Evidence by allelic association-dependent methods for a type 1 diabetes polygene (IDDM6) on chromosome 18q21
Evidence by allelic association-dependent methods for a type 1 diabetes polygene ( IDDM6 ) on chromosome 18q21Tony Merriman, Rebecca Twells, Marilyn Merriman, Iain Eaves, Roger Cox, Francesco Cucca, Patricia McKinney1, Julian Shield2, David Baum2, Emanuele Bosi3, Paolo Pozzilli4, Lorenza Nisticò4, Raffaella Buzzetti4, Geir Joner5, Kjersti Rønningen6, Erik Thorsby7, Dag Undlien7, Flemming Pociot8, Jørn Nerup8, Stephen Bain9, Anthony Barnett9 and John Todd*
The Wellcome Trust Centre for Human Genetics, Nuffield Department of Surgery, University of Oxford, Windmill Road, Headington, Oxford OX3 7BN, UK, 1Paediatric Epidemiology Group, Research School of Medicine, University of Leeds, Leeds LS2 9LN, UK, 2Institute of Child Health, University of Bristol, Royal Hospital for Sick Children, Bristol BS2 8BJ, UK, 3Department of Internal Medicine, Istituto Scientifico San Raffaele, University of Milan, Milan, Italy, 4Endocrinologia, Istituto Clinica Medica II, University of Rome `La Sapienza', Viale de Policlinico, 151, 00161 Rome, Italy, 5Aker Diabetes Research Centre, Aker University Hospital, Oslo, Norway, 6Department of Population Health Sciences, National Institute of Public Health, PO Box 4404, N-0403 Oslo, Norway, 7Institute of Transplantation Immunology, The National Hospital, Oslo, Norway, 8Steno Diabetes Center, DK-2820, Gentofte, Denmark and 9Department of Medicine, University of Birmingham, Birmingham Heartlands Hospital, Birmingham B9 5SS, UK
Received February 14, 1997;Revised and Accepted April 23, 1997
Type 1 diabetes is a common polygenic disease. Fine mapping of polygenes by affected sibpair linkage analysis is not practical and allelic association or linkage disequilibrium mapping will have to be employed to attempt to detect founder chromosomes. Given prior evidence of linkage of the Jk-D18S64 region of chromosome 18q12-q21 to type 1 diabetes, we evaluated the 12 informative microsatellite markers in the region for linkage with disease by the transmission disequilibrium test (TDT) in a UK data set of type 1 diabetic families (n = 195). Increased transmission of allele 4 of marker D18S487 to affected children was detected (P = 0.02). Support for this was extended in a total of 1067 families from four different countries by isolating, and evaluating by the TDT, two novel microsatellites within 70 kb of D18S487. Evidence for linkage and association was P = 5 * 10-5 and 3 * 10-4, respectively. There was no evidence for increased transmission of associated alleles to nonaffected siblings. Analysis of an additional 390 families by the TDT did not extend the evidence further, and reduced support in the total 1457 families to P = 0.001 for linkage and P = 0.003 for association. However, evidence for linkage by affected sibpair allele sharing was strong (P = 3.2 * 10-5) in the second data set. Heterogeneity in TDT results between data sets was, in part, accounted for by the presence of more than one common disease-associated haplotype (allelic heterogeneity) which confounds the analysis of individual alleles by the TDT. Guidelines for strategies for the mapping of polygenes are suggested with the emphasis on collections of large numbers of families from multiple populations that should be as genetically homogeneous as possible.
Whole genome scanning has shown that type 1, or insulin-dependent, diabetes mellitus is a polygenic disease with the major locus (IDDM1) in the major histocompatibility complex on chromosome 6p21 (1 ,2 ). Aetiological mutations of IDDM1 and IDDM2 have been identified as, respectively, allelic variants of the peptide binding sites of the HLA-DQ and -DR molecules (3 ), and a variable number of tandem repeats sequence in the 5' regulatory region of the insulin gene on chromosome 11p15 (4 ,5 ). The localisation of these genes and the CTLA-4 region (IDDM12) (6 ), was made possible only through their selection as functional candidate genes. None of the other three confirmed loci (IDDM4, IDDM5 and IDDM8) (7 ,8 ) detected by genome linkage scanning (1 ,2 ) has been fine-mapped. To map a typical polygene, which has a similar effect in families as IDDM2 ([lambda]s <= 1.25) for example, using standard linkage analysis to a region of <1 cM would require 9250 affected sibpairs (9 ). Moreover, linkage to a region may be observed in one data set but it may take several more data sets to observe the same linkage if there are several genes acting to cause the disease (10 ). Here we aimed to use linkage disequilibrium to provide evidence for a type 1 diabetes polygene on chromosome 18q12-q21. This strategy has been productive in positional cloning of highly penetrant monogenic disease loci in genetically isolated populations such as in Finland (11 -13 ), in which most present day cases are related to the original case and remnants of the original founder chromosome can easily be detected in modern-day populations. In these situations measures of linkage disequilibrium such as [delta] or Pexcess (13 ) can exceed 0.6 (maximum 1.0). The question addressed here is whether, in the absence of functionally relevant candidate genes, a similar approach can be applied to a polygenic disease caused by multiple low penetrance susceptibility genes in populations of mixed ancestry in which disease predisposing allele frequencies may be high (>50%).
Thirteen dinucleotide repeat microsatellites flanking and including D18S64 were genotyped in 195 UK multiplex families (which consisted of the 93 families originally typed for D18S64 and a further 102 families available at the time) (Materials and Methods) (Fig. 1 ). All markers except D18S57 were tested by the TDT for linkage in the presence of association. Transmission of only the three most common alleles at each locus from heterozygous parents to affected offspring was examined in order to reduce the problem of multiple testing (19 ) (the three most common alleles at each of the 12 markers accounted for an average 75% of the total alleles). Of the 36 microsatellite alleles tested, only allele 4 of D18S487 showed increased transmission at a 0.05 level of significance [192 transmissions (T) versus 150 nontransmissions (NT); TDT = 5.2, %T = 56.1, P = 0.02].
Linkage of the 13 markers to type 1 diabetes was also assessed by affected sibpair allele sharing (Fig. 1 ). Although D18S487 was not at the peak of linkage and had an MLS of only 0.27 (Fig. 1 ; P > 0.05), it was in the region of increased identical-by-descent allele sharing (MMLS = 0.75, P = 0.05). This was consistent with the possibility that D18S487 was in tight linkage with a type 1 diabetes susceptibility gene. In a polygenic disease recombination cannot be relied upon to rule out the location of a gene since less allele sharing could reflect the fact that an affected individual did not require that gene for disease development.
D18S487 was then typed in a 133 family US multiplex data set and a largely simplex Sardinian (SD) data set of 181 families (Materials and Methods). The TDT was applied specifically and only to allele 4 of D18S487 in these data sets. Positive transmission of allele 4 of D18S487 in the US (96T versus 78N;, TDT = 1.9, %T = 55.2, P = 0.17) and SD (60T versus 48NT; TDT = 1.3, %T = 55.6, P = 0.25) data setswas observed which increased the overall evidence of linkage. The total for the three data sets (n = 509 families) was 348T versus 276 NT (TDT = 8.3, %T = 55.8, P = 0.004). Importantly, transmission of allele 4 was neutral to healthy siblings within the total data set (101T versus 98NT; TDT = 0.05, %T = 50.8).
. Linkage disequilibrium of D18S487, A181,2 and IO43,56 with type 1 diabetes in the UK, US, SD and NOR data sets (1067 families)
D18S487
A181,2
IO43,56
T
NT allele 4
%T
Freq
T allele 2
NT
%T
Freq allele 2
T
NT
%T
Freq
UK multiplex
A
264
221a
54.4
0.25
187
138b
57.5
0.15
148
124
54.4
0.12
NA
44
36
21
17
19
21
US multiplex
A
96
78
55.2
0.21
77
59
56.6
0.14
46
42
52.3
0.09
NA
26
17
13
13
10
5
Sardinia
A
60
48
55.6
0.16
36
20c
64.3
0.08
33
24
57.9
0.08
NA
36
48
17
28
16
22
Norway
A
133
123
52.0
0.20
90
68
57.0
0.10
80
66
54.5
0.10
NA
150
160
88
86
s79
95
Total
A
553
470*
54.1
21.0
390
285d
57.8
11.7
307
256d
54.5
10.0
NA
256
261
139
144
124
143
A, affected; NA, non-affected; T, transmitted; NT, not transmitted. %T, percent transmission (Materials and Methods).
aP = 0.05;
bP = 0.007;
cP = 0.03 dTDT [chi]2 (1 df) values and P values for transmission of associated alleles of D18S487, A181,2 and IO43,56 to affected offspring in the combined data sets are 6.7 (P = 0.009), 16.3 (P = 0.00005) and 4.6 (P = 0.03) respectively.
It was possible that D18S487 was some distance from the putative type 1 diabetes susceptibility gene and that cloning, genotyping and analysis by the TDT of microsatellite markers in closer proximity would help substantiate contribution of the region to disease. A bacterial artificial chromosome (BAC) clone positive for D18S487 was identified (HBAC 22_d_8; Materials and Methods) and two polymorphic microsatellite markers (A181,2 and IO43,56, on either side of D18S487 by 60 and 70 kb respectively) were isolated using an efficient PCR-based method (Materials and Methods). D18S487, A181,2 and IO43,56 were analysed in the 509 families plus a further 558 families available at the time: 138 UK affected sibpair families and a 420 family Norwegian (NOR) data set (Table 1 ) (Materials and Methods). Importantly, only those allele(s) of A181,2 and IO43,56 in linkage disequilibrium with allele 4 of D18S487 were evaluated for linkage with type 1 diabetes by the TDT (Fig. 2 ). Allele 2 of both A181,2 and IO43,56 had D' values supporting linkage disequilibrium with allele 4 of D18S487 of at least 0.73 in each of the four populations. Overall, by the TDT allele 2 of A181,2 was in stronger linkage with type 1 diabetes in the 1067 family combined data set (TDT = 16.3, %T = 57.8, P = 5 * 10-5) than both D18S487 (TDT = 6.7, %T = 54.1, P = 0.009) and IO43,56 (TDT = 4.6, %T = 54.5, P = 0.03). By analysing transmission to probands only (Materials and Methods) there was also evidence for association of allele 2 of A181,2 with disease (250T versus 175NT; TDT = 13.2, Pproband = 3 * 10-4). The transmission of allele 2 of A181,2 to unaffected siblings was close to neutral (139T versus 144NT).
Involvement of the D18S487 region in type 1 diabetes susceptibility was assessed in an additional six data sets totalling 390 families. The data sets were: 56 multiplex families from the US, 66 multiplex families from the UK, 32 simplex families from South-West UK, 80 simplex families from the County of Yorkshire (UK), 108 families from Continental Italy (of which 46 were multiplex) and 48 multiplex families from Denmark (Methods). Transmission of allele 2 of A181,2 to affected children was specifically evaluated in these families. In none of the data sets was there any increased transmission of this allele (23T versus 29NT, 25T versus 21NT, 6T versus 3NT, 11T versus 25NT, 24T versus 21NT and 22T versus 19NT respectively: total 111T versus 118NT). When added to the equivalent data from the initial 1067 families (a total of 1457 families) there was still significant transmission of allele 2 of A181,2 to diabetic children (TDT = 10.6, 501T versus 403NT, P = 0.001, Pproband = 0.003).
Of the six additional data sets, four contained multiplex families-the UK, US, Italian and Danish data sets (n = 216). In three of the four data sets there was positive evidence for linkage (Table 2 ), with a total MLS for the 216 families of 3.7 (P = 3.2 * 10-5). This suggested that the D18S487 region may be contributing to diabetes susceptibility in these data sets. Therefore, it was hypothesised that the lack of evidence by the TDT for allelic association of allele 2 of A181,2 with type 1 diabetes in the second 390 family data set could be explained by the presence of additional diabetogenic haplotype(s) not containing allele 2 of A181,2 which were confounding analysis by the TDT of this allele. To test this hypothesis, transmission of the five most common haplotypes of markers A181,2-D18S487- IO43,56: 5-3-8 (9.2% frequency in parents in the 1457 families), 2-4-2 (5.8%), 5-8-10 (4.7%), 5-8-11 (4.4%) and 2-4-3 (4.4%), from heterozygous parents to affected offspring was analysed in the initial 1067 families, the second 390 families and the entire 1457 families (Table 3 ). In addition to the expected positive transmission of the 2-4-2 haplotype in the 1067 families (P = 0.001) the most common haplotype, 5-3-8, was also positively transmitted (P = 0.05). In the second 390 families, however, no haplotypes were significantly positively transmitted. Notably, in the 390 family data set, the 2-4-2 haplotype tended to be negatively transmitted (%T = 43.5), which would help account for the failure to observe positive transmission of allele 2 of A181,2 in this data set. In the combined 1457 families three of the five most common haplotypes were positively transmitted to affected children: the 5-3-8 haplotype (P = 0.04), 2-4-2 haplotype (P = 0.03) and the 2-4-3 haplotype (P = 0.05). Transmission of the other two common haplotypes (5-8-10 and 5-8-11) did not deviate significantly from neutral in either the 1067 families or 390 families or total 1457 families (data for the 1457 families was 188T versus 173NT and 183T versus 165NT respectively).
. Summary of linkage of the D18S487 region to type 1 diabetes in five data sets. The three markers typed (A181,2, D18S487 and IO43,56) were combined to make the locus informative in 96% of meioses
IBD sharing
IBD sharing
n
2
1
0
1
0
MLS
P
[lambda]s
UKa
399
116
180
88
416
361
1.1
0.02
1.09
USb
189
42
93
49
179
193
0
1.00
Norway
40
13
18
3
45
24
1.6
0.005
1.78
C Italy
46
14
21
3
49
28
1.5
0.006
2.63
Denmark
48
16
16
10
49
36
0.8
0.04
1.20
390 data set
216
65
101
25
235
157
3.7
3.2 * 10-5
1.74
Total
722
201
328
153
738
642
1.6
0.005
1.11
MLS, maximum lod score; [lambda]s, contribution to familial clustering (21).aData for the second set of UK families analysed (Materials and Methods) (n = 66) were 23,31,4 (2,1,0 IBD sharing), 80,42 (1,0 IBD sharing), MLS = 3.0, P = 0.0002, [lambda]s = 2.16. bData for the second set of US families analysed (n = 56) were 12,33,8 (2,1,0 IBD sharing), 57,51 (1,0 IBD sharing), MLS = 0.1, [lambda]s = 1.17.Allele sharing linkage data for the multiplex families (n = 216) within the second data set of 390 families (see text) is also shown: this subset of the total 722 families comprises the second UK and US data sets (above) and the C Italian and Danish data sets.
. Transmission of haplotypes 5-3-8, 2-4-2 and 2-4-3 of markers A181,2, D18S487 and IO43,56 respectively
5-3-8
2-4-2
2-4-3
T
N
%T
Freq
T
N
%T
Freq
T
N
%T
Freq
1067 families
278
234a
54.3
9.4
202
141b
58.9
6.2
130
109
54.4
4.4
390 families
93
82
53.1
8.4
47
61
43.5
4.3
42
29
59.2
4.4
UKc
143
129
52.6
9.4
115
78d
59.6
7.7
74
61
54.8
5.4
US
64
59
52.0
9.7
50
42
54.3
7.2
25
25
50.0
3.3
Sardinia
23
14
62.2
6.7
17
7e
70.8
3.1
13
6
68.4
3.4
Norway
92
80
53.5
10.0
47
35
57.3
5.2
34
30
53.1
3.7
Yorkshire (UK)
22
7f
75.9
12.0
8
14
36.4
7.1
3
11g
21.4
6.2
C Italy
17
15
53.1
7.1
5
14e
26.3
2.2
13
3h
81.3
5.4
Denmark
10
12
45.5
7.1
7
12
36.8
5.5
10
2i
83.3
4.7
Total
371
316j
54.0
9.2
249
202j
55.2
5.8
172
138j
55.5
4.4
In the top two lines transmissions to the two separate large data sets (the initial 1067 families and second 390 families) are presented, whereas in the remainder of the Table transmissions are presented to the seven constituent populations (Materials and Methods). aP = 0.05; bP = 0.001; cconsists of the 399 multiplex families and 32 simplex families from the South-West of the UK; dP = 0.008; eP = 0.04; fP = 0.005; gP = 0.03; hP = 0.01; iP = 0.02; jTotal P values for transmission of the haplotypes are 0.04, 0.03 and 0.05 respectively.
Transmission of haplotypes 5-3-8, 2-4-2 and 2-4-3 was then examined in the data sets pooled according to region or country of origin (Table 3 ). Positive transmission was observed for the 2-4-2 haplotype in the UK multiplex and Sardinian data sets (P = 0.008 and 0.04 respectively), the 5-3-8 haplotype in the Yorkshire simplex data set (P = 0.005) and the 2-4-3 haplotype in the Italian and Danish data sets (P = 0.01 and 0.02 respectively). Negative transmission was observed only for the 2-4-3 haplotype in the Yorkshire data set (P = 0.03) and the 2-4-2 haplotype in the Italian data set (P = 0.04).
We found evidence by the TDT for linkage (P = 5 * 10-5) and association (Pproband = 3 * 10-4) of allele 2 of microsatellite marker A181,2 with a type 1 diabetes susceptibility locus in a large data set of 1067 families drawn from four separate populations. However, this was not replicated in a second sample of 390 families drawn from six separate populations, so that statistical support obtained by the TDT for linkage and association in the combined 1457 families was reduced (P = 0.001, Pproband = 0.003 respectively). The appropriate level of statistical support required before claiming linkage disequilibrium of a positionally defined locus with a polygenic disease is uncertain. A level of significance of P = 5 * 10-8 (22 ) has been suggested for a genome-wide linkage disequilibrium search using one million markers. In our case, however, we had a prior hypothesis of linkage and linkage data in Table 2 (MLS = 1.6, P = 0.005 for the D18S487 region in 722 families) provides prior support for the existence of a type 1 diabetes susceptibility locus on chromosome 18q21, with four out of the five data sets tested here showing linkage to disease. Therefore, to take account of multiple testing, we corrected by a factor of 36 (the number of alleles initially tested in the 195 UK families for linkage disequilibrium with type 1 diabetes), giving a corrected P = 0.002 (corrected Pproband = 0.01) for the initial 1067 families and a corrected P = 0.036 (corrected Pproband = 0.11) for the total 1457 families. It is emphasised that transmission of allele 2 of A181,2 was neutral to unaffected siblings (Table 1 ; 139T versus 144NT). Furthermore, the presence of a type 1 diabetes polygene in the D18S487 region is also supported by positive transmission to affected children of three of the five most common haplotypes (as defined by A181,2, D18S487 and IO43,56) in the 1457 families (Table 3 ). Therefore, we have provisionally designated the type 1 diabetes susceptibility locus in the D18S487 region IDDM6.
Positive transmission of three of the five most common haplotypes to diabetic children in the total 1457 families (Table 3 ) is evidence consistent with IDDM6 being present on more than one distinct founder chromosome and this may have confounded the TDT analysis in the 390 family data set. However, it should be noted that there is a possibility that the significant positive transmission of one of the haplotypes in the 1457 families could be due to type 1 error. Investigation of transmission of the 5-3-8/2-4-2/2-4-3 haplotypes in additional large data sets will be required to determine whether these haplotype results are true or false. If there are multiple founder chromosomes (i.e., allelic heterogeneity) then estimates, based on the assumption of a single disease allele, of the number of families required for fine-mapping polygenes (22 ) will be too small.
The positive transmission of the 5-3-8, 2-4-2 and 2-4-3 haplotypes could reflect allelic heterogeneity at IDDM6 where each haplotype could be in linkage disequilibrium with distinct aetiological mutations at IDDM6. Alternatively, the haplotypes could be in linkage disequilibrium with the same IDDM6 aetiological mutation and appear as distinct founder chromosomes in the modern day type 1 diabetic population for two reasons. First, they could have evolved from a single founder chromosome by microsatellite mutation (for example, the 2-4-2 and 2-4-3 haplotypes differ apparently by only one microsatellite mutation event between allele 2 and 3 of the third marker). This could be tested by typing the 130 kb region defined by the three marker haplotype with a set of more stable, single nucleotide polymorphisms. A second reason is that the 130 kb region could be some distance from IDDM6 and the distinct founder chromosomes may have resulted from ancestral recombination events with IDDM6. This can be tested by cloning further microsatellite markers flanking the 130 kb region (up to 1 Mb on either side would probably be sufficient) and genotyping the diabetic families. The separate positively transmitted haplotypes should be in linkage disequilibrium with the same allele of markers spanning IDDM6.
In the total data set of 1457 families none of the individual alleles of the 5-3-8 haplotype is significantly associated with type 1 diabetes by the TDT, and of the 2-4-2 and 2-4-3 haplotypes, only allele 2 of A181,2 and allele 4 of D18S487 were significantly associated with disease (TDT = 10.6, P = 0.001, %T = 55.4 and TDT = 4.6, P = 0.03, %T = 52.9, respectively). Allele 2 of A181,2 is present at a frequency of 11.5% in parents in the 1457 families, allele 5 at 74.3%, allele 3 of D18S487 at 27.0%, allele 4 at 20.7%, allele 2 of IO43,56 at 9.8%, allele 3 at 12.5% and allele 8 at 21.6%. Thus a possible reason why allele 2 of A181,2 and allele 4 of D18S487 were the only significant alleles could be due to them being present on both the 2-4-2 and 2-4-3 haplotypes. Allele 2 of A181,2 was in stronger linkage disequilibrium with IDDM6 than allele 4 of D18S487 in the 1457 families. This could be because allele 2 of A181,2 has a relatively low frequency (11.5%), thus lessening the chances of it being on other non-associated haplotypes, whereas allele 4 of D18S487 has a higher frequency (20.7%) thus increasing the chances that it is on other non-associated haplotypes. Furthermore, it is possible that allele 2 of A181,2 arose at a similar time to the IDDM6 mutation it `marks', thus being a more specific marker of the IDDM6 founder chromosome. The age and frequency of marker alleles are key parameters in assessing linkage disequilibrium with disease (23 ). The strength of association of a marker allele with disease is a function of the relative frequency of the allele on associated haplotype(s) versus non-associated haplotypes. This conclusion has also been drawn from analysis of association of MHC marker alleles with disease (35 ).
Genome-wide linkage scanning of multifactorial diseases is revealing that many of them have a polygenic basis (1 ,2 ,24 ). Combined with locus, allelic and clinical heterogeneity the challenge to actually identify polygenes is enormous. Based on the results from IDDM6 presented here, we can make some recommendations for strategies. Evidence of linkage in affected sibpair families (the most common pedigree configuration in common polygenic diseases) should be sought in several different populations. The ideal family data sets are those drawn from isolated populations (Sardinia, Finland) or from countries such as Denmark, Sweden, Spain, Italy or Norway which are more homogeneous than those of the USA or UK. Families can be collected from defined regions within countries (for example, the 80 simplex families from the Yorkshire region used here), again optimising the chances that the families have a common ancestry. If only one data set shows positive evidence of linkage to a chromosome region, evidence of replication should be sought by collecting more families from the same country. Given consistent evidence of linkage, linkage disequilibrium should be evaluated specifically in the linked region using all available microsatellites with an appropriate level of polymorphism (Materials and Methods). It is necessary to assume the presence of multiple founder chromosomes and therefore haplotype analysis is advised, but this means that the numbers of families required will be large. Large numbers of simplex families can be collected from a specific region. Haplotype analysis will also help alleviate the problem of variation in estimates of linkage disequilibrium which arise from variation in the frequency and age of alleles between markers. [Here, %T of the 5-3-8 and 2-4-2 haplotypes was very similar (Table 3 ) whereas transmission of the individual alleles differed widely.] Hopefully by defining different founder chromosomes in different populations, regions of strongest and most consistent linkage disequilibrium with multiple markers can be defined. Such regions can then be subjected to cloning, physical mapping, genomic DNA sequencing, gene characterisation and identification of the polymorphisms associated with genes, particularly those found more frequently on disease-associated haplotypes.
All families used in this study were Caucasian with at least one affected sibling per family and both parents included. The initial 333 multiplex families recruited from the UK were restricted to Caucasians with grandparents born in the UK, and were affected sibpair families in which at least one sibling was diagnosed with type 1 diabetes under age 17 years and the other under age 29 years (25 ). The 189 US multiplex families were selected according to the same age-of-onset criterion and were from the Human Biological Data Interchange (HBDI) repository of type 1 diabetic families (26 ). An additional 66 multiplex families from the UK had at least one affected sibling diagnosed under the age of 29 years, and families with more than two affected siblings were included in this data set (25 ). There were 80 simplex families collected from the Yorkshire region of the UK, and a further 32 families collected from the South West region of the UK, and in all cases the single affected sibling was diagnosed under the age of 17 years. The Sardinian data set comprised 175 simplex families and six multiplex families with all diabetics diagnosed under the age of 17 years. The Norwegian data set comprised 380 simplex families with all cases diagnosed under the age of 15 years, and 40 multiplex families with at least one sibling diagnosed under age 23 years and the other under age 36 years. The Continental Italian data set comprised 62 simplex families and 46 multiplex families and all cases were diagnosed under the age of 29 years. The 48 Danish multiplex families are described elsewhere (27 ).
The 13 markers initially used in this study for linkage analysis and preliminary linkage disequilibrium detection were obtained from public databases and were all dinucleotide repeats. The microsatellites were typed on the diabetic families using methods already described (28 ). MLS values (21 ,29 ) were calculated using the MAPMAKER/SIBS program (30 ). Where families contained more than two affected children, only the first two diagnosed were tested for linkage. Theoretical P values were assigned to the MLS scores (29 ).
The TDT, which is a test of linkage in the presence of association, was used to assess linkage disequilibrium between common marker alleles and disease (18 ). This test is family based so it distinguishes between association due to linkage and association that may arise in the absence of linkage, such as that due to population stratification. The test statistic (the `TDT') is a [chi]2 (1df) statistic, which tests deviation of transmission from the expected 50% transmission of an allele from heterozygous parents to offspring. Extent of linkage disequilibrium of a marker allele with disease was quantitated by percent transmission (%T). %T = number of times an allele is transmitted from heterozygous parents to affected children divided by the total number of transmissions, expressed as a percentage. We chose the TDT because measures of linkage disequilibrium such as [delta] (or Pexcess) (13 ,20 ) are not informative because associated marker allele frequencies are too high in the general population. For example, the [delta] value for allele 2 of A181,2 in the UK population is 0.04 ([delta] varies between 0, no association; and 1, complete association). In the TDT we analysed transmissions to the first two diagnosed individuals in multiplex families. Because meioses are not strictly independent between affected siblings showing linkage to disease, this will introduce some bias into the TDT statistic (31 ). However, the degree of this bias is uncertain and, in the case of IDDM6 where increased allele sharing in sibpairs is modest[MLS = 1.6, [lambda]s = 1.11 (Table 2 )], the bias may be minimal-not enough to warrant discarding a large proportion of the available data (50% of families were multiplex). Therefore, P values quoted are derived from transmissions to both affected siblings in multiplex families and are reported as evidence for linkage in the presence of association, and P values where transmission has been analysed to the probands only (Pproband) are quoted as support for allelic association of a marker: this is a completely valid use of the TDT as a test of allelic association or linkage disequilibrium (31 ).
HBAC 22_d_8 (containing D18S487) was isolated from a library obtained from Research Genetics, Inc. Microsatellites were cloned from the BAC clone using a method, termed `microsatellite rescue', developed here but also reported independently (32 ). A vectorette library was made from the clone (33 ) and PCR was carried out using the Not1-A and (AC)11N (where N is not A) primers at an annealing temperature of 65oC. PCR products were gel purified and cycle sequenced using a PRISMTM Ready Reaction DyeDeoxyTM Terminator Cycle Sequencing Kit (Perkin Elmer) with the manufacturer's instructions. Sequence data was collected on a Model 373A Sequencer (ABI). From the sequence a unique primer was designed such that the 3' end was oriented towards the dinucleotide repeat. The opposite side of the dinucleotide repeat was cloned by performing PCR on the vectorette libraries using the newly designed unique primer and the Not1-A primer, and sequencing the product as before with both primers. The sequences from either side of the microsatellite were merged and PCR primers designed for genotyping. Microsatellites were genotyped over 24 families to determine their polymorphism. Four microsatellites, in addition to D18S487, were cloned from HBAC 22_d_8. One had a major allele with a frequency >0.80, and another had >25 alleles. Both were unsuitable, therefore, for linkage disequilibrium mapping. Genotyping of markers that are insufficiently polymorphic (major allele frequency >0.80) and markers that are too polymorphic (>15 alleles-this can be indicative of a higher mutation rate and large numbers of alleles can lead to ambiguities in defining allele sizes) is inefficient for linkage disequilibrium mapping. The other two (A181,2 and IO43,56) were considered suitable for linkage disequilibrium mapping. The newly developed microsatellites were positioned on HBAC 22_d_8 by standard restriction enzyme mapping techniques. Primer sequences for amplifying A181,2 are TCTCCTCTTTCAGAGACGCTG and TTAACCCCACAATAGCCTTACG and for amplifying IO43,56 are GCCCCCAACTTCATTTCTTT and CAAAGGCCAAAGGGACATAT.
We thank L. Pritchard, S. Bennett, W. Ewens, A. Wilson, S. Jenkins, Z. Atta, B. Rowe, J. Carr-Smith and C. Smyth for their help and contribution to this work. We thank the IMDIAB study group physicians involved in collecting the Italian families. This work was funded by the Wellcome Trust, British Diabetic Association, UK Medical Research Council, Juvenile Diabetes Foundation International (JDFI), Lilly Industries UK, Telethon-Italy and The Norwegian Diabetes Association. The BDA, HBDI and The Norwegian Study Group for Childhood Diabetes are thanked for the collection of families. T.R.M. was a Wellcome Trust-New Zealand Health Research Council Overseas Postdoctoral Fellow and is currently a JDFI Postdoctoral Fellow, I.A.E is the recipient of a Wellcome Trust Prize Studentship and J.A.T. is a Wellcome Trust Principal Research Fellow.
1 Davies, J.L., Kawaguchi, Y., Bennett, S.T., Copeman, J.B., Cordell, H.J., Pritchard, L.E., Reed, P.W., Gough, S.C.L., Jenkins, S.C., Palmer, S.M., Balfour, K.M., Rowe, B., Farrall, M., Barnett, A.H., Bain, S.C. and Todd, J.A. (1994) A genome-wide search for human type 1 diabetes susceptibility genes. Nature371, 130-136.MEDLINE Abstract
2 Hashimoto, L., Habita, C., Beressi, J.P., Delepine, M., Besse, C., Cambon-Thomsen, A., Deschamps, I., Rotter, J.I., Djoulah, S., James, M.R., Froguel, P., Weissenbach, J., Lathrop, G.M. and Julier, C. (1994) Genetic mapping of a susceptibility locus for insulin-dependent diabetes mellitus on chromosome 11q. Nature371, 161-164.MEDLINE Abstract
3 Cucca, F. and Todd, J.A. (1996) in Browning, M.J. and McMichael, A.J. (eds), HLA/MHC: Genes, Molecules and Function. bIOS Scientific Publishers, Oxford, pp. 383-406.
4 Bell, G.I., Horita, S. and Karam, F.H. (1984) A polymorphic locus near the human insulin gene is associated with insulin-dependent diabetes mellitus. Diabetes 33, 176-183. MEDLINE Abstract
5 Bennett, S.T., Lucassen, A.M., Gough, S.C.L., Powell, E.E., Undlien, D.E., Pritchard, L.E., Merriman, M.E., Kawaguchi, Y., Dronsfield, M., Pociot, F., Nerup, J., Bouzekri, N., Cambon-Thomsen, A., Rønningen, K.S., Barnett, A.H., Bain, S.C. and Todd, J.A. (1995) Susceptibility to human type 1 diabetes at IDDM2 is determined by tandem repeat variation at the insulin gene minisatellite locus. Nature Genet. 9,284-292.MEDLINE Abstract
6 Nisticò, L., Buzzetti, R., Pritchard, L.E., Van der Auwera, B., Giovannini, C., Bosi, E., Larrad, M.T.M., Rios, M.S., Chow, C.C., Cockram, C.S., Jacobs, K., Mijovic, C., Bain, S.C., Barnett, A.H., Vandewalle, C.L., Schuit, F., Gorus, F.K., Belgian Diabetes Registry, Tosi, R., Pozzilli, P. and Todd, J.A. (1996) The CTLA-4 gene region of chromosome 2q33 is linked to, and associated with, type 1 diabetes. Hum. Mol. Genet. 5, 1075-1080.MEDLINE Abstract
7 Davies, J.L., Cucca, F., Goy, J.V., Atta, Z.A.A., Merriman, M.E., Wilson, A., Barnett, A.H., Bain, S.C. and Todd, J.A. (1996) Saturation multipoint linkage mapping of chromosome 6q in type 1 diabetes. Hum. Mol. Genet. 5, 1071-1074.MEDLINE Abstract
8 Luo, D.F., Buzzetti, R., Rotter, J.I., Maclaren, N.K., Raffel, L.J., Nisticò, L., Giovannini, C., Pozzilli, P., Thomson, G. and She, J.-X. (1996) Confirmation of three susceptibility genes to insulin-dependent diabetes mellitus: IDDM4, IDDM5 and IDDM8. Hum. Mol. Genet. 5, 693-698.
9 Kruglyak, L. and Lander, E.S. (1995) High-resolution genetic mapping of complex traits. Am. J. Hum. Genet. 56, 1212-1223.MEDLINE Abstract
10 Suarez, B.K., Hampe, C.L. and van Eerdewegh, P. (1994) in Gershon, E.S. and Cloninger, C.R. (eds), Genetic Approaches to Mental Disorders. American Psychiatric Press, Washington DC, pp. 23-46.
11 Hästbacka, J., de la Chapelle, A., Kaitila, I., Sistonen, P., Weaver, A. and Lander, E.S. (1992) Linkage disequilibrium mapping in isolated founder populations: diastrophic dysplasia in Finland. Nature Genet.2, 204-211.MEDLINE Abstract
12 Schleutker, J., Laine, A.-P., Haataja, L., Renlund, M., Weissenbach, J., Aula, P. and Peltonen, L. (1995) Linkage disequilibrium utilized to establish a refined genetic position of the Salla disease locus on 6q14-q15. Genomics27, 286-292.MEDLINE Abstract
13 Lehesjoki, A.-E., Koskiniemi, M., Norio, R., Tirrito, S., Sistonen, P., Lander, E. and de la Chapelle, A. (1993) Localization of the EPM1 gene for progressive myoclonus epilepsy on chromosome 21: linkage disequilibrium allows high resolution mapping. Hum. Mol. Genet. 2, 1229-1234.
14 Geitvik, G.A., Hoyheim, B., Gedde-Dahl, T., Grzeschilk, K.H., Lothe, R., Tomter, H. and Olaisen, B. (1987) The Kidd (JK) blood group locus assigned to chromosome 18 by close linkage to a DNA-RFLP. Hum. Genet. 77, 205-209.MEDLINE Abstract
15 Hodge, S.E., Anderson, C.E., Neiswanger, K., Field, L.L., Spence, M.A., Sparkes, R.S., Sparkes, M., Christ, M., Terasaki, P.I., Rimoin, D.L. and Rotter, J.I. (1982) Close genetic linkage between diabetes mellitus and Kidd blood group. Lancet 2, 893.
16 Hodge, S.E., Anderson, C.E., Neiswanger, K., Sparkes, R.S. and Rimoin, D.L. (1983) The search for heterogeneity in insulin-dependent diabetes mellitus (IDDM): linkage studies, two-locus models, and genetic heterogeneity. Am. J. Hum. Genet. 35, 1139-1155.MEDLINE Abstract
17 Dunsworth, T.S., Rich, S.S., Swanson, J. and Barbosa, J. (1982) No evidence for linkage between diabetes and the Kidd marker. Diabetes 31, 991-993.MEDLINE Abstract
18 Spielman, R., McGinnis, R. and Ewens, W. (1993) Transmission test for linkage disequilibrium: the insulin gene region and insulin-dependent diabetes mellitus (IDDM). Am. J. Hum. Genet.52, 506-516.MEDLINE Abstract
19 Copeman, J.B., Cucca, F., Hearne, C.M., Cornall, R.J., Reed, P.W., Rønningen, K.S., Undlien, D.E., Nisticò, L., Buzzetti, R., Tosi, R., Pociot, F., Nerup, J., Cornélis, F., Barnett, A.H., Bain, S.C. and Todd, J.A. (1995) Linkage disequilibrium mapping of a type 1 diabetes susceptibility gene (IDDM7) to human chromosome 2q31-q33. Nature Genet. 9, 80-85.MEDLINE Abstract
20 Devlin, B. and Risch, N. (1995) A comparison of linkage disequilibrium measures for fine-scale mapping Genomics 29, 311-322.MEDLINE Abstract
21 Risch, N. (1990) Linkage strategies for genetically complex traits. I. Multilocus Models. Am. J. Hum. Genet.46, 222-228.MEDLINE Abstract
22 Risch, N. and Merikangas, K. (1996) The future of genetic studies in complex disease. Science 273, 1516-1517.MEDLINE Abstract
23 Allamand, V., Broux, O., Richard, I., Fougerousse, F., Chiannilkulchai, N., Bourg, N., Brenguier, L., Devaud, C., Pasturaud, P., Pereira de Souza, A., Roudaut, C., Tischfield, J.A., Conneally, P.M., Fardeau, M., Cohen, D., Jackson, C.E. and Beckman, J.S. (1995) Preferential localisation of the limb-girdle muscular dystrophy type 2A gene in the proximal part of a 1-cM 15q15.1-q15.3 interval. Am. J. Hum. Genet. 56, 1417-1430.MEDLINE Abstract
24 Todd, J.A. and Farrall, M. (1996) Panning for gold: genome-wide scanning for linkage in type 1 diabetes. Hum. Mol. Genet. 5, 1443-1448.MEDLINE Abstract
25 Bain, S.C., Todd, J.A. and Barnett, A.H. (1990) The British Diabetic Association-Warren Repository. Autoimmunity 7, 83-85.MEDLINE Abstract
26 Lernmark, A., Ducat, L., Eisenbarth, G., Ott, J., Permutt, M.A., Rubenstein, P. and Spielman, R. (1990) Family cell lines available for research. Am. J. Hum. Genet. 47, 1028-1030.MEDLINE Abstract
27 Pociot, F., Norgaard, K., Hobolth, N., Andersen, O. and Nerup, J. (1993) A nationwide population-based study of the familial aggregation of type 1 (insulin-dependent) diabetes mellitus in Denmark. Diabetologia 36, 870-875.MEDLINE Abstract
29 Holmans, P. (1993) Asymptotic properties of affected-sib-pair linkage analysis. Am. J. Hum. Genet. 52, 362-374.MEDLINE Abstract
30 Kruglyak, L. and Lander, E.S. (1995) Complete multipoint sib-pair analysis of qualitative and quantitative traits. Am. J. Hum. Genet. 57, 439-454.MEDLINE Abstract
31 Spielman, R.S. and Ewens, W.J. (1996) The TDT and other family-based tests for linkage disequilibrium and association. Am. J. Hum. Genet. 59, 983-989.MEDLINE Abstract
32 Lench, N.J., Norris, A., Bailey, A., Booth, A. and Markham, A.F (1996) Vectorette PCR isolation of microsatellite repeat sequences using anchored dinucleotide repeat primers. Nucleic Acids Res. 24, 2190-2191.MEDLINE Abstract
33 Munroe, D.J., Haas, M., Bric, E., Whitton, T., Aburatani, H., Hunter, K., Ward, D. and Housman, D.E. (1994) IRE-bubble PCR: a rapid method for efficient and representative amplification of human genomic DNA sequences from complex sources. Genomics 19, 506-514.MEDLINE Abstract
34 Olivès, B., Merriman, M., Bailly, P., Bain, S., Barnett, A., Todd, J., Cartron, J.-P. and Merriman, T. (1997) The molecular basis of the Kidd blood group polymorphism and its lack of association with type 1 diabetes susceptibility. Hum. Mol. Genet., 6, 1017-1020.
35 Tay, G.K., Leeayuwat, C., Chorney, M.J., Cattley, S.K., Hollingsworth, P.N., Witt, C.S., Daly, L.N., Hughes, A. and Dawkins, R.L. (1997) The MHC contains multiple genes potentially relevant to hemochromatosis. Immunogenetics 45, 336-340.MEDLINE Abstract
*To whom correspondence should be addressed. Tel: +44 1865 740014; Fax: +44 1865 742193; Email: john.todd{at}well.ox.ac.uk
-->
This page is maintained by OUP admin. Last updated Tue Jun 10 19:01:42 BST 1997. Part of the OUP Journals World Wide Web service.
Copyright
Oxford University Press, 1996
