Research ArticleType 1 Diabetes

Mixed Chimerism and Growth Factors Augment β Cell Regeneration and Reverse Late-Stage Type 1 Diabetes

See allHide authors and affiliations

Science Translational Medicine  09 May 2012:
Vol. 4, Issue 133, pp. 133ra59
DOI: 10.1126/scitranslmed.3003835

Abstract

Type 1 diabetes (T1D) results from an autoimmune destruction of insulin-producing β cells. Currently, islet transplantation is the only curative therapy for late-stage T1D, but the beneficial effect is limited in its duration, even under chronic immunosuppression, because of the chronic graft rejection mediated by both auto- and alloimmunity. Clinical islet transplantation is also restricted by a severe shortage of donor islets. Induction of mixed chimerism reverses autoimmunity, eliminates insulitis, and reverses new-onset but not late-stage disease in the nonobese diabetic (NOD) mouse model of T1D. Administration of gastrin and epidermal growth factor (EGF) also reverses new-onset but not late-stage T1D in this animal model. Here, we showed that combination therapy of induced mixed chimerism under a radiation-free nontoxic anti-CD3/CD8 conditioning regimen and administration of gastrin/EGF augments both β cell neogenesis and replication, resulting in reversal of late-stage T1D in NOD mice. If successfully translated into humans, this combination therapy could replace islet transplantation as a long-term curative therapy for T1D.

Introduction

Type 1 diabetes (T1D) results from destruction of insulin-secreting islet β cells by autoimmunity, which arises from defects in both central and peripheral immune tolerance (1). Therapies such as administration of anti-CD3 to restore the peripheral balance of effector T and regulatory T cells have been shown to ameliorate new-onset T1D in mouse models (2) and in humans (3), but the therapeutic effect in humans was transient (4). Similar transient effects were observed with depletion of B cells by administration of anti-CD20 (5). This situation has led to the recent proposition that therapies that can simultaneously correct multiple defects in T1D individuals will be required for efficient long-term reversal of T1D (4).

Currently, induction of mixed chimerism is the only approach that has been found to simultaneously correct multiple defects in T1D nonobese diabetic (NOD) mice, including restoring both central and peripheral tolerance (69). In non-autoimmune humans, Strober and colleagues have induced mixed chimerism via conditioning with total lymphoid irradiation and anti-thymocyte globulin in the context of transplantation of human leukocyte antigen (HLA)–matched donor T and hematopoietic cells (HCs); this protocol resulted in immune tolerance to kidney transplants from the HC donor without causing graft-versus-host disease (GVHD) (10). These data demonstrate the feasibility for induction of mixed chimerism in humans; however, HLA-matched mixed chimerism may be unable to reverse autoimmunity, because studies with autoimmune mouse models indicate that reversal of autoimmunity requires major histocompatibility complex (MHC)–mismatched mixed chimerism (6). Indeed, induction of mixed chimerism with HLA-matched BM was not able to reverse autoimmune systemic lupus in a patient (11). Moreover, although induction of mixed chimerism with MHC-mismatched BM transplants has been shown to eliminate insulitis, augment residual β cell proliferation, and reverse new-onset diabetes in NOD mice (7), induction of mixed chimerism alone was not able to reverse late-stage T1D (7, 9).

Thus, cure of late-stage T1D may require both reversal of autoimmunity and β cell replacement (islet transplantation) or β cell regeneration. In patients, islet transplantation is currently the only therapy capable of reversing late-stage T1D (12). Unfortunately, under the current protocol of immunosuppression for preventing rejection, the therapeutic effect lasts only for about 3 years (13). This limited effect results from chronic rejection mediated by allo- and autoimmunity, toxicity of immunosuppressants, and a nonideal graft site (9, 13, 14). In addition, islets from two to three donors are required for one recipient to reach insulin independence under most current immunosuppressant protocols. Therefore, the severe shortage of donor islets, combined with the short-lived therapeutic effect, has somewhat dampened enthusiasm for islet transplantation as a curative therapy for late-stage T1D.

Alternatively, β cell regeneration has been considered an attractive approach. Studies with non-autoimmune mice suggest that β cell regeneration can occur via β cell replication (15), β cell neogenesis from pancreatic ductal epithelial progenitors or mesenchymal progenitors (1618), as well as transdifferentiation from acinar cells or α cells (19, 20). Pancreatic tissues from non-autoimmune humans were shown to give rise to β cells through neogenesis in vitro and in severe combined immunodeficient (SCID) mice (2123). However, it is unclear how β cells can be regenerated in NOD mice and humans with T1D due to the existence of autoimmunity.

In the current studies, we have developed a radiation-free anti-CD3/CD8 conditioning regimen that allows for induction of MHC-mismatched mixed chimerism without any signs of GVHD, when applied to NOD mice with late-stage T1D. The mixed chimerism, in combination with administration of gastrin and epidermal growth factor (EGF) that have been shown to augment both mouse and human β cell regeneration (2324), was able to augment β cell neogenesis and replication, resulting in reversal of late-stage diabetes in NOD mice. This approach may provide a replacement of islet transplantation as a long-term curative therapy for late-stage T1D.

Results

Combination therapy of induction of mixed chimerism under an anti-CD3/CD8 conditioning regimen and administration of gastrin and EGF resulted in reversal of late-stage T1D in NOD mice

We previously reported that both prediabetic and late-stage diabetic NOD mice can be induced to develop chimerism without GVHD by conditioning with anti-CD3/CD8 and then transplantation of donor BM and CD4+ T-depleted spleen cells (69). However, a cytokine storm syndrome caused by elevation of serum tumor necrosis factor–α (TNF-α) and interleukin-2 (IL-2) triggered by intact Fc receptor (FcR)–binding anti-CD3 (FB-anti-CD3) prevents the clinical application of this conditioning regimen (25, 26). Genetically modified FcR-nonbinding anti-CD3–IgG3 (immunoglobulin G3) (FNB-anti-CD3) caused little T cell activation and does not cause a cytokine storm (fig. S1A) (27). Unfortunately, conditioning with FNB-anti-CD3 alone did not allow induction of mixed chimerism (fig. S2A). Serum levels of TNF-α and IL-2 usually peaked at 1 to 2 hours after injection of FB-anti-CD3 (25). We found that injection of FNB-anti-CD3 12 hours before injection of FB-anti-CD3 and anti-CD8 was able to markedly reduce the peak serum levels of IL-2 and TNF-α triggered by FB-anti-CD3 (P < 0.01; fig. S1A). This was associated with FNB-anti-CD3–mediated internalization of T cell receptor (TCR)–CD3 complex and significant blockade of T cell activation triggered by FB-anti-CD3 (P < 0.01; fig. S1, B and C). Conditioning with the sequential injection of FNB-anti-CD3 and FB-anti-CD3/CD8 was still able to induce mixed chimerism with no GVHD in late-stage diabetic NOD mice as judged by no significant body weight changes (P > 0.1; fig. S2E).

Mixed chimerism alone or administration of gastrin/EGF alone reverses new-onset but not late-stage T1D in NOD mice (7, 9, 24, 28, 29). With the new conditioning regimen of sequential injection of FNB- and FB-anti-CD3, we tested whether combination therapy involving induction of mixed chimerism and administration of gastrin/EGF would reverse late-stage T1D in NOD mice. As shown in Fig. 1A, late-stage diabetic NOD mice, that is, 3 weeks after diagnosis of diabetes onset as defined in our previous publications (7, 9), were conditioned on day −8 and transplanted with BM and CD4+ T-depleted spleen cells from C57BL/6 donors on day 0. Starting on the following day, the BM recipients were injected daily intraperitoneally with gastrin (3 μg/kg) and EGF (1 μg/kg) for up to 60 days. This group was defined as the “Chimerism + GE” group. There were two control groups: “Chimerism + PBS,” where the BM recipients were injected daily with phosphate-buffered saline (PBS), and “Conditioning + GE,” where the mice were conditioned and received daily injections of gastrin and EGF. Blood glucose levels were measured in all groups every 3 days. Gastrin and EGF were withdrawn once recipients showed normal glycemia for 3 consecutive days. Before day 60, mice with blood glucose levels above 500 mg/dl (the maximal reading of our meter) were implanted with 1/2 insulin pellet. By 60 days, treatment of gastrin and EGF was withdrawn from all mice, and all implanted insulin pellets were removed. Mice with blood glucose levels above 500 mg/dl were injected subcutaneously with long-term insulin (1 U) every other day.

Fig. 1

Combination therapy with induction of mixed chimerism and administration of gastrin/EGF resulted in reversal of late-stage T1D. Late-stage diabetic NOD mice were given conditioning alone or induction of mixed chimerism. One day after BM transplantation, gastrin (3 μg/kg) and EGF (1 μg/kg) were intraperitoneally injected daily for 60 days. Insulin pellet or insulin (1 U) was given to the mice if blood glucose was higher than 500 mg/dl. (A) Experimental scheme. (B) Blood glucose levels of mice from the Conditioning + GE (n = 6), Chimerism + PBS (n = 6), and Chimerism + GE (n = 12) groups. (C) GTT blood glucose curve of each group and additional control NOR mice (n = 4 to 6). (D) Mean ± SE of insulin level before and after glucose injection (n = 4). (E) Representative staining patterns of insulin (brown) and insulin (red)/glucagon (green). One representative is shown of four replicate experiments, except the Conditioning + GE group, which is not available. (F) Mean ± SE of percentage of β cell surface (n = 4). (G) Ratio of insulin/glucagon mRNA expression at different time points after treatment and mean ± SE of five replicate experiments.

We found that about 60% (7 of 12) of mixed chimeric recipients from the Chimerism + GE group showed normal glycemia (below 200 mg/dl) for up to 150 days after 60 days of gastrin and EGF treatment. The recovered recipients had blood glucose fluctuation from 200 to 350 mg/dl for ~2 to 4 weeks before reaching normal glycemia (Fig. 1B and fig. S3). In contrast, none of the mice in the Conditioning + GE or Chimerism + PBS group showed reversal of hyperglycemia (P < 0.01; Fig. 1B). These results indicate that the combination therapy involving induction of mixed chimerism and administration of gastrin and EGF can reverse late-stage T1D.

β Cell regeneration and increase of insulin sensitivity contributed to reversal of late-stage diabetes in mixed chimeric recipients after gastrin/EGF treatment

Next, we measured the insulin secretion capacity of the mice from different groups 150 days after treatment using a fasting glucose tolerance test (GTT). As shown in Fig. 1C, the peak level and recovery rate of the recipients that had normal glycemia from the Chimerism + GE group were significantly improved compared to recipients that still have hyperglycemia from the same group (P < 0.01). The peak levels and recovery rate of the recipients that had hyperglycemia from the Chimerism + GE group were similar to that of recipients from the Chimerism + PBS group, but the peak levels of blood glucose of the two groups were significantly lower and the recovery rate was significantly higher compared to that of the Conditioning + GE group (P < 0.01).

Furthermore, serum levels of insulin before and 10 min after glucose injection were compared. As shown in Fig. 1D, serum levels of insulin from all mice before glucose injection were similar (~0.2 ng/ml). However, compared with the levels before glucose injection, serum insulin levels of recipients that had normal glycemia from the Chimerism + GE group were increased by 3.5-fold (P < 0.01); the levels of the recipients that still had hyperglycemia from the same group were increased by 1.9-fold (P < 0.01); and the levels of recipients from the Chimerism + PBS group were increased by 1.5-fold (P < 0.05). However, levels of mice from the Conditioning + GE group were not significantly increased (Fig. 1D). Thus, this combination therapy significantly augments the insulin secretion capacity of T1D mice compared to either single therapy.

Next, the pancreas of recipients in the aforementioned groups before and after treatment were stained for insulin-secreting β cells and measured for β cell surface as previously described (7). There were only a few insulin-secreting β cells among residual islet remnants that mainly or predominantly consisted of α cells in the pancreas of mice before treatment or from the Conditioning + GE or Chimerism + PBS group, and those remnants in the former two groups had severe infiltration (Fig. 1E). In contrast, there were islets that predominantly consisted of insulin-secreting β cells in the recipients with or without normalization of blood glucose from the Chimerism + GE group (Fig. 1E). The total β cell surface of mice before treatment and mice from the Conditioning + GE and Chimerism + PBS groups was similar and represented less than 0.005% of total tissue surface (Fig. 1F). In contrast, the β cell surface of recipients with or without normal glycemia from the Chimerism + GE group was increased by 10- and 6-fold, respectively (P < 0.01; Fig. 1F). Taken collectively, combination therapy of induction of mixed chimerism and administration of gastrin/EGF augments β cell regeneration in late-stage diabetic NOD mice.

Additionally, we compared the insulin-secreting capacity of the recovered mice from the Chimerism + GE group with that of age-, sex-, and weight-matched normal NOR mice. As shown in Fig. 1, C and D, although both groups had peak blood glucose level of 250 mg/dl 15 min after injection of glucose, the former had reduced recovery at 30 and 60 min after glucose injection (P < 0.05). Consistently, the serum insulin levels of the former group were somewhat lower than those of the latter group (0.80 ± 0.17 ng/ml versus 0.96 ± 0.08 ng/ml), although not significantly different (P > 0.1). Finally, the percentage of β cell surface of mice from the Chimerism + GE group was 0.048 ± 0.005%, and the latter was 0.194 ± 0.034% (mean ± SE, n = 4). The former is ~25% of the latter. This is consistent with a previous theory that ~20% recovery of total β cell mass can maintain normal glycemia after extreme β cell loss (20). This may be because β cell regeneration requires triggering by increased blood glucose levels (30). Once blood glucose reaches normal, β cell mass will not expand further, even when the β cell mass is still much lower than the original. These results also indicate that the recovered late-stage T1D recipients have less of a reservation of insulin secretion capacity compared to normal mice.

Next, we analyzed the kinetics of β cell regeneration by comparing the ratio of insulin mRNA versus glucagon mRNA of isolated islets at different time points after treatment, as previously described (9). Reliability of the method was first confirmed with islets from C57BL/6, NOD-SCID, and new-onset and late-stage diabetic NOD mice (fig. S4). Islets from the pancreas of late-stage diabetic mice before treatment (day 0) and from the Conditioning + GE, Chimerism + PBS, and Chimerism + GE groups were isolated at 15, 30, and 60 days after treatment. As shown in Fig. 1G, before treatment (day 0) and 15 days after treatment, there was no difference among the groups, and the insulin/glucagon ratio was less than 0.01. By 30 days after treatment, although the insulin/glucagon ratio in the Conditioning + GE and Chimerism + PBS groups was still less than 0.01, the insulin/glucagon ratio in the Chimerism + GE group reached about 0.3. By 60 days after treatment, the insulin/glucagon ratio in the Conditioning + GE and Chimerism + PBS groups was ~0.01 and 0.02, respectively, but the insulin/glucagon ratio in the Chimerism + GE group had reached ~0.9 (P < 0.01; Fig. 1G). This represented a 90- or 45-fold increase compared to the Conditioning + GE or Chimerism + PBS group. These results indicate that β cell regeneration in the late-stage diabetic NOD mice receiving the combination therapy is a slow and time-dependent process.

There was little increase in insulin-secreting capacity, β cell surface, or insulin/glucagon ratio in mice from the Chimerism + PBS group compared to mice from the Conditioning + GE group (Fig. 1, D, F, and G), but the former had significantly reduced peak levels of blood glucose and significantly increased recovery rate (Fig. 1C). Thus, we compared the insulin sensitivity in these two groups by an insulin sensitivity test. Insulin sensitivity was significantly improved in the Chimerism + PBS group compared with the Conditioning + GE group (P < 0.01; Fig. 2A). Anti-insulin autoantibody was reported to be associated with a reduction of insulin sensitivity (31). Consistently, mice from the Conditioning + GE group had high levels of anti-insulin in their serum, but this autoantibody was not detectable in the mice from the Chimerism + PBS group (P < 0.01; Fig. 2B). Reversal of autoimmunity by induction of mixed chimerism was further supported by elimination of sialitis, an autoimmune property of NOD mice, in chimeric late-stage diabetic NOD mice (Fig. 2C). Together, increased insulin sensitivity by eliminating autoimmunity and anti-insulin autoantibody also contribute to normalization of blood glucose in chimeric NOD recipients, although it may not be a major contributor.

Fig. 2

Induction of mixed chimerism improved insulin sensitivity by eliminating autoimmunity. One hundred fifty days after treatment, insulin sensitivity test was applied to Conditioning + GE and Chimerism + PBS groups before salivary gland and serum were harvested. (A) Percentage of change of glucose baseline (n = 6). (B) Mean ± SE of anti-insulin antibody levels in serum by enzyme-linked immunosorbent assay (n = 8 to 10). OD, optical density. (C) Representative hematoxylin and eosin staining patterns of salivary gland (n = 6).

Induction of mixed chimerism that reverses autoimmunity allowed newly generated β cells to survive and accumulate

Next, we dissected the mechanisms of β cell regeneration under the combination therapy. Because newly generated β cells are reported to be more susceptible to inflammatory cytokine-induced apoptosis (32), we first tested whether induction of mixed chimerism allowed newly generated β cell survival. Because there were so few residual β cells left in the late-stage diabetic NOD mice, and β regeneration process was so slow in those mice after the combination therapy, it would be difficult to directly test the hypothesis using late-stage diabetic NOD mice after the combination therapy. Because partial pancreas duct ligation (PDL) triggered rapid β cell regeneration in non-autoimmune mice within 2 weeks (17), we instead tested whether induction of mixed chimerism that reverses autoimmunity augmented newly generated β cell survival and accumulation in late-stage diabetic NOD mice after PDL treatment.

Accordingly, late-stage diabetic NOD mice were induced to develop mixed chimerism or treated with conditioning only. Two weeks after treatment, PDL was applied. Surprisingly, PDL did not result in a significant increase in β cell surface in either group. There were only small insulin+ clusters among the islet remnants, and no significant difference was observed between the two groups (fig. S5A). There was no difference between the two groups in Ngn-3 mRNA expression, which is associated with β cell progenitor expansion in pancreatic tissues (17) (fig. S5B). These data suggest that there is nearly no β cell regeneration in late-stage diabetic NOD mice after induction of mixed chimerism, even in combination with PDL treatment.

Alternatively, we applied PDL to prediabetic NOD mice 2 weeks after induction of mixed chimerism or treated with conditioning alone. We compared the insulitis and β cell surface 3 and 14 days after PDL in the pancreas head and the ligated pancreas tail. Three days after PDL, there was no significant difference between β cell surface in the head area of mice from the two groups, but β cell surface in the ligated tail area of chimeric group was significantly higher than that of the conditioning group (P < 0.05; Fig. 3A). The islets of the conditioning group were severely infiltrated, but the islets of the chimerism group were infiltration-free (Fig. 3B, upper panels); the former islet β cells showed apoptosis as indicated by terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining (P < 0.05; Fig. 3B, lower panels). The frequency of TUNEL+ β cells appears to be very low (one to three apoptotic cells per islet), but it is consistent with a previous report (33). Fourteen days after PDL, islets were rarely seen in the tail area of the conditioning alone group, but the β cell surface in the ligated tail area of the chimerism group was markedly increased compared to its head area (P < 0.01; Fig. 3C). These results indicate that mixed chimerism that reverses autoimmunity allows the survival and accumulation of newly generated β cells.

Fig. 3

Induction of mixed chimerism allowed newly formed β cells to survive and accumulate in PDL-treated prediabetic NOD mice. Two weeks after BM transplantation, PDL was applied to prediabetic NOD mice in both “Conditioning” and “Chimerism” groups. Three and 14 days after PDL, pancreatic tissues from the head and ligated tail were stained for insulin. (A) Representative insulin staining patterns of the head and ligated tail 3 days after PDL. Mean ± SE of percent β cell surface of the head and ligated tail is shown (n = 4). (B) Representative insulin staining patterns of the ligated pancreas tail in prediabetic NOD mice from both groups 3 days after PDL. Representative staining patterns of 4′,6-diamidino-2-phenylindole (DAPI) (blue), insulin (red), and TUNEL (green) 3 days after PDL are shown below. Mean ± SE of TUNEL+insulin+ cells per islet is shown (n = 4). (C) Representative insulin staining patterns of the head and ligated tail 14 days after PDL. Mean ± SE of percent β cell surface of the head and ligated tail is shown (n = 4).

Administration of gastrin/EGF augmented both β cell neogenesis and replication in chimeric late-stage diabetic NOD mice

We next explored how gastrin/EGF augmented β cell regeneration in the mixed chimeric recipients. The major β cell regeneration pathways include replication and neogenesis (1517). We used [Rip-CreER; R26R-YFP] (Rip-YFP) NOD mice to trace the origin (replication versus neogenesis) of the regenerated β cells, as previously described (15). As shown in the diagram (Fig. 4A), Rip-YFP NOD mice were treated with tamoxifen to induce yellow fluorescent protein (YFP) expression and label the preexisting β cells. The YFP+ β cells derive from the replication of preexisting β cells, and the YFP β cells are de novo developed from β cell progenitors after tamoxifen treatment. We first tested the applicability of our β cell tracing assay by treating 10-week-old prediabetic NOD mice with tamoxifen that induces YFP expression and then treating the mice with PDL, which augments both replication and neogenesis.

Fig. 4

Both β cell neogenesis and replication arose in PDL-treated chimeric prediabetic Rip-YFP NOD mice. (A) Experimental scheme. (B and C) Prediabetic Rip-YFP NOD mice were given conditioning alone (B) or mixed chimerism was induced (C). Two weeks later, mice were given PDL treatment. Before and 14 days after PDL, the tail area of pancreatic tissues was stained for insulin (red), YFP (green), and DAPI (blue) to measure the percentage of YFP+insulin+ cells in each islet. One representative staining pattern of an islet is shown. The percentage of YFP+ cells among insulin+ cells in each islet of the mouse is also shown. There are four mice in each group. In (C), the bottom three rows represent islets with 90, 40, and 3% of YFP+ β cells, respectively. (D) The parallel slide of islets with 90 and 3% YFP+ β cells from PDL-treated chimeric recipients in (C) was stained for insulin (red), glucagon (green), and DAPI (blue). Representative staining patterns of the islets are shown (n = 4).

We observed that in the control mice receiving conditioning alone, more than 70% of β cells in each islet were YFP+ before PDL treatment (Fig. 4B, upper row), consistent with a previous report in non-autoimmune mice (15). Fourteen days after PDL, however, most islets disappeared; there were only small insulin-secreting β cell clusters that consisted almost entirely of YFP β cells among severe inflammatory infiltration (Fig. 4B, lower row). This is consistent with our observation in Fig. 3 that newly generated β cells cannot survive or accumulate in autoimmune NOD mice. Next, we performed this test again in chimeric prediabetic NOD mice. Consistently, most β cells in the islets were YFP+ in the chimeric mice before PDL treatment (Fig. 4C, upper row). After PDL treatment, ~60% of islets had more than 70% YFP+ β cells, ~30% of islets had 10 to 50% of YFP+ β cells, and ~10% of islets had almost no YFP+ β cells (Fig. 4C, lower rows). Furthermore, the islets with mostly YFP+ β cells consisted of either insulin or glucagon single-positive cells (Fig. 4D, upper row), but the islets with mostly YFP β cells consisted of insulin or glucagon single-positive as well as double-positive cells (Fig. 4D, lower row). The presence of insulin+glucagon+ cells indicates that the YFP islets are from neogenesis.

Therefore, using Rip-YFP NOD mice and tamoxifen-induced YFP expression, we are able to distinguish β cells derived from replication or from neogenesis. After PDL stimulation, nonchimeric prediabetic NOD mice have the capacity for β cell neogenesis, but the newly formed β cells cannot survive because of ongoing autoimmunity. After induction of mixed chimerism and reversal of autoimmunity, the newly generated β cells from both replication and neogenesis can form new islets in prediabetic NOD mice upon PDL stimulation. This further supports our observation that elimination of insulitis is required to allow β cell regeneration in autoimmune individuals.

Next, we determined the role of gastrin/EGF in β cell regeneration in late-stage diabetic Rip-YFP NOD mice after induction of mixed chimerism. As mentioned above, the late-stage diabetic NOD mice were defined as 3 weeks after diabetes onset. Thus, 1 day after diagnosis, the mice were injected with tamoxifen to label the preexisting β cells. Three weeks after onset, the mice were induced to develop mixed chimerism and injected with gastrin/EGF for up to 60 days, as described in Fig. 1A. Sixty days after treatment, the percentage of YFP+ β cells in each islet was measured. It was very difficult to find enough islets for quantification comparison in untreated late-stage diabetic NOD mice, although in an islet remnant found, 83% of residual β cells were still YFP+ (fig. S6). Alternatively, we measured the percentage of YFP+ β cells in control new-onset diabetic NOD mice. Ten-week-old NOD mice were treated with tamoxifen to label the preexisting β cells, and consistently, more than 80% of β cells in the islet remnants of the new-onset diabetic mice were YFP+ (Fig. 5A, top row).

Fig. 5

Administration of gastrin/EGF augmented both β cell neogenesis and replication in chimeric late-stage diabetic Rip-YFP NOD mice. (A) Mixed chimerism was induced in late-stage diabetic Rip-YFP NOD mice in combination with gastrin and EGF administration. Pancreatic tissues were harvested on day 60 after treatment and stained for insulin (red), YFP (green), and DAPI (blue) to measure the percentage of YFP+insulin+ cells in each islet. Untreated new-onset diabetic Rip-YFP NOD mice were used as control group. One representative of an islet from each group is shown. The percentage of YFP+ cells among insulin+ cells in each islet is also shown. There are four mice in each group. Bottom four rows of staining pattern represent islets with 90, 40, and 0% of YFP+ β cells. (B) The parallel slide of islets with more than 70 and 0% YFP+ β cells from chimeric late-stage diabetic recipients treated with gastrin/EGF in (A) was stained for insulin (red), glucagon (green), and DAPI (blue). Representative staining patterns of the islets are shown (n = 4). The percentage of glucagon+ cells among insulin+ cells in each islet with more than 70 and 0% YFP+ β cells is also shown.

Islets containing β cells were easily identified in mixed chimeric late-stage diabetic NOD recipients after treatment with gastrin/EGF. Among them, ~30% of islets had more than 70% YFP+ β cells, ~25% of islets had 10 to 50% YFP+ β cells, and ~45% of islets had no YFP+ β cells (Fig. 5A). The YFP+ and YFP islets had similar insulin staining intensity as judged by two islets in the same field (fig. S7), although some YFP islets appeared to have reduced insulin staining intensity, and there were insulindim and insulinbright cells in some islets (Fig. 5A). Furthermore, in the YFP+ islets (>70% YFP+), almost all the insulin+ cells were glucagon or somatostatin (Fig. 5B and fig. S8). In contrast, in two-thirds of YFP islets, insulin+ cells were also glucagon+; in the other one-third of YFP islets, insulin+ cells were either glucagon+ or glucagon (Fig. 5B). In some YFP islets, the insulin+glucagon+ cells appeared to be insulindim, and some of them appeared to be somatostatin+, as judged by staining parallel slides (Fig. 5B and fig. S8). These results suggest that the de novo developed YFP islet cells may be at different developing stages. These results also demonstrate that administration of gastrin and EGF augments both β cell replication and neogenesis in late-stage diabetic NOD mice with mixed chimerism.

Discussion

We have shown that conditioning with a radiation-free nonmyeloablative sequential injection of FNB-anti-CD3 and FB-anti-CD3/CD8 avoids a cytokine storm and allows for induction of mixed chimerism with no signs of GVHD in late-stage diabetic NOD mice. We have also demonstrated that combination therapy of induction of mixed chimerism and administration of gastrin/EGF can reverse late-stage diabetes in NOD mice, although either alone cannot. The major role of induction of mixed chimerism is to eliminate autoimmunity and allow newly generated β cells to survive and accumulate, whereas the major role of administration of gastrin and EGF is to augment β cell neogenesis and replication.

Conditioning with FB-anti-CD3/CD8 can facilitate induction of mixed chimerism without causing any signs of GVHD, but the cytokine storm triggered by FB-anti-CD3 prevents its clinical application. In the current studies, sequential injection of FNB- and FB-anti-CD3 was shown to avoid a cytokine storm while still allowing for induction of mixed chimerism with no GVHD in late-stage diabetic NOD mice. Thus, it may be possible to induce mixed chimerism with a radiation-free nontoxic approach in late-stage autoimmune recipients. Translation of this sequential injection of FNB- and FB-anti-CD3 into clinical application may be challenging, although not impossible, because clinical use of FcR-binding OKT3 has been largely replaced by nonbinding anti-CD3, and clinical use of OKT3 will need to be reintroduced.

In addition, although our anti-CD3/CD8 conditioning regimen for induction of mixed chimerism without causing any signs of GVHD in diabetic NOD mice, GVHD may still be a concern for clinical translation because autoimmune diabetes mainly affects children and young adults. Therefore, to translate this approach into T1D patients, we need to take the roadmap proposed by Sykes and colleagues (34). First, we need to test whether this anti-CD3–based conditioning can prevent GVHD in patients with hematological malignancies as well as test whether anti-CD3–based conditioning can induce mixed chimerism without causing any signs of GVHD in nonhuman primates. Then, we need to test this regimen in patients with life-threatening autoimmune diseases such as multiple sclerosis and systemic lupus, and finally in late-stage refractory T1D patients.

We observed that induction of mixed chimerism, but not treatment with anti-CD3/CD8 conditioning in combination with gastrin/EGF, was able to eliminate autoimmunity and insulitis as well as allow newly generated β cell survival in PDL-treated prediabetic NOD mice. This is consistent with previous reports that newly generated β cells are more susceptible to apoptosis induced by proinflammatory cytokines (32). Both administration of anti-CD3 alone and administration of EGF/gastrin alone were previously reported to ameliorate autoimmunity and reverse new-onset T1D in NOD mice (2, 24, 35); however, no strong evidence of β cell regeneration was reported, although increased β cell surface was shown in some reports (24). Thus, the reversal of new-onset diabetes may occur through suppressing the ongoing autoimmunity, subsequently reducing β cell death, and improving the function of the residual β cells. Late-stage diabetic NOD mice have too few β cells left, and therefore, regeneration of β cells is required for reversal of late-stage T1D. Therefore, reversal of autoimmunity and elimination of insulitis are necessary but not sufficient first steps for reversal of late-stage T1D. This might be the reason that most clinical trials that aim to reverse diabetes in T1D patients did not reach the expected endpoints (4).

Induction of mixed chimerism in late-stage diabetic NOD mice did not result in significant increase of β cell surface, even in combination with PDL treatment that induces pancreatic tissue release of growth factors (36). Although there were still some β cells in the islet remnants of late-stage diabetic NOD, it is unclear yet why the clusters of β cells do not expand like those in the new-onset diabetic mice after elimination of insulitis. It is also unclear why induction of mixed chimerism in combination with PDL did not augment β cell regeneration in late-stage diabetic NOD, but it did augment in prediabetic NOD mice. However, administration of gastrin/EGF was able to augment the β cell neogenesis and replication, resulting in reversal of late-stage T1D. This observation suggests that in late-stage diabetic individuals, the microenvironment for β cell regeneration may have been damaged and that exogenous growth factors are required for restoring β cell regeneration ability.

Most reports regarding β cell regeneration after β cell destruction in mice are about β cell replication (15), although β cell neogenesis took place in PDL-treated mice (16, 17). However, β cell replication in humans has not been observed, although neogenesis has been reported (37). Thus, it has become doubtful whether mechanisms of β cell regeneration revealed with mouse models are related to humans. Using Rip-YFP NOD mice, we observed that the regenerated β cells were from both β cell neogenesis and replication in late-stage diabetic NOD mice after induction of mixed chimerism and administration of gastrin/EGF. To this end, we believe ours to be among the first reports that β cell neogenesis usually observed in humans occurs in autoimmune diabetic mice. Because gastrin/EGF has been proved to augment human β cell neogenesis in non-autoimmune SCID mice (23), it makes this animal model and the therapeutic regimen highly relevant to humans. In conclusion, combination therapy of induction of mixed chimerism and administration of growth factors represents an approach that could potentially replace islet transplantation as a curative therapy for refractory T1D.

Materials and Methods

Female NOD/LtJ, C57BL/6, NOR, and NOD-SCID mice were purchased from The Jackson Laboratory. Rip-CreER NOD mice were established by Wilson laboratory at University of Florida with the original Rip-CreER mice from Melton laboratory at Harvard University (15). R26R-YFP NOD mice were established by Bluestone laboratory at University of California, San Francisco. [RIP-CreER; R26R-YFP] NOD mice were generated by crossing the two strains of NOD mice at City of Hope Research Animal Facilities (Duarte, CA). All animals were maintained in a pathogen-free room, and the animal use procedures were approved by institutional committee of City of Hope. Other procedures are described in the figure legends or Supplementary Materials.

Supplementary Materials

www.sciencetranslationalmedicine.org/cgi/content/full/4/133/133ra59/DC1

Materials and Methods

Fig. S1. A new anti-CD3/CD8 conditioning regimen that prevented cytokine storm.

Fig. S2. Induction of mixed chimerism in late-stage diabetic NOD mice under anti-CD3/CD8 conditioning regimen.

Fig. S3. Diabetes recovery curves of late-stage diabetic NOD mice after combination therapy of induction of mixed chimerism and administration of gastrin/EGF.

Fig. S4. Ratio of insulin and glucagon mRNA expression in the pancreatic tissues of different mice.

Fig. S5. Lack of β cell regeneration in late-stage diabetic NOD mice after PDL treatment.

Fig. S6. YFP+insulin+ cells in an islet remnant in a late-stage diabetic Rip-YFP NOD mouse.

Fig. S7. Comparison of insulin staining intensity in YFP+ and YFP islets in late-stage diabetic Rip-YFP NOD mice treated with the combination therapy.

Fig. S8. Comparison of somatostatin staining in YFP+ and YFP islets in the pancreas of Rip-YFP NOD mice treated with the combination therapy.

References and Notes

  1. Acknowledgments: We are grateful to A. Riggs for his continuous encouragement and support of this research. We thank P. Martin for his critical review of the manuscript, J. Bluestone for providing Rosa26-YFP NOD mice and FcR-nonbinding anti-CD3 hybridoma, and D. Melton for providing Rip-CreER mice. Funding: Supported by a grant from Iacocca Family Foundation and private donations from Todd & Karen Wanek and the Davis family. Author contributions: M.W. designed and performed the experiments and wrote the manuscript; I.T. performed immunofluorescence staining of β cells and reviewed the manuscript; J.J.R. established transgenic mice and assisted in the experiments; X.S. established the RIP-CreER NOD mouse model; X.L., H.L., I.N., A.A.-M., H.F.J., C.L., and C.S. assisted in the experiments; B.W. provided RIP-CreER NOD breeders and critically reviewed the manuscript; M.A. advised on experimental design and critically reviewed the manuscript; F.K. and S.F. supported the research project and critically reviewed the manuscript; D.Z. designed the research, supervised the project, and wrote the manuscript. Competing interests: B.W. is on the scientific advisory board of NKT Therapeutics. The other authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article