Induction of Mixed Chimerism Depletes Pre-existing and De Novo–Developed Autoreactive B Cells in Autoimmune NOD Mice

The autoimmune response that gives rise to type 1 diabetes (T1D) is caused by autoreactive T and B cells that damage the insulin-producing β-cells of the pancreas (1). The importance of B cells in disease pathogenesis was demonstrated by the following: The appearance of autoantibodies directed against islet antigens have long been used as a diagnostic in prediabetic patients, and anti-insulin antibodies are commonly found in NOD mice (2). B-cell knockout mice are highly resistant to diabetes development (3,4), and withdrawal of B cell–depleting antibodies can retrigger insulitis progression (5). B cells are critical for the initial priming of certain autoreactive clones (6,7), although T cells alone are sufficient to transfer diabetes into secondary NOD-SCID recipients (8). These studies indicate that B cells play an important role in early priming and expansion of autoimmune responses. However, therapies that target B cells, including blockade of B-cell receptor (BCR) signaling (9) and depletion of B cells by anti-CD20 or anti–CD22-cal, have yielded variable results (10–15). On the other hand, targeting T cells by anti-CD3 was also found to have limited effect in new-onset diabetic patients (16). Therefore, it has been recently proposed that a therapy that can simultaneous tolerize both T and B cells is required for effective reversal of autoimmunity in T1D (1,17,18).

We have demonstrated that induction of mixed chimerism via hematopoietic cell transplantation (HCT) is an effective therapy to reverse autoimmune T1D in NOD mice, using a radiation-free nontoxic anti-CD3/CD8 conditioning regimen (19–26). Induction of chimerism in new-onset diabetic mice could eliminate insulitis and allow residual β-cell replication to reverse diabetes (25). Although induction of chimerism alone was not able to reverse late-stage T1D that has little or no residual islet β-cells, combination therapy with administration of growth factors was able to augment β-cell neogenesis and cure a majority of late-stage T1D (23). In addition, induction of chimerism allowed a small numbers of donor islets (1 of 20 of regular dose) to reverse late-stage T1D after implanting islet grafts in the liver or native pancreas (26). Therefore, induction of mixed chimerism is an important curative therapy for late-stage T1D.

The cure of autoimmunity by induction of major histocompatibility complex (MHC)–mismatched mixed chimerism indicates that mixed chimerism can simultaneously correct both T- and B-cell autoimmunity in NOD mice. We recently observed that mixed chimerism mediates thymic negative selection of host-type autoreactive T cells in a mismatched MHC II–dependent manner (22), but how mixed chimerism impacts autoreactive B-cell tolerance remains unclear. The current study was designed to seek answers to this important question.

Female NOD/LtJ, thymectomized NOD/LtJ, wild-type (WT) C57BL/6, Rag2^−/− C57BL/6, and Igμ^−/− C57BL/6 were purchased from The Jackson Laboratory (Bar Harbor, ME). 125Tg-NOD mice that harbor anti-insulin Ig transgenes were obtained from Dr. James W. Thomas (Vanderbilt University, Nashville, TN). All mice were maintained in a pathogen-free room at City of Hope Research Animal Facilities. All animal procedures were approved by our institutional committee.

These procedures have been described in our previous publications (23,25) and in the current figure legends.

Monoclonal antibodies for CD45.2(104), CD45.1(A20), CD21/35(7G6), BrdU, Mac1(M1/70), Gr1(RB6-8C5), CD4(RM4-5), IgM(R6-60.2), TCRβ(H57-597), CD24(M1/69), CD138(281-2), CD8α(53-6.7), and B220(RA3-6B2) were all purchased from BD PharMingen (San Diego, CA). Antibodies for CD24(30-F1), CD4(GK1.5), IgM(II/41) IgD (11–26), BP1(6C3), CD19(eBio1D3), CD43(eBioR2/60), and CD23(B3B4) as well as efluor 450–labeled streptavidin were purchased from eBioscience (San Diego, CA). Dead cells were excluded using DAPI or aqua fluorescent reactive dye (Invitrogen, Carlsbad, CA). FACS analyses were performed using the CyAn immunocytometry system (Dako Cytomation, Fort Collins, CO), and data were analyzed using FlowJo software (TreeStar, San Carlos, CA) as described in our previous publications (19–32).

As anti-CD3/CD8–based conditioning regimen is a nonmyeloablative regimen, host-type B cells present in the mixed chimeras include both pre-existing and de novo–developed B cells. We first evaluated the impact of induction of mixed chimerism on pre-existing autoreactive B cells. Accordingly, prediabetic NOD mice were induced to develop mixed chimerism by conditioning with anti-CD3/CD8 and transplanting with bone marrow (BM) and CD4⁺ T-depleted spleen cells from C57BL/6 donors as previously described (22,23). Mixed chimerism status was determined upon sacrifice by coexistence of de novo–developed CD45.2⁺ donor-type and CD45.1⁺ host-type thymocytes in the thymus as well as CD45.2⁺ and CD45.1⁺B220⁺ or Mac1⁺Gr1⁺ mononuclear cells in BM, as previously described (22,23) and shown in Supplementary Fig. 1.

Since there is not yet a marker for identifying memory B cells in mice, and the pre-existing autoreactive B cells can include CD19^LoCD138^Hi plasma cells and CD19^HiCD138^Lo preplasma cells (33), we kinetically measured the changes of percentage and yield of CD19^LoCD138^Hi plasma and CD19^HiCD138^Lo preplasma cells in the spleen of mixed chimeric NOD mice on days 30, 45, and 60 after HCT. There were ∼0.5% CD19^LoCD138^Hi plasma cells and 7.5% CD19^HiCD138^Lo preplasma cells and a total of ∼8% in the spleen of control NOD mice given anti-CD3/CD8 conditioning only (Fig. 1A and B). In contrast, 30 and 45 days after HCT, the total percentage of host-type CD45.1⁺ plasma and preplasma B cells was reduced to <0.5%, a >10-fold reduction, and the total yield was reduced ∼100-fold (P < 0.01) (Fig. 1A and B). Interestingly, by 60 days after HCT, the percentage of CD19^HiCD138^Lo preplasma cells among residual host-type B cells recovered to levels similar to control mice, although the yield was still lower than that in controls (Fig. 1A and B).

In BM, CD19^HiCD138^Lo B cells can be immature developing B cells (34). We observed that the CD45.1⁺ host-type CD19^LoCD138^Hi plasma B cells in BM were reduced ∼10-fold in percentage as well as in yield at 30, 45, and 60 days after HCT (P < 0.01) (Fig. 1C and D). Whereas the yield and percentage of CD19^HiCD138^Lo cells on days 30 and 45 are significantly lower in mixed chimeras (P < 0.001), by day 60 after transplantation, these had returned to the levels seen in anti-CD3/8 conditioning alone controls (Fig. 1C and D).

In addition, by 60 days after HCT, whereas serum anti-insulin autoantibody was high in the control mice, it was not detectable in the mixed chimeric NOD mice that had recovery of host-type B cells (P = 0.012) (Fig. 1E, left panel); in contrast, the total serum IgG of the mixed chimeric mice was similar to control mice (Fig. 1E, right panel). Concurrent with reversal of B-cell autoimmunity, analysis of pancreatic sections 60 days after HCT revealed elimination or marked reduction of insulitis, compared with anti-CD3/CD8–conditioned controls (Fig. 1F). Taken together, induction of mixed chimerism with MHC-mismatched BM depletes host-type pre-existing autoreactive B cells without causing an overall B-cell immunodeficiency.

Next, we evaluated the impact of induction of mixed chimerism on de novo development of host-type B cells in the BM. BM B220⁺ B cells can be divided into CD43⁺ and CD43⁻ groups; the B220⁺CD43⁺ group is the earliest precursor of B cells (pro-B), and the B220⁺CD43⁻ group includes IgM⁻IgD⁻ pre-B, IgM^HiIgD^Lo immature B, and IgM^LoIgD^Hi mature recirculating B cells (35,36) (Fig. 2). We compared the percentage and yield changes of the BM B220⁺ B subsets at days 30, 45, and 60 after HCT. Compared with control mice, by 30 days after HCT, the percentage (∼40%) of CD43⁻ B cells (including pre-B, immature B, and mature B) among host-type B220⁺ cells in the mixed chimeras was approximately twofold lower, and the percentage (∼60%) of CD43⁺ pro-B cells was approximately threefold higher (P < 0.05) (Fig. 2A and B). In control mice, although the percentage of CD43⁺ versus CD43⁻ B cells among CD45.1⁺B220⁺ cells remained constant, ∼20 vs. 80% (Fig. 2A and B), the yields of each subset started high (∼0.5 × 10⁶ vs. 2.5 × 10⁶) and declined to ∼0.2 × 10⁶ vs. 1 × 10⁶ by 60 days after conditioning (Fig. 2B, right panels). In mixed chimeric recipients, whereas the percentage of CD43⁺ vs. CD43⁻ host-type B cells was ∼50 vs. 50% at day 30, the former decreased and the latter increased as time went on, and it became 30 vs. 70% (Fig. 2A and B, left panels). The yield of CD43⁺ pro-B cells of the chimeras (∼0.2 × 10⁶) at day 30 was twofold lower than that of control mice, but it increased to ∼0.5 × 10⁶ and became ∼2.5-fold higher than that of the control mice (P < 0.05) (Fig. 2B, right). The yield of CD43⁻ B cells (including pre-B, immature B, and mature B) of the chimeras (∼0.2 × 10⁶) at day 30 was 10-fold lower than that of control mice (P < 0.001), but it increased to the same level (∼1 × 10⁶) by 60 days after HCT (Fig. 2B, right).

Furthermore, we analyzed the host-type CD43⁻ B-cell subsets by their expression of IgM versus IgD and divided the CD43⁻ B cells into IgM^LoIgD^Hi recirculating mature B cells, IgM^HiIgD^Lo immature B cells, and IgM^LoIgD^Lo pre-B cells. The percentage and yield of the recirculating mature B cells in mixed chimeras was markedly lower than that of control mice (P < 0.001) (Fig. 2C and D) at 30–60 days after HCT. The percentage of immature B cells in the chimeras was similar to that of control mice 30–45 days after HCT, but the yield of the immature B cells in the chimeras was significantly lower than that of control mice at the same time period. However, by 60 days after HCT, the percentage of the immature B cells became higher (P < 0.05) in the chimeras and the yield became similar (Fig. 2C and D). The percentage of pre-B cells in the chimeras was significantly higher than control mice at 30–60 days after HCT. Although the yield of the pre-B cells in the chimeras was significantly lower at 30–45 days after HCT, it recovered and became similar to that of control mice by 60 days after HCT (Fig. 2C and D). Therefore, the reduction of B220^HiCD43^Lo cells in the BM of MHC-mismatched mixed chimeras results from loss of IgM^LoIgD^Hi recirculating mature B cells but not immature or pre-B cells in the BM.

Next, we evaluated the impact of induction of mixed chimerism on immature B cells in the spleen. Different approaches have been reported for subgrouping splenic immature B cells. Although none of the gating strategies are likely ideal (with overlap of populations), we chose to use the gating strategy that has been used by several groups (37–39). As shown in Fig. 3A, splenic immature B cells were divided into CD24^HiCD21^Lo transitional 1 (T1), CD24^IntCD21^Hi transitional 2 (T2)/marginal zone (MZ), and CD24^LoCD21^Lo follicular (FO) subsets. The CD24^IntCD21^Hi fraction was further split into CD23⁺ T2 and CD23⁻ MZ subsets. It was reported that there was a relative expansion of T2 and MZ subsets in NOD spleen, which has been associated with an expansion of autoreactive B cells (37,40). Consistently, we found that, compared with nonautoimmune C57BL/6 mice, there was a reduction of T1 and expansion of T2 and MZ subsets by percentage and yield in NOD mice. Whereas the ratio of T1 to T2 in C57BL/6 was ∼3, it became ∼1 in NOD mice (Supplementary Fig. 2A and B). The abnormal T1-T2 ratio in NOD mice is an intrinsic B-cell defect, as a similar T1-T2 ratio was observed in T cell–depleted NOD, T cell–depleted thymectomized NOD, CD4⁺ T-deficient MHC II^−/− NOD, as well as nonautoimmune NOR mice (Supplementary Fig. 2A and B).

We also observed that, as compared with nonchimeric control NOD mice, induction of MHC-mismatched mixed chimerism with C57BL/6 donor BM increased the percentage of T1 and decreased the percentage of T2 and MZ subsets among host-type B cells in the spleen of mixed chimeras (P < 0.01). The proportion of T1, T2, and MZ B subsets among host-type B cells was similar to that of donor-type B cells in the chimera, and no significant changes of the percentage of FO subset were observed (Fig. 3A, top panel). On the other hand, the yield of T1 and FO subsets of host-type B cells in the chimera was markedly reduced as compared with the donor-type B in the chimeric mice or control NOD mice (P < 0.01), although the T2 and MZ yield was similar to donor type (Fig. 3B, middle panel). Finally, as compared with control NOD mice, the ratio of T1 to T2 among host-type B cells was significantly increased from 1 to 5 (P < 0.01) (Fig. 3B, lower panel), and the latter was similar to the ratio in nonautoimmune donor mice (Supplementary Fig. 2B). Taken together, the results indicate that 1) induction of MHC-mismatched mixed chimerism corrects the intrinsically abnormal ratio of T1 to T2 in the autoimmune NOD mice, and 2) in the mixed chimera, donor-type B cells dominate repopulation of splenic B cells.

Since host-type pre-B and immature B cells in the BM of mixed chimera were not reduced (Fig. 2), but immature T1 B cells were markedly reduced in the spleen, although the T1-T2 ratio was increased (Fig. 3), we attempted to find out the cause for the reduction of T1 cell numbers. Although T1 B cells are generally more susceptible to apoptosis after BCR cross-linking than T2 B cells (39), NOD T1 B cells were shown to be less susceptible to apoptosis than T1 B cells from nonautoimmune mice (37). Thus, we tested whether, with the relative increase in T1 B cells among host-type B cells in mixed chimeric animals, there was increase in background apoptosis in this subset compared with T1 B cells in anti-CD3/CD8–conditioned controls. We found that MHC-mismatched mixed chimerism did significantly augment host-type T1 B-cell apoptosis (P = 0.0001) (Fig. 3C and D).

In order to directly examine what happens to an autoreactive B-cell clone, we used transgenic 125Tg-NOD mice that express BCR with transgenic heavy and light chains specific for insulin (41). Almost all B cells in 125Tg-NOD mice are insulin-reactive B cells (41). We also observed that de novo–developed B cells in lethal total body irradiation (TBI)–conditioned NOD mice given T and B cell–depleted (TBCD) BM from 125Tg-NOD mice were all insulin reactive (Supplementary Fig. 3), indicating that HCT procedures do not interfere with the development of insulin-specific 125Tg B cells, and that almost all host-type B cells in the chimeric 125Tg recipients are expected to be insulin-reactive B cells.

Accordingly, we induced mixed chimerism in transgenic 125Tg-NOD mice by anti-CD3/8 conditioning regimen and transplants from MHC-mismatched C57BL/6 donors (Supplementary Fig. 4). Sixty days after transplantation, in MHC-mismatched mixed chimeras, ∼40% of splenic host-type B cells were T1 B cells; this was a fourfold increase compared with control mice (P < 0.001), but similar to that among donor-type B cells (Fig. 4A and B, top panel). In contrast, the percentage of T2 and MZ B cells among host-type B cells was significantly reduced (P < 0.01) and became similar to that among donor-type B cells in the chimera. There was no significant difference for the FO B subset (Fig. 4B, top panel). On the other hand, the yields of the T1, T2, MZ, and FO subsets of host-type B cells were all markedly lower than that of donor-type B cells in the chimera or the control NOD mice (Fig. 4B, middle panel). The ratio of T1 to T2 (5:1) in the chimeras was markedly increased (P < 0.01), as compared with control mice (1:3), and the former was similar to the ratio of T1 to T2 of the donor-type B cells in the chimeras (Fig. 4B, lower panel). Furthermore, T1 B cells in the mixed chimeras had significantly increased apoptosis as compared with control mice (P = 0.001) (Fig. 4C and D). These results indicate that induction of MHC-mismatched mixed chimerism can induce apoptosis of autoreactive B cells.

Induction of mixed chimerism under the radiation-free anti-CD3/CD8 conditioning regimen requires infusion of donor CD8⁺ T cells to facilitate engraftment, and the donor CD8⁺ T cells can kill host T and B cells and possibly cause the apoptosis of host-type B cells. Unfortunately, we could not directly test the role of donor CD8⁺ T cells in the apoptosis of de novo–developed B in the anti-CD3/CD8–conditioned NOD mice, as no chimerism can be established in the absence of donor CD8⁺ T cells (21). Alternatively, we established mixed chimerism with TBCD donor and host BM cells in lethally irradiated NOD mice with or without addition of donor CD8⁺ T cells. Accordingly, lethally irradiated NOD mice were reconstituted with TBCD BM from syngeneic WT NOD or TBCD BM from syngeneic WT NOD and MHC-mismatched C57BL/6 mice. Thirty days after HCT, half of the mixed chimeric recipients were injected with donor-type CD8⁺ T cells (2 × 10⁶).

Sixty days after HCT, the mixed chimerism of the recipients with or without infusion of donor CD8⁺ T cells was confirmed (Supplementary Fig. 5). We found that the apoptosis of the T1 B cells of the chimeras was significantly increased as compared with that of T1 B of control mice (P < 0.001) (Fig. 5A), but no significant difference was observed between the mixed chimeras with or without addition of donor CD8⁺ T cells (Fig. 5A). Furthermore, the increase of apoptosis of host-type T1 B cells in the chimeras was associated with the increase of percentage of donor-type B cells among total B cells (Fig. 5B), and >50% of B cells should be of donor type in order for this augmented apoptosis to occur (Fig. 5E). In contrast, the percentage of donor B cells had no impact on the apoptosis of host-type follicular B cells (Fig. 5C and D). These results indicate that de novo–developed donor B cells instead of donor CD8⁺ T cells in transplants may mediate the augmentation of apoptosis of host-type immature T1 B cells in the MHC-mismatched mixed chimeras.

To test this hypothesis, lethal TBI-conditioned 125Tg-NOD mice were transplanted with syngeneic TBCD BM with or without TBCD BM from MHC-mismatched WT, T and B cell–deficient Rag-2^−/−, or B cell–deficient Igµ^−/− C57BL/6 donors. Sixty days after transplantation, mixed chimeric 125Tg-NOD recipients (Supplementary Fig. 6) were compared for the apoptosis of T1 B cells. We first evaluated WT BM and Rag-2^−/− BM chimeras and found that MHC-mismatched mixed chimeras with WT BM but not Rag-2^−/− BM had increased host-type T1 B-cell apoptosis (P < 0.0001) (Fig. 6A and C). Then, we compared chimeras with WT and Igµ^−/− donor BM and found that mixed chimeras with Igµ^−/− donor BM did not have enhanced host-type T1 B apoptosis (Fig. 6B and C). Taken together, the results demonstrate that de novo–developed donor-type B cells augment apoptosis of host-type autoreactive T1 B cells in MHC-mismatched mixed chimeras.

We have shown that induction of MHC-mismatched mixed chimerism depleted both pre-existing and de novo–developed autoreactive B cells in autoimmune NOD mice. The depletion of pre-existing plasma and preplasma B cells (including autoreactive B cells) early after HCT in the MHC-mismatched mixed chimeric NOD mice established with radiation-free anti-CD3/CD8 conditioning should result from graft versus autoimmunity effect mediated by alloreactive CD8⁺ T cells in transplants. The recipients were injected with CD4⁺ T-depleted spleen cells that contained ∼10% CD8⁺ T cells. Alloreactive CD8⁺ T cells were able to effectively kill host-type T and B cells (20,25), but they also caused graft versus host disease (GVHD) in TBI-conditioned recipients (42). However, in the anti-CD3/CD8–conditioned recipients, the alloreactive donor T cells are confined to lympho-hematopoietic tissues and are prevented from migration into GVHD target tissues to cause GVHD. The infused alloreactive CD8⁺ T cells in the anti-CD3/CD8–conditioned recipients gradually became tolerized or regulatory (19,24). This can explain our observation that although 30–45 days after HCT, host-type plasma and preplasma cells and IgM^LoIgD^Hi mature B cells in the periphery were depleted, they started to increase 60 days after HCT. In addition, the presence of host-type immature B and precursor cells in the BM and de novo–developed host-type CD4⁺CD8⁺ thymocytes in the thymus indicate that the injected CD8⁺ T cells only transiently attacked host-type hematopoietic cells early after HCT and they become tolerized by 30 days after HCT.

We observed that induction of MHC-mismatched mixed chimerism in the absence of donor CD8⁺ T cells in TBI-conditioned WT or insulin-specific BCR transgenic NOD recipients augmented host-type T1 B-cell apoptosis, and addition of donor CD8⁺ T cells did not have significant impact. However, induction of mixed chimerism with T and B cell–deficient Rag-2^−/− or B cell–deficient Igµ^−/− donor BM failed to augment NOD T1 B-cell apoptosis. These demonstrate that the presence of donor B but not injected donor CD8⁺ T cells or de novo–developed donor T cells augments NOD immature autoreactive T1 B-cell apoptosis. The latter effect may result from nonautoreactive donor B cells competing with NOD autoreactive B cells for survival factors such as BAFF, which makes autoreactive B cells more susceptible to apoptosis (43).

NOD mice have a collapse of T1 immature B cells in the spleen, such that autoreactive B cells at the T1 stage have less survival competition with nonautoreactive B cells and rapidly progress into T2 and MZ B cells, and become mature autoreactive B cells (37,40). The presence of nonautoreactive B cells, such as nonautoreactive donor B cells in the mixed chimeric NOD recipients, could outcompete the autoreactive B cells for survival factors, causing autoreactive B cells to go through apoptosis, as indicated by previous publications (43–49). In the mixed chimeric NOD mice, the presence of nonautoreactive donor B cells can take away survival factors and result in apoptosis of autoreactive T1 B cells. We should point out that Meyer-Bahlburg et al. (39), however, did further separate the T1 subset into CD21^Lo and CD21^Int and identified a distinct population of “T2” cells that are highly susceptible to BAFF levels. As our “T1” gating does include this subset, it could help explain the increased apoptosis observed in our T1 B cells.

We should also point out that although the immature host-type B-cell numbers in the BM of the mixed chimeric NOD recipients were not reduced, and although there was a marked increase in the percentage of T1 B cells among host-type B cells in the spleen, the host-type immature B-cell numbers in the spleen were markedly reduced, and there was a dominance of donor B cells. Therefore, although host-type autoreactive T1 B cells are more susceptible to apoptosis induced by donor B cells, we cannot rule out the possibility that splenic donor B cells also augment the apoptosis of nonautoreactive host-type T1 B cells at a certain level, especially when donor-type B cells are predominant.

In summary, as described in Fig. 7, we propose that early after HCT and before donor stem cell engraftment, the infused donor alloreactive CD8⁺ T cells kill almost all pre-existing host-type plasma and preplasma auto- and nonautoreactive B cells. Thereafter, the injected donor CD8⁺ T cells become tolerant to the host cells. After donor stem cell engraftment, ∼15–30 days after HCT, both donor- and host-type de novo–developed B cells start to develop in the mixed chimeric recipients. The dominance of donor B cells takes away growth factors, resulting in apoptosis of host-type B cells, especially host-type autoreactive B cells, since they are more susceptible than nonautoreactive B cells. Therefore, the autoimmunity mediated by host-type B cells is prevented. In addition, the mixed chimerism also tolerized the host-type autoreactive T (22). Taken together, induction of mixed chimerism tolerizes not only autoreactive T but also autoreactive B cells, and induction of mixed chimerism represents one novel regimen that can simultaneously restore both T- and B-cell tolerance in autoimmune recipients.

Acknowledgments. The authors are grateful to Dr. Arthur Riggs (City of Hope [COH]) for his encouragement and support of this research. The authors thank Lucy Brown and her staff at COH Flow Cytometry Facility, Sofia Loera and her staff at COH Anatomic Pathology Laboratory, and Dr. Richard Ermel and his staff at COH Research Animal Facility for technical and logistical support. The authors are also grateful to Chrys Hulbert and Dr. James W. Thomas for providing 125Tg-NOD mice and for Dr. Thomas’ review of the manuscript.

Funding. This work was supported in part by a private donation from Mr. and Mrs. Arthur and Judith Lubin.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.J.R. designed and performed experiments, contributed to data analysis, and wrote the manuscript. M.W. and M.Z. assisted in experiments. D.Z. designed experiments, contributed to data analysis, and wrote the manuscript. D.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.