Transgenic rescue implicates β₂-microglobulin as a diabetes susceptibility gene in nonobese diabetic (NOD) mice

Abstract

Type 1 diabetes in both humans and nonobese diabetic (NOD) mice results from T-cell-mediated autoimmune destruction of insulin-producing pancreatic β cells. Linkage studies have shown that type 1 diabetes in NOD mice is a polygenic disease involving more than 15 chromosomal susceptibility regions. Despite extensive investigation, the identification of individual susceptibility genes either within or outside the major histocompatibility complex region has proven problematic because of the limitations of linkage analysis. In this paper, we provide evidence implicating a single diabetes susceptibility gene, which lies outside the major histocompatibility complex region. Using allelic reconstitution by transgenic rescue, we show that NOD mice expressing the β₂ microglobulin (β₂M)^a allele develop diabetes, whereas NOD mice expressing a murine β₂M^b or human allele are protected. The murine β₂M^a allele differs from the β₂M^b allele only at a single amino acid. Mechanistic studies indicate that the absence of the NOD β₂M^a isoform on nonhematopoietic cells inhibits the development or activation of diabetogenic T cells.

The protein β₂M is encoded on human chromosome 15 and mouse chromosome 2 (1, 2). It is a 12-kDa protein with an amino acid sequence homologous to a single Ig domain. Apart from its association with the products of class I major histocompatibility complex (MHC) responsible for presentation of peptides to the immune system, β₂M is associated with a number of homologues of MHC molecules that have diverse roles. These include presentation of lipid antigens (CD1), transport of immunoglobulins (neonatal F_c receptor), regulation of iron metabolism (HFE), and deception of the host immune system (by viral homologues) (3). β₂M is relatively nonpolymorphic. Although seven alleles have been identified in mice, they differ only in a limited number of residues (4, 5). Of note, the three isoforms that have been identified in inbred strains differ only at a single residue, 85 (4, 6). Lack of polymorphism in β₂M has led to the view that β₂M has purely structural roles in ensuring correct protein folding and transport to the cell surface (7), although there is some evidence that β₂M polymorphism may influence antigen presentation (8).

The development of type 1 diabetes (IDDM) in nonobese diabetic (NOD) mice results from the destruction of pancreatic β cells by autoreactive T-cell responses that are mediated by both the class I (K^d, D^b) and class II (A^g7) gene products of the H2^g7 MHC haplotype (9). In addition to the H2^g7 MHC haplotype, genes within a minimum of 14 other chromosomal regions also contribute to IDDM development in NOD mice (10, 11). β₂M maps within the 24-centimorgan segment on chromosome 2 originally defined as Idd13 on the basis of recessive IDDM resistance in NOD mice congenic for this genomic interval derived from the related nonobese resistant (NOR) strain (12). More recent studies on NOD mice that carry NOR-derived subcongenic intervals of Idd13 have shown that this region consists of at least two allelically variable genes that contribute to IDDM susceptibility or resistance, although the identity of these genes has remained unknown (13). One of the NOR-derived subcongenic intervals that is associated with a decreased incidence of IDDM in NOD mice contains the β₂M locus. On the basis of the role of β₂M in MHC class I and CD1 expression, we investigated β₂M as a candidate susceptibility gene. The ideal approach would be direct replacement of the NOD β₂M allele with another variant through an embryonic stem (ES) cell-based gene “knock-in” strategy. However, this was not possible, because no ES cell lines of NOD origin with efficient germ line transmission capacity are yet available. A β₂M knock in could be done by using ES cells of 129 strain origin, but in the course of transferring this allele to the NOD background, other linked genes of 129 origin would also be cotransferred, which could affect IDDM development. Hence, we chose the best currently available alternative approach of determining whether transgenic rescue of β₂M-deficient NOD mice with various β₂M isoforms engendered IDDM susceptibility or resistance. An advantage of this approach over the knock in strategy by using non-NOD ES cells is that the parental β₂M-deficient NOD genotype used for transgenic rescue is internally controlled i.e., the various β₂M transgenic NOD mice were generated in the same parental strain.

Methods

Mice.

C3H.H-2^o(K^d, D^k) NOD/Lt, β₂M^−/− NOD and β₂M transgenic NOD mice were maintained under specific pathogen-free conditions at the John Curtin School of Medical Research. NOD/Lt, β₂M^−/− NOD, human β₂M (hβ₂M) transgenic NOD, and NOD-severe combined immunodeficient (SCID) mice were maintained under specific pathogen-free conditions at The Jackson Laboratories.

Sequencing.

The β₂M exon and promoter regions were amplified from NOD/Lt mice by PCR by using Pwo polymerase. DNA from three independent PCR products was sequenced by using ABI PRISM Dye Terminator Cycle Sequencing Ready Reaction Kit (Perkin–Elmer Applied Biosystems), according to the manufacturer's instructions.

Generation of β₂M Transgenic Mouse Lines.

The previously described hβ₂M transgene (two to four copies) (14) was backcrossed for nine generations to the previously described N11 stock of NOD.β₂M^−/− mice (15). Backcross segregants carrying the hβ₂M transgene were identified by flow cytometry for positive staining of peripheral blood leukocytes with the FITC-conjugated mAb BM63 (Sigma). By using previously described methods (16), this NOD.β₂M^−/−.hβ₂M transgenic stock was shown to be homozygous for linkage markers delineating all known Idd loci of NOD origin.

The β₂M^b gene derived from the plasmid pKC7-β₂M^b (17) was converted to the β₂M^a form of the gene by excision of the EcoRI-BglI fragment of exon II, and replacement with a 53-bp oligonucleotide identical to the excised fragment except the codon GCC of β₂M^b was changed to the GAC of β₂M^a. The integrity of the insertion site was checked by sequencing, as described above. To distinguish the transgenic β₂M gene from the endogenous gene, 150-bp dormant loxP sites were inserted into the NdeI site in intron A and the BamHI site in intron C. The converted β₂M^a gene and the unmodified β₂M^b gene were used to generate NOD transgenic mice according to standard methods (18). Transgenic founders were crossed to β₂M^−/− NOD mice, which were at the 11th backcross, and then intercrossed for homozygosity. β₂M^a transgenic offspring (two to three copies) were genotyped by a PCR, which detected the 150-bp insertion into intron A by using the oligonucleotides TCAATTGTCATGGTCCTCACATCTC and GCAGGCGTATGTATCAGTCTCAGT. β₂M^b transgenics (one to two copies) were genotyped by BglI digestion of a PCR product amplified from exon 2 by using the oligonucleotides GAGAATGGGAAGCCGAACATACT and TTAAAGTCAAGAAACCTACCATCAT. The BglI site is unique to exon II of β₂M^b (4), and both the β₂M^a and β₂M^b PCRs differentiate the disrupted β₂M gene, which has a 1.1-kb insert in exon 2 (19).

Disease Assessment in β₂M Transgenic Mice.

Female β₂M^+/− NOD, β₂M^a transgenic β₂M^−/−NOD, β₂M^b transgenic β₂M^−/−NOD, and β₂M^b transgenic β₂M^+/−NOD, described in Fig. 3a, were regularly tested for glycosuria by using TesTape (Lilly Research Laboratories, Indianapolis) and were scored as diabetic after two positive readings 2 days apart. After 250–300 days or development of diabetes, mice were killed and pancreata fixed and stained with hematoxylin/eosin. Islets were scored as: 0, no infiltrate; 1, peri-insulitis; 2, circumferential insulitis; 3, intraislet infiltration; and 4, severe structural derangement. Between 10 and 100 islets from each mouse were examined. Alternatively pancreatic tissue was frozen in Tissue-Tek OCT compound (Bayer, Elkhart, IN), and acetone-fixed sections were stained with primary antibodies as indicated, followed by horseradish peroxidase-conjugated anti-rat Ig (Dako). Color was developed with diaminobenzidine tetrahydrochloride, and sections were counterstained with hematoxylin.

Figure 3.

(a) Incidence of diabetes in the John Curtin School of Medical Research colony of female β₂M^a and β₂M^b transgenic NOD mice. β₂M^a transgenic β₂M^−/− NOD (diamonds, n = 11), β₂M^b transgenic β₂M^−/− NOD (squares, n = 27), littermate control β₂M^b transgenic β₂M^+/− NOD (circles, n = 34), β₂M^−/− NOD (open squares, n = 16), and β₂M^+/− NOD mice (triangles, n = 12) were followed for 250 days for development of diabetes. The incidence in β₂M^b transgenic β₂M^−/− NOD mice was dramatically reduced (P < 0.002 vs. β₂M^+/− NOD mice). (b) Incidence of diabetes in The Jackson Laboratory colony of female hβ₂M transgenic NOD mice. hβ₂M transgenic β₂M^−/− NOD mice (squares, n = 19), β₂M^−/− NOD mice (open squares, n = 11), hβ₂M transgenic NOD mice (circles, n = 19), β₂M^+/− NOD (triangles, n = 12), and NOD mice (diamonds, n = 21) were followed for 250 days for development of diabetes. Diabetes incidence in hβ₂M transgenic NOD β₂M^−/− mice was 25% that of NOD or hβ₂M transgenic NOD mice (P < 0.0001).

Female NOD and hβ₂M transgenic NOD mice, described in Fig. 3b, were maintained at The Jackson Laboratory and were monitored weekly for the development of glycosuria with Ames Diastix (kindly supplied by Miles Diagnostics, Elkhart, IN), and were scored as diabetic after two positive readings at least 2 days apart.

Flow-Cytometric Analysis.

Islets were isolated as previously described (20), then cultured for 18 h in RPMI 1640 culture medium containing 100 units of γIFN, 10% FCS, penicillin, gentamicin, and streptomycin at 37°C with 10% CO₂. Single-cell islet suspensions were made by digestion for 10 min with trypsin, followed by passage through a 20-gauge needle five times. Islet, spleen, mesenteric lymph node, thymus, and peripheral blood lymphocyte cells were stained for H2-K^d with SF1–1.1 and for H2-D^b with 28–14-8 (PharMingen). Dead cells were excluded by propidium iodide staining, islet β cells were identified by sorting before fixation with acetone and immunohistochemical staining for insulin by using peroxidase-antiperoxidase methodology (Dako), and lymphocytes were identified by anti-CD3 staining.

⁵¹Cr Release Cytotoxicity Assay.

C3H.H-2^o, NOD, β₂M^+/− NOD, β₂M^a transgenic β₂M^−/− NOD, β₂M^b transgenic β₂M^−/−NOD, and β₂M^−/− NOD male mice greater than 12 weeks of age were immunized for 7 days with 10⁷ plaque-forming units of vaccinia virus or 10⁴ haemagglutination units of influenza virus strain A/JAP. Splenocytes were then restimulated for 5 days with nuclear protein peptide (NPP) or vaccinia virus, as described (21), to generate secondary cytotoxic T lymphocytes. P815 (H2^d, β₂M^a), H2-K^d-transfected RMA-S (H-2^b-K^d, β₂M^b), and fibroblasts OH (K^d,D^k, β₂M^a) and HTG (K^dD^b) targets were labeled with ⁵¹Cr and treated with synthetic NPP, A/JAP, vaccinia virus, or mock peptide, as described (21). ⁵¹Cr release cytotoxicity assay was performed as described with an assay length of 18 h. Percentage specific lysis was calculated by the formula: percentage specific lysis = (experimental release—medium release)/(maximum release—medium release).

Generation of Bone Marrow Chimeras.

Females from the indicated strains were lethally irradiated (1,200 R from a ¹³⁷Cs source) at 4 weeks of age, and then reconstituted, as previously described (13), with 5 × 10⁶ bone marrow cells isolated from the indicated 8-week-old female donors. Bone marrow chimeras were then monitored through 21-week postreconstitution for the development of IDDM, as assessed by the appearance of glycosuria. In some experiments, splenic leukocytes were isolated from the indicated bone marrow chimeras at 6 weeks postreconstitution and injected i.v. into 4- to 6-week-old female NOD-SCID mice (10⁷ cells/recipient). The NOD-SCID recipients were then monitored for IDDM development for up to 15 weeks. Recipients were assessed for CD4 and CD8 T cell repopulation by flow cytometric analyses of splenocytes at the onset of IDDM or at the end of the 15-week observation period.

Results and Discussion

The Promoter Region of the NOD β₂M^a Allele Does Not Contribute to IDDM.

The NOD β₂M gene was sequenced to determine whether either the promoter or the coding region carried a rare mutation. Although we found the coding sequence to be identical to that of the common “a” isoform, there were four changes in the promoter region that differed from published sequences (Fig. 1). Three of these changes were found in a cluster between the TATA box and the start codon.

Figure 1.

Sequence of the NOD β₂M promoter. TATA box and ATG are underlined. The 226-bp fragment reported to have enhancer activity (30) is marked with stars. The previously published β₂M^a sequence is aligned beneath the NOD sequence (30).

To test whether the changes found in the NOD β₂M promoter could confer susceptibility to IDDM, we replaced the β₂M^a gene in NOD mice with a β₂M^a gene derived from a nondiabetes prone strain. We made NOD mice transgenic for the nondiabetes prone β₂M^a construct and crossed this transgene into β₂M^−/− NOD mice, which completely lack expression of β₂M. Resultant β₂M^a transgenic β₂M^−/− NOD mice were shown to express MHC class I in a pattern, and at a level that is not significantly different from NOD mice (Fig. 2a). The β₂M^a molecule was shown to function normally in terms of its ability to pair with MHC class I and to select for CD8 T cells. β₂M^a transgenic β₂M^−/− NOD mice were shown to have normal CD8 T-lymphocyte function by their ability to reject class I disparate skin grafts (data not shown) and their ability to clear viral infection (Fig. 2b).

Figure 2.

(a) Expression of transgenic mouse β₂M is normal. Single-cell suspensions were made from peripheral blood, spleen, mesenteric lymph nodes, and thymus of NOD, β₂M^+/−, β₂M^a transgenic, β₂M^b transgenic, and β₂M^−/− NOD female 10- to 17-week-old mice. Cells were stained for H2-K^d and analyzed by FACS. One of several experiments is shown. (b) Secondary cytotoxic T lymphocyte (CTL) responses to NPP. CTL activity of cells from primed C3H.H2^o, β₂M^−/− NOD, β₂M^+/− NOD, β₂M^a transgenic, β₂M^−/− NOD, and β₂M^b transgenic β₂M^−/− NOD mice was assessed (n = 2 per group, one of two experiments). P815 target killing was reduced in β₂M^b transgenic derived cytotoxic cells, compared with β₂M^a derived cytotoxic cells. The lower cytotoxicity of β₂M^b transgenic derived cells was corrected by inclusion of β₂M^b on the target cells, showing a differential peptide binding between the two β₂M isoforms.

A colony of these mice was established and assessed for disease in terms of mononuclear cell infiltrate into the islet tissue (insulitis) and in terms of hyperglycemia. We found that the β₂M^a transgene was able to restore both insulitis and diabetes to the normally resistant β₂M^−/− NOD mice. There was no difference in the extent of insulitis or the incidence of hyperglycemia (Fig. 3a) between β₂M^−/− NOD mice carrying the β₂M^a transgene and the β₂M^+/− littermates, which carried a copy of the endogenous NOD β₂M^a gene. We therefore conclude that, although the promoter for the β₂M gene in NOD mice is different from the published sequence, it does not contribute to IDDM susceptibility.

The NOD β₂M^a Structural Variant Confers Dominant Susceptibility to IDDM.

Although some of the susceptibility loci that contribute to diabetes may represent rare mutations that confer aberrant gene products, it is becoming increasingly clear that many of the susceptibility genes contributing to this polygenic disorder may represent common allelic variants that manifest a diabetes phenotype only in a combinatorial way (22). Having established that the coding region of the β₂M gene in NOD mice is identical to the published sequence for the common “a” isoform of β₂M, we next assessed the possibility that this particular isoform was associated with diabetes susceptibility. We made NOD mice transgenic for the “b” isoform of the β₂M gene by using the same nondiabetes-prone promoter that was used for the β₂M^a transgenic NOD mice. The “b” isoform characterizing the IDDM-resistant NOR strain differs from the “a” isoform only at amino acid 85, in that it has an alanine instead of an aspartate. Similarly, NOD mice were made transgenic for the hβ₂M, which is ≈70% homologous with murine β₂M^a. Both the β₂M^b and hβ₂M transgenes were then crossed into β₂M^−/− NOD mice, as was done for the β₂M^a transgene. Both β₂M^b (Fig. 2a) and hβ₂M transgenic β₂M^−/− NOD mice expressed functional β₂M, as evidenced by normal β₂M-dependent MHC class I expression. Staining for MHC class I expression in NOD and hβ₂M transgenic β₂M^−/− NOD mice, respectively, showed a mean fluorescence intensity of 191 ± 7 and 235 ± 17 (K^d) and 121 ± 7 and 160 ± 10 (D^b). Expression of MHC class I was significantly higher in hβ₂M transgenic β₂M^−/− NOD than NOD mice (P < 0.01). Both hβ₂M and β₂M^b transgenic β₂M^−/− NOD mice were also shown to have normal CD8 T-lymphocyte numbers and function. Splenocytes from hβ₂M transgenic β₂M^−/− NOD responded normally in an MHC class I-specific mixed lymphocyte reaction (data not shown). β₂M^b transgenic β₂M^−/− NOD mice rejected class I disparate skin grafts (data not shown) and were able to mount a normal cytotoxic response to both vaccinia virus and influenza virus (Fig. 2b).

Colonies of β₂M^b transgenic and hβ₂M transgenic β₂M^−/− NOD mice and appropriate control mice were independently established at the John Curtin School of Medical Research and The Jackson Laboratories and were assessed for disease in terms of insulitis and hyperglycemia. Although control β₂M^+/− NOD mice and β₂M^a transgenic β₂M^−/− NOD mice, both of which expressed the “a” isoform of β₂M, developed diabetes, the β₂M^b transgenic and hβ₂M transgenic β₂M^−/− NOD mice, which lacked the β₂M^a isoform, showed markedly reduced susceptibility to diabetes (Fig. 3). That diabetes susceptibility is markedly reduced in both non-β₂M^a lines of NOD mice analyzed indicates such protection is not because of a nonspecific position effect of transgene insertion. Furthermore, the finding that diabetes protection was not observed when the β₂M^b or hβ₂M transgenes were expressed on a background in which the endogenous NOD β₂M^a allele was coexpressed (Fig. 4) indicates the former alleles do not provide dominant resistance, which also supports the fact they are unlikely to confer protection through a nonspecific position effect. Thus, β₂M^a is implicated as a dominant diabetes susceptibility gene in NOD mice.

Figure 4.

Degree of insulitis is not reduced in infiltrated β₂M^a or β₂M^b transgenic mice. The mean scores of female β₂M^+/− (n = 8), β₂M^−/− (n = 7), β₂M^a β₂M^−/− (n = 6), and β₂M^b β₂M^−/− (n = 9, mice with infiltrate only) NOD mice are shown. Black bars, 0; diagonal stripes, 1; vertical stripes, 2; gray, 3; and hatched, 4. The proportion of islets exhibiting each level of insulitis severity was not significantly different between β₂M^a, β₂M^b, and β₂M^+/− mice.

Replacing β₂M^a with Other Variants Does Not Impair MHC Class I Expression, but Does Limit Insulitis Development in NOD Mice.

In an attempt to understand the role of β₂M in the disease process, the severity of the insulitis lesion was compared histologically between β₂M^a and β₂M^b transgenic β₂M^−/− NOD mice. Although 100% of β₂M^a transgenic β₂M^−/− NOD mice and β₂M^+/− NOD mice develop insulitis, only 38% of β₂M^b transgenic β₂M^−/− NOD mice showed signs of infiltration; the remainder were completely free of insulitis. Of those mice that did develop insulitis, it was no less severe than that seen in β₂M^a transgenic β₂M^−/− NOD mice (Fig. 4). Further, no differences were found in the relative proportions of CD8 T lymphocytes, CD4 T lymphocytes, or macrophages in the insulitis lesion (Fig. 5a). The absence of insulitis in the majority of β₂M^b mice suggests that β₂M^a confers susceptibility at a very early stage of autoimmunity.

Figure 5.

(a) Content of lymphocytic infiltrate in β₂M transgenic mice. Consecutive frozen sections taken from 250- to 300-day-old female mice were stained with the antibodies 53.6 (CD8 specific), GK1.5 (CD4 specific), and F4/80 (macrophage specific). Sections from β₂M^+/− NOD, β₂M^a transgenic β₂M^−/− NOD, and β₂M^b transgenic β₂M^−/− NOD mice are shown. (Bar = 10 mm.) (b) Expression of transgenic β₂M on islet β cells is normal and correlates with level of lymphocytic infiltration. Islet cells from female, 13- to 17-week-old β₂M^b transgenic β₂M^−/− NOD mice, and age-matched control β₂M^a transgenic β₂M^−/− NOD mice were analyzed for H2-K^d expression. The β-cell population was identified by sorting, followed by immunohistochemical insulin staining, and the lymphocyte population was identified by double staining for CD3. Several experiments were performed, and a typical result is shown.

The level of MHC class I expression on islet β cells isolated from β₂M^a and β₂M^b transgenic mice was compared. Islet cells and the associated insulitis lesions were isolated and cultured in vitro with γIFN before analyzing their relative levels of MHC class I by FACS. Consistent with our earlier analysis of the lymphoid population, we found that the γIFN-induced level of class I expression on lymphoid cells from the insulitis lesion was no different between β₂M^a and β₂M^b transgenic NOD mice. This was also true for the γIFN-induced level of class I expression on islet β cells from the two groups. However, we did find that β₂M^b transgenic NOD mice generally had fewer islet β cells with up-regulated class I than β₂M^a transgenic NOD mice, and this was directly correlated with the severity of insulitis (Fig. 5b). It has previously been shown that MHC class I expression on β cells increases with severity of insulitis (23). Hence, the quantitative reduction in infiltrating leukocytes most likely accounts for the overall lower levels of MHC class I expression on β cells from β₂M^b transgenic NOD mice. However, it is important to note that MHC class I expression was equivalent on β cells from β₂M^a and β₂M^b transgenic NOD mice with matched levels of insulitis.

Replacement of β₂M^a with Another Isoform Inhibits the Development of Diabetogenic T-Cell Responses in NOD Mice.

Previous studies indicated IDDM resistance in a stock of NOD mice that express β₂M^b rather than β₂M^a, because of a homozygous NOR-derived chromosome 2 congenic interval, results from a reduced ability of nonhematopoietically derived cell types to support the development of diabetogenic T cells and/or target them to pancreatic β cells (13). However, it was possible that this decrease in the development or targeting of diabetogenic T cells resulted not from the loss of β₂M^a but rather through the effects of some other NOR-derived gene(s) in the chromosome 2 congenic interval. Thus, we tested whether the IDDM resistance induced in NOD mice by transgenic replacement of β₂M^a by another isoform also results from a reduced ability of nonhematopoietically derived cell types to support development of diabetogenic T cells or target them to pancreatic β cells. This was initially done by comparing the rate of IDDM development in reciprocal bone marrow chimeras between female standard NOD mice and the hβ₂M transgenic β₂M^−/− NOD stock. Controls consisted of female NOD mice reconstituted with syngeneic bone marrow. As expected, IDDM developed in 91.6% (11/12) of NOD control females over a 21-week period after reconstitution with syngeneic marrow (Table 1). In contrast, reconstitution of hβ₂M transgenic β₂M^−/− NOD recipients with NOD marrow elicited a significantly lower incidence of IDDM (30.0%, 3/10). Conversely, reconstitution of standard NOD recipients with hβ₂M transgenic β₂M^−/− NOD bone marrow elicited a very high rate of IDDM (81.3%, 13/16). Thus, the IDDM resistance that results from replacing the NOD β₂M^a variant with another isoform can be mechanistically explained by alternations in nonhematopoietically derived cell types that inhibit the development and/or targeting of β-cell autoreactive T cells.

Table 1.

Replacement of the murine β₂M^a isoform with a human variant inhibits diabetes development in NOD mice at the level of nonhematopoietically derived cells

Marrow donor	Recipient (1,200 R) treated	% IDDM 21-wk postreconstitution
NOD	NOD	91.6%  (11/12)
NOD	hβ2M transgenic β2M−/− NOD	30/0%  (3/10)*
hβ2M transgenic β2M−/− NOD	NOD	81.3%  (13/16)

Female recipients were lethally irradiated (1,200 R) at 4 weeks of age and reconstituted with 5 × 10⁶ T-cell-depleted bone marrow cells from the indicated donors. The chimeras were then monitored for IDDM development through 25 weeks of age. 

^*Significantly less (P < 0.005, χ² analysis) than in NOD recipients of syngeneic marrow. 

We next determined whether T cells originally generated in an environment where β₂M^a vs. hβ₂M was expressed on nonhematopoietically derived cell types differed in ability to subsequently transfer IDDM to lymphocyte-deficient NOD-SCID mice in which the MHC class I molecules expressed on β cells would all be associated with β₂M^a. At 6 weeks postreconstitution, splenic leukocytes from reciprocal chimeras between standard NOD mice and the hβ₂M transgenic β₂M^−/− NOD stock were transferred into NOD-SCID recipients. As expected, splenocytes from NOD mice that had been reconstituted with marrow from the hβ₂M transgenic stock efficiently transferred IDDM to 7/10 NOD-SCID recipients (Table 2). In contrast, splenocytes from hβ₂M transgenic β₂M^−/− NOD mice that had been reconstituted with standard NOD marrow failed to transfer IDDM to any NOD-SCID recipients (0/9). Equivalent levels of CD4 and CD8 T-cell repopulation were observed in NOD-SCID recipients of splenocytes from each type of reciprocal chimera. Collectively, these results demonstrate that replacing the NOD β₂M^a variant with another isoform induces alterations in nonhematopoietically derived cell types that inhibit the development or activation, rather than the targeting, of autoreactive diabetogenic T cells.

Table 2.

Replacement on nonhematopoietic cells of the murine β₂M^a isoform with a human variant blocks the selection, rather than the targeting, of autoreactive diabetogenic T cells

Chimeric splenocyte donor*	% IDDM in NOD-SCID recipients by 15-wk postsplenocyte repopulation	Splenic CD4 T levels in NOD-SCID recipients (%±sem)†	Splenic CD8 T levels in NOD-SCID recipients (%±sem)†
NOD BM→hβ2M Tg β2M−/− NOD	0%  (0/9)	33.2  ± 1.4  (n = 9)	5.8  ± 0.4 (n = 9)
hβ2M Tg β2M−/− NOD BM→NOD	70.0%  (7/10)	28.5  ± 1.2  (n = 10)	7.4  ± 0.5 (n = 10)

Splenocytes were pooled from the indicated reciprocal bone marrow chimeras (five each) and injected i.v. into 4 to 6-week-old female NOD-SCID mice (10⁷ cells/recipient) that were then monitored for diabetes development over a 15-week followup period. 

^*Donor splenocytes from the NOD BM→hβ₂M Tg β₂M^−/− NOD chimeras contained 31.1% CD4 and 9.0% CD8 T cells. Donor splenocytes from the hβ₂M Tg β₂M^−/− NOD BM→NOD chimeras contained 20.4% CD4 and 5.2% CD8 T cells. 

^† Levels of splenic CD4 and CD8 T-cell repopulation in the secondary NOD-SCID recipients were assessed at the onset of diabetes or at the end of the 15-week observation period. 

Conclusion

Using the technique of allelic reconstitution by transgenic rescue, we have implicated β₂M^a as one of the diabetes susceptibility genes within the NOD Idd13 region on chromosome 2. Because β₂M pairs with a number of functional molecules, there are many possible explanations for its role in susceptibility to IDDM. It is possible that β₂M confers susceptibility through its interaction with viral products or neonatal Fc receptor or, perhaps less likely, through its interaction with HFE. More likely is its role in pairing with either CD1 or the heavy chain of classical MHC class I molecules, both of which have an established role in IDDM via interaction with NKT cells and CD8 T lymphocytes, respectively (24, 25). However, we found no differences between β₂M^a and β₂M^b transgenic NOD mice in thymic NKT cell numbers. Furthermore, because IDDM resistance elicited by transgenic β₂M^b or hβ₂M isoforms is not dominant in mice that also express β₂M^a and is not mediated by hematopoietically derived antigen-presenting cells, protection is unlikely to result from the clonal deletion (26) or anergy (27) of autoreactive T lymphocytes. Instead, our results indicate that replacing the NOD β₂M^a variant with another isoform elicits its effect through nonhematopoietically derived cell types that are unable to support the development of β-cell autoreactive T-cell responses. This might be explained by the previously published finding that, when dimerized with different isoforms of β₂M, MHC class I molecules could present an altered peptide repertoire to CD8 T lymphocytes (8). Such alterations in antigen presentation may be because of alterations in the structural conformation of NOD MHC class I molecules upon dimerization with hβ₂M or β₂M^b rather than the β₂M^a isoform. Regardless of the mechanism, we have implicated β₂M as a susceptibility gene in NOD mice.

It is not yet possible to determine whether subtle variations in β₂M may also contribute to autoimmune diabetes in humans, because the extent of polymorphism within this gene has yet to be extensively investigated. However, that the β₂M^a allele that we have implicated as a dominant diabetes susceptibility gene in NOD mice is not a biologically aberrant variant has a potentially important implication for the pathogenic basis of this disease in humans. This implication is that our current findings strongly support the hypothesis that many diabetes susceptibility genes may actually represent common physiologically normal alleles, which exert pathogenic functions only in certain combinatorial contexts (22). Although not yet definitively proven, further support for this concept is that strong linkage disequilibrium implicates a number of other physiologically normal cytokine variants as candidate susceptibility genes (28, 29). Our current findings implicating β₂M as a susceptibility gene in NOD mice, in combination with these strong candidates, suggests that the search for type 1 diabetes susceptibility genes in humans should not be restricted solely to functionally defective variants.

Abbreviations

β₂M

β₂-microglobulin

hβ₂M

human β₂M

MHC

major histocompatibility complex

IDDM

type 1 diabetes

SCID

severe combined immunodeficient