Targeting Cells Causing Split Tolerance Allows Fully Allogeneic Islet Survival With Minimal Conditioning in NOD Mixed Chimeras


Colin C. Anderson,


Donor-specific tolerance induced by mixed chimerism is one approach that may eliminate the need for long-term immunosuppressive therapy, while preventing chronic rejection of an islet transplant. However, even in the presence of chimerism it is possible for certain donor tissues or cells to be rejected whereas others from the same donor are accepted (split tolerance). We previously developed a nonmyeloablative protocol that generated mixed chimerism across full major histocompatability complex plus minor mismatches in NOD (nonobese diabetic) mice, however, these chimeras demonstrated split tolerance. In this study, we used radiation chimeras and found that the radiosensitive component of NOD has a greater role in the split tolerance NOD mice develop. We then show that split tolerance is mediated primarily by preexisting NOD lymphocytes and have identified T cells, but not NK cells or B cells, as cells that both resist chimerism induction and mediate split tolerance. Finally, after recognizing the barrier that preexisting T cells impose on the generation of fully tolerant chimeras, the chimerism induction protocol was refined to include nonmyeloablative recipient NOD T cell depletion which generated long-term mixed chimerism across fully allogeneic barriers. Furthermore, these chimeric NOD mice are immunocompetent, diabetes free and accept donor islet allografts.


anti-CD40 ligand


antigen presenting cell


bone marrow cell


bone marrow transplant




cytotoxic T-lymphocyte-associated antigen 4


green fluorescent protein


hematopoietic stem cells






major histocompatability complex


nonobese diabetic


phosphate buffered saline


recombinase activating gene 1


tissue specific antigen


wild type


Clinical islet transplantation is a minimally invasive approach to restore functional β cell mass and normoglycemia (1). Regardless of insulin independence, islet transplantation can stabilize some secondary complications of diabetes when compared to optimal medical therapy (2,3). However, islet transplantation subjects recipients to a lifetime of immunosuppression, thereby increasing risk of infection and malignancy (4). Furthermore, despite immunosuppressive medications, the recurrence of autoimmunity after pancreas transplant is at least 5–6%, and may contribute to the inexorable loss of endocrine function years after transplantation (5).

Donor-specific tolerance induction, through the generation of mixed chimerism, is one approach that may eliminate need for long-term immunosuppressive therapy, while also preventing chronic rejection. Owen first associated chimerism with tolerance after demonstrating fraternal cattle twins were natural chimeras (6). In mixed chimeras, tolerance involves both central (7–10) and peripheral mechanisms (11–13), however, even in the presence of chimerism, it is possible for certain donor tissues or cells to be rejected, a phenomenon known as split tolerance (14–17). In addition to its occurrence in mice, in both large animal models (18–20) and human trials (21,22) chimerism induction and solid organ grafting can be associated with split tolerance. In these cases, split tolerance usually develops such that peripheral blood chimerism is lost, with the preservation of the solid organ graft; however, the reverse can also occur (23). Although the relevance of the maintenance of chimerism in the presence of solid organ tolerance is controversial (21,23–25), rejection of an organ allograft is clearly not desirable. Therefore, a thorough understanding of the mechanisms of tolerance operating in mixed chimerism and the subtleties of why tolerance develops toward some tissues and not others would facilitate design of chimerism induction therapies that ensure the recipient will indeed be tolerant of relevant donor tissues.

Nonobese diabetic (NOD) mice (26) are resistant to tolerance induction (27,28) and represent the most challenging inbred mouse model to induce chimerism using nonmyeloablative protocols (14,29,30). A clinically feasible nonmyeloablative protocol that generates lasting multilineage and fully allogeneic chimerism in NOD mice has yet to be achieved. We previously developed a nonmyeloablative protocol that generated mixed allogeneic chimerism across full major histocompatability complex (MHC) plus minor mismatches (14,29), however, the long-lasting chimerism was not multilineage. Furthermore, in contrast to chimeric C57BL/6 (B6) mice, chimeric NOD mice demonstrated multiple levels of split tolerance, such that islets, skin and some hematopoietic lineages were rejected. Understanding this autoimmune model may help identify barriers to chimerism induction that will aid in tailoring efficient, clinically relevant, protocols capable of generating fully tolerant chimeras. In this study, we isolated the contributions of both the radiosensitive and radioresistant cellular compartments to the development of split tolerance and found split tolerance was caused primarily by T lymphocytes. We then developed a refined chimerism induction protocol based on recipient NOD T cell depletion. This clinically relevant, nonmyeloabalative protocol was successful at generating chimeric NOD mice that are diabetes free and accept donor islet allografts while remaining immunocompetent.

Materials and Methods


Adult female B6.NOD-(D17Mit21-D17Mit10) mice, abbreviated here as B6.g7 (H-2g7), were purchased from the Jackson Laboratory (Bar Harbor, ME, USA). Wild type NOD (NOD.wt; H-2g7), NOD mice on the recombinase activating gene (RAG) 1-knockout (KO) background, B cell deficient NOD.129S2(B6)-Igh-6tm1Cgn (NOD.μMT) and C3H (H-2k), originally from Jackson Laboratory, were bred in-house. NOD mice expressing green fluorescent protein (GFP; NOD-Tg [UBC-GFP] 30Scha), originally a gift from Dr R. Gill (Barbra Davis Center for Diabetes; University of Colorado), were bred in-house. NOD.GFP, NOD.wt and C3H day 15 gestation fetuses were generated by timed pregnancies onsite. All mice used for chimerism induction were females between 8–10 weeks of age. All care and handling of animals were carried out in accordance with the guidelines of the Canadian Council on Animal Care.

Creation of lethal irradiation full bone marrow chimeras

Bone marrow cells (BMC) were isolated from the femurs and tibias of donor NOD.GFP and B6.g7 mice. Recipient NOD.GFP and B6.g7 mice were lethally irradiated with 1200 RAD in split doses over 2 days, and then immediately transplanted intravenously (i.v.) with 20 × 106 unmodified BMCs from the opposite strain. Alternatively, NOD.GFP FLC→B6.g7 full chimeras were generated by the i.v. injection of lethally irradiated B6.g7 mice with 20 × 106 unmodified NOD.GFP FLC. Irradiated mice were administered TMX/SMP antibiotics in their drinking water for the duration of the experiment.

Induction of allogeneic nonmyeloablative chimerism

Transplantation of 20 × 106 allogeneic BMCs by a nonmyeloablative protocol into NOD.wt, NOD.μMT, NOD.GFP→B6.g7 and B6.g7→NOD.GFP chimeric mice was described in depth previously (29).

Chimerism induction based on T cell depletion was performed by administering anti-CD4 (GK1.5) and anti-CD8 (53.6.7) on days –5 and –1 (each at 0.25 mg i.p.; Bio X Cell, West Lebanon, NH, USA). Busulfan (BUS) was administered 1 day before the day 0 transplantation of 20 × 106 allogeneic C3H BMCs i.v. along with a single i.p. injection of anti-CD40L. After BMT, rapamycin was injected daily for 28 days.

To generate chimeras in immunodeficient hosts, recipient NOD.RAG1-KO mice were conditioned with BUS (day –1) and anti-asialo GM1 (days –3, 0 and 3). Mixed chimeras were generated by an i.v. injection of a 1:1 mixture of allogeneic (C3H) and syngeneic (NOD.wt) FLCs, at 15 × 106 cells each. To generate full chimeras, recipient NOD.RAG1-KO mice were transplanted i.v. with 100 million unmodified C3H BMCs.

Flow cytometry

Antibodies against TCR, CD4, CD8, CD11b, CD11c, CD19, CD49b, CD122, H-2Kk and H-2Db were purchased from BD Pharmingen (San Diego, CA, USA) and eBioscience (San Diego, CA, USA). Donor C3H cells were identified by anti-H-2Kk antibodies and recipient cells were identified by anti-H-2Db or GFP. Data were acquired using a LSR II (Becton Dickson, Sunnyvale, CA, USA) flow cytometer and analyzed with FCS Express™ (De Novo Software, Los Angeles, CA, USA).

Skin transplantation

Full thickness trunk skin was transplanted onto the dorsum of recipient mice. Briefly, approximately 1 cm2 of donor skin was secured with sutures to the recipient graft bed. The skin grafts were then bandaged and left intact for 7 days. The grafts were inspected daily and considered rejected at the time when >90% surface area was necrotic.

Islet isolation and transplantation

Islet isolation was carried out as previously described (31). Briefly, 3 days before transplantation, recipients were made diabetic by an i.p. injection of streptozotocin (Sigma-Aldrich Canada, Oakville, ON, Canada) at 185 mg/kg. Diabetes was confirmed by a blood glucose of > 20.0 mmol/L. Five hundred donor islets were transplanted into the renal subcapsular space. Grafts were considered rejected when blood glucose exceeded 15 mmol/L on two consecutive readings over 2 days.

Statistical analysis

An unpaired two-tailed Student's t-test was used for comparison of means between two groups and a log-rank test was used to compare survival curves. All statistical analyses were done using Prism 5 (GraphPad Software, San Diego, CA, USA) with statistical significance defined as p < 0.05.


NOD bone marrow has the capacity to generate split tolerance toward donor hematopoietic lineages but not donor islets

To isolate the effects NOD radiosensitive and radioresistant tissues have on split tolerance, reciprocal radiation bone marrow chimeras were generated between female tolerance resistant NOD.GFP and MHC matched tolerance competent B6.g7 mice (Table 1). NOD.GFP mice develop diabetes with a similar incidence to NOD.wt mice (Figure S1A) and demonstrate split tolerance when made chimeric with fully allogeneic C3H BMT by our nonmyeloablitive regimen (data not shown). After 5 weeks, NOD.GFP→B6.g7 and B6.g7→NOD.GFP full chimeras were assessed for hematopoietic cell reconstitution (Figures S1B and C). Although less than 2% of the NOD.GFP BMC's used to generate the radiation chimeras were composed of T cells (data not shown), we were concerned that these few, preexisting NOD.GFP derived T cells may be sufficient to cause split tolerance when transferred into B6.g7 mice. Therefore, to ensure all NOD.GFP cells developed in the B6.g7 host, we also generated a cohort of NOD.GFP→B6.g7 chimeras using NOD.GFP day 15 gestation fetal liver cells (FLC), which have no mature T cells. The hematopoietic cells in these reconstituted B6.g7 mice were 78.6 ± 1.4% NOD.GFP in origin. In agreement with prior studies (32), none of the NOD.GFP→B6.g7 or B6.g7→NOD.GFP full chimeras developed diabetes, however insulitis was present in the NOD.GFP→B6.g7 group (data not shown).

Table 1.  Rationale for use of various radiation chimeras as recipients of a C3H BMT
Radiation chimeraRadiosensitive cellsRadioresistant cellsRationale
NOD.GFP→B6.g7NOD B6  Isolate the effect that NOD hematopoietic cells have on the development of split tolerance
NOD.GFP FLC→B6.g7NOD B6  Isolate the effect NOD hematopoietic precursors have on the development of split tolerance
B6.g7→NOD.GFPB6.g7NODIsolate the effect NOD radioresistant cells have on the development of split tolerance
NOD.GFP→NOD.GFPNOD NODChimera to control for any effects radiation and BMT may have on the development of split tolerance
B6.g7→B6.g7B6.g7B6.g7Chimera to control for any effects radiation and BMT may have on the development of split tolerance

Using our irradiation free, costimulation based, nonmyeloablative protocol (14), we induced allogeneic (C3H donor) hematopoietic chimerism in NOD.GFP→B6.g7, NOD.GFP FLC→B6.g7, B6.g7→NOD.GFP and in control B6.g7→B6.g7 and NOD.GFP→NOD.GFP radiation chimeras (Figure 1A). In all groups, initial C3H chimerism was multi-lineage, consisting phenotypically of T, B and NK cells, macrophages and dendritic cells. The multilineage nature of the C3H chimerism was stable long-term in B6.g7→NOD.GFP and B6.g7→B6.g7 groups, suggesting that these mice were tolerant of all hematopoietic lineages (data shown for CD4 T and B cells; Figure 1B). This tolerance is in contrast to the split tolerance seen in NOD.GFP→NOD.GFP mice where it was especially noticeable that donor B cell frequency decreased and was lost, whereas CD4 T cell frequency increased (Figure 1B). The NOD.GFP→B6.g7 chimeras could be divided into two subgroups, those with high or low initial chimerism levels. Interestingly, NOD.GFP→B6.g7 mice with high C3H chimerism maintained all hematopoietic lineages; however, mice with low C3H chimerism, demonstrated split tolerance toward donor B cells, similar to NOD.GFP→NOD.GFP mice. We then reanalyzed the B6.g7→NOD.GFP and B6.g7→B6.g7 mice with low chimerism levels and confirmed that split tolerance toward hematopoietic cells did not develop (Figure 1B). Thus, split tolerance was not simply a reflection of low chimerism levels but instead additionally required hematopoietic cells to be of the NOD genetic background.

Figure 1.

Split tolerance develops toward progeny of HSCs only in NOD.GFP→B6.g7 mice with low C3H chimerism levels. The full chimeras (donor and host as indicated) were created by lethal irradiation and BMT. One–three weeks after full chimerism was confirmed they were given nonmyeloablative conditioning and a C3H BMT. (A) Left, 8 weeks post-BMT, peripheral blood C3H chimerism was assessed. Mixed chimerism was generated in 19/19 B6.g7→NOD.GFP, 13/15 NOD.GFP→B6.g7 (this group generated mice with either high or low chimerism levels; above or below solid line at 15%, respectively), 6/6 NOD.GFP FLC→B6.g7, 3/3 B6.g7→B6.g7 and 3/5 NOD.GFP→NOD.GFP mice. Right, in all chimeric animals, peripheral blood was analyzed by flow cytometry for the presence of multilineage C3H chimerism. Total chimerism levels, CD4, CD8, CD19 and CD49bCD122 cell proportions are taken from the lymphocyte gate, whereas CD11b and CD11c proportions are taken from the live cell gate. (B) Long-term monitoring of C3H CD4 T and B cell chimerism. Top panels, mice that had high (> 15%) initial chimerism levels and bottom panels, mice that had low (<15%) initial chimerism levels. Fifteen percent was chosen as the cutoff because no mice with greater than this level of chimerism in the NOD.GFP→B6.g7 group showed evidence of split tolerance.

Next, we wanted to determine if tolerance or split tolerance toward C3H hematopoietic cell progeny in NOD.GFP→B6.g7 and B6.g7→NOD.GFP chimeras extended to islet transplants. Because the C3H chimerism levels had a wide range (Figure 1A), we transplanted C3H islets into mice of various levels of C3H chimerism to rule out any confounding effect the level of chimerism may have. As seen in Figure 2A, all NOD.GFP→B6.g7, B6.g7→NOD.GFP and B6.g7→B6.g7 chimeric with C3H accepted donor C3H islets long term. In contrast, a NOD.GFP→NOD.GFP chimera rejected donor islets, despite the maintenance of chimerism. Control non-C3H chimeric NOD.GFP→B6.g7 and B6.g7→NOD.GFP mice (that did not receive C3H bone marrow) rejected their C3H islet transplant. The quick rejection of allogeneic islets by most of the NOD.GFP→B6.g7 and B6.g7→NOD.GFP mice that received conditioning, but not C3H bone marrow, demonstrates that these mice are immunocompetent. In addition, immunocompetence was demonstrated in mice that accepted C3H islets by the rejection of third-party islets (Figure 2A).

Figure 2.

NOD.GFP→B6.g7 and B6.g7→NOD.GFP chimeras accept donor C3H islets but are split tolerant toward C3H skin grafts. (A) Survival of islet transplants. Between 10 and 12 weeks after conditioning and C3H BMT, NOD.GFP→B6.g7, B6.g7→NOD.GFP, B6.g7→B6.g7 and NOD.GFP→NOD.GFP chimeric mice were made diabetic with streptozotocin and then given C3H islet transplants. As controls, NOD.GFP→B6.g7 and B6.g7→NOD.GFP mice that received conditioning, but no C3H BMT, were similarly given a C3H islet transplant. The outcome of these groups were similar and combined as “NOD.GFP⇔B6.g7 no C3H”. After greater than 100 days, a donor nephrectomy was performed to ensure the recurrence of hyperglycemia and for islet histological analysis (data not shown). After donor nephrectomy, some mice in each group were then transplanted with third party (B6.RAG1-KO) islets. The outcome of these groups was similar and combined as “NOD.GFP⇔B6.g7 3rd party.” (B) Skin was transplanted onto NOD.GFP→B6.g7, NOD.GFP FLC→B6.g7 and B6.g7→NOD.GFP mice during the nonmyeloablative allogeneic chimerism induction protocol. As both the NOD.GFP→B6.g7 (n = 7) and NOD.GFP FLC→B6.g7 (n = 4) groups rejected donor skin transplants with similar kinetics, these groups were combined (labeled as NOD.GFP→B6.g7). Skin survival was significantly different between NOD.GFP→B6.g7 and B6.g7→NOD.GFP (p = 0.018). Some NOD.GFP→B6.g7 and B6.g7→NOD.GFP chimeras also received a third party (B6.RAG1-KO) skin transplant. The outcome of these groups was similar and combined as “NOD.GFP⇔B6.g7 third party.”

We further tested the tolerance of NOD.GFP→B6.g7, NOD.GFP FLC→B6.g7 and B6.g7→NOD.GFP radiation chimeras toward skin grafts. During the nonmyeloablative allogeneic chimerism induction protocol, we transplanted skin from donor (C3H), syngeneic and third party mice. As expected, both the NOD.GFP→B6.g7 and B6.g7→NOD.GFP chimeras quickly rejected third party skin. NOD.GFP→B6.g7 chimeras rejected donor skin, indicating that the capacity for split tolerance to skin can be conferred with NOD bone marrow alone (Figure 2B). This split tolerance was not because of contaminating T cells in NOD bone marrow, as NOD FLC also had the capacity to generate split tolerance to skin. Interestingly, the B6.g7→NOD.GFP chimeras also demonstrated split tolerance toward C3H skin grafts; however, the rate of rejection was significantly slower than the NOD.GFP→B6.g7 mice (p = 0.018). The slower rate of rejection may be related to higher levels of chimerism in the B6.g7→NOD.GFP mice (47 ± 11) than the NOD.GFP→B6.g7 (34 ± 4) mice. However, irrespective of the rate of rejection or chimerism level, all NOD.GFP→B6.g7 and B6.g7→NOD.GFP chimeras demonstrated split tolerance toward skin. Together these results suggest that both the radiosensitive and resistant components of the NOD genetic background are essential for islet rejection in mixed chimeras. In contrast, either of the NOD components is sufficient for donor skin rejection, although the radiosensitive hematopoietic cells make a greater contribution to split tolerance.

NOD lymphocytes developing in the presence of chimerism do not generate split tolerance

Although we have shown evidence that the split tolerance phenotype in NOD chimeras resides primarily in hematopoietic cells, we next wanted to identify the cells involved. To isolate the effects on the generation of split tolerance of NOD lymphocytes that have developed in the presence of chimerism, immunodeficient NOD.RAG1-KO mice had their immune system reconstituted with a mixture of allogeneic C3H and syngeneic NOD.wt FLCs. Therefore, all NOD T and B cells developed in the presence of C3H chimerism. These mice remained stable mixed chimeras for greater than 20 weeks (Figure 3A). Furthermore, the chimerism was multi-lineage, with no significant loss of any cell lineage (data shown for CD4 T and B cells; Figure 3A). The stability of the chimerism in these mice is in contrast to the split tolerance seen in NOD.wt mice where overall chimerism levels decline and CD4 T cells become the predominant cell lineage (See previous sections). Although unlikely, the maintenance of chimerism in this experiment could be because of a competitive advantage of C3H FLCs over NOD FLCs. Therefore, in an additional experiment, NOD.RAG1-KO mice were made full C3H chimeras by transplanting 100 × 106 BMCs. These C3H→NOD.RAG1-KO full chimeras were then challenged solely with 20 × 106 NOD.wt FLCs. Similar to the results obtained with the mixture of C3H and NOD FLCs transplanted into NOD.RAG1-KO mice, the transfer of NOD FLCs alone into C3H→NOD.RAG1-KO full chimeras generated a state of mixed chimerism with no signs of split tolerance toward hematopoietic cells (Figure 3B). These data suggest that the split tolerance that NOD mice develop under nonmyeloablative conditioning is generated from preexisting lymphocytes that did not become tolerant to donor cells during the conditioning protocol and not from cells that have developed in the presence of chimerism.

Figure 3.

Stable mixed chimerism develops in NOD.RAG1-KO mice when recipient lymphocytes develop in the presence of allogeneic cells. (A) Top, peripheral blood was assessed for the presence of allogeneic and syngeneic lymphocytes in immunodeficient NOD.RAG1-KO mice that had their immune system reconstituted with a mixture of allogeneic C3H and syngeneic NOD.wt FLCs (n = 4). Bottom, Long-term monitoring of donor lymphocyte gated CD4 T and B cells in these chimeric mice. (B) Top, C3H→NOD.RAG1-KO full chimeras were injected with BUS and then 20 million NOD.wt FLCs and had their peripheral blood monitored long-term for the presence of allogeneic and syngeneic lymphocytes (n = 4). Bottom, long-term monitoring of donor lymphocyte gated CD4 T and B cells in these chimeric mice.

NOD CD4 and CD8 T cells mediate split tolerance

Because preexisting lymphocytes appeared to be the cause of split tolerance toward C3H hematopoietic cells, we next determined the specific lineage involved. Through modifications to the chimerism induction protocol used above, we isolated the effects that eliminating NK, B and T cells have on split tolerance development. Beginning with NK cells, we tested whether anti-asialo GM1 treatment on days –3, 0 and +3 relative to C3H BMT converts split tolerance into full tolerance. In agreement with our previous data (29), we were able to induce chimerism at a greater frequency and level than without NK cell depletion (Figure 4A). However, these chimeric mice still showed split tolerance toward both certain hematopoietic cell lineages and islets (Figures 4B and 5A). Donor CD4 cells had the highest frequency whereas B cells appeared rejected. In addition, C3H islets were rejected as quickly as in controls. Therefore, it appears that NK cells are not central to the development of split tolerance. To isolate the effects of B cells, we used as the recipient of our conditioning protocol and BMT, the NOD.μMT mouse that does not develop endogenous B cells (33,34). These mice became chimeric with similar frequency and levels as NOD.wt mice. Similar to the results with NK depletion, these mice also demonstrated split tolerance toward donor hematopoietic cells (Figure 4). Through a significant modification of our induction protocol, the last lymphocyte subgroup that we tested was T cells. Similar to the results of others, we could not induce chimerism in NOD mice with CD4 or CD8 depletion alone (30). Therefore, we used a protocol that involved both CD4 and CD8 depletion on days –5 and –1 relative to BMT as well as a single dose of anti-CD40L antibody on the day of transplant. This protocol was successful at inducing C3H chimerism at a high incidence and a significantly higher level than protocols that did not include T cell depletion. None of the chimeric NOD mice developed diabetes (data not shown). The chimerism levels and donor hematopoietic cell subsets remained constant long-term (Figure 4B); with no trends to T cell dominance and B cell elimination. These mice were also tolerant of donor islets, although retaining the ability to reject third party islets (Figure 5A). However, the T cell depletion protocol was insufficient to render chimeric mice tolerant toward donor skin grafts transplanted during the conditioning period (Figure 5B). Taken together, the elimination of NOD T cells along with a single dose of anti-CD40L antibody and a short course of rapamycin can induce chimerism with high incidence, and also render these chimeras operationally tolerant to all donor hematopoietic cells lineages and islets; however, split tolerance is still seen toward skin grafts.

Figure 4.

NOD T cells resist chimerism induction and cause split tolerance. (A) C3H chimerism level in peripheral blood at eight weeks post-BMT. Wild type NOD and B cell deficient NOD mice treated with a nonmyeloablative, costimulation-based chimerism induction protocol before the transplantation of 20 million C3H BMCs. Other groups of mice received a modified chimerism induction protocol that included NK or T cell depletion. (B) At 20 weeks post-BMT, peripheral blood was analyzed for donor lymphocyte composition by flow cytometry. (C) Peripheral blood was monitored over time for the presence of allogeneic lymphocytes. The mean and standard error was calculated by including only mice that had chimerism levels above zero for each chimerism induction protocol. Asterisks denote statistical significance.

Figure 5.

Chimerism induction that includes T cell depletion allows tolerance to donor islets. (A) At 10–12 weeks post-BMT, chimeric mice were treated with streptozotocin to induce diabetes. Left, once diabetic, chimeric mice generated by either our costimulation blockade-based protocol or a protocol that also included NK depletion were transplanted with donor (C3H) islets. As a control, a single mouse treated with costimulation blockade that did not receive a BMT was also transplanted with C3H islets. After the C3H islets were rejected, syngeneic (NOD) islets were transplanted into these chimeric mice. Right, After diabetes induction, mice made chimeric via the T cell depletion protocol received either donor (C3H) or third party (B6.RAG-KO) islets. (B) During BMT conditioning, some mice in each protocol group received donor (C3H) skin grafts.


Generating mixed allogeneic chimerism has long been recognized as a method of inducing operational tolerance toward allografts (35). Mixed chimerism is also clinically applicable, with trials attempting to take advantage of the robust tolerance associated with chimerism (21,36). In addition, in the case of transplantation for an autoimmune disease, chimerism may be able to “reeducate” the immune system to avoid recurrent autoimmunity (37). However, although chimerism is usually associated with tolerance, even in the setting of mixed chimerism, it is possible for rejection of allografts or the chimerism itself to occur. Previously, we showed that the propensity for such split tolerance is much greater in autoimmune-prone NOD recipients. Many mechanisms behind the occurrence of split tolerance have been proposed; reviewed in (38). Our goal was to isolate the cells/tissues involved in causing split tolerance in NOD mice and develop a protocol to generate fully-tolerant chimeras.

Through the use of radiation chimeras between NOD.GFP and B6.g7 mice, we have shown the potential to develop split tolerance after fully allogeneic BMT resides primarily in the NOD radiosensitive cellular compartments, although a contribution by radioresistant cells is also suggested. The radiosensitive compartment implicates NOD hematopoietic cells in the development of split tolerance, whereas high on the list of candidate cells in the radioresistant compartment are dermal APCs (discussed below) or thymic stromal cells. NOD mice display defective central tolerance (39–41) (although this has been recently contested [42]) and NOD thymic epithelium is sufficient to generate autoreactive T cells (43). Therefore, despite chimerism, a defect in central tolerance may allow the export of mature T cells capable of mediating split tolerance. There is a greater contribution to the generation of split tolerance from the radiosensitive compartment of NOD mice because despite chimerism, some NOD.GFP→B6.g7 mice were capable of rejecting donor B cells and skin, whereas B6.g7→NOD.GFP chimeras only rejected donor skin (the most immunogenic tissue tested). Different from both of these groups were NOD.wt mice (14) and control NOD.GFP→NOD.GFP mice, which demonstrated split tolerance toward B cells, skin and islets. The most likely explanation for these findings is that radioresistant and radiosensitive cellular compartments have differential effects on the generation of split tolerance, and both compartments are required together for the rejection of less immunogenic tissues. Subsequent experiments implicated preexisting lymphocytes as the cause of split tolerance, as NOD lymphocytes developing in the presence of chimerism were tolerant of donor HSCs. These data indicate that if all recipient T cells have a chance at central tolerance to the donor, then there is no split tolerance, and furthermore suggest that central tolerance is sufficiently intact in chimeric NOD mice. Consistent with a key role for central tolerance in preventing split tolerance, elimination of T, but not B or NK cells could abrogate the occurrence of split tolerance after allogeneic BMT. Only if T cell depletion was part of the chimerism induction protocol were we able to generate tolerance to islets in NOD chimeras. These results indicate NOD split tolerance is partially a result of preexisting T cells; however, other mechanisms of split tolerance must be present as T cell depletion did not allow for tolerance toward donor skin grafts. The continual display of split tolerance toward skin demonstrates that despite the central tolerance offered by mixed chimerism, peripheral tolerance may still be overwhelmed by the presence of polymorphic tissue-specific antigens (TSA). Tolerance to TSA requires peripheral tolerance, of which there are defects in the NOD mouse (44). Although polymorphic TSAs may be present in other tissues, the high susceptibility of skin to indirect rejection (15) makes it difficult to induce tolerance toward. Therefore, we hypothesize that unless some degree of donor-recipient antigen matching is performed, there will be split tolerance toward skin grafts in mice strains with defects in peripheral tolerance.

NOD T cells are known to resist tolerance induction (27,28,45) and may therefore resist both central (7–9) and peripheral (11,12,46) tolerance mechanisms acting in mixed chimeras. These nontolerized T cells can then mediate split tolerance through the recognition of polymorphic TSAs (47) or against targets that have a susceptibility to indirect rejection (15). T cells are indeed capable of generating split tolerance as previous work in our lab has shown that monoclonal CD4 T cells alone are able to eliminate certain allogeneic targets such as B cells, but not others (15), a split tolerance remarkably similar to that seen in NOD mice. Although we have identified T cells as a major contributor to the development of split tolerance in NOD mice, T cells must be activated by antigen presented on APCs. NOD APCs display a defect in the ability to induce peripheral tolerance (44), which may be important in the development of full versus split tolerance after costimulation-based chimerism induction protocols. Although chimerism takes advantage of central tolerance (7–9), with costimulatory-based induction protocols, more of the recipient cellular compartment is left intact and must be subject to peripheral tolerance mechanisms (11,12,46). Dermal APCs are resistant to radiation (48); therefore, NOD APCs may indeed be the radioresistant cell responsible for the development of split tolerance in B6.g7→NOD GFP chimeras. If this is the case, split tolerance may be solely because of NOD hematopoietic cells (T cells and APCs), as a limitation of our radiation chimera approach is an inability to remove all hematopoietic derived cells.

In our study, we found that the depletion of CD4 and CD8 cells along with a single dose of anti-CD40L were necessary to induce chimerism and long-term acceptance of all hematopoietic cell lineages and islets. Experiments preformed with only CD8 depletion or CD4 depletion were unsuccessful at inducing chimerism. Similarly, studies where we omitted anti-CD40L were unsuccessful at inducing chimerism (data not shown). Perhaps anti-CD40L is required in this protocol to tolerize any nondeleted T cells as our data suggest preexisting T cells are primarily responsible for the development of split tolerance in NOD mice. Previously, others have found 3Gy TBI, anti-CD8 and anti-CD40L without CD4 depletion could induce chimerism in 50% of mice and make them tolerant toward skin grafts (30). In this case, they concluded the effect of anti-CD40L is to tolerize preexisting CD4 T cells. The mice in this experiment appeared fully tolerant; however, the donor–recipient strain combinations were partially MHC matched (B6 to NOD), which may reduce split tolerance (47). In contrast, we have used the fully mismatched C3H strain as a donor, and both NOD T and NK cells act as potent barriers to successful BMT in this combination (29). The requirement of anti-CD40L in our experiments argues against the previous interpretation that NOD mice resist costimulation-based tolerance inducing therapy (27,49). Upon review of these previous studies, NOD resistance to tolerance induction may be instead because of a resistance of the donor-specific transfusion aspect of these protocols; as there was no difference in allogeneic skin graft survival given to NOD and B6 mice when DST was omitted and only anti-CD40L administered (49). Although chimerism induction protocols involving T cell depletion are currently more clinically relevant (21) than the use of anti-CD40L (50,51), the potential to avoid split tolerance and powerful tolerizing effects of interrupting the CD40-CD40L pathway cannot be ignored. As such, novel costimulation blockade therapies using anti-CD40 are being tried with success at inducing chimerism (52,53). It will be of interest to see if the eventual loss of chimerism seen in this protocol could be avoided if some amount of T cell depletion is performed.

Aside from being successful at generating chimerism across full MHC barriers in NOD mice, which are both autoimmune and have a resistance to tolerance induction, the chimerism induction protocol we used, based on T cell depletion, has significant advantages over other approaches. First, the protocol is irradiation-free, minimalistic and clinically feasible. T cell depletion and rapamycin are currently used for solid organ transplant (54,55), and busulfan has been used in human patients (56,57) and in nonhuman primate bone marrow transplant recipients (53). Although anti-CD40L is not available clinically because of risk of thromboembolic complications, alternative promising approaches with costimulatory blockade are being developed (52,53). Second, our protocol generates mixed chimerism across fully mismatched barriers and autoimmune barriers. Many other protocols for generating chimerism were only tested in the setting of some degree of MHC matching between donor and recipient (30,58,59); a less clinically feasible combination. Although there has been previous successes at generating chimerism in NOD mice using fully mismatched donor–recipient strains, there was a tendency for chimerism to become full (60), or eventually lost if not induced with megadoses of bone marrow and the infusion of donor CD8 T cells (61). Third, the chimerism obtained with our protocol is multilineage and stable long term. Although the clinical relation of sustained chimerism to solid organ tolerance is controversial, and often associated with split tolerance (21,22), we have shown long-term acceptance in NOD mice of both allogeneic hematopoietic cells and islet grafts.

We have identified a potentially clinically relevant induction protocol based on T cell depletion that generates NOD mixed chimeras and have also implicated preexisting T cells as a major cause of split tolerance in the NOD mice. The need to overcome additional factors that enhance the likelihood of split tolerance, such as transplantation into highly sensitized recipients, can be anticipated.


We thank Drs Mark Cattral, Thomas Mueller and John Elliott for their critical review of the manuscript. We thank Baoyou Xu for his technical assistance. This work was funded by a grant from the Canadian Diabetes Association (to CCA) and the Edmonton Civic Employees Charitable Assistance Fund and awards from the Alberta Heritage Foundation for Medical Research (to DPA and CCA).


The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation. DPA participated in research design, performance of research, data analysis and writing the paper. RP and AMJS participated in performance of research. CCA participated in research design, data analysis and writing the paper.