Further Mapping of the Idd5.1 Locus for Autoimmune Diabetes in NOD Mice

The Idd5 locus was initially mapped to the proximal half of murine chromosome 1 (1,2). It was associated with susceptibility to both spontaneous and cyclophosphamide-induced diabetes (1) and also with inflammatory infiltration of the islets of Langerhans, including periinsulitis and insulitis (2,3). To further study the Idd5 locus, we generated NOD mice congenic for a 30-cM interval of C57BL/6 (B6) origin extending from D1Mit478 to D1Mit26 (NOD-B6.Bcl2) (see Table 1). An intercross between heterozygous mice was performed at the 7th generation of backcrossing to confirm the existence of Idd5 and to initiate its fine mapping. Progeny mice were genotyped, and development of spontaneous diabetes in female mice was monitored for 9 months. Incidence of diabetes was markedly influenced by genotypes at the D1Mit18 (Fig. 1A) and D1Mit5 (not shown) markers and at the proximal end of the selected region, but not at D1Mit44 and the more distal loci (not shown). The protective effect of the B6 haplotype was partially dominant (P = 0.008). Similar results were obtained when diabetes was induced by injection of cyclophosphamide in progeny mice from heterozygous parents at the 11th generation of backcross. Genotypes at D1Mit18 (Fig. 1B) and D1Mit5 but not at more distal loci (not shown) significantly affected susceptibility to this form of diabetes, which is otherwise specifically induced in NOD mice (4). Interestingly, the protective effect of the B6 haplotype was dominant (P = 0.008).

The Idd5 interval was recently circumscribed by the use of congenic and recombinant congenic mice (5). Evidence was also obtained for two loci, Idd5.1 and Idd5.2, that were closely linked and both predisposing to autoimmune diabetes. Importantly, the human synteny equivalent to Idd5.1 is on chromosome 2q33 and overlaps the IDDM12 locus for human type 1 diabetes (5).

Here, we have further narrowed down the Idd5.1 interval using new recombinant congenic lines. The R1 line harbored a segment of B6 origin (Table 1) between D1Mit478 and D1Mit178 (both markers included) that overlaps with that defined by the Idd5R467 recombinant congenic strain (5). As shown on Fig. 2A, R1 mice had a significantly reduced incidence of spontaneous diabetes (P = 1 × 10⁻⁵) that was approximately half of that of their NOD littermates. These mice were also protected from diabetes induced by cyclophosphamide (P = 9 × 10⁻⁵) (Fig. 3A). Additional recombinant congenic strains were selected (Table 1). The R6 and R16.14 subcongenics retained various segments of B6 origin in the proximal part of the R1 segment. These mice had an incidence of diabetes similar to that of NOD mice (Fig. 2B). Conversely, the R67 and R39 recombinants, which retained a B6 chromosomal segment in the distal part of the R1 interval, were protected from diabetes (P < 1 × 10⁻⁵ for both R67 and R39) (Fig. 2A).

Altogether, we now map the Idd5.1 locus to the interval defined by the R39 recombinant congenic strain. The centromeric boundary of this interval lies between the D1Mit279 and D1Mit302/D1Mit22 markers. This location therefore rules out the segment comprising the region between D1Mit478 and D1Mit302/D1Mit22 that was previously part of the Idd5.1 interval (5). The distal boundary of R39 is more loosely limited and maps between the D1Mit178 and D1Mit19 markers, and thus it cannot be distinguished with currently available markers from the distal boundary defined by the Idd5R467 line (5). As a result, the size of the Idd5.1 interval ranges from 3.0 to 5.0 cM.

The newly restricted Idd5.1 locus excludes Cflar and Casp8, two important candidate genes (5). It still retains the two costimulatory receptors Cd28 and Ctla4. No coding polymorphism between NOD and B6 was identified in these two genes (6; S.-E.L.-C., O.B., unpublished observations). By endonuclease restriction of both genomic DNA (Southern blot analysis) and polymerase chain reaction (PCR)-amplified intronic sequences, restriction fragment–length polymorphism (RFLP) between the two strains were characterized (see research design and methods). Typing of recombinant haplotypes for these RFLPs allowed for the placement of Ctla4 telomeric to Cd28.

Such noncoding polymorphisms could have a regulatory function that results in differential expression of Ctla4 or Cd28. Indeed, Ctla4 is induced in activated T-cells at a lower level in NOD than in B6 mice (6) (Fig. 4). This is consistent with a contribution of the Ctla4 gene to the autoimmune predisposition of NOD mice, because Ctla4 is viewed as a negative regulator of T-cell activation (7), and Ctla4-null mice develop early lethal pleiotropic dysimmune manifestations (8). However, activated T-cells from R67 congenic mice expressed Ctla4 at a low level, like their NOD parent (Fig. 4). The decrease of inducible Ctla4 expression in NOD T-cells is therefore unlikely to be determined at the Ctla4 locus itself or at a closely linked gene included in the Idd5.1 interval.

Inducible costimulator (Icos), which facilitates T-cell activation, is another important regulatory molecule homologous to Ctla4 and Cd28 (9). The Icos gene was located in the vicinity of Cd28 and Ctla4 (10). A polymorphic dinucleotide repeat was identified in the second intron of the Icos sequence. Using this marker, Icos was located in the R39 interval, telomeric to Cd28 and tightly linked to Ctla4, D1Nds25, and D1Mit249. Because the expression and nucleotide sequence of Icos have not yet been characterized in NOD mice, it is an important candidate that warrants additional investigation.

The Idd5 region defined by the large Idd5R8 segment did not affect inflammatory infiltration of the islets of Langerhans, as determined from tissue sections stained with hematoxylin and eosin (5). This was also the case for the smaller Idd5.1 interval harbored by the R67 recombinant congenic line (data not shown). Transfer experiments were performed to determine whether diabetes protection occurred before or after the entry of these inflammatory cells into the islets of Langerhans. Prediabetic spleen cells from NOD or R67 donor mice were infused into immunodeficient NOD.scid/scid recipient mice. As shown in Fig. 3B, spleen cells from both donors were equally capable of transferring diabetes. This finding is consistent with an association of Idd5.1 with resistance to induction of diabetes by cyclophosphamide, because this drug is thought to require the presence of insulitis to be effective. The Idd5.1 gene might therefore act at a late stage of disease development, subsequent to the onset of insulitis and possibly taking place in the islets of Langerhans. For example, it might affect the late differentiation of inflammatory cells into cytotoxic cells or modulate β-cell sensitivity to autoimmune aggression.

With the rapid progress of the Mouse Genome Sequencing program, the current size of the Idd5.1 interval makes the identification of the Idd5.1 gene a feasible task. Construction and sequencing of a contig covering the region of interest should be carried out concurrently with the search for DNA polymorphisms. Knowing the site of expression of Idd5.1 should be helpful in testing correlations between these polymorphisms and quantitative variation of candidate gene expression. Such quantitative variation could be the basis of Idd5.1 if the pathogenic DNA alteration maps outside the coding region.