© 2000 Oxford University PressHuman Molecular Genetics, 2000, Vol. 9, No. 20 2967–2972
Confirmation of the DRB1-DQB1 loci as the major
component of IDDM1 in the isolated founder population
Patrizia Zavattari1, Rosanna Lampis1, Annapaola Mulargia1,2, Miriam Loddo1,2,
Efisio Angius2, John A. Todd3and Francesco Cucca1,+
1Dipartimento di Scienze Biomediche e Biotecnologie, University of Cagliari, Via Jenner, Cagliari 09121, Italy,
2Servizio di Diabetologia Pediatrica, Ospedale G. Brotzu, Via Peretti, Cagliari 09121, Italy and3Wellcome Trust
Centre for Molecular Mechanisms in Disease, University of Cambridge, Addenbrooke’s Hospital, Cambridge
CB2 2XY, UK
Received 18 August 2000; Revised and Accepted 13 October 2000
concerning the application of linkage disequilibrium
(LD) mapping in common multifactorial diseases,
including the choice of population and the density of
the marker map. Previously, it has been shown that,
in the large cosmopolitan population of the UK, the
established type 1 diabetes IDDM1 locus in the HLA
region could be mapped with high resolution by LD.
The LD curve peaked at marker D6S2444, 85 kb from
the HLA class II gene DQB1, which is known to be a
major determinant of IDDM1. However, given the
many unknown parameters underlying LD, a validation
of the approach in a genetically distinct population is
necessary. In the present report we have achieved
this by the LD mapping of IDDM1 in the isolated
founder population of Sardinia. Using a dense map of
microsatellite markers, we determined the peak of LD
to be located at marker D6S2447, which is only 6.5 kb
from DQB1. Next, we typed a large number of SNPs
defining allelic variation at functional candidate
genes within the critical region. The association
curve, with both classes of marker, peaked at the loci
conclusive evidence that the class II loci DRB1-DQB1
dominate the association of the HLA region to type 1
diabetes, provide empirical support for LD mapping.
It is anticipated that linkage disequilibrium (LD) mapping can
be used to locate disease susceptibility genes in common
multifactorial diseases (1). However, even though many
genome-wide linkage studies have been carried out in complex
diseases, so far no convincing sublocalization of a disease-
associated region, containing markers in LD with the disease
phenotype, under the peak of linkage has been reported. The
main factors accounting for these failures are represented by
the locus and allelic heterogeneity and by the low penetrance
of the polygenes involved in complex traits (2,3). However,
part of the explanation lies also in technical barriers, which
include the absence of the human genome sequence, a dense
mapof polymorphismseach witharobustassay and an affordable
high throughput method of scoring them in data sets large
enough to provide substantial statistical power. Rapid progress
is being made towards lowering these barriers but the question
remainsthat even if we had a very dense map of single nucleotide
polymorphisms (SNPs), could we find the aetiological SNPs in
the face of the likely model of multiple disease susceptibility
genes with multiple disease-associated alleles within one
linked chromosome region?
Many of the features of this ‘worst-case’ scenario for the
the human leukocyte antigen (HLA) region on chromosome
6p21 with the autoimmune disease type 1 diabetes mellitus
(T1DM), namely multiple genes and allelic heterogeneity
[insulin-dependent diabetes mellitus 1 (IDDM1)] (4). We
therefore used IDDM1 as a model locus to test the possibility
of mapping the aetiological polymorphism to any degree of
confidence. An advantage of IDDM1 is that it is a major
disease locus in every population studied so far in which
affected sib pairs have been genotyped and from which the
locus-specific sibling risk:population prevalence ratio (λs) can
be estimated (IDDM1 λs= 2.7 in UK families). Hence, in a
modest number of families (n = 200–400), very significant LD
values between marker alleles and disease can be obtained.
One possible criticism of the choice of the HLA region is the
‘unusual’ degree of LD observed in the region. However,
given that several other regions in the genome show even
greater LD (5) and that, in general, systematic studies of the
LD patterns in the human genome are in their infancy, it is
perhaps premature to reach this conclusion.
We previously carried out such an LD mapping analysis of
IDDM1 in 385 UK affected sib pair (ASP) families, the UK
being a large cosmopolitan population (6). Transmission analysis
of alleles of 25 polymorphic microsatellite markers allowed
the sublocalization of IDDM1, or at least its major determinant
in these UK families, to a 570 kb region. The peak of LD with
disease was at marker D6S2444. D6S2444 is within 85 kb of
the HLA-DQB1 locus, which is, given a combination of genetic
+To whom correspondence should be addressed. Tel: +39 070 6095681; Fax: +39 070 6095558; Email: firstname.lastname@example.org
2968 Human Molecular Genetics, 2000, Vol. 9, No. 20
and biological data, an established primary component of
IDDM1. However, LD between markers and the disease poly-
morphism depends not only on patterns and frequency of
recombination along chromosomes but also on several largely
including effective population size, population admixture,
mutation, drift, breeding system and selection (2). Therefore, it
was crucial to validate the approach in a population genetically
and demographically different from the UK population. In
order to increase mapping resolution and to rule out the possi-
bility that the LD map obtained in the UK population was a
specific property of polymorphic microsatellite markers we
also incorporated SNPs into the map.
The population of Sardinia offered several advantages for
this analysis. Sardinia is not a mixed population. It is historically
well separated from the UK population by at least 15 000 years
and 750 generations, and indeed it is well known that several
HLA haplotypes vary significantly in frequency between the
UK and Sardinians. The demographic features of the two
populations could not be more different, whilst still demanding
that both be white European. If factors other than the activity
and frequency of recombination points were influencing the
UK mapping results or if the HLA class II loci are not the
major determinants of IDDM1, in Sardinia or even in the UK,
then LD mapping in Sardinia could well give very different
results. However, a clear confirmation is provided by this
study of Sardinian families, in which we are able to accurately
map the main HLA disease component to the loci DRB1 and
DQB1. Indeed, in Sardinia the most associated marker was
6.5 kb from DQB1. Our empirical data are discussed considering
the implications for the application of LD mapping strategies
in multifactorial disorders.
as population demography,
Twenty-two microsatellite markers covering 9.452 Mb of the
extended HLA region were scored in 257 Sardinian T1DM
families (Fig. 1). The peak of association was at D6S2447,
6.5 kb telomeric of HLA-DQB1 [extended transmission dise-
quilibrium test (ETDT) Pc= 7.6 × 10–31]. Note that marker
D6S265 is located close to the classical HLA locus HLA-A and
2478 kb telomeric of DQB1, and was still strongly associated
with the disease (Pc= 5.4 × 10–5), whereas marker D6S2444,
located only 85 kb centromeric of DQB1, showed no associa-
tion. Remarkably, the overall shape of the curve was very
similar to that obtained previously in UK families (shown in
Fig. 1 asa dashed line). Next, we typed 106 SNPs, including 94
in the HLA-DPB1, -DQB1 and -DRB1 genes, in order to
increasetheinformativity of themapand toprovide acomparison
of the associations of markerswith those of the actual aetiological
SNPs (in the exon 2 sequences of the HLA-DQB1 and -DRB1
genes). The other SNPs were chosen because they were in
functional candidate genes (tapasin, DMA, DMB, LMP2,
LMP7, DOB and HSP70–2) and had common enough minor
allele frequencies to provide statistical power in the association
study. Ostensibly, the shape of the plot with all 33 loci
markers did not change (Fig. 2) except that statistical
significance of the main peak in the class II region increased
from –log10(Pc) = 30.1 to –log10(Pc) = 42.9 by inclusion of the
aetiological SNPs in the exon 2 sequences of the HLA-DQB1
and -DRB1 genes.
Two other features of the mapping results are notable: first,
the relation between the association of the markers with the
disease and their patterns of global LD with the disease loci
DRB1-DQB1 and, second, the importance of the allele specific
LD patterns of the neutral markers with the disease loci.
Figure 3 shows, as a dashed line, the pairwise D′ multiallelic
values detected between each marker and DRB1-DQB1, here
considered as a superlocus, and the respective association with
Figure 1. Single-point association of microsatellite markers with T1DM in the
6p21 region. Distances are given in megabases proceeding from the most cen-
tromeric marker (D6S291) to the most telomeric (D6S2223). Association is
presented as the negative log of the P value for the extended transmission dis-
equilibrium test (ETDT). The bold horizontal line corresponds to the threshold
significance of 0.05. Single point scores for the Sardinian sample set are con-
nected by the solid line. For the purpose of comparison, corresponding values
for the UK population, obtained from Herr et al. (6), are shown by the dotted
lines. Selected microsatelite loci are labelled.
Figure 2. Single-point association of SNPs and microsatellite markers with
T1DM in the 6p21region. Physical distance is shown as megabases proceeding
from the most centromeric marker (D6S291) to the most telomeric (D6S2223).
Association is presented as the negative log of the P values obtained with the
ETDT. The bold horizontal line shows the threshold significance of 0.05. Sin-
gle point scores for SNPs in various expressed genes are indicated by the open
symbols, and filledsymbolsare presentedfor microsatellites. Selectedmarkers
Human Molecular Genetics, 2000, Vol. 9, No. 20 2969
the disease, expressed as the –log of the P value previously
obtained with the ETDT. The associations of the various loci
with the disease tend to follow those with DRB1-DQB1. In
particular, LD of the various loci with DRB1-DQB1 was
stronger in the telomeric side than in the centromeric part of
the map. Accordingly, the allelic association curve drops more
suddenly in the region centromeric of DQ-DR and more linearly
on the telomeric side in the chromosomal region located
between the DRB1 and the class I loci, which encompasses the
whole class III region. Additionally, we evaluated LD between
the various loci and DR3 (DRB1*0301-DQB1*0201) and DR4
(DRB1*0405-DQB1*0302), which are the main predisposing
haplotypes in this population (Fig. 4 and Materials and
Methods). There was a remarkable similarity in the shape of
these curves except for the class III (markers D6S273 and
TNFc) and the most telomeric part of the map (markers
D6S2223), where LD of the various markers with DR3 was
stronger than with DR4.
The importance of the allele-specific LD patterns is illus-
trated by the behaviour of marker D6S2444. This marker,
located 85 kb centromeric of DQ, was in very significant LD
with the DRB1-DQB1 disease superlocus (D′= 0.55,P < 1 × 10–7)
but was not associated with the disease itself (ETDT Pc= 0.19 ).
Allele 2 (153 bp) of D6S2444 is the most common allele in the
Sardinian sample but was in positive LD both with (DR3)
DRB1*0301-DQA1*-0501-DQB1*0201 (D′ = 0.5, Pc< 1 × 10–7)
and (DR5) DRB1*11–12-DQA1*-0501-DQB1*0301 (D′ = 0.7,
Pc< 1 × 10–7), which represent, respectively, the most common
predisposing and the most common protective haplotypes in
the Sardinian population. This marker allele distribution, i.e.
the same allele present on protective and predisposing haplo-
types, results in a cancelling out of the marker association with
the disease. Note that marker D6S2444, despite its lack of
association with the disease using the single point analysis, was
strongly associated using a two-point analysis, in conjunction
with D6S2447 (ETDT Pcvalue = 1.23 × 10–25).
Using a high-resolution allelic association study of the HLA
region in Sardinian T1DM families, we have demonstrated that
the HLA class II genes DRB1 and DQB1 are the main determi-
nants of IDDM1 in this population. Remarkably, the overall
shape of the plot obtained in this study of Sardinian families is
very similar to that previously observed in the distantly related
population of the UK. Taken together the Sardinian and UK
results are conclusive evidence that the DRB1-DQB1 HLA
classII loci dominate the association of the major histocompat-
ibility complex (MHC) region with the disease.
Overall, our empirical results support the feasibility of LD
mapping in candidate regions. Nevertheless, based on our data
some recommendations for strategies are advised. As
illustrated in Figure 2, the allelic association curve is heavily
influenced by the global LD patterns of the various markers
with the disease loci. As the multiallelic D′ declines to 0.6, the
degree of association with the disease decreases rapidly. These
results suggest that marker density in the critical region must
be high in order to ensure strong LD between the markers
employed in the map and theaetiologic SNPs (7).Our data also
suggest that the underlying pattern of LD between the neutral
markers and the disease locus determines the general shape of
the association curve. In the case of IDDM1, differences in the
LD patterns of the various loci with DRB1-DQB1 explain why
bothintheUKandin the Sardiniansamplestheallelicassociation
curve drops more steeply in the region centromeric of DQ-DR,
and more linearly on the telomeric side. The pattern of LD of
the region with disease correlates well with the recognized hot
spots of recombination, particularly in the region just centromeric
of the HLA-DQB1 (8–10).
Allele-specific LD patterns and marker informativity, i.e. the
distribution of marker alleles on disease-associated chromo-
somes, are also critical parameters. This distribution could
vary considerably between two populations, particularly two
as historically distinct asSardinia and the UK. A good example
of this was provided by marker D6S2444. In the Sardinians, no
association with the disease is observed with this marker which
is located only ∼85 kb centromeric of the main disease locus
Figure 3. Overlay of single-point association of loci with T1DM (solid line)
and strength of LD (dotted line) over the 6p21 region. Association is presented
as the negative log of the P value for the ETDT. Strength of LD with DRB1-
DQB1, here considered as a superlocus, is measured as a normalized D′ value
(26,27). Physical distance is shown as megabases proceeding from the most
centromeric marker (D6S291) to the most telomeric (D6S2223).
Figure 4. Comparison of haplotype-specific strength of LD versus physical
distance for the6p21 region. Dashed lines show pairwise D′ valuesobservedin
DR3 haplotypes between thevarious markers and DRB1-DQB1, and solidlines
show corresponding pairwise LD in DR4 haplotypes. Physical distance is
shown as megabases proceeding from the most centromeric marker (D6S291)
to the most telomeric (D6S2223).
2970 Human Molecular Genetics, 2000, Vol. 9, No. 20
HLA-DQB1. In striking contrast this marker was the most
strongly associated in the UK study (6). This conflicting obser-
vation was explained by the fact that the most common allele
at this marker in the Sardinian population is simultaneously in
very significant LD with the most common predisposing and
protective DRB1-DQB1 haplotypes. In contrast, allelic variation
at D6S2444 was informative in the UK families owing to its
distinct allelic distribution on DRB1-DQB1 predisposing and
protective haplotypes (data not shown). These results underline
how LD patterns between markers and disease can be
complicated. Two point haplotype analysis alleviates these
problems while also increasing the information content of the
map by raising the number of heterozygous parents (11–14).
In conclusion, we have mapped the main components of
IDDM1 to the loci DRB1 and DQB1 in the isolated founder
population of Sardinia. Our empirical observations are applicable
to association studies of disease loci with weaker genetic
effects except that the sample sizes required to detect significant
associations will be proportionally higher (7).
MATERIALS AND METHODS
Subjects and HLA typing
The data set consisted of 257 Sardinian T1DM families, all
from a paediatric department of the southern part of the island.
The sample set included 273 affected children, their parents
and 241 unaffected siblings (average age of the patients at
disease onset was 8.15 ± 4.5 years (females: 8.2 ± 3.9; males:
8.1 ± 4.2). Overall the whole data set was genotyped for 22
microsatellite markers. The primer sequences for D6S291,
D6S439, D6S1629, D6S1560, D6S1568, D6S2445, D6S2444,
D6S2447, D6S273, C1-2-A, D6S265, D6S258, D6S1683 and
D6S306 were obtained from Foissac and Cambon-Thomsen
(15). Sequences for TNFa, TNFc, TNFd and TNFe were
obtained from Udalova et al. (16). Sequences for D3A, 82-1
and 82-2/9N-1 were established by Hsieh et al. (17). The
primer sequence for D6S2223 was obtained from the Genome
by separating fluorescently tagged polymerase chain reaction
(PCR) products on a polyacrylamide gel using ABI 373 and
ABI 377 DNA sequencersand the GeneScan 3.1 andGenotyper
2.0 software (Perkin-Elmer Applied Biosystems, Warrington,
UK). PCR product standards, consisting of the amplification
product of two different standards for each marker were loaded
on each gel for correct allele assignment. The two standards
consisted of the Centre d’Etude des Polymorphisms Human
(CEPH) individual no. 1347.02 and of a pool of DNAs. The
alleles at each microsatellite were given a numerical value
(1, 2, 3 etc.) starting with the allele with the lowest number of
base pairs. The physical map of the region with relative order,
map position and distances between markers was obtained from
and physical map positions of the microsatellite markers are
shown in Table 1. Markers were PCR amplified and genotyped
a second time when showing failures during the first round of
amplifications. On average 74.2% of all parents were hetero-
zygous for the microsatellites investigated, with outlying
markers D6S2445, D6S2444, TNFe, TNFc and D6S2223
showing heterozygosity < 60%. The average number of alleles
per microsatellite was 10, but it was only 3 when we consid-
ered alleles with a parental frequency of at least 10%.
The expressed genes considered in this study were typed as
follows. The polymorphic second exons of the HLA-DRB1,
-DQBl and -DPB1 genes were amplified and the amplified
products were dot-blot analysed using primers and sequence-
specific oligonucleotide (SSO) probes previously described
(18–20). This includes characterization of 51, 25 and 18 SNPs
for the DRB1, DQB1 and DPB1 loci, respectively. Alleles at
the polymorphic second exon of DQA1 were inferred by the
known patterns of LD with DQB1 and DRB1 in the Sardinians.
LMP2 (1 SNP), LMP7 (1 SNP), DMA (4 SNPs) and DMB
(3 SNPs) polymorphisms were typed using primers and condi-
tions previously described (21,22). DOB (1 SNP) was typed
with amplification of the fourth exon and subsequent dot-blot
analysis of the amplified products using 5′-GTGTCTAG-
TACAGATTCTG-3′ and 5′-CACTCCTCACAGGCTCAT-3′
as PCR primers and 5′-GTGGGAATCATCATCCAG-3′ and 2
(1 SNP) gene was typed using primers and conditions
previously described (23).
The tapasin (1 SNP) gene was typed with amplification
of the fourth exon and subsequent dot-blot analysis of the
amplified products using 5′-AAATGGGACCTTCTGGCTGC-3′
and 5′-AAGCTCCAGGGTGACCTGTC-3′ as PCR primers and
5′-GGCTGCCTAGAGTTCAACCC-3′ and 5′-GGCTGCCTA-
The degree of association of the various loci with T1DM was
established using the ETDT (24). This test takes into account
the transmission or non-transmission of alleles of a marker
relative to the alleles of the marker present on the other
parental chromosome. The ETDT takes multiple alleles into
accountand obtainsaglobalP valueindicativeof thesignificance
of the association with the disease at each individual locus.
Therefore, the ETDT could be considered a particularly suitable
method for the analysis of a region showing a multiallelic, two-
sided association. The P values were corrected for number of
loci considered and the –log of the corrected P values were
plotted versus the physical position of the loci in the map. For
this analysis only one affected child was randomly selected
from all the families with more than one affected sibling.
Haplotyping was performed in all analyses considered in this
manuscript following the method proposed by Clayton (25)
and using computer programs written by Frank Dudbridge
(CIMR, Cambridge University, UK; available by anonymous
ftp at http://diesel.cimr.cam.ac.uk ). When haplotypes are not
certain from parental genotype data, all possibilities are
considered with weighting proportional to their population
frequencies, which are estimated by the expectation-maximi-
The total normalized disequilibrium (total D′) between
DRB1-DQB1 and the various marker loci was calculated using
a multiallelic extension of Lewontin’s standardized measure of
disequilibrium (26,27) and ranges from 0 to 1, with 0 reflecting
perfect independence between alleles at the two loci compared
and 1 reflecting complete LD. The respective P values were
calculated using the Markov chain method described by Guo
and Thompson (28). The LD between the various loci
Human Molecular Genetics, 2000, Vol. 9, No. 20 2971
and DR3 (DRB1*0301-DQA1*0501-DQB1*0201) and DR4
(DRB1*0405-DQA1-0301-DQB1*0302) was computed, using
the methods described above, by selecting 100 positive haplo-
types for each category, evaluated respectively in the same
background of 100 non-DR3-DR4 haplotypes.
We wish to thank Antonio Cao, Stefano De Virgiliis, Mario
Silvetti, Mathias Herr, Elisabetta Deidda and Michael Whalen
for help, advice and support, Cesare Zavattari for writing a
program that allows assignment of the allele sizes into their
appropriate allele bins, Frank Dudbridge for statistical advice,
Margi Chessa, Paola Frongia and Rossella Ricciardi for help in
collecting the Sardinian T1DM families, James Copeman for
information about tapasin and Andrew Mungall (Sanger
Center) for the establishment of the physical map. We would
also like to thank the Italian Telethon, the Regione Autonoma
Sardegna(L.R.11, 30–4-90) and theWellcome Trust for financial
support. F.C. and J.A.T. are recipients of a Wellcome Trust
Biomedical Research Collaboration grant and J.A.T. was a
Wellcome Trust Principal Research Fellow.
Table 1. Marker characteristics
Marker NumberPosition on physical map (Mb)Number of alleles Heterozygosity
2 1.6 11 0.687
3 1.87 0.766
5 2.8 120.795
6 2.882 0.513
8 3.234 0.287
9 3.2445 0.299
10 3.293 110.558
12 3.3342 0.442
15 3.51715 0.784
16 3.523511 0.741
17 3.5367 0.653
19 4.00110 0.614
21 4.3982 0.445
22 4.40811 0.790
24 4.54127 0.618
28 4.6202 13 0.760
29 5.995212 0.679
2972 Human Molecular Genetics, 2000, Vol. 9, No. 20 Download full-text
1. Lander, E.S. and Schork, N.J. (1994) Genetic dissection of complex traits.
Science, 265, 2037–2048. [Erratum (1994) Science, 266, 353.]
2. Terwilliger, J.D. andWeiss, K.M. (1998) Linkage disequilibriummapping of
complexdisease: fantasyor reality? Curr. Opin. Biotechnol., 9, 578–594.
3. Risch, N.J. (2000) Searchingforgeneticdeterminants in the newmillennium.
Nature, 405, 847–856.
4. Cucca, F. and Todd, J.A. (1996) HLA susceptibility to type 1 diabetes:
methods and mechanisms. In Browning, M. and McMichael, A.M. (eds),
HLA and MHC—Genes, Molecules and Function. BIOS Scientific,
Oxford, UK, pp. 383–406.
5. Huttley, G.A., Smith, M.W., Carrington, M. and O’Brien, S.J. (1999) A
scan for linkage disequilibrium across the human genome. Genetics, 152,
6. Herr, M., Dudbridge, F., Zavattari, P., Cucca, F., Guja, C., March, R.,
Campbell, R.D., Barnett, A.H.,Bain,S.C., Todd, J.A. etal. (2000)Evaluation
of fine mapping strategies for a multifactorial disease locus: systematic link-
age and association analysis of IDDM1 in the HLA region on chromosome
6p21. Hum. Mol. Genet., 9, 1291–1301.
7. McCarthy, J.J. and Hilfiker, R. (2000) The use of single-nucleotide poly-
morphism maps in pharmacogenomics. Nature Biotechnol., 18, 505–508.
8. Cullen, M., Noble, J., Erlich, H., Thorpe, K., Beck, S., Klitz, W.,
Trowsdale, J. and Carrington, M. (1997) Characterization of recombina-
tion in the HLA class II region. Am. J. Hum. Genet., 60, 397–407.
9. Carrington, M. (1999) Recombination within the human MHC. Immunol.
Rev., 167, 245–256.
10. Jeffreys, A.J., Ritchie, A. and Neumann, R. (2000) High resolution
analysis of haplotype diversity and meiotic crossover in the human TAP2
recombination hotspot. Hum. Mol. Genet., 9, 725–733.
11. Merriman, T.R., Eaves, I.A., Twells, R.C., Merriman, M.E., Danoy, P.A.,
Muxworthy, C.E., Hunter, K.M., Cox, R.D., Cucca, F., McKinney, P.A.
et al. (1998) Transmission of haplotypes of microsatellite markers rather
than single marker alleles in the mapping of a putative type 1 diabetes
susceptibility gene (IDDM6). Hum. Mol. Genet., 7, 517–524.
12. Dudbridge, F., Koeleman, B.P.C., Todd, J.A. and Clayton, D.G. (2000)
Unbiased application of the transmission/disequilibrium test to multilocus
haplotypes. Am. J. Hum. Genet., 66, 2009–2012.
13. Beckman, L.E., Van Landeghem, G.F., Sikstrom, C. and Beckman, L.
(1998) DNA polymorphisms and haplotypes in the human transferrin
gene. Hum. Genet., 102, 141–144.
14. Martin, E.R., Lai, E.H., Gilbert, J.R., Rogala, A.R., Afshari, A.J., Riley, J.,
Finch, K.L., Stevens, J.F., Livak, K.J., Slotterbeck, B.D. et al. (2000)SNPing
away at complex diseases: analysis of single-nucleotide polymorphisms
aroundAPOE inAlzheimer disease. Am. J. Hum. Genet., 67, 383–394.
15. Foissac, A. and Cambon-Thomsen, A. (1998) Microsatellites in the HLA
region: 1998 update. Tissue Antigens, 52, 318–352.
16. Udalova, I.A., Nedospasov, S.A., Webb, G.C., Chaplin, D.D. and
Turetskaya, R.L. (1993) Highly informative typing of the human TNF
locus using six adjacent polymorphic markers. Genomics, 16, 180–186.
17. Hsieh, S.L., March, R.E., Khanna, A., Cross, S.J. and Campbell, R.D. (1997)
Mapping of 10 novel microsatellites in the MHC class III region: application
tothe study of autoimmune disease. J. Rheumatol., 24, 220–222.
18. Horn, G.T., Bugawan, T.L., Long, C.M. and Erlich, H.A. (1988) Allelic
sequence variation of the HLA-DQ loci: relationship to serology and to
insulin-dependent diabetes susceptibility. Proc. Natl Acad. Sci. USA, 85,
19. Wordsworth, B.P., Allsopp, C.E., Young, R.P. and Bell, J.I. (1990) HLA-
DR typing using DNA amplification by the polymerase chain reaction and
sequential hybridization to sequence-specific oligonucleotide probes.
Immunogenetics, 32, 413–418.
20. Bugawan, T.L., Begovich, A.B. and Erlich, H.A. (1991) Rapid HLA-DPB
typing using enzymatically amplified DNA and nonradioactive sequence-
specific oligonucleotide probes. Immunogenetics, 34, 413.
21. Deng, G.Y., Muir, A., Maclaren, N.K. and She, J.X. (1995) Association of
LMP2 and LMP7 genes within the major histocompatibility complex with
insulin-dependent diabetes mellitus: population and family studies. Am. J.
Hum. Genet., 56, 528–534.
22. Djilali-Saiah, I., Benini, V., Schmitz, J., Timsit, J., Assan, R., Boitard, C.,
Bach, J.F. and Caillat-Zucman, S. (1996) Absence of primary association
between DM gene polymorphism and insulin-dependent diabetes mellitus
or celiac disease. Hum. Immunol., 49, 22–27.
23. Vinasco, J., Beraun, Y., Nieto, A., Fraile, A., Pareja, E., Mataran, L. and
Martin, J. (1997) Heat shock protein 70 gene polymorphisms in rheuma-
toid arthritis. Tissue Antigens, 50, 71–73.
24. Sham, P.C. and Curtis, D. (1995) An extended transmission/disequilibrium
test (TDT) for multi-allelemarker loci. Ann. Hum. Genet., 59, 323–336.
25. Clayton, D. (1999) A generalization of the transmission/disequilibrium test
for uncertain-haplotype transmission. Am. J. Hum. Genet., 65, 1170–1177.
26. Hedrick, P.W. (1987) Gametic disequilibrium measures: proceed with
caution. Genetics, 117, 331–341.
27. Lewontin, R.C. (1988) On measures of gametic disequilibrium. Genetics,
28. Guo, S.W. and Thompson, E.A. (1992) Performing the exact test of
Hardy–Weinberg for multiple alleles. Biometrics, 48, 361–372.