean ± SD for four independent experimental groups. Statistical analysis was performed using SPSS for Mac Ver. 20.0 (SPSS, Inc., Chicago, IL, United States). All graphs were made using the Prism program (GraphPad, San Diego, CA, United States). The significance (P < 0.05) among different groups was evaluated using one-way ANOVA followed by the Tukey test.

To assess the potential additive effect of combined use of GABA and sitagliptin in preventing hyperglycemia, we administrated GABA, sitagliptin, or GABA plus sitagliptin orally 1 week before STZ injection (Figure 1A). Two weeks after STZ injection, the control diabetic mice developed serious hyperglycemia (16.8 ± 3.1 mM) (Figure 1B). Mice in the three treatment groups, however, showed significantly lower ambient blood glucose levels (<11 mM). At 2 weeks after STZ injection and thereafter, mice from the GABA+sitagliptin group showed lower blood glucose levels when compared with those that received a monotherapy. Whereas the non-treated control mice developed severe hyperglycemia (30.5 ± 0.9 mM) after a 6-weeks period, mice receiving GABA or sitagliptin monotherapy showed significantly lower glycemic levels (15.3 ± 0.5 mM and 13.0 ± 0.6 mM). Moreover, blood glucose level in the GABA+sitagliptin group was at an even lower range (10.6 ± 0.3 mM, P < 0.001 between GABA+Sita group and GABA group; P < 0.01 between GABA+Sita group and Sita group) (Figure 1B).

To further demonstrate that the combined therapy has a superior effect on lowering glucose levels, we also examined high-dose STZ injection, and this revealed larger and more significant differences between the combined therapy group and monotherapy groups (Figure 1C). After 6-weeks treatment, the glucose levels of HD-STZ+Water, HD-STZ+GABA, HD-STZ+Sita, and HD-STZ+GABA+Sita groups are 51.0 ± 8.3, 31.6 ± 3.4, 37.9 ± 3.4, and 21.0 ± 7.7 mM, respectively (P < 0.05 between HD-STZ+GABA+Sita group and HD-STZ+GABA group; P < 0.05 between HD-STZ+GABA+Sita group and HD-STZ+Sita group).

To test whether this combined therapy leads to a better glucose and insulin challenge response, IPGTT and IPITT were performed before (indicated as baseline in Figures 1D–G) and 5 weeks after the drug treatment. As shown in Figures 1D,E, GABA or sitagliptin monotherapy partially improved glucose tolerance while the combined therapy improved the tolerance even further. Figures 1F,G shows that insulin sensitivity is increased by either GABA or sitagliptin monotherapy, while combined therapy tended to increase this even further, albeit without statistical significance.

We then determined circulating insulin and glucagon levels in each of the four groups of mice before and after drug treatment. Both GABA and sitagliptin monotherapies increased plasma insulin levels and decrease plasma glucagon levels (Table 1). An additive effect on the elevation of plasma insulin levels but not on the repression of plasma glucagon levels was observed with the combined therapy. As anticipated, plasma GLP-1 levels in mice that received sitagliptin or GABA plus sitagliptin were significantly elevated (Table 1).

Expected therapeutic effects on T1D include the reduction of water consumption and urine volume, as well as an improvement in body weight gain. Although we did not find any effects of GABA, sitagliptin, or combined therapy on body weight changes or food intake (Figures 2A,B), we observed a repression of water consumption in mice that received GABA or sitagliptin treatment (Figure 2D). Importantly, combined therapy decreased water consumption even further (Figure 2D). Figure 2C shows that the daily urine volume was lower in mice of the GABA group, the sitagliptin group, and the GABA+sitagliptin group when compared with the non-treatment diabetic controls.

In this MDSD model, body weight differences among the four groups of mice were very marginal. Nevertheless, we have also tested the therapeutic effect of GABA, sitagliptin, and GABA+sitagliptin in mice that received a large-dose of STZ (Supplementary Figure S1A). As shown in Supplementary Figure S1B, each of the three treatments increased the body weight of the mice with time.

To initiate the exploration on mechanisms underlying the additive improvement with the combined therapy, we determined beta cell and alpha cell mass in the four groups of mice. As shown in Figure 3A, STZ injection destroyed nearly all beta cells, while the residual islets contained mostly alpha cells. GABA or sitagliptin monotherapy partially mitigated such changes, while the combined therapy altered the changes the most. Figures 3B,C shows that either one of the monotherapies increased beta cell mass and decreased alpha cell mass. Importantly, combined therapy increased beta cell mass by nearly twofold when compared with either of the mono-therapeutic approaches, and generated a trend in further reducing alpha cell mass.

As GABA or GLP-1 can promote mouse beta cell survival and replication (Wang and Brubaker, 2002; Soltani et al., 2011), we assessed whether combined therapy could generate an additive effect in this MDSD model. We used a Ki67/insulin dual immunofluorescence staining approach to determine the beta cell replication rates. In the MDSD mice that received no drug treatment, the rate of beta cell proliferation was 0.40 ± 0.03%, consistent with our previous report under the similar experimental conditions (Soltani et al., 2011). In contrast, the rates of beta cell proliferation in mice that received GABA or sitagliptin treatment were ∼1.12 and 0.94%, respectively (Figures 4A,C). Moreover, mice that received the combined therapy showed further increase in the rate of beta cell proliferation (∼1.45%).

To further elucidate the underlying mechanism of beta cell replication under our treatment conditions, we performed dual staining for insulin and PDX-1, one of the key beta cell developmental markers. Consistent with the Ki67 results, the PDX-1⁺insulin⁺ cell counts are significantly increased in GABA (22.14 ± 1.97%) and sitagliptin (20.32 ± 5.45%) treated groups compared with non-treated diabetic mice (12.84 ± 2.17%); while combined therapy increased this even more (29.83 ± 3.49%) (Figures 4B,D). In addition, we also demonstrated that the small-size islet numbers (beta cell count less than 50) are significantly higher in the combined therapy group (Figure 4E), which might partially explain the pro-regenerative effects of combined use of GABA and sitagliptin.

The Tunel and insulin double immunostaining method was then utilized to determine beta cell apoptosis. As shown in Figures 5A,B, GABA or sitagliptin monotherapy decreased the Tunel positive beta cells, while GABA+sitagliptin treatment generated an additive effect on reducing beta cell apoptosis. As previously reported by our group and others (Prud’homme and Chang, 1999; Maksimovic-Ivanic et al., 2002; Soltani et al., 2007, 2011), beta cell apoptosis is a slower process in an MDSD model compared with high-dose STZ-induced diabetes. Notably, beta cell death in MDSD is largely dependent on inflammation (insulitis), and apoptosis is still apparent weeks after STZ administration, as shown by our current results. Thus, our Tunel staining results in MDSD suggest that combined GABA and sitagliptin treatment exerts superior and long-term protection against beta cell death.