Induction of a regulatory response by xenogeneic SC in diabetic mice. (A) Cytofluorometric analyses of Foxp3, CTLA-4, and GITR expression in CD4⁺ cells from spleens (left) or PLN (right) of mice (n = 3) treated with EC or SC for 1–3 wk. Isotype controls were included in all assays for normalization, and dot plots were gated on CD4⁺ cells. Preimplant percentages of double-positive cells for each marker (at week 0) are shown at the top. Line graphs depict fold changes (mean and SD from three experiments) in normalized marker expression from SC-treated mice relative to control counterparts on EC treatment (in which fold change = 1). Error bars indicate SD. *, P < 0.05–0.01. (B) The peri-islet infiltrate observed by immunofluorescence analysis in successfully treated NOD mice consisted of CD4⁺ cells, of which ∼20% expressed Foxp3, the transcriptional factor associated with T reg cells. Islets from normoglycemic NOD mice were characterized by intense lymphocytic peri-islet infiltrates, which is consistent with a nonprogressive insulitis (not depicted). Arrows indicate double-positive cells (>50% peri-insular Foxp3/CD4 cells, as opposed to <2% intrainsular in EC-treated controls). Bars, 50 µm.

Effects of SC therapy on cytokine secretion profile of Th cells from diabetic mice. (A) Sorted CD4⁺ T cells from spleens (Sp) or PLNs, as well as unfractionated pancreatic-infiltrating leukocytes (PIL), were assayed for cytokine release in response to 1 µg/ml anti-CD3 after SC or control EC therapy. Data are means ± SD from four experiments. *, P < 0.001–0.01; **, P < 0.05. (B) Intracellular TGF-β and IL-17A staining profiles of CD4⁺CD25⁺ and CD4⁺CD25⁻ cells from SC-treated NOD mice upon in vitro stimulation with proinsulin-related P3UmPI or control OVA. Percentages of double-positive cells are shown in the top right quadrants. One experiment is shown representative of three.

IDO expression by SC is instrumental in their therapeutic efficacy. (A) In vitro differentiation of SC progressively increases mRNA putatively encoding porcine IDO. SCs were cultured for 3, 5, or 7 d, and IDO-encoding mRNA was analyzed by real-time PCR with β-actin transcript normalization. Data (mean values and SD from four independent experiments) are presented as fold change in normalized transcript expression in cultured SC relative to freshly harvested cells (in which fold change = 1; dashed line). (B) Detection of predicted IDO protein by immunoblot with rabbit anti–mouse or mouse anti–human IDO. Freshly harvested SCs (day 0), as well as cells cultured in vitro for 3 or 7 d, were analyzed using β-actin normalization. SCs were also recovered from successfully treated mice (ex vivo) and assayed with the anti–human reagent. Indo-transfected (+) and mock-transfected (−) P1.HTR cells were assayed in combination with anti–murine IDO as the respective positive and negative controls. Human HeLa cells, treated (+) or not (−) with IFN-γ (a transcriptional inducer of INDO), were a second pair of controls used in combination with the anti–human reagent. Note that with both reagents the mature form of the protein was detected as a doublet (this has been described for human IDO as well). One experiment is shown representative of three. (C) Functional IDO expression by either freshly harvested SC (day 0) or cells cultured in vitro for 3 or 7 d. SCs were also recovered from successfully treated mice (ex vivo) and assayed for functional IDO expression. Treatments included cell exposure in vitro to the IDO inhibitor 1-methyl-tryptophan (1-MT) or silencing IDO-encoding transcripts by siRNA technology. Nc siRNA represents the negative control treatment. Enzyme activity (means ± SD; n = 4) was measured in terms of the ability of SC (3 × 10⁶/ml) to metabolize tryptophan to kynurenine, measured by high-performance liquid chromatography. (D) Kaplan-Meier plot for survival in diabetic mice receiving SC with silenced IDO-encoding mRNA expression. Control indicates diabetic mice on EC treatment. Nc, negative control. P < 0.001 for comparison between the control and IDO siRNA treatment groups. One experiment is shown representative of three.

Effects of treatment with encapsulated SC on disease progression and pancreatic histology in diabetic NOD mice. (A) Kaplan-Meier plot for survival. Diabetic NOD females were treated with an implant of EC or SC, and survival probability was plotted over time. Data are from 18 and 21 animals that received EC or SC. P < 0.001 for comparison between the two treatment groups. (B) Blood glucose concentrations in three NOD recipients successfully treated with SC (for mouse ID, see Table I). (C) Pancreatic histology. Three mice (Numbers 2, 7, and 15) were killed 18 wk after grafting. Sections of the pancreata were stained with hematoxylin and eosin, showing both islets (mostly periductal; indicated by arrows) free of infiltrate and islets with innocuous-looking infiltrates and no signs of invasive insulitis. Prediabetic and age-matched diabetic controls are shown for comparison (indicated). Bars, 100 µm. (D) Islets from hyperglycemic control or SC-treated mice (n = 3) were scored at different weeks after transplant (on time 0), and percentages were recorded that represent numbers of islets of a given score over total number of islets (30–40 per pancreas). Control mice (top) were either sham-transplanted or transplanted with EC. Each experiment was repeated with reproducible results two to four times and one representative experiment is shown in each panel.

In diabetic NOD mice, SC therapy causes reduced expression of the ROR-γt–encoding gene and augmentation of Foxp3, with negligible effects on Tbet and Gata3. Rorc, Foxp3, Tbet, and Gata3 transcripts were evaluated in CD3⁺ T cells from PLN of mice treated with EC or SC for 3–4 wk. 5 × 10⁵/ml of sorted CD3⁺ lymphocytes were activated with 1 µg/ml anti-CD3 for 24 h. Rorc, Foxp3, Tbet, and Gata3 mRNAs were quantified by real-time PCR using Gapdh normalization. Data (means ± SD from three experiments) are presented as fold change in normalized transcript expression in mice grafted with EC or SC relative to prediabetic 4-wk-old controls (in which fold change = 1; dotted line).

Induction but not effector phase of protection by SC xenografts requires TGF-β and is cell transferable to NOD-SCID recipients. (A) 10⁶ 8–12-wk-old NOD/BDC2.5 splenocytes were injected intravenously alone (control [C]) or in combination with 10⁶ sorted PLN CD4⁺ cells from NOD mice implanted with EC (EC/CD4) or SC (SC/CD4) for 3 wk. Groups of SC/CD4–treated mice were given anti–TGF-β (mouse IgG2b 1D11), anti–CTLA-4 (hamster IgG 4F10; either antibody at 0.5 mg/mouse twice weekly for 3 wk), or the respective control antibodies (not depicted). The incidence of diabetes in the NOD-SCID recipients is indicated over time for the different groups (n = 10), as revealed by glycosuria and hyperglycemia (P < 0.001, SC versus EC or no treatment). (B) Protection from diabetes was also studied by transfer of splenocytes (5 × 10⁶) from diabetic NOD/Mrk mice injected alone (control [C]) or in combination with 10⁶ sorted PLN CD4⁺ cells from NOD mice implanted with EC (EC/CD4) or SC (SC/CD4) for 3 wk. Additional groups included anti–TGF-β or anti–CTLA-4 treatments as in A. The incidence of diabetes in the NOD-SCID recipients is indicated over time for the different groups (n = 10), as revealed by glycosuria and hyperglycemia (P < 0.001, SC versus EC or no treatment). (C) The regulatory response initiated by implanted SC requires TGF-β for induction in the primary host. NOD recipients of protective SC implants were treated with TGF-β–neutralizing monoclonal 1D11 (0.5 mg/mouse twice weekly for 3 wk) or the isotype control, and the incidence of diabetes was monitored over time (n = 10). Additional experimental groups received treatments (0.5 mg/mouse, twice weekly for 3 wk) with anti–IL-10 (JES5.2A5; rat IgG1), anti–CTLA-4 (4F10), or anti-GITR (DTA-1; rat IgG2b; in alternative, rat/hamster control IgG antibodies; not depicted). P < 0.001, for anti–TGF-β versus control treatment. The percentages of pancreatic Foxp3⁺ cells in the CD25⁺ fraction were 35% in the EC group, 62% in SC-treated mice, and 29% in the latter group on additional anti–TGF-β. (D) Purified CD4⁺CD25⁺ T cells from SC-treated NOD mice were cultured for 3 d with fusion protein–pulsed splenic CD11c⁺ cells as antigen-presenting cells (a mixture of proinsulin-related P3UmPI, GAD-related P3UmG, and IA-2-related P3UhIA). The recovered T cells were assayed (at the different regulatory/effector cell ratios indicated) for suppression of 72-h proliferation in CFSE-labeled CD4⁺CD25⁻ cells from diabetic donor mice. In selected cultures, control (mouse IgG2b) or anti–TGF-β antibody was added. Controls included CFSE-labeled CD4⁺CD25⁻ cells cultured in the absence of the regulatory fraction and/or anti-CD3. The percentages of red-marked proliferating cells are shown. The corresponding percentages of control CD4⁺CD25⁻ cells cultured with OVA-primed CD4⁺CD25⁺ T cells from SC-treated mice or with P3UmPI/P3UmG/P3UhIA–primed CD4⁺CD25⁺ T cells were consistently in the 85–90% range (see also Fig. S3 B). One experiment is shown representative of three.

Gene expression pattern and insulin production in islets from successfully treated mice. (A) Real-time PCR analysis of pancreatic gene expression in diabetic mice treated with EC or SC for 2 wk. SCs were treated with either IDO siRNA or negative control (nc) siRNA. mRNA was extracted from digested pancreata and genes encoding factors were examined that initiate (Ngn3) or contribute to islet cell differentiation. Data (mean values ± SD from three experiments) are presented as fold change in normalized transcript expression in pancreata from diabetic mice on EC or SC therapy relative to prediabetic 4-wk-old controls (in which fold change = 1; dashed lines). (B) Immunoblot analysis of Ngn3 expression in diabetic mice grafted with EC or SC for different times. Pancreas tissue extracts from age-matched untreated mice (diabetic), as well as transplanted hosts over time, were reacted with Ngn3 A-19 affinity-purified goat polyclonal antibody, using β-actin normalization. Prediabetic mice were included as a control (indicated as such), together with sham-transplanted mice at 2 wk (there were no survivors at 16 wk among the EC recipients). One experiment is shown representative of three. Using scanning densitometry analysis and mean values from the three experiments, the differences between the EC and SC treatment groups were significant at both 2 wk (P = 0.006) and 4 wk (P = 0.009). (C) Immunostaining for insulin, glucagon, and somatostatin in islets from three successfully treated mice. Frozen pancreatic sections from diabetic controls and SC-treated mice (indicated) were post-fixed with formaldehyde, and fluorescent images, in which nuclei had been stained with DAPI (blue), were acquired by light microscopy. Bars, 50 µm. One experiment is shown representative of three.