Hum Genet (2005)
DOI 10.1007/s00439-005-0053-5
O RI GI N AL IN V ES T IG A T IO N
Toralf Bernig Æ Willemijn Breunis Æ Nannette Brouwer
Amy Hutchinson Æ Robert Welch Æ Dirk Roos
Taco Kuijpers Æ Stephen Chanock
An analysis of genetic variation across the MBL2 locus in Dutch
Caucasians indicates that 3¢ haplotypes could modify circulating
levels of mannose-binding lectin
Received: 19 May 2005 / Accepted: 3 August 2005
Springer-Verlag 2005
Abstract Mannose-binding protein (MBL) is a critical
component of innate immunity and provides first-line
protection against pathogens. Both circulating MBL
serum levels and functional activity have been correlated
with common genetic variants in the MBL2 gene.
Associations between MBL deficiency and severe infections have been reported in immuno-incompetent patients and for autoimmune disorders; however,
measured MBL serum levels do not fully correlate with
the ‘secretor haplotypes’. Previously, the MBL2 locus
was resequenced and determined that a recombination
hotspot divides MBL2 into two haplotype blocks. It was
sought to investigate whether additional variants, in either block structure could associate with MBL serum
levels. Therefore, 31 common variants were analysed
across the locus in 212 DNA samples of healthy Caucasian individuals with known MBL serum concentraElectronic Supplementary Material Supplementary material is
available for this article at http://dx.doi.org/10.1007/s00439-0050053-5
T. Bernig Æ S. Chanock (&)
Section on Genomic Variation, Pediatric Oncology Branch,
Center for Cancer Research, National Cancer Institute,
National Institutes of Health,
Bethesda, MD 20892-4605, USA
E-mail: [email protected]
Tel.: +1-301-4357559
Fax: +1-301-4023134
W. Breunis Æ T. Kuijpers
Emma Children’s Hospital, Academic Medical Centre,
University of Amsterdam, 1105, Amsterdam,
The Netherlands
N. Brouwer Æ D. Roos
Sanquin Research Institute at the Central Laboratory
of the Blood Transfusion Service, University of Amsterdam,
1066, Amsterdam, The Netherlands
A. Hutchinson Æ R. Welch
Core Genotyping Facility, Advanced Technology Center,
National Cancer Institute, National Institutes of Health,
Bethesda, MD 20892-4605, USA
tions. The additional 5¢ variants were in strong linkage
to the elements of the ‘secretor haplotypes’; functional
alleles B, C and D also lie on restricted haplotypes. Four
variants in the 3¢ block (Ex4-1483T>C, Ex4-1067G>A,
Ex4-901G>A and Ex4-710G>A) are components of a
distinct haplotype block. The results of this study suggest that additional 5¢ variants as well as markers of
distinct 3¢ haplotype blocks in MBL2 may contribute to
circulating protein levels, but further studies are
required to confirm these observations. Last, there could
be a selective advantage for diversification of the 3¢
region of the gene.
Keywords Mannose-binding lectin Æ Innate immunity Æ
Genetic association Æ Haplotype association Æ
Haplotype-trait interaction
Introduction
Mannose-binding protein (MBL) is a major component
of the innate immune response and specifically provides
first-line protection against pathogens by binding to Nacetyl -glucosamine and mannose residues on the surface
of microorganisms (bacteria, fungi and parasites) (Jack
and Turner 2003). MBL is a C-type collectin patternrecognition molecule, critical for the immediate response
to invading pathogens. In turn, opsonisation or activation of the lectin pathway is effected through the MBLassociated serine proteases (MASPs) (Ikeda et al. 1987;
Kuhlman et al. 1989; Thiel et al. 1997). Both circulating
MBL serum levels and functional activity have been
correlated with common genetic variants in the MBL2
gene on chromosome 10 (10q11.2-11). Three polymorphisms in exon 1 of MBL2 that are known as the D-, Band C-alleles coding for structurally variant proteins in
codons 52, 54 or 57 have been identified (Sumiya et al.
1991; Lipscombe et al. 1992; Madsen et al. 1994). These
components form together with two linked promoter
SNPs (i.e. 550 H/L and 221 Y/X) as well as a 5¢UTR
SNP (i.e. +4 P/Q) seven well-characterized ‘secretor
haplotypes’ (HYPA, LYPA, LYQA, LXPA, HYPD,
LYPB, LYQC), which partially account for alterations
in complement activation and decreased circulating
levels of MBL2 (Madsen et al. 1995, 1998; Wallis 2002;
Garred et al. 2003). A number of studies have reported
associations between MBL deficiency and recurrent or
severe infections, especially in immuno-incompetent
patients (i.e. BMT recipients, cancer patients and infants) or in auto-immune disorders; however, measured
MBL serum levels do not fully correlate with the
‘secretor haplotypes’ (Kilpatrick 2002; Eisen and
Minchinton 2003; Biezeveld et al. 2003).
Previously, a resequencing analysis of the 10.0 kbp
region that includes the MBL2 gene was conducted and
a high degree of heterozygosity was observed (Bernig
et al. 2004). Moreover, the pattern of linkage disequilibrium revealed a probable recombination hotspot in
the 3¢ end of the gene dividing MBL2 into two separate
haplotype blocks (Bernig et al. 2004). There appears to
be two possible gene conversion tracts within the gene;
one cassette includes exon 2, whereas the second one
overlaps with the recombination hotspot. Each of the
common ‘secretor haplotype’ resides on a unique extended 5¢ haplotype that is distinct from the 3¢ haplotype
block, which could be a hallmark of selective pressure
acting to separate regions of the gene. In this regard,
balancing selection has probably maintained the restricted haplotype pattern of MBL2 leading to a high
preservation of heterozygosity. The resulting changes in
the circulating levels and function of variant MBL
proteins could have a selective advantage in response to
environmental pressures, such as infection (Takahashi
et al. 2002; Soborg et al. 2003).
It was therefore investigated in a population of
unrelated Dutch Caucasians whether the extended block
structure might identify additional SNPs that contribute
to MBL serum levels. Not only haplotype-tagging SNPs
were analysed, but also additional common SNPs with
minor allele frequencies of greater than 5% in both
blocks defined in the Caucasian population in the previous resequencing study (Bernig et al. 2004). The data
establish the foundation for investigating the contribution of additional 5¢ variants as well as separate 3¢
haplotype blocks to circulating levels of MBL.
Materials and methods
Study population
The study population consists of a cohort of total 212
unrelated Dutch Caucasian individuals (age range 2.4 –
52.7 years, median 12.5 years; 120 males, 90 females)
who were referred to the Emma Children’s Hospital and
the Academic Medical Center Amsterdam, The Netherlands, for minor surgery or as follow-up for miscellaneous non-infectious diseases. All participants provided
written informed consent. The local ethics committee of
the Emma Children’s Hospital, Academic Medical
Center, Amsterdam, approved the study. Plasma was
separated from blood cells and frozen; DNA was isolated and stored at
80C at Sanquien Research
Amsterdam, The Netherlands.
Genotype analysis
Twenty known SNPs across MBL2 were directly genotyped. They were chosen to capture both haplotype
diversity in the 5¢ and 3¢ block and the defined two gene
conversion elements in the Caucasian population
(n=31) of SNP500cancer project. The six markers of the
‘secretor haplotypes’ were pre-assigned as tagging SNPs.
The set also contains further polymorphisms that have
shown a strong LD to the original ‘secretor haplotypes’
in the pilot resequencing study. Finally, a set of 13 SNPs
for the 5¢ block ( 2701, 2477, 1964, 1401, 618,
289, 65, Ex1-34, Ex1-27, Ex1-18, IVS1-242, IVS1-90,
IVS2-250), 4 SNPs for the 3¢ block (Ex4-1483, Ex4-1067,
Ex4-901, Ex4-710) and 3 SNPs for the conversion elements (IVS2-630, IVS3-28 and Ex4+5) were analysed
by Taqman assays (http://snp500cancer.nci.nih.gov) or
direct sequencing (i.e tri-allelic SNP IVS1-242) (Packer
et al. 2004; Bernig et al. 2004).
In addition to the genotype analysis, DNA sequence
analysis was performed in a highly conserved region 5¢
upstream, adjacent to the region previously sequenced;
evolutionarily conserved sequences were identified on the
basis of at least 75% sequence similarity over 75 bp
window
between
mouse
and
man
(http://
www.gsd.lbl.gov/vista) (Mayor et al. 2000; Dubchak
et al. 2000; Bray et al. 2003). M13-tagged PCR primers
were designed for four contigs (contig 1: forward 5¢- GGG
GCCTAAGGTAGTCCTTG, reverse 5¢- GGATAGGGTG
CTGTCCGTAA, 548 bp; contig 2: forward 5¢- ATGCCC
CTCTCCTGCTTAAT, reverse 5¢- CATGTGAGAGGCA
TGGTTTG, 543 bp; contig 3: forward 5¢- GCTCTGTGG
AAAAGGGAATG, reverse 5¢- CACAACAAATCCACC
TCACC, 578 bp; contig 4: forward 5¢- GCCAGGATGGT
CTCAATCTC, reverse 5¢- CTTTAGCAGCCAGAGCTT
CC, 491 bp) from the human MBL2 genomic sequence
(RefSeq NM_000242, NCBI build 35, locus ID 4135)
using the web based version of Primer3 software (http://
www.genome.wi.mit.edu/cgi-bin/primer/primer3_www.cgi)
(Rozen and Skaletsky 2000). Bi-directional sequence
analysis was performed on amplified fragments generated by standard PCR technique using Big Dye version 3
(ABI Perkin-Elmer) under previously described conditions (Bernig et al. 2004). Sequence reactions were
analysed on an ABI Perkin-Elmer 3700 platform with
Sequencher software 4.0.5 (Gene Codes Corporation,
Ann Arbor, MI, USA). A genetic variant was verified
only if it was observed in both the forward and reverse
orientation. Detected common variants (MAF ‡ 0.05)
were included in subsequent analyses.
Hereafter, mutation nomenclatures are presented in
regard to the NCBI contig NM_000242 as recommended
by Dunnen JT den and Antonarakis SE (Dunnen and
Antonarakis 2001).
Measurement of MBL concentration
MBL concentrations were measured in a solid-phase
enzyme-linked immunosorbent assay (ELISA) by using
monoclonal biotinylated mouse anti-human MBL IgG
(aMBL-1, 10lg/ml). Plasma samples and MBL standards (standard serum 1.5 lg/ml MBL), which were
diluted in TTG/Ca2+ (20 mM Tris pH 7.4, 150 mM
NaCl, 0.02% Tween-20/0.2% gelantine, 10 mM CaCl2)
and 10 U/ml heparin, were added to mannan-coated
micro titer plates and incubated at room temperature for
1 h. After washing, the plates were incubated with
aMBL-1 for 1 h before incubation with streptavidin
poly-HRP 1:10,000 in TBS/Ca2+/2% milk (20 mM Tris
pH 7.4, 150 mM NaCl, 10 mM CaCl2, 2% milk) for
30 min. Colorimetric determination was analysed using
0.11 M CH3COONa pH 5.5, 100 lg/ml TMB, 0.0033%
H2O2. The reaction was terminated with 2 M H2SO4
and measured at 405 nm in an ELISA plate reader
(BioAssay Reader, Sunrise, Tecan, Salzburg, Austria).
Statistical analysis
Genotype frequencies for each marker were tested for
deviation from Hardy–Weinberg equilibrium (HWE)
(Chi-squared statistic with one df) (Weir 1996). In a first
step preliminary haplotypes across the entire locus were
inferred from unphased genotype data using the Baysian
statistical method in PHASE 2.1 (Stephens et al. 2001b;
Stephens and Donnelly 2003). Subsequently, haplotype
block boundaries were inferred from the phased genotype data (probability threshold for correct phase call at
each site: 0.95) by D¢ confidence limits (upper confidence
limit >0.97, lower confidence limit > 0.70, fraction of
informative pairs in strong LD: 0.95) using Haploview
(http://www.broad.mit.edu/personal/jcbarret/haploview/) (Gabriel et al. 2002; Barrett et al. 2004).
Consecutively, the relationship between haplotypes
and MBL serum concentration was analysed using the
haplo.stats software (version 1.1.1 written for R 1.7.1.)
and each haplotype block was analysed independently as
well as in combination. A log transformation of the
measured MBL levels was performed to reach the required distribution of continuous variants. The embedded expectation-maximization (EM) algorithm was used
to infer the final haplotype frequencies for each defined
haplotype block on the basis of genetic markers with an
unknown linkage phase. The score statistics provides a
global and haplotype-specific test in a co-dominant
model taking linkage phase ambiguity into account
(Schaid et al. 2002). Therefore, haplotype frequencies
were estimated by EM algorithm independently of the
trait and covariates; the latter can be included in the
analysis in order to develop score statistics adjusted for
covariates (i.e. age as continuous variable and gender as
binary variable). Statistical significance is expressed as a
permutation P-value (minimal simulations: 10,000, significance level < 0.05) (Besag and Clifford 1991). The
function haplo.glm computes the regression of a trait on
haplotypes as well as other covariates (age, gender) that
could interact with the haplotypes in reference to a
baseline haplotype (Lake et al. 2003; Stram et al.
2003b). A co-dominant model was used for the analyses.
The most common haplotype in the sample population
was chosen as reference in each analysis. Finally, haplotype tagging SNPs for each block were determined as
per the method of Stram (Stram et al. 2003a) by using
the imported haplotype frequencies inferred in Haploview before. The chosen tagging SNPs represent the set
of SNPs that maximize the minimum R2H, which is a
coefficient of determination for common haplotypes
(frequency ‡0.01) in both blocks.
We also reconstructed the most likely haplotype pair
of both blocks for each subject to compare the performed statistics with diplotype data using the conditional probability values for estimated haplotype pairs
(probability threshold for correct phase call at each site:
0.95). The Statview package for Windows, version 5.0.1
(SAS Institute Inc., USA) was used for the statistical
analysis (non-parametric tests: Kruskal–Wallis for three
or more groups, Mann–Whitney U-test for two groups;
two-tailed P-value) and graphical illustration (box-plot)
of the MBL serum concentrations, which were stratified
according to the individual haplotype pairs.
The unrooted statistical parsimony network for 5¢
‘secretor haplotypes’ were estimated under the assumption of a limit of parsimony equal to one, namely, the
lineage of haplotypes differing by one nucleotide can be
assessed (Templeton et al. 1992). Insertions/deletions as
well as multi-allelic variants were not included in the
analysis. The final network was estimated using the TSC
software (Clement et al. 2000).
Results
Linkage disequilibrium pattern and haplotypes
Sequencing analysis of a highly conserved region 5¢
upstream (roughly 1,500 bp) has identified 11 additional
common polymorphisms, including a 1-bp in/del
( 4461delG) and 10 SNPs with a minor allele frequency
(MAF) >0.05. Together with the 20 directly genotyped
loci, a total number of 31 variants were further analysed
in 212 unrelated Caucasians (Supplementary Table 1).
All variant sites conformed to HWE. The observed
frequencies for the structural mutations in exon 1 (Ex127, B-allele 0.15; Ex1-18, C-allele 0.04; Ex1-34, D-allele
0.06) reflect the expected distribution of these variants
on the basis of previously published Caucasian populations (Madsen et al. 1998; Crosdale et al. 2000). Eightythree individuals were heterozygous for one of the three
exon 1 mutations, and 12 individuals were homozygous
Table 1 MBL2 common 5¢ and 3¢ haplotypes in accordance with the underlying haplotype block structure
Haplotypes were separately inferred from unphased genotype data inside the two haplotype blocks using the EM-algorithm (haplo.score).
Haplotype blocks were generated by Haploview using the LD confidence intervals (5¢ block—25 SNPs, 3¢ block—4 SNPs). Each square
represents an allele at the position placed on the top; major allele gray, minor allele white. The observed relative frequencies are shown in
accordance to the total number of chromosomes (n=424); the percentage of coverage for all observations is shown on the bottom. Only
haplotypes with a frequency more than 0.01 are presented. Six markers form the ‘secretor haplotypes’: 618 (aka-550), 289 (aka-221), 65
(aka+4), Ex1-34 (codon 52), Ex1-27 (codon 54) and Ex1-18 (codon 57). Lines display the connection between both blocks; value of
multiallelic D prime between both blocks is 0.94. The htSNPs were selected on a block-by-block basis and marked with a tick beneath the
SNP identifier; 7 htSNPs for the 5¢ block (min R2H 1.000), 2 htSNPs for the 3¢ block (min R2H 1.000) IVS1-242G>T/A - tri-allelic SNP
or compound heterozygous. Analysis confirmed two
distinct haplotype blocks based on the upper and lower
confidence limit of LD (Fig. 1) (Gabriel et al. 2002). The
extended 5¢ block, which is edged by SNP 4466 and
Ex4+5, includes 25 polymorphisms across at least
7.0 kbp. At the 3¢ end four SNPs are linked in a distinct
haplotype block that is flanked by Ex4-1483 and Ex4710. However, a multiallelic D¢ = 0.94 represents the
low level of diversification between the two blocks.
The border of the 5¢ block was extended to include
the 25 common SNPs (from SNP - 4466 to Ex4+5) for
all further analyses, whereas the six SNPs of the ‘secretor
haplotypes’ as well as the tagging SNPs of the two gene
conversion elements are part of this 5¢ haplotype block.
Within the 5¢ block, the EM-algorithm inferred 41 distinct haplotypes from the genotype data of 25 variants;
eight haplotypes had a frequency ‡0.01, which capture
92.7% of all the haplotypes in this block (Table 1). The
seven ‘secretor haplotypes’ (HYPA 0.27, LYPA 0.05,
LYQA 0.23, LXPA 0.21; B-, C- and D-allele see previous para) were observed to lie on restricted haplotypes
without evidence for novel recombination sites in this
population. Only ‘secretor haplotype’ LXPA resides on
two distinct extended 5¢ haplotypes with a frequency
‡0.01; the two SNPs in the putative gene conversion
cassette in exon 4 (SNPs IVS3-28G>C and Ex4+
5G>C) separate the two haplotypes. Furthermore, the
tagging alleles of the second conversion cassette (IVS2630) in MBL2 were exclusively seen in linkage with
specific ‘secretor haplotypes’: IVS2-630A – HYP[A/D],
LYP[A/B]; IVS2-630G – LYQ[A/C], LXPA.
Overall, the haplotype heterozygosity (HET) of the
seven ‘secretor haplotypes’ was only slightly less than the
HET of the 30 extended haplotypes with 25 variants
(HET6 0.8418 versus HET25 0.8414) (Stephens et al.
2001a; Niu et al. 2002). The high haplotype heterozygosity of the seven ‘secretor haplotypes’ reflects strong
linkage disequilibrium for the 25 variants within the 5¢
block. The second haplotype block includes 4 variants,
beginning in the 3¢ UTR. Three major haplotypes were
observed within the 3¢ block accounting for 99.5% of all
observations (HET4 0.5786), TAAA (0.57), CGGG
(0.28) and TGAA (0.15).
Several additional markers are linked to certain
‘secretor haplotypes’ (e.g.
4459delG and SNP
4453T to HYP[A/D], P < 0.0001) or in strong LD
to structural mutations in exon 1 (SNP 1964C to the
B-allele, P < 0.0001; SNPs 3427T and 3201G to
the C-allele, P<0.0001). Furthermore, the three alleles
of IVS1 242 are in linkage with SNPs 289 and 65
(Fisher exact for each allele: P < 0.0001, see Table 1).
For the non-synonymous SNP at position Ex1-34 (i.e.
D-allele), which is a component of the classical
‘secretor haplotypes’, no other marker in strong LD
has been identified so far. Accordingly, in addition to
this SNP six new, alternative ‘‘haplotype tagging’’
variants capture the observed eight haplotypes with a
frequency ‡0.01 inside the 5¢ block (minimal R2H 1.000;
see Table 1). Otherwise, the six traditional markers of
the ‘secretor haplotypes’ are not able to distinguish
between the two extended 5¢ haplotypes, which are
bearing the LXPA markers (LXPA – CC: R2H 0.742;
LXPA – GG: R2H 0.176). The three common 3¢ haplotypes can be tagged completely by two SNPs (SNP
Ex4-1483 and SNP Ex-1067) in this Caucasian population (min. R2H 1.000).
Fig. 1 MBL2—map of the variant sites with comparison of mouse
and human sequence. Analysed polymorphic sites (w/o triallelic
IVS1-242) are mapped on the genomic structure of the MBL2 gene
and its extended 5¢ promoter region (11.0 kbp). Mutation
nomenclatures are presented in regard to the NCBI contig
NM_000242 (NCBI build 35, locus ID 4153) as recommended by
Dunnen JT den and Antonarakis SE (Dunnen et al. 2001)
considering the reverse orientation of MBL2 on chromosome 10.
The VISTA plot between human MBL2 and its mouse homologous
mblC is located above the gene structure. (mblC: chromosome 14;
NW_039687, NCBI build 32, October 2003). Regions with a
conservation of sequence of more than 75% are shaded (window
size: 75 bp). Pair wise LD (ID’I) between all bi-allelic markers is
shown at the bottom of the figure. ID’I values of 1.0 are not
presented. The two haplotype blocks (black triangles) are defined
by D¢ confidence limits (upper confidence limit > 0.97, lower
confidence limit >0.70) using Haploview software (5¢ block: 25
variants; 3¢ block: 4 variants). Color scheme: dark gray strong
evidence of LD, light gray uninformative, white strong evidence of
recombination
Serum MBL levels
haplotypes’ have detected that the promoter polymorphism at position 289 (i.e. 221, Y/X) in cis to a
normal exon 1 (A-allele) has the most significant impact
on the MBL serum level (Madsen et al. 1995, 1998).
Thus, it is justifiable to ease the interpretation by
pooling the structural mutations B, C and D into one 0,
whereas the promoter variant at 289 (Y/X) is only
linked to a normal exon 1 allele (YA or XA) (Garred
et al. 2003). Table 2 shows significantly different MBL
levels for each observed diplotype (Kruskal–Wallis,
P<0.0001). As expected, the highest mean value was
seen for YA/YA with 4.78 (SD 0.59, median 3.52 mg/l),
whereas individuals homozygous or compound heterozygous for one structural mutation had the lowest
measured MBL levels (mean 0.05 mg/l, SD 0.02, median 0.02 mg/l).
The test of haplotype association between MBL
concentration and the eight extended 5¢ haplotypes with
The mean MBL level of all 212 individuals was
2.55 mg/l (SD 0.24, median 1.26 mg/l), which is comparable with the results of previous studies in unselected populations (Terai et al. 1993; Madsen et al.
1995; Minchinton et al. 2002). Supplementary Table 2
shows the measured MBL serum levels according to the
individual genotypes for each variant site. Six-teen
variants (15 SNPs and 1 deletion), including all four tag
SNPs of the 3¢ block, have significantly different MBL
levels for each genotype (P < 0.05 after correction for
multiple comparisons). SNP-618 (i.e. 550, H/L) as
well as the non-synonymous mutations at position Ex127 (B-allele) and Ex1-18 (C-allele) represent the only
markers of the ‘secretor haplotypes’ of which individual
genotypes influence the protein level. In contrast to a
single locus approach, previous analyses of the ‘secretor
Table 2 Mean, median and range of MBL levels stratified according to truncated MBL2 genotypes
Secretor genotype
YA/YA
YA/XA
XA/XA
YA/0
XA/0
0/0
Total
n
64
45
8
57
26
12
212
MBL serum concentration (mg/l)
Mean ± SE
Range
Median
4.78±0.59
2.83±0.34
1.39±0.55
1.58±0.35
0.19±0.05
0.05±0.02
2.55±0.24
0.21–25.73
0.16–8.30
0.05–3.60
0.09–16.54
0.01–1.10
0.01–0.183
0.01–25.73
3.52
2.22
0.46
0.78
0.11
0.02
1.26
The secretor genotypes (diplotypes) are stratified according to the genotype at position 289 (aka-221, Y/X) and the structural mutations
in exon 1, which were pooled in one 0 (A represent a normal coding region). The MBL concentration was measured in the plasma of 212
Dutch Caucasian individuals by ELISA (Kruskal–Wallis, P<0.001)
a frequency >0.01 was analysed using the haplo.stats
software. This haplotype-based approach allows defining an association between select haplotypes and circulating levels directly without preceding reconstruction of
best pairs of haplotypes (i.e. diplotypes); and, in turn it
considers haplotype ambiguity in the statistical model.
The result of the adjusted global score statistics suggests
a strong association between MBL serum levels and
inferred haplotypes (global score value 106.93, 8 df, Pvalue <0.0001; adjusted for age and gender). Consistent
with this result, the haplotype-specific score values in
Table 3 demonstrate that the haplotypes 5¢H-1 and 5¢H5, which are bearing the HYPA and LYQA ‘secretor
haplotypes’, are positively associated with the MBL
serum concentration, whereas the haplotypes 5¢H-4
(‘LYPB’) and 5¢H-6 (‘LYQC’) are negatively associated
with the measured circulating protein levels. To analyse
the association further, measured MBL concentrations
were regressed on the estimated haplotype counts (adjusted for age and gender); and 5¢H-1 (‘HYPA’), the
most common haplotype, was chosen as the required
reference. In the past, numerous studies, including
functional promoter analyses, have shown, that the
‘HYPA – secretor haplotype’, which resides on 5¢H-1
has been associated with high MBL secretion (Madsen
et al. 1995; Naito et al. 1999a). The results shown in
Table 3 indicate that the extended 5¢ haplotypes bearing
an exon 1 mutation (5¢H-2, 5¢H-4, 5¢H-6) as well as the
haplotypes 5¢H-7 and 5¢H-8, on which the LXPA
‘secretor haplotype’ resides, have a significantly lower
MBL level as compared to the chosen reference 5¢H-1.
No difference was seen in the comparison to 5¢H-3
(‘LYPA’) and 5¢H-5 (‘LYQA’). In addition, the interaction model does not yield any significant interaction
between 5¢ haplotypes and gender or age (P>0.05 for
each individual haplotype). Consecutively, extended 5¢
haplotype pairs were reconstructed by selecting the most
likely pair. In Fig. 2 the reconstructed individual diplotypes are plotted against individual MBL serum levels
(eight common haplotypes of the MBL2 5¢ block as
shown in Table 1; all haplotypes with a frequency <0.01
were pooled in one ‘U’). Although the haplotypes
contain an extended number of variants and the MBL
serum concentrations were significantly different
(Kruskal–Wallis, P < 0.001), wide intervals of MBL
were still seen for each established diplotype. Moreover,
in a simplified analysis of the common extended haplotypes considering only the polymorphism at –289 (YA
or XA) and the structural mutations (pooled in one 0),
which reside on these haplotypes, the serum level distribution is similar to the concentrations stratified according to the eased classical ‘secretor haplotypes’ (grayshaded plot areas).
A separate association test for the 3¢ haplotypes was
performed to investigate the possibility of a correlation
between 3¢ haplotypes and the MBL circulating levels.
Consistent with the global score statistic (global score
value 59.58, 3 df, P value <0.0001; adjusted for age and
gender) haplotype-specific score values show a positive
association for haplotype 3¢H-1 (TAAA; score value
7.45, P value <0.0001), whereas haplotype 3¢H-2
(CGGG) is associated with depressed levels (score value
6.72, P value <0.0001). Furthermore, the analysis of
the reconstructed best pair haplotypes for all study individuals suggests also a strong correlation between the
3¢ haplotypes and the circulating levels of the protein
(Fig. 3, Kruskal–Wallis, P <0.001).
To determine if the 3¢ haplotype contributes to MBL
levels, the distribution of the 3¢ haplotypes was stratified
according to the ‘secretor haplotypes’ in the 5¢ end,
which have a well-defined impact on the MBL2 expression (simplified as YA, XA or 0). Table 4 illustrates that
the three distinct 3¢ haplotype were more often seen in
linkage with a particular 5¢ allele: TAAA and YA;
TGAA and XA; CGGG and 0. Moreover, analyses of
the combined extended 5¢ and 3¢ haplotypes detected an
analogous ‘‘biased haplotype usage’’ in MBL2 suggesting that the main effect of the association is linked to the
5¢ haplotype (see Table 3). For instance, haplotype 3H-1
together with 5H-1 is positively associated with the
protein concentration (score value 5.78, P value
<0.0001), whereas the same 3¢ haplotype in linkage to
5¢H-4 bearing the codon-54 mutation (B-allele) is associated with a low MBL level (score value 3.74, P value
= 0.001). Similar to the analysis of the 5¢ haplotype
alone, the association test of the combined haplotypes
revealed a statistically significant difference between the
chosen reference 5¢H-1/3¢H-1 (most common haplotype)
Table 3 MBL2 haplotype association analysis
Haplotype
Secretor
haplotype
fa
Score statisticsb
Value
5 ¢ haplotype
5¢H-1
5¢H-2
5¢H-3
5¢H-4
5¢H-5
5¢H-6
5¢H-7
5¢H-8
Gender
Age
3 ¢ haplotype
3¢H-1
3¢H-2
3¢H-3
Gender
Age
Both
5¢H-1/3¢H-1
5¢H-2/3¢H-1
5¢H-3/3¢H-2
5¢H-4/3¢H-2
5¢H-4/3¢H-1
5¢H-5/3¢H-1
5¢H-6/3¢H-2
5¢H-7/3¢H-3
5¢H-8/3¢H-2
5¢H-7/3¢H-2
Gender
Age
Regression analysis
c
P value
Coefficient
P value
HYPA
HYPD
LYPA
LYPB
LYQA
LYQC
LXPA
LXPA
0.262
0.051
0.038
0.143
0.210
0.027
0.153
0.043
5.8334
0.9724
1.5753
8.1975
4.5158
3.0149
2.2219
0.2068
<0.0001
0.3358
0.1253
<0.0001
<0.0001
0.0032
0.0266
0.8314
Reference
0.5693
0.1644
1.1032
0.0709
0.9548
0.5008
0.4283
0.0065
0.0772
0.0002
0.3480
<0.0001
0.4430
<0.0001
<0.0001
0.0135
0.1180
0.3660
–
–
–
0.575
0.276
0.144
7.4536
6.7217
1.7217
<0.0001
<0.0001
0.0851
Reference
0.6540
0.3965
0.1671
0.0008
<0.0001
<0.0001
0.0736
0.8470
HYPA
HYPD
LYPA
LYPB
LYPB
LYQA
LYQC
LXPA
LXPA
LXPA
0.260
0.045
0.039
0.119
0.021
0.210
0.028
0.139
0.041
0.011
5.7806
0.8366
1.5350
7.1340
3.7346
4.5561
3.0264
2.2272
0.3824
0.7369
<0.0001
0.399
0.127
<0.0001
0.001
<0.0001
0.003
0.025
0.700
0.465
Reference
0.5231
0.1634
1.0562
1.3533
0.0643
0.9263
0.5026
0.4050
0.5576
0.0850
0.0064
0.0009
0.3540
<0.0001
<0.0001
0.4900
<0.0001
<0.0001
0.0202
0.1180
0.3270
0.1370
Haplotype association analysis was performed for the 5¢ and 3¢ MBL2 haplotypes separately as well as for their combinations using
haplo.stats software. The haplotypes in both blocks are labelled according to Table 1. Only inferred haplotypes with a frequency > 0.01
are shown
a
The haplo.score function was used to infer the haplotype frequencies and to perform the score test statistics
b
The simulated P values are based on at least 10,000 simulation and maximal 10,000 simulations; results are adjusted for age and gender
c
The haplo.glm function was applied to both the 5¢ and 3¢ haplotypes as well as to the combined haplotypes using the most common
haplotype as reference (5¢H-1, 3¢H-1 or 5¢H-1/3¢H-1); gender and age were included as covariates. Following six markers of the 5¢
haplotype form the ‘secretor haplotypes’: 618 (aka-550), 289 (aka-221), 65 (aka+4), Ex1-34 (codon 52), Ex1-27 (codon 54) and Ex118 (codon 57)
and all haplotypes bearing an exon 1 mutation (5¢H-2/
3¢H-1, 5¢H-4/3¢H-1, 5¢H-4/3¢H-2, 5¢H-6/3¢H-2) as well as
the haplotypes 5¢H-7/3¢H-3 and 5¢H-8/3¢H-2, on which
the LXPA ‘secretor haplotype’ resides. The applied
statistical model did not reveal any significant interactions with age and gender as covariates (P>0.05 for
each individual combination of 5¢ and 3¢ haplotypes). In
the context of these findings and the observed high value
of LD between both blocks, it is not clear if either the
‘secretor haplotype’ alone accounts for the observed
differences in the MBL serum concentration or particular markers of distinct 3¢ haplotypes contribute partially to circulating levels.
Discussion
In a previous resequencing analysis of the MBL2 gene,
two major points were determined: (1) a high degree of
heterozygosity across the gene, which is divided into two
distinct blocks with a possible recombination hot spot in
the intervening region; and (2) the classical ‘secretor
haplotypes’, which consists of six SNPs residing on a
restricted number of extended haplotypes (Bernig et al.
2004). This study has been followed-up to determine if
there are additional variants, in either block or within
the interpolated possible conversion tracts that could
have functional implications, namely, an effect on circulating MBL levels. In this study, which includes 212
unrelated Dutch Caucasians, the two-block structure
and the restricted pattern of haplotype structure for the
common B, C and D alleles have been confirmed.
Additional analysis of the region in the 5¢ block indicates
that linkage disequilibrium extends upstream and includes a highly conserved region, but maintains the restricted pattern of haplotype diversity.
A dense set of SNP markers were studied to determine if additional markers could be associated with
Fig. 2 MBL serum concentrations (n=212) stratified according to
the extended 5¢ haplotypes of MBL2. The MBL serum concentrations measured by ELISA are stratified according to the observed
diplotypes of the 5¢ extended haplotypes (25 variant sites). The
eight common haplotypes (‡ 0.01; capturing 93% of all inferred
haplotypes) are labelled 1 – 8 according to inferred haplotypes in
Table 1 (gray-shaded area); all rare haplotypes are pooled in one
Fig. 3 MBL serum
concentrations (n=212)
stratified according to the
haplotypes in the 3¢ block of
MBL2. The MBL serum
concentrations measured by
ELISA are stratified according
to the observed diplotypes of
the 3¢ common haplotypes
(‡0.01; capturing 99% of all
inferred haplotypes) TAAA,
CGGG and TGAA (four SNPs:
7825, 8241, 8407, 8598).
Median, interquartile and range
are shown (Kruskal–Wallis,
P < 0.001)
‘‘U’’ (31 individuals). Median, interquartile and range are shown
(Kruskal–Wallis, P < 0.001). The dashed lines indicate the distinct
‘secretor haplotypes’ (diplotypes) whose marker reside on the
extended 5¢ haplotypes; simplified by considering only the 289
polymorphism (YA/XA) and the structural mutations, which are
pooled in one 0 (Kruskal–Wallis, P < 0.001)
Table 4 Distribution of 3¢ block haplotypes in regard to linked truncated ‘secretor haplotypes’
5¢ haplotypea
YA
XA
0
Total
3¢ haplotypeb
3¢H-1 (TAAA)
3¢H-2 (CGGG)
3¢H-3 (TGAA)
204
3
33
240
22
22
73
117
0.19
0.19
0.62
0.276
0.85
0.01
0.14
0.571
other
4
61
–
65
0.06
0.94
–
0.149
1
–
1
2
0.50
–
0.50
0.005
a
5¢ Haplotypes represent ‘secretor haplotypes’, which are presented as truncated promoter haplotypes in cis to a normal coding region
(YA or XA) or as pooled structural mutation (0). Frequencies are shown with regard to the truncated ‘secretor haplotype’ or total number
of chromosome n = 424
b
3¢ Haplotypes are based on following four SNPs: Ex4-1483, Ex4-1067, Ex4-901 and Ex4-710
circulating MBL levels. In an analysis of the effect of the
six SNPs within the ‘secretor haplotype’ on MBL levels
determined by ELISA assays, it was observed that
individuals who were either heterozygous or homozygous or compound heterozygous for the structural
mutations in exon 1 have reduced circulating levels of
MBL (Terai et al. 1993; Madsen et al. 1998; Minchinton
et al. 2002). However, circulating levels of MBL in A/D
donors were significantly higher than in individuals
heterozygous for either the B- or C-allele. It was observed that the promoter allele at 289 (i.e. 221, Y/X)
linked to a normal or variant allele in exon 1 probably
accounts most for circulating MBL levels. In the past,
numerous studies have shown that the correlation between circulating MBL levels and the ‘secretor haplotypes’ is adequate, but could misclassify as many as 15%
(Madsen et al. 1995, 1998; Chanock and Taylor 2002;
Minchinton et al. 2002; Garred et al. 2003). Thus, it is
possible that the additional variants contribute to serum
levels. Although individual genotypes significantly
influence the MBL level in a single locus analysis, the
observed wide range of circulating protein concentrations for each of the known ‘secretor haplotypes’ cannot
be explained by extending the analysis to additional
markers in the 5¢ block because of the strong linkage
disequilibrium between markers in this block. The
inclusion of additional markers in a gene notable for
such a high degree of heterozygosity did not significantly
increase the total number of extended 5¢ haplotypes with
observed frequency ‡0.01; eight distinct haplotypes
capture roughly 93% of haplotypes in the 5¢ block
(Bernig et al. 2004). On the contrary, the SNPs at 618,
Ex1-27 and Ex1-18 show a significant impact on the
measured serum MBL in a separate analysis for each
marker of the ‘secretor haplotypes’. Although these
variants account for the extreme variations in the level, a
study that ignores the restricted 5¢ haplotype pattern
would misclassify the effect of single variants on the
backbone of different haplotypes (e.g. 289: YA vs Y0)
(Madsen et al. 1995, 1998; Minchinton et al. 2002;
Garred et al. 2003). In this regard it is noteworthy that
association studies on the basis of haplotypes turned out
to be more powerful than individual SNP analyses
(Morris and Kaplan 2002; Chapman et al. 2003; Clark
2004). However, it was recognized that the sample size
may not be large enough to detect smaller effects nor to
determine a possible impact on the MBL level by rare 5¢
block haplotypes (frequency <0.01). This study does
not provide evidence that the previously described gene
conversion elements in MBL2 are associated with circulating levels of MBL.
The investigation of the functional consequences of
additional variants in the 5¢ haplotypes block could be
informative in elucidating regulatory elements in the
MBL2 promoter region. In previous analyses of the
functional promoter elements, it was shown that
the variant at position 289 (i.e. 221; Y/X) downregulates the protein concentration; and, in turn the
presence of a negatively controlling cis-acting element in
this region has been suggested (Naito et al. 1999a). The
previous resequencing study together with the current
results demonstrate that the SNPs of the ‘secretor
haplotypes’ reside on distinct extended 5¢ haplotypes,
which could be in strong LD with additional promoter
or intronic variants that possess functional effects. The
three variant alleles in codons 52, 54 and 57, directly
affect the folding and stability of the protein in the
homozygous or compound heterozygous situation,
respectively and lead to a low oligomerized protein that
does not activate complement (Larsen et al. 2004).
Diminished serum concentration can be partly explained
by a reduced half-life for the B-allele transcript (codon
54) (Naito et al. 1999b). On the other hand, the higher
MBL serum concentration in A/D donors and the observed higher capacity to activate complement by D
variant chains (codon 52) suggest that oligomerization
into higher order structures is optimal in the presence of
the normal allele A (Minchinton et al. 2002; Garred
et al. 2003). The data reveal that the mutations in codon
54 (B-allele) and 56 (C-allele) are in LD with additional
promoter variants, which do not occur on the ancestral
normal alleles. Interestingly, the mutation at codon 52
(D-allele) represents the only difference between the extended 5¢ haplotype bearing this variant and the normal
allele. Except for the D-allele, the classical markers of
the ‘secretor haplotypes’ can be completely replaced
with linked variants to tag common haplotypes. These
findings raise the question, whether the intensively
studied variants at positions 618 (i.e. 550), 289 (i.e.
221) and 65 (i.e. +4) are functionally important or
could be in linkage with additional regulative elements
not yet characterized. Furthermore, the additional
variants, which are strongly linked to either B or C could
contribute to reduced circulating level in individuals
carrying at least one copy of these alleles.
An important assumption for intervening recombination hot spots is that they contribute to the generation
of separate haplotype blocks (Phillips et al. 2003; Wall
and Pritchard 2003; Crawford et al. 2004). Recently, a
recombination hot spot in MBL2 that divides the gene
into two distinct blocks was described (Bernig et al.
2004). In the current study the observed level of
recombination between the two haplotype blocks was
apparent, but with a degree of association between 5¢
and 3¢ blocks. In the context of this finding, it is possible
that the 3¢ haplotypes could contribute to serum levels,
but at this time, it is still speculative whether the 3¢ block
has an independent impact on the MBL concentration.
In an unrooted haplotype network it was observed that
numerous mutation steps are required to separate the
backbone haplotypes for the structural mutations on
exon 1 (see Fig. 4). However, it was also noted that the
‘secretor haplotypes’ bearing the B- and C-alleles are
strongly linked to the same 3¢ haplotype (CGGG). The
fact that two distinct 5¢ haplotypes, which have very
similar functional consequences occur at different frequencies in human populations, suggest that there could
be an independent selective advantage for diversification
of the 3¢ region of the MBL2 gene, perhaps altering
regulation of translation and mRNA stability. Thus, it is
possible that both linked 5¢ variants and the CGGG 3¢
haplotype contribute towards reducing serum levels,
especially in individuals homozygous or heterozygous
for B or C. However, the small number of observations
for the latter indicates that the observation of the effect
of 3¢ haplotypes on circulating levels is preliminary.
In summary, our results have confirmed and extended
previous findings that the MBL2 gene is divided into two
distinct haplotype blocks by a possible recombination
hot spot. The variant sites inside the two blocks
Fig. 4 Estimated haplotypes network of the 5¢ haplotype block in
MBL2. The unrooted haplotype network (statistical parsimony
network) was estimated under the null hypothesis of no recombination inside the 5¢ block. Twenty-three variants were included in
the analysis (the 1-bp in/del and the tri-allelic SNP IVS1-242 were
excluded). The eight common haplotypes of the 5¢ block in MBL2
(5¢H) are indicated by a number in regard to Table 1. Small circles
indicate intermediate haplotype states, which were not found in the
population. Additionally, haplotypes are clustered and labelled
according to the marker of the ‘secretor haplotypes’. The observed
frequencies of the linked three common 3¢ haplotypes (TAAA,
CGGG and TGAA) are presented next to each particular 5¢
haplotype (n=424 chromosomes). The dashed line between 5¢H-8
and 5¢H-7 indicates an ambiguous branch, which is more likely
caused by gene conversion as by independent mutation events
(Bernig et al. 2004)
including the six markers of the well-studied seven
‘secretor haplotypes’ were in strong LD to each other.
Further work is required to determine the functional
impact of the extended 5¢ haplotypes, which might include additional regulative elements. The results provide
a powerful tool for more sophisticated functional promoter analyses in the future. Moreover, in spite of
moderate diversification, it is possible that select 3¢
variants could be selected on the basis of functional effects, such as modifying translation or mRNA stability.
Future studies examining larger data sets are needed to
assess the contribution of additional variants in MBL2,
in both 5¢ and 3¢ blocks, for infections and autoimmune
outcomes.
Acknowledgments The authors are grateful to S. Berndt for her
helpful assistance in the computation of the haplotype association
tests. We thank E. Tarazona-Santos, M. Yeager, B. Staats, B.
Packer, S. Savage, J.G. Taylor VI. and H.C. Erickson for discussion of the data; M. Kiley, E. Schatzkin, M. Brown, A. Eckert, A.
Crenshaw and D. Kressley for their excellent technical assistance.
We also want to thank the two anonymous reviewers for their
critical comments.
References
Barrett JC, Fry B, Maller J, Daly MJ (2004) Haploview: analysis
and visualization of LD and haplotype maps. Bioinformatics
21:263–265
Bernig T, Taylor JG, Foster CB, Staats B, Yeager M, Chanock SJ
(2004) Sequence analysis of the mannose-binding lectin (MBL2)
gene reveals a high degree of heterozygosity with evidence of
selection. Genes Immun 5:461–476
Besag J, Clifford P (1991) Sequential Monte Carlo p-values. Biometrika 78:301–304
Biezeveld MH, Kuipers IM, Geissler J, Lam J, Ottenkamp JJ, Hack
CE, Kuijpers TW (2003) Association of mannose-binding lectin
genotype with cardiovascular abnormalities in Kawasaki disease. Lancet 361:1268–1270
Bray N, Dubchak I, Pachter L (2003) AVID: a global alignment
program. Genome Res 13:97–102
Chanock S, Taylor JG (2002) Using genetic variation to study
immunomodulation. Curr Opin Pharmacol 2:463–469
Chapman JM, Cooper JD, Todd JA, Clayton DG (2003) Detecting
disease associations due to linkage disequilibrium using haplotype tags: a class of tests and the determinants of statistical
power. Hum Hered 56:18–31
Clark AG (2004) The role of haplotypes in candidate gene studies.
Genet Epidemiol 27:321–333
Clement M, Posada D, Crandall KA (2000) TCS: a computer
program to estimate gene genealogies. Mol Ecol 9:1657–1659
Crawford DC, Bhangale T, Li N, Hellenthal G, Rieder MJ,
Nickerson DA, Stephens M (2004) Evidence for substantial
fine-scale variation in recombination rates across the human
genome. Nat Genet 36:700–706
Crosdale DJ, Ollier WE, Thomson W, Dyer PA, Jensenious J,
Johnson RW, Poulton KV (2000) Mannose binding lectin
(MBL) genotype distributions with relation to serum levels in
UK Caucasoids. Eur J Immunogenet 27:111–117
Dubchak I, Brudno M, Loots GG, Pachter L, Mayor C, Rubin
EM, Frazer KA (2000) Active conservation of noncoding sequences revealed by three-way species comparisons. Genome
Res 10:1304–1306
Dunnen Jd, Antonarakis S (2001) Nomenclature for the description
of human sequence variations. Hum Genet 109:121–124
Eisen DP, Minchinton RM (2003) Impact of mannose-binding
lectin on susceptibility to Infectious Diseases. Clin Infect Dis
37:1496–1505
Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, Higgins J, DeFelice M, Lochner A, Faggart M, LiuCordero SN, Rotimi C, Adeyemo A, Cooper R, Ward R,
Lander ES, Daly MJ, Altshuler D (2002) The structure of
haplotype blocks in the human genome. Science 296:2225–2229
Garred P, Larsen F, Madsen HO, Koch C (2003) Mannose-binding
lectin deficiency—revisited. Mol Immunol 40:73–84
Ikeda K, Sannoh T, Kawasaki N, Kawasaki T, Yamashina I (1987)
Serum lectin with known structure activates complement
through the classical pathway. J Biol Chem 262:7451–7454
Jack DL, Turner MW (2003) Anti-microbial activities of mannosebinding lectin. Biochem Soc Trans 31:753–757
Kilpatrick DC (2002) Mannan-binding lectin: clinical significance
and applications. Biochim Biophys Acta 1572:401–413
Kuhlman M, Joiner K, Ezekowitz RA (1989) The human mannosebinding protein functions as an opsonin. J Exp Med 169:1733–
1745
Lake SL, Lyon H, Tantisira K, Silverman EK, Weiss ST, Laird
NM, Schaid DJ (2003) Estimation and tests of haplotypeenvironment interaction when linkage phase is ambiguous.
Hum Hered 55:56–65
Larsen F, Madsen HO, Sim RB, Koch C, Garred P (2004) Diseaseassociated mutations in human mannose-binding lectin compromise oligomerisation and activity of the final protein. J Biol
Chem 279: 21302–21311
Lipscombe RJ, Sumiya M, Hill AV, Lau YL, Levinsky RJ, Summerfield JA, Turner MW (1992) High frequencies in African
and non-African populations of independent mutations in the
mannose binding protein gene. Hum Mol Genet 1:709–715
Madsen HO, Garred P, Kurtzhals JA, Lamm LU, Ryder LP, Thiel
S, Svejgaard A (1994) A new frequent allele is the missing link
in the structural polymorphism of the human mannan-binding
protein. Immunogenetics 40:37–44
Madsen HO, Garred P, Thiel S, Kurtzhals JA, Lamm LU, Ryder
LP, Svejgaard A (1995) Interplay between promoter and
structural gene variants control basal serum level of mannanbinding protein. J Immunol 155:3013–3020
Madsen HO, Satz ML, Hogh B, Svejgaard A, Garred P (1998)
Different molecular events result in low protein levels of mannan-binding lectin in populations from southeast Africa and
South America. J Immunol 161:3169–3175
Mayor C, Brudno M, Schwartz JR, Poliakov A, Rubin EM, Frazer
KA, Pachter LS, Dubchak I (2000) VISTA : visualizing global
DNA sequence alignments of arbitrary length. Bioinformatics
16:1046–1047
Minchinton RM, Dean MM, Clark TR, Heatley S, Mullighan CG
(2002) Analysis of the relationship between mannose-binding
lectin (MBL) genotype, MBL levels and function in an Australian blood donor population. Scand J Immunol 56:630–641
Morris RW, Kaplan NL (2002) On the advantage of haplotype
analysis in the presence of multiple disease susceptibility alleles.
Genet Epidemiol 23:221–233
Naito H, Ikeda A, Hasegawa K, Oka S, Uemura K, Kawasaki N,
Kawasaki T (1999a) Characterization of human serum mannan-binding protein promoter. J Biochem (Tokyo) 126:1004–
1012
Naito H, Ma Y, Uemura K, Arano Y, Kawasaki T (1999b) Metabolic properties of normal and mutant mannan-binding proteins in mouse plasma. Biochem Biophys Res Commun
256:231–234
Niu T, Qin ZS, Xu X, Liu JS (2002) Bayesian haplotype inference
for multiple linked single-nucleotide polymorphisms. Am J
Hum Genet 70:157–169
Packer BR, Yeager M, Staats B, Welch R, Crenshaw A, Kiley M,
Eckert A, Beerman M, Miller E, Bergen A, Rothman N,
Strausberg R, Chanock SJ (2004) SNP500Cancer: a public
resource for sequence validation and assay development for
genetic variation in candidate genes. Nucleic Acids Res
32:D528-D532
Phillips MS, Lawrence R, Sachidanandam R, Morris AP, Balding
DJ, Donaldson MA, Studebaker JF, Ankener WM, Alfisi SV,
Kuo FS, Camisa AL, Pazorov V, Scott KE, Carey BJ, Faith J,
Katari G, Bhatti HA, Cyr JM, Derohannessian V, Elosua C,
Forman AM, Grecco NM, Hock CR, Kuebler JM, Lathrop JA,
Mockler MA, Nachtman EP, Restine SL, Varde SA, Hozza
MJ, Gelfand CA, Broxholme J, Abecasis GR, Boyce-Jacino
MT, Cardon LR (2003) Chromosome-wide distribution of
haplotype blocks and the role of recombination hot spots. Nat
Genet 33:382–387
Rozen, Skaletsky HJ (2000) Primer3 on the WWW for general
users and for biologist programmers. In: Krawetz S, Misener S
(eds) Bioinformatics methods and protocols: methods in
molecular biology. Humana Press, Totowa, pp 365–386
Schaid DJ, Rowland CM, Tines DE, Jacobson RM, Poland GA
(2002) Score tests for association between traits and haplotypes
when linkage phase is ambiguous. Am J Hum Genet 70:425–434
Soborg C, Madsen HO, Andersen AB, Lillebaek T, Kok-Jensen A,
Garred P (2003) Mannose-binding lectin polymorphisms in
clinical tuberculosis. J Infect Dis 188:777–782
Stephens M, Donnelly P (2003) A comparison of bayesian methods
for haplotype reconstruction from population genotype data.
Am J Hum Genet 73:1162–1169
Stephens JC, Schneider JA, Tanguay DA, Choi J, Acharya T,
Stanley SE, Jiang R, Messer CJ, Chew A, Han JH, Duan J,
Carr JL, Lee MS, Koshy B, Kumar AM, Zhang G, Newell WR,
Windemuth A, Xu C, Kalbfleisch TS, Shaner SL, Arnold K,
Schulz V, Drysdale CM, Nandabalan K, Judson RS, Ruano G,
Vovis GF (2001a) Haplotype variation and linkage disequilibrium in 313 human genes. Science 293:489–493
Stephens M, Smith NJ, Donnelly P (2001b) A new statistical
method for haplotype reconstruction from population data.
Am J Hum Genet 68:978–989
Stram DO, Haiman CA, Hirschhorn JN, Altshuler D, Kolonel LN,
Henderson BE, Pike MC (2003a) Choosing haplotype-tagging
SNPS based on unphased genotype data using a preliminary
sample of unrelated subjects with an example from the multiethnic cohort study. Hum Hered 55:27–36
Stram DO, Leigh PC, Bretsky P, Freedman M, Hirschhorn JN,
Altshuler D, Kolonel LN, Henderson BE, Thomas DC (2003b)
Modeling and E-M estimation of haplotype-specific relative
risks from genotype data for a case-control study of unrelated
individuals. Hum Hered 55:179–190
Sumiya M, Super M, Tabona P, Levinsky RJ, Arai T, Turner MW,
Summerfield JA (1991) Molecular basis of opsonic defect in
immunodeficient children. Lancet 337:1569–1570
Takahashi K, Gordon J, Liu H, Sastry KN, Epstein JE, Motwani
M, Laursen I, Thiel S, Jensenius JC, Carroll M, Ezekowitz RA
(2002) Lack of mannose-binding lectin-A enhances survival in a
mouse model of acute septic peritonitis. Microbes Infect 4:773–
784
Templeton AR, Crandall KA, Sing CF (1992) A cladistic analysis
of phenotypic associations with haplotypes inferred from
restriction endonuclease mapping and DNA sequence data. III.
Cladogram estimation. Genetics 132:619–633
Terai I, Kobayashi K, Fujita T, Hagiwara K (1993) Human serum
mannose binding protein (MBP): development of an enzymelinked immunosorbent assay (ELISA) and determination of
levels in serum from 1085 normal Japanese and in some body
fluids. Biochem Med Metab Biol 50:111–119
Thiel S, Vorup-Jensen T, Stover CM, Schwaeble W, Laursen SB,
Poulsen K, Willis AC, Eggleton P, Hansen S, Holmskov U,
Reid KB, Jensenius JC (1997) A second serine protease associated with mannan-binding lectin that activates complement.
Nature 386:506–510
Wall JD, Pritchard JK (2003) Assessing the performance of the
haplotype block model of linkage disequilibrium. Am J Hum
Genet 73:502–515
Wallis R (2002) Structural and functional aspects of complement
activation by mannose-binding protein. Immunobiology
205:433–445
Weir BS (1996) Disequilibrium. In: Genetic daya analysis II:
methods for discrete population genetic data. Sinaur Associates, Sunderland, pp 91–139