We have searched the human genome for genes that predispose to type 1 (insulin-dependent) diabetes mellitus using semi-automated fluorescence-based technology and linkage analysis. In addition to IDDM1 (in the major histocompatibility complex on chromosome 6p21) and IDDM2 (in the insulin gene region on chromosome 11p15), eighteen different chromosome regions showed some positive evidence of linkage to disease. Linkages to chromosomes 11q (IDDM4) and 6q (IDDM5) were confirmed by replication, and chromosome 18 may encode a fifth disease locus. There are probably no genes with large effects aside from IDDM1. Therefore polygenic inheritance is indicated, with a major locus at the major histocompatibility complex.

Newly assembled major histocompatibility complex (MHC) class I molecules, together with the endoplasmic reticulum chaperone calreticulin, interact with the transporter associated with antigen processing (TAP) through a molecule called tapasin. The molecular cloning of tapasin revealed it to be a transmembrane glycoprotein encoded by an MHC-linked gene. It is a member of the immunoglobulin superfamily with a probable cytoplasmic endoplasmic reticulum retention signal. Up to four MHC class I-tapasin complexes were found to bind to each TAP molecule. Expression of tapasin in a negative mutant human cell line (220) restored class I-TAP association and normal class I cell surface expression. Tapasin expression also corrected the defective recognition of virus-infected 220 cells by class I-restricted cytotoxic T cells, establishing a critical functional role for tapasin in MHC class I-restricted antigen processing.

To facilitate large-scale genetic mapping of the human genome, we have developed chromosome-specific sets of microsatellite marker loci suitable for use with a fluorescence-based automated DNA fragment analyser. We present 254 dinucleotide repeat marker loci (80% from the Généthon genetic linkage map) arranged into 39 sets, covering all 22 autosomes and the X chromosome. The average distance between adjacent markers is 13 centiMorgans, and less than 4% of the genome lies more than 20 cM from the nearest marker. Each set of microsatellites consists of up to nine marker loci, with allele size ranges that do not overlap. We selected marker loci on the basis of their reliability in the polymerase chain reaction, polymorphism content, map position and the accuracy with which alleles can be scored automatically by the Genotyper program.

The role of human chromosome 2 in type 1 diabetes was evaluated by analysing linkage and linkage disequilibrium at 21 microsatellite marker loci, using 348 affected sibpair families and 107 simplex families. The microsatellite D2S152 was linked to, and associated with, disease in families from three different populations. Our evidence localizes a new diabetes susceptibility gene, IDDM7, to within two centiMorgans of D2S152. This places it in a region of chromosome 2q that shows conserved synteny with the region of mouse chromosome 1 containing the murine type 1 diabetes gene, Idd5. These results demonstrate the utility of polymorphic microsatellites for linkage disequilibrium mapping of genes for complex diseases.

The Tapasin molecule is a member of the immunoglobulin (Ig) superfamily required for the association of TAP transporters and MHC class I heterodimers in the endoplasmic reticulum. In this study, the Tapasin gene was precisely mapped in relation to the MHC. The gene was centromeric of the HLA-DP locus between the HSET and HKE1.5 genes and within 500 kbp of the TAP1 and TAP2 genes. A homologous mouse EST was mapped to a syntenic position on chromosome 17, centromeric of the H-2 K locus. Similarly, the rat Tapasin gene was shown to be in an equivalent location with respect to the RT1.A locus. The localization of Tapasin, TAP, LMP and class I genes within such a short distance of each other on the chromosome implies some regulatory or functional significance. We determined the Tapasin gene sequence for comparison of its structure to that of other Ig superfamily members, such as MHC class I genes. The IgC domain was encoded by a separate exon. However, the positions of the other introns were not characteristic of other Ig superfamily genes, indicating that Tapasin has a distinct phylogeny.

Chromosome locations of non-major histocompatibility complex (MHC) genes contributing to insulin-dependent diabetes mellitus (IDDM) in mice have been determined by outcrossing NOD mice to other inbred strains congenic for the NOD MHC haplotype (H2g7). At least nine non-MHC IDDM susceptibility genes (Idd) were previously identified at first backcross (BC1) after outcross of NOD to C57BL/10.H2g7 congenic mice (B10.H2g7). We investigated whether the same set of Idd loci segregated with IDDM susceptibility after outcross of NOD to NON.H2g7 congenic mice. Since the outcrosses to NON.H2g7 and B10.H2g7 were performed in the same vivarium, direct comparisons were made of the chromosomal locations and relative strengths of Idd alleles in diabetic progeny from the two different outcrosses. In comparison with the NOD x B10.H2g7 outcross, the NOD x NON.H2g7 outcross produced significantly higher IDDM frequencies in F1, F2, and BC1 generations. The high F2 diabetes frequency allowed evaluation of the effects of homozygous expression of both the susceptibility and the resistance allele at Idd loci. This analysis demonstrated that no single non-MHC Idd locus was essential for the onset of diabetes in this cross. After outcross to NON.H2g7, Idd4 (chromosome [Chr] 11), Idd5 (Chr 1), and Idd8 (Chr 14) did not segregate with IDDM in either the BC1 or the F2 generation. Diabetogenic NOD-derived alleles at Idd2 (Chr 9), Idd3 (Chr 3), and Idd10 (Chr 3) were segregating in the BC1. An NON-derived allele contributing to susceptibility on Chr 7 (Idd7) was also detected. Dominant traits, detectable only in the F2 cross, were encoded by Chr 4 (Idd9) and two newly mapped loci on Chr 13 (Idd14) and 5 (Idd15). A third dominant trait was encoded by Chr 6 (possibly Idd6), but here, in contrast to Idd9, Idd14, and Idd15, the NON allele was diabetogenic. Stepwise logistic regression analysis of the BC1 and F2 data confirmed that the ability to identify certainty of the non-MHC Idd loci was contingent on the extent of homozygosity for NOD background genes. This study shows that the diabetogenic phenotype can be achieved through the actions of variable combinations of MHC-unlinked genes and a diabetogenic MHC haplotype.

Antigenic peptides are presented to cytotoxic T lymphocytes by heterodimers of MHC class I molecules and g 2 -microglobulin. Peptides are generated in the cytosol and translocated into the endoplasmic reticulum (ER) through the transporter associated with antigen processing (TAP). Optimal binding of peptides to class I molecules is facilitated by the physical association between class I heterodimers and TAP. This association is mediated largely by the glycoprotein tapasin. Analysis of tapasin function has relied on a mutant cell line, .220, which is defective in tapasin expression and antigen presentation. We have investigated the genetic basis of these defects. In .220 cells, Tapasin transcripts lack exon two. This is caused by a single nucleotide substitution, disrupting the 5' splice site of the second intron. A tapasin protein is produced in .220 cells, but has a truncated signal peptide and lacks the N-terminal 49 amino acids encoded by full-length transcripts. Nonetheless, this truncated form is translocated into the ER and interacts with TAP. As a result of alternative splicing, transcripts lacking exon two are also present in wild-type cells, although no truncated protein was detected. Additionally we describe a polymorphism in the Tapasin gene, with two alleles encoding arginine or threonine at peptide position 240.

The CD3 gene region on chromosome 11q23 has been implicated in susceptibility to type 1 (insulin-dependent) diabetes mellitus. Using semi-automated fluorescence-based technology, we have undertaken association and linkage analysis of a dinucleotide microsatellite in the CD3 delta (CD3D) gene. We have also performed a large case-control analysis of a restriction fragment length polymorphism (RFLP) in the CD3 epsilon (CD3E) gene, 26 kb from CD3D. We found no evidence for the previously reported association between the 8 kb allele of the RFLP and disease in a UK dataset of 403 diabetic patients and 446 nondiabetic controls. Furthermore, the use of the transmission/disequilibrium test (TDT) showed no evidence of linkage or association to type 1 diabetes at either marker locus. We conclude that the CD3 gene region does not contribute significantly to IDDM susceptibility. We have successfully applied semi-automated, fluorescence-based technology to undertake association analysis on the CD3D microsatellite. Moreover, by analysing 94 other dinucleotide repeat markers, we conclude that fluorescence-based methodology can generally be applied to large-scale, semi-automated association studies with most microsatellite markers.