Type 1 Diabetes: Cellular, Molecular & Clinical Immunology

Chapter 7 - Type I Diabetes Mellitus of Man: Genetic Susceptibility and Resistance
Andrea K. Steck¹, Alberto Pugliese², and George S. Eisenbarth^1
1Barbara Davis Center for Childhood Diabetes, University of Colorado Health Sciences Center
²Diabetes Research Institute, University of Miami

Updated 7/08, slides updated 9/09 Click to download Powerpoint slide set
Additional Slideset: Chapter7_Sardinia.ppt Type 1 Diabetes and Related ADs in Sardinia (M. Songini) - Slides added 1/06
Additional Slideset: Chapter7_Japan.ppt Genetics Type 1A Diabetes in Japan (H. Ikegami) - Slides added 10/05

Besides inherited alleles, other mechanisms regulating gene expression including epigenetic and parent-of-origin effects influence susceptibility by modifying the transmission and transcription of inherited genes. It is also an intriguing possibility that additional genetic factors or their expression may be acquired after birth, perhaps through environmental exposures. Thus, a variety of genetic mechanisms may influence the autoimmune responses leading to ß-cell destruction. This chapter will review the current knowledge about the genetics of type 1 diabetes in humans.

Figure 7.2. Odds ratios for a series of identified “genes/genetic loci” from recent genome screens and replication studies.

Inherited Susceptibility Loci
Both association studies and linkage analysis using various analytical methods have been used to identify IDDM susceptibility loci. These are conventionally noted using the abbreviation IDDM and a number, e.g. IDDM1, IDDM2, etc., with the number usually reflecting the order in which such loci were reported (Table 7.1 and Figure 7.2). Many of the early IDDM loci appear at present to have been “false positives” and are generally being replaced by identified genes (figure 7.2), though small influences on diabetes susceptibility that cannot presently be proven for such loci with the current large genome screens is a possibility. Using the candidate gene approach, association studies provided evidence for the first two susceptibility loci, the HLA region (IDDM1) and the insulin gene (INS) locus (IDDM2). These two loci contribute the great majority of familial clustering (Figure 7.2), suggesting the existence of additional loci. One estimate is that the MHC alone contributes 41% of the familial clustering of type 1 diabetes of the 48% estimated to be accounted for with all known genes (22). The next most potent locus for type 1 diabetes of man was also discovered using a candidate gene approach, namely the PTPN22 (LYP) gene with an odds ratio of approximately 1.7 for a “missense” mutation that creates susceptibility to multiple autoimmune disorders (28-30). Figure 7.3 illustrates allele frequencies in case and control populations from the Wellcome Trust Consortium genome analysis, combining both control population and a series of non-autoimmune disorders to provide the reference population. The ratio of differences in frequencies, except for PTPN22 are relatively small, making it unlikely that the other indicated loci will contribute to the genetic prediction of type 1A diabetes, in contrast to the HLA and insulin region genes. For instance the HLA DR3/4-DQ2/8 genotype is present in 2.3% of newborns in Colorado, but more than 30% of children developing diabetes, providing “extreme” risk, as will be discussed subsequently. Compared to a population prevalence of type 1 diabetes of approximately 1/300, DR3/4-DQ2/8 newborns from the general population have a 1/20 genetic risk (31). As will be discussed subsequently additional loci within or linked to the MHC (Major Histocompatibility Complex) can increase this risk for first degree relatives of DR3/4-DQ2/8 newborns to as high as 80% (23). Such extreme risk, suggests that for this major subgroup of children, the bulk of familial aggregation is determined by alleles of genes within or linked to the classic MHC, and the search for additional (non-DR and DQ) genetic determinants in this region is underway (32-37).

Figure 7.3. Allele frequencies for case versus control association studies with “significant” associations outside of the major histocompatibility complex.

Prior to the whole genome SNP analyses that have recently been reported, a number of genome-wide scans using families and affected sibling-pairs have been performed since the mid 1990’s in an attempt to identify susceptibility loci using linkage analysis (38). Linkage analysis confirmed linkage with IDDM1 and IDDM2 and further provided evidence for the existence of approximately 20 susceptibility loci. Many of these loci show modest linkage and linkage is often not confirmed in all genome scans. Sample size and composition, genetic heterogeneity and analytical methods underlie much of the variability observed in these studies. A coordinated effort to investigate the genetics of the disease, the Type 1 Diabetes Genetics Consortium (T1DGC) (www.t1dgc.org), has been launched and involves the study of patients and their families from around the world. In 2005 the consortium published its first report, an interim, combined linkage analysis of four datasets, three previously published genome scans, and a new dataset of 254 families. This analysis included 1,435 families with 1,636 affected sibling pairs, representing one of the largest linkage studies ever performed for any common disease and involving families from the U.S., U.K. and Scandinavia (39). Given the average map information content (67%, >400 polymorphic microsatellite markers in each scan), this dataset had ~95% power to detect a locus with λS λ1.3 and p= 10-4. With this analytical power, more than 80% of the genome was found not to harbor susceptibility genes of modest effect that could be detected by linkage. The study confirmed linkage with IDDM1 (nominal P = 2.0 x 10–52). Moreover, nine non–HLA-linked regions showed some evidence of linkage (nominal P < 0.01), including three at (or near) genome-wide significance (P < 0.05): 2q31-q33, 10p14-q11, and 16q22-q24. In addition, after taking into account the linkage at the 6p21 (HLA) region, there was evidence of linkage with the 6q21 region (IDDM15). The published literature on these loci is discussed in detail in the following paragraphs. A comprehensive list of these initial susceptibility loci is shown in Table 7.1 with LOD scores and λS from the 2005 T1DGC scan (39).

Table 7.1. Susceptibility Loci for Type 1 Diabetes as of “2005”.

The recent whole genome screens, with increasing power suggest as indicated above that many of the prior loci are either false positives, have such small effects that they were not detected in the genome screens, or are related to only specific populations, as for instance is suggested for the SUMO4 gene for only Asian patients (40). Table 7.2 summarizes “significant” regions for the whole Wellcome Trust case control study using the combined “control” reference population of 7,670 controls compared to 2,000 patients with type 1 diabetes (The locus for IFIH1 did not reach “significance” in this Wellcome whole genome analysis with the SNPs analyzed, but is included in Table 7.2 related to a follow-up study (22)).

Table 7.2. Post-whole genome screens (2007) - Susceptibility Loci for Type 1 Diabetes; Bolded loci <10 (-7);underline <10 (-5) WTCCC analysis.

The Major Histocompatibility Complex, IDDM1
The major locus for type 1 diabetes susceptibility is located within the HLA (Human Leukocyte Antigen) region (41) on the short arm of chromosome 6 (42) and is calculated to provide up to 40-50% of the inheritable diabetes risk (43), though this calculation is based upon certain assumptions, including negligible recombination between susceptibility loci in the region. We believe this assumption may not be valid, if loci, far telomeric or centromeric of DR and DQ class II genes, contribute to genetic susceptibility. The HLA complex was first linked to diabetes when associations with several HLA class I antigens (HLA-B8, -B18, and -B15) were discovered by serological typing and affected sib-pairs showed evidence of linkage (44-46). With the development of novel typing reagents, HLA class II genes (DQ, DR, and DP in that order of risk) were shown to be even more strongly associated with the disease (46-48). However, several loci within or near the HLA complex appear to modulate diabetes risk, and add further complexity to the analysis of IDDM1-encoded susceptibility (23;49). Alleles, modes of inheritance and putative mechanisms of susceptibility encoded for at the IDDM1 locus are discussed below. A schematic representation of the HLA region and its association with IDDM is shown in Figure 7.4. A recent manuscript analyzes a large number of families of the Type 1 Diabetes Genetics consortium an documents the influence of multiple class II DR and DQ alleles and genotypes influencing risk of type 1A diabetes (50).

Figure 7.4. The HLA Region and IDDM Susceptibility. Schematic representation of the HLA region showing microsatellite markers, loci, and alleles associated with IDDM susceptibility. Distances between loci are grossly approximated.

The HLA Class II Region
The great majority of Caucasian patients have the HLA-DR3 or -DR4 class II alleles and approximately 30% to 50% of patients are DR3/DR4 heterozygotes (51). The DR3/DR4 genotype confers the highest diabetes risk with a synergistic mode of action, followed by DR4 and DR3 homozygosity, respectively (52) [See attached teaching slides by J. Noble summary of HLA nomenclature]. Following the development of DNA-based sequencing and typing technology, the HLA-DQ locus was found to be the most strongly associated with diabetes susceptibility. This locus encodes for multiple variants of the HLA-DQ molecule, a heterodimer consisting of two chains (a and ß) involved in immune recognition and antigen presentation to CD4 T cells. In Caucasians, the HLA-DQ heterodimers (the α­chain genes are labeled DQA1 and the ß-chain genes DQB1) encoded by the DQA1*0301, DQB1*0302 and DQA1*0501, DQB1*0201 alleles have the strongest association with diabetes. These alleles are in linkage disequilibrium with the HLA-DR4 and -DR3 alleles (Table 7.3), respectively (53). Linkage disequilibrium often extends centromeric and telomeric of the class II region (54-56).

Table 7.3. HLA Class II DR-DQ Linkage Patterns and IDDM Susceptibility in Whites.

Allelic variation at the DQB1 locus differentiates diabetes susceptibility among the two most common HLA-DR4 haplotypes found in Caucasians based on the presence of the DQB1*0302 or DQB1*0301 allele. Most patients with DR4 carry the DQB1*0302 allele, while the DQB1*0301 and *0302 alleles are more evenly distributed in the general population. An independent effect has not been demonstrated for DQB1*0201 because of the strong linkage disequilibrium between DQB1*0201 and DRB1*0301 on Caucasian DR3 haplotypes. However, DQB1*0201 does not confer increased susceptibility in association with DRB1*0701 on DR7 haplotypes. The different risk conferred by DQB1*0201 when on a chromosome with DR3 or DR7 may be explained by the different DQA1 alleles associated with DQB1*0201 on such haplotypes (DQA1*0501 on DR3, DQA1*0201 on DR7) (57;58). In addition, other susceptibility loci may be in linkage disequilibrium with DQB1*0201 on DR3 haplotypes (43), although the class II region may be the primary risk determinant on DR3 haplotypes (59).  Trans-complementation of DQ α- and ß-chains from opposite haplotypes has been demonstrated, and this significantly increases the diversity of class II antigens participating in the immune response and the potential for HLA-DQ contribution to IDDM susceptibility. A trans-complementing DQ molecule would be unique to a heterozygous individual and usually it would not be expressed in his parents. Thus, this phenomenon has been proposed as an explanation for the increased diabetes risk observed in DR4, DQA1*0301, DQB1*0302/DR3, DQA1*0501, DQB1*0201 heterozygotes (58).
DQB1*0302 differs from DQB1*0301 at position 57, where it lacks an aspartic acid residue, similar to the I-A molecule of the NOD mouse (reviewed in ref. (60)). The DQB1*0201 allele also lacks aspartic acid at position 57, and it has been proposed that this residue may be involved in the molecular mechanism underlying IDDM1-encoded susceptibility (53;56;56). In fact, the amino acid residue at position 57 of the DQ-ß chain appears to be critical for peptide binding and recognition (61). Other residues of the DQ-ß chain may influence peptide binding and diabetes susceptibility, and in particular the combined variation of residues at positions 57 and 70 seem to more strongly correlate with diabetes risk (62;63;63). An arginine residue at position 52 of the DQ-α chain also correlates with diabetes susceptibility (57). The importance of the residue at position 57 has been disputed by trans-racial studies showing that DQB1 alleles found with increased frequency in Japanese patients carry instead of lack an aspartic acid residue at this position (64). Patients carrying similar Asp57 high risk alleles are also found among Caucasians (65-74) (Figure 7.5). Moreover, certain low risk DQB1 genotypes also lack aspartic acid at position 57, including DQB1*0302/DQB1*0201 (DR7), and DQB1*0201 (DR3)/DQB1*0201 (DR7).  
It is important to recognize that even the class II MHC genes with the greatest impact on diabetes susceptibility have a complex inheritance and their effect on risk cannot be explained by relatively simple rules (for instance, based on the presence of certain amino acid residues in the DQ genes). As illustrated below (Figure 7.5), the rule that lack of aspartic acid at position 57 of the DQB1 gene is strongly associated with risk is not consistent with relatively potent diabetogenic DQ alleles such as DQB1*0303 and DQB1*04 (usually 0401 in Caucasians and 0402 in Korean and Japanese patients) (75).

Figure 7.5. High-risk genotypes in population based Oxford, England study with Asp57+ high risk DQB alleles marked with an asterisk.

However, there is clear evidence that certain residues have a functional role in determining binding and presentation of certain peptides (76). By using X-ray crystallography, investigators have determined the three-dimensional structure of the HLA-DQ8 molecule (encoded by DQA1*0301/DQB1*0302) complexed with an immunodominant peptide of the insulin molecule (insulin B:9-23) (77). The DQ8 structure suggests that the residue at position 57 contributes to the shaping of the P9 pocket, which together with the P1 and P4 pockets appear relevant to diabetes susceptibility. The P4 pocket is deeper in DQ8 compared to DR1, DR2, DR3, DR4 but predictably similar in HLA-DQ2 (DQA1*0501, DQB1*0201) and the diabetes protective HLA-DQ6 (DQA1*0102, DQB1*0602), thus not directly correlating with susceptibility. Moreover, the binding pockets of HLA-DQ8 were similar to those of HLA-DQ2 and to those of the I-Ag7 molecule (corresponding to human DQ), the main genetic susceptibility locus in NOD mice. This finding suggests that diabetes may depend on antigen-presentation event (s) that may be similar in humans and NOD mice. In further support of this hypothesis, it has been shown that HLA-DQ8 and I-Ag7 select common peptides, use the same binding register, which is not promiscuous and is rather selective and dominated by the P9 pocket (78;79).
IDDM1-encoded susceptibility is mostly conferred by alleles of the HLA-DQ locus in the class II region. The above conclusion is also supported by the fact that the DQA1*0102, DQB1*0602 alleles, encoding for the HLA-DQ6 heterodimer found on HLA-DR2 haplotypes, confer dominant protection from the development of type 1 diabetes (reviewed in ref. (80)). Among four common DR2 haplotypes observed in Caucasians, the DQA1*0102, DQB1*0602, DRB1*1501 haplotype is negatively associated with type 1 diabetes and is reported in less than 1% of patients in most populations studied, including those of Caucasian (both European and North-American) (70;71;77;81-86), Asian (64;87;88), African-American (69;70), and Mexican-American origin (89).
The DQB1*0602 allele in particular is the only class II allele exclusively found on protective DR2 haplotypes while all the other alleles (DQA1*0102, DQA1*0103, DQB1*0601, DQB1*0502, DRB1*1501, DRB1*1502, DRB1*1601) can be found on neutral or moderately predisposing DR2 haplotypes. Moreover, a few rare patients with type 1 diabetes have been described carrying mutated DQB1*0602 alleles or unusual DQA1/DQB1 alleles in cis with the usual DRB1*1501 allele. Thus, the available evidence suggest that the diabetes-protective effect associated with DR2 haplotypes may be mostly mapped within the DQ locus and in particular to the DQB1*0602 allele. Although a number of patients with DQB1*0602 have been identified (86), the overall number is small (approximately 1% of children developing diabetes and perhaps 5% of adults from the Swedish population) (90). Protection appears to be dominant since DQB1*0602 protects from diabetes even in the presence of high-risk HLA alleles (68;81).
However, a subset of HLA-DR2, DQB1*0602 haplotypes marked by alleles at the D6S265 locus has been identified as less protective (but still markedly protective) in a Swedish cohort. HLA-DR2 (DRB1*15), DQB1*0602 haplotypes carrying D6S265*15 have a ten-fold higher odds ratio (OR) than those carrying other alleles and thus confer reduced protection (OR with D65265*15 0.186 (.074 to .472) versus .017 (.005 to .062)). Marker D6S265 maps 100 kb telomeric of the HLA-A locus, which has been previously associated with diabetes susceptibility. Associations between D6S265 and other autoimmune diseases have been reported, including an association with multiple sclerosis and D6S265 specifically on HLA-DRB1*15, DQB1*0602 haplotypes (91). Thus, genetic variation at D6S265 can influence or is linked to a locus that can influence susceptibility to or protection from the autoimmunity conferred by HLA-DRB1*15, DQB1*0602 haplotypes. Known genes that may be marked by D6S265 include HLA-A, HLA-B, MICA, TNF and BAT1. Polymorphisms at these loci may have important effects on the function of cytotoxic T cells and cytokine secretion. Moreover, possible effects on transcriptional regulation may perhaps influence the expression of the HLA-DQ molecule encoded by DQA1*0102, DQB1*0602. Further characterization of this region will be needed to identify the loci that contribute to the genetic protection from type 1 diabetes conferred by DRB1*15, DQB1*0602.
Studies in transgenic mice have provided direct evidence that the DQ locus, and the DQB1*0302 allele in particular, can engender an immune response leading to the development of diabetes (69). However, the mechanism by which the HLA-DQ locus influences diabetes susceptibility is the subject of intense speculation. Since HLA-DQ molecules are known to play a role in antigen presentation, allelic variation at this locus may affect the binding and functional properties of DQ heterodimers and in turn the presentation of islet cell antigen-derived peptides to immunocompetent cells. Protective HLA molecules may have higher affinity for one or several peptides than predisposing molecules. This is also suggested by the report that the NOD mouse I-Ag7 molecule, homologous to the human DQ8 molecule encoded by DQA1*0301/DQB1*0302, is generally a poor peptide binder (70). Therefore, it is suggested that predisposing HLA molecules may be ineffective at binding and presenting peptides derived from islet cell antigens. Indeed, the HLA-DQ molecules encoded by the protective DQA1*0102, DQB1*0602 and predisposing DQA1*0301, DQB1*0302 alleles appear to differ in their affinity and specificity for peptides derived from the insulin, glutamic acid decarboxylase (GAD), and tyrosine phosphatase IA-2 autoantigens (71;72;72). Similar findings were reported for DR molecules, with protective HLA-DR2 (DRB1*1501) molecules displaying stronger affinity for (pro) insulin peptides than susceptible HLA-DR3 (DRB1*0301) molecules (73).
It is unclear whether genetically determined differences in peptide binding and presentation affects the shaping of the T-cell repertoire in the thymus or modulates immune responses in the extra-thymic periphery. A poor presentation in the thymus could impair mechanisms of negative selection allowing autoreactive T cells to escape deletion. In contrast, a protective HLA-DQ molecule could promote tolerance to ß-cell molecules by eliciting more efficient antigen presentation and negative selection in the thymus. Although several studies involving the transgenic expression of MHC molecules in mice did not support this hypothesis (74;92), a study provided novel evidence for thymic deletion as a mechanism of protection associated with MHC genes in transgenic mice (93). Moreover, the demonstration that insulin and other islet cell antigens are ectopically expressed in human thymus (83;84) indirectly supports the hypothesis that thymic self-antigen presentation and deletional mechanisms may be affected by the affinity and binding properties of HLA-DQ and HLA-DR molecules.
An alternative hypothesis is that DQB1*0602-associated protection may be mediated outside the thymus through the stimulation of regulatory immune responses associated with peripheral tolerance. The predominance of Th2 responses is usually associated with lack of progression to overt diabetes (reviewed in ref. (85)) and regulatory T-cells are essential for the prevention of autoimmunity. There is indeed evidence that a non-diabetogenic immune response, mostly limited to the production of autoantibodies against the GAD autoantigen, may occur in first degree relatives with DQB1*0602 (68), and in whom the presence of DQB1*0602 and DQA1*0102 has been confirmed by direct sequencing of the second exon (86). A similar response has been reported in patients with type 1 autoimmune polyendocrine syndrome who do not invariably progress to overt diabetes (94;95). Moreover, a similar protective effect has also been reported in first degree relatives participating in the ongoing Diabetes Prevention Trial (DPT-1), although different degrees of protection may occur in different ethnicities (96). The presence of GAD autoantibodies, often at high titers, may reflect the predominance of Th2 responses in relatives with DQB1*0602. Finally, the two hypotheses are not mutually exclusive and DQB1*0602-associated protection could be mediated both in the thymus and the periphery.
Two additional strongly protective haplotypes are DRB1*1401, DQA1*0101, DQB1*0503 and DRB1*0701, DQA1*0201, DQB1*0303. The DRB1*1401 haplotype is particularly interesting in that it is a HLA-DR allele with an apparent lack of transmission to affected children as dramatic as for DQB1*0602 (Both DRB1*1401 and DQA1*0201/DQB1*0303 are relatively infrequent but strongly protective) (97). (Fig. 7.6).

.

Figure 7.6. Transmission of DR/DQ haplotypes to patients with type 1 diabetes. Note that DQ6 (DQB1*0602) containing haplotypes and DRB1*1401 containing haplotypes are not/rarely transmitted to diabetics, while the usual DR or DQ alleles associated are transmitted when DQB1*0602 (e.g. DRB1*1501) or DRB*1401 (DQB1*0503) are not present in the haplotype.

Other loci in the class II region have been associated with diabetes susceptibility besides HLA-DQ. Several studies indicate that DRB1 alleles (Figure 7.7) significantly contribute and modulate diabetes susceptibility (55;98-105). The DRB1*0405 and *0401 alleles have been reported as predisposing, *0402 and *0404 as mostly neutral, while *0403, *0406, and *0407 appear to be protective.

Figure 7.7. Modified risk of diabetes relative DRB1*04 alleles, with DRB1*0403/0403 even when combined with high risk DQB1*0302, decreasing risk to background population (approximately 1/300).

There is evidence for contribution to risk from the DPB1 locus, confirmed in an extensive analysis by Valdes et al. (106-108). An independent association has been observed in Mexican American (109) and Caucasians with DPB1*0301 (110). The frequency of DPB1*0101 is increased in patients (almost exclusively found on DR3 haplotypes). The maternal transmissions of DRB1*0301-DPB1*0101 haplotypes to affected children occurred twice as frequently as do paternal transmissions (43). Transmissions of DR3 haplotypes carrying other DPB1 alleles occurred at approximately equal maternal and paternal frequencies. A recent analysis indicates that the DPB1*0402 allele, previously associated with decreased diabetes risk, is associated with dominant protection from development of anti-islet autoantibodies and diabetes in young children (32) amongst children having the highest risk DR3/4-DQ2/DQ8 genotype.
It is controversial whether loci (TAP1, TAP2) encoding for peptide transporter genes associated with antigen processing and localized centromeric to the DQ loci may also affect IDDM susceptibility (36;111). Homozygosity for the TAP2*0101 allele was associated with increased IDDM risk independent of HLA-DQ susceptibility in a French study (112), but other studies have failed to show such independent effect and suggested linkage disequilibrium between the HLA-DQ and TAP2 loci (113;114). Of note, a mutation at the same locus has been implicated as the cause of the class I deficiency associated with IDDM in studies in humans and NOD mice (115-119).
The HLA Class I Region
A number of observations indicate that class II genes cannot explain all of the HLA association with IDDM. A role for HLA complex genes other than the DR-DQ or other class II genes was first demonstrated by Robinson et al. (120). They examined affected sib pairs with parents homozygous for the DR3 haplotype and used the HLA class I B locus to distinguish between the two DR3 haplotypes of the homozygous parent. Under the null hypothesis that no HLA region variation additional to that defined by the DR3 haplotype is involved in IDDM, the affected sib pairs should share the two parental DR3 haplotypes at the same frequency. Significant deviation from 50% sharing was observed. Since the DR3 haplotypes examined in this study could be assumed to be homogeneous for their DR-DQ alleles at the molecular level (DRB1*0301 DQA1*0501 DQB1*0201), this test implicated other HLA loci in IDDM susceptibility. Several other reports suggest that HLA class I genes, and in particular the HLA-A24 allele, may also influence susceptibility and particular clinical aspects of the disease such as age of onset (121;122) and the rate of  ß-cell destruction (93;123-127). Besides HLA-A24 (*2401), other class I alleles are independently associated with susceptibility (HLA-A*0101 and *3002) (128). There is also evidence that several alleles at the class I HLA-B and C loci modulate susceptibility and influence age of onset (129). A risk-modifying locus may lie between HLA-B and marker D6S2702, which is located 970 kb telomeric of HLA-B (130).
Another diabetes-associated locus has been found in the class I region, telomeric to HLA-F (36). By considering the transmission ratios of microsatellite variation from parents homozygous for the HLA class II DR-DQ genes (using the Homozygous Parent Transmission Disequilibrium Test), the possible confounding effect of linkage disequilibrium was removed. Evidence for a second IDDM locus in this region was demonstrated, near the HFE (hemochromatosis) gene and 8.5 Mb distal to the HLA class II loci. Analyses from three independent family data sets from Norway (100 families), Denmark (51 families), and UK (74 families) suggested the presence of additional type 1 diabetes gene (s). Allele 3 of marker D6S2223, 5.5 Mb telomeric of the class II region, was associated with type 1 diabetes when the haplotype was fixed for HLA-DRB1*03, DQA1*0501, DQB1*0201. In a case-control study allele 3 at D6S2223 was found to be reduced among DRB1*03, DQA1*0501, DQB1*0201 homozygous patients with type 1 diabetes compared to DR-DQ matched controls, thus corroborating the results of the family analysis (37). The protective effects seem to be inherited as a recessive trait. An association with D6S2223 has been reported in a Dutch family dataset (131).
The HLA Class III Region
Moghaddam et al. (113) analyzed 11 markers in the HLA region in IDDM patients and controls fully matched for the highest risk DQA1*0501, DQB1*0201/HLA-DQA1*0301, DQB1*0302 (DR3/DR4) genotype. Their study provided strong evidence that another critical region for IDDM susceptibility, approximately 200 kb in size, lies around the microsatellite locus D6S273 which is located between the TNF and HSP70 genes. Another study has independently confirmed linkage with marker D6S273 showing evidence for non-random transmission from DRB1*03-DQA1*0501-DQB1*0201 homozygous parents (36). Further studies in multiplex families from the US indicate that allele D6S273*2 marks an extended DR3-B18 haplotype associated with increased susceptibility. On DR3 haplotypes, other D6S273 alleles were significantly associated with both increased transmission (D6S273*5; P < 0.02) and decreased transmission (D6S273*7; P < 0.05) to affected individuals. The differential transmission was most evident among DR3-B8 haplotypes. Thus, these data indicate that D6S273 marks a susceptibility locus that increases diabetes risk associated with DR3 haplotypes (132). Linkage disequilibrium analysis suggested that "diabetogenic haplotypes" might have resulted from a recombination telomeric of D6S1014 in the region of D6S273 and TNF. The TNF gene is a strong candidate since polymorphisms of this gene may affect the production of TNFa (Tumor Necrosis Factor), a potent cytokine, and in turn the magnitude of immune responses. It has also been reported that TNFa polymorphisms are associated with age-of onset and may influence the inflammatory process leading to the destruction of pancreatic ß-cell and the development of IDDM (114).
In addition, the class I chain-related MIC-A and MIC-B genes, located between the HLA-B and the TNFa genes, may also affect IDDM susceptibility. MIC-A polymorphisms are associated with disease susceptibility in several populations (133-140). In a case-control study of Italian patients, the frequency of the MIC-A5 allele was increased in patients while none of the TNFa alleles were statistically significantly associated with the disease. In this study, the MIC-A5 allele was associated with IDDM independently of class II alleles, suggesting an independent contribution of this locus to diabetes risk (133). MIC-A alleles have a strong effect on development of Addison’s disease and a weaker apparent influence on the development of type 1 diabetes. However, homozygosity for the MIC-A 5.1 allele (with a premature stop codon) was associated with increased diabetes risk and faster progression to diabetes in young children followed from infancy in the DAISY study, especially in those with the HLA-DR3-DQ2/DR4-DQ8 genotype (141). Finally, HSP70-2 and HSP70-Hom genes are also located in the class III region although there is no evidence for an independent association with IDDM from studies that could not circumvent linkage disequilibrium (121;142-146).
Clinical Heterogeneity of Type 1 Diabetes in Relation to the IDDM1 Locus
Age dependent HLA heterogeneity has been observed in Caucasian IDDM patients, indicating that high risk HLA genotypes occur at a higher frequency among the younger age onset groups (70;147-149), whereas older age at diagnosis is associated with an increased heterogeneity of DRB1 and a decreased heterogeneity of DPB1 (101). Caillat-Zucman and coworkers have found a decreased frequency of DR3 and DR4 haplotypes and of DR3/DR4 heterozygosity amongst patients who had developed diabetes after age 15 (123). Similar findings were reported by Tait et al. (107). Earliest development of diabetes is strongly associated with the DR3/DR4-DQ8 genotype and such a genotype is preferentially followed in the population based DAISY (Diabetes Study of the Young) study (150). The immunogenetic analysis of islet cell antibody (ICA) positive first-degree relatives from our family study has confirmed that DR3 (in the absence of DR4/DQB1*0302) is associated with a slower rate of progression to diabetes. This slower rate may be in part explained by a recessive lack of humoral response to insulin during the prediabetic period that was noted in a subset of first degree relatives at increased IDDM risk. This lack of humoral anti-insulin autoimmunity is mostly associated with DR3 homozygosity, but it was observed also in relatives with DR5 or DR8 haplotypes. All these haplotypes carry DQA1 alleles from the evolutionary lineage 4 (125;126) sharing glutamic acid and phenylalanine amino acid residues at position 40 and 51 of the second exon. Thus, lack of or reduced humoral responsiveness to insulin during the prediabetic period may be associated with this particular subset of DQA1 alleles rather than with DR3 itself (151).
Extended (Ancestral) MHC Haplotypes
Early studies by Alper and co-workers and Dawkins and co-workers documented the existence of haplotypes in the MHC region that were very large and relatively common where all polymorphic markers were essentially all conserved (152;153). With high throughput SNP analysis or extensive sequencing the remarkable size and conservation of these haplotypes has been confirmed (154;155). In particular, the HLA-A1, B8, DR3 haplotype is more than 2.7 megabases, within which greater than 99.9% of SNPs or sequences are identical. Another remarkable haplotype is the A30, B18, DR3 “Basque” haplotype (156). Overall the 8.1 haplotype does not show enhanced transmission to diabetics compared to non 8.1 DR3 haplotypes (59) while the HLA-A30, B18, DR3 haplotype does show enhanced transmissions. It is likely that careful analysis of these haplotypes will aid in localization of additional genes contributing to type 1 diabetes susceptibility. We have evidence from analysis in the DAISY study that there is a major gene linked to DR-DQ such that for siblings with DR3-DQ2, DR4-DQ8 sharing both MHC haplotypes with their proband the risk of islet autoimmunity exceeds 80% (23) with type 1 diabetes following the appearance of islet autoantibodies by several years in this high risk young population. In contrast siblings with the same HLA DR and DQ alleles, but sharing only one or no HLA haplotypes (despite being DR3/4-DQ2/DQ8) have a risk of activating anti-islet autoimmunity of “only” 20% (figure 7.8). This strongly implicates non-DR/DQ loci, linked to or within the major histocompatibility complex contributing to diabetes risk. It also suggests that for this DR-DQ genotype, environmental factors essential to activate anti-islet autoimmunity are unlikely to be rare, given the extremely high penetrance of disease.

Figure 7.8. Extreme risk of type 1A diabetes for siblings of patients with type 1 diabetes who share both HLA haplotype identical by descent with their sibling proband (23).

The Insulin Gene Locus, IDDM2
Insulin was the only autoantigen in humans for which expression within the pancreatic islet is specifically restricted to ß-cells. Recently John Hutton and coworkers have discovered that the Zinc Transporter 8 is a beta cell specific autoantigen recognized by anti-islet autoantibodies of the majority of new onset and prediabetic individuals. Nevertheless insulin remains a beta cell specific autoantigen that is expressed at very high levels. Autoantibodies against insulin have been reported in individuals at increased diabetes risk or patients at onset (157-160). Moreover, autoreactive T cells against various proinsulin and insulin peptides have been demonstrated in patients and prediabetic subjects, including CD8 T cells (161-169). Major epitopes include the A chain epitope A1-A15 (166), the B chain/C-peptide junction of proinsulin and the B chain B9-23 epitope. T cells against the B9-B23 epitope represent approximately 40% of the lymphocytic infiltrate observed in the endocrine pancreas of nonobese diabetic (NOD) mice before diabetes onset (170), and both CD4 and CD8 T cells recognizing insulin as their target have been reported in NOD mice (171;172). Knocking out the insulin 2 gene in NOD mice greatly accelerates the development of diabetes, and knocking out the insulin 1 gene decreases the development of diabetes by 90% (173;174).  In addition, there is evidence that insulin B9-23 is a critical trigger for the development of diabetes through the study of insulin gene mutant mice (175;176), and specific T cell receptor transgenic mice targeting this molecule can either cause or prevent NOD diabetes (177;178).
The insulin gene (INS) is therefore an obvious candidate susceptibility locus. Its role in disease susceptibility was easily demonstrated by association studies and was replicated by linkage analysis (39;179). Indeed, the 4.1 Kb region containing INS and its flanking regions contain several polymorphisms in linkage disequilibrium that have been associated with diabetes risk (180). Extensive studies involving polymorphisms in the neighboring HUMTHO1 (tyrosine hydroxylase) and IGF2 genes provided strong evidence that INS is the main susceptibility determinant in this region (157;158;181-184). All of the polymorphisms lie outside coding sequences, confirming that diabetes susceptibility must derive from modulation of INS transcription. Susceptibility in the INS region, or the IDDM2 locus, has been primarily mapped to a variable number of tandem repeats (VNTR) located ~0.5 kb upstream of INS (185-187) (Figure 7.8). The VNTR may not explain all of the susceptibility in this region (172;188) and at least two other polymorphisms (-23HphI and +1140A/C) may contribute to the etiological effect (189) (Figure 7.8)
The VNTR
This polymorphic repeat, also known as the insulin gene minisatellite or ILPR (insulin-linked polymorphic region), consists of a 14-15 bp unit consensus sequence (ACAGGGGTCTGGGG) with slight variations of the repeat sequence. Any number from 30 to several hundred repeats has been observed, but allele frequencies tend to cluster in the 30-60 repeats range (class I alleles) or at 120-170 repeats (class III alleles). The intermediate class II alleles are rare in Caucasians, and less rare in individuals of African descent (190;191). The sequence of the VNTR is particularly G-rich, and it tends to form unusual DNA structures in vitro and in vivo, presumably through the formation of G-quartets (130;192). Shortly after its discovery (185), the insulin VNTR was found to be associated with type 1 diabetes (171). Homozygosity for the short class VNTR I alleles is found in ~75-85% of the patients compared to a frequency of 50-60% in the general population, suggesting that it predisposes to type 1 diabetes. In contrast, homozygosity for the longer class III VNTR alleles is rarely seen in patients and the class III VNTR is believed to confer a dominant protective effect (171;193;194). The relative risk ratio of the I/I genotype vs. I/III or III/III has been reported to be moderate (in the 3-5 range) and it accounts for about 10% of the familial clustering of type 1 diabetes (195). Moreover, by measuring the HphI polymorphism (in tight linkage disequilibrium with the VNTR) (180), Metcalfe et al. (196) showed that homozygosity for the predisposing INS genotype increases the likelihood that identical twins will be concordant for the development of autoimmunity and diabetes in the BabyDiab study, in which offspring of affected parents are followed prospectively from birth (Fig. 7.9) (197). Halminen et al. (198) reported that IDDM2-encoded susceptibility is associated with reduced insulin secretory capacity found in autoantibody-positive first-degree relatives (siblings) from the Childhood Diabetes Study in Finland. This finding may be explained with more aggressive autoimmunity against insulin in subjects with the high-risk genotype and is consistent with the hypothesis that IDDM2 may modulate immune responsiveness to insulin.

Figure 7.9. Insulin gene VNTR (Variable Nucleotide Tandem Repeats) polymorphisms increase risk of developing anti-islet autoimmunity in the BabyDiab study.

VNTR heterogeneity
Although VNTR alleles cluster in two main classes with divergent associations with type 1 diabetes, there is evidence that VNTR alleles are quite heterogeneous and may differ in their ability to modulate disease susceptibility. Further classification of VNTR alleles is indeed possible according to size differences, and at least 21 class I and 15 class III VNTR alleles were described by fluorescence-based DNA fragment sizing technology (199) (Fig. 7.7). Bennett et al. grouped the 15 class III VNTR alleles identified according to two main modes of transmission based on the linkage disequilibrium pattern with alleles at the HUMTHO1 locus on chromosome 11p15. Thus, by taking both size and flanking haplotypes into account class III VNTR alleles linked to the HUMTH01 Z-8 allele were found more protective (very protective haplotype or VPH) than those linked to the HUMTH01 Z allele (protective haplotype or PH) (Fig. 7.7) (199;200). However, certain VNTR alleles can be found in linkage disequilibrium with either the Z or Z-8 alleles. The variable degree of protection observed for these alleles may also be influenced by sequence heterogeneity and its effects on the VNTR physical state and transcriptional activity (130;192;195;199;201). Sequencing studies have indeed identified several variants of the commonest VNTR repeat sequence that characterize yet another level of heterogeneity (185;190;191;195;202).
Studies have also analyzed the variant repeat distribution within the VNTR using minisatellite variant repeat mapping by PCR (MVR-PCR) (203). Some of the variation within the repetitive sequence most probably arises from mitotic replication slippage at an estimated frequency of 10-3 per gamete. However, sperm DNA analysis revealed a second class of mutation occurring at a frequency of approximately 2 x 10-5 that involved highly complex intra- and inter-allelic rearrangements which are probably meiotic in origin (204). These events may help explain the heterogeneity of the VNTR locus. The combined analysis of the variant repeat distribution and of the haplotypes flanking the VNTR has allowed defining five new ancestral allele lineages (183). By this approach, class III VNTR alleles can be divided into two diverging lineages, IIIA and IIIB (Fig. 7.10). These two lineages correspond to the PH and VPH haplotypes previously defined by Bennett et al. (199). Class I alleles can also be divided into three newly defined lineages, IC+, ID+ and ID-. The lineage denomination reflect the class of alleles, noted by the letter “I”, while the letters “C” or “D” identify two different lineages defined by the very different distribution of variant repeats noted by multi-dimensional scaling (183). The notation “+” or “-“ refers to the presence (+) or absence (-) of a MspI restriction site at position +3,850, so that depending on this haplotypic analysis lineages could either be IC+, ID+ or ID- (183). IC+ and ID+ alleles are predisposing to type 1 diabetes.
In contrast, ID- alleles are protective when transmitted from ID-/III heterozygous fathers. Similar findings had been previously reported for the class I allele termed 814 (42 repeats), which is included in the ID- lineage (see next paragraph). The analysis of class ID- alleles into those of 42 repeats and those of other sizes suggested that the protective effect was a feature of all ID- alleles, irrespective of size. However, ID- alleles are clearly distinguished from all other alleles by a MspI variant within the IGF2 gene. This suggests that at least for class I ID- alleles the susceptibility conferred by the VNTR may be modified by nearby sequences, and that in this case IDDM2 susceptibility may have a multi-locus origin (145). All together, the above studies suggest that the VNTR locus is extremely polymorphic, and that not only size of the VNTR but also sequence variation may play a significant role in modulating INS transcription and diabetes susceptibility.

Figure 7.10. The IDDM2 Susceptibility Locus. Top to bottom, the figure shows the HUMTH01, INS and IGF2 loci, as well as a schematic structure of the insulin gene with the approximate location of some of the most characterized polymorphic loci (VNTR, HphI, DraIII, PstI). Also shown are a schematic representation of the two main VNTR classes, their association with diabetes, as well as the VNTR alleles and allele lineages that have been defined with the variety of approaches described in the figure and in the main text.

VNTR Effects On Transcription
Several studies have investigated the effects of the VNTR on INS transcription. Transfection of rodent ß-cell lines with reporter constructs representing the INS promoter flanked by class I or class III alleles resulted in three-fold differences going in opposite directions in reports from different laboratories (186;187). These discrepant results may be due to species-specificity, differences among specific alleles within each class, or the absence of the genomic context necessary for the VNTR to have its physiologic effects. Studies on the transcriptional effects of the VNTR in vivo produced more meaningful results. In fetal pancreas RNA, the INS transcript in cis with the class III VNTR was expressed at lower levels (15-20%) than the class I transcript, a small but statistically significant difference (182). Bennett et al. (199) found a somewhat larger difference in adult pancreas. Moreover, single nucleotide differences in the VNTR sequence can affect INS transcription and correlate with the ability to form unusual DNA structures, both at the inter- and intra-molecular levels (201). These findings led to the hypothesis that VNTR variants may differ in their ability to stimulate transcription as a function of the binding of inter- and intra-molecular quartets with the transcription factor Pur1. However, the transcriptional activity of these variants observed in vitro may not always correspond to that in vivo, where overall transcription may depend on the interaction with other proteins involved in the transcriptional machinery and on differences among the various cell types that actively transcribe the insulin gene. The studies described above report only marginal differences in pancreatic INS transcription, and the lower transcription associated with diabetes-protective class III VNTR alleles does not fit well with their dominant protective effect. It seems unlikely that such minor differences in pancreatic INS transcription may influence susceptibility to a form of diabetes resulting from the autoimmune destruction of pancreatic ß-cells. It was later discovered that INS is actively transcribed in the thymus in mouse (188), rat (205), and humans (83;84). The human thymus was found to express low levels of INS message throughout fetal development and childhood but also during adulthood (193). Overall, genes encoding for several self-molecules have been found to be expressed in the thymus, including pancreatic and thyroid hormones, neuroendocrine molecules and other peripheral proteins (206). Functional studies in transgenic mice and fetal organ thymic cultures have provided both in vivo and in vitro data showing that thymic expression of self-antigens and their levels of expression can dramatically affect the development of self-tolerance (reviewed in (207)).  The fact that negative selection of autoreactive thymocytes is dose-dependent suggested the hypothesis that different VNTR alleles may modulate tolerance to insulin by affecting insulin expression levels in the thymus. Consistent with this hypothesis, INS mRNA levels in the thymus were found to correlate with VNTR alleles in opposite fashion to that observed in the pancreas (199). INS transcripts in cis with class III VNTR alleles are transcribed at much higher levels (on average 2-3 fold) than those in cis with class I VNTR alleles (83). The increased transcription levels detected in thymus fit well with the dominant protective effect associated with class III VNTR alleles, as higher insulin levels in the thymus may more efficiently induce negative selection of insulin-specific T-lymphocytes (or improved selection of regulatory T cells). In contrast, homozygosity for diabetes-associated class I VNTR alleles determines lower insulin levels that may be associated with a less efficient deletion of insulin-specific autoreactive T-cells (or impaired selection of regulatory T cells). Proinsulin appears to be the main product of the insulin gene in the thymus (84;193). This is not surprising since thymus cells expressing proinsulin are not likely to possess the refined machinery necessary to process proinsulin to mature insulin. Proinsulin expression may be sufficient to obtain tolerance to insulin since most of the known immunodominant epitopes identified as targets of the insulin autoimmune responses in type 1 diabetes are shared by both insulin and proinsulin. Both thymic epithelial cells and bone marrow derived dendritic cells have been shown to transcribe INS and other genes coding for self-molecules (193;207-211). Similar cells and INS transcription have also been demonstrated in peripheral lymphoid organs, suggesting that insulin expression in lymphoid organs may also play a role in maintaining peripheral self-tolerance throughout life (193;212). Direct support for the hypothesis that levels of INS expression in thymus and lymphoid organs can influence type 1 diabetes susceptibility is provided by studies in insulin gene knockout mice and transgenic mice (173;175;213-215), which were recently reviewed (207).
Other Effects at the IDDM2 Locus
It is also possible that other loci in the 11p15 region may also contribute to IDDM2-encoded susceptibility, as suggested for certain class I VNTR allele lineages (ID-) (183). The four promoters of the IGF2 gene are situated only 5-20 kb downstream of the INS VNTR, a distance that would allow enhancer effects. IGF-II is a growth factor with ubiquitous expression and has been implicated in biological functions that could be relevant to autoimmune diabetes. These include inhibition of apoptosis (216) and the stimulation of ß-cells proliferation (217), functions that are of potential importance in resistance to immune injury and regeneration, respectively. IGF-II is also produced by T-lymphocytes (218) and can influence an activation-induced autocrine loop that would amplify clonal expansion of autoreactive T cells. However, the demonstration that the INS-VNTR does not influence IGF2 transcription in the human thymus, pancreas, and leukocytes argues against a role for IGF2 as a major contributor to IDDM2-encoded susceptibility (219). It has also been suggested that IGF-II expression in the thymus may be important for the development of self-tolerance to IGF-II and other proteins of the IGF/insulin family, including insulin (220). There also appears to be a thymus defect in IGF-II expression in the thymus of diabetes-prone BB rats, and this has led to the hypothesis that defective IGF-II expression in the thymus of BB rats may impair tolerance to insulin and favor diabetes development in BB rats (221-223). However, it is controversial whether insulin is an autoantigen in the autoimmune diabetes of BB rats (224). An alternative hypothesis is that the thymic IGF-II deficiency reported in this model may perhaps determine some more generic defects of the immune system and not necessarily affect tolerance to insulin.
Overall, the studies reviewed here suggest that the IDDM2 is a quantitative trait resulting from allelic variation and, as discussed in a later paragraph, from complex parental and epigenetic effects at the VNTR locus. IDDM2-associated susceptibility and resistance may derive from quantitative differences in INS transcription in the specialized antigen presenting cells found in thymus and peripheral lymphoid tissues, where production of self-antigens such as proinsulin may be crucial for the shaping and maintaining of a self-tolerant T cell repertoire (193;225;226). Such mechanisms may influence the probability of developing autoimmune responses to insulin, a key autoantigen in type 1 diabetes.
PTPN22 (Lyp)
This gene was also identified through the candidate gene approach. Bottini and coworkers evaluated a functional polymorphism in the lyp gene (no relation to the lymphopenia gene of the BB rat) in two series of patients with type 1 diabetes, one from Denver and one from Sardinia (28). The odds ratio was approximately 1.7 (Figure 7.11), making this polymorphism the most potent after IDDM1 and IDDM2.The Lyp molecule, coded for by  PTPN22, is a lymphoid tyrosine phosphatase located on chromosome 1p13. The relevant diabetes associated polymorphism appears to be a missense mutation that changes an arginine at position 620 to a tryptophan and thereby abrogates the ability of the molecule to bind to the signaling molecule Csk (28;29;227). The lyp-Csk complex downregulates T cell receptor signaling and thus loss of this interaction was thought to enhance T cell receptor signaling, though a recent study by Bottini and colleagues indicates a gain of function with the missense mutation and inhibition of T-cell receptor signaling. Consistent with a general effect on immune function is the finding that the minor tryptophan encoded allele is associated with a series of autoimmune disorders including type 1 diabetes, rheumatoid arthritis (29) and lupus erythematosus (30). Multiple studies have confirmed the association of this missense mutation with type 1 diabetes including a large study from Great Britain (Figure 7.1). It is of interest that PTPN22 has an effect similar in magnitude to the insulin gene polymorphism and similar to IDDM2 it was not identified through linkage studies, but rather through the candidate gene approach. It is likely that further polymorphisms in the pathways controlling T cell receptor signaling will be important, but it is also likely that many genes with this magnitude of effect will not be found at a population level, given the likely multiplicative models for disease susceptibility. Evidence for linkage was not even obtained in the largest genome scan to date, performed by the T1DGC in 2005 (39). It has been estimated that given the magnitude of the association of PTPN22 with diabetes it would require >8,000 sibling pairs to detect linkage, while this study had less than 1,500 families. It is possible that polymorphisms in linkage disequilibrium determine increased risk of autoimmunity rather than the R620W polymorphism, but this seems unlikely given the rapid confirmation of this polymorphism’s association with multiple forms of autoimmune disease in multiple populations and its dramatic functional significance. A gain of function with a missense polymorphism probably also explains why the R620W change is the only PTPN22 polymorphism clearly associated with type 1 diabetes despite extensive analysis.

Figure 7.11. PTPN22 (Lyp) genotypes. The minor T allele is associated with type 1 diabetes.

CTLA-4 (IDDM12)
Linkage with markers on chromosome 2q33 was initially reported in a group of Italian families (228). This chromosomal region contains the CTLA-4 (cytotoxic T lymphocyte associated-4) and CD28 genes, which encode for two molecules that are intimately involved in the regulation of T-cell activation and proliferation. Differential regulation of these molecules could easily affect T-cell function and hence the regulation of immune responses. The CTLA-4 gene is a strong candidate gene for autoimmune diseases since it encodes for a molecule that functions as a key negative regulator of T-cell activation, and the linked markers encompass a region containing an (AT)n microsatellite located in the 3' UTR of the CTLA-4 gene. Moreover, the analysis of an A-G transition in the first exon of the CTLA-4 gene, coding for a Thr/Ala substitution in the leader peptide, also showed preferential transmission to affected siblings (229). Although linkage was not observed in families from Sardinia, U.K., and U.S.A., preferential transmission was observed considering all of the above families together (n= 818).
Further confirmation of association with the IDDM12-CTLA-4 locus came through linkage disequilibrium (association) analysis using a multi-ethnic collection of families with one or more affected children, which included families from Spain, France, China, Korea, and Mexican-Americans. In this study, the TDT revealed a highly significant deviation for transmission of alleles at the (AT)n microsatellite marker in the 3' untranslated region as well as the A/G polymorphism in the first exon of the CTLA-4 gene. The overall evidence for transmission deviation of the CTLA-4 A/G alleles remained highly significant even combining data sets (669 multiplex and 357 simplex families) from this study and the above families from Italy, U.K., U.S.A., Spain and Sardinia. However, significant heterogeneity was observed: U.K., Sardinian, and Chinese families did not show any deviation for the A/G polymorphism, while the U.S.A. families showed a weak transmission deviation. In contrast, strong deviation for transmission was seen in the three Mediterranean-European populations (Italian, Spanish and French), in Mexican-Americans, and in Koreans (230). Linkage has also been reported in Japanese patients with IDDM and patients with autoimmune thyroid disease (231;232), although not in all studies of Japanese patients (233). Further evidence for linkage in the CTLA-4/CD28 region was obtained in a study of 960 families from Italy, Sardinia, U.K., and U.S.A. (234), but not in the Danish population (235). The CTLA4 predisposing variant was increased diabetes risk in synergy with HLA-DR3 but not with HLA-DR4 in a German population (236).
A multiethnic (U.S. Caucasian, Mexican-American, French, Spanish, Korean, and Chinese) collection of 178 simplex and 350 multiplex families was typed for 10 polymorphic markers within a genomic interval of approximately 300 kb containing the candidate genes CTLA4 and CD28. The transmission disequilibrium test revealed significant association/linkage with three markers within CTLA4 and two immediate flanking markers (D2S72 and D2S105) on each side of CTLA4 but not with more distant markers including the candidate gene CD28. Because these markers are contained within a phagemid artificial chromosome clone of 100 kb, the IDDM12 locus is likely to correspond to either CTLA4 itself or to or an unknown gene in very close proximity (237). Moreover, the Idd5 susceptibility locus in NOD mice overlaps with CTLA-4 (238).
In a very large combined analysis of more than 3,600 families, Ueda and coworkers reported that a CTLA-4 polymorphism was transmitted to 53.3% (versus the expected transmission of 50%) to affected individuals, with a relative risk 1.14 (figure 7.10) (239). Susceptibility was mapped to a polymorphism in the non-coding 6.1 kb 3’ end associated with lower messenger RNA levels of a soluble form of CTLA-4, which results from alternative splicing. The 49 exon 1 G/G variant is associated with decreased expression of a soluble variant of CTLA-4 that may have an influence on immune function, especially in light of CTLA-4 polymorphism associated with diabetes of the NOD mouse (240), Graves’ disease (241), and Addison’s disease (242). Evidence for linkage was also obtained in 2005 scan of the T1DGC, albeit the region is likely to contain also IDDM7 (39). Because the effect of the reported CTLA-4 polymorphism in human diabetes is so small, lack of confirmation in smaller studies or discordant results among studies is to be expected (241). However, the biological contribution of this polymorphism remains to be assessed by functional studies. Overall, CTLA-4 appears to be a stronger determinant for Graves’ disease than for type 1 diabetes (figure 7.12). Concordant with its relatively weak effect in the Wellcome Trust Case Control Consortium genome-wide association study of 2,000 patients and 3,000 controls, no SNPs at 2q33 were significant at either the 5X10 (-7) cutoff (243). Of note, CTLA4 G allele of rs3087243 was more strongly associated with patients with type 1 diabetes plus thyroid peroxidase autoantibodies (OR=1.49) compared to those without the thyroid autoantibodies (OR 1.16) (244)

Figure 7.12. Summary of CTLA-4 association with type 1 diabetes (man and mouse) and Grave’s disease.

IL2RA (CD25)
In the Wellcome Trust Case Control Consortium Study SNP rs2104286 was analyzed with rs706778 referenced as the prior reported SNP (HApMAp r2=.25) with finding of an association trend value of 8.0X10 (-6), and p value of approximately 10 (-8) with expanded control series. The odds ratio was 1.30 for heterozygotes and 1.57 for homozygotes (245). There is evidence of two different SNPs associated with type 1A diabetes for the IL2RA region (ss52580109 (P=7.8X10 (-11)) and rs11597367 (P=8.19X10 (-7)). The odds ratio for the minor A allele of rs11597367 was 0.78 (1.0/0.78=1.28). The odds ratio for the ss52580109 minor A allele was 0.68 (1/0.68=1.47) (246). The associated SNPs are located in region of intron 1 of IL2RA and 5’ intergenic sequence between IL2Ra and RBM17 (RNA binding motif protein 17). There is an interesting observation (Figure 7.13) that soluble IL2 receptor correlates with the genotype of the SNPs, though with very extensive overlap, leading to the hypothesis that influence on diabetes might relate to lower immune responsiveness contributing to type 1 diabetes (246). The association of the two different SNPs appear to have different relations to soluble IL2RA by genotype.

Figure 7.13. Soluble IL2 receptor levels (serum) relative to IL2RA SNP genotypes. Lowe et. al., Nature genteics 39:1074, 2007.

IFIH1
The minor allele of rs1990760 of the Interferon Induced Helicase region (IFIH1) was reported to be associated with type 1 diabetes with a risk ratio of 0.86 in a large study of 4,253 cases, 5,842 controls, and with an additional 2,134 parent-child trio analysis. The Wellcome Trust analysis utilizing a different SNP (rs3788964) found a genotypic P-value of 7.6X10 (-3) and Trend p-value of 1.9X10 (-3), suggesting a very modest association in this study (247). The gene is of particular interest in that it may relate the innate immune system to the development of a disease presumably mediated by the adaptive immune system, and animal models are available where activation of innate immunity, and interferon alpha, is associated with induction of autoimmune diabetes (248;249).
KIAA0350 (16p13)
The association of KIAA0350 with type 1 diabetes was discovered with the Wellcome Trust Case Control Consortium study with a Trend P Value of 9.2X10 (-8) and a heterozygous odds ratio of 1.19 and homozygous of 1.55 (SNP rs12708716) (250). The association was confirmed with analysis of 4,000 patients and 5,000 controls (10 (-8)) and in a family analysis (trios, 10 (-6)). There are only two genes in the region, the lectin KIAA0350 and dexamethasone-induced transcript. KIAA0350 is a putative C-type lectin, and also has an immunoreceptor tyrosine-based activation motif (ITAM) (22).
PTPN2 (18p11)
In the Wellcome Trust the region associated with PTPN2 (protein tyrosine phosphatase, non-receptor type 2) was associated with all of the autoimmune disorders studied, namely Crohn’s disease, rheumatoid arthritis and type 1 diabetes (for type 1 diabetes P=1.9X10 (-6) with a heterozygote odds ratio of 1.30 and homozygote odds ratio of 1.62 (251). Follow up study gave a P value of 3.36X10 (-10) with an odds ratio of 1.29 (252). The molecule is a member of the same family as PTPN22, an allele of which is strongly associated with type 1 diabetes (R620W) (28;253).
The Wellcome Trust Case Control Consortium (WTCCC) primary genome-wide association (GWA) scan (254) on seven diseases, including the multifactorial autoimmune disease type 1 diabetes (T1D), has recently shown associations at P < 5 λ10-7 between T1D and six chromosome regions: 12q24, 12q13, 16p13, 18p11, 12p13 and 4q27. Four of those regions have been replicated by another big study including 4,000 individuals with T1D, 5,000 controls and 2,997 family trios (22;23): there was strong evidence for disease association for chromosomes 12q24, 12q13, 16p13 and 18p11 in independent cases and controls (P λ1.82 λ10-6), in families (P = 5.23 λ10-3 to 1.07 λ10-6) and overall (P = 1.15 λ10-14 to 1.52 λ10-20).
12q13
In the WTCCC, SNP rs11171739 showed strong evidence of association with T1D with heterozygote odds ratio (OR) of 1.34 and a homozygote OR of 1.75 and a genotypic P value of 9.71 x 10-11. This SNP rs11171739 is close to the ERRB3 gene (v-erb-b2 erythroblastic leukemia viral oncogene homolog 3) and another SNP rs2292239, in the ERRB3 gene, has also shown association with T1D in the study by Todd et al (22;23). The ERRB3 gene encodes a member of the epidermal growth factor receptor (EGFR) family of receptor tyrosine kinases. Amplification of this gene and/or overexpression of its protein have been reported in numerous cancers, including prostate, bladder (255) and breast tumors (256). In diabetic rats, expression of both ERRB2 and ERRB3 is enhanced during oral oncogenesis, possibly resulting in promotion of cell proliferation and inhibition of apoptosis (257).
The VDR gene is located on chromosome 12q12-q14. Four common single nucleotide polymorphisms (SNPs) in the VDR gene have been studied: FokI T>C (rs10735810), BsmI A>G (rs1544410), ApaI G>T (rs7975232), and TaqI C>T (rs731236). FokI polymorphism in exon 2 results in an alternative transcription initiation site, leading to a protein variant with 3 additional amino acids (258). SNPs BsmI and ApaI are located in intron 8, and TaqI is a silent SNP in exon 9. Several studies reported association of type 1 diabetes with one of these four SNPs. Pani et al. genotyped 152 Caucasian families for these four polymorphisms and suggested an association with T1D susceptibility in Germans (259). Guja et al studied 204 Romanian diabetic families and found that VDR FoqI F allele seemed to be predisposing while TaqI T allele seemed to be protective (260). However, these associations have not been confirmed in more recent and bigger studies: one study in the Finish population (1000 cases and 2000 controls) (261), another report by Todd et al. with up to 3,763 T1D families from the UK, Finland, Norway, Romania and US and 3414 case-control subjects from the UK (262), and finally a recently conducted meta-analysis also found no evidence of association (263).
Recently, several studies have reported associations of type 1 diabetes and other autoimmune diseases with polymorphisms in the CYP27B1 gene on chromosome 12q13.1-q13.3, which encodes 1[alpha]-hydroxylase, the enzyme that converts 25-hydroxyvitamin D (25OHD3) into 1,25-dihydroxyvitamin D (1,25diOHD3). Lopez et al. (264) report a significant association between allelic variation of the promoter (-1260) C/A polymorphism and Addison's disease, Hashimoto's thyroiditis, Graves' disease and type 1 diabetes mellitus (P=0.0062, P=0.0173, P=0.0094 and P=0.0028 respectively). Todd et al. (265) studied 7,854 patients with type 1 diabetes, 8,758 control subjects from the U.K., and 2,774 affected families and found evidence that the promoter polymorphism CYP27B1 -1260 is associated with type 1 diabetes in both the case-control (P = 9.6 x 10-4; CC genotype OR 1.22 [95% CI 1.10-1.36]) and the family (P = 3.9 x 10-3; CC genotype RR 1.33 [95% CI 1.12-1.58]) collections.
Lower serum concentrations of 1,25-dihydroxyvitamin D (1,25diOHD3), the hormonally active form of vitamin D, and of its precursor 25-hydroxyvitamin D (25OHD3) have been reported at the diagnosis of type 1 diabetes compared with normal control subjects (266;267). Epidemiological studies have suggested that vitamin D supplementation in early childhood is associated with a decreased risk of developing type 1 diabetes (268;269). In the immune system, vitamin D and its analogs have been shown to function by stimulating Th-2 T-helper cells to produce transforming growth factor-λ1 and IL-4 that might serve to suppress the TNF-λ and interferon-λ production by Th-1 cells (270). In the animal models, 1,25diOHD31 and its analogs have been effective in prevention diabetes in NOD mice (271;272).
12q24
The SNP (from the WTCCC study) rs17696736 in C12orf30 (Chr. 12 open reading frame 30) maps to regions of extensive linkage disequilibrium covering more than ten genes. Several of these represent functional candidates genes because of their presumed roles in immune signaling, considered to be a major feature of T1D-susceptibility. These include SH2B3/LNK (SH2B adaptor protein 3), TRAFD1 (TRAF-type zinc finger domain containing 1) and PTPN11 (protein tyrosine phosphatase, non-receptor type 11). In the study by Todd et al. (22;23), rs3184504, a nsSNP in exon 3 of SH2B3 encoding a pleckstrin homology domain (R262W), had the highest association (P = 1.73 λ10-21; OR = 1.33, 95% CI = 1.26–1.42). SH2B3 is an adaptor protein that regulates growth factor and cytokine receptor-mediated pathways implicated in lymphoid, myeloid and platelet homeostasis. It has been shown to negatively regulates TNF-alpha expression in endothelial cells (273). TRAFD1, also known as FLN29, is a novel interferon- and LPS-inducible gene that acts as a negative regulator of toll-like receptor signaling (274). PTPN11 is probably the most attractive candidate gene in the region given a major role in insulin and immune signaling (275). It is also a member of the same family of regulatory phosphatases as PTPN22, already established as an important susceptibility gene for T1D and other autoimmune diseases.
12p13
In the multilocus analysis of the WTCCC, there was increased support for a region on chromosome 12p13 containing several candidate genes, including CD69 (CD69 antigen (p60, early T-cell activation antigen)) and multiple CLEC (C-type lectin domain family) genes. The SNP rs3764021 is located in the CLEC2D (C-type lectin domain family, member D) gene, also known as LLT1 (lectin-like transcript). The LLT1 receptor induces IFN-gamma production by human NK cells (276). CD69 is involved in lymphocyte proliferation and functions as a signal-transmitting receptor in lymphocytes, NK cells and platelets. CD69 appears to be the earliest inducible cell surface glycoprotein acquired during lymphoid activation. Locus 12p13 has not been replicated so far.
22q13
SNP rs229541 close to the IL2RB gene shows evidence of T1D association in the WTCCC study (P = 2.18 λ 10-6), but has not been replicated so far. The IL-2 receptor, which is involved in T cell-mediated immune responses, is a trimeric molecule of alpha (IL2RA), beta (IL2RB, also known as CD122) and gamma (IL2RG) chains. IL2RB is an interesting candidate as IL2Ra on chr. 10p15 is already a known susceptibility gene for T1D. IL-2 plays a major role in the proliferation of cell populations during an immune reaction (277).
18q22
In the study by Todd et al. (22;23), another locus showed association with T1D: rs763361 in the T lymphocyte costimulation gene CD226 on chromosome 18q22 (Poverall = 1.38 λ10-8). CD226 is a glycoprotein expressed on the surface of NK cells, platelets, monocytes and differentiated Th1 cells. Anti-CD226 treatment has been shown to delay the onset and reduce the severity of experimental autoimmune encephalomyelitis, a Th1-mediated autoimmune disease (278).
IDDM3
The initial evidence for linkage with marker D15S107 on chromosome 15q26 was initially reported in 250 Caucasian families from the U.K., USA, and Canada (279). Families lacking the typical HLA predisposition provided most of the evidence for linkage, with sibling pair disease concordance or discordance being strongly affected by allele sharing at the D15S107 locus (280). Linkage was confirmed in additional studies of 104 U.S.A. Caucasian families (281) and 81 Danish families (282). In the Danish data set, the D15S107*130 allele provided increased susceptibility with a relative risk of 3.55 and the D15S107 locus was found to contribute up to 16% of the familial clustering of type 1 diabetes (282). However, linkage was not confirmed in another data set of 265 Caucasian families (283). Studies of Han chinese nationality found allele A5 at D15S657 increased in frequency in patients (284). At present there is no known candidate gene in the 15q26 region, and evidence for IDDM3 has not been replicated in the 2005 T1DGC scan (39).
IDDM4
Several studies reported evidence for the existence of the IDDM4 susceptibility locus. This locus is tightly linked to the FGF3 marker on chromosome 11q13 (281;283;285-288). Evidence was found for a decreased transmission (46.4%) to the affected offspring of a 15 Kb stretch of DNA containing two tightly linked alleles (D11S1917*03 and H0570polyA*02) (289). In contrast, the D11S1917*03-H0570polyA*02 haplotype showed increased transmission (56.6%) to unaffected siblings. These results suggest that IDDM4 susceptibility may derive from a gene very close to the D11S1917 marker. Moreover, similar to that discussed for IDDM1 and IDDM2, these findings show that analysis of both predisposing and non-predisposing alleles may be of value when mapping genes for common polygenic diseases (289). A subsequent study provided further evidence for linkage with a peak LOD score of 3.4 at the D11S913 marker (290). Moreover, the extended transmission disequilibrium test (ETDT) revealed significant association/linkage with the marker D11S987 (P= 0.0004) within an interval of approximately 6 cM between D11S4205 and GALN. Several candidate genes can be found in this chromosomal region. MDU1, encoding a cell-surface cell protein regulating intracellular calcium, and ZFM1, a nuclear protein, are both expressed in the pancreas. The RT6 gene lies also in this region, coding for a T-cell protein that is deficiently expressed in the BB rat animal model of diabetes (286). The interleukin-converting enzyme (ICE) and CD3 genes were also proposed as candidates to explain IDDM4 susceptibility. However, the CD3 gene has been excluded by both association and linkage analysis (291). A previous report of an association of the CD3 gene with T1D could have been due to population stratification (292). The gene coding for the low-density lipoprotein receptor related protein 5 (LRP5) has been mapped within the boundaries of the IDDM4 locus and proposed as yet another candidate. However, its functional role in the pathogenesis of T1D remains unclear (293). A large study of markers in the LRP5 region involving 1,106 T1D families provided no further evidence for disease association at LRP5 or at D11S987. In the same study, the analysis of 1,569 families from Finland failed to replicate linkage at LRP5 (239). Three additional genes have been identified in the LRP5 region: the CGI-85 gene and two novel genes, C11orf24 and C11orf23. The C11orf24 gene has no known similarity to other genes, and its function is unknown. C11orf23 has similarity which genes involved in regulation of the cell cycle (294). Finally, the gene coding for the Fas-associated death domain protein FADD/MORT1 has been mapped to chromosome 11q13.3 (295). Both its chromosomal localization and function in apoptosis, a mechanism of cell death implicated in the autoimmune destruction of ß-cells (296;297;297), make it another plausible candidate for a susceptibility gene at the IDDM4 locus. However, polymorphisms in the FADD and GALN genes were not found to be associated or in linkage with diabetes (290). Supporting evidence for IDDM4 has not been replicated in the 2005 T1DGC scan (39).
IDDM5
Several but not all studies have shown linkage with the a046xa9 and ESR markers on chromosome 6q25 (156;283;285;288;298;299). The Mn-superoxide dismutase (MnSOD) is a candidate gene for susceptibility at the IDDM5 locus. Polymorphisms affecting the function of MnSOD could render ß-cells more susceptible to free oxygen radical damage. This region may contain a susceptibility gene that is common to several autoimmune diseases (300). Bohren et al. (301) identified a novel gene, SUMO-4, coding for a Small Ubiquitin-like Modifier 4 protein. The authors also identified a single nucleotide polymorphism involving a highly conserved methionine with a valine residue (M55V). The SUMO-4 variant carrying the methionine was associated with diabetes susceptibility in families from the US and UK (301). Guo and coworkers (40) almost simultaneously reported an association of SUMO4 with type 1 diabetes, and provided evidence that SUMO4 conjugates to I kappa B alpha and negatively regulates NF kappa B transcriptional activity. The M55V substitution results in 5.5 times greater NF kappa B transcriptional activity and approximately 2 times greater expression of IL12B, an NF kappa B-dependent gene. Such functional effects could be relevant to immune function. However, the variant associated with diabetes susceptibility was the one carrying the valine residue, in obvious contrast with the findings of Bohren et al. (301). Further studies have challenged the validity of these associations and whether these differences could be explained by genetic heterogeneity (302-305). Since then evidence for an association of SUMO4 with type 1 diabetes has been reported in a study of Asian families but has not been confirmed in subsequent studies (306), including the 2005 T1DGC scan (39).
IDDM6
Linkage with the 18q12-q21 region, and in particular with the Kidd Blood group locus (JK), was suggested almost 20 years ago (307) and linkage to the JK-D18S64 marker was initially confirmed by the very first genome-wide scan performed in 1994 (285). The Transmission Disequilibrium Test (TDT) provided evidence for increased transmission of allele 4 of marker D18S487 to affected children in a total of 1,067 families from four different countries. Analysis using the TDT also provided evidence for genetic heterogeneity, which can often play as a confounding factor when mapping susceptibility genes in complex diseases (308). Additional evidence for the existence of the IDDM6 susceptibility locus near D18S487 was provided by another large study of 1,708 families from seven different countries (309). There is evidence that this region may predispose to several autoimmune diseases (300). Later studies in the Finnish population and the 2005 T1DGC genome wide scan did not confirm evidence for linkage at this locus (39). A candidate gene has been reported in this region, ZNF236, a gene coding for a Kruppel-like zinc-finger protein. ZNF236 is ubiquitously expressed in all human tissues tested. Its expression levels are highest in skeletal muscle and brain, intermediate in heart, pancreas, and placenta, and lowest in kidney, liver, and lung. Two alternative spliced forms of the ZNF236 transcript have been found to be up-regulated in human mesangial cells in response to elevated levels of glucose, suggesting that ZNF236 may be a candidate gene for diabetic nephropathy (310).
IDDM7
Linkage on chromosome 2q near the marker D2S326 was initially reported in U.K. families (285) and was later observed for the D2S152 marker (2q31-q33) in 348 affected sibling pairs and 107 simplex families from three different populations (311). Further analysis and expansion of the above data sets did not reproduce evidence for a susceptibility gene in this region (234), similar to studies of families from the U.S.A. and China (283;312). However, linkage has been replicated in 241 Danish families (313). Analysis of a combined 831 affected sib pairs by Cox and coworkers gave suggestive evidence for 2q31 (IDDM7) (179), and this has been confirmed in the 2005 T1DGC genome scan (39), in a region that also comprises IDDM12. IDDM7 lies within two centiMorgans of D2S152, a chromosomal region that is synthenic with the nonobese diabetic (NOD) mouse chromosome 1 region containing the Idd5 susceptibility gene (288;311). The HOXD8 gene has been proposed as a possible susceptibility gene at the IDDM7 locus (314). Another potential candidate gene in this chromosomal region is NEUROD, another transcription factor regulating the expression of the insulin gene and playing an important role in the development of pancreatic ß-cells. The NEUROD gene has been mapped to the long arm of human chromosome 2 (2q32). A polymorphism consisting of a nucleotide G-to-A transition results in the substitution of alanine to threonine at codon 45 (Ala45Thr). The analysis of this polymorphism in Japanese and Danish patients suggested an association with type 1 but not type 2 diabetes (315;316), but a case-control study in France did not find a similar association (317). The frequency of the Ala45 allele was 70.3% in Polish patients and 62.9% in controls (p= 0.04) but a TDT analysis with 209 trio families did not show significant distortion of transmission (318). Another candidate gene in this region is GALNT3, which encodes a polypeptide N-acetyl-galactosaminyltransferase-T3 (GalNAc-T3) and was  mapped to a region 5-25 cM from D2S152 (319). GalNAc transferases may influence autoimmunity by glycosylating autoantigens. However, both a marker corresponding to GALNT3 (D2S2363) and the T284A polymorphism in the GALNT3 3'UTR (untranslated region) were not found to be linked with diabetes in Danish families (313).
IDDM8
Several groups reported evidence for linkage with markers D6S264, D6S446 (281;285), and D6S281 (283;298;299) on chromosome 6q25-q27 . At present there is no known candidate gene in the 6q25-q27 region. Owerbach has defined a linkage disequilibrium map of nearly 1 Mb in the 6q27 region and identified multiple haplotypes associated with IDDM8, suggesting localization of this putative susceptibility locus to the terminal 200 kb of chromosome 6 (320). The IDDM8 locus may also be subject to parental effects (321) and may confer susceptibility to rheumatoid arthritis as well (322). Analysis of 831 affected sib pairs in the study of Cox and coworkers implicated IDDM8 only after stratification by HLA genotype (179). Owerbach et al. examined five potential candidate genes in the IDDM8 region using 36 genetic markers in 478 families and detected evidence for an association of a CAG/CAA polymorphism in exon 3 of the TATA box-binding protein gene (323). There is also evidence that the IDDM8 region contains polymorphisms in the insulin-growth factor II receptor gene that are associated with increased susceptibility when maternally transmitted (324).
IDDM9
Initial evidence suggested a susceptibility locus on chromosome 3q21-q25 in linkage with marker D3S1303 (285). IDDM9 appears to be distinct from a susceptibility locus for Rheumatoid Arthritis reported on chromosome 3q (325). Laine et al. (326) analyzed 22 microsatellite markers in 121 Finnish type 1 diabetes multiplex families in the IDDM9 region and detected LOD scores of 3.4 and 2.5 with markers D3S1589 and D3S3606, respectively. Two additional markers showed association using the TDT in 384 Finnish type 1 diabetes simplex families. Marker AFM203wd10 showed association with type 1 diabetes. Interestingly, there was evidence of interaction with IDDM2. There was no strong evidence of linkage in the 2005 T1DGC genome scan (39).
IDDM10
Another susceptibility locus may exist on chromosome 10p11-q11 (marker D10S193), and has been termed IDDM10 (285). Additional support for the existence of IDDM10 was provided by the TDT analysis of 1, 159 families with at least one affected child from the U.K., the U.S.A., Norway, Sardinia, and Italy (327). A study in Russian patients gave a multipoint LOD score (MLS) of 2.2 between markers D110S1733 and D10S1780, while Todd and coworkers analyzed 418 United Kingdom sib pairs and did not confirm linkage (328;329). Evidence for linkage was confirmed by the 2005 scan of the T1DGC with D10S1426 (39). The IDDM10 locus may be subject to parental effects (321) and may play a stronger role in younger patients (330). A possible candidate gene may be Stromal-cell derived factor-1 (SDF-1) (331). Nejentsev and colleagues sequenced candidate genes in the region, CREM and SDF1, and then analyzed the region identifying 12,058 SNPs, and genotyped 1,612 patients compared to 1,828 controls, and only D10S193 microsatellite near the PAD1 gene gave nominal evidence of association (p=.03) (332).
IDDM11
IDDM11 appears to lie on chromosome 14q24.3-q31 and was linked to the microsatellite D14S67 using both maximum likelihood methods and affected sib pair methods. This represents the strongest evidence for linkage to any locus outside the HLA region. Similar to IDDM3, the strongest linkage (with the D14S67 marker) was obtained in a subset of families lacking increased HLA sharing among the affected offspring, suggesting that IDDM11 may be an important susceptibility locus in families lacking strong HLA region predisposition (333). Supporting evidence for IDDM11 has not been replicated in the 2005 T1DGC scan (39). Two candidate genes have been mapped to this chromosomal region. The ENSA gene encodes alpha-endosulfine, an endogenous regulator of ß-cell K (ATP) channels (334). The recombinant alpha-endosulfine has been shown to inhibit sulfanylurea binding to ß-cell membranes, to reduce cloned K (ATP) channel currents, and to stimulate insulin secretion from ß-cells. The SEL-1L gene encodes for a negative regulator of the NOTCH, LIN-12, and GLP-1 receptors, which are required for differentiation and maturation of cells as well as cell-to-cell interactions during development (294). SEL-1L is abundantly expressed only in the pancreas, and appears to be involved in the down-regulation of mammalian Notch signaling, shown to be critical for the development of the pancreas and ß-cells (295). However, a study of families from Denmark and Sardinia found no evidence that SEL-1L is directly linked to diabetes (335).
IDDM13
The IDDM13 susceptibility locus lies in the 2q34 region and is linked to the D2S164 marker in Caucasian families from Australia and the U.K. (BDA Repository) (336). Similarly to IDDM3 and IDDM11, IDDM13 may be of particular interest since it was detected in non-HLA identical siblings, suggesting that yet another locus may be an important susceptibility factor in subjects lacking the typical HLA predisposition. It has also been suggested that IDDM13 may be active early during the evolution of diabetes since linkage was found also in prediabetic subjects. Moreover, it appears that IDDM13 may favor diabetes development predominantly in males (337). Several biological explanations are possible for these findings, including X and Y linkage, effects of sex hormones on gene expression, and quasi-linkage between sex chromosomes and autosomes. Another study in Japanese families has confirmed linkage to the D2S137 microsatellite in siblings lacking HLA predisposition (338). In contrast, little evidence for IDDM13 was found in a data set including 352 U.K. families and 94 U.S.A. families (234). Some evidence for linkage and association of the IDDM13/D2S137-D2S1471 region (approximately 3.5 cM) was found in Danish families (235). It is of interest that IDDM13, IDDM7, and IDDM12, are all located on chromosome 2q31-35. This region may correspond to the mouse Idd5, possibly a multigenic susceptibility locus in the NOD mouse. Candidate genes in the region include the insulin growth factor-2 and ­5 binding proteins (IGFBP2, IGFBP5), which are expressed at decreased levels in patients with type 1 diabetes (339). A number of polymorphisms of IGFBP2, IGFBP5 and other genes in the region (including NEUROD, HOXD8, and CTLA4) were not associated with diabetes in a case-control study (340).
IDDM14
This denomination has not been assigned to any locus.
IDDM15
Linkage with the microsatellite D6S283 on chromosome 6q21 has been reported in families from France, Denmark, and the U.S. An Analysis of 408 multiplex families from Scandinavia confirmed HLA, INS, and IDDM15 (341). This locus is linked to HLA in males but not in females (299). IDDM15 is the third locus localized on 6q together with IDDM5 and IDDM8, but there is no evidence that these loci interact or are linked to IDDM1 on chromosome 6p. Multilocus analysis shows that linkage decreases with increasing distance from IDDM1 (Lod Score IDDM15>IDDM5>IDDM8) (299). As discussed later in this chapter, parental effects may influence susceptibility at the IDDM15 locus, and it has been suggested that the susceptibility gene at this locus may correspond to an imprinted gene associated with transient neonatal diabetes mellitus (299;342;343). Evidence for IDDM15 has been replicated in the latest genome scan performed by T1DGC, and this study provided evidence that this locus confers susceptibility independently of IDDM1 (39).
IDDM16
Field and coworkers analyzed immunoglobulin heavy chain (IGH) region microsatellites in 351 North American and British families and 241 families from Denmark with affected sibling pairs. Linkage was obtained for three markers close to the IGH gene cluster using affected sib-pair analysis but not using family-based methods. There was no linkage in the Danish data set but significant evidence for association, suggesting the IGH region may influence susceptibility to type 1 diabetes (344). The study raises the possibility that an immunoglobulin heavy chain gene may contribute to an autoimmune disorder with anti-islet autoantibodies. There was no evidence for IDDM16 in the 2005 T1DGC scan (39).
IDDM17
Unlike all of the preceding susceptibility loci, which have been mostly pinpointed by studying large collection of families with affected sibling pairs, evidence for the IDDM17 locus has been found studying a large Bedouin Arab family with 19 affected individuals (345). IDDM17 maps to the long arm of chromosome 10 (10q25). Recombination events occurring on this haplotype place IDDM17 within an 8-cM interval between markers D10S1750 and D10S1773. Two other markers, D10S592 and D10S554, showed evidence of linkage disequilibrium with the disease locus. Remarkably, one chromosome 10 haplotype, the B haplotype, was transmitted from a heterozygous parent to 13 of 13 affected offspring compared to 10 of 23 unaffected siblings. A 273-bp allele at D10S592 was transmitted to 8 of 10 affected offspring compared to 3 of 14 unaffected siblings, and a 151-bp allele at D10S554 was transmitted to 15 of 15 affected offspring compared with 10 of 24 unaffected siblings. Moreover, all of the affected members in this family carry one or two high-risk HLA-DR3 haplotypes that are rarely found in other family members. Thus, the study of this family suggests the alternative hypothesis that type 1 diabetes may be a oligogenic rather than polygenic disease, and that perhaps just two or three genes may suffice to explain all of the inherited susceptibility in a given family. This family has members affected by both celiac disease and type 1 diabetes. The region of association has been studied in detail with definition of more than 100 SNPs, with several SNPs in more than a single gene associated with significant distorted transmission to affected individuals. At present there is not enough genetic information to distinguish which polymorphism is primarily responsible for the disease association in this family (346;347), while there was no linkage with this region in the 2005 T1DGC scan (39).
IDDM18
Morahan and coworkers (348) reported linkage dysequilibrium between a single base pair change in the 3’ UTR of the IL12B gene (5q31.1-q33.1) and type 1 diabetes in two Australian cohorts . This gene encodes for the p40 subunit of interleukin-12 (IL-12). IL-12 is a disulphide-linked heterodimer composed of a heavy chain (p40, 40 kDa) and a light chain (p35, 35 kDa). The IL12A gene located on chromosome 3 encodes the light chain. The resulting heterodimer (p70 or p75) is the biologically active form of IL-12. IL-12p40 has been shown to stimulate Th-1 differentiation and IL-12 accelerates diabetes development in NOD mice. Thus, the IL-12B gene appears to be an important candidate gene in terms of immune function. Unfortunately, multiple studies of family datasets from the U.K., U.S. and Scandinavian countries did not reproduce evidence for linkage (39). A possible functional influence of the 3’ UTR on the mRNA expression levels of Il-12p40 remains unconfirmed, but it mostly relied on mRNA analysis in EBV cell lines without stimulation. While later studies used peripheral blood lymphocytes, a clear functional significance was not found (349). Further validation of the original findings reported by Morahan and whether the IL12B locus is a bona fide susceptibility seems critical, while there was no further evidence in the 2005 T1DGC scan (39).
Other Inherited Susceptibility Loci
Linkage has been reported with a few other loci that have not received an official denomination. The glucokinase gene (GCK) on chromosome 7 was linked to IDDM in 339 affected sib-pair families, but this finding has not been reproduced in other studies (350). Linkage was also reported for the D1S1617 marker on chromosome 1q (D1S1617) (351), and yet another locus may lie on chromosome X linked to markers DX6678 and DXS1068. This locus may influence the male-female bias in HLA-DR3-positive patients (352). The combined analysis of multiple data sets showed the most dramatic linkage (LOD=3.83) after IDDM1 and IDDM2 with a region on chromosome 16q22-q24 in association with D16S3098. This was the only “significant” LOD score (outside of IDDM1 and IDDM2) in this study of 767 multiplex families of Cox et al. (179). Evidence for linkage at this locus has been confirmed in the 2005 T1DGC genome scan. In this study, additional chromosomal regions with linkage to diabetes were 3p13-p14 (D3S1261), 9q33-q34 (D9S260), 12q14-q12 (D12S375), 16p12-q11.1 (D16S3131, 16q22-q24 (D16S504) and 19p13.3-p13.2 (INSR) (39).
Parental Effects On Inherited Genes Expression
The genetics of type 1 diabetes is further complicated by the possible existence of parental effects acting on the transmission and expression of inherited genes. Several studies have shown that diabetes risk differs in the offspring of diabetic mothers and fathers, although the results of different studies have been discrepant (353;354). It is also controversial whether parent of origin effects influence the transmission of IDDM1 alleles to the diabetic offspring (355-357). Moreover, there is evidence that parental origin effects may be operative at the IDDM8, IDDM10, and IDDM15 loci (299;321).
Parental effects also influence the transmission of the VNTR alleles at the IDDM2 locus, and probably this is the most studied locus in this regard. The first report of linkage at the IDDM2 locus found evidence, in a small subset of families that were informative for parental origin, that the excess allele sharing was exclusively paternal (358). Most of the subsequent studies of intra-familial association demonstrated a statistically significant difference only for paternally transmitted alleles (148;359) (360). These observations may be explained by imprinting, a mechanism that regulates gene expression by silencing either the maternal or the paternal allele. The silencing effect results from the epigenetic modification (probably mediated by methylation) of the DNA during the passage from the male or the female germline. This modification of the DNA marks the genetic material as maternal or paternal (parental imprint). Of note, the insulin gene is located in a region of the human genome that is known to be subject to parental imprinting (360). The IGF2 gene, which is adjacent to the insulin gene on chromosome 11p15,was the first human gene found to be imprinted and it is expressed exclusively from the paternal chromosome (361). Several other genes in the region are expressed from the paternal or maternal chromosomes only, at least in some tissues or developmental stages (362). INS is expressed from both copies in the pancreas of mice (363), human fetuses of 7-20 weeks gestation (182) and adult humans (200;364). However, monoallelic INS expression was observed in the pancreas of a 40-week old female fetus (364). INS is also expressed monoallelically, and specifically from the paternal chromosome in the mouse yolk sack (199). In addition, evidence has been presented for the imprinted paternal expression of INS in the human yolk sac (365). Thus, imprinted expression can depend on the tissue and possibly the developmental stage (366). The effects of imprinting on insulin expression may influence insulin expression during development and susceptibility to insulin/growth-related diseases in later life, such as insulin resistance and type 2 diabetes (365). More importantly, it has been shown that INS can be expressed monoallelically in the thymus (83;84). In all instances identified, the silenced allele was the one in cis with a class III VNTR. Such monoallelic expression resulting from the silencing of class III VNTR transcripts in the thymus may prevent the protective effect associated with the class III VNTR and explain the parent-of-origin effects discussed above. Vafiadis et al. studied in more detail the class III alleles that were silenced in the thymus (367). They developed a DNA fingerprinting method for identifying the type of alleles corresponding to the class III VNTR alleles that were found silenced in two thymus samples (S1, S2), and then analyzed the parental transmission of these type of class III alleles in a set of 287 diabetic children. Twelve of 18 possible transmissions of alleles matching the fingerprint of the S1 or S2 alleles were transmitted to the diabetic offspring, at a frequency of 0.67, which is significantly higher than the frequency of 0.38 seen in the remaining 142 class III alleles. These findings suggest that certain class III alleles may be predisposing instead of protective, and presumably these alleles are silenced in the thymus with obvious effects on the development of tolerance to insulin. Moreover, monoallelic INS expression was reported in the spleen of an 18 year-old Caucasian male, again preventing the expression of the INS transcript in cis with the class III VNTR allele (364). Assuming that monoallelic expression in this subject was mediated by imprinting (parents were unavailable to determine the parental origin of the silenced allele), this finding suggests that the imprint status may be maintained beyond development and perhaps throughout life. There is also evidence for even more complex mechanisms regulating INS transcription. Bennett et al. (368) studied more than 1,300 triads (two parents and affected child) and showed that the most common class I VNTR allele among Caucasians, termed 814 in arbitrary electrophoresis’ mobility units, has a protective effect similar to that of class III VNTR alleles. A protective effect of the 814 allele was independently confirmed in Basque families (228). This protective effect was apparent only when the 814 allele was inherited from fathers with an 814/class III VNTR genotype. In contrast, fathers with an 814/class I VNTR genotype transmitted both the 814 and other class I VNTR alleles to their diabetic children at similar frequency. This unusual transmission pattern suggests that this allele may behave differently in the offspring depending on the father's non-transmitted allele. This phenomenon could be explained by paramutation, a mechanism initially described in plants that requires some kind of physical interaction between homologous chromosomal domains in the pre-meiotic nucleus of the male germline (229). According to this mechanism, the function of the 814 class I VNTR allele can be modified by some interactions with the paternal class III allele that is not transmitted to the offspring. This hypothesis finds additional support in the finding that both cis- and trans-allelic interactions influence imprinting at the Ins2 locus in the mouse (369). Morphological evidence for a similar interaction has been presented for another imprinted locus (the Prader-Willi/Angelman syndrome locus on chromosome 15) in somatic cells (370). The data presented here suggest that besides allelic variation, parent-of-origin effects and complex epigenetic phenomena can dramatically influence INS transcription. It is important to notice that these phenomena can only be studied by evaluating INS transcription in selected tissues and correlating these data to the VNTR genotype. Thus, expression studies and genotyping must be combined to fully dissect the contribution of the insulin gene to diabetes susceptibility.
Other Non-Mendelian Regulatory Mechanisms
Besides parent of origin effects and other epigenetic phenomena, there is also evidence that alternative splicing can affect gene expression in a tissue specific manner and predispose to certain conditions (371). These include type 1 diabetes, multiple sclerosis, and other neurological diseases (372). In the case of type 1 diabetes, alternative splicing may affect the probability that one would mount autoimmune responses to the autoantigen IA-2. IA-2 is a tyrosine-phosphatase-like protein enriched in the secretory granules of islet and neuroendocrine cells (373), and consists of a single transmembrane (TM) region (residues 577-600) and extra- and intra-cellular domains (374;375).  An alternatively spliced variant of the IA-2 transcript has been discovered through the sequencing of a clone (ICA512.bdc) derived from a human pancreas library that is routinely used as a source of antigen in a specific assay for the detection IA-2 autoantibodies (376). This alternatively spliced transcript lacks exon 13 (Dexon 13), which codes for 73 amino acids (aa 557-629) encompassing the TM and juxta-membrane domains. The evaluation of the IA-2 expression in islets, thymus and spleen from non-diabetic human tissue donors revealed that thymus and spleen specimens exclusively express the Dexon 13 transcript and lack expression of the full-length transcript. Both transcripts are expressed in the pancreas. Another alternatively spliced IA-2 transcript in which 129 bp of exon 14 are spliced out, resulting in the deletion of 43 amino acids (aa 653-695) in the intracellular domain, was detected in about 50% of the pancreatic samples studied but essentially in none of the thymus and spleen specimens. Thus, alternative splicing causes differential IA-2 mRNA and protein expression in pancreas compared to lymphoid organs. Such differences may affect immune responsiveness to specific epitopes and help explain why IA-2 and not many other islet proteins become targets of autoimmunity in IDDM. Tolerance to linear or conformational epitopes typical of the full-length protein or of the Dexon 13 variant may not be achieved if these epitopes are expressed in islets but not in lymphoid organs. The specific lack of expression of the TM/Juxta-membrane domains (exon 13) in lymphoid organs helps in explaining why epitopes from these domains are often targeted by autoimmune responses in IDDM (372). Autoantibodies against IA-2 epitopes encoded by exons 13 and 14 have been reported in patients and can precede the appearance of autoantibodies against other intra-cellular epitopes (epitope spreading) (377-380). There is evidence that the HLA-DR4 restricted, naturally processed 654-674 epitope (exon 14) is recognized by autoreactive T-cells (381). Similar to the parent-of-origin effects affecting insulin gene expression in thymus (83) and peripheral lymphoid organs (84;364), differential IA-2 splicing appears to function as mechanism regulating gene expression independent of inherited alleles at the insulin and IA-2 loci. Although investigations had excluded linkage with IA-2 polymorphisms (234), these findings suggest that expression studies for selected candidate genes in tissues relevant to the disease process can help dissect the complex genetics of a multi-factorial disease such as type 1 diabetes.
"Acquired" Genetic Polymorphisms
Factors other than inherited genes must play a role in determining progression to overt disease in those individuals carrying predisposing genes. Environmental factors (viruses, diet) are suspected to be such determinants (reviewed in ref. (382)). The ability to identify genetic risk is aiding the search for environmental factors. It has been suggested that early introduction of cereals into infant diets dramatically increases development of anti-islet autoimmunity of high-risk (HLA/family history) individuals (383;384). Viruses could trigger specific autoimmune responses through mechanisms of molecular mimicry or by mediating a direct insult to ß-cells. It is also an intriguing and yet unproven possibility that novel genes may be acquired through, or their expression stimulated by, environmental factors (viruses or diet) after birth. Unlike more common viruses, retroviruses can integrate in the human genome. Retroviral genes can be either inherited or acquired after birth, and common viral infections and/or sex hormone changes associated with puberty may activate quiescent retroviruses. Such acquired expression may trigger the development of diabetes in genetically predisposed individuals either via cross-reactivity or immunity against novel viral antigens previously unknown to the immune system. This could drive immunity against the tissue that is expressing the novel gene, or to any tissue expressing molecules with significant cross-reactivity. Thus, environmental factors may provide or activate genes that could act as "disease genes". This hypothesis was supported by the finding that a human endogenous retrovirus, termed IDDMK1, 222, is apparently expressed and released from leukocytes in patients with type 1 diabetes but not in control individuals (385). Yet it is unclear whether this or similar retroviruses could be expressed in the endocrine pancreas. It was also suggested that IDDMK1, 222 could drive the same T-cell receptor restriction observed in T-lymphocytes infiltrating the endocrine pancreas of two children who died at the onset of diabetes (386), and act as a superantigen. However, the role of IDDMK1, 222 has been questioned by later studies. In fact, IDDMK1, 222 was found expressed at similar frequencies in patients and controls in several studies and no evidence for autoreactivity against this virus has been reported (387;388). The analysis of polymorphisms in the region of the endogenous retrovirus HERV-K18 or the DNA flanking it, including the CD48 gene, provided evidence for association of three variants belonging to a single haplotype. Genotype analysis suggested a dominantly protective effect of this haplotype. Further genetic and functional analyses are required to confirm these findings (389).
Genetics In Disease Prediction
With the current knowledge, high-risk individuals can be identified in the general population (31). Extremely high-risk individuals can be identified in families. In particular a number of large population based studies have been carried out stratifying individuals at birth by HLA genotype and insulin gene polymorphisms. Children born in Denver with the highest risk genotype DR3/4-DQ8 (further increase in risk can be provided by DRB1*04 sub-typing) comprise 2.4% of newborns and almost 50% of children developing anti-islet autoimmunity by age five (DAISY study) (150;390). The BabyDiab study of offspring of patients with type 1 diabetes in Germany and the DIPP study from Finland provide similar information concerning the risk associated with specific HLA genotypes and insulin gene polymorphisms (197;391-394).
In the DAISY study, siblings of patients with type 1 diabetes who have the highest risk HLA genotype (DR3/4-DQ8) have a risk of activating anti-islet autoimmunity of approximately 50% versus a risk of approximately 5% for the general population with the same class II HLA alleles. This dramatic difference in risk is at present unexplained and we have termed it the “relative paradox”. A risk approaching 50% for children who at birth are characterized only by high-risk class II HLA alleles (and as high as 80% if both the highest risk DR-DQ alleles and identical by descent for HLA haplotypes (23)) and relation to a proband with type 1 diabetes suggests that if environmental factors are of importance they are ubiquitous or “family” based. There may be ubiquitous environmental factors but possibly they play a minor role in determining familial aggregation of type 1 diabetes. The difference between relatives and the general population with the same class II HLA alleles could also be explained by additional genetic polymorphisms outside the HLA complex. Combined analysis of polymorphisms of the insulin and PTPN22 genes may further refine prediction (395). Among first-degree relatives with the high-risk HLA genotype that were followed for 3 years, 9 of 43 (28.1%) with the high-risk -23HphI polymorphism developed anti-islet autoantibodies versus two of 36 (5.6%) relatives with the lower-risk -23HphI genotypes. However, PTPN22 polymorphisms did not show a significant difference in risk by genotype in a study of 85 relatives. Overall, these results highlight the multiplicative risk of combined high-risk genotypes at different loci in terms of time to autoantibody and autoimmune disease development.
In addition, it is plausible that polymorphisms linked to the HLA complex or modulating the effects of the primary HLA determinants may have a greater impact on familial aggregation. This hypothesis stems from the observation that DR3/4-DQ8 siblings of patients with type 1 diabetes in the Denver DAISY study are almost always HLA identical to their sibling with diabetes. Namely, they share the complete HLA region by descent with their affected sibling, and thus all polymorphisms in this region are inherited together. There is growing evidence that polymorphisms of genes such as DP (396), class I HLA, and other genes within this region can modulate and contribute to risk and these would in families be shared with patients. In contrast, DR3 and DR4 haplotypes in the general population may not always carry the full complement of susceptibility alleles. A major effort to further dissect risk associated with the HLA region remains therefore crucial.
At present we can predict greatly increased risk of type 1 diabetes and a series of other autoimmune disorders by genetic typing at birth, using primarily information provided by HLA DNA based typing. The importance of this information will primarily be driven by our ability to use that information to prevent morbidity and mortality. For some disorders such as celiac disease, strongly associated with HLA-DR3-DQ2 haplotypes, altering the intake of gliadin is an effective therapy, and timing of gliadin introduction may be an important risk factor given genetic susceptibility. For type 1 diabetes we do not at present have a preventive therapy, but participation in studies such as DAISY decrease morbidity at the time of diagnosis (150). Whereas only 1 child of 30 in the DAISY study (HLA typing at birth followed by anti-islet autoantibody determination and metabolic follow up) required hospitalization at the onset of diabetes, approximately 40% of children presenting with diabetes of the general population (without screening) presented with ketoacidosis and required hospitalization (150). As illustrated in Fig. 7.14, many of the children from the general population had glucose greater than 1,000 mg% at diagnosis, and there is an important risk of death from cerebral edema when diagnosis of diabetes is delayed. Of note, even children from the general population with a relative with type 1 diabetes presented with severe metabolic abnormalities. This prevention of onset morbidity will need to be balanced against increased anxiety in families where a child is identified with increased disease risk. We believe it is likely that as the major efforts for prevention and rational treatment of a series of autoimmune diseases are developed, the balance will weigh toward identification, similar to newborn screening for a series of diseases in developed countries.

Figure 7.14.

“Monogenic" Forms Of Immune Mediated Diabetes
Dramatic progress in the understanding of the immunogenetics and pathogenesis of immune-mediated diabetes has come with the definition of a series of genes in animal models and man that underlie Mendelian forms of the disease. In particular, two very rare syndromes are now genetically characterized with plausible mechanistic hypotheses, namely APS-I (Autoimmune Polyendocrine Syndrome Type 1 (also termed APECED: Autoimmune Polyendocrinopathy-candidiasis-ectodermal dystrophy; OMIM 240300) (397;398) and IPEX (immune dysregulation, polyendocrinopathy, enteropathy, X-linked), also termed the XPID syndrome (X-linked Polyendocrinopathy, Immune Dysfunction and Diarrhea). The APECED or APS-I syndrome results from mutations of the AIRE (Autoimmune Regulator) gene. AIRE is a transcription factor acting as a major (but probably not the only one) determinant of the development of central thymic tolerance to “peripheral antigens” (399-401), which is mediated by the transcription of genes coding for peripheral proteins, for example, insulin, in medullary thymic epithelial cells and dendritic cells (212;402). The IPEX syndrome results from mutations in the Foxp3 gene. This is essential for the development of regulatory T lymphocytes (403). Both the APECED and IPEX syndromes are characterized by the development of immune mediated diabetes. Neonatal diabetes develops in patients with the IPEX syndrome while 18% of patients with APS-I develop diabetes as young children or even adults. Both syndromes are covered in detail in Chapter 8 of this web book. There is much to learn from these diseases about the pathophysiology of autoimmunity and key function such as thymic expression of self-molecules (AIRE) and the generation of regulatory cells (Foxp3) (400). At present there is no or little indication that polymorphisms at these two loci contribute to common forms of T1D susceptibility. While one small case-control study has reported an association of the Foxp3 gene with T1D in Japanese patients, another study in Sardinian families and a case-control cohort have not found evidence for linkage or an association with Foxp3 (404;405). Further genetic manipulation of diabetes prone nonobese diabetic (NOD) mice suggest that Foxp3 does not play a major role in the spontaneous development of diabetes in a model that closely resembles human type 1 diabetes (406). However, the administration of Foxp3+CD4+CD25+ regulatory T cells or the administration of T cells transduced with Foxp3 are reported to antagonize diabetes development in experimental rodent models, suggesting therapeutic potential even though Foxp3 may be a less specific marker of regulatory T cells in man (407;408).
Concluding Remarks
A large body of evidence indicates that genetic factors influence both susceptibility to and resistance to type 1 diabetes. Several chromosomal regions have been associated with the disease, suggesting that this is a polygenic disorder in most families. Coordinated efforts with large datasets combined with whole genome analyses, are now providing further insight into the genetic factors associated with type 1 diabetes. Mendelian mutations affecting certain genes result in rare monogenic syndromes, the study of which has led to better understanding of the molecular basis of autoimmunity and autoimmune diabetes. These are candidate genes for type 1 diabetes, as polymorphisms may affect their expression and function (albeit less dramatically than in the syndromes) and predispose to type 1 diabetes. Predictably, some of the susceptibility genes for type 1 diabetes are shared with other autoimmune diseases (e.g. PTPN22, CTLA4), while others appear to be disease specific. Based on the information generated so far, almost all of the loci appear to control immune function. It is still possible that some loci may have an effect on selected functions in pancreatic ß-cells, though genetic loci such as TCF7L2 that influences insulin secretion and development of type 2 diabetes is not associated with type 1 diabetes (409). It remains an open question whether no other loci with a major effect exist, similar in risk determination to that of HLA, or whether such loci may exist but be highly variable among families. In addition we believe it is likely that additional loci with effects potentially larger than those found in the recent Wellcome Trust Whole Genome analysis are present within or linked to the Major Histocompatibility Complex. Defining such loci is complicated by the extensive linkage dysequilibrium in this region that can extend for millions of base pairs. A number of groups are actively pursuing genetic candidates in this region. The ability to predict diabetes with the greatest accuracy based on genetic testing is a critical pre-requisite for the success of primary prevention strategies and, given the dramatic ability to predict risk of type 1A diabetes amongst relatives with the highest risk DR/DQ genotypes,  trials for primary prevention with for instance oral insulin (to induce mucosal tolerance: PrePoint) are about to begin based on algorithms identifying extreme genetic risk (determined by having multiple first degree relatives and HLA DR3/4-DQ2/8 or HLA haplotype identity by descent with sibling proband and HLA DR3/4-DQ2/8). A major goal is to define such extreme genetic risk in the general population, and this will almost certainly be dependent upon a fuller understanding of additional polymorphisms contributing to disease that are within or linked to the major histocompatibility complex.

Reference List - links to PubMed available in Reference List.

Chapter 7 Powerpoint slide set - Updated 9/09

Additional Slideset: Chapter7_Sardinia.ppt Type 1 Diabetes and Related ADs in Sardinia (M. Songini) - Slides added 1/06
Additional Slideset: Chapter7_Japan.ppt Genetics Type 1A Diabetes in Japan (H. Ikegami) - Slides added 10/05

For comments, corrections or to contribute teaching slides, please contact Dr. Eisenbarth at: george.eisenbarth@ucdenver.edu

Back to Type 1 Diabetes Table of Contents