A Major Histocompatibility Class I Locus Contributes to
Multiple Sclerosis Susceptibility Independently from
Bruce A. C. Cree1*, John D. Rioux2, Jacob L. McCauley3, Pierre-Antoine F. D. Gourraud1, Philippe
Goyette2, Joseph McElroy1, Philip De Jager4,5, Adam Santaniello1, Timothy J. Vyse6, Peter K. Gregersen7,
Daniel Mirel8, David A. Hafler9,10, Jonathan L. Haines10, Margaret A. Pericak-Vance11, Alastair
Compston12, Stephen J. Sawcer12, Jorge R. Oksenberg1, Stephen L. Hauser1, IMAGEN"a, IMSGC"b
1Department of Neurology, University of California San Francisco, San Francisco, California, United States of America, 2Montreal Heart Institute, Montre ´al, Que ´bec,
Canada, 3Dr. John T. MacDonald Foundation, Department of Human Genetics, Miami University School of Medicine, Miami, Florida, United States of America,
4Department of Neurology, Brigham and Women’s Hospital, Boston, Massachusetts, United States of America, 5Partners Healthcare Center for Genetics and Genomics,
Harvard Medical School, Cambridge, Massachusetts, United States of America, 6Department of Rheumatology, Hammersmith Hospital, Imperial College London, London,
United Kingdom, 7Robert S. Boas Center for Genomics and Human Genetics, Feinstein Institute for Medical Research, Manhasset, New York, United States of America,
8Broad Institute of MIT and Harvard, Cambridge, Massachusetts, United States of America, 9Department of Neurology, Yale University School of Medicine, New Haven,
Connecticut, United States of America, 10Department of Molecular Physiology and Biophysics, Vanderbilt University Medical Center, Nashville, Tennessee, United States
of America, 11Hussman Institute for Human Genomics, Miami University School of Medicine, Miami, Florida, United States of America, 12Neurology Unit, Department of
Clinical Neurosciences, University of Cambridge, Cambridge, United Kingdom
Background: In Northern European descended populations, genetic susceptibility for multiple sclerosis (MS) is associated
with alleles of the human leukocyte antigen (HLA) Class II gene DRB1. Whether other major histocompatibility complex
(MHC) genes contribute to MS susceptibility is controversial.
Methodology/Principal Findings: A case control analysis was performed using 958 single nucleotide polymorphisms (SNPs)
spanning the MHC assayed in two independent datasets. The discovery dataset consisted of 1,018 cases and 1,795 controls
and the replication dataset was composed of 1,343 cases and 1,379 controls. The most significantly MS-associated SNP in
the discovery dataset was rs3135391, a Class II SNP known to tag the HLA-DRB1*15:01 allele, the primary MS susceptibility
allele in the MHC (O.R.=3.04, p,1610278). To control for the effects of the HLA-DRB1*15:01 haplotype, case control analysis
was performed adjusting for this HLA-DRB1*15:01 tagging SNP. After correction for multiple comparisons (false discovery
rate=.05) 52 SNPs in the Class I, II and III regions were significantly associated with MS susceptibility in both datasets using
the Cochran Armitage trend test. The discovery and replication datasets were merged and subjects carrying the HLA-
DRB1*15:01 tagging SNP were excluded. Association tests showed that 48 of the 52 replicated SNPs retained significant
associations with MS susceptibility independently of the HLA-DRB1*15:01 as defined by the tagging SNP. 20 Class I SNPs
were associated with MS susceptibility with p-values #161028. The most significantly associated SNP was rs4959039, a SNP
in the downstream un-translated region of the non-classical HLA-G gene (Odds ratio 1.59, 95% CI 1.40, 1.81, p=8.45610213)
and is in linkage disequilibrium with several nearby SNPs. Logistic regression modeling showed that this SNP’s contribution
to MS susceptibility was independent of the Class II and Class III SNPs identified in this screen.
Conclusions: A MHC Class I locus contributes to MS susceptibility independently of the HLA-DRB1*15:01 haplotype.
Citation: Cree BAC, Rioux JD, McCauley JL, Gourraud P-AFD, Goyette P, et al. (2010) A Major Histocompatibility Class I Locus Contributes to Multiple Sclerosis
Susceptibility Independently from HLA-DRB1*15:01. PLoS ONE 5(6): e11296. doi:10.1371/journal.pone.0011296
Editor: Christoph Kleinschnitz, Julius-Maximilians-Universita ¨t Wu ¨rzburg, Germany
Received March 12, 2010; Accepted June 4, 2010; Published June 25, 2010
Copyright: ? 2010 Cree et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted
use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported primarily by a grant from the National Institutes of Allergy and Infectious Diseases (AI067152). Additional support was
received from the National Institute of Neurological Disease and Stroke (NS21799) to SLH and (K23 NS048869) to BACC. PLD was supported by a Harry Weaver
Neuroscience Scholar Award of the National Multiple Sclerosis Society (NMSS). The IMSGC is supported by RO1NS049477. The authors also thank the NMSS and
Nancy Davis Foundation for support of DNA collections. They acknowledge use of DNA from the British 1958 Birth Cohort collection (D. Strachan, S. Ring, W.
McArdle and M. Pembrey), funded by the Medical Research Council grant G0000934 and Wellcome Trust grant 068545/Z/02. The Broad Institute Center for
Genotyping and Analysis is supported by grant U54 RR020278 from the National Center for Research Resources. The funders had no role in study design, data
collection, and analysis, decision to publish, or preparation in the manuscript.
Competing Interests: The authors have declared that no competing interests exist.
* E-mail: email@example.com
"a Membership of the International MHC and Autoimmunity Genetics Network (IMAGEN) is provided in the Acknowledgments.
"b Membership of the International Multiple Sclerosis Genetics Consortium (IMSGC) is provided in the Acknowledgments.
PLoS ONE | www.plosone.org1June 2010 | Volume 5 | Issue 6 | e11296
The autoimmune disease multiple sclerosis (MS) is one of the
leading causes of neurological disability in young adults.
Pathologically the disease is characterized by focal areas of
inflammation and demyelination (plaques) within the central
nervous system with ensuing axonal damage. Although the
etiology is not fully understood, MS is a complex genetic disorder
and whole genome studies indicate that the major histocompat-
ibility complex (MHC) on chromosome 6p21 represents the
strongest genome-wide MS susceptibility locus [1,2].
In both Northern European and African descended popula-
tions, MS susceptibility is associated with alleles of the HLA Class
II gene DRB1 [2–5] whereas the contribution of other genes
within the extended MHC has been controversial [6–8].
Extensive linkage disequilibrium (LD) operating in the region
[9–11], as well as marked polymorphism and high gene density,
have complicated efforts to fully resolve the roles of HLA and
non-HLA genes in MS susceptibility. Due to these inherent
challenges, a comprehensive approach is needed to refine the
contributions of the MHC to genetic risk for MS that includes a
large and well-characterized dataset, dense concentration of
markers, and appropriate methods to control for the extensive
LD across the region.
A panel of single nucleotide polymorphisms (SNPs) selected for
moderate LD across the 29 to 34 Mb region of the MHC was
employed to map both HLA and non-HLA disease susceptibility
signals . Here we present the results of an analysis of two
independent case control MS datasets using 958 SNPs adjusting
for the effect of HLA-DRB1*15:01 whose extended haplotype
spans the MHC.
Case control study
Following quality control, 958 markers were genotyped in both
datasets. In the discovery dataset the average number (standard
deviation) of missing genotypes for cases was .0040 (.0331) and for
controls was .0027 (.0325). In the replication dataset, the average
number (standard deviation) of missing genotypes for cases was
.0020 (.0060) for controls was .0022 (.0080). There was not a
statistically significant difference in missing genotypes between
cases and controls in either dataset.
Case control analysis was performed in the discovery dataset
composed of 1018 cases and 1795 controls (Table S1) using 958
MHC spanning SNPs (Table S2, see Figure S1 for study design).
Population stratification effects were controlled for by including
sex and location of subject recruitment (United States versus
United Kingdom) in the regression analyses. The Cochran
Armitage trend test was used to identify MS associated SNPs
and the false discovery rate (FDR=.05) was used to correct for
multiple comparisons . The most highly associated SNP was
rs3135391 (odds ratio=3.04, p,1610278), a Class II SNP known
to tag the primary MS susceptibility allele HLA-DRB1*15:01 with
very high sensitivity and specificity .
Using the trend test in the discovery dataset, a total of 501 SNPs
in Class I, II and III regions showed statistically significant
association with MS susceptibility; most of these associations were
likely due to LD within extended haplotypes, particularly the one
anchored by the HLA-DRB1*15:01 allele (Figure1A). To correct
for the effect of this haplotype, the trend test was performed
adjusting for rs3135391 using the 958 SNPs (FDR=.05) and the
number of significantly associated SNPs was reduced to 87
A second independent dataset consisting of 1343 cases and 1379
controls was then used to replicate these associations (Table S1).
All 958 markers were assessed in the replication dataset with the
same association strategy adjusting for the HLA-DRB1*15:01
tagging SNP rs3135391 (FDR=.05). Only markers that were
significantly associated in both cohorts, and had the same direction
of association, were studied further. 52 such SNPs were
significantly associated with MS susceptibility in both datasets
The merged HLA-DRB1*15:01(-) dataset
A merged cohort was next created by combining the discovery
and replication datasets. The MAF for each SNP is reported for
cases and controls in the merged dataset as well as the strength of
association using the trend test (Table S3). A SNP in the
downstream non-coding region of HLA-G (rs4959039) was the
most significantly associated marker (p,8.65610212) in the
merged cohort analysis, after adjusting for the HLA-DRB1*15:01
tagging SNP rs3135391 and potential stratification effects caused
by sex, location (US versus UK), and dataset (discovery versus
To further demonstrate that these 52 replicated SNP associa-
tions were independent from effects of the extended HLA-
DRB1*15:01 haplotype, all subjects carrying at least one copy of
this allele, as defined by the tagging SNP rs3135391, were dropped
from the merged dataset to create a ‘‘HLA-DRB1*15:01(-)’’
dataset. This excluded a total of 2088 subjects (1277 cases and
811 controls) leaving a HLA-DRB1*15:01(-) dataset that consisted
of 1075 cases and 2363 controls. Association tests were performed
in this merged HLA-DRB1*15:01(-) dataset and significant
associations were found for 48 of the 52 SNPs identified in the
case control screens including all previously identified Class I and
Class III SNPs (Table S4).
Using the genotypetestfor
DRB1*15:01(-) dataset, 20 Class I SNPs had p-values #1028
(Table 1). The HLA-G linked rs4959039:A.G allele (rs4959039)
continued to have the strongest association in this HLA-
DRB1*15:01(-) dataset (odds ratio 1.59, 95% confidence intervals
1.40, 1.81, p,8.45610213). Importantly, rs4959039 and the other
Class I SNPs associated with MS susceptibility are poorly
correlated with the SNPs in the Class III and Class II regions as
illustrated by the LD map (Figure 2). For example, the average
(range) r2for rs4959039 with the Class III SNPs was .081 (.014
to .149) and for the Class II SNPs was .024 (.002 to .085).
In contrast to the poor correlations with the Class III and Class
II SNPs, the LD map (Figure 2) shows that some of the associated
Class I SNPs are closely linked. SNPs in the Class I region with
p-values #161028that are in moderate to strong LD with each
other (as defined by LD-R2$0.5) include: rs2523822, rs2517701
(HLA-80), rs4713270 (HCG2PG), rs4713274 (MICD), rs2523946
(MICD), rs3823355 (MICD), rs4959039 (in between HLA-G and
HLA-A), rs4713281 (HLA-J), rs9357092 (HCG9), and rs9393989
(RNF39). Using an algorithm to define haplotype blocks by LD-
R2$0.5 an apparently separate Class I SNP cluster (rs1362126,
rs2523393, rs2743951) emerges that includes a tagging SNP for
the HLA-B*44:02 allele (rs2523393), a recently identified MS
protective allele [12,14].
Tests for independent association using logistic
To confirm that the contribution to MS susceptibility of the
rs4959039 SNP was independent of any residual Class II
associations, logistic regression models were constructed. Because
many of the 48 SNPs associated with MS susceptibility in the HLA-
A MHC Class I MS Locus
PLoS ONE | www.plosone.org 2 June 2010 | Volume 5 | Issue 6 | e11296
Figure 1. Association test results for 958 SNPs spanning the MHC in the discovery dataset are shown. The location of the SNPs is
depicted on the X-axis and the statistical significance of the association is depicted on the Y-axis. A: Discovery dataset (1018 cases and 1795 controls),
958 common SNP subset, FDR=.05, adjusted for sex and center (US versus UK), trend test. B: Discovery dataset, 958 common SNP subset, FDR=.05,
adjusted for the HLA-DRB1*15:01 tagging SNP rs3135391), sex and center (US versus UK), trend test.
A MHC Class I MS Locus
PLoS ONE | www.plosone.org3 June 2010 | Volume 5 | Issue 6 | e11296
DRB1*15:01(-) dataset are in moderate to strong LD with each
other a clustering algorithm was used to group the 48 SNPs into 20
clusters (LD-R2$.05) and identify SNPs that tagged each cluster
(Table S5) .
Logistic regression with backwards stepwise selection was then
used with the 20 tagged SNPs and covariates to control for
population stratification, i.e. sex and dataset (discovery versus
replication). Using the trend model, the rs4959039 SNP was
significantly associated with MS susceptibility (p=3.70610210,
odds ratio =1.54), despite controlling for the cumulative effects of
Class II SNPs. Further logistic regression modeling showed that
the rs4959039 MS association was also independent of the Class
III associated SNPs. When the Class II and Class III SNPs were
included in the logistic regression model, the rs4959039 SNP
retained a highly significant association with MS susceptibility
(p=9.70610210, odds ratio =1.52).
To estimate the contributions of the 20 Class I, II and III SNP
clusters to MS susceptibility a model was constructed entering all
20 SNPs, plus covariates to control for stratification effects.
Backwards stepwise selection was used to refine the model so that
only variables with p-values #.01 were retained in the model. In
the final model, SNP rs4959039 maintained the most statistically
significant contribution (p,4.80610210, odds ratio=1.53). Three
Class II SNPs rs3132963 (p,1.5961025, odds ratio=1.65),
(p,.00125, odds ratio=1.28) were retained in the model
suggesting residual independent Class II contributions. The area
under the receiver operator curve for this model was .634 whereas
the area under the receiver operator curve modeling the
rs4959039 SNP alone was .617 showing that the contribution of
these Class II SNPs is modest. Importantly, during the backward
stepwise selection process all other Class I SNP clusters were
dropped from the model suggesting that the Class I contribution to
MS susceptibility is driven by the SNP cluster tagged by
Logistic regression was used to determine whether the
association of the rs4959039 SNP was dependent on the
rs2523393 SNP (tags HLA-B*44:02). Despite their close physical
proximity, the association of the rs4959039 SNP remained highly
significant (p=6.1061026, odds ratio =1.43) after adjusting for
the effect of the rs2523393 SNP whereas the association of the
rs2523393 SNP (that tags the MS protective allele HLA-B*44:02)
was attenuated (p=.015, odds ratio =.85).
Two-locus Class I haplotypes
To further understand the contributions of these loci to MS
susceptibility two-locus haplotypes were constructed for SNPs
rs2523393 (the HLA-B*44:02 tagging SNP) and rs4959039 (Table
S6). This analysis defined a MS risk haplotype as rs2523393:T.C
with rs45959039:A.G and the converse MS protective haplotype
as rs2523393:C.T with rs45959039:G.A. Due to LD this
analysis could not definitively prove that the influence of these loci
on MS risk was independent. However, the heterozygous
haplotype appears to be protective for MS risk (odds ratio=.71,
p,9.7361025) indicating that the protective haplotype is
Table 1. SNPs associated with MS susceptibility with genome-wide statistical significance in the merged dataset excluding all
subjects who carry the HLA-DRB1*15:01 allele listed in order of highest to lowest statistical significance using the Cochran-Armitage
trend test for association.
Merged Cohort HLA-DRB1*15:01(-) Subjects
MS AssociatedTrendOdds 95% CI
SNPPositionClassGene AlleleP Values Ratio LowerUpper
rs4959039 30065047 Class IHLA-GA 8.45610213
rs9393989 30148062Class I RNF39A 9.84610213
rs471327430045471Class IMICDC 5.11610212
rs173693629902295 Class IHCG4P8C2.22610211
0.70 0.63 0.78
rs382335530050061Class I MICDC7.74610211
rs273497129942427 Class I3.8–1.4C2.07610210
rs2239530 30260093Class ITRIM26C2.96610210
rs252339329813637Class IFLJ35429C 6.04610210
rs225626629740296Ext Cls I MOGA2.6761029
rs2743951 29817212Class IFLJ35429C3.5561029
rs251770130033950 Class IHLA-80A5.6261029
rs2523946 30049921Class IMICDC8.6961029
rs2256543 30045811Class IMICDA9.1561029
A MHC Class I MS Locus
PLoS ONE | www.plosone.org4June 2010 | Volume 5 | Issue 6 | e11296
Transmission disequilibrium test in HLA-DRB1*15:01(-) trio
As an additional test of association, the rs4959039 was assessed
using the transmission disequilibrium test in a subset of the
discovery dataset for whom parental genotyping was available. 347
trio families (affected individual plus both parents) that did not
carry the HLA-DRB1*15:01 allele were genotyped for the
rs4959039 SNP. The chromosome carrying the allele of
rs4959039:A.G was transmitted 112 times and not transmitted
81 times in heterozygous trio families. Despite the small size of this
Figure 2. LD map and associations for the 48 SNPs in the merged dataset that excludes all HLA-DRB1*15:01 subjects. Each SNP’s
position in the MHC is shown on the X-axis with the most telomeric SNPs on the left and the most centromeric SNPs on the right. The lower portion
of the figure depicts the strength of LD is in intensity from black to grey to white. Multiple SNPs in the Class I region associated with MS susceptibility
independently from HLA-DRB1*15:01 are in moderate to strong LD with each other. These SNPs are in much weaker LD with the MS associated SNPs
in the Class III and Class II regions. The degree of statistical significance is depicted in the upper portion of the figure where each SNP’s –log10
transformed p-value is depicted on the Y-axis. The most significant associations with MS susceptibility are in the Class I region and the Class III and
Class II signals, although statistically significant, are considerably weaker. An algorithm used to cluster SNPs based on LD-R2 grouped together
SNPs in the Class III NOTCH4 gene (rs2071285, rs206015, rs384247) and Class II gene TSBP (rs9268148, rs3132958, rs3129904, rs3132963, rs2050191).
A MHC Class I MS Locus
PLoS ONE | www.plosone.org5 June 2010 | Volume 5 | Issue 6 | e11296
family based dataset, a borderline level of statistical significance
was observed (p=.046) supporting the validity of this SNP as an
MS susceptibility locus using a family-based association test.
To determine whether the rs4959039:A.G allele adds to the
risk of MS in HLA-DRB1*15:01 subjects, bi-allelic haplotypes for
rs3135391:T.C (the SNP that tags HLA-DRB1*15:01) and
rs4959039:A.G individuals were constructed in the merged
dataset (Table 2). Each bi-allelic haplotype was treated as a
dichotomous variable in this analysis. The presence of the
rs4959039:A.G allele contributed to MS susceptibility both in
subjects who carry the HLA-DRB1*15:01 allele as well as those
that do not. In addition, the rs4959039:A.G allele appears to be
additive to the effect of HLA-DRB1*15:01 increasing the odds ratio
for MS from 5.89 to 6.46, although the confidence intervals for the
odds ratios of these haplotypes overlap.
HLA-G SNP associations from a meta-analysis genome-
wide association study
Depending on the reference sequence the SNP rs4959039 maps
to non-coding regions centromeric to HLA-G or HLA-A. The
chromosome 6 cox reference sequence places this SNP in the
intergenic non-coding region centromeric to HLA-G whereas the
chromosome 6 qb1 reference sequence maps the SNP centromeric
to HLA-A. It appears that this SNP tags a possible ancestral
duplication near both genes . This observation raises the
question as to whether the MS susceptibility signal associated with
this SNP arises from alleles of HLA-G, HLA-A, or other nearby
genes. Indeed, as presented above, many of the Class I SNPs
identified in this study are in moderate to strong LD with each
A panel of different SNPs in the HLA-G locus was assessed using
a dataset described in a recent genome wide association scan
(GWAS) meta-analysis . Although the published GWAS meta-
analysis included subjects from the discovery dataset, these subjects
were excluded from the following analysis to create an
independent dataset consisting of 1606 MS cases and 5425
controls. In the GWAS meta-analysis 167 SNPs mapped to the
HLA-G locus. After adjusting for HLA-DRB1*15:01 using a tagging
SNP and sex 63 of the 167 SNPs were associated with MS
susceptibility with p-values #.01 (Table S7). The majority of the
SNPs mapped to the untranslated region centromeric to HLA-G,
some with p-Values #161026(rs1611715, rs3115627, rs2734982,
rs2975033). 6 SNPs map within the HLA-G gene itself with
p-Values #161024. SNPs rs1611627, rs915668, rs 1736920 and
rs1632933 are intronic SNPs whereas SNP rs1063320 maps to the
39 end of the last exon of HLA-G and is transcribed but not
translated. These data are consistent with the proposition that a
MHC Class I MS susceptibility locus that is independent of the
extended HLA-DRB1*15:01 haplotype maps to the region of the
This comprehensive SNP based analysis spanning the 29 to
34 kb region of the MHC shows that 52 SNPs in Class I, II and III
regions of the MHC were associated with MS susceptibility in two
independent datasets. Moreover, 20 of these SNPs were associated
with MS susceptibility with p-values ,161028in a dataset that
does not carry the extended HLA-DRB1*15 haplotype. The most
significant association was with rs4959039, a class I SNP near
HLA-G. The association of this SNP with MS susceptibility
appears to be independent of the effects of the other identified
Class II and Class III SNPs.
Using two case control datasets and a panel of SNPs specifically
selected to capture the genetic variation within the MHC region
we found that the MHC locus contributes to MS susceptibility, not
only through the well recognized effect of HLA-DRB1*15:01, but
also through independent contributions from a Class I locus. This
study proves that, after the HLA-DRB1*15:01 extended haplotype,
the Class I region is the most significant contributor to MS
susceptibility within the MHC. Importantly, these observations
contrast with an earlier publication of a Canadian cohort which
concluded that all Class I associations with MS susceptibility were
due to LD with HLA-DRB1*15:01 . Although genetic
heterogeneity might account for these differences, it is more likely
that the structure of the current study, specifically the large dataset
and denser set of informative markers, made possible the detection
of independent effects of Class I and Class III genes.
Class I genes and MS susceptibility
The strongest HLA-DRB1*15:01 independent MS association
was with rs4959039, a SNP near the non-classical HLA-G gene.
Several other SNPs in neighboring pseudogenes HLA-80,
HCG2P6, MICD and HLA-J were also associated with MS
susceptibility and are in LD with the rs4959039 SNP. These
SNPs are not strongly linked to the SNP that tags the recently
identified Class I MS protective allele HLA-B*44:02 [12,14] and
are independent of the major MS susceptibility allele HLA-
DRB1*15:01. Because of the prohibitive cost we were unable to
genotype classical HLA alleles in these large datasets to control for
the possible contributions of HLA-DRB1*0301  or other HLA-
DRB1 alleles. Nevertheless, logistic regression models that
controlled for the 10 most statistically significant Class II SNPs,
as well as the 8 Class III SNPs identified in this study,
Table 2. Paired marker analysis for HLA-DRB1*15:01 and rs4959039 haplotypes in the merged dataset.
HLA-DRB1rs4959039N O.R.95 C.I. lower95 C.I. upperp-Value
HLA-DRB1*15:01‘‘G’’ 1295.89 3.62 9.59 1.036610212
HLA-DRB1*X ‘‘A’’ 2161.901.35 2.67.0002
Two locus haplotypes were constructed and the odds ratio for association with MS susceptibility for each haplotype was tested in a logistic regression model treating
each haplotype as a categorical variable. The odds ratio for the HLA-DRB1*15:01 allele in the merged dataset was 3.50 (p,1.466102100). All results are adjusted for
stratification effects caused by sex, location (US versus UK) and dataset (discovery versus replication). DRB1*X refers to subjects who do not carry the HLA-DRB1*15:01
A MHC Class I MS Locus
PLoS ONE | www.plosone.org6June 2010 | Volume 5 | Issue 6 | e11296
demonstrated an independent allelic contribution of rs4959039 to
Although this association study cannot exclude the possibility
that another closely linked MHC Class I gene, or genes, gives rise
to the MS susceptibility signal detected by the rs4959039 SNP it is
clearly of interest that this SNP is in the 39 un-translated region of
HLA-G. We conclusively demonstrated that this SNP’s association
with MS susceptibility is independent of HLA-DRB1*15:01 and
provided evidence that this SNP is not tightly linked to any of the
Class III or Class II associations identified in this screen.
However, the rs4959039 SNP also maps to a duplication that is
near HLA-A. HLA-A alleles were previously associated with MS
susceptibility: the HLA*03 allele is thought to increase MS risk in
HLA-DRB1*15:01 subjects [18,19] whereas the HLA-A*02 allele is
thought to reduce MS risk . Several lines of evidence suggest
that the rs4959039 SNP’s association with MS might be through
HLA-G rather than HLA-A*03. First, the HLA-A*03 allele is part of
the extended HLA-DRB1*15:01 haplotype that was effectively
excluded in this study. Second, the HLA-A*03 allele that was
imputed in the discovery dataset is not tightly correlated with the
rs4959039 SNP (r2=.002). Third, SNPs in HLA-A were not
identified as disease-associated in either the discovery or the
replication datasets. Lastly, using a different panel of HLA-G
imputed SNPs from a genome wide meta-analysis in an
independent dataset, multiple SNPs in the HLA-G locus were
significantly associated with MS susceptibility after adjusting for
HLA-DRB1*15:01. Thus we interpret our results as suggesting that
the rs4959039 SNP association with MS risk is not through HLA-
A*03. However, because typing of class I genes was unavailable for
nearly the entire dataset we were unable to further analyze the
relationship between the rs4959039 SNP and HLA-A*02, or other
HLA-A alleles. Given that SNP rs4959039 tags a large haplotype
block that includes HLA-A, mapping the class I susceptibility gene,
or genes, will not only require classical typing of HLA-A but also
could require an even larger dataset that excludes HLA-
DRB1*15:01 carriers. For this reason, functional studies of HLA-
A and HLA-G associated variants in MS patients will likely be
useful to understand how alleles of these genes influence MS risk.
HLA-G is a biologically interesting candidate gene because of its
prominent function in immune tolerance. HLA-G is a non-
classical, HLA Class I molecule characterized by relatively limited
polymorphism and alternate splice sites that result in several
membrane bound and soluble isoforms . The HLA-G gene
includes 42 alleles at the DNA level, 14 alleles at the protein level,
and 2 null alleles based on sequence variation in exons 2–4 (the _1
to _3 domains) . In theory, polymorphisms affecting the HLA-
G primary sequence, differences in alternate splicing and
expression pattern, could promote or reduce immune tolerance
and in this manner influence MS susceptibility. Prior genetic
studies of HLA-G in MS susceptibility found conflicting results.
One study found no association of three HLA-G alleles and MS
susceptibility  whereas another found an association of an
HLA-G promoter polymorphism with MS susceptibility by the
transmission distortion test . Both studies were limited by
relatively small sample sizes and few genetic markers.
In contrast to the ubiquitous expression of HLA-A, HLA-B and
HLA-C, HLA-G is found primarily in extravillous trophoblasts:
fetal cells that invade the maternal decidua during placenta
formation [24,25]. These fetal trophoblasts are thought to play a
role in inducing maternal tolerance for the fetus. HLA-G probably
does not function in antigen presentation to HLA class restricted T
cells . Rather, HLA-G binds to and stimulates signaling via the
leukocyte immunoglobulin-like receptors (LILRB1/ILT2/CD85j)
as well as LILRB2/ILT4/CD85d) and KIR2DL4 (CD158d) .
These cell surface receptors are expressed on antigen presenting
cells such as dendritic cells, macrophages and B cells and are also
found on natural killer (NK) cells, T cells, eosinophils, and
osteoclasts. Although not well understood, LILRB signaling
inhibits co-stimulation of T cell responses during antigen
presentation . When expressed on target cells HLA-G inhibits
NK cell killing of the target cell by stimulation of inhibitory
pathways . These observations suggest that HLA-G has an
important role in inducing maternal-fetal tolerance. Additional
support for role of HLA-G in immune tolerance comes from
murine allogenic tissue graft experiments in which HLA-G
expression prolongs graft survival .
Whether HLA-G is involved in induction of immune tolerance
in other body tissues, or disease states, is somewhat controversial.
Some authors challenge the idea that HLA-G is expressed
anywhere other than the trophobast . However, a growing
body of evidence suggests that HLA-G has an important role in
preventing immunological targeting of malignant cells .
Furthermore, HLA-G may have important roles in inflammatory
skin conditions  and myopathies .
A role for HLA-G in multiple sclerosis pathogenesis was first
proposed based on the observation that sHLA-G levels were
elevated in MS patients relative to healthy controls .
Furthermore, sHLA-G is down-regulated in patients who have
actively inflamed MS plaques as evidenced by gadolinium-DPTA
enhancement on brain MRI imaging . HLA-G is known to be
strongly expressed in brain specimens from MS patients where it is
present in acute inflammatory demyelinating plaques, chronic
active plaques, peri-plaque areas and normal appearing white
matter . In MS, HLA-G is expressed primarily on microglia,
macrophages, and endothelial cells. In addition to HLA-G, one of
its receptors, LILRB1/ILT2, is also found in MS brain tissue
suggesting that HLA-G expression in MS brain is functionally
relevant, possibly through an inhibitory feedback pathway directed
at down regulating pro-inflammatory T cells. Recently, HLA-Gpos
Tregcells were identified in MS cerebrospinal fluid, as well as in
inflammatory brain tissue, and these cells are thought to function
as suppressor cells, counterbalancing the tissue destructive effects
of autoimmune inflammation . Taken together, these
observations suggest that HLA-G may have a fundamental role
in limiting tissue injury in MS by regulating auto-reactive immune
cells within the central nervous system.
Thus, it is possible that a HLA-G associated haplotype could
contribute to MS risk by influencing signaling via LILRB1/ILT2 or
the KIR2DL4 natural killer (NK) receptors . Polymorphisms in
HLA-G or KIR2DL4 could influence CD56brightNK cell function
whose corresponding immunoregulatory pathway involves the
already established MS susceptibility genes, the interleukin 2
receptor (IL2RA) and interleukin 7 receptor (IL7R) .
Other Class I loci
In addition to the rs4959039 (near HLA-G) association, several
other Class I SNPs associated with MS susceptibility were
identified, replicated and shown to have HLA-DRB1*15:01
independent effects. One group of SNPs tags the HLA-B*44:02
allele. Tagging SNPs for the closely linked HLA-C*0501 allele 
did not survive the stringent criteria for association used in this
study. These SNPs narrowly missed the cutoff for inclusion as
candidates in the discovery dataset screen but were associated with
MS susceptibility in the replication dataset screen. When these
SNPs were included in the merged HLA-DRB1*15:01(-) dataset,
tagging SNPs for HLA-C*05:01  were significantly associated
with MS susceptibility (data not shown). Logistic regression
modeling suggested that the primary signal in the Class I region
A MHC Class I MS Locus
PLoS ONE | www.plosone.org7June 2010 | Volume 5 | Issue 6 | e11296
arises from the locus identified by SNP rs4595039 although it
remains possible that there could be independent contributions
from other Class I loci.
Both HLA-C*05:01 and HLA-B*44:02 are reportedly protective
alleles for MS susceptibility [7,12,14]. These neighboring alleles
are in tight LD making discrimination between the effects of each
allele challenging. In addition, different alleles of HLA-A may
influence MS susceptibility in opposite directions. HLA-A*0301
may in crease MS risk; however, this allele is part of the expanded
haplotype shared by HLA-DRB1*15:01 and its proposed influence
on MS susceptibility may be confounded by linkage to HLA-
DRB1*15:01 . In contrast, HLA-A*02:01 appears to have a
protective effect . This allele is also linked to the SNP
identified in the present study, rs4595039. Functional studies, or
fine mapping studies in populations with different patterns of LD,
will be needed to determine whether the protective effect proposed
for HLA-A*02:01 is mediated by linkage to an allele of HLA-G or
other neighboring genes.
In summary, we found MHC SNP associations with MS
susceptibility, independent from the primary influence of HLA-
DRB1*15:01, in the Class I, Class II and Class III regions. The
most significant contribution arises from the Class I region in the
vicinity of the HLA-G gene. HLA-G, or another closely linked gene
such as HLA-A, contributes to MS risk independently from the
recently identified Class I allele HLA-B*44:02, as well as other
Class II and Class III SNPs identified in the present study. Thus a
Class I locus near HLA-G/HLA-A is a replicated locus within the
MHC that contributes to MS risk independently of HLA-
DRB1*15:01. The possible HLA-G association is particularly
interesting because HLA-G is thought to function in induction
of immune tolerance and is highly expressed in MS brain tissue.
Further studies of functional polymorphisms in HLA-G, classical
HLA typing, as well as studies in populations with different
patterns of LD within the MHC, will help further define this
locus’s contribution to MS risk.
All study subjects signed written informed consent forms
approved by the following local institutional review boards in
accordance with the Declaration of Helsinki: Committee on
Human Research (UCSF), CERDNT (MHI), Human Subjects
Research Office (University of Miami), Partners Healthcare IB/
Human Research Office, North Thames MREC, The North
Shore - LIJ Health System IRB, Vanderbilt HRPP and Berkshire
Research Ethics Committee.
The MS discovery dataset consists of 1018 cases (520 from the
US and 498 from the UK) and 1795 controls (1049 from the US
and 746 from the UK). All MS subjects met International Panel
criteria for multiple sclerosis . The control population was
composed of samples from the United Kingdom 1958 birth cohort
as well as a cohort of healthy subjects form The New York Cancer
Project. The family based trio analysis was conducted on a subset
of 347 trio families (MS patient and both parents) from the
discovery cohort who did not carry the HLA-DRB1*15:01 tagging
The genetic marker analysis used for the discovery cohort was a
custom Illumina array that composed of 1337 SNPS to tag
common SNP variation across the 3.44 Mb of the MHC. These
SNPs were selected using the Tagger algorithm for having
relatively low LD from approximately 7000 SNPs spanning the
MHC [11,40]. Overall this set of SNPs captured variation of
common ($5%) HLA markers, less-common (,5%) HLA
markers, common non-HLA markers, and less-common non-
HLA markers, with an average maximum r2of 0.80, 0.64, 0.90,
and 0.62, respectively.  The genetic marker analysis used for
the replication cohort was a custom Illumina array that included
the 1337 SNPS used to tag common SNP variation across the
3.44 MB of the MHC, 29 to 44 Mb as well as other SNPs in genes
of interest that are neither described nor analyzed in the present
manuscript. The HLA-DRB1*15:01 (negative) dataset had .98
power (a=.05) to detect the association of the rs4959039 SNP
with the class I MS susceptibility locus, assuming that this SNP was
tightly linked to the locus with D’=.8 and using the odds ratio and
minor allele frequencies associated with this SNP in this dataset.
A multi-step quality control (QC) strategy was employed for the
samples and SNPs using the following strategy for both discovery
and replication cohorts.
1. Samples whose call rate was ,75% were removed
2. SNPs whose call rate was ,60% in each group were removed
3. Samples in which there was evidence of contamination as
estimated by p$0.1 using IBD/IBS statistics were removed
4. SNPs with minor allele frequency (MAF) ,1% were removed
5. SNPs where HWE ,0.01 in the datasets (cases and controls)
6. Only SNPs that passed QC in both the discovery and
replication datasets were included
Following the QC strategy, 16 MS cases were removed, yielding
a total of 1018 cases available for the discovery case control study.
The replication dataset consisted of an additional 1343 cases and
1379 controls from the US and UK. Of the 1337 Illumina SNPs,
958 passed QC in both datasets.
Marker Trait Association
Associations with MS susceptibility with SNPs and imputed
alleles were assessed by the Cochran Armitage trend tests using the
false discovery rate method to control for multiple comparisons
. SAS, JMPH genomics (Cary, NC) and STATA 9 (North
Fork, TX) were used to perform statistical analyses. Population
stratification caused by differences in markers between the sexes,
the country of the subject’s origin (United States versus United
Kingdom) and dataset (discovery versus replication) was controlled
for by inclusion of these covariates as fixed effects in the regression
Following identification of SNPs that were significantly
associated with MS susceptibility in both datasets the discovery
and replication datasets were merged and subjects carrying the
rs3135391:T.C SNP that tags the HLA-DRB1*15:01 allele were
excluded. MS associated SNPs where (Hardy Weinberg Equilib-
rium) HWE ,0.01 in the control population of the merged dataset
were dropped. 52 SNPs were significantly associated with MS
susceptibility in both datasets.
In the discovery dataset previous 2- or 4-digit typing of HLA-
DRB1 was available for 27.6% of the dataset (N=777).  In this
subset, the tagging SNP rs3135391 was 100% sensitive and 100%
specific for correctly calling HLA-DRB1*15:01. HLA typing was
performed by different methodologies, including PCR-based
sequence-specific oligonucleotide probe reverse-line blot assay,
sequence-specific oligonucleotide (LABType) typing, and exons 2/
3 sequence based typing.
MHC used in the initial screens are listed in Supplemental Table
2. The 48 SNPs associated with MS in both datasets are listed in
Study design summary. The 958 SNPs spanning the
A MHC Class I MS Locus
PLoS ONE | www.plosone.org8 June 2010 | Volume 5 | Issue 6 | e11296
Supplemental Table 3 and the 48 SNPs with p-values #161028in
the merged HLA-DRB1*15:01(-) dataset are listed in Table 1.
Found at: doi:10.1371/journal.pone.0011296.s001 (0.10 MB TIF)
women to men in the control populations was well matched at the
two study centers. However, the proportion of women to men in
the MS subjects was significantly increased in the UK dataset.
Found at: doi:10.1371/journal.pone.0011296.s002 (0.07 MB
Table S1. Case control datasets: The proportion of
replication datasets. Ext=extended.
Found at: doi:10.1371/journal.pone.0011296.s003 (0.17 MB
Table S2: 958 SNPs genotyped in both discovery and
susceptibility in the discovery and replication datasets using
Cochran Armitage trend test, FDR=.05, adjusted for sex, center
(US versus UK) and HLA-DRB1*15:01. The SNPs are listed in
order of chromosomal position from telomere to centromere. The
p-values for the merged dataset are unadjusted. rs2523393 is a
tagging SNP for HLA-B*44:02 [12,14].
Found at: doi:10.1371/journal.pone.0011296.s004 (0.17 MB
Table S3: 52 SNPs significantly associated with MS
susceptibility in the merged HLA-DRB1*15:01 (-) dataset, using the
trend test and adjusting for sex, center (US versus UK) and dataset
(discovery versus replication). SNPs are listed in order of most to
least statistical significance. Four class II SNPs identified in the
discovery and replication datasets were no longer significantly
associated with MS susceptibility in the HLA-DRB1*15:01 (-)
dataset: rs3129961, rs3135352, rs3135391, and rs3135388.
Found at: doi:10.1371/journal.pone.0011296.s005 (0.14 MB
Table S4: 48 SNPs significantly associated with MS
susceptibility in the HLA-DRB1*15:01(-) dataset are grouped
together using an algorithm to define SNP clusters based on
LD-R2$.05 (moderate to strong LD) . The 48 SNPs can be
grouped into 20 SNP clusters and tagging SNPs for each cluster
are designated by an asterisk. The SNPs are listed in order of
cluster size with the largest cluster including 10 SNPs and the
smallest SNP clusters include only single SNPs.
Found at: doi:10.1371/journal.pone.0011296.s006 (0.09 MB
Table S5: 48 SNPs that are associated with MS
rs2523393 (tags HLA-B*44:02) and SNP rs459039 (near HLA-G).
A MS risk haplotype is rs2523393:T.C with rs4959039:A.G and
rs4959039:G.A. The heterozygous haplotype is appears to be
protective suggesting a dominant effect of the protective
haplotype. The p-values and odds ratios are adjusted for the
Table S6: Two locus haplotypes for the SNPs
is rs2523393:C.T with
covariates sex (men versus women) and cohort (discovery versus
replication) to control for stratification.
Found at: doi:10.1371/journal.pone.0011296.s007 (0.05 MB
susceptibility in the HLA-G locus typed in an independent dataset
used for a genome wide meta-analysis.
Found at: doi:10.1371/journal.pone.0011296.s008 (0.12 MB
Table S7: SNPs significantly associated with MS
International MHC and Autoimmunity Genetics Network: John D. Rioux,
Philippe Goyette, Timothy J. Vyse, Lennart Hammarstro ¨m, Michelle M.
A. Fernando, Todd Green, Philip L. De Jager, Sylvain Foisy, Joanne
Wang, Paul I. W. de Bakker, Stephen Leslie, Gilean McVean, Leonid
Padyukov, Lars Alfredsson, Vito Annese, David A. Hafler, Qiang Pan-
Hammarstro ¨m, Ritva Matell, Stephen J. Sawcer, Alastair D. Compston,
Bruce A. C. Cree, Daniel B. Mirel, Mark J. Daly, Tim W. Behrens, Lars
Klareskog, Peter K. Gregersen, Jorge R. Oksenberg, and Stephen L.
The International Multiple Sclerosis Genetics Consortium: Clinical
and Sample Collection Groups (in order of the number of samples
collected): University of Cambridge School of Clinical Medicine, Cam-
bridge, United Kingdom — S. Sawcer (project coleader), M. Ban, A.
Compston; University of California at San Francisco, San Francisco —
J.R. Oksenberg (project coleader), B. Cree, S.L. Hauser; Brigham and
Women’s Hospital, Boston— P.L. De Jager (project coleader), H.L.
Weiner, D.A. Hafler. Project Management and Genotyping
Centers: Harvard Center for Neurodegeneration and Repair, Boston
— A.J. Ivinson (project leader); Brigham and Women’s Hospital, Boston —
D.A. Hafler; Broad Institute of Harvard University and Massachusetts
Institute of Technology, Cambridge, MA — S.B. Gabriel, D.B. Mirel;
Duke University Medical Center, Durham, NC — S.G. Gregory, M.A.
Pericak-Vance. Analysis Group: Massachusetts General Hospital,
Boston — M.J. Daly (project coleader), P.I.W. de Bakker; Brigham and
Women’s Hospital, Boston — P.L. De Jager, L.M. Maier; University of
California at Berkeley, Berkeley — L.F. Barcellos, J.R. Oksenberg;
University of Cambridge School of Clinical Medicine, Cambridge, United
Kingdom — S. Sawcer; University of Miami School of Medicine, Miami
— M.A. Pericak-Vance, J.L. McCauley; and Vanderbilt University
Medical Center, Nashville —J.L. Haines (project leader). The authors
would like to thank all study participants and acknowledge Jordan Hiller
for his consultations regarding use of JMP Genomics.
The authors would like to thank all study participants and acknowledge
Jordan Hiller for his consultations regarding use of JMP Genomics.
Conceived and designed the experiments: BACC JDR JLM PG PLDJ TJV
PG IMAGEN IMSGC DBM DAH JLH MAPV AC SS JRO SLH.
Performed the experiments: BACC JDR JLM PG IMAGEN IMSGC
DBM JRO SLH. Analyzed the data: BACC JDR JLM PAG JPM AS SS
JRO SLH. Contributed reagents/materials/analysis tools: BACC JDR
JLM PAG PG JPM PLDJ AS TJV PG IMAGEN IMSGC DBM DAH
JLH MAPV AC SS JRO SLH. Wrote the paper: BACC JDR JLM PAG
PG JPM PLDJ AS TJV PG DBM DAH JLH MAPV AC SS JRO SLH.
1. Sawcer S, Ban M, Maranian M, Yeo TW, Compston A, et al. (2005) A high-
density screen for linkage in multiple sclerosis. Am J Hum Genet 77: 454–467.
2. Hafler DA, Compston A, Sawcer S, Lander ES, Daly MJ, et al. (2007) Risk
alleles for multiple sclerosis identified by a genomewide study. N Engl J Med
3. Barcellos LF, Oksenberg JR, Green AJ, Bucher P, Rimmler JB, et al. (2002)
Genetic basis for clinical expression in multiple sclerosis. Brain 125: 150–158.
4. Oksenberg JR, Barcellos LF, Cree BA, Baranzini SE, Bugawan TL, et al. (2004)
Mapping multiple sclerosis susceptibility to the HLA-DR locus in African
Americans. Am J Hum Genet 74: 160–167.
5. Lincoln MR, Montpetit A, Cader MZ, Saarela J, Dyment DA, et al. (2005) A
predominant role for the HLA class II region in the association of the MHC
region with multiple sclerosis. Nat Genet 37: 1108–1112.
6. Chao MJ, Barnardo MC, Lui GZ, Lincoln MR, Ramagopalan SV, et al. (2007)
Transmission of class I/II multi-locus MHC haplotypes and multiple sclerosis
susceptibility: accounting for linkage disequilibrium. Hum Mol Genet 16:
7. Yeo TW, De Jager PL, Gregory SG, Barcellos LF, Walton A, et al. (2007) A
second major histocompatibility complex susceptibility locus for multiple
sclerosis. Ann Neurol 61: 228–236.
8. Burfoot RK, Jensen CJ, Field J, Stankovich J, Varney MD, et al. (2008) SNP
mapping and candidate gene sequencing in the class I region of the HLA
complex: searching for multiple sclerosis susceptibility genes in Tasmanians.
Tissue Antigens 71: 42–50.
9. Horton R, Wilming L, Rand V, Lovering RC, Bruford EA, et al. (2004) Gene
map of the extended human MHC. Nat Rev Genet 5: 889–899.
A MHC Class I MS Locus
PLoS ONE | www.plosone.org9 June 2010 | Volume 5 | Issue 6 | e11296
10. Miretti MM, Walsh EC, Ke X, Delgado M, Griffiths M, et al. (2005) A high-
resolution linkage-disequilibrium map of the human major histocompatibility
complex and first generation of tag single-nucleotide polymorphisms. Am J Hum
Genet 76: 634–646.
11. de Bakker PI, McVean G, Sabeti PC, Miretti MM, Green T, et al. (2006) A
high-resolution HLA and SNP haplotype map for disease association studies in
the extended human MHC. Nat Genet 38: 1166–1172.
12. Rioux JD, Goyette P, Vyse TJ, Hammarstrom L, Fernando MM, et al. (2009)
Mapping of multiple susceptibility variants within the MHC region for 7
immune-mediated diseases. Proc Natl Acad Sci U S A 106: 18680–18695.
13. Benjamini Y, Drai D, Elmer G, Kafkafi N, Golani I (2001) Controlling the false
discovery rate in behavior genetics research. Behav Brain Res 125: 279–284.
14. De Jager PL, Jia X, Wang J, de Bakker PI, Ottoboni L, et al. (2009) Meta-
analysis of genome scans and replication identify CD6, IRF8 and TNFRSF1A as
new multiple sclerosis susceptibility loci. Nat Genet 41: 776–782.
15. Carlson CS, Eberle MA, Rieder MJ, Yi Q, Kruglyak L, et al. (2004) Selecting a
maximally informative set of single-nucleotide polymorphisms for association
analyses using linkage disequilibrium. Am J Hum Genet 74: 106–120.
17. Marrosu MG, Murru MR, Costa G, Cucca F, Sotgiu S, et al. (1997) Multiple
sclerosis in Sardinia is associated and in linkage disequilibrium with HLA-DR3
and -DR4 alleles. Am J Hum Genet 61: 454–457.
18. Fogdell-Hahn A, Ligers A, Gronning M, Hillert J, Olerup O (2000) Multiple
sclerosis: a modifying influence of HLA class I genes in an HLA class II
associated autoimmune disease. Tissue Antigens 55: 140–148.
19. Harbo HF, Lie BA, Sawcer S, Celius EG, Dai KZ, et al. (2004) Genes in the
HLA class I region may contribute to the HLA class II-associated genetic
susceptibility to multiple sclerosis. Tissue Antigens 63: 237–247.
20. Brynedal B, Duvefelt K, Jonasdottir G, Roos IM, Akesson E, et al. (2007) HLA-
A confers an HLA-DRB1 independent influence on the risk of multiple sclerosis.
PLoS One 2: e664.
21. Hunt JS, Petroff MG, McIntire RH, Ober C (2005) HLA-G and immune
tolerance in pregnancy. FASEB J 19: 681–693.
23. Kroner A, Grimm A, Johannssen K, Maurer M, Wiendl H (2007) The genetic
influence of the nonclassical MHC molecule HLA-G on multiple sclerosis. Hum
Immunol 68: 422–425.
24. Kovats S, Main EK, Librach C, Stubblebine M, Fisher SJ, et al. (1990) A class I
antigen, HLA-G, expressed in human trophoblasts. Science 248: 220–223.
25. Hunt JS (2006) Stranger in a strange land. Immunol Rev 213: 36–47.
26. Bainbridge D, Ellis S, Le Bouteiller P, Sargent I (2001) HLA-G remains a
mystery. Trends Immunol 22: 548–552.
27. Brown D, Trowsdale J, Allen R (2004) The LILR family: modulators of innate
and adaptive immune pathways in health and disease. Tissue Antigens 64:
28. Pazmany L, Mandelboim O, Vales-Gomez M, Davis DM, Reyburn HT, et al.
(1996) Protection from natural killer cell-mediated lysis by HLA-G expression on
target cells. Science 274: 792–795.
29. Ristich V, Liang S, Zhang W, Wu J, Horuzsko A (2005) Tolerization of dendritic
cells by HLA-G. Eur J Immunol 35: 1133–1142.
30. Apps R, Gardner L, Moffett A (2008) A critical look at HLA-G. Trends
Immunol 29: 313–321.
31. Rouas-Freiss N, Moreau P, Menier C, LeMaoult J, Carosella ED (2007)
Expression of tolerogenic HLA-G molecules in cancer prevents antitumor
responses. Semin Cancer Biol 17: 413–21.
32. Urosevic M (2007) HLA-G in the skin–friend or foe? Semin Cancer Biol 17:
33. Wiendl H, Behrens L, Maier S, Johnson MA, Weiss EH, et al. (2000) Muscle
fibers in inflammatory myopathies and cultured myoblasts express the
nonclassical major histocompatibility antigen HLA-G. Ann Neurol 48: 679–684.
34. Fainardi E, Rizzo R, Melchiorri L, Vaghi L, Castellazzi M, et al. (2003) Presence
of detectable levels of soluble HLA-G molecules in CSF of relapsing-remitting
multiple sclerosis: relationship with CSF soluble HLA-I and IL-10 concentra-
tions and MRI findings. J Neuroimmunol 142: 149–158.
35. Fainardi E, Rizzo R, Melchiorri L, Castellazzi M, Paolino E, et al. (2006)
Intrathecal synthesis of soluble HLA-G and HLA-I molecules are reciprocally
associated to clinical and MRI activity in patients with multiple sclerosis. Mult
Scler 12: 2–12.
36. Wiendl H, Feger U, Mittelbronn M, Jack C, Schreiner B, et al. (2005)
Expression of the immune-tolerogenic major histocompatibility molecule HLA-
G in multiple sclerosis: implications for CNS immunity. Brain 128: 2689–2704.
37. Huang YH, Zozulya AL, Weidenfeller C, Metz I, Buck D, et al. (2009) Specific
central nervous system recruitment of HLA-G(+) regulatory T cells in multiple
sclerosis. Ann Neurol 66: 171–183.
38. Goris A, Dobosi R, Boonen S, Nagels G, Dubois B (2009) KIR2DL4 (CD158d)
polymorphisms and susceptibility to multiple sclerosis. J Neuroimmunol 210:
39. McDonald WI, Compston A, Edan G, Goodkin D, Hartung HP, et al. (2001)
Recommended diagnostic criteria for multiple sclerosis: guidelines from the
International Panel on the diagnosis of multiple sclerosis. Ann Neurol 50:
40. de Bakker PI, Yelensky R, Pe’er I, Gabriel SB, Daly MJ, et al. (2005) Efficiency
and power in genetic association studies. Nat Genet 37: 1217–1223.
A MHC Class I MS Locus
PLoS ONE | www.plosone.org10 June 2010 | Volume 5 | Issue 6 | e11296