| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Department of Diabetes and Vascular Medicine, School of Postgraduate Medicine and Health Sciences, University of Exeter (T.M.F., S.E., A.T.H.), EX2 5AX Exeter; the Division of Medicine, Imperial College School of Medicine, St. Mary’s Hospital (M.I.McC.), London; the Department of Medicine, School of Medicine (M.W.), Newcastle-upon-Tyne; Diabetes Research Laboratories, Radcliffe Infirmary (J.C.L.), Oxford; the Departments of Medicine and Clinical Biochemistry, Addenbrooke’s Hospital (S.O.), Cambridge; the Department of Diabetes and Metabolic Medicine, St. Bartholomew’s and the Royal London School of Medicine and Dentistry (G.A.H.), London; and the Wellcome Trust Center for Human Genetics, University of Oxford (P.V.S.R., A.J.B., E.C.J., S.M.), Oxford, United Kingdom
Address all correspondence and requests for reprints to: Dr. T. M. Frayling, Department of Diabetes and Vascular Medicine, School of Postgraduate Medicine and Health Sciences, Barrack Road, Exeter, United Kingdom EX2 5AX. E-mail: t.m.frayling{at}exeter.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
(HNF-1) and HNF-4. Two studies examining the HNF-1
region have demonstrated evidence for linkage at genome-wide levels of
significance, whereas four studies examining the HNF-4 locus have
resulted in evidence for linkage at more suggestive levels of
significance. The demonstration of linkage to these regions in
additional patient series will strengthen the evidence that
susceptibility alleles exist at these loci. We therefore assessed the
evidence for linkage to these regions using a large cohort of United
Kingdom Caucasian type 2 diabetes-affected sibling pairs.
A maximum total of 315 affected full sibling pairs were typed for
microsatellite markers across the MODY regions and, in a subset of
families, for markers spanning the whole of chromosome 20. Evidence for
linkage was assessed using a multipoint, mode of inheritance-free
method. Linkage analysis did not reveal any significant evidence for
excess allele sharing at any of the regions studied. Loci contributing
sibling recurrence risks, relative to the general population risk, of
1.75 and 1.25 could be excluded for the HNF-1 and HNF-4 regions,
respectively.
We have not confirmed in United Kingdom Caucasians the evidence for linkage previously reported on 12q and 20q. Our results highlight further the problems of replicating previous positive linkage results across different ethnic groups.
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
(HNF-1) and HNF-4, mutations in which cause
maturity-onset diabetes of the young (MODY) (10, 11), a rare autosomal
dominant form of type 2 diabetes. Secondly, several studies have shown
evidence for linkage at these loci.
Two studies have identified evidence for linkage to the chromosome
12/HNF1 region with genome-wide levels of significance, defined as a
logarithm of the odds (LOD) score of 3.6 or more (12). In the first
study, of large Finnish families with type 2 diabetes, linkage to this
region was only seen after stratification by insulin response to an
oral glucose challenge and the 6 families in the quartile with the
poorest insulin response selected (4). The second study was of a single
large Pacific Island family (6). In addition, evidence of linkage in
non-MODY type 2 diabetes came from 21 U.S.-European Caucasian
families with a history of diabetic nephropathy (maximum LOD score,
1.45) (5). Mutations within the coding region of HNF-1 do not appear
to account for the evidence of linkage (6). Several studies have failed
to identify excess allele sharing in this region, including those
specifically looking for evidence of linkage to the HNF-1 locus (13)
and genome-wide linkage studies (2, 3, 14).
The existence of a susceptibility gene on chromosome 20q is supported
by studies in several different populations, including 43 extended
U.S.-European Caucasian families (maximum nonparametric linkage score,
3.3) (7), 148 French families (maximum LOD score, 1.31) (8), studies of
21 U.S.-European Caucasian families with a history of diabetic
nephropathy (the same families linked to chromosome 12; maximum LOD
score, 1.48) (5), and 477 Finnish families from the Finland and United
States Investigation of Noninsulin-Dependent Diabetes (FUSION; two
peaks with maximum LOD scores of 3.08 and 2.06) (9). The linked regions
identified by these studies all cover broad sections of the chromosome.
All of these peaks, with the exception of one on 20p, include to some
extent the HNF-4 gene on 20q, although variants within the coding
region or 5'-promoter region of this gene are unlikely to account for
the linkage evidence (9, 15). As with the HNF-1 region, several
other studies have failed to find evidence for linkage to the
chromosome 20 regions previously identified (2, 3, 14). Ascertainment
and ethnic differences may account for this apparent discrepancy. Even
in the absence of such differences, however, discrepancies should be
expected when trying to replicate initial linkage results in
multifactorial traits and do not necessarily indicate that positive
findings are simply due to type 1 error (16).
Linkage to these candidate regions in additional type 2 diabetes
patient cohorts will strengthen the evidence that susceptibility genes
exist at these loci. Replication of previous linkage results may also
help narrow susceptibility loci. We therefore sought evidence for
linkage to the HNF-1 and HNF-4 loci in a large cohort of United
Kingdom (UK) Caucasian type 2 diabetes families. In a subset of
families we sought evidence for linkage across the entire length of
chromosome 20.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Families were from the British Diabetic Association-Warren collection, consisting of at least 2 siblings of British/Irish origin with type 2 diabetes diagnosed between 35–75 yr of age. Where affected family members were treated with diet alone, diagnosis was confirmed by historical and/or contemporary glucose measurements. Clinical exclusion of type 1 diabetes, MODY, and maternally inherited diabetes and deafness was made on the basis of personal and family history. In addition, screening for glutamic acid decarboxylase and IA2A autoantibodies revealed that 2.5% (14 of 550) of subjects had titers exceeding the normal range. Measurements of fasting insulin and postglucose challenge glucose and insulin levels were not available.
For the analysis of the HNF-1 and HNF-4 loci we used subjects
from 265 families (635 individuals consisting of 228 affected pairs, 25
3-affected-sibling families, and 2 4-affected-sibling families, 73
unaffected siblings, 12 parents, and 11 affected half-siblings,
resulting in a maximum 315 affected full sibling pairs and 13 affected
half-sibling pairs) in whom sufficient markers had been typed to verify
family relationships (17). Table 1
shows
the clinical details of the affected subjects. For analysis of the full
length of chromosome 20, we used a subset of 78 families (184 affected
individuals consisting of 52 affected pairs, 24 3-sibling-affected
families, 2 4-sibling-affected families, 68 unaffected siblings, and 7
parents; maximum sibling pairs, 136).
|
All 635 individuals from the 265 families in which family
relationships had been confirmed were genotyped using a panel of 11
microsatellite markers. Five markers flanking the HNF-1 gene
(D12S366, D12S321, D12S807, D12S820, and D12S342) and 6 flanking the
HNF-4 gene (D20S170, D20S96, ADA, D20S119, D20S17, and D20S197) were
used. Forward primers were labeled at the 5'-end with fluorescent dyes
to facilitate the pooling of the 11 products into a single ABI-377 gel
lane. PCR was carried out in a 10-µL volume containing 1 x PCR
buffer (Perkin-Elmer Corp., Norwalk, CT), 200 µmol/L of
each deoxy-NTP, 1.5 mmol/L MgCl2, 250 µmol/L of
each primer, 0.2 U AmpliTaq Gold, and 40 ng genomic DNA. Cycling
conditions were 12 min at 96 C followed by 12 cycles of 30 s at 94
C, 60 s at annealing temperature, and 60 s at 72 C followed
by 23 cycles of 30 s at 89 C, 60 s at annealing temperature,
and 60 s at 72 C, finishing with 10 min at 72 C. The annealing
temperature was 55 C, with the exception of D12S807 and D20S96 for
which annealing temperatures of 60 and 58 C were used, respectively.
PCR products were pooled with a ratio of 1:2:1 for
TET:HEX:FAM-labeled products, respectively, before
electrophoresis on an ABI 377 automated sequencer. Alleles were sized
and identified using GENESCAN and GENOTYPER software
(ABI-Perkin-Elmer Corp.).
The following panel of 10 markers spread across the entire length of chromosome 20 was used to genotype the 265 individuals from the subset of 78 families: D20S103, D20S482, D20S851, D20S898, D20S477, D20S107, D20S481, D20S197, D20S480, and D20S171. These markers formed the panel of framework markers used by the Type 2 Diabetes Linkage Consortium (http://www.sph.umich.edu/group/statgen/consortium) and were amplified using the same PCR protocol described above (with an annealing temperature of 55 C used for all markers) and with fluorescent labels designed to facilitate pooling of all products into a single ABI-377 gel lane.
Statistical analyses
For assessment of linkage to candidate loci, we used multipoint
nonparametric linkage analysis using GENEHUNTER software (version 2.0)
(18). All possible sibling pairs were used in the analysis with
sibships of s sibling pairs being weighted by s/2
as an approximate correction for the dependence among siblings. Allele
frequencies were calculated from the whole dataset. For analysis of the
entire length of chromosome 20, genetic map distances estimated from
the Type 2 Diabetes Linkage Consortium were used. For the chromosome
12/HNF-1 and chromosome20/HNF-4 regions, distances from the
Marshfield map (http://www.marshmed.org/genetics/) were used where
available. The markers D12S820, D20S17, and ADA do not appear on
published genetic maps. The marker D12S820 has previously been placed
close to D12S807 (19), and using our data was placed 1.2 cM distal of
D12S807 and 2.6 cM proximal of D12S342 using the function sibmap of the
ASPEX program; this computes multipoint maximum likelihood estimates of
map orders and distances between markers
(ftp://lahmed.stanford.edu/pub/aspex). Markers D20S17 and ADA had
previously been ordered in relation to D20S170, D20S96, D20S119, and
D20S197 on a physical map (20), and sibmap was used to estimate
distances for the intervals including these markers. Resulting overall
map distances (D12S366 to D12S342 for chromosome 12 and D20S170 to
D20S197 for chromosome 20) were consistent with published distances,
and the estimated total number of recombinations, as computed by
GENEHUNTER, was consistent with the order and distances between markers
used. Exclusion mapping was performed using the exclude command of
GENEHUNTER, under an additive model and at several locus-specific
values of 
s (risk to sibling of affected/risk to member of general
population). Exclusion analysis compares the likelihood of the data
arising under specified parameters (in this case s values that
equate directly to the proportion of sibling pairs sharing zero, one,
or two alleles identical by descent) to the likelihood under the null
hypothesis of no linkage (21). Allele sharing is therefore expressed as
a LOD score.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
, HNF-4, and chromosome 20 regions, respectively.
Nonparametric linkage analysis did not reveal any significant increases
in allele sharing across any of the chromosome 12 and 20 regions
analyzed (Fig. 1).
Excluding the 14 affected individuals with glutamic acid
decarboxylase antibody titers exceeding the normal range did not
significantly affect the degree of allele sharing. For the HNF-1
chromosome 12 region, we observed a marginal increase in allele sharing
(maximum nonparametric linkage z-score, 0.44) between markers D12S366
and D12S321, an interval that includes the HNF-1 gene
(Genemap99-http://www.ncbi.nlm.nih.gov/genemap). On the basis of
exclusion mapping we could not exclude (LOD score, less than -2) a
locus with a s of less than 1.75 for this interval.
|
|
region flanked by markers D20S170 and D20S119, we
could exclude a locus of more modest effect at s of 1.25 or more.
For the chromosome 20 region flanked by markers D20S477 and D20S480 (31
cM), a locus with a s of 1.75 or more could be excluded. For the
remainder of the majority of chromosome 20, loci contributing s
values of 2.0 could be excluded, with the exception of the region
between D20S103 and D20S851, a result that probably reflects the
reduced amount of information (due to a less dense spread of
microsatellite markers) extracted from this region. |
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
region and across
chromosome 20. This contrasts with a number of recent reports, mainly
in Caucasian cohorts, describing linkage to these regions. Our study
represents the second largest cohort used to investigate these
candidate loci; the FUSION study is larger for the chromosome
20/HNF-4 region (9), and the study of Lesage et
al. (13) is larger for the chromosome 12/HNF-1
region.
In assessing the significance of our results, it is important to
consider the power available to exclude loci contributing specified
increases in relative risk. In this study we used the risk to a sibling
compared to the general population risk, s. Overall s values for
type 2 diabetes are estimated to range from 3–4 (22, 23). Given the
multiple genes and environmental factors that may be contributing to
this figure, it is unlikely that any single locus is contributing more
than a s of 2.0. The peak close to the HNF-4 gene identified by
the FUSION study (
3 cM proximal of D20S170) was estimated to have a
s of 1.25 (9). Although we could not formally exclude (at LOD less
than -2) this point, a LOD score of -1.74 at s = 1.25
suggests that we had good power to replicate increased allele sharing
at this locus, and further explanation is needed to explain the
apparent discrepancy between the two studies. There are several other
possible reasons why we have not reproduced the evidence for linkage in
the HNF-4 region besides the possibility that our result is a false
negative or the FUSION result is a false positive. Ethnic
heterogeneity, such as possible differences in susceptibility allele
frequencies, ascertainment, and phenotypic differences, may all result
in apparent discrepancies between studies. Age at diagnosis and BMI are
similar between the two groups, and both studies consist primarily of
affected sibling pairs; ethnic heterogeneity, therefore, represents the
most likely explanation for the difference.
For the regions of chromosome 20 outside of the HNF-4 region
(D20S477 to D20S480), we were unable to exclude loci with relatively
large contributions (s values of 1.75–2.0) to type 2 diabetes
susceptibility. The lack of significant linkage across the remainder of
chromosome 20 may therefore represent a lack of power to detect loci of
relatively weak effect. Alternatively, the lack of any evidence for
linkage could be due to any of the above-stated reasons, including the
possibility that previous positive results are due to type 1 error.
For the region containing the HNF-1 gene we were unable to
exclude loci at s values of less than 1.75. The modest increase in
allele sharing between D12S366 and D12S321 may therefore reflect the
presence of a susceptibility gene contributing a s value of less
than 1.75. With the present cohort, however, we cannot determine
whether this is the case or the increase in allele sharing is due to
stochastic variation. Previous studies that identified linkage in this
region did not estimate s values due to the highly selected nature
of the families studied. In contrast to a number of the previous
studies, we have not stratified results according to any intermediate
traits. Mahtani et al. identified evidence for linkage to
the HNF-1 region only when stratifying families according to glucose
tolerance and selecting the quartile of families with the poorest
insulin response (4). In addition, Bowden et al. identified
linkage to the HNF-1 region using families with a history of
nephropathy (5). Measures of complications and glucose tolerance were
not available in our subjects, and it remains a possibility that
significant linkage would be identified if subsets of families were
used.
In summary, we have not provided any further evidence that type 2
diabetes susceptibility loci exist on chromosomes 12 and 20. Our study
provides further indication that ethnic differences occur in the
genetic etiology of type 2 diabetes. It is unlikely that genes
contributing a relative sibling risk of greater than 1.75 and 1.25 to
type 2 diabetes susceptibility exist at the HNF-1 and HNF-4 loci,
respectively, in our cohort of UK patients.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received September 17, 1999.
Revised November 2, 1999.
Accepted November 8, 1999.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
gene in maturity-onset
diabetes of the young (MODY3). Nature. 384:455–458.[CrossRef][Medline]
gene in maturity-onset diabetes
of the young (MODY1). Nature. 384:458–460.[CrossRef][Medline]
This article has been cited by other articles:
 |
|
 |
|
  A. L. Gloyn, F. Reimann, C. Girard, E. L. Edghill, P. Proks, E. R. Pearson, I. K. Temple, D. J.G. Mackay, J. P.H. Shield, D. Freedenberg, et al. Relapsing diabetes can result from moderately activating mutations in KCNJ11 Hum. Mol. Genet., April 1, 2005; 14(7): 925 - 934. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. N. Weedon, K. R. Owen, B. Shields, G. Hitman, M. Walker, M. I. McCarthy, L. D. Love-Gregory, M. A. Permutt, A. T. Hattersley, and T. M. Frayling Common Variants of the Hepatocyte Nuclear Factor-4{alpha} P2 Promoter Are Associated With Type 2 Diabetes in the U.K. Population Diabetes, November 1, 2004; 53(11): 3002 - 3006. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. van Tilburg, T. W van Haeften, P. Pearson, and C. Wijmenga Defining the genetic contribution of type 2 diabetes mellitus J. Med. Genet., September 1, 2001; 38(9): 569 - 578. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
