The new england journal of medicine [601868]

The new england journal of medicine
n engl j med 360;16 nejm.org april 16, 2009 1646review article
Mechanisms of Disease
Genetics of Type 1A Diabetes
Patrick Concannon, Ph.D., Stephen S. Rich, Ph.D.,
and Gerald T. Nepom, M.D., Ph.D.
From the Center for Public Health Genom –
ics (P.C., S.S.R.) and the Departments of
Biochemistry and Molecular Genetics
(P.C.) and Public Health Sciences (S.S.R.),
University of Virginia, Charlottesville; and the Diabetes Research Program, Ben –
aroya Research Institute, Seattle (G.T.N.). Address reprint requests to Dr. Concan –
non at the Center for Public Health Ge –
nomics, University of Virginia, Charlottes –
ville, VA 22908, or at [anonimizat].
N Engl J Med 2009;360:1646-54.
Copyright © 2009 Massachusetts Medical Society.In 1976, the noted human geneticist James Neel titled a book chap –
ter “Diabetes Mellitus: A Geneticist’s Nightmare.”1 Over the past 30 years, how –
ever, the phenotypic and genetic heterogeneity of diabetes has been painstakingly
teased apart to reveal a family of disorders that are all characterized by the disrup –
tion of glucose homeostasis but that have fundamentally different causes. Recently,
the availability of detailed information on the structure and variation of the human genome and of new high-throughput techniques for exploiting these data has ge –
neticists dreaming of unraveling the genetic complexity that underlies these disor –
ders. This review focuses on type 1 diabetes mellitus and includes an update on
recent progress in understanding genetic factors that contribute to the disease and how this information may contribute to new approaches for prediction and thera –
peutic intervention.
Type 1 diabetes becomes clinically apparent after a preclinical period of varying
length, during which autoimmune destruction reduces the mass of beta cells in the pancreatic islets to a level at which blood glucose levels can no longer be main –
tained in a physiologic range. The disease has two subtypes: 1A, which includes the common, immune-mediated forms of the disease; and 1B, which includes nonim –
mune forms. In this review, we focus on subtype 1A, which for simplicity will be re –
ferred to as type 1 diabetes.
Although there are rare monogenic, immune-mediated forms of type 1 diabetes,
2,3
the common form is thought to be determined by the actions, and possible inter –
actions, of multiple genetic and environmental factors. The concordance for type 1
diabetes in monozygotic twins is less than 100%, and although type 1 diabetes ag –
gregates in some families, it does not segregate with any clear mode of inheri –
tance.4-7 Despite these complexities, knowledge of genetic factors that modify the
risk of type 1 diabetes offers the potential for improved prediction, stratification of
patients according to risk, and selection of possible therapeutic targets. As germ-line
factors, genetic risk variants are present and amenable to study at all times ― be –
fore, during, and after the development of diabetes. Thus, genetic information can
serve as a potential predictive tool and provide insights into pathogenetic factors occurring during the preclinical phase of the disease, when preventive measures might be applied.
Genetic Studies
Because of the uncertainty regarding the number and action of genes involved in
type 1 diabetes, genetic studies have tended to focus on approaches that require few assumptions about the underlying model of disease risk. The two primary approach –
es have been linkage studies (using pairs of affected relatives, typically siblings) and
association studies (using either case–control or family-based designs). Linkage stud –
ies using affected sibling pairs seek to identify regions of the genome that are shared
Copyright © 2009 Massachusetts Medical Society. All rights reserved. Downloaded from www.nejm.org by JESUS Z. VILLARREAL PEREZ MD on April 20, 2009 .

mechanisms of disease
n engl j med 360;16 nejm.org april 16, 2009 1647more frequently than by chance alone among sib –
lings who share the phenotype of type 1 diabe –
tes. Nuclear families, or even just the affected sib –
ling pairs themselves, are genotyped with panels
of markers spanning the genome at a modest den –
sity. Linkage between a marker and a susceptibil –
ity locus for type 1 diabetes is determined by ac –
cumulating evidence across families. Since affected
sibling pairs are relatively rare in type 1 diabetes, data from linkage studies are collected from a rather unique subgroup of families with type 1 dia –
betes. In general, linkage studies are the method of choice when the risk factors being sought have large effect sizes but are relatively rare. As risk fac –
tors become more common and have smaller ef –
fect sizes, association methods emerge as a poten –
tially more powerful approach (Fig. 1). Since the genetic basis of type 1 diabetes is probably a com –
plex mixture of small, moderate, and large genet –
ic effects, multiple strategies are needed and vary according to the population being studied and their exposure to unknown environmental factors.
Until recently, association studies in type 1 dia –
betes have focused on candidate genes, pathways, or chromosomal regions. Specifically selected mar

k

ers in genes of interest and the regions sur –
rounding those genes are genotyped in case sub –
jects and unaffected control subjects or, in some
studies, in case subjects and their parents, and the frequencies of marker alleles are compared between affected and unaffected chromosomes. However, association studies have recently been revolutionized by genomewide association stud –
ies,
8 as have linkage studies (to a lesser extent) for
a number of years.
Genetic Linkage Studies
The results of several genomewide searches for
linkage between genetic markers and type 1 dia –
betes have been reported previously.9-15 The stud –
ies have consistently reported significant evidence of linkage between the HLA region on chromo –
some 6p21 and type 1 diabetes. Although many
studies have shown suggestive evidence of link –
age at additional, non-HLA loci, findings at these loci have been inconsistent. The most likely ex –
planation is the limited size of these studies that
individually provide power to adequately detect only loci with large effects on the risk to siblings (such as HLA). A meta-analysis of data combined from most of the genomewide studies of linkage to type 1 diabetes has been carried out under the auspices of the Type 1 Diabetes Genetics Consor -tium (T1DGC)
15 (Table 1 ). This meta-analysis dem –
onstrated overwhelming evidence supporting link –
age to type 1 diabetes in the HLA region and
suggestive evidence at a small number of other
regions in the genome. In general, the emerging picture from linkage studies is that the class II genes encoding HLA-DR and HLA-DQ, as well as one or more additional genes within the HLA re –
gion, confer most of the genetic risk for type 1 dia –
betes. Genes outside the HLA region also con –
tribute to the risk of type 1 diabetes, but their
individual contributions are much smaller than that of HLA.
Candidate-Gene Association Studies
Although linkage studies have pointed to a num –
ber of regions of the genome that contain novel
genes that may contribute to the risk of type 1 dia –
betes, most identif ications of actual risk loci have 22p3Effect SizeUnlikely to exist
Unlikely to be foundAssociation studiesLinkage studies
Frequency in Population
AUTHOR:
FIGURE:
JOB:4-C
H/TRETAKE
SIZEICM
CASE
EMailLine
H/T
ComboRevised
AUTHOR, PLEASE NOTE:
Figure has been redrawn and type has been reset.
Please check carefully.REG F
Enon1st
2nd
3rdConcannon
1 of 3
04-1609ARTIST: ts
36016 ISSUE:Figure 1. Relative Strengths of Linkage and Association Approaches
for Mapping Genes in Complex Disorders.
The chart shows the effect of a disease allele’s frequency in the population
and its effect size on the optimal choice of study design. A disease allele that occurs frequently in the population and that has a large effect on dis –
ease risk is unlikely to exist. At the opposite end of the spectrum, a disease allele that is rare and has a small effect size is likely to exist but is unlikely to be found — and such alleles would be of limited public health interest. In general, linkage studies are most effective in disorders in which disease alleles are anticipated to have a large effect size but occur infrequently. As –
sociation studies are most effective for the detection of alleles that occur frequently but have a small effect size. These are general trends, and there are no specific boundaries in efficacy between the two approaches.
Copyright © 2009 Massachusetts Medical Society. All rights reserved. Downloaded from www.nejm.org by JESUS Z. VILLARREAL PEREZ MD on April 20, 2009 .

The new england journal of medicine
n engl j med 360;16 nejm.org april 16, 2009 1648come from studies of candidate genes. This, in
part, reflects differences in the approaches. Asso –
ciation depends on linkage disequilibrium between
specific alleles at an unknown causative variant
and testing with known polymorphic markers. Linkage disequilibrium extends over relatively short genomic distances in human populations (typi –
cally, tens to hundreds of kilobases) and is de –
pendent on ethnic background, ancestral history,
admixture, and local recombination frequencies. Thus, although a significant result in a genome
–
wide linkage scan may implicate a region span –
ning many megabases of DNA and subsequently
require substantial fine-mapping studies, a signifi –
cant result from an association study may impli –
cate a region of only a few hundred kilobases or less. The associated region may contain either a single gene or a few genes or be located in an ap –
parent “gene desert.”
A major exception to this pattern of few can –
didates in a disease-associated genomic region is chromosome 6p21 (the HLA region), where signi
–
ficant linkage disequilibrium spreads over several
megabases encoding hundreds of genes, many of which are reasonable candidates for involvement in type 1 diabetes. Specific genes in this region were originally investigated as candidate genes for
type 1 diabetes because of the roles of their prod -ucts in the presentation of antigens to the cel –
lular immune system.
16 Subsequent candidate-
gene studies have identified and confirmed other
risk loci for type 1 diabetes, including the gene
for insulin ( INS), a major autoantigen in type 1
diabetes, and CTLA4, which plays a role in T-cell
development and antigen recognition17-20 (Fig. 2).
Since the status of these genes as risk loci is well
established and they have been reviewed else –
where, they will not be discussed in detail here.
An examination of the loci in Figure 2 might
raise the question as to whether such loci, many
of which are predicted to have only modest indi –
vidual effects on risk, could have a clinically rel –
evant effect on phenotype or disease progression.
A relatively recent addition to the list of replicated candidate-gene associations to type 1 diabetes, PTPN22, is an excellent example of the insights
that can be gained through the identification of such loci, the phenotypic effects that might be
associated with such a gene, and its potential use
as a target for intervention. PTPN22 encodes the
lymphoid protein tyrosine phosphatase (LYP).
21
In T cells, LYP acts in a complex with C-terminal
Src kinase (CSK) to negatively regulate signaling
from the T-cell receptor. Specifically, LYP dephos –
phorylates positive regulatory tyrosines on LCK,
VAV, ZAP-70, and CD3 zeta chains, thereby caus -Table 1. Current Status of Linkage Data for Type 1 Diabetes.*
Chromosomal
RegionPosition
(cM)Closest
MarkerLod
ScoreSibling
Risk RatioLod-1
Interval†Nominal
P Value
2q31-33 192 D2S2167 3.34 1.19 177–204 9.0×10−5
3p13-p14 98 D3S1261 1.52 1.15 78–112 8.2×10−3
6p21 47 TNFA 116.3 3.35 46–48 4.9×10−52
6q21 80 D6S283 22.39 1.56 ND ND
9q33-q34 150 D9S260 2.20 1.13 138–161 1.5×10−3
10p14-q11 61 D10S1426 3.21 1.12 52–66 1.2×10−4
11p15 2 D11S922 1.87 1.16 0–14 3.4×10−3
12q14-q12 81 D12S375 1.66 1.10 77–83 5.8×10−3
16p12-q11.1 56 D16S3131 1.88 1.17 26–71 3.3×10−3
16q22-q24 108 D16S504 2.64 1.19 100–121 4.9×10−4
19p13.3-p13.2 25 INSR 1.92 1.15 0–43 3.0×10−3
* Data are from the Type 1 Diabetes Genetics Consortium.15 The abbreviation cM denotes centimorgan, lod logarithmic
odds, and ND not done.
†

The lod-1 interval is the size of the interval (in centimorgans) in which the lod score is greater than or equal to the
maximum minus 1.0.
Copyright © 2009 Massachusetts Medical Society. All rights reserved. Downloaded from www.nejm.org by JESUS Z. VILLARREAL PEREZ MD on April 20, 2009 .

mechanisms of disease
n engl j med 360;16 nejm.org april 16, 2009 1649ing down-regulation of signals emanating from
the T-cell receptor.22 The minor allele of a single-
nucleotide polymorphism (SNP) in the coding re –
gion of PTPN22, rs2476601 (1858C→T), results in
an arginine-to-tryptophan substitution at residue
620 in LYP (R620W), disrupting its ability to in –
teract with CSK.
In the case of PTPN22, it might be anticipated
that autoimmunity would arise when a genetic
variant resulted in reduced LYP activity and con –
sequent T-cell hyperactivity. This hypothesis is consistent with the expansion of T-cell popula –
tions in mice in which the orthologue of LYP is knocked out.
23 However, the PTPN22 1858T allele
is associated with reduced T-cell activation. T cells
from heterozygous carriers of this allele have re –
duced phosphorylation of LYP targets and de –
creased T-cell signaling as assayed by antigen-
stimulated calcium flux or cytokine secretion.24,25
There is also a suggestion of a dose effect of
PTPN22 1858T in some studies.26-28 These results
are consistent with studies using mice engineered with different genetic defects in T-cell signaling. Severe, inactivating mutations tend to result in im –
munodeficiency, whereas more subtle missense
mutations result in dysregulation and, in some cases, autoimmunity.
29 Thus, T-cell signaling may
be similar to a quantitative trait with thresholds
for different phenotypes that might be amenable
to pharmacologic manipulation. The apparent
“gain of function” associated with the PTPN22
R620W variant and the putative effect of gene dose raise the possibility that selective inhibitors could target PTPN22 as a possible therapeutic approach.
Such an approach is further encouraged by numer –
ous reports that the same allelic variant (R620W)
modifies risk in several other common autoim –
mune diseases, including rheumatoid arthritis, systemic lupus erythematosus, and Graves’ dis –
ease.
26,30,31
Genomewide Association Studies
Genomewide association studies take advantage
of newly developed, high-throughput SNP geno –
typing platforms and the development of dense
maps of SNPs from the human genome. The re -33p9
n
n2.50Odds Ratio1.502.002.25
1.00
0.500.751.75
0.251.25
0.00
INS
PTPN22IL2RASH2B3ERBB3
CIQTNF6CCR5IFIH1CTSHCD226 IL2RAIL2
BACH2
UBASH3ARGS1IL7RAPRKCQ PTPN2IL18RAPCTLA4CLEC16APTPN2TNFAIP3TNFAIP3TAGAP
Locus
AUTHOR:
FIGURE:
JOB:4-C
H/TRETAKE
SIZEICM
CASE
EMailLine
H/T
ComboRevised
AUTHOR, PLEASE NOTE:
Figure has been redrawn and type has been reset.
Please check carefully.REG F
Enon1st
2nd
3rdConcannon
2 of 3
04-16-09ARTIST: ts
36016 ISSUE:Insulin production
and metabolismImmunity Protection from beta-
cell apoptosisUnknown function
Figure 2. Putative Functions of Non–HLA-Associated Loci in Type 1 Diabetes.
The y axis indicates the best estimate of the odds ratio for risk alleles at each of the indicated loci on the basis of
currently published data. Although not shown, the HLA region has a predicted odds ratio of approximately 6.8. On the x axis are indicated possible candidate genes within genomic regions in which convincing associations with type 1 diabetes have been reported. On the basis of the known functions of these candidate genes, the corresponding bars in the graph depicting odds ratios have been color-coded to suggest possible roles of these loci in susceptibility to type 1 diabetes. At IL2RA and TNFAIP3, there is evidence of two independent effects on risk with different odds ra –
tios, so these loci both appear twice in the figure. An excellent resource for current information on all aspects of genes implicated in type 1 diabetes is T1DBase (www.t1dbase.org).
Copyright © 2009 Massachusetts Medical Society. All rights reserved. Downloaded from www.nejm.org by JESUS Z. VILLARREAL PEREZ MD on April 20, 2009 .

The new england journal of medicine
n engl j med 360;16 nejm.org april 16, 2009 1650sults of three association scans of type 1 diabetes
have been reported, including an initial study that genotyped only nonsynonymous coding SNPs and
two later studies that used much denser panels of
SNPs (≥300,000 per subject) distributed across the genome.
32-35 The results of these studies replicate
findings for confirmed loci cited earlier in this
review but, more important, provide evidence for
a number of novel loci. A subsequent meta-analy
–
sis of genomewide association studies36 and fol –
low-up studies37 has further added to the list of
new loci.
Among the genes implicated in the scan of
nonsynonymous coding SNPs and replicated in
a separate population were the previously identi –
fied 1858C→T SNP in PTPN22 a n d a S N P i n
IFIH1.32,38,39 IFIH1 is a gene that encodes an in –
terferon-induced helicase, also known as Mda-5,
which plays a role in innate immunity through
the recognition of the RNA genomes of picorna –
viruses.40 Numerous attempts have been made to
link various infectious agents to the risk of type 1
diabetes, making the identification of a gene spe –
cifically involved in viral defenses as a risk factor
for type 1 diabetes particularly intriguing. Prom –
inent among the viruses that have been proposed as potential environmental triggers for type 1 dia –
betes is coxsackievirus B4, an enterovirus belong –
ing to the picornavirus family.
41 The coding SNP
in IFIH1 at which association with type 1 diabe –
tes is detected predicts an alanine-to-threonine
substitution. Whether it is this specific substitu –
tion in the helicase protein that confers a risk for
type 1 diabetes has yet to be determined.
The high-density genomewide association stud –
ies in type 1 diabetes provide confirmatory evi –
dence for previously identified loci such as INS,
PTPN22, CTLA4, and IL2RA, as well as significant
findings for a number of new regions. Although
considerable fine mapping and characterization of these new regions remain to be performed, likely candidate genes within the regions suggest
a prominent role for effects on immunity (Fig. 2).
Notable among the genes contained within these regions are PTPN2, a second protein tyrosine phos –
phatase. PTPN2 is expressed ubiquitously but at highest levels in hematopoietic cells, where it acts, in part, to regulate signaling by dephosphorylating multiple JAK and STAT molecules. One region that is implicated in genomewide association studies of type 1 diabetes contains a gene of unknown function, CLEC16A, that has been annotated as a
possible C-type lectin.
33,34,38 A SNP in this gene had a significant association with multiple scle –
rosis in a separate genomewide study,42 which also
showed evidence of association at two other loci
— IL7R and IL2RA — that are implicated in type 1
diabetes. These data provide suggestive evidence,
beyond that provided by the examples of HLA and PTPN22, of common genetic risk factors and
common mechanisms that may lead to autoim –
munity.
Disease Prediction
Current approaches for the prediction of type 1
diabetes take advantage of the major genetic risk factors, genotyping for HLA-DR and HLA-DQ loci
(which is then combined with family history), and
screening for autoantibodies directed against is –
let-cell antigens.
43,44 The individual distribution
of specific risk alleles correlates with gradations
in disease penetrance, enabling a tiered staging
strategy for the prediction of type 1 diabetes. For
example, children who carry both of the highest-risk HLA haplotypes (DR3–DQ2 and DR4–DQ8) have a risk of approximately 1 in 20 for a diagno –
sis of type 1 diabetes by the age of 15 years.
45 If
the child has a sibling who has diabetes and the
same haplotypes, the risk is even higher (approx –
imately 55%).46 Since this haplotype combination
occurs in only 2.3% of the white population, it is possible to envision universal screening strategies
that pinpoint this highest-risk group. Inclusion
of additional moderate HLA risk haplotypes and screening for autoantibodies would add cost and complexity to a population-screening approach but have the potential to identify the majority of all
children with diabetes before the onset of the dis –
ease. If this were possible, then tests of potential preventive strategies could be performed, as out –
lined later in this article. The large number of new risk loci for type 1 diabetes that were recently iden –
tified from genomewide association studies could
be added to these prediction schemes. These ge –
netic factors are relatively easy, inexpensive, and
noninvasive to measure and can be detected well before other features, such as autoantibodies, would typically develop.
As true risk variants for type 1 diabetes are
fine mapped, identified, and characterized, their functional use for prediction and prevention should become clearer. Even based on the current collec –
tion of implicated risk loci, it is obvious that mul –
tiple distinct biochemical pathways are involved. Not all pathways are likely to influence the risk
Copyright © 2009 Massachusetts Medical Society. All rights reserved. Downloaded from www.nejm.org by JESUS Z. VILLARREAL PEREZ MD on April 20, 2009 .

mechanisms of disease
n engl j med 360;16 nejm.org april 16, 2009 1651of type 1 diabetes in the same way (Fig. 3). Some
may be associated with an earlier (or a later) age of onset, a slower or faster rate of loss of beta
cells, or a different pattern of epitope spreading
in the autoimmune destruction of islets. Although some variants make small individual contributions to risk, they may cluster in pathways so that func -tional assays targeting these processes may have
useful predictive value.
future Genetic studies
Despite the increasing number of potential target genes, considerable work remains to develop these
Pancreas
Cellular immunityHealthy islet
Autoimmune diabetes
Genes that expand self-reactive cells
Genes that modify immune function
Genes that interfere with immune regulation
Genes that influence beta-cell survival
3
Schwar tz3/26/09
AUTHOR PLEASE NOTE:
Figure has been redrawn and type has been reset
Please check carefullyAuthor
Fig #
Title
ME
DE
Artist
Issue dateCOLOR FIGURE
Draft 7
Concannon
Knoper
4/16/09 Differential roles of risk
loci in T1D pathogenesisBeta cell
Damaged beta cell
Figure 3. Differential Roles of Risk Loci in the Pathogenesis of Type 1 Diabetes.
Current data suggest that many risk loci for type 1 diabetes may exert their effects through the immune system.
Within the immune response, these genes can act at multiple levels, affecting the establishment of the immune rep –
ertoire, the function of cell types in the immune system, or the regulation of cellular responses that can lead to au –
toimmunity.
Copyright © 2009 Massachusetts Medical Society. All rights reserved. Downloaded from www.nejm.org by JESUS Z. VILLARREAL PEREZ MD on April 20, 2009 .

The new england journal of medicine
n engl j med 360;16 nejm.org april 16, 2009 1652findings into a better understanding of the cause
of type 1 diabetes and to translate these findings into clinical applications. A first step in following
up on association results is detailed fine mapping
of the region. This step is necessary to determine whether the primary association is confined to a gene of interest in the region and whether the association can be attributed to an allele of the
original SNP, to a group of alleles that are com –
monly coinherited on a single haplotype, or to alleles at multiple SNPs in the region that inde –
pendently contribute to disease risk. Additional genotyping may also provide evidence that the as –
sociation is stronger in a flanking gene or an in –
tergenic region.
Once a credible risk variant has been identified,
a second step is required to determine the proxi –
mal effect of the SNP — that is, the immediate effect of genetic variation at this position on gene expression or on the function of a specified pro –
tein. Possible effects might include moderation of transcription levels, differential splicing, direct ef –
fects of nonsynonymous substitutions on protein function, or indirect effects mediated through microRNAs.
The third step is to determine whether the
presence of the risk variant is associated with a
discernible phenotype in patients with type 1 dia –
betes. Identifying such endophenotypes may re –
quire a substantial number of subjects genotyped at all known risk loci in order to dissect the ef –
fects of individual loci as well as carefully planned clinical research (perhaps based on genotype). An advantage in doing this type of study with type 1 diabetes is that many of the implicated loci ap –
pear to function primarily in cells of the immune system, which allows for access to the involved cell populations in subjects with known genotypes.
It is also important to recognize that although
linkage studies are perhaps not currently as fash –
ionable as genomewide association approaches, the regions that are identified through a family-based linkage approach still merit follow-up. In addition, family studies are useful in establishing effects of any variant identified as a risk factor for type 1 diabetes. Mendelian transmission of the causal variant cosegregating with an endopheno –
type that clusters family members who are at risk from those who are protected against the disease would have important biologic, clinical, and thera –
peutic implications.Diabetes and Personalized
Medicine
In type 1 diabetes, there is an extended preclini –
cal period during which escalating autoimmune
destruction depletes beta cells and thereby reduces
insulin secretion and the ability to maintain glu –
cose homeostasis. This period provides a window
for interventions that could prevent overt diabetes by slowing or halting the progression of beta-cell loss. Early attempts at prevention using broadly
immunosuppressive treatments have progressed to targeted approaches that seek to induce immune tolerance.
47-50 The design of such trials requires
a very careful risk–benefit analysis that puts a pre –
mium on prediction of disease risk and potential
outcome. Ideally, therapies with a higher risk of
adverse effects would be matched to patients with a higher predicted risk of type 1 diabetes. As ther –
apies become more targeted, the likelihood of dif –
ferential responsiveness to therapy among patients
will increase. Biomarkers that could more accu –
rately predict response to specific types of thera –
pies would increase the efficiency of trials. There is also suggestive evidence that for at least some therapies, earlier intervention, at a stage at which an increased beta-cell mass remains, may improve
outcomes.
50
Similarly, there may be biochemical pathways
that are affected by particular risk variants that
play a role in response to certain preventive thera –
pies. As one example, a short course of treatment
with humanized non–Fc-receptor–binding mono –
clonal antibodies to the T-cell receptor component CD3 has been shown to have long-term effects by slowing the loss of insulin secretion in patients
with newly diagnosed diabetes.
49 This effect does
not appear to result from T-cell depletion, since
treated patients attain normal levels of circulating
lymphocytes within 2 weeks after cessation of
treatment, but may result from tolerization through the induction of regulatory T cells.
51 Since this
therapy targets T-cell activation through the T-cell
receptor, the recruitment of intracellular signal –
ing proteins, such as LCK, FYN, and SYK, is re –
quired. These signaling pathways are potentially
affected by the products of several of the currently identified risk loci, including LYP, which normally acts to dephosphorylate LCK and down-regulates
signaling from the T-cell receptor; SH2B3, an adapter protein that is a phosphorylation target of
Copyright © 2009 Massachusetts Medical Society. All rights reserved. Downloaded from www.nejm.org by JESUS Z. VILLARREAL PEREZ MD on April 20, 2009 .

mechanisms of disease
n engl j med 360;16 nejm.org april 16, 2009 1653LCK, which binds to the CD3 zeta chain52; and
UBASH3A (also called TULA and Sts2), which
together with the related UBASH3B protein sup –
presses T-cell signaling, in part, through the de –
phosphorylation of ZAP70, FYN, and SYK.53 Risk
loci such as these, whose products act within T-cell signaling pathways, could be candidate biomark –
ers for predicting responsiveness to therapies with
agents such as anti-CD3 monoclonal antibody that are directed at T-cell activation.
conclusions
What general conclusions can be drawn from our
current state of understanding of the genetics of type 1 diabetes? Genes within the HLA region, pre –
dominantly those that encode antigen-presenting
molecules, confer the greatest part of the genetic risk of type 1 diabetes. The existence of other loci with individual effects on risk of a similar mag –
nitude is very unlikely. The remaining non-HLA
loci will make only modest individual contribu –
tions to risk; most will probably have odds ratios of 1.3 or less. A majority of these other loci ap –
pear to exert their effects in the immune system, particularly on T cells, affecting antigen-driven T-cell activation and cytokine signaling, prolifera –
tion, or maturation. Careful dissection of the bio –
chemical pathways in which the products of these loci are known to function should allow an under -standing of how they act to confer a risk of type 1 diabetes. Refinement of our genetic mapping of these loci may improve our ability to predict the risk of type 1 diabetes and facilitate the testing of more aggressive preventive therapies. Dissection of the phenotypic effects of variation at these loci should provide new insights into the preclinical
period of type 1 diabetes and potentially suggest new, rationally designed therapies.
It has long been anticipated that loci contrib –
uting in some generalized manner to the develop –
ment of autoimmunity would be identified. The
apparent identification of multiple common risk loci in recent independent genomewide associa –
tion studies in different autoimmune disorders appears to fulfill this prediction.
54 Although these
loci are identified because of their association with specific autoimmune disorders, such as type 1 dia –
betes, it will be desirable to study their effect on
human health prospectively by following large co –
horts of genotyped subjects to understand the broader range of immune variation, including re –
sponses to infection and vaccines.
Supported by grants (DK062418 and DK46635) from the Na –
tional Institutes of Health and from the Juvenile Diabetes Re –
search Foundation.
No potential conflict of interest relevant to this article was
reported.
We thank the many patients and their families who contrib –
uted to the genetic studies summarized here and Sarah Field and John Todd for sharing data that were adapted for Figure 2.
References
Neel J. Diabetes mellitus: a geneticist’s 1.
nightmare. In: Creutzfeldt W, Kobberling J,
Neel JV, eds. The genetics of diabetes mel –
litus. Berlin: Springer-Verlag, 1976:1-11.
Nagamine K, Peterson P, Scott HS, et al. 2.
Positional cloning of the APECED gene. Nat Genet 1997;17:393-8.
Bennett CL, Christie J, Ramsdell F, et
3.
al. The immune dysregulation, polyendo –
crinopathy, enteropathy, X-linked syndrome (IPEX) is caused by mutations of FOXP3.
Nat Genet 2001;27:20-1.
Barnett AH, Eff C, Leslie RDG, Pyke
4.
DA. Diabetes in identical twins: a study of
200 pairs. Diabetologia 1981;20:87-93.
Hyttinen V, Kaprio J, Kinnunen L, Ko – 5.
skenvuo M, Tuomilehto J. Genetic liability of type 1 diabetes and the onset age among 22,650 young Finnish twin pairs: a nation –
wide follow-up study. Diabetes 2003;52:
1052-5.
Rich SS. Mapping genes in diabetes:
6.
genetic epidemiological perspective. Dia –
betes 1990;39:1315-9.
Redondo MJ, Yu L, Hawa M, et al. Het – 7. erogeneity of type I diabetes: analysis of monozygotic twins in Great Britain and the United States. Diabetologia 2001;44:
354-62. [Erratum, Diabetologia 2001;44:
927.]
McCarthy MI, Abecasis GR, Cardon
8.
LR, et al. Genome-wide association stud –
ies for complex traits: consensus, uncer –
tainty and challenges. Nat Rev Genet 2008;
9:356-69.
Davies JL, Kawaguchi Y, Bennett ST, 9.
et al. A genome-wide search for human type 1 diabetes susceptibility genes. Na –
ture 1994;371:130-6.
Hashimoto L, Habita C, Beressi JP, et
10.
al. Genetic mapping of a susceptibility locus for insulin-dependent diabetes mel –
litus on chromosome 11q. Nature 1994;
371:161-4.
Concannon P, Gogolin-Ewens KJ,
11.
Hinds DA, et al. A second-generation screen of the human genome for suscepti –
bility to insulin-dependent diabetes mel –
litus (IDDM). Nat Genet 1998;19:292-6.
Mein CA, Esposito L, Dunn MG, et al.
12. A search for type 1 diabetes susceptibility genes in families from the United King –
dom. Nat Genet 1998;19:297-300.
Nerup J, Pociot F. A genomewide scan
13.
for type 1-diabetes susceptibility in Scan-dinavian families: identification of new loci with evidence of interactions. Am J Hum
Genet 2001;69:1301-13.
Cox NJ, Wapelhorst B, Morrison VA,
14.
et al. Seven regions of the genome show
evidence of linkage to type 1 diabetes in a consensus analysis of 767 multiplex fami –
lies. Am J Hum Genet 2001;69:820-30.
Concannon P, Erlich HA, Julier C, et
15.
al. Type 1 diabetes: evidence for suscepti –
bility loci from four genome-wide linkage scans in 1,435 multiplex families. Diabe –
tes 2005;54:2995-3001.
Nerup J, Platz P, Andersen OO, et al.
16.
HL-A antigens and diabetes mellitus. Lan –
cet 1974;2:864-6.
Bell GI, Horita S, Karam JH. A polymor – 17.
phic locus near the human insulin gene is associated with insulin-dependent diabetes mellitus. Diabetes 1984;33:176-83.
Copyright © 2009 Massachusetts Medical Society. All rights reserved. Downloaded from www.nejm.org by JESUS Z. VILLARREAL PEREZ MD on April 20, 2009 .

n engl j med 360;16 nejm.org april 16, 2009 1654mechanisms of disease
Nisticò L, Buzzetti R, Pritchard LE, et 18.
al. The CTLA-4 gene region of chromosome
2q33 is linked to, and associated with, type 1 diabetes. Hum Mol Genet 1996;5:
1075-80.
Ueda H, Howson JM, Esposito L, et al.
19.
Association of the T-cell regulatory gene CTLA4 with susceptibility to autoimmune disease. Nature 2003;423:506-11.
Lowe CE, Cooper JD, Brusko T, et al.
20.
Large-scale genetic fine mapping and ge – notype-phenotype associations implicate polymorphism in the IL2RA region in
type 1 diabetes. Nat Genet 2007;39:1074-
82.
Bottini N, Musumeci L, Alonso A, et al.
21.
A functional variant of lymphoid tyrosine
phosphatase is associated with type I dia –
betes. Nat Genet 2004;36:337-8.
Wu J, Katrekar A, Honigberg LA, et al. 22.
Identification of substrates of human pro –
tein-tyrosine phosphatase PTPN22. J Biol Chem 2006;281:11002-10.
Hasegawa K, Martin F, Huang G, Tu –
23.
mas D, Diehl L, Chan AC. PEST domain-enriched tyrosine phosphatase (PEP) regu –
lation of effector/memory T cells. Science
2004;303:685-9.
Vang T, Congia M, Macis MD, et al.
24.
Autoimmune-associated lymphoid tyrosine
phosphatase is a gain-of-function variant. Nat Genet 2005;37:1317-9.
Rieck M, Arechiga A, Onengut-Gumus –
25.
cu S, Greenbaum C, Concannon P, Buck –
ner JH. Genetic variation in PTPN22 cor –
responds to altered function of T and B
lymphocytes. J Immunol 2007;179:4704-10.
Kyogoku C, Langefeld CD, Ortmann 26.
WA, et al. Genetic association of the R620W
polymorphism of protein tyrosine phos –
phatase PTPN22 with human SLE. Am J
Hum Genet 2004;75:504-7.
Lee AT, Li W, Liew A, et al. The 27.
PTPN22 R620W polymorphism associates
with RF positive rheumatoid arthritis in
a dose-dependent manner but not with
HLA-SE status. Genes Immun 2005;6:129-
33.
Smyth D, Cooper JD, Collins JE, et al. 28.
Replication of an association between the
lymphoid tyrosine phosphatase locus (LYP/
PTPN22) with type 1 diabetes, and evi –
dence for its role as a general autoimmu –
nity locus. Diabetes 2004;53:3020-3.
Siggs OM, Miosge LA, Yates AL, et al. 29.
Opposing functions of the T cell receptor kinase ZAP-70 in immunity and tolerance differentially titrate in response to nu –
cleotide substitutions. Immunity 2007;27:
912-26.Begovich AB, Carlton VE, Honigberg
30.
LA, et al. A missense single-nucleotide poly –
morphism in a gene encoding a protein tyrosine phosphatase (PTPN22) is associ –
ated with rheumatoid arthritis. Am J Hum
Genet 2004;75:330-7.
Criswell LA, Pfeiffer KA, Lum RF, et
31.
al. Analysis of families in the Multiple Au –
toimmune Disease Genetics Consortium
(MADGC) collection: the PTPN22 620W
allele associates with multiple autoimmune
phenotypes. Am J Hum Genet 2005;76:
561-71.
Burton PR, Clayton DG, Cardon LR, et 32.
al. Association scan of 14,500 nonsynony –
mous SNPs in four diseases identifies au –
toimmunity variants. Nat Genet 2007;39:
1329-37.
Wellcome Trust Case Control Consor – 33.
tium. Genome-wide association study of
14,000 cases of seven common diseases and 3,000 shared controls. Nature 2007;
447:661-78.
Hakonarson H, Grant SF, Bradfield JP,
34.
et al. A genome-wide association study identifies KIAA0350 as a type 1 diabetes gene. Nature 2007;448:591-4.
Hakonarson H, Qu HQ, Bradfield JP,
35.
et al. A novel susceptibility locus for type 1 diabetes on Chr12q13 identified by a genome-wide association study. Diabetes
2008;57:1143-6.
Cooper JD, Smyth DJ, Smiles AM, et
36.
al. Meta-analysis of genome-wide associa –
tion study data identifies additional type 1
diabetes risk loci. Nat Genet 2008;40:
1399-401.
Grant SF, Qu HQ, Bradfield JP, et al. 37.
Follow up analysis of genome-wide asso –
ciation data identifies novel loci for type 1 diabetes. Diabetes 2009;58:290-5.
Todd JA, Walker NM, Cooper JD, et al.
38.
Robust associations of four new chromo –
some regions from genome-wide analyses of type 1 diabetes. Nat Genet 2007;39:857-
64.
Smyth DJ, Cooper JD, Bailey R, et al.
39.
A genome-wide association study of non –
synonymous SNPs identifies a type 1 dia-
betes locus in the interferon-induced heli –
case (IFIH1) region. Nat Genet 2006;38:
617-9.
Kato H, Takeuchi O, Sato S, et al. Dif – 40.
ferential roles of MDA5 and RIG-I heli –
cases in the recognition of RNA viruses. Nature 2006;441:101-5.
Clements GB, Galbraith DN, Taylor
41.
KW. Coxsackie B virus infection and on –
set of childhood diabetes. Lancet 1995;
346:221-3.The International Multiple Sclerosis 42.
Genetics Consortium. Risk alleles for mul –
tiple sclerosis identified by a genomewide study. N Engl J Med 2007;357:851-62.
Barker JM, Barriga KJ, Yu L, et al. Pre –
43.
diction of autoantibody positivity and pro –
gression to type 1 diabetes: Diabetes Auto –
immunity Study in the Young (DAISY).
J Clin Endocrinol Metab 2004;89:3896-902.
Eisenbarth GS. Prediction of type 1 44.
diabetes: the natural history of the predi-abetic period. Adv Exp Med Biol 2004;
552:268-90.
Aly TA, Ide A, Humphrey K, et al. Ge –
45.
netic prediction of autoimmunity: initial oligogenic prediction of anti-islet autoim –
munity amongst DR3/DR4-DQ8 relatives
of patients with type 1A diabetes. J Auto –
immun 2005;25:Suppl:40-5.
Aly TA, Ide A, Jahromi MM, et al. Ex –
46.
treme genetic risk for type 1A diabetes.
Proc Natl Acad Sci U S A 2006;103:14074-9.
Bougneres PF, Carel JC, Castano L, et al. 47.
Factors associated with early remission of type I diabetes in children treated with cy –
closporine. N Engl J Med 1988;318:663-70.
Silverstein J, Maclaren N, Riley W,
48.
Spillar R, Radjenovic D, Johnson S. Im –
munosuppression with azathioprine and prednisone in recent-onset insulin-depen –
dent diabetes mellitus. N Engl J Med 1988;
319:599-604.
Herold KC, Hagopian W, Auger JA, et
49.
al. Anti-CD3 monoclonal antibody in new-onset type 1 diabetes mellitus. N Engl J Med 2002;346:1692-8.
Keymeulen B, Vandemeulebroucke E,
50.
Ziegler AG, et al. Insulin needs after CD3-antibody therapy in new-onset type 1 dia –
betes. N Engl J Med 2005;352:2598-608.
Bisikirska B, Colgan J, Luban J, Blue –
51.
stone JA, Herold KC. TCR stimulation with modified anti-CD3 mAb expands CD8+
T cell population and induces CD8+CD25+
Tregs. J Clin Invest 2005;115:2904-13.
Li Y, He X, Schembri-King J, Jakes S,
52.
Hayashi J. Cloning and characterization
of human Lnk, an adaptor protein with pleckstrin homology and Src homology
2 domains that can inhibit T cell activa –
tion. J Immunol 2000;164:5199-206.
Carpino N, Turner S, Mekala D, et al.
53.
Regulation of ZAP-70 activation and TCR signaling by two related proteins, Sts-1 and Sts-2. Immunity 2004;20:37-46.
Xavier RJ, Rioux JD. Genome-wide as –
54.
sociation studies: a new window into im –
mune-mediated diseases. Nat Rev Immu –
nol 2008;8:631-43.
Copyright © 2009 Massachusetts Medical Society.
Copyright © 2009 Massachusetts Medical Society. All rights reserved. Downloaded from www.nejm.org by JESUS Z. VILLARREAL PEREZ MD on April 20, 2009 .