Formation of pancreatic β-cells from precursor cells contributes to the reversal of established type 1 diabetes

Abstract

Ig-GAD2, an antigen-specific immune modulator, requires bone marrow (BM) cell transfer in order to restore beta (β)-cell formation and induce recovery from established type 1 diabetes (T1D). The BM cells provide endothelial precursor cells (EPCs) that give rise to islet resident endothelial cells (ECs). This study shows that, during development of T1D, the immune attack causes collateral damage to the islet vascular network. The EPC-derived ECs repair and restore islet blood vessel integrity. In addition, β-cell genetic tracing indicates that the newly formed β-cells originate from residual β-cells that escaped the immune attack and, unexpectedly, from β-cell precursors. This indicates that the rejuvenated islet microenvironment invigorates formation of new β-cells not only from residual β-cells but also from precursor cells. This is twofold significant from the perspective of precursor cells as a safe reserve for restoration of β-cell mass and its promise for therapy of T1D long after diagnosis.

Graphical Abstract

1. Introduction

T1D is a chronic disease in which cells of the immune system infiltrate the pancreatic islets causing an inflammatory process that destroys insulin-producing β-cells [1]. The non-obese diabetic (NOD) mouse develops T1D spontaneously and has always been the model of choice to investigate the pathogenesis of the disease [2, 3]. Because T1D is mediated by T lymphocytes reactive with β-cell-associated antigens (Ag) [4], it was logical to envision specific inactivation or elimination of these aggressive T cells as a means to halt the disease [5–8]. Indeed, reagents that target T cells have been developed and were able to suppress T1D [5, 6, 9–14]. Our own Ag-specific studies have shown that Ig-GAD2, an immunoglobulin (Ig) molecule genetically engineered to carry the I-A^g7-restricted T cell epitope corresponding to amino acid sequence 206–220 of glutamic acid decarboxylase (GAD) is able to serve as an immune modulator and suppress T1D when given at the prediabetic stage [15]. However, despite the ability of Ig-GAD2 to clear islet infiltration and sustain formation of new β-cells, the regimen could not induce recovery from disease when intervention was made at the diabetic stage [16]. Intriguingly, supplementation of Ig-GAD2 treatment with bone marrow (BM) cells from health mice led to immune modulation, β-cell formation and recovery from established T1D [16]. The BM cells contained endothelial precursor cells (EPCs) which give rise to mature endothelial cells (ECs) that populated the pancreatic islets [17]. In fact, EPCs were able to substitute for whole BM during Ig-GAD2 treatment and recovery from disease [16, 17]. The role EPCs play in the recovery from established T1D remains unclear. This study used the Ig-GAD2+BM combination therapy and β-cell genetic tracing to address two fundamental questions as to the settling of EPCs in the pancreatic islets and the origin of newly formed β-cells that drive recovery from the disease. The insight to be gained from these should inform the development of therapies for established T1D long after diagnosis, a matter of relevance to human T1D.

2. Material and Methods

2.1. Mice

All mouse experiments were approved by the University of Missouri, Columbia Animal Care and Use Committee. All animals were maintained in the Animal Facility at the Medical Science building under barrier conditions. Mice handling was guided by the Veterinarian overseeing the Medical School Vivarium. The NOD (H-2^g7), NOD.RIP-CreER (NOD.Cg-Tg(Ins2-cre/ERT)1Dam/SbwJ), and B6.Rosa26-YFP (B6.129X1-Gt(ROSA)26Sor^tm1(EYFP)Cos/J) were purchased from The Jackson Laboratories (Bar Harbor, ME). The Rosa26-YFP NOD mice were generated by breeding B6. Rosa26-YFP onto the NOD background via speed congenic technology based on 58 microsatellite markers on sequences between the B.6 donor and NOD recipient strains for a total of 8 backcrosses. To generate RIP-CreER/ROSA26-YFP NOD mice, NOD.RIP-CreER mice were bred with Rosa26-YFP NOD mice. The resulting offspring which were heterozygous for Rosa26-YFP and hemizygous for RIP-CreER were bred with Rosa26-YFP NOD mice to obtain female mice homozygous for Rosa26-YFP and hemizygous for RIP-CreER.

Mice were maintained at 20 – 26°C, 30 – 70% humidity, and under a 12/12 h light-dark cycle in a barrier condition. Mice had constant access to water and standard chow. All experiments were performed in female mice with 4 – 8 mice per experimental group. To avoid bias, mice generated from indiscriminate breeders were randomly assigned to control or experimental groups at week 4 of age. At week 6 of age, the experimental group RIP-CreER/ROSA26-YFP NOD mice were given tamoxifen to genetically label insulin-producing β-cells with YFP. At 10 weeks of age, all mice began blood glucose level (BGL) monitoring using Nova Max Plus (Nova Biomedical, Waltham, MA). A mouse was considered diabetic when the blood glucose levels were ≥300 mg/dl for two consecutive days. Mice were then enrolled in the treatment regimen depicted in Figure 1. Every mouse in each experimental group was included in the final analysis. Each animal served as a biological replicate.

Figure 1.

Schematic representation of Ig-GAD2+BM combination therapy. At the onset of diabetes (BGL ≥ 300mg/dl for 2 consecutive days) the mice receive two insulin pellet implants (LinShin, Toronto, ON, Canada) subcutaneously to maintain normoglycemia for three weeks. This is considered day 1 of the therapy regimen. The next day the mice are given i.p. 300 μg Ig-GAD2 in saline per mouse. Another injection is given every other day totaling 3 injections per week for 5 weeks. In parallel, the mice received intravenously (i.v.) 10 ×10⁶ BM cells on week 2, 3, and 4. Subsequently, one Ig-GAD2 injection was given per week for the last 5 weeks.

2.2. Genetic labelling of insulin-producing β-cells with YFP

Tamoxifen (Sigma, Saint Louis, MO) was dissolved in corn oil at 20 mg/ml and used to induce CreER relocation to the nucleus and removal of the stop codon allowing for insulin-producing β-cells to express YFP. Tamoxifen was given intraperitoneally (i.p.) to RIP-CreER/ROSA26-YFP NOD mice every other day three times in total at 3mg or 7mg per mouse.

2.3. Ig-GAD2+BM combination therapy

This was performed as previously described [16] with slight modifications as indicated in the legend to Figure 1. The experimental groups included mice receiving Ig-GAD2 only, BM cells only or a combination of Ig-GAD2+BM. The control groups were healthy and diabetic mice. At the end of the treatment, mice were euthanized according to the 2020 AVMA euthanasia guidelines and the pancreas, pancreatic lymph nodes and spleen were harvested for further analysis. Prior to commencement of treatment, diabetic mice received two insulin implants to sustain normoglycemia for approximately three weeks. To administer the insulin implant, cotton gauze soaked with 1cc isoflurane (VET ONE, Boise, ID) was put inside a wire mess and placed in a closed 1L anesthetic glass jar. Mouse was closely monitored for ~1 min and removed. To ensure adequate anesthesia, the mouse was pinched on the toe for noxious stimulus followed by decontamination of implantation site (underbelly) with 50 ul of 10% betadine solution (Purdue Products, Stamford, CT). Two insulin implants (LinShin, Toronto, ON, Canada) were sterilized by immersion in 2% Betadine solution and inserted subcutaneously using a 12G insertion needle. The mouse was put back into the home cage and closely monitored every day for the next five days.

2.4. Analysis of insulitis

Pancreata were harvested, weighed, fixed in 4% formaldehyde for 6 h at 4°C, and immersed in 30% sucrose overnight prior to freezing at −80°C. Frozen pancreata were transferred into tissue-freezing medium (Sakura Finetek, USA), and cut into 5-μm-thick non-serial sections 150 μm apart to avoid repeated counting of the same islet. Sections were stained with hematoxylin and eosin and cell infiltration was analyzed by light microscopy. Insulitis was scored based on the following criteria: mild insulitis, <50% of the islet area is infiltrated; peri-insulitis, infiltration restricted to islet periphery; severe insulitis, ≥50% of the islet area is infiltrated; and no insulitis, absence of islet infiltration. Four to six non-serial sections per pancreas were used for each experiment and every islet/section was scored.

2.5. Analysis of islet blood vessel integrity

Immunohistochemistry.

Integrity of pancreatic islet blood vessels was analyzed by light microscopy upon staining with antibodies to platelet endothelial cell adhesion molecule 1 (PECAM1/CD31) and pericyte marker, PDGFRβ. Briefly, pancreatic sections were stained with primary rabbit anti-CD31 (1:50; Abcam) or rat anti-PDGFRβ (1:100; Thermo Fisher) antibodies at 4°C overnight, followed by incubation with horseradish peroxidase-conjugated goat anti-rabbit (1:2000; Abcam), or goat anti-rat (1:5000; Thermo Fisher) antibodies respectively, for 30 mins at room temperature. The sections were then incubated with the IHCWORLD substrate and 3-amino-9-ethylcarbazole (AEC) chromogen. Cell nuclei were counterstained with hematoxylin. The images were visualized and captured with a Leica DM5500 B microscope.

Immunofluorescence.

Analysis of blood vessel integrity was also carried out by Confocal Microscopy upon vessel immunostaining of mechanically isolated islets. The islets were isolated from pancreata with minimal toxicity as previously described [18]. In brief, mice were euthanized and pancreata were perfused with collagenase, and digestion was carried out at 37°C to release islet from connective tissues. The islets were then treated with zinc-chelating dye (Dithizone, Sigma) and hand-picked under a dissecting microscope.

The islets were preincubated with blocking buffer (1% BSA, 1% goat serum, 1% donkey serum, and 0.2% Triton X-100) at RT for 1 hour and then stained with primary guinea pig anti-insulin (1:100; Abcam), rabbit anti-CD31 (1:10; Abcam) and rat anti-PDGFRβ (1:10; ThermoFisher) antibodies at 4°C overnight. The islets were washed and stained overnight at 4°C with the secondary antibodies; Alexa Fluor 647–conjugated goat anti-guinea pig (1:100; Abcam), Alexa Fluor 488–conjugated donkey anti-rabbit (1:50; Abcam) and Alexa Fluor 594-conjugated chicken anti-rat (1:40; Invitrogen). The islets were then washed, resuspended in mounting medium, and pipetted unto a glass slide, followed by a cover slip and 4°C overnight cure. The images were visualized and acquired with a Leica TCS SP8 confocal microscope. For quantitative analysis of the vascular networks of each individual islet, AngioTool software was used (http://rsbweb.nih.gov/ij/). Briefly, images from confocal microscopy was input into the system and then blood vessels were identified using the multiple Hessian analysis and skeletonization followed by automated analysis of morphological and spatial parameters.

2.6. Measurement of YFP-positive (YFP⁺) and YFP-negative (YFP⁻) β-cells

Pancreatic sections were blocked with a PBS solution containing 1% BSA, 1% goat or donkey serum, and 0.2% Triton X-100 at RT for 1 hour. The sections were then incubated overnight at 4°C with primary antibodies, including rabbit anti-GFP (1:500; Abcam), guinea pig anti-insulin (1:300; Abcam), rabbit anti-CD31 (1:100; Abcam), rat anti-PDGFRβ (1:100; ThermoFisher) and eFluor 570-conjugated Ki-67 monoclonal Ab (1:200; Invitrogen). The slides were washed with three changes of 0.02% Triton X-100 in PBS and then stained for 1 hour at RT with the corresponding secondary antibodies, including Alexa Fluor 647–conjugated goat anti-guinea pig (1:300; Abcam), Alexa Fluor 488–conjugated donkey anti-rabbit, (1:500; Abcam) and Alexa Fluor 594–conjugated chicken anti-rat (1:400; Abcam). DAPI mounting medium (Sigma) was used for nuclear staining. The images were visualized and acquired with a Leica TCS SP8 confocal microscope. The number of β-cells and islets was scored with a computer-assisted Image Pro Plus program (Media Cybernetics, Silver Spring, MD). To determine the number of β-cell and islets per mouse, each pancreas was sectioned at every 150 μm, spanning both proximal and distal areas of the pancreas, and all the resulting sections were used for the analysis.

2.7. Measurement of β-cell mass

Pancreata were harvested, weighed and nonserial sections covering both proximal and distal areas were prepared. The sections were stained with primary guinea pig anti-insulin (1:300; Abcam) antibodies at 4°C overnight, incubated with horseradish peroxidase-conjugated goat anti-guinea pig (1:2500; Abcam) antibodies respectively, for 30 mins at room temperature. Insulin expression was identified by incubating sections with DAB chromogen and substrate (IHCWORLD). Cell nuclei were counterstained with hematoxylin. Β-cell mass was determined by pixel-based technology as previously described [16]. In brief, cross-sectional areas with insulin positive cells were measured at 150μm intervals and normalized to total pancreatic area on Image-Pro Plus program. Β-cell mass (mg) was derived following normalization to total pancreatic mass. Four to six non-serial sections per mouse were stained and at least 120 islets were assessed for each group.

2.8. Flow Cytometry

Cell surface staining.

For detection of CD45 and PECAM1 (CD31), peripheral blood samples were collected, incubated with red blood lysis buffer and resulting cells were stained with allophycocyanin (APC)-cy7-conjugated anti-CD45 (BD Biosciences) and phycoerythrin (PE)-cy7-conjugated anti-CD31 (Invitrogen). Dead cells were detected by 7-AAD (Invitrogen) staining.

Intracellular staining.

For detection of intracellular interferon-γ (IFN-γ) and IL-17 in CD4⁺ T cells, splenic and pancreatic lymph node cells were stimulated with phorbol myristic acid (PMA) (50 ng/mL) and ionomycin (500 ng/mL) for 4 hour in the presence of brefeldin A (10 μg/mL) and then stained with peridinin-chlorophyll-protein (PerCP)-cy5.5–conjugated anti-CD4 (RM4–5) and fluorescein isothiocyanate (FITC)–conjugated anti-CD3 (145–2C11) antibodies (all from BD Biosciences) for cell surface staining. Subsequently, the cells were fixed and permeabilized with Fix/Perm buffer (eBioscience) and incubated with allophycocyanin-conjugated anti-IFN-γ and PE-conjugated anti–IL-17 Abs (BD Biosciences) for intracellular staining. The samples were read using a Beckman Coulter CyAn ADP, and data were analyzed using FlowJo 10.0.8 software (TreeStar).

2.9. Statistical analysis

The p values were calculated using the two-tailed Student t test. Comparison of more than two groups was performed using One-way ANOVA. All data are presented as mean ± SEM, and p < 0.05 indicated significance. Prism Software v4.0c (GraphPad) was used in all statistical analyses.

3. Results

3.1. Exogenous EPCs give rise to ECs that repair islet vasculature during recovery from T1D

We have previously shown that recovery from overt T1D requires immune modulation and donor bone marrow (BM) cells from healthy mice [16]. Histologic analysis indicated that the BM cells gave rise to endothelial cells (ECs) that migrated to the pancreatic islets but did not colocalize with insulin producing β-cells [16]. Also, endothelial precursor cells (EPCs) substitute for BM cells and assist immune modulation for induction of recovery from overt T1D [16]. Given that the pancreatic islets are highly vascularized and the inflammatory process may damage the vascular network [19], we sought to determine whether the BM-derived ECs contribute a repair process to blood vessels within the pancreatic islets. To this end, diabetic NOD mice were treated with Ig-GAD2+BM combination therapy as indicated in Figure 1 and on day 70 (week 10) of recovery from disease the pancreatic islet were examined for vessel integrity both by light and confocal microscopy. The immunohistochemistry staining results obtained for endothelial PECAM1 and pericyte PDGFRβ markers show that the islet vasculature in mice recipient of the combination therapy exhibited fine arrangement of the vascular network similar to healthy young animals while untreated recently diagnosed diabetic controls had a degenerated vascular network (Fig. 2A and B). Furthermore, 3D confocal microscopy analysis of vascular integrity within the same insulin-producing islet shows similar structural findings as light microscopy for both PECAM1 and PDGFRβ staining (Fig. 2C). Because PECAM1 is specific for endothelial cells the 3D images were suitable for fine and quantitative analysis of vessel integrity. This analysis shows that vessel length, density and lacunarity in the islets from Ig-GAD2+BM-treated mice are similar to islets from healthy controls (Fig. 2D). However, islets from diabetic mice showed high lacunarity, low density and the vessels were shorter (Fig. 2D). In order for blood vessels to regain integrity the treatment must restore the frequency of endothelial cells (ECs) to meet the needs of the repair process. This proved to be correct as the number of ECs in mice recipient of Ig-GAD2+BM combination therapy increased progressively while those in control untreated diabetic mice decreased gradually over time (Fig. 3A). In fact, in Ig-GAD2+BM treated mice the ECs rose to 62% relative to 4-week-old healthy NOD mice on day 60 post treatment while those in untreated diabetic mice declined to 20% during the same period of time (Fig. 3B). Overall, the endothelial precursor cells (EPCs) from BM transfer during Ig-GAD2 treatment proliferate and mature into ECs [16] that serve to repair the damaged islet vascular network.

Figure 2.

Restoration of vascular integrity in the islets during recovery from overt T1D. (A, B) Show representative (A) PECAM1 and (B) PDGFRβ AEC-labeled (red) and hematoxylin counterstained (blue) images of pancreatic sections from 6-week-old healthy (BGL ≤ 140 mg/dl), recently diabetic (day 2 of diagnosis), and Ig-GAD2+BM-treated (day 70 post treatment) mice. The right panels in both (A, B) show higher magnifications of the boxed areas depicted in the left panels as indicated. (C) Mechanically purified islets from healthy, diabetic and Ig-GAD2+BM-treated mice (4 per group) were immunostained with PECAM1 (green), PDGFRβ (yellow) and insulin (red) and representative images were obtained by 3D reconstructions of confocal z-stacks. White scale bars, 20 μm. (D) Shows graphical representations of vessel density (top), total vessel length (middle) and lacunarity (bottom) obtained from PECAM1 confocal analyses for all 3 groups of mice. At least 100 islets were analyzed for each group mice. Error bars represent mean ± SEM. ***p < 0.0005 as determined by two-tailed Student t test.

Figure 3.

Recovery from T1D correlates with increase in peripheral blood ECs. Ig-GAD2+BM-treated mice were bled once every 10 days for a period of 60 days and the peripheral blood cells were stained with anti-CD45, anti-PECAM1 and 7-AAD. Diabetic mice that received insulin implants on day 2 and 30 post diagnosis but no Ig-GAD2+BM treatment were used as control. (A) Shows the mean ± SEM number of live (7-AAD^-) ECs (PECAM1⁺ CD45^-). *p < 0.05, **p < 0.005 as determined by two-tailed Student t test. (B) Shows the ratio of average number of ECs in diabetic or Ig-GAD2+BM mice divided by the average number of ECs obtained from healthy mice times 100. *p < 0.05 as determined by Mann-Whitney U test. Each group included four mice.

3.2. NOD mice with genetically labeled insulin-producing β-cells recover from overt T1D upon treatment with Ig-GAD2+ BM combination therapy

During recovery from T1D, restoration of β-cell mass may result from division of residual β-cells or differentiation of precursor cells. To address this premise, we sought to generate a mouse model where insulin-producing β-cells can be genetically labeled with yellow fluorescent protein (YFP) prior to development of T1D which will be then used to determine the origin of β-cells during recovery from disease. To this end, RIP-CreER transgenic NOD mice in which the insulin promoter drives the tamoxifen-dependent expression of Cre-recombinase-estrogen receptor (ER) fusion gene [20] were bred to ROSA26-YFP NOD mice yielding a RIP-CreER/ROSA26-YFP NOD strain (Fig. 4A). These mice were responsive to tamoxifen as their insulin-producing pancreatic β-cells stained positive with anti-YFP antibodies whether the treatment was carried out with 3 or 7 mg tamoxifen regimen (Fig. 4B). Control mice not treated with tamoxifen stained negative with the same anti-YFP antibody (Fig. 4B). The 3 mg tamoxifen dose led to YFP staining of 50% of insulin-producing cells (Fig. 4B, middle panel of the bar graph) while the 7mg dose yielded YFP staining of almost all insulin-producing cells (Fig. 4B, lower panel of the bar graph). These results indicate that the mouse model sustains optimal genetic labeling of insulin-producing β-cells. In addition, the RIP-CreER/ROSA26-YFP NOD mice, while become diabetic earlier than wild-type NOD mice, the pattern of disease at the plateau phase is similar to wild-type mice (Fig. 4C). This is perhaps due to the fact that the wild-type NOD mice represent an established colony while the RIP-CreER/ROSA26-YFP NOD mice come from a fresh colony that remains in the adaptation phase to the conditions in our vivarium. Overall, the disease susceptibility was not compromised by the genetic labeling with YFP. To ensure that RIP-CreER/ROSA26-YFP NOD mice are responsive to treatment with Ig-GAD2+BM [16], the animals that developed overt T1D (BGL ≥ 300 mg/dl) were subjected to the treatment regimen illustrated in Figure 1 and monitored for recovery from disease. Figure 5A shows that most of the mice recipient of Ig-GAD2+BM combination therapy are no longer diabetic and the mean BGL was 240 ± 38 mg/dl on day 70 which is the last day of monitoring period, while those given BM or Ig-GAD2 alone could not recover from T1D (Fig. 5A, left panel). In fact, while 6/8 (75%) of the recipients of the combination therapy were free of diabetes for the duration of the monitoring period, 8/8 (100%) of the animals given BM or Ig-GAD2 alone returned to diabetic BGL by day 25 post treatment and remained diabetic until the last day (day 70) of monitoring (Fig. 5A, right panel). Histological analyses confirm the clinical data as Ig-GAD2+BM recipient mice had mostly islets with no (no insulitis) or reduced (Peri) infiltration while the islets from recent diabetic mice displayed mild (20%) or severe (80%) infiltration (Fig. 5B). Also, the combination therapy restored the pancreatic β-cell mass which was comparable to healthy mice (Fig. 5C). Thus, mice undergoing genetic labeling with YFP recover from T1D upon treatment with Ig-GAD2+BM combination therapy. Analysis of immune modulation driven by Ig-GAD2+BM therapy shows a low frequency of IL-17-producing Th17 cells in both the spleen and pancreatic lymph nodes (Fig. 5D, bottom panel). On the other hand, Ig-GAD2+BM-treated mice had significantly more IFNγ-producing Th1 cells in both the spleen and pancreatic lymph nodes than the diabetic-untreated cohorts (Fig. 5D, top panel). This is consistent with our previous report demonstrating that recovery from T1D parallels with decline of IL-17-producing Th17 cells in lymphoid organs [15, 16] while Th1 cells remain elevated but innocuous [15]. These findings suggest that immune modulation driven by Ig-GAD2+BM is operative in mice undergoing genetic labeling and the underlying mechanism is similar to wild type NOD mice [16].

Figure 4.

Treatment with tamoxifen induces effective genetic labeling of insulin-producing β-cells with yellow fluorescent protein (YFP). (A) Shows a schematic representation of the generation of NOD.RIP-CreER/ROSA26-YFP mice. In this mouse tamoxifen binding to ER induces specific expression of cre in insulin-producing β-cells. This leads to removal of loxP-flanked STOP codon in front of YFP allowing for β-cell labeling. (B) Shows representative confocal microscopy images of pancreatic section stained for INS (red), YFP (green) and merged INS/YFP/Dapi in mice recipient of zero (0), three (3), or seven (7) mg tamoxifen. White scale bar, 10 μm. The bar graphs represent the number of cells per islet that are INS⁺, YFP⁺ or INS⁺YFP⁺. The data represent the mean number ± SEM of cells per islet calculated from at least 120 islets in each group of mice (n=6). ***p < 0.0005 as determined by one-way ANOVA. (C) Shows the mean ± SEM weekly blood glucose levels (BGL) in female wild type NOD and NOD.RIP-CreER/ROSA26-YFP mice (7 per group) starting at week 10 of age.

Figure 5.

Treatment of diabetic β-cell reporter mice with Ig-GAD2+BM induces immune modulation that supports recovery from disease. (A) Diabetic NOD.RIP-CreER/ROSA26-YFP mice recipient of insulin implants (8 per group) were treated with Ig-GAD2, BM, or Ig-GAD2+BM combination therapy as indicated in Figure 1 and monitored for BGL daily for 70 days. The left panel shows the mean ± SEM BGL while the right panel illustrates the percentage of recovery during the 70-day monitoring period. *p < 0.0005 to 0.05 as determined by two-tailed Student t test. (B) NOD.RIP-CreER/ROSA26-YFP mice were given tamoxifen at week six of age and one group (healthy, n = 4) was sacrificed upon completion of the tamoxifen regimen, another group (diabetic, n = 4) was sacrificed two days after diagnosis of T1D, and a third group was sacrificed after completing a 10-week Ig-GAD2+BM treatment and recovery from T1D (n = 7). Four to six non-serial pancreatic sections per mouse were then prepared and stained with hematoxylin-eosin and scored for insulitis severity according to the pattern of infiltration observed across the groups. These patterns include islets with no, peri, mild, and severe insulitis (left panel). The bars show the percent of islets with no (green), peri (purple), mild (blue), and severe (red) insulitis for healthy, diabetic, and Ig-GAD2+BM treated mice. At least 120 islets were counted for each group (C) Shows the mean mg β-cell mass ± SEM for the three groups of mice; healthy (n = 4), diabetic (n = 4) and Ig-GAD2+BM (n = 7). Four to six non-serial sections per mouse were stained and at least 120 islets were counted for each group. **p < 0.005 as determined by two-tailed Student t test. (D) The splenic (SP) and pancreatic lymph node (PLN) cells from the diabetic and Ig-GAD2+BM groups were stimulated with PMA and ionomycin, and stained for surface CD3 and CD4 as well as intracellular IFN-γ and IL-17. The mean percentage ± SEM of CD3⁺CD4⁺ T cells producing IFN-γ or IL-17 by individual mouse are illustrated in top and bottom panels, respectively. *p < 0.05, **p < 0.005 as determined by two-tailed Student t test.

3.3. Newly formed β-cells during recovery from T1D originate from residual and β-cell precursors

During restoration of β-cell mass and recovery from T1D, the newly formed β-cells may develop from division of residual β-cells and/or differentiation of β-cell precursors. To address this question, 6-week-old healthy RIP-CreER/ROSA26-YFP NOD mice were first subjected to YFP genetic labeling of β-cells by treatment with tamoxifen. Subsequently, upon diagnosis of overt T1D, the mice were enrolled in Ig-GAD2+BM combination therapy and monitored for long-lasting recovery from disease. On day 70 (week 10) of treatment the pancreata were harvested and the islets were assessed for YFP and insulin expression. Confocal microscopy analysis shows three different types of islets (Fig. 6). The first type of islets contained equal numbers of insulin- and YFP-expressing cells that colocalize suggesting that these islets harbor only double-positive (INS⁺YFP⁺) β-cells (Fig. 6A). The second type of islets comprised only insulin-positive cells which were negative for YFP (INS⁺YFP^-) (Fig. 6B). The third type of islets contained double-positive (INS⁺YFP⁺) as well as insulin-single positive (INS⁺YFP^-) cells (Fig. 6C). Both YFP⁺ (Fig. 6D) and YFP⁻ (Fig.6E) β-cells stained positive for Ki-67 marker indicating that the cells are proliferative. The proliferative trait of both types of β-cells is uniform as all islets containing insulin-producing β-cells had cells staining positive for Ki-67 (Fig. 6 D and E, bar graphs). All together, these observations indicate that YFP⁺β-cells emanate from residual β-cells that survived the immune attack while YFP^-β-cells originate from precursor cells that were not subject to immune attack as they were not producing insulin and had no mature β-cell phenotype.

Figure 6.

Newly formed β-cells originate from both residual and β-cell precursors during recovery from T1D. (A-E) Diabetic NOD.RIP-CreER/ROSA26-YFP reporter mice that have undergone tamoxifen regimen at 6 weeks of age were implanted with insulin pellets upon T1D diagnosis and treated with Ig-GAD2+BM combination therapy as in Figure 1. On week 10 of treatment and recovery from T1D, the mice were sacrificed and their pancreatic islets were assessed for expression of insulin, YFP and Ki-67 by confocal microscopy. White scale bar, 10 μm. (A-C) the images depict a representative islet that harbors (A) INS⁺ YFP⁺, (B) INS⁺ YFP⁻, and (c) both INS⁺YFP⁺ and INS⁺YFP⁻ β-cells. The bar graphs show the mean number ± SEM of total INS⁺ and YFP⁺ cells per islet in comparison to the number of INS⁺YFP⁺ and INS⁺YFP⁻ cells in the same islet. (D, E) The images depict a representative islet stained for insulin and Ki-67. The bar graphs show the mean number ± SEM of total INS⁺ and INS⁺Ki67⁺ cells. The results are based on four to six nonserial pancreatic sections from six mice. At least 150 islets and 7400 INS⁺ cells were counted for each panel. *p < 0.05, **p < 0.005 as determined by two-tailed Student t test.

3.4. Β-cells of precursor origin develop only during recovery from T1D

The development of β-cells from precursor cells during recovery from disease may represent a safe guard mechanism for circumstances where insulin-producing β-cells could not survive the immune attack. To test this premise, healthy RIP-CreER/ROSA26-YFP NOD mice were treated with tamoxifen and their pancreatic islets were assessed for YFP-labeled β-cells before diagnosis of disease (Healthy), at the onset of T1D (Diabetic) and upon recovery from disease by treatment with Ig-GAD2+BM therapy. The results show that mice recipient of Ig-GAD2+BM therapy, like healthy animals, had relative to diabetic mice, significant number of islets per section that encompass insulin-producing β-cells with positive stain for YFP (INS⁺YFP⁺) (Fig. 7A, left panel). Interestingly, the mice treated with Ig-GAD2+BM had insulin-producing cells that are negative for YFP (INS⁺YFP⁻) while healthy mice like diabetic animals showed no detectable INS⁺YFP⁻ β-cells (Fig. 7B, right panel). Similar patterns were observed with all islets as INS⁺YFP⁺ β-cells were detectable in both healthy and Ig-GAD2+BM treated mice, while INS⁺YFP⁻ β-cells arose only in Ig-GAD2+BM treated mice (Fig. 7B). Again, when the analysis was made at the cell level, while both healthy and Ig-GAD2+BM treated mice had INS⁺YFP⁺ double-positive β-cells, only Ig-GAD2+BM recipient mice showed INS⁺YFP⁻ single-positive β-cells (Fig. 7C). These results suggest that newly formed β-cells during recovery from T1D emanate from both residual and precursor cells. INS⁺YFP⁻ cells could have theoretically come from β-cells that did not label with YFP because they were not sensitive to tamoxifen. This is unlikely, because the regimen was devised to label all insulin-producing β-cells (Fig. 4) and a second exposure to tamoxifen during treatment with Ig-GAD2+BM induces YFP labeling of all insulin-producing β-cells suggesting that INS⁺YFP⁻ cells are indeed sensitive to tamoxifen (Fig. 7D and E).

Figure 7.

Β-cells of precursor origin form only in Ig-GAD2+BM treated mice. (A-C) NOD.RIP-CreER/ROSA26-YFP mice were given tamoxifen at week six of age and one group was sacrificed upon completion of the tamoxifen regimen (healthy), another group was sacrificed two days after diagnosis of T1D (Diabetic), and a third group was sacrificed after completing a 10-week Ig-GAD2+BM treatment and recovery from T1D (Ig-GAD2+BM). Pancreatic sections were then analyzed for insulin and YFP expression by confocal microscopy. (A, B) show the mean number ± SEM of islets (A) per section and (B) per mouse that are INS⁺ and YFP⁺ (left panel) or YFP⁻ (Right panel) among all 3 groups of mice. (C) shows the mean number ± SEM of INS⁺YFP⁺ (Left panel) and INS⁺YFP⁻ (right panel) cells per mouse among all 3 groups of mice. *p < 0.05, **p < 0.005, ***p < 0.0005 as determined by two-tailed Student t test (left panels) and one-way ANOVA (right panels). (D, E) Two groups of NOD.RIP-CreER/ROSA26-YFP mice were given tamoxifen at week six of age and upon diagnosis of T1D they were treated with Ig-GAD2+BM for 10 weeks as in Figure 1. One week later, one group of mice received a second tamoxifen regimen (2X TM) while the other group did not (1X TM). On week 12, both groups were sacrificed and their pancreata were analyzed for insulin (red) and YFP (green) expression by confocal microscopy. (D) the upper panel (1X TM) depicts a representative islet that harbors only INS⁺ YFP⁻ and the lower panel (2X TM) depicts a representative that contains only INS⁺ YFP⁺ (2X TM). White scale bar, 10 μm. (E) Shows the mean number ± SEM of islets per section with INS⁺YFP⁺ and INS⁺YFP⁻ cells. The results are based on four to six nonserial sections per pancreas from three to six mice. At least 80 islets were counted for each group. **p < 0.005 as determined by two-tailed Student t test.

4. Discussion

This report demonstrates that antigen-specific/EPC combination therapy, which eradicates the inflammatory immune cells from the pancreatic islets [16], repairs the islet-vascular network and enables both residual and β-cell precursors to regenerate, thrive and support long-lasting recovery from established T1D. In a prior study we have shown that while Ig-GAD2 drives eradication of Th1 and Th17 pathogenic cells from the pancreas, the exogenous EPCs migrate to the pancreatic islets and mature into ECs [16]. This work demonstrates that the immune attack against β-cells causes collateral damage to the islet vascular network. The EPCs from the bone marrow give rise to ECs that repair the islet vasculature; which explains the requirement for healthy EPCs to overcome T1D [16] as the differentiation and function of endogenous EPCs are impaired in diabetics [21].

Recovery from T1D is dependent on clearance of immune cells from pancreatic islets [17]. This work show that the recovery is also dependent on the repair of the islet vascular network which is necessary for formation of new β-cells. It has previously been shown that pre-existing β-cells can renew by self-duplication following mechanical pancreatic injury [22]. The findings here show that residual pancreatic β-cells that survive the immune attack can renew, thrive and support recovery from spontaneous autoimmune T1D. The renewal of residual β-cells is most likely related to restoration of microenvironment integrity which supports symbiotic interplay between β-cells and endothelial cells [23, 24]. Surprisingly however, a few healthy islets were observed where the newly formed β-cells had no YFP labeling suggesting that these cells did not originate from residual β-cells. It has previously been shown that mechanical injury of the pancreas in healthy mouse strains that are not prone to T1D, new β-cells can form as a result of beta-cell progenitor proliferation [25]. It is therefore, possible that the newly formed YFP-negative cells originate from β-cell precursors. Forced trans-differentiation from alpha (α) [26] or exocrine cells [27] to β-cells is unlikely in this case because no genetic modification of specific transcription factors was applied. Natural conversion from α cells is, however, plausible [28] but remains to be assessed in the NOD model where the loss of β-cells is progressive with an ongoing autoimmune inflammation rather than extreme targeted ablation. The fact that both residual and precursor cells contribute to the formation of new β-cells represents a significant clinical value to therapy of T1D as intervention would perhaps extend beyond the availability of residual β-cells. Furthermore, the combination therapy, although involves healthy EPCs, remains Ag-specific and targets diabetogenic T cells with limited impact on the function of the immune system [6, 17, 29].

In summary, while Ig-GAD2 clears the islets from cell infiltration, the EPC-derived ECs repair the islet vascular network. Under these circumstances, the microenvironment invigorates residual β-cells to regenerate and precursor cells to differentiate and give rise to additional new β-cells. The end result is restoration of β-cell mass and recovery from T1D. In all, this study pinpoints that Ig-GAD2+BM combination therapy against established T1D reverses the disease by promoting the formation of new β-cells from the duplication of residual β-cells and, unexpectedly from the differentiation of precursor cells. Although, it is unclear whether these precursors for β-cells represent unused reserve precursors or freshly committed stem cells, their maturation into insulin-producing β-cells upon repair of the islet vascular network is promising especially for T1D patients where the disease is usually diagnosed when there is little or no residual insulin-producing β-cells.

5. Conclusions

Ig-GAD2+BM combination therapy drive immune modulation and restores islet vascular integrity leading to the formation of new β-cells from duplication of residual β-cells and differentiation of β-precursors cells. This is promising for T1D patients as the disease is usually diagnosed when there is little or no residual insulin-producing β-cells and suggests that treatment of T1D would be possible long after diagnosis.

Highlights.

• T1D causes collateral damage to the islets vascular network.

• Bone marrow-derived EPCs mature into ECs that restore islet blood vessel integrity

• Restored vasculature invigorates the formation of new β-cells

• New β-cells emanate from residual and β precursor cells

Abbreviations:

7-AAD

7-aminoactinomycin D

Ag

antigen

BGL

blood glucose level

BM

bone marrow

CreER

Cre recombinase fused to an estrogen receptor

EC

endothelial cell

EPC

endothelial precursor cell

Ig-GAD2

immunoglobulin with genetically incorporated glutamic acid carboxylase sequence 206–220

INS

insulin

PDGFRβ

platelet-derived growth factor receptor beta

PECAM1

platelet endothelial cell adhesion molecule 1

PLN

pancreatic lymph node

RIP

rat insulin promoter

SP

spleen

TM

tamoxifen

T1D

type 1 diabetes

YFP

yellow fluorescent protein