Loss of tolerance is considered to be an early event that is essential for the development of autoimmune disease. In contrast to this expectation, autoimmune (type 1) diabetes develops in NOD mice that harbor an anti-insulin Ig transgene (125Tg), even though anti-insulin B cells are tolerant. Tolerance is maintained in a similar manner in both normal C57BL/6 and autoimmune NOD mice, as evidenced by B cell anergy to stimulation through their Ag receptor (anti-IgM), TLR4 (LPS), and CD40 (anti-CD40). Unlike B cells in other models of tolerance, anergic 125Tg B cells are not arrested in development, and they enter mature subsets of follicular and marginal zone B cells. In addition, 125Tg B cells remain competent to increase CD86 expression in response to both T cell-dependent (anti-CD40) and T cell-independent (anti-IgM or LPS) signals. Thus, for anti-insulin B cells, tolerance is characterized by defective B cell proliferation uncoupled from signals that promote maturation and costimulator function. In diabetes-prone NOD mice, anti-insulin B cells in this novel state of tolerance provide the essential B cell contribution required for autoimmune β cell destruction. These findings suggest that the degree of functional impairment, rather than an overt breach of tolerance, is a critical feature that governs B cell contribution to T cell-mediated autoimmune disease.
Tolerance in the B cell repertoire is governed by the nature of the self-Ag and by the stage in B cell development at which Ag encounter occurs. Many self-reactive B cells escape clonal deletion and receptor editing in the bone marrow and are permitted to enter the peripheral repertoire. For autoreactive B cells that exit the bone marrow, engagement of their BCR by self-Ag induces a program of gene expression that maintains peripheral B cell tolerance (1). Tolerant B cells fail to differentiate during Ag-specific immune responses (functional silencing) and are unable to respond to signals delivered through their BCR (anergy) (2, 3). In model systems studied to date, silencing and anergy are associated with arrested B cell development, which prevents tolerant B cells from completing their normal maturation in the periphery. For example, B cells that are rendered tolerant in the anti-hen egg lysozyme (anti-HEL)3/HEL model fail to develop normally in the periphery, as manifested by a lack of marginal zone (MZ) B cells and by low levels of expression of CD23 on follicular (FO) B cells (4, 5, 6). Developmental arrest may facilitate removal of tolerant B cells from the repertoire by impairing their ability to compete with normal B cells for FO niches or by increasing their susceptibility to apoptosis after activation (6, 7, 8). Studies of B cells in mice that harbor anti-DNA H chain transgenes also show that developmental arrest culls autoreactive B cells in nonautoimmune strains (9, 10). When the same anti-DNA transgenes are present in autoimmune strains, the autoreactive B cells are not developmentally arrested, and anti-dsDNA B cells enter a functional autoimmune repertoire (11). Thus, genetic predisposition to systemic autoimmune disease includes mechanisms that alter the fate of autoreactive B cells and permit maturation of B cells that are normally arrested in their development.
In contrast to autoimmunity to Ags expressed systematically, organ-specific autoimmune diseases, such as type 1 diabetes mellitus (T1DM), are characterized by breaches of immune tolerance to specific proteins and polypeptides, leading to a selective attack on target organs. These disorders share many features of systemic autoimmune diseases, such as lupus, including associations with MHC alleles, regulation of coreceptor molecules, and profiles of expressed genes (12, 13). Thus, genetic factors that alter B cell tolerance in systemic autoimmune diseases may also contribute to the loss of tolerance in organ-specific autoimmune disease. Alternatively, organ-specific autoimmune disease, by virtue of its T cell-mediated nature, may not be associated with overt breaches in B cell tolerance that accompany systemic autoimmune disorders. Because many autoantigens in organ-specific autoimmune disease are present only at low levels, the fate of B cells that encounter such physiologic autoantigens will probably differ from that in model systems, such as anti-DNA and anti-HEL, in which Ag exposure is continuous.
To understand how tolerance is maintained for a physiologic autoantigen and to determine how the fate of such B cells is governed by an organ-specific autoimmune disease, we introduced anti-insulin BCR transgenes (125Tg) into autoimmune NOD mice and into normal C57BL/6 (B6) mice. NOD mice are a spontaneous model of T1DM in which insulin-producing β cells are destroyed by autoimmune T cells. Detection of autoantibodies to insulin and other islet Ags in the prodrome of T1DM indicates that loss of tolerance in the B cell repertoire is a harbinger of β cell destruction (14, 15). Although Abs do not directly cause disease, a number of studies clearly demonstrate that B lymphocytes are essential for the autoimmune process in NOD (16, 17, 18). The potential roles for B cells in T1DM include Ag presentation, maintenance of T cell homeostasis, and transplacental passage of Ig (19, 20, 21). In addition, NOD mice that harbor irrelevant Ig transgenes (e.g., anti-HEL) do not develop diabetes, suggesting that Ag specificity within the B cell repertoire is also important for disease progression (22, 23). Although anti-insulin 125Tg binds rodent insulin with modest affinity (5 × 10−6 M−1), we have previously documented that this interaction is sufficient to functionally silence anti-insulin B cells and prevent Ab production after T cell-dependent immunization of B6 mice (24).
Investigating the fate and function of anti-insulin 125Tg B cells in NOD mice provides a unique opportunity to understand how B cell tolerance is regulated in the context of an organ-specific autoimmune disease. Using this approach, we have discovered a novel state of B cell tolerance in which BCR signals that maintain anergy are uncoupled from the mechanisms that govern peripheral B cell maturation and expression of costimulator molecules. Although the degree of anergy in 125Tg B cells is comparable to that found in major models of B cell tolerance, the fate of anti-insulin B cells differs markedly from that in other models, in that 125Tg B cells mature into phenotypically normal MZ and FO subsets. The significance of this novel tolerance state is apparent in 125Tg NOD mice, in which anergic anti-insulin B cells provide the necessary B cell functions for NOD mice to develop diabetes. These findings indicate that for B lymphocytes to contribute to T1DM, they do not have to be fully functional; rather, the peripheral tolerance for B cells may include states in which retained function is sufficient to support progression of the autoimmune disease without an overt breach of tolerance. Understanding functional differences within the tolerant states will be the key for successful use of tolerance regimens for the treatment of autoimmune disorders.
Materials and Methods
The Ig transgenic mice used in this study harbor conventional anti-insulin transgenes on B6 or NOD backgrounds as described previously (23, 25). NOD and B6 mice were purchased from Taconic Farms. Hemizygous lines expressing H and L chain transgenes derived from anti-insulin mAb125 were maintained by backcrosses to the parental strain and monitored by PCR on tail DNA. Anti-insulin 125Tg mice were generated by intercrossing H and L chain-expressing lines, and nontransgenic littermates were used as controls. Original founder lines were selected on the basis of low copy number (one or two) on Southern blot, and when the phenotypes of the lines proved identical, studies were consolidated to single lines. All data are derived from lines that have been backcrossed >18 generations to B6 or NOD, and for NOD mice microsatellite markers were used to verify that known insulin-dependent diabetes loci were present (16, 23). Female NOD mice were used for diabetes outcome studies, and mice were considered diabetic if two consecutive blood sugars were >200 mg/dl using a One Touch glucometer (Lifescan). All mice were housed under specific pathogen-free conditions, and all studies were approved by the institutional use and animal care committee of Vanderbilt University.
Proliferation and cell culture
For proliferation assays, B cells were purified from spleen after RBC lysis in Tris-NH4Cl using Miltenyi beads (Miltenyi Biotec) and anti-CD43 mAb (BD Pharmingen). Purified B cells were then cultured in complete medium at 2 × 10−5 cells/well in 96-well, flat-bottom plates (Corning). Complete medium consisted of RPMI 1640 (Invitrogen Life Technologies) containing 10% FBS (HyClone), glutamine, gentamicin, and 2 × 10−5 M 2-ME (Invitrogen Life Technologies). Wells were pulsed with 1 μCi of [3H]thymidine (NEN) on day 2 and harvested on day 3 using a semiautomated cell harvester (Skatron). [3H]Thymidine uptake was measured by scintillation counting after stimulation. Mitogens include F(ab′)2 goat anti-mouse μ-chain (Jackson ImmunoResearch Laboratories). LPS, Escherichia coli B (Difco), anti-CD40 (HM40-3; BD Pharmingen), and IL-4 (gift from Dr. T. M. Aune, Vanderbilt University, Nashville, TN). Results are reported as the mean ± SD for triplicate determinations. For analysis of activation markers, purified B cells were cultured in 1-ml wells in the presence or the absence of mitogen for 24–96 h. Cells were washed twice before staining for flow cytometry.
Flow cytometry and Abs
For flow cytometric analysis, spleen cells or in vitro cultured B cells were washed in PBS containing 1% BSA and 0.1% sodium azide and stained with the indicated mAb. Analysis was performed on a FACSCalibur flow cytometer (BD Biosciences). Ab reagents (BD Pharmingen) were reactive with B220 (6B2), IgMa (DS-1), IgMb (AF6-78), CD19 (1D3), CD21/CD35 (7G6), CD22 (Cy34.1), CD23 (B3B4), CD40 (HM40-3), CD44 (IM7), CD69 (H1.2F3), CD80 (16-10A1), CD86 (GL-1), I-Ab (25-9-3), and I-Ak (10-3.6). WinMDI 2.8 software (Dr. J. Trotter, Scripps Institute, San Diego, CA) was used for analysis.
For immunofluorescence microscopy, spleens were removed and soaked in 30% sucrose for 4 h, then snap-frozen in OCT (Sakura Finetek). Cryosections (8 μm) were adhered to slides and blocked with PBS containing 1% BSA and 5% normal goat sera. Sections were then stained for B cells using biotinylated anti-B220 (BD Pharmingen; 6B2) and avidin-Texas Red (Molecular Probes). To delineate the MZ, metallophilic macrophages were stained with FITC-conjugated MOMA-1 (Serotec). After washing in PBS, slides were air-dried and mounted using ProLong anti-fade (Molecular Probes), and sections were visualized using an Olympus BX60 epifluorescence microscope. Images were captured using a CCD camera and MagnaFire software (Optronics).
Anti-insulin transgenes are effectively expressed in NOD mice
IgM transgenes derived from anti-insulin mAb125 (125Tg) were introduced into NOD mice by multiple backcrosses. Previous studies suggest that the B cell repertoire of NOD mice is subject to selection defects and may not be generated normally (26, 27); therefore, the expression of IgM alleles on B cells from NOD and B6 mice harboring 125Tg anti-insulin (IgMa) was examined. As shown in Fig. 1, transgene IgMa was expressed efficiently in splenic B cells (B220+) from B6 and NOD mice. The few IgMa-negative B cells present corresponded to B cells that expressed IgMb (not shown). Surface expression of the transgene was not depressed in either strain and was comparable to that observed in nontransgenic controls. Allelic exclusion was highly efficient in both strains, but the transgene was consistently expressed in slightly more B cells in NOD (97.6 ± 1.8%; n = 12) compared with B6 (94.7 ± 2.1%; n = 14) mice (Fig. 1, upper panels). In addition, the mean fluorescence intensity (MFI) for surface IgMa (125Tg) was increased when 125Tg was expressed in NOD B cells (MFI, 225 ± 8) compared with B cells from B6 mice (MFI, 180 ± 7). These differences are shown in representative histograms (Fig. 1, lower left) that compare surface IgMa expression in 125Tg B6 and 125Tg NOD B cells. The difference in surface IgM expression was not simply strain-related, because IgMb expression did not differ when B cells from nontransgenic NOD and B6 mice were compared (Fig. 1, lower right). These observations indicate that anti-insulin 125Tg is expressed somewhat more efficiently in NOD than in B6 mice, but the overall selection of 125Tg B cells is preserved in NOD mice. In addition, there was no evidence for decreased surface IgM expression, as seen in some models of tolerance (2).
B cells that harbor anti-insulin transgenes are anergic in both B6 and NOD mice
To determine whether B cells harboring anti-insulin 125Tg differ in their functions when expressed in B6 or NOD mice, we measured B cell proliferation in response to different stimuli. B cells were purified from the spleens of transgenic mice and nontransgenic controls, and the kinetics and dose response to anti-IgM (F(ab′)2), LPS, anti-CD40, and IL-4 were determined. The dose response of B cells to anti-IgM in NOD and B6 was determined by [3H]thymidine incorporation on day 3 (Fig. 2,A). Nontransgenic B cells from both B6 and NOD mice mounted strong proliferative responses to stimulation through BCR at all concentrations of anti-IgM, whereas B cells that expressed anti-insulin 125Tg were anergic in both strains. In addition, compared with controls, 125Tg B cells on both B6 and NOD backgrounds showed impaired responses to stimulation with LPS, anti-CD40, and IL-4 (Fig. 2,B). Nonetheless, anti-insulin 125Tg B cells remained competent to respond to the combination of anti-CD40 and IL-4 (Fig. 2 B) and to combinations of PMA and ionomycin (not shown). These findings indicate that 125Tg B cells have functionally impaired proliferative responses to signals delivered through the BCR, TLR4, and CD40 in both 125Tg B6 and 125Tg NOD mice. The degree of anergy to BCR stimulation in 125Tg B cells was comparable to that in high affinity anti-HEL/HEL and other models of B cell tolerance (10, 28). The finding of anergy is consistent with our previous study (24), which showed that 125Tg B6 mice fail to respond to T cell-dependent immunization, and most circulating IgM in 125Tg mice is derived from endogenous B cells (IgMb). Similar results were found in 125Tg NOD mice (not shown). Therefore, we concluded that anti-insulin 125Tg B cells on an autoimmune-prone NOD background are maintained in a tolerant state that does not differ from that of 125Tg B cells on a nonautoimmune background.
Tolerant B cells in 125Tg NOD mice support the development of T1DM
B lymphocytes are recognized to play an essential role in the development of insulitis and diabetes in the NOD model of T1DM (16, 17, 18, 22). However, the functional status of B cells that can capture islet-derived Ags and contribute to the development of T1DM in NOD mice is not known. To investigate how a repertoire of anergic B cells would influence progression to diabetes in NOD mice, cohorts of female NOD mice that harbored anti-insulin 125Tg (n = 20) and their nontransgenic littermates (n = 40) were monitored for the development of diabetes (Fig. 3). Strikingly, the age of onset and frequency of diabetes were similar in 125Tg NOD mice and their nontransgenic littermates. These data markedly contrast the lack of disease in NOD mice that were either B cell deficient (16, 17) or expressed a BCR directed at an irrelevant (nonislet) Ag (22). Thus, a B cell repertoire directed at a single β cell Ag is competent to promote the development of diabetes in the NOD model, even when B cells are defective in their responses to BCR, TLR4, and CD40 engagement.
B cell development is similar in 125Tg B6 and NOD mice
To determine whether the fate of anti-insulin B cells differs in autoimmune NOD mice compared with B6 mice, flow cytometry was used to identify phenotypic markers expressed by anti-insulin B cells from age- and sex-matched mice. B220, CD19, and CD22 were measured as indicators of B cell development, and CD44, CD69, CD80, CD86, and class II MHC were measured as indicators of their activation status. The results of a representative dataset are shown in Fig. 4. The top row shows histograms on B220+ B cells in spleen, and lower panels are histograms gated on the B cell (B220+) population. In 125Tg B6 mice, the number of B cell events (B220+, top row) was decreased ∼20% compared with that in nontransgenic controls (44.8 ± 16.0 vs 36.5 ± 15.0 × 106; n = 15). A decrease in B cell numbers is common in mice that harbor Ig transgenes, because they rapidly transit the proliferative stages of development in the bone marrow. The data also show that the number of B cell events (B220+) in nontransgenic NOD mice was slightly decreased, and the presence of anti-insulin 125Tg corrected this mild B lymphopenia (30.3 ± 11.5 vs 40.3 ± 14.9 × 106; n = 10). A similar correction of B cell numbers was recently noted when anti-HEL transgenes were expressed in NOD mice (22). Within the B cell populations, the expression of CD19, a positive regulator of BCR signaling, and CD22, a negative regulator of BCR signaling, was not influenced by the presence of the transgene and did not differ between the two strains. In addition, we did not find a difference in the surface expression (MFI) of the activation markers CD44 and CD86 (lower two rows) when transgenic mice were compared with nontransgenic littermates from the same strain. No differences were observed in the surface expression of other activation markers, including CD69, CD80, and class II MHC (not shown). These data are highly reproducible and show that in adult mice anti-insulin B cells do not undergo phenotypic changes associated with an arrest in their development and are not activated as a consequence of encountering insulin in vivo. The findings also show that overall B cell development proceeds in a near-normal fashion in nonautoimmune B6 mice and in autoimmune-prone NOD mice even when an anti-insulin Ig transgene is present.
Anti-insulin 125Tg B cells remain competent to enter mature B cell subsets
Newly emerging B cells enter the peripheral repertoire in an immature state and complete their development in the periphery to become FO or MZ B cells. A block in this maturation process is characteristic of tolerant B cells (6, 9). To better understand the fate of anergic anti-insulin B cells that nonetheless promote disease, we assessed the mature B cell compartments in the spleens of NOD and B6 125Tg mice. In this analysis, B cells (B220+) were examined for CD23 (IgE FcεRII) and CD21 (Cr2) expression. This approach identifies mature FO B cells (CD21int, CD23high) and MZ B cells (CD21high, CD23low). Studies were conducted on age- (8 wk) and sex-matched 125Tg B6 and 125Tg NOD mice and on nontransgenic controls. A representative dataset is shown in Fig. 5,A. The data show the expected normal distribution of adult B cell populations in nontransgenic B6 mice (upper left) compared with nontransgenic NOD mice (lower left). NOD mice, as previously shown (29), had proportionally more MZ B cells than B6 mice. The histograms for B6 and NOD mice that harbor anti-insulin 125Tg are shown in Fig. 5,A, upper and lower panels, respectively. Anti-insulin 125Tg B cells entered MZ (CD21high, CD23int) and FO (CD21int, CD23high) compartments in both strains, and the proportion of 125Tg MZ B cells was increased compared with that of MZ B cells in nontransgenic controls. Although the relative proportion of MZ B cells was greatest in 125Tg NOD mice (Fig. 5,A, lower right), the relative increase (2-fold) in MZ B cells was similar to that in 125Tg B6 compared with their nontransgenic B6 controls. This outcome for anti-insulin B cells was strikingly different from the anti-HEL/HEL model of tolerance, in which MZ B cells failed to develop (4, 6). Additional evidence for the normal maturation of anti-insulin B cells was the high level of CD23 expression that is characteristic of mature FO B cells. Decreased CD23 expression is typical of developmentally arrested B cells in other models of tolerance, such as anti-HEL and anti-dsDNA (5, 9). All mature subsets of 125Tg B cells remained competent to bind insulin (Fig. 5 B) (24). Insulin binding is dependent on the presence of both H and L chains of mAb125, because B cells in mice harboring only VH125 as a transgene pairing with endogenous L chains do not bind insulin (23). In addition, in cDNA libraries amplified from anti-insulin 125Tg B cells, only Ig transgenes and not endogenous VH or VL genes, were found, indicating that the expression of endogenous V genes is of very low frequency (E. J. Woodward and J. W. Thomas, unpublished observations). The entry of anergic B cells into mature subsets stands in striking contrast to other models of tolerance and shows that anergy maintained by exposure to a physiologic self-Ag does not prevent B cell maturation in the periphery.
B cells that harbor anti-insulin 125Tg are targeted to the MZ in increased numbers
To verify that B cells expressing MZ markers (CD21high, CD23int) are truly residents of the MZ, immunofluorescence was used to examine spleens from nontransgenic and 125Tg B6 and NOD mice (Fig. 6). Metallophilic macrophages (MOMA-1+) that delineate the marginal sinuses in the spleen are shown in green (FITC), and B cells (B220+) are shown in red (Texas Red). Fig. 6,A (arrow) shows the normally thin layer of MZ B cells (red) that reside outside MOMA-1+ macrophages in the marginal sinuses (green) of nontransgenic B6 spleen. Fig. 6,B shows a notable increase in the size of the MZ in 125Tg B6 mice relative to nontransgenic B6. A similar increase in the size of the MZ is observed in the spleen of a nontransgenic NOD mouse (Fig. 6,D, bracket) compared with nontransgenic B6. Fig. 6,C shows a very large MZ (bracket) that is typical of 125Tg NOD spleen. These data are consistent with the surface phenotypes observed by flow cytometry (Fig. 5) and confirm that anti-insulin 125Tg B cells are targeted to the MZ in increased number in both transgenic NOD and B6 mice.
125Tg B cells increase CD86 in response to T cell-dependent and T cell-independent stimulation
Because the development of T1DM in NOD is mediated by T lymphocytes, and B cells are believed to play a key Ag-presenting role in the disorder, we examined the potential for anti-insulin B cells to up-regulate molecules that mediate T-B interactions. Spleen B cells were cultured in the presence of the same stimuli to which 125Tg B cells are anergic, and the expression of CD86 on B cells (B220+) was measured after 6–72 h of stimulation. Fig. 7 shows the peak (24-h) response of CD86 in B cells cultured with anti-IgM (5 μg/ml), LPS (1 μg/ml), and anti-CD40 (0.1 μg/ml). Nontransgenic B cells (left column) and 125Tg B cells (right column) were observed to increase CD86 expression to comparable levels in response to all three stimuli. The unimpeded nature of the CD86 response in 125Tg B cells was notable because it occurred at mitogen concentrations that are suboptimal for B cell proliferative responses. The data shown correspond to B6 mice. NOD B cells responded in the same manner with a >10-fold increase in MFI in response to different stimuli, although the maximal MFI was lower in NOD (150 ± 15). Other activation markers (e.g., class II MHC, CD44, and CD69) were also comparably induced in transgenic and nontransgenic B cells (data not shown). CD80 expression did not change after stimulation in transgenic or nontransgenic mice from either strain. Thus, B cells that are rendered anergic by insulin in vivo remain competent to increase key T cell costimulator molecules in response to both T cell-independent stimulation (LPS or anti-IgM) and engagement of CD40 typically provided by CD154 on T cells.
This study shows that anti-insulin B cells exposed to their endogenous Ag are permitted to enter the repertoire in a unique state of tolerance. In this state, anti-insulin 125Tg B cells fail to proliferate when stimulated by BCR (anti-IgM), TLR4 (LPS), or CD40 (anti-CD40), yet they remain competent to enter mature subsets of MZ and FO B cells. These findings contrast with other models of tolerance (e.g., anti-HEL/HEL and anti-dsDNA) in which anergy is accompanied by an arrest in development that impairs peripheral B cell maturation (5, 6, 9, 10). This tolerant phenotype does not differ in B6 and autoimmune NOD mice, and in both strains anti-insulin 125Tg B cells are targeted to the MZ in increased numbers. A striking feature of anergic anti-insulin B cells is that they remain fully capable of increasing CD86 expression, a key molecule required for T cell costimulation. Thus, the failure of anti-insulin 125Tg B cells to proliferate after stimulation is disassociated from their ability to enter mature B cell subsets and to increase CD86 expression. The functional significance of this split anergic state is observed in NOD mice that harbor 125Tg B cells. In the NOD model of T1DM, B cells are required for disease progression, and anti-insulin 125Tg B cells provide this essential component even though they are anergic. Anergy is recognized as a state in which Ag receptors become uncoupled from downstream signaling pathways (30). The discovery of a split form of anergy in anti-insulin 125Tg B cells indicates that the mechanisms that maintain functional silencing may selectively target different intracellular signaling pathways. For anti-insulin 125Tg B cells, the result is a repertoire that is blocked for proliferation after BCR, TLR4, and CD40 engagement, whereas signals that mediate peripheral maturation and expression of costimulator molecules remain intact. This novel uncoupling of anti-insulin 125Tg B cells may be governed by the nature of BCR’s encounter with insulin. Structural studies of insulin show that it behaves as a molten globule, and insulin interaction with mAb or BCR results in an induced fit that maintains insulin in association with the Ag binding site (31). For this reason, BCRs occupied by insulin may be able to sustain signals and maintain tolerance, whereas similar levels of ligand (1 ng/ml) in the anti-HEL/HEL model resulted in clonal ignorance, rather than tolerance (32). Consistent with this possibility, we found that most BCRs in 125Tg mice are occupied by endogenous insulin, but this interaction does not result in sufficient cross-linking of surface IgM to reduce 125Tg expression (24). Thus, basal signals emanating from a tolerized anti-insulin BCR may contribute to functional silencing in other signaling pathways, such as TLR4 and CD40, but these signals do not halt B cell development. Recent data showing that Lyn deficiency restores LPS responsiveness in anti-dsDNA B cells that remain tolerant to BCR stimulation are consistent with this type of mechanism (33). Our findings suggest that BCR encounters with soluble self-Ags may generate a variety of outcomes and that anergy is not a uniformly fixed state. Studies are in progress to investigate the nature of signals delivered through the BCR that maintain tolerance in anti-insulin B cells.
The failure to arrest development of anti-insulin 125Tg B cells and their ability to express CD86 may be key elements that promote the T cell interactions necessary for T1DM. Previous studies in tolerant B cells that fail to express costimulator molecules showed that the defective second signal induces tolerance in the T cell compartment (34). This important mechanism by which B cell-T cell interactions maintain tolerance is missing in 125Tg NOD mice, in which B cells retain CD86 expression, so that 125Tg B cells can promote T cell interactions that drive the progression of diabetes. Elegant studies in mice harboring anti-dsDNA transgenes show that a distinguishing feature of anti-dsDNA B cells in lupus-prone mice is their failure to undergo developmental arrest, whereas B cells harboring the same transgenes are arrested in normal mice (11). Thus, arrested development of tolerant B cells is an important barrier that must be breached for systemic autoimmune disease to proceed. For 125Tg B cells in which anergy is uncoupled from maturation arrest, this important barrier to autoimmune disease is not present, and anti-insulin B cells enter mature MZ and FO compartments. MZ B cells may be especially important in this regard because their location allows them to monitor products of islet cell destruction released into the blood, and they are potent APCs for T cell responses (35). Thus, unimpaired entry of self-reactive B cells into mature subsets combined with their potential to induce T cell activation, rather than tolerance, poses a threat for promoting an organ-specific autoimmune disease, just as it does for systemic autoimmune disease. Accordingly, checkpoints preventing B cell maturation and T cell-B cell interactions may be more important targets for intervention in autoimmune disease than induction of anergy. Studies are underway to examine the kinetics of early development of anti-insulin B cells and their genetic regulation.
The data showing that a fully functional B cell repertoire is not essential for T1DM in NOD mice may be relevant to suggestions that the role of B cells is different in human T1DM. This stems from a case report of a child with Xid (BTK) who developed T1DM (36). However, our data suggest an alternate hypothesis, that B cells may contribute to an autoimmune process even when their ability to proliferate and differentiate into Ab-forming cells is impaired. In this way, tolerant anti-insulin B cells resemble B cells in lupus-prone mice in which the BCR is restricted to a transmembrane-only isoform (37). In those mice, membrane-only IgM is adequate to promote many aspects of the autoimmune disease without terminal B cell differentiation. Thus, residual B cells in an immunodeficient subject may contribute to an autoimmune process even when their ability to proliferate and differentiate into Ab-forming cells is impaired. However, it is possible that the inability of 125Tg B cells to switch from IgM to IgG serves to maintain the anergic state of these cells, because the switch to anti-insulin IgG is critical in the progression to T1DM. We are currently developing knockin mice in which anti-insulin V genes have been targeted to the Ig locus to examine this possibility.
We thank Drs. Mark Boothby, Wasif Khan, and Geraldine Miller for discussions of the data and for critical review of the manuscript. The secretarial assistance of Elaine Beeler is appreciated.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported in part by National Institutes of Health Grants DK43911 and AI051448.
Abbreviations used in this paper: HEL, hen egg lysozyme; FO, follicular; MZ, marginal zone; MFI, mean fluorescence intensity; T1DM, type 1 or autoimmune diabetes mellitus.