The highly selective nature of organ-specific autoimmune disease is consistent with a critical role for adaptive immune responses against specific autoantigens. In type 1 diabetes mellitus, autoantibodies to insulin are important markers of the disease process in humans and nonobese diabetic (NOD) mice; however, the Ag-specific receptors responsible for these autoantibodies are obscured by the polyclonal repertoire. NOD mice that harbor an anti-insulin transgene (Tg) (VH125Tg/NOD) circumvent this problem by generating a tractable population of insulin-binding B cells. The nucleotide structure and genetic origin of the endogenous κ L chain (Vκ or IgL) repertoire that pairs with the VH125Tg were analyzed. In contrast to oligoclonal expansion observed in systemic autoimmune disease models, insulin-binding B cells from VH125Tg/NOD mice use specific Vκ genes that are clonally independent and germline encoded. When compared with homologous IgL genes from nonautoimmune strains, Vκ genes from NOD mice are polymorphic. Analysis of the most frequently expressed Vκ1 and Vκ9 genes indicates these are shared with lupus-prone New Zealand Black/BINJ mice (e.g., Vκ1–110*02 and 9–124) and suggests that NOD mice use the infrequent b halpotype. These findings show that a diverse repertoire of anti-insulin B cells is part of the autoimmune process in NOD mice and structural or regulatory elements within the κ locus may be shared with a systemic autoimmune disease.
Type 1 diabetes mellitus is an organ-specific autoimmune disease in which T cells mediate the destruction of insulin-producing pancreatic β cells. Directly, T cell involvement is well established by adoptive transfer experiments in animal models (1, 2, 3, 4, 5, 6) and indirectly in humans by strong genetic linkage to specific class II MHC alleles (7, 8, 9, 10, 11, 12). The role of B cells in this disease is both necessary (13, 14) and complex. Experiments in the nonobese diabetic (NOD)4 murine model of autoimmune diabetes indicate that B cells function in an Ag presentation capacity that is essential for disease progression (15, 16, 17, 18, 19). In this model, B cells specific for islet Ag capture and process autoantigens, resulting in presentation of peptides to cognate T cells. These T-B interactions most likely result in two outcomes: 1) autoaggressive T cells would undergo clonal expansion and ultimately target pancreatic islets, and 2) activated B cells would produce class-switched (IgG), islet specific autoantibodies. Consistent with this process, islet-specific autoantibodies, particularly ones reactive with insulin, are recognized as sensitive indicators of disease (20, 21, 22, 23, 24, 25, 26). To date, tolerance-inducing therapies initiated after the appearance of IgG anti-insulin autoantibodies fail to halt disease progression (27, 28). Thus, earlier predictive methods are required to maintain or restore lymphocyte tolerance before β cell destruction.
The presence of IgG autoantibodies to insulin and other islet Ag in the prodrome of type 1 diabetes mellitus is assumed to be the product of clonal expansion via interaction with autoreactive T cells (29). However, this has not been definitively shown. Additionally, initial interaction of the BCR (surface Ig) with islet Ag necessitates the existence of identifiable molecular characteristics inherent in that receptor that promote Ag binding. To address these gaps, we generated NOD mice harboring an Ig H chain transgene (Tg) (VH125Tg/NOD) derived from an anti-insulin mAb. These mice are unique among Ig transgenic NOD, in that not only do they support the development of diabetes, but it may be accelerated (30). Other Ig transgenic NOD mice that harbor specificities for nonislet Ag (18, 31) exhibit protection from diabetes, further supporting the importance of BCR specificity to diabetogenesis. VH125Tg/NOD mice generate a polyclonal B cell repertoire by using endogenous IgL. This polyclonal repertoire contains a subset (1–2%) of highly insulin-reactive B cells that are not observed in VH125Tg/C57BL/6 mice. By analyzing the nucleotide sequences of the IgL that pair with the VH125Tg, we are able to draw conclusions about a model B cell population that recognizes a key diabetes-associated Ag, insulin. In contrast to expectations, our data demonstrate that anti-insulin B cells are not the product of oligoclonal expansion. Instead, they are independently seeded into the peripheral repertoire. Additionally, they exhibit no evidence of Ag-driven selection or hypermutation. In addition, IgL genes from NOD encode multiple germline polymorphisms distinct from nonautoimmune prone mouse strains such as C57BL/6, C3H, and BALB/c. Rather, certain NOD Vκ genes, such as Vκ1–110*02 and Vκ9–124, are identical with those present in autoimmune-prone New Zealand Black (NZB)/BINJ mice that are of the rare Vκ b haplotype. Thus, the polymorphic residues in structural, coding regions and intervening regulatory elements characteristic of the b haplotype of NZB/BINJ may contribute to autoimmune features in both systemic and organ-specific autoimmune disease.
Materials and Methods
VH125Tg/NOD mice were described previously (30). Lines were maintained as heterozygotes by backcrosses to wild-type (WT) NOD (>20 generations). All mice were housed under pathogen-free conditions, and all experiments were approved by the Institutional Animal Care and Use Committee of Vanderbilt University.
Splenocytes and bone marrow cells were analyzed on a FACSCalibur flow cytometer (BD Biosciences). mAb (BD Pharmingen) used were reactive with: IgMa (DS-1), IgMb (AF6-78), CD23 (B3B4), and B220 (RA3-6B2). Biotinylated insulin (50 ng/ml; NOVO) was used to detect insulin-binding B cells with PerCP-streptavidin. Specificity was confirmed by inhibition with excess human insulin (32). WinMDI 2.8 software (Dr. J. Trotter, Scripps Institute, San Diego, CA) was used for data analysis.
Selection of anti-insulin B cells
VH125Tg/NOD spleens were depleted of T cells via anti-Thy-1.1 and complement. Insulin-binding B cells were selected by MACS or adherence to insulin-coated plates. For MACS, T cell-depleted splenocytes (107 cells/90 μl) were incubated with biotinylated insulin (50 ng/ml/106 cells) in buffer (2 mM EDTA, 0.5% BSA in 1× PBS) for 10 min at 4°C, washed, and incubated with streptavidin-conjugated magnetic beads (20 μl beads/107 cells; Miltenyi Biotec) for 15 min at 4°C. Cells were resuspended (108 cells/500 μl) and passed over an LS column (Miltenyi Biotec). Insulin-binding cells were eluted from the column and lysed in TRI Reagent (Molecular Research Center). For plate binding, dishes (Corning Glass) were precoated with human insulin (1 μg/ml in PBS overnight at 4°C) and blocked with BSA. Unbound cells were thoroughly washed from the plate with PBS. Cells were removed by scraping in TRI reagent.
Vκ amplification and analysis
To analyze expressed Vκ genes, RNA was isolated from T cell-depleted splenocytes (total Vκ) or from insulin-selected B cells. First strand cDNA was generated from total RNA using Superscript II RT (Invitrogen Life Technologies) and 0.67 μg of oligo(dT) primer (Amersham Biosciences) in a standard cDNA synthesis protocol. Vκ sequences were amplified from first strand cDNA using the following primers: murine κ C region primer, 5′-GGA TAC AGT TGG TGC AGC ATC-3′; murine VκA, 5′-ATT GTK MTS ACM CAR TCT CCA-3′; murine VκB, 5′-GAT RTT KTG RTR ACB CAR RM-3′; murine VκC, 5′-AYA TYN WGM TGA CHC ARW CTM M-3′. Vκ sequences were amplified using AmpliTAQ DNA polymerase (2 U/reaction) (Applied Biosystems) and the following: 200 nM dNTP, 1.25 mM MgCl2, 13.35 mM constant primer, and 13.35 mM of one of the three Vκ primers. The PCR protocol was 94°C/1 min, 42°C/1 min, and 72°C/2 min for 35 cycles. PCR product was ligated into pGEM-T easy plasmid (pGEM-T Easy Vector System I; Promega). Positive clones were sequenced using an Applied Biosystems 3730xl DNA Analyzer (Vanderbilt-Ingram Cancer Center). Analysis, homologies, and germline gene segment assignment were accomplished with blastn (www.ncbi.nlm.nih.gov/BLAST), the ImMunoGeneTics database (http://imgt.cines.fr:8104/), and BioEdit (www.mbio.ncsu.edu/BioEdit/bioedit.html). Statistical significance for Vκ families and individual genes in the insulin-selected and unselected groups (see Figs. 2 and 3) is derived from a χ2 test of independence.
NOD germline Vκ genes
Germline Vκ gene segments were amplified from NOD tail DNA. Primers were designed based on known C57BL/6 germline genes: Vκ9–120/4FWD, 5′-ATG GAC ATG AGG GYT CCT GC-3′; Vκ9–120FWD2, 5′-GGG CTC CTG CAC AGA TTT TTG-3′; Vκ9–120iFWD, 5′-GGG GGA TGT CCT CTT TTC TC-3′; Vκ9–120/4REV, 5′-CAC TGT GGG AGG AKA ACT AG-3′; Vκ1–132/3FWD, 5′-ATG ATG AGT CCT GTC CAG TTC C-3′; Vκ1–110FWD, 5′-ATG AAG TTG CCT GTT AGG CTG TTG G-3′; Vκ1–132/3REV, 5′-CAC TGT GTG AGG AWA ATR TGT ACC-3′; and Vκ1–110REV, 5′-CAC TGT GGG AGG AAC ATG TGT AC-3′. Germline Vκ sequences were amplified using AmpliTAQ DNA polymerase (2 U/reaction) and the following: 500 ng of genomic DNA, 2.5 mM MgCl2, 250 nM each primer, and 200 nM dNTP. Reactions were cycled at 94°C/1 min; 53°C (Vκ9), 55°C, and (Vκ1–132/3) 58°C (Vκ1–110)/1 min; and 72°C/1 min for 40 cycles. PCR products were cloned, sequenced, and analyzed, as described above.
VH125Tg/NOD mice generate a detectable population of anti-insulin B cells
VH125Tg/NOD mice develop diabetes at an accelerated rate compared with their WT littermates (30). In these mice, 1–2% of B cells bind biotinylated insulin with a mean fluorescence intensity (MFI) >200 (Fig. 1,B). The specificity of this population is confirmed by inhibition with soluble, unlabeled insulin (32). Insulin-specific B cells are undetectable in C57BL/6 mice expressing VH125Tg (Fig. 1,D) as well as in WT NOD and C57BL/6 (Fig. 1, A and C, respectively) by this method. These findings are consistent with the low frequency (10−5) of anti-insulin B cells predicted by T cell-independent responses to insulin (33). The presence of insulin autoantibodies in WT NOD indicates that anti-insulin B cells are present in these animals, and the VH125Tg increases the frequency of this population.
The Vκ gene families expressed by insulin-binding B cells in VH125Tg/NOD mice are heterogeneous
To understand their molecular origins, we isolated insulin-binding B cells from VH125Tg/NOD mice and examined their expressed Vκ genes. Vκ genes were cloned and sequenced from B cells either selected for insulin binding or unselected (total). Vκ identity was assigned based on nucleotide homology to germline reference sequences in the ImMunoGeneTics database. The findings do not differ for B cells captured on insulin-coated plates or MACS columns, and the results of six independent experiments are pooled. Histograms show the frequency of Vκ families used by anti-insulin VH125Tg/NOD B cells (Fig. 2,A) compared with unselected VH125Tg/NOD B cells (Fig. 2,B). The most frequently used family in this repertoire was Vκ1, representing 45% of isolates from insulin-binding VH125Tg/NOD B cells and 31% of the unselected population. The high frequency of Vκ1 in this study is consistent with the representation of this family in other primary repertoires (34). In addition to Vκ1, insulin-binding B cells also use genes from Vκ families 9, 10, and 19, while unselected B cells also use Vκ2, 3, and 6 (Fig. 2, compare A with B). Surprisingly, Vκ4 usage was not dramatically increased in the insulin-binding population even though the original partner of VH125 is Vκ4–74∗01 (35). These findings demonstrate that multiple IgL families contribute to the insulin-binding repertoire. The distribution of Vκ families used by the insulin-selected and unselected groups did not differ from those expected based on a χ2 test for independence. To corroborate the IgL heterogeneity exhibited by VH125Tg/NOD B cells, we surveyed Vκ gene usage in a panel of VH9-containing mAb collected from GenBank. The variety of IgL used by these VH9 H chains, as shown in Fig. 2 C, suggests that the heterogeneity exhibited by the VH125Tg is representative of similar H chains.
Specific Vκ genes are expressed by insulin-binding B cells in VH125Tg/NOD mice
The previous analysis demonstrates that a variety of Vκ families can pair with the VH125Tg to generate anti-insulin BCR, but does not indicate how any single gene is used. Therefore, we examined the frequency of specific Vκ genes. Using this approach, marked differences in Vκ gene usage between insulin-binding and unselected B cells were revealed (Fig. 3). For example, within the Vκ1 family, the Vκ1–110*02 gene is significantly overrepresented in sequences derived from insulin-selected B cells (p = 0.0044 by χ2), while Vκ1–135 is preferred by the unselected population (not significant; p = 0.086). Likewise, the Vκ9–120 gene is more frequently associated with insulin-selected B cells than the Vκ9–124 gene. Thus, within the diverse family repertoire, a subset of Vκ genes preferentially generates anti-insulin BCR. Therefore, insulin binding is not governed by a single Vκ family, but is determined by unique features of individual genes from different families.
Anti-insulin Vκ1 genes expressed in VH125Tg/NOD mice are germline encoded, clonally independent, and polymorphic to nonautoimmune strains
As shown in Fig. 3, Vκ1–110*02 (V1B) is exclusively expressed by insulin-binding VH125Tg/NOD B cells (p = 0.0044). To investigate the roles of Ag-driven clonal expansion and somatic mutation in the generation of anti-insulin B cells, we compared the nucleotide sequences of expressed anti-insulin genes with known germline Vκ1 genes (Fig. 4). The *02 allele from lupus-prone NZB mice (36), haplotype b (37), is allelic to Vκ1–110*01 germline genes from C57BL/6 and BALB/c mice (haplotype c). Polymorphic residues associated with the *02 allele are located principally in the CDR and account for most of the structural differences exhibited by NOD Vκ1–110. The frequency of the other nucleotide differences does not exceed that which is anticipated from errors in amplification and sequencing. Because VH125Tg/C57BL/6 mice do not exhibit anti-insulin B cells in the periphery, this suggests that the Vκ1–110*02 allele may favor insulin binding. Additionally, Vκ1 genes expressed by anti-insulin VH125Tg/NOD B cells are germline encoded and clonally independent. Lack of clonality is deduced from the nucleotide sequence of each VκJκ joining site in CDR3 (Fig. 4). Clones 39 and 92 use the same joining sequence, but were derived from different donor mice. These data demonstrate that anti-insulin Vκ1 genes in VH125Tg/NOD mice are principally derived from independent clones rather than recurrent expansion of the same clone. In 58 independently derived isolates, only one pair exhibited evidence of potential clonality.
Vκ9 genes expressed in VH125Tg/NOD mice are also polymorphic and germline encoded
The preceding data on Vκ1 genes suggest that polymorphisms in the IgL locus may contribute to B cell autoreactivity in NOD mice. Therefore, we extended our analysis to Vκ9 genes (Fig. 5). A subset of Vκ9 sequences from VH125Tg/NOD is most closely related to Vκ9–120*01 (C3H and C57BL/6 strains). Vκ9–120 clones from NOD differ at 11 nt from the reference sequences (*01). To confirm these polymorphisms, we amplified genomic DNA from NOD mice (Materials and Methods). Ten independent isolates from four separate experiments identified a Vκ9 gene that is polymorphic to the known Vκ9–120*01 germline sequences. When compared with the Vκ9–120 genes expressed in VH125Tg/NOD B cells, the germline polymorphisms account for all the nucleotide differences between expressed NOD Vκ9–120 and the reference Vκ9–120*01 sequence. Thus, the Vκ9 gene used in VH125Tg/NOD mice is a novel Vκ9–120 allele. These data support the conclusion that germline NOD Vκ are highly polymorphic when compared with those in other strains.
Polymorphisms are not limited to insulin-binding Vκ genes
As shown in Fig. 3, Vκ9–124 is highly expressed in the unselected VH125Tg/NOD B cell population. The sequences in Fig. 5 suggest that numerous polymorphisms occur in genes used by the unselected population. To further cement this observation, we analyzed the nucleotide sequence of expressed and germline Vκ9–124 genes (Fig. 6). The six isolates (A4, genomic DNA; 17, 59, 226, 227, and 244, expressed cDNA) exhibit limited polymorphisms in CDR 2 and 3. The A4 group represents allelic homologues of Vκ9–124*01, and further exhibits germline polymorphisms in both the insulin-specific and nonspecific repertoires. A search of the public database has revealed that germline NOD Vκ9–124 is identical with two expressed mAb IgL from NZB (GenBank accession numbers AF321948 and Z22118).
Germline nucleotide polymorphisms in NOD Vκ confer amino acid changes
To determine whether the polymorphic nucleotide residues seen in NOD-expressed and germline Vκ genes could have functional implications, their predicted amino acid sequences were analyzed (Table I). Polymorphisms conferring no amino acid change are not detailed. The total number of polymorphisms and their segregation into CDR and framework regions are indicated. Only one-half of the amino acid changes are conservative, while the rest constitute alterations in size and/or charge. These unique NOD polymorphisms could thus influence Ag specificity and BCR assembly, consequently governing the primary B cell repertoire.
|NOD Vκ Gene .||Polymorphisms .||.||Nonsilent changesab .|
|CDR||5||K110/111R, S227F, S232G|
|CDR||4||G37/38Y, S40G, A98G|
|FWR||7||V4I, L55F, E64K, R86L, V200A|
|CDR||4||A130/131S, A133T, Y263S|
|NOD Vκ Gene .||Polymorphisms .||.||Nonsilent changesab .|
|CDR||5||K110/111R, S227F, S232G|
|CDR||4||G37/38Y, S40G, A98G|
|FWR||7||V4I, L55F, E64K, R86L, V200A|
|CDR||4||A130/131S, A133T, Y263S|
Silent amino acid changes are not indicated.
Numbers indicate nucleotide positions (not amino acid numbers), as indicated in Figs. 4–6.
FWR, framework region.
Anti-insulin B cells are generated in the bone marrow of VH125Tg/NOD mice
The presence of anti-insulin B cells in the spleens of VH125Tg/NOD, but not in VH125Tg/C57BL/6 mice could result from two processes. First, anti-insulin B cells could be generated on both backgrounds, but only be removed (by deletion or receptor editing) from the repertoire of C57BL/6 mice. Alternatively, germline polymorphisms unique to NOD could favor the production of B cells that recognize insulin. To address these possibilities, we analyzed the bone marrow of VH125Tg/C57BL/6 and VH125Tg/NOD for insulin-binding B cells. As shown in Fig. 7, anti-insulin B cells (indicated by M1) are only detected in the bone marrow of VH125Tg/NOD mice. The cells shown are B220+, IgMa+, and CD23−, representing immature and newly formed bone marrow B cells and not mature recirculating B cells from the spleen. The specificity of this population was confirmed by inhibition with excess unlabeled insulin (for VH125Tg/NOD B cells in the M1 gate: uninhibited MFI = 293, inhibited MFI = 151; data not shown). These data demonstrate that germline Ig polymorphisms in NOD mice favor the production of autoantigen-specific B cells. Anti-insulin B cells are generated in the bone marrow of NOD mice in the absence of peripheral positive selection. Additionally, anti-insulin B cells are not obviously generated in VH125Tg/C57BL/6 bone marrow and then negatively selected in the periphery.
NOD mice that carry the IgH chain of anti-insulin mAb125 (VH125Tg/NOD) generate a small population of insulin-binding B cells that are not observed in VH125Tg/C57BL/6 mice. In this study, we have isolated anti-insulin B cells from VH125Tg/NOD and demonstrate that their BCRs are independently generated by the recombination of several different Vκ and Jκ genes. Genes from the Vκ1 and Vκ9 families constitute the majority of anti-insulin BCR, and their nucleotide sequences do not show evidence of somatic hypermutation. In-depth analysis of germline and expressed Vκ genes isolated from NOD mice demonstrates germline-encoded polymorphisms that are allelic (e.g., Vκ1–110*02) to nonautoimmune prone strains of the c haplotype (C57BL/6, C3H, and BALB/c). Because anti-insulin B cells are not observed in the bone marrow or spleen of VH125Tg/C57BL/6 (Figs. 1 and 7), these data support the hypothesis that the κ locus of the NOD strain facilitates skewing of the primary repertoire toward this autoantigen.
In systemic autoimmune disease, such as lupus, oligoclonal expansion and somatic mutation typify the anti-DNA response even when Ig H chain Tg are present (38, 39, 40, 41, 42). Because IgG insulin autoantibodies are recognized in the prodrome of autoimmune diabetes (22), we anticipated that anti-insulin B cells in adult NOD mice would exhibit oligoclonal expansion and somatic mutation as a consequence of T cell help. However, the data suggest that anti-insulin B cells in adult VH125Tg/NOD mice are not derived from this process. Rather, seeding of a diverse group of anti-insulin B cells into the repertoire may provide an important source of APCs that capture insulin-related Ag and contribute to expansion of the autoreactive T cell pool. An important goal for future studies is to understand the relationship between the primary repertoire of anti-insulin B cells and that which differentiates to produce IgG autoantibodies. This will require the production of NOD mice in which fully functional Tg are targeted into the Ig H chain locus.
In NOD mice, multiple genes, particularly from the Vκ1 and Vκ9 families, are capable of forming anti-insulin BCRs (Figs. 2 and 3). Although multiple families contribute to insulin binding, specific genes within each family preferentially segregate into the insulin-binding pool. For example, Vκ1–110 segregates exclusively with insulin binding, whereas Vκ1–135 predominates in the unselected population. Comparison of Vκ1–110 to published germline counterparts clearly demonstrates that the NOD and NZB/BINJ strains share this allele (Fig. 4). To our knowledge, Vκ1–110*02 (36) is the only germline NZB Vκ gene published in the database. We therefore compared our germline NOD Vκ9–120 and Vκ9–124 genes with NZB Vκ from published mAb. NOD Vκ9–124 (clone A4) is identical to an anti-peptide Vκ (GenBank accession number AF321948) and differs by one nucleotide from an anti-DNA Vκ (GenBank accession number Z22118) both from NZB × NZW F1 mice. Earlier work in our lab has demonstrated substantial (>99%) identity between spontaneous mAb generated from NOD and NZB hybridomas (43). In 1988, D’Hoostelaere et al. (37) used RFLP analysis to conclude that NZB/BINJ mice have the Igκ b haplotype, whereas C3H, BALB/c, and C57BL/6 are all of the c haplotype. Although NOD was not included in that study, our data strongly support the hypothesis that NOD and NZB share the Igκ b haplotype. This is not the case for Ig H chains, as studies clearly indicate that the H chain loci of NOD and C57BL/6 mice are highly similar, if not identical (44, 45). Interestingly, in the D’Hoostelaere study, NZB was the only strain of the 55 investigated that exhibited the b haplotype. This observation raises the possibility that the Igκ b allelic group may predispose susceptible strains to autoreactivity. A number of genetic features are shared by systemic and organ-specific autoimmune diseases (46), and our findings suggest that the Igκ b haplotype may be included among these. It is of note that two diabetes susceptibility loci, Idd6 and Idd19, map to chromosome 6 distal to the Igκ locus, but the relationship of these loci to the b haplotype is not known.
The finding that VH125Tg/NOD, and not VH125Tg/C57BL/6 mice have anti-insulin B cell populations in both their immature bone marrow and splenic repertoires suggests intrinsic differences in B cell generation between NOD and C57BL/6 strains. Because the polymorphisms observed in NOD Vκ alleles encode alterations in primary structure, it is possible that these structures may skew the repertoire of Ag recognized by NOD B cells. However, the original partner of VH125 is a member of the largest family, Vκ4, recombined with Jκ5 (35). The Vκ4Jκ5 configuration is considered an indicator of receptor editing (47). It is possible that Vκ1 and Vκ9 genes are rapidly edited in VH125Tg/C57BL/6 mice and that this process is less efficient in NOD mice. Because Ag-driven receptor editing takes place chiefly at the L chain loci, it is also possible that polymorphisms in NOD Vκ alleles extend to regulatory elements that impact L chain replacement. Recent studies using fixed Ig Tg also reveal a selection defect in NOD mice that permits autoreactive B cells to enter the peripheral repertoire (18). Thus, the Vκ haplotype may synergize with other defects in NOD B cell generation (48, 49), resulting in the maintenance of autoreactive specificities in the repertoire. Further studies confirming the Vκ haplotype of NOD as well as directly assessing the role of the b haplotype in predisposition to autoimmunity are clearly required.
We thank the Vanderbilt DNA Sequencing Facility. Additionally, the statistical advice of Dr. S. Haneuse and the critical reviews of Drs. G. G. Miller, M. R. Boothby, and J. Crowe as well as C. A. Acevedo Suárez, R. A. Henry, and C. Hulbert are greatly appreciated.
The authors have no financial conflict of interest.
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.
These studies were supported in part by grants from the National Institutes of Health (AI47763). E.J.W. was supported sequentially by a Graduate Assistance in Areas of National Need Fellowship and the Immunobiology of Blood and Vascular Systems Training Program (5 T32 HL069765). The Vanderbilt DNA Sequencing Facility is supported by National Institutes of Health Grants CA68485 (Vanderbilt-Ingram Cancer Center), DK20593 (Vanderbilt Diabetes Research and Training Center), and HL65962 (Vanderbilt Pharmacogenomics Research Center).
Abbreviations used in this paper: NOD, nonobese diabetic; Tg, transgene; MFI, mean fluorescence intensity; NZB, New Zealand Black; FWR, framework region; WT, wild type.