Islet Recovery and Reversal of Murine Type 1 Diabetes in the Absence of Any Infused Spleen Cell Contribution

See allHide authors and affiliations

Science  24 Mar 2006:
Vol. 311, Issue 5768, pp. 1775-1778
DOI: 10.1126/science.1124004


A cure for type 1 diabetes will probably require the provision or elicitation of new pancreatic islet β cells as well as the reestablishment of immunological tolerance. A 2003 study reported achievement of both advances in the NOD mouse model by coupling injection of Freund's complete adjuvant with infusion of allogeneic spleen cells. It was concluded that the adjuvant eliminated anti-islet autoimmunity and the donor splenocytes differentiated into insulin-producing (presumably β) cells, culminating in islet regeneration. Here, we provide data indicating that the recovered islets were all of host origin, reflecting that the diabetic NOD mice actually retain substantial β cell mass, which can be rejuvenated/regenerated to reverse disease upon adjuvant-dependent dampening of autoimmunity.

Type 1 diabetes, an autoimmune disease that targets pancreatic islet β cells, is an important and increasing health problem. It is generally believed that effective therapy for autoimmune diabetes will require two scientific advances: (i) restoration of insulin production by providing or eliciting new β cells, and (ii) repair of the breakdown in immunological tolerance that precipitated the disease in the first place.

Recently, successful achievement of both of these advances was reported, culminating in disease abrogation in a fraction of severely diabetic NOD mice (1), a widely used murine model of the human type 1 disorder. The protocol incorporated injection of a single dose of Freund's complete adjuvant (FCA) into severely diabetic mice, coupled with repeated infusion of allogeneic splenocytes, resulting in restoration of normoglycemia and permanent disease extinction. The conclusion of this study was that the FCA had eliminated anti-islet autoimmunity and that the donor splenocytes had differentiated into insulin-producing (presumably β) cells, ultimately leading to islet regeneration.

Although FCA has been used for a number of years in a variety of experimental protocols to modulate diabetes in NOD mice (2, 3), the concept of donor splenocytes giving rise to islets in a host pancreas was a novel and exciting one, prompting the suggestion that the function of the spleen should be radically reappraised to include a role as a reservoir of multilineage stem cells (4). We set out to further dissect the underlying mechanisms of diabetes reversal by this treatment regime.

As a first step, we attempted to replicate the original findings, following closely the protocol published by Kodama et al. (1) and incorporating additional details from more extensive protocols provided by the authors (5). Severely diabetic NOD mice (blood glucose >400 mg/dl) were transplanted under the kidney capsule with syngeneic islets in order to maintain insulin levels; engraftment was deemed a failure when hyperglycemia reappeared before 2 days. Thirty successfully transplanted mice were injected with FCA and infused with live allogeneic [C57BL/6 × Balb/c (referred to as CB6 F1)] splenocytes (Table 1). The majority (21 of 30) developed diabetes before the end of the 120-day observation period (2 to 60 days after islet transfer). We do not know why the 70% reversion rate observed here was higher than the 8% reported by Kodama et al. (1). After removal of the transplanted islets from the nine mice that had completed the 120-day observation period, four animals became diabetic shortly thereafter, indicating that the insulin that permitted their survival prenephrectomy was produced mainly by the grafted islet cells. The other five animals remained normoglycemic to termination, indicating that they carried insulin-producing cells outside the islet graft, which the immune system no longer destroyed. This 56% of long-term survivors that were insulin–self-sufficient was similar to the 67% reported by Kodama et al. (1).

Table 1.

Response profiles of treated mice. The treatment protocol is described in (5). BG, blood glucose.

Mouse IDAge at onset (weeks)Maximum BG before transplantationDays with BG >400 mg/dlNumber of transplanted isletsBG on first day after transplantation (mg/dl)Number of days with BG <250 mg/dl after transplantationView inlineOverall course
Successful transplantation for 120 days → nondiabetic after nephrectomy 132 23 476 3 650 129 151 Nephrectomy on day 120, still nondiabetic at day 151
138 30 499 3 800 147 186 Nephrectomy on day 120, still nondiabetic at day 186
154 26 401 3 800 120 180 Nephrectomy on day 120, still nondiabetic at day 180
155 35 406 7 800 74 151 Nephrectomy on day 124, still nondiabetic at day 151
89 19 459 7 800 135 144 Nephrectomy on day 124, still nondiabetic at day 144
Successful transplantation for 120 days → diabetic after nephrectomy 78 37 403 3 650 51 122 Nephrectomy on day 120, diabetic 3 days later
114 39 458 7 800 105 120 Nephrectomy on day 120, diabetic 1 day later
11 29 597 7 800 83 123 Nephrectomy on day 123, diabetic 1 day later
76 27 458 3 800 50 123 Nephrectomy on day 124, diabetic 1 day later
Recurrent diabetes before day 120 15 29 586 3 800 68 60 Spontaneously diabetic
174 20 >600 3 800 139 46 Spontaneously diabetic
74 24 584 3 800 101 46 Spontaneously diabetic
78 18 547 3 800 136 25 Spontaneously diabetic
84 24 591 14 650 134 24 Spontaneously diabetic
128 27 505 3 800 125 22 Spontaneously diabetic
120 44 >600 7 650 93 11 Spontaneously diabetic
10 25 532 3 800 104 11 Spontaneously diabetic
81 18 450 3 800 97 11 Spontaneously diabetic
159 26 502 3 800 97 5 Spontaneously diabetic
141 21 513 1 650 86 4 Spontaneously diabetic
72 17 514 3 800 64 4 Spontaneously diabetic
89 25 >600 8 650 134 2 to 7 Spontaneously diabetic
86 26 >600 12 650 125 2 to 7 Spontaneously diabetic
98 26 >600 1 650 90 2 to 7 Spontaneously diabetic
136 19 541 7 650 83 2 to 3 Spontaneously diabetic
170 16 581 7 800 172 2 to 3 Spontaneously diabetic
36 30 543 3 800 84 2 to 3 Spontaneously diabetic
81 40 >600 7 650 81 2 Spontaneously diabetic
23 25 432 7 800 113 2 Spontaneously diabetic
47 24 412 3 800 86 2 Spontaneously diabetic
  • View inline* A few mice showed transient blood glucose values >250 mg/dl during the follow-up period, but these normalized in the following days, and these excursions were not considered in the evaluation of posttransplant success or failure.

  • Histology of the pancreas revealed islets of various sizes (fig. S1A), including a hyperplastic subset found in both the “cured” and “revertant” long-term survivors, although more frequent in the former (e.g., fig. S1B). All mice had insulitis (fig. S1, A to D) as well as some islets free of infiltrate (fig. S1, A and E). Many islets exhibited an innocuous-looking infiltrate like that described in a number of contexts wherein insulitis does not progress to overt diabetes (6). Indeed, there was a striking similarity to the islets of recent-onset diabetic NOD mice treated with anti-CD3 monoclonal antibody (mAb) (fig. S1F). Insulitis was also observed in the syngeneic islet graft (fig. S1G).

    We next evaluated the possibility that pancreatic islets were regenerated through differentiation of donor splenocytes. Because the fluorescence in situ hybridization (FISH) assays used in the previous study (1) can be challenging and their results can be equivocal, we established a more robust test based on single-nucleotide polymorphism (SNP) analysis of DNA derived from islet tissue isolated from histological sections by laser-capture microdissection (LCM). Islet tissue and any associated inflammation could be definitively identified by staining sections with hematoxylin and/or anti-insulin mAb (Fig. 1A) and could be cleanly excised (Fig. 1B). The SNP analysis assessed three polymorphic loci on three separate chromosomes (Fig. 1C) and clearly distinguished cells of NOD and CB6 F1 origin. In all “cured” animals tested, all tissue was derived from NOD host or graft cells; none came from CB6 F1 donor splenocytes (Fig. 1D).

    Fig. 1.

    Laser-capture microdissection and genotyping of islets from “reverted” mice. (A) Anti-insulin staining of an islet from mouse 154. (B) An adjacent section before (left) and after (right) microdissection. (C) Typical genotyping by fluorogenic polymerase chain reaction for SNP rs4151928, detecting the alleles present in NOD or CB6 F1 mice as FAM or VIC fluorescence, respectively. Controls shown are DNA from microdissected islets of a nondiabetic NOD mouse and from an F1 mouse. Solid triangles denote several samples microdissected from islets of mouse 154. (D) Summary of SNP genotyping of microdissected islets from mice 132, 138, and 154 (all from the first group in Table 1). Nucleotides detected for three different SNP positions (rs4223216, rs4230624, and rs4151928) in microdissected islets are shown, with the corresponding nucleotides for control NOD and F1 DNA (x, unsuccessful amplification; n.d., not done).

    This result is actually the expected one, given that the NOD host should make a strong alloresponse to the splenocytes' large major histocompatibility complex (MHC) divergence (Kb, Db, Ld, Ad, Ed, and Ab in the CB6 F1s). We found no evidence of chimerism in the spleen or lymph nodes subsequent to the splenocyte infusions (fig. S2). Instead, strong alloreactivity was evident from the development of antibodies to the allogeneic splenocytes at titers up to 1/5000 (7).

    The fleeting presence of the splenocytes raised the possibility that the protocol's beneficial effects might derive primarily from the FCA injection. This adjuvant is known to have immunomodulatory effects in the NOD context (2, 3), but a FCA-alone control was not included in the report by Kodama et al. (1). Therefore, under otherwise the same protocol, we treated 13 severely diabetic NOD mice with FCA and a syngeneic islet transplant but not with allogeneic splenocytes. Four of these animals maintained normal glucose levels during the 120-day observation period, and three of these remained normoglycemic after removal of the transplanted islets (table S1).

    A likely source of the insulin that permitted certain long-term survivors to remain nondiabetic is residual islets that expanded through either differentiation or replication once autoimmunity had been muted. Such a scenario is consistent with recent reports that substantial β cell mass can be detected in diabetic NOD mice (8). To test this notion, we performed a morphometric analysis of pancreatic β cell mass both in the prediabetic state (n = 3) and in the first weeks after the development of severe hyperglycemia (n = 23). Residual β cell mass was detected in essentially all of the diabetic mice (fig. S3), although it was variable in quantity and was always diminished relative to that of prediabetic comparators. There was no correlation between the blood glucose value and residual β cell mass (9). These results indicate that host β cells can contribute to islet regeneration promoting disease reversal in severely diabetic NOD mice.

    Our experiments, like those of Kodama et al. (1), yielded a fraction of previously diabetic NOD mice that survived for a long term after syngeneic islet transplantation, a proportion of which remained normoglycemic even after the removal of the grafted islets. On the other hand, we found no evidence that the source of insulin underlying the reversal of diabetes was islet cells derived from donor splenocytes. Rather, we found that the diabetic hosts had substantial residual β cell mass and that the recovered/expanded islets were all of NOD origin rather than splenocyte CB6 F1 origin. Our conclusions match very well those of two accompanying reports (10, 11). Given recent reports of active β cell replication in both nondiabetic (12) and diabetic (8) states, the most likely explanation for islet recovery in this context is that the dampening of autoimmunity permitted β cell replication to outdo β cell death. However, other explanations remain possible: differentiation of new host β cells or seeding and expansion of islet graft–derived cells (also of NOD genotype).

    Should the autonomous efficacy of FCA demonstrated in our study prompt a reconsideration of the use of the analogous reagent Bacille Calmette-Guúrin (BCG), in isolation, to treat diabetic humans? Although a preliminary study with 17 newly diagnosed individuals appeared to show some therapeutic efficacy (13), three more extensive, double-blind trials failed to demonstrate a positive effect (14). Likely explanations for the divergent outcomes are simply that FCA and BCG are different reagents, or that the footpad injection of FCA used for mice and the subcutaneous injection of BCG applied to humans are radically different interventions, the former provoking a massive systemic inflammatory response (15) difficult to contemplate for human patients.

    Supporting Online Material

    Materials and Methods

    Figs. S1 to S3

    Table S1

    References and Notes

    View Abstract

    Stay Connected to Science

    Navigate This Article