The n e w e ng l a n d j o u r na l of m e dic i n e

original article

NALP1 in Vitiligo-Associated Multiple

Autoimmune Disease
Ying Jin, M.D., Ph.D., Christina M. Mailloux, B.S., Katherine Gowan, B.S.,
Sheri L. Riccardi, B.S., Greggory LaBerge, M.S., Dorothy C. Bennett, Ph.D.,
Pamela R. Fain, Ph.D., and Richard A. Spritz, M.D.

A BS T R AC T

BACKGROUND
From the Human Medical Genetics Pro- Autoimmune and autoinflammatory diseases involve interactions between genetic
gram (Y.J., C.M.M., K.G., S.L.R., G.L., risk factors and environmental triggers. We searched for a gene on chromosome
P.R.F., R.A.S.) and the Barbara Davis
Center for Childhood Diabetes (P.R.F.), 17p13 that contributes to a group of epidemiologically associated autoimmune and
University of Colorado at Denver and autoinflammatory diseases. The group includes various combinations of general-
Health Sciences Center, Aurora; and the ized vitiligo, autoimmune thyroid disease, latent autoimmune diabetes in adults,
Division of Basic Medical Sciences, St.
Georges, University of London, London rheumatoid arthritis, psoriasis, pernicious anemia, systemic lupus erythematosus,
(D.C.B.). Address reprint requests to Dr. and Addisons disease.
Spritz at the Human Medical Genetics
Program, University of Colorado at Den-
ver and Health Sciences Center, P.O. Box
METHODS
6511, Mailstop 8300, Aurora, CO 80045, We tested 177 single-nucleotide polymorphisms (SNPs) spanning the 17p13 linkage
or at [email protected]. peak for association with disease and identified a strong candidate gene. We then
N Engl J Med 2007;356:1216-25. sequenced DNA in and around the gene to identify additional SNPs. We carried out
Copyright 2007 Massachusetts Medical Society. a second round of tests of association using some of these additional SNPs, thus
elucidating the association with disease in the gene and its extended promoter re-
gion in fine detail.

RESULTS
Association analyses resulted in our identifying as a candidate gene NALP1, which
encodes NACHT leucine-rich-repeat protein 1, a regulator of the innate immune
system. Fine-scale association mapping with the use of DNA from affected families
and additional SNPs in and around NALP1 showed an association of specific vari-
ants with vitiligo alone, with an extended autoimmune and autoinflammatory dis-
ease phenotype, or with both. Conditional logistic-regression analysis of NALP1
SNPs indicated that at least two variants contribute independently to the risk of
disease.

CONCLUSIONS
DNA sequence variants in the NALP1 region are associated with the risk of several
epidemiologically associated autoimmune and autoinflammatory diseases, impli-
cating the innate immune system in the pathogenesis of these disorders.

1216 n engl j med 356;12 www.nejm.org march 22, 2007

Downloaded from www.nejm.org on June 15, 2010 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 Association of NALP1 with Multiple Autoimmune Disease

A
utoimmune and autoinflammato- tribute to this epidemiologic association, includ-
ry diseases are a group of about 80 disor- ing one on chromosome 17p13.10 This genomic
ders that can involve almost any tissue, region also appears to contribute to systemic lupus
organ, or system.1 A major source of illness and erythematosus in members of families who in-
death, these diseases together affect 15 to 25 herit lupus together with either vitiligo11 or vari-
million people in the United States,2 particularly ous other autoimmune and autoinflammatory
women,3,4 in whom they rank among the top 10 diseases.12 This finding suggests that chromo-
causes of death.5 The risk of autoimmune and some 17p13 is involved in the susceptibility to
autoinflammatory diseases is thought to depend multiple autoimmune disease. To identify the auto-
on interactions between environmental factors and immunity susceptibility gene in the 17p13 region,
specific variants of specific genes, some of which we performed fine-scale genetic association and
may confer a risk that an individual disease will DNA sequence analyses in 114 families with viti-
develop, and others a risk that several different ligo and associated autoimmune and autoinflam-
diseases will develop.6 matory diseases.
Indeed, many patients eventually have more
than one autoimmune or autoinflammatory dis- Me thods
ease. Various names have been applied to different
combinations of so-called multiple autoimmune SUBJECTS
disease, such as Schmidts syndrome and auto- We obtained DNA samples from 656 persons
immune polyglandular syndromes. A few very rare from 114 extended families with multiple auto-
multiple autoimmune disease syndromes result immune disease associated with vitiligo from the
from mutations in single genes7; however, most United States and United Kingdom between 1996
cases of multiple autoimmune disease do not fol- and 2005. All families were white (as self-reported
low mendelian patterns of inheritance, but rather on multiple-answer questionnaires; see the Sup-
have complex inheritance patterns. Susceptibility plementary Appendix, available with the full text
genes in these cases probably fall into two catego- of this article at www.nejm.org) and were selected
ries: some may specifically predispose patients to on the basis of having two or more family mem-
one or more of the component diseases, whereas bers with generalized vitiligo (multiplex families)
others may affect the susceptibility of patients to and at least one having one or more of the other
autoimmune and autoinflammatory disease in autoimmune and autoinflammatory diseases epi-
general. The latter type of gene may represent a demiologically associated with vitiligo (autoim-
target in the treatment or even prevention of sev- mune thyroid disease, rheumatoid arthritis, latent
eral different diseases. autoimmune diabetes in adults, psoriasis, perni-
We8,9 and others have observed that, among cious anemia, systemic lupus erythematosus, and
patients with generalized vitiligo, there is an in- Addisons disease).8,9 We studied two series of fam-
creased frequency of several other autoimmune ilies. The first series comprised the 51 extended
and autoinflammatory diseases, particularly auto- families (333 persons) we previously studied to
immune thyroid disease (Graves disease and map a vitiligomultiple autoimmune disease locus
autoimmune hypothyroidism), latent autoimmune to 17p13,10 and the second series comprised 63
diabetes in adults, rheumatoid arthritis, psoriasis, similar, independent families (323 persons). Clin-
pernicious anemia, systemic lupus erythemato- ical information on all families from the first se-
sus, and Addisons disease. There is also an in- ries and on about half the families from the sec-
creased frequency of these same disorders among ond have been reported previously.9
first-degree relatives of patients with vitiligo, sug- Available affected and unaffected family mem-
gesting that some families have a genetic predis- bers completed a detailed questionnaire to pro-
position to this group of autoimmune and auto- vide the clinical history with regard to approxi-
inflammatory diseases. By testing for genetic mately 50 autoimmune and autoinflammatory
linkage between disease and polymorphic DNA diseases and immune-related diseases, including
markers spanning the whole genome in families vitiligo, autoimmune thyroid disease (Graves dis-
with vitiligo and other autoimmune and autoin- ease and autoimmune hypothyroidism), rheuma-
flammatory diseases, we have identified several toid arthritis, psoriasis, pernicious anemia, sys-
chromosomal regions (or loci) that appear to con- temic lupus erythematosus, Addisons disease, and

n engl j med 356;12 www.nejm.org march 22, 2007 1217

Downloaded from www.nejm.org on June 15, 2010 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

diabetes (for which type, age of onset, and course single-nucleotide polymorphisms (SNPs) selected
of treatment were specified). All patients with viti- from the Illumina SNP Knowledge Resource. Each
ligo provided a skin-lesion map. SNP had a minor allele frequency exceeding 0.10
The study investigators reviewed all data from (i.e., the less common variant of the SNP occurred
the questionnaires and the lesion maps and the on at least 10% of chromosomes) in whites; alto-
study staff examined most family members, both gether, these SNPs captured approximately 18%
affected and unaffected. Inclusion criteria for fam- of the common genetic variation (r20.5) across
ily members with generalized vitiligo were the the genetic-linkage region of chromosome 17p (ap-
presence of depigmented patches of skin that were proximately 11.3 cM, or 6.19 Mb). We genotyped
acquired; that had changed in extent and bound- family members in the second series for the 23
aries over time; that were nonfocal and bilateral; SNPs that were significantly associated with dis-
that typically initially involved the fingers, hands, ease in the first series according to both the
feet, face, or crural areas; and that were not as- pedigree disequilibrium test14 and the family-based
sociated with concurrent underlying eczema or association test.15 We then genotyped two inser-
psoriasis or with exposure to depigmenting chem- tiondeletion polymorphisms and 78 additional
icals. Family members with diagnoses that were SNPs in all 114 families (both series combined).
considered to be questionable on the basis of stan- We identified most of these 78 SNPs by sequenc-
dard diagnostic criteria13 were excluded. Table 1 ing NALP1, the gene for NACHT leucine-rich-re-
provides a summary of autoimmune diseases in peat protein 1, and its extended promoter region
the 114 study families (see Table 1 in the Supple- in 15 genetically informative family members. Ad-
mentary Appendix for details of the clinical diag- ditional details of genotyping are provided in the
noses). Supplementary Appendix.
Our study was approved by the Colorado Mul-
tiple Institutional Review Board and the South DNA SEQUENCING
Thames Regional Multicentre Research Ethics Having narrowed the 17p13 autoimmunity locus
Committee. Written informed consent was ob- to NALP1 and its extended promoter region, we
tained from all participants. sought to identify DNA sequence variants within
this region that we could use for further tests of
GENOTYPING association and for identification of the causal
DNA was prepared from peripheral-blood speci- SNP or SNPs. We therefore sequenced 82.9 kb of
mens with the use of a genomic DNA purifica- the NALP1 gene and its extended promoter region
tion kit (Puregene, Gentra Systems) or from saliva in each of four parents (two from each series) who
specimens with the use of a DNA self-collection were heterozygous for a haplotype (a set of SNP
kit (Oragene, DNA Genotek). We initially used the alleles that occur on a single chromosome) of three
Illumina genotyping service to genotype family adjacent Illumina SNPs (rs3926687, rs2733359, and
members in the first series, assaying 177 known rs878329) that initially appeared to confer a high
risk (haplotype 1) and who had transmitted this
Table 1. Autoimmune and Autoinflammatory Disease Phenotypes haplotype to at least one affected offspring. We
in 114 Multiplex Families.*
sequenced the same region in 11 unrelated patients
Disease Phenotype No. of Cases who were homozygous for haplotype 1 (8 from the
Vitiligo only 219 first series and 3 from the second series). Most of
Vitiligo and autoimmune thyroid disease 70
these patients had vitiligo and at least one other
autoimmune disease. The sequenced regions in-
Vitiligo, autoimmune thyroid disease, and other, nonthyroid 20
autoimmune disease cluded a contiguous segment of 69.1 kb that
Vitiligo and other autoimmune disease 60
spanned the extended NALP1 promoter region and
exons 1, 2, and 3, as well as 11 individual segments
Autoimmune thyroid disease only 86
containing exons 4 through 18 with 90 to 500 bp
Autoimmune thyroid disease and other autoimmune disease 23 of adjacent intronic sequences. We sequenced
Other autoimmune disease only 89 several small introns completely. The regions that
we sequenced define the five known alterna-
* A total of 567 family members reported at least one autoimmune and auto
inflammatory disease; 175 reported more than one disease. tively spliced isoforms of NALP1 messenger RNA
(mRNA); approximately 77% of the sequence was

1218 n engl j med 356;12 www.nejm.org march 22, 2007

Downloaded from www.nejm.org on June 15, 2010 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 Association of NALP1 with Multiple Autoimmune Disease

determined by analyzing both DNA strands. Nu- improved error-checking algorithm,20 the maxi-
cleotide positions were obtained from the human mum multipoint lod score was 4.59 (P=2.14106)
genome sequence for chromosome 17 from the an improvement on the previously obtained
National Center for Biotechnology Information lod score of 4.00.10 The multipoint maximum lod
(NCBI) (Build 36). score occurred at approximately 4.3 cM on chro-
The predicted effect of the substitution of his- mosome 17p; the linkage region encompassing a
tidine for leucine at position 155 (Leu155His) 1-lod reduction from the maximum (thus, the re-
on the secondary structure of the NALP1 protein gion that probably contains the causal gene)
was assessed with the use of the protein structure spanned 11.3 cM (Fig. 1A).
prediction server (PSIPRED).16 We investigated the
potential effects of both alleles of all promoter- FAMILY-BASED ASSOCIATION STUDIES
region variants on transcription-factor binding There are approximately 80 known or predicted
motifs predicted with the use of the Transcription genes within the 11.3-cM region spanning the
Element Search System (TESS)17 and rVista 2.0 linkage peak on chromosome 17p. To determine
software.18 which of these genes is most likely to contribute
to multiple autoimmune disease associated with
STATISTICAL ANALYSIS vitiligo, we genotyped the families in the first
Details on preliminary analyses, genetic-linkage series for the 177 SNPs spanning the linkage
analyses, and conditional logistic-regression anal- peak. The frequencies of all SNPs in founders and
ysis are given in the Supplementary Appendix. We spouses who had married into the family were
tested for HardyWeinberg equilibrium in found- consistent with HardyWeinberg equilibrium. Ge-
ers (the earliest specified persons in lineages) and netic-linkage analysis incorporating both SNP and
in persons not in the lineage, such as spouses, in microsatellite data did not result in a significant
all 114 families. Calculation of linkage disequi- narrowing of the linkage peak (results not shown).
librium between markers of the NALP1 region was We then tested for an allelic association of these
carried out with Haploview software,19 version SNPs with vitiligo alone and with the entire group
3.32. We also calculated the association of each of autoimmune and autoinflammatory diseases as-
marker with vitiligo or with an expanded auto- sociated with vitiligo, considering a family mem-
immune phenotype considering as affected per- ber with any of these diseases as affected. We
sons with any autoimmune or autoinflammatory used both the pedigree disequilibrium test14 and
disease associated with vitiligo using the family- the family-based association test,15 which yielded
based association test,15 version 1.5.5, and the ped- generally similar results (Table 2 in the Supplemen-
igree disequilibrium test,14 version 5.1, which pro- tary Appendix). Both tests showed that 23 SNPs
vides a combined statistic accounting for allele were associated with the vitiligo phenotype, with
transmission to both affected offspring and un- the expanded autoimmune and autoinflammatory
affected offspring. Haplotype-based transmission- disease phenotype, or with both, including a clus-
disequilibrium statistics were calculated with the ter of five adjacent SNPs rs2301582, rs9889625,
use of the family-based association test, version rs3926687, rs2733359, and rs878329 spanning
1.5.5. P values of less than 0.05 were considered a 117-kb region (Fig. 1B).
to indicate statistical significance. To assess the reproducibility of these candi-
date association signals, we carried out an inde-
R e sult s pendent analysis in which we genotyped the 23
significant SNPs in a second series of 63 extended
GENETIC-LINKAGE ANALYSIS families with multiple autoimmune disease as-
We previously reported10 genetic linkage between sociated with vitiligo. The results of this analysis
a locus on chromosome 17p13 and multiple auto- provided support for an association of three of
immune disease associated with vitiligo, a result the five adjacent SNPs rs3926687, rs2733359,
obtained by genotyping microsatellite markers and rs878329 with the vitiligo phenotype and
across the genome in 51 extended families (the with the expanded autoimmune and autoinflam-
first series). In other families with vitiligo only, matory disease phenotype, both in the second
there was no linkage to this chromosomal re- series and in all 114 families (Fig. 1B, and Table
gion.10 With the use of these same data but an 2 in the Supplementary Appendix).

n engl j med 356;12 www.nejm.org march 22, 2007 1219

Downloaded from www.nejm.org on June 15, 2010 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Figure 1. Association of Autoimmune Disease A

with the Chromosome 17 Linkage Region and NALP1. 5
Panel A shows the nonparametric multipoint lod scores
for markers across chromosome 17 for 51 multiplex 4

Multipoint Lod Score

families with vitiligo-associated multiple autoimmune
3
and autoinflammatory disease. Panel B shows the dis-
tribution of 177 single-nucleotide polymorphisms
2
(SNPs) genotyped across the 17p linkage region, from
0.04 to 6.24 Mb, in the same 51 families. The 23 SNPs 1
that were each significantly associated with disease in
these 51 families (red bars and red circles) were then 0
genotyped in a replicate series of 63 similar families 0 30 60 90 120
(red circles indicate SNPs that were significantly associ- Location on Chromosome 17 (cM)
ated with disease in these families) (see Table 2 in the
Supplementary Appendix). Panel C shows the negative B
log10 of the P values from the pedigree disequilibrium 0.04 Mb 6.24 Mb
test for the 85 variants (83 SNPs and 2 insertiondele-
51 Families
tion polymorphisms) genotyped across the NALP1 re- 63 Families
gion of chromosome 17p in all 114 families. Green cir-
cles indicate the vitiligo phenotype, and orange circles C
3.5
indicate the extended autoimmune and autoinflamma-

P Value (negative log10)

tory disease phenotype. The 18 exons of the NALP1 3.0
structural gene are indicated by the black bars (tran- 2.5
scriptional orientation shown from right to left, blue ar- 2.0
row). Panel D shows pairwise r2 values for linkage dis-
1.5
equilibrium (with darker boxes indicating stronger
disequilibrium) among the 19 of the 85 NALP1 region 1.0
markers shown in Panel C that were most consistently 0.5
associated with disease and thus were used in condi-
0.0
tional logistic-regression analyses, graphed against the 5.34 5.39 5.44 5.49 5.54 5.59 5.64 5.69
physical positions of the markers. Stars indicate mark-
Nucleotide Position on Chromosome 17p (Mb)
ers for which potential independent effects could not
be excluded through regression analysis.
NALP1

D
These three high-risk SNPs span a 61-kb seg-
ment that includes the proximal coding region
and the extended promoter region of NALP1, which 1.0
encodes a key regulator of the innate immune 2

system (Fig. 1C). The genomic region around r

NALP1 is gene-sparse; SNPs located downstream

0.0
of NALP1 were not associated with disease, and
the closest upstream gene, KIAA0523, is more than
486 kb toward the centromere from NALP1. The With regard to correction for multiple testing, the
results of the family-based association test showed 177 SNPs constituted approximately 100 indepen-
that a preliminary haplotype defined by these three dent tests (21 blocks of linkage disequilibrium and
SNPs (haplotype 1) had the most significant as- the 79 remaining SNPs), with a corrected P value
sociation with both the vitiligo phenotype and of 0.04
ICM
forAUTHOR:
vitiligo-associated
Jin (Spritz) autoimmune
RETAKE and
1st
the expanded autoimmune and autoinflamma- autoinflammatory
REG F FIGURE: 1 ofdiseases.
2 2nd
3rd
tory disease phenotype in the first series (P=0.01 CASE Revised
Line 4-C
and P=0.009, respectively), in the second series SEQUENCE
EMail analysiS AND HIGH-DENSITY
ARTIST: ts
SIZE
(P=0.03 and P=0.02, respectively), and in both association
Enon analysiS H/T
Combo
H/T 16p6

series combined (P<0.001 for both comparisons; Sequence analysis of NALP1

AUTHOR, and
PLEASE its extended pro-
NOTE:
odds ratio for vitiligo, 1.85; 95% confidence in- Figure has
moter region wasbeenperformed
redrawn and type
in has
11 been reset. with
persons
Please check carefully.
terval [CI], 1.25 to 2.71; odds ratio for autoimmune vitiligo who were homozygous for haplotype 1 and
and autoinflammatory disease, 1.79; 95% CI, 1.25 in JOB:
4 unaffected
35612 heterozygotes who ISSUE: had transmit-
03-22-07
to 2.56). Of the 177 SNPs initially tested, 98 clus- ted haplotype 1 to at least one affected offspring.
tered into 21 blocks of linkage disequilibrium. The analysis yielded a total of 261 sequence vari-

1220 n engl j med 356;12 www.nejm.org march 22, 2007

 Association of NALP1 with Multiple Autoimmune Disease

ants (Table 3 in the Supplementary Appendix), 54 largest individual effects both for the expanded
of which were newly discovered. We also identi- autoimmune and autoinflammatory disease phe-
fied a segment of 524 bp that was missing from notype (odds ratios, 1.93, 1.81, and 1.82, respec-
the NCBI human chromosome 17 sequence, im- tively; P<0.001 for all three markers) and for the
mediately after nucleotide 5,466,866. vitiligo phenotype (odds ratios, 2.08, 2.01, and
To define more precisely the NALP1 genomic 2.01, respectively; P<0.001 for all three markers).
region that confers susceptibility to autoimmune The inclusion of rs6502867 significantly improved
and autoinflammatory disease, we genotyped all the fit of logistic-regression models that included
114 families for 78 additional SNPs (identified any 1 of the 18 other markers; conversely, the fit
by means of sequence analysis) (Fig. 1C) and two of the model including rs6502867 was signifi-
small insertiondeletion polymorphisms (Table 4 cantly improved by the inclusion of any 1 of 15 of
in the Supplementary Appendix), selected on the the 18 other markers (Table 5 in the Supplemen-
basis of their physical positions, HapMap tag-SNP tary Appendix). These results provide further sup-
status, minor allele frequencies, and potential port for the existence of at least two independent
functional significance. The genotype frequen- variants in the NALP1 region associated with the
cies of all variants tested were consistent with risk of disease: one variant tagged by rs6502867
HardyWeinberg equilibrium and were similar in and the other located further upstream, in the
the two series (data not shown). As shown in Fig- proximal coding region or in the transcriptional
ure 1C, many NALP1 region variants were associ- promoter.
ated with vitiligo (with P values ranging from The markers rs878329, rs7223628, rs8182352,
0.048 to <0.001 and odds ratios ranging from and rs4790796 are in almost perfect linkage dis-
1.39 to 2.08) or with associated autoimmune and equilibrium with rs4790797 (Table 6 in the Sup-
autoinflammatory disease (with P values ranging plementary Appendix), indicating that these five
from 0.04 to <0.001 and odds ratios ranging SNPs, which span only 2107 bases, all represent
from 1.25 to 1.93), in a pattern broadly distrib- the same signal. The colinearity among the five
uted across the proximal portion of the NALP1 markers precludes logistic-regression analyses that
structural gene and its extended promoter region. include any two of them. Inclusion of rs4790797
In general, we observed a stronger association significantly improved the fit of models that in-
for the expanded autoimmune and autoinflam- cluded any 1 of the 14 remaining markers, except
matory disease phenotype than for the smaller for rs12150220 and rs2670660, whereas none of
vitiligo subgroup (Table 4 in the Supplementary the 14 remaining markers, except for rs6502867,
Appendix). improved the fit of the model that included
Apparent associations of disease with multiple rs4790797. These results suggest that an associa-
markers in the NALP1 region may reflect multiple tion of 11 of the 14 markers (rs961826, rs11078575,
independent causal variants or may be a conse- rs1877658, rs925597, rs925598, rs3926687, the
quence of linkage disequilibrium between multi- 12-bp deletion, rs2733359, rs35658367, rs2716914,
ple markers and one true causal variant. The align- and rs8182354) with disease may be secondary to
ment of the genomic positions (Fig. 1C) and the linkage disequilibrium with the cluster of 5 SNPs
linkage-disequilibrium pattern (Fig. 1D and Ta- represented by rs4790797. Overall, the marker
ble 2) of the 19 NALP1 region markers (17 SNPs most significantly associated with disease was
and 2 insertiondeletion polymorphisms) for rs4790797, but it could not be distinguished from
which an association with disease was replicated rs12150220 and rs2670660 in the logistic-regres-
in the two series (by means of the pedigree dis- sion analysis (Table 5 in the Supplementary Ap-
equilibrium test and the family-based association pendix). The results for the vitiligo phenotype were
test) suggested that at least two markers might similar to those for the expanded autoimmune
be independently associated with disease. To dis- and autoinflammatory disease phenotype, except
tinguish markers that might reflect independent that the association of rs12150220 with vitiligo
variants from those that merely reflect linkage also appeared to be secondary to linkage disequi-
disequilibrium, we carried out conditional logis- librium with rs4790797.
tic-regression analyses21 of these 19 markers (Ta- Thus, we detected at least two independent sig-
ble 2, and Table 5 in the Supplementary Appen- nals associated with autoimmune and autoinflam-
dix). On the basis of this analysis, three markers matory diseases: one located within the NALP1
(rs6502867, rs8182352, and rs4790797) had the structural gene, tagged by SNP rs6502867, and at

n engl j med 356;12 www.nejm.org march 22, 2007 1221

Downloaded from www.nejm.org on June 15, 2010 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

Table 2. Association of 19 NALP1 Variants with Vitiligo and Autoimmune and Autoinflammatory Disease
in 114 Multiplex Families.*

Variant/Allele P Value Odds Ratio (95% CI)

Pedigree Family-Based Conditional Logistic-
Disequilibrium Test Association Test Regression Analysis
Autoimmune and autoinflam-
matory disease
rs6502867/A <0.001 0.002 <0.001 1.93 (1.322.88)
rs961826/A 0.001 0.001 0.008 1.60 (1.142.24)
rs12150220/A 0.001 0.001 0.003 1.66 (1.192.31)
rs11078575/C 0.003 0.002 0.02 1.50 (1.082.08)
rs1877658/T 0.003 0.005 0.04 1.40 (1.011.93)
rs925597/A <0.001 <0.001 0.005 1.62 (1.162.27)
rs925598/A 0.001 0.002 0.004 1.63 (1.172.26)
rs3926687/T 0.002 0.003 0.006 1.59 (1.152.21)
12-bp deletion 0.006 0.01 0.01 1.51 (1.102.09)
rs2670660/C <0.001 <0.001 0.003 1.68 (1.202.35)
rs2733359/G 0.002 0.001 0.005 1.64 (1.172.30)
rs35658367/ATGA 0.001 0.001 0.004 1.64 (1.172.31)
rs2716914/C 0.003 0.004 0.02 1.49 (1.082.07)
rs878329/G 0.005 0.001 0.003 1.63 (1.172.26)
rs7223628/G 0.002 <0.001 0.002 1.68 (1.212.35)
rs8182352/G 0.001 <0.001 <0.001 1.81 (1.282.56)
rs4790796/A 0.003 <0.001 0.001 1.73 (1.242.42)
rs4790797/T 0.002 <0.001 <0.001 1.82 (1.292.56)
rs8182354/A 0.004 0.001 0.002 1.69 (1.222.36)
Vitiligo
rs6502867/A 0.005 0.004 <0.001 2.08 (1.373.15)
rs961826/A 0.002 0.003 0.007 1.67 (1.152.41)
rs12150220/A 0.002 0.002 0.004 1.69 (1.182.41)
rs11078575/C 0.005 0.005 0.02 1.55 (1.092.19)
rs1877658/T 0.004 0.007 0.03 1.49 (1.042.10)
rs925597/A 0.002 0.003 0.005 1.69 (1.182.44)
rs925598/A 0.003 0.003 0.005 1.66 (1.162.36)
rs3926687/T 0.004 0.004 0.008 1.61 (1.142.29)
12-bp deletion 0.001 0.009 0.007 1.66 (1.162.36)
rs2670660/C <0.001 0.001 0.002 1.80 (1.252.61)
rs2733359/G 0.001 0.004 0.004 1.75 (1.212.53)
rs35658367/ATGA <0.001 0.001 0.002 1.82 (1.242.67)
rs2716914/C 0.002 0.008 0.007 1.64 (1.152.35)
rs878329/G 0.002 0.003 0.002 1.75 (1.222.51)
rs7223628/G <0.001 0.002 0.001 1.82 (1.262.63)
rs8182352/G <0.001 <0.001 <0.001 2.01 (1.362.95)
rs4790796/A 0.001 0.002 0.001 1.83 (1.262.63)
rs4790797/T <0.001 0.001 <0.001 2.01 (1.392.91)
rs8182354/A 0.001 0.002 0.001 1.83 (1.272.64)

* The variants listed are those for which the association with disease was replicated by the pedigree disequilibrium test
and the family-based association test in both series of families.
Odds ratios were calculated from the coefficients of the regression equation.
The 12-bp deletion includes nucleotides 5457169 through 5457180 (TATGACTATGTG).

1222 n engl j med 356;12 www.nejm.org march 22, 2007

Downloaded from www.nejm.org on June 15, 2010 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 Association of NALP1 with Multiple Autoimmune Disease

least one other, located within a 64.7-kb linkage disease in the 114 families (Fig. 1C), and only
disequilibrium block tagged by the nonsynony- rs12150220 (Leu155His) showed evidence of an
mous coding SNP rs12150220 (Leu155His) and association, both with vitiligo alone and with
six promoter-region SNPs (rs2670660, rs878329, autoimmune and autoinflammatory diseases (P=
rs7223628, rs8182352, rs4790796, and rs4790797). 0.002 and P=0.001, respectively, by the pedigree
The significance (P=0.0001) of a model that in- disequilibrium test). Leu155His is a nonconser-
cludes two SNPs, rs6502867 and rs4790797, each vative substitution located between the N-termi-
of which represents one of the two independent nal pyrin and NACHT domains of the human
association signals, was greater than that of ei- NALP1 polypeptide. This region contains no known
ther individual SNP (P<0.001 for each), and the peptide motifs, but its amino acid sequence, in-
haplotype carrying both high-risk alleles con- cluding residue Leu155, has been highly conserved
ferred the highest risk (odds ratio, 2.77; P<0.001) through primate evolution, from bush baby to
(Table 7 in the Supplementary Appendix). human (Fig. 2) suggesting that the region is
We assessed the haplotype-specific effects of critical to protein function. Indeed, the predicted
rs6502867 and rs4790797 (representing the clus- secondary structure of the region has been even
ter of five SNPs with perfect linkage disequilib- more highly conserved than its sequence.
rium) by comparing logistic-regression models
that included the additive effects of both loci, NALP1 PROMOTER-REGION VARIANTS
with and without accounting for linkage phase. We observed a total of 205 variants in the ex-
We did not detect haplotype-specific effects, which tended promoter region, all of which were as-
indicates that it makes no difference whether the sessed for predicted effects on transcription-fac-
two variants in the NALP1 region are cis or trans tor binding motifs. Five of the six tightly linked
to one another. We also used logistic-regression SNPs in the promoter region that were associated
models to evaluate the mode of inheritance of with disease were found to affect high-probabil-
risks individually associated with rs6502867 and ity mammalian transcription-factor binding sites
rs4790797. These analyses favored an additive (Table 3). Furthermore, rs2670660 occurs within
model with no dominant or recessive effects. a segment that is remarkably conserved in the
human, chimpanzee, macaque, bush baby, cow,
ANALYSIS OF NALP1 NONSYNONYMOUS mouse, and rat, suggesting that this variant is
CODING-REGION VARIANTS functionally significant. It alters predicted bind-
A total of 15 SNPs in the NALP1 region predicted ing motifs for the transcription factors HMGA1
nonsynonymous amino acid substitutions in the [HMG-I(Y)] and MYB. MYB regulates transcrip-
NALP1 protein (Table 4 in the Supplementary Ap- tion during the differentiation, proliferation, and
pendix). These 15 SNPs were included in the sec- apoptosis of erythroid, myeloid, and lymphoid cell
ond set of 78 SNPs tested for an association with lineages. Whether any of these sequence variants

Human GCTQGSERRVLRQLPDTSGRRWREISASLLYQALPSSPDHESPSQESPNAPTSTAVLGSWGS
Chimpanzee GCTQGSERRVLRQLPDTSGRRWREISASLLYQALPSSPDHESPSQESPNAPTSTAVLGSWGS
Rhesus Monkey ECTQGSERRVLRQLPDTSGRRWREISSSLLYQALPSSPDFESPSQESPNAPTSTAVLASWGS
Bush Baby RLGTGSGERLLNRSVPTSGRSWGRESRSLLSYQDLSSSPDHQSVRQESPNTSTSTVVLLGRKP

Figure 2. Alignment of Predicted Primate NALP1 Amino Acid Sequences Surrounding the Leu155His Substitution
(Arrowhead).
Residues in red are completely conserved among known primates; those in black are not. Dashes indicate gaps in-
troduced to maximize sequence alignments.
ICM
AUTHOR: Jin (Spritz)
The NALP1 sequence of humans RETAKE (Homo1st sapiens) is from
FIGURE: 2 of 2 2nd
ENSG00000091592 from the National REG FCenter for Biotechnology Information genome sequence (Build 36). The se-
3rd
quence of the chimpanzee (Pan troglodytes)
CASE is from ENSPTRG00000008637 from
Revised the PanTro, version 2.1, database.
The sequence of the rhesus monkey Line 4-C
(Macaca mulatta) was obtained through SIZE
EMail manual annotation of the entry for
ARTIST: ts
ENSMMUG00000030453 in the MMUL, Enon version 1.0, database.H/T The H/Tsequence 33p9
of the bush baby (Otolemur garnettii)
Combo
was obtained through manual assembly and annotation of data from scaffolds 14080 and 113389 from the bush
baby, version 1.0, prerelease database. NALP1 inAUTHOR,
the cow,PLEASE NOTE: rat, and ground squirrel lacks the entire
dog, mouse,
Figure has been redrawn and type has been reset.
NH4-terminal segment found in the primate protein, Pleaseand sequences
check carefully.orthologous to NALP1 could not be identified
in any other early-release genomic databases.
JOB: 35612 ISSUE: 03-22-07

n engl j med 356;12 www.nejm.org march 22, 2007 1223

Downloaded from www.nejm.org on June 15, 2010 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 The n e w e ng l a n d j o u r na l of m e dic i n e

NALP1, also known as CARD7, DEFCAP, and NAC,

Table 3. Predicted Transcription-Factor Binding Motifs Affected by NALP1
Promoter-Region SNPs Associated with Disease.* is thought to mediate activation of the innate im-
mune system in response to so-called pathogen-
Nucleotide Transcription associated molecular patterns such as bacterial
SNP Position Factor La Lq
peptides.22-24 NALP1 is widely expressed at low
rs2670660 5,459,730 levels but is expressed at a high level in immune
T HMGA1 [HMG-I(Y)] 22.00 0.917 cells, particularly T cells and Langerhans cells,25,26
C (high risk) MYB 12.00 1.000 patterns that are consistent with the particular
rs878329 5,493,974 involvement of NALP1 in skin autoimmunity.
NALP1 recruits the adapter protein ASC, caspase
C PEBP2 18.00 0.900
1, and caspase 5 to a complex termed the NALP1
G (high risk)
inflammasome,23,24,26 which activates the proin-
rs7223628 5,495,192 flammatory cytokine interleukin-1. Serum in-
A AP-1 18.00 0.900 terleukin-1 levels are elevated in patients with
NFAT-1 14.00 1.000 generalized vitiligo,27 suggesting the involvement
of this pathway in the pathogenesis of disease.
G (high risk) TCF2 (HNF-1B) 14.00 1.000
NALP1 also appears to play a role in cellular apop-
PU.1 14.00 1.000
tosis, its overexpression stimulating caspase-medi-
rs8182352 5,495,711 ated apoptosis in a variety of cell types.22,28,29
A Mutations in at least two other NALP-related
G (high risk) PR 12.00 1.000 genes involved in the innate immune system are
rs4790797 5,496,043
associated with autoinflammatory diseases. Vari-
ants in NOD2/CARD15 are associated with inflam-
C FOXF1 24.00 0.923
matory bowel disease30,31 and the Blau syn-
T (high risk) drome.32 Mutations in NALP3/CIAS1, a homologue
of NALP1 and a component of the NALP3 inflam-
* The five SNPs are among the six promoter-region SNPs whose potential in
dependent effects could not be ruled out by conditional logistic-regression masome,24,26 result in several autoinflammatory
analyses. La denotes the log-likelihood score. Lq, a measure of the goodness- phenotypes, including the cold autoinflamma-
of-fit of the DNA sequence to the consensus binding motif, was calculated by tory syndrome, the MuckleWells syndrome, and
dividing La by the maximum La possible for the site model; the best possible
Lq was 1.000. neonatal-onset multisystem inflammatory dis-
The NALP1 exon 1 (the site of the start of translation) begins at nucleotide ease.33-35 Administration of an interleukin-1 in-
5,428,550. hibitor36,37 or a caspase-1 inhibitor38 ameliorates
symptoms in patients with these disorders. If the
affects NALP1 transcription in humans requires association between NALP1 and autoimmune and
further investigation. autoinflammatory diseases is confirmed, and if
NALP1 variants are found to result in the activa-
Dis cus sion tion of interleukin-1, then inhibitors of interleu-
kin-1 and caspase might be effective in the
Our study shows that variants of NALP1 confer treatment or prevention of NALP1-associated auto-
susceptibility to autoimmune and autoinflam- immune and autoinflammatory diseases.
matory diseases that are associated with vitiligo. Supported by grants from the National Institutes of Health
Confirmation of this finding will require addi- (AR45584, AI46374, and DK57538), the United Kingdom Vitiligo
Society, and the U.S. National Vitiligo Foundation.
tional studies in other patient groups, including Dr. Spritz reports serving on the scientific advisory board of,
analysis of the extent to which NALP1 is specifi- and receiving stock options from, GammaCan International. No
cally involved in the component autoimmune other potential conflict of interest relevant to this article was
reported.
disorders. Furthermore, the SNPs that we have We thank the many families who participated in this study,
implicated may not be the causal variants; iden- the United Kingdom Vitiligo Society, the U.S. National Vitiligo
tification of such variants will require the dem- Foundation, and Vitiligo Support International for their enthu-
siastic help in family ascertainment; Anita Amadi-Myers, Pau-
onstration of specific effects on NALP1 transcrip- lene Holland, Saunie Hutton, and Angela Wooden for their in-
tion, mRNA, or protein function. valuable assistance; and Tasha Fingerlin for intellectual input.

1224 n engl j med 356;12 www.nejm.org march 22, 2007

Downloaded from www.nejm.org on June 15, 2010 . Copyright 2007 Massachusetts Medical Society. All rights reserved.
 Association of NALP1 with Multiple Autoimmune Disease

References

1. Marrack P, Kappler J, Kotzin BL. Auto- Kaplan NL. A test for linkage and associa- 27. Tu C-X, Gu J-S, Lin X-R. Increased in-
immune disease: why and where it occurs. tion in general pedigrees: the pedigree terleukin-6 and granulocyte-macrophage
Nat Med 2001;7:899-905. disequilibrium test. Am J Hum Genet colony stimulating factor levels in the
2. NIH Autoimmune Diseases Coordi- 2000;67:146-54. sera of patients with non-segmental viti-
nating Committee. Progress in autoim- 15. Horvath S, Xu X, Lake SL, Silverman ligo. J Dermatol Sci 2003;31:73-8.
mune diseases research. Bethesda, MD: EK, Weiss ST, Laird NM. Family-based 28. Chu ZL, Pio F, Xie ZH, et al. A novel
National Institutes of Health, 2005. (Ac- tests for association haplotypes with gen- enhancer of the Apaf1 apoptosome in-
cessed February 21, 2007, at http://www. eral phenotype data: application to asth- volved in cytochrome c-dependent cas-
niaid.nih.gov/dait/pdf/ADCC_Final.pdf.) ma genetics. Genet Epidemiol 2004;26: pase activation and apoptosis. J Biol Chem
3. Jacobson DL, Gange SJ, Rose NR, 61-9. 2001;276:9239-45.
Graham NM. Epidemiology and estimat- 16. Bryson K, McGuffin LJ, Marsden RL, 29. Liu F, Lo CF, Ning X, et al. Expression
ed population burden of selected autoim- et al. Protein structure prediction servers of NALP1 in cerebellar granule neurons
mune diseases in the United States. Clin at University College London. Nucleic Acids stimulates apoptosis. Cell Signal 2004;16:
Immunol Immunopathol 1997;84:223-43. Res 2005;33(Web Server issue):W36-W38. 1013-21.
4. Cooper GS, Stroehla BC. The epidemi- 17. Schug J, Overton GC. TESS: Tran- 30. Hugot JP, Chamaillard M, Zouali H, et
ology of autoimmune diseases. Autoim- scription Element Search system. Techni- al. Association of NOD2 leucine-rich re-
mun Rev 2003;2:119-25. cal report CBIL-TR-1997-1001-v0.0. Phila- peat variants with susceptibility to Crohns
5. Walsh SJ, Rau LM. Autoimmune dis- delphia: University of Pennsylvania, 1997. disease. Nature 2001;411:599-603.
eases: a leading cause of death among (Accessed February 21, 2007, at http:// 31. Ogura Y, Bonen DK, Inohara N, et al.
young and middle-aged women in the www.cbil.upenn.edu/cgi-bin/tess/tess.) A frameshift mutation in NOD2 associated
United States. Am J Public Health 2000; 18. Loots GG, Ovcharenko I. rVISTA 2.0: with susceptibility to Crohns disease.
90:1463-6. evolutionary analysis of transcription fac- Nature 2001;411:603-6.
6. Brand O, Gough S, Heward J. HLA, tor binding sites. Nucleic Acids Res 32. Miceli-Richard C, Lesage S, Rybojad
CTLA-4, and PTPN22: the shared genetic 2004;32(Web Server Issue):W217-W221. M, et al. CARD15 mutations in Blau syn-
master-key to autoimmunity? Expert Rev 19. Barrett JC, Fry B, Maller J, Daly MJ. drome. Nat Genet 2001;29:19-20.
Mol Med 2005;7:1-15. Haploview: analysis and visualization of 33. Hoffman HM, Mueller JL, Broide DH,
7. Gregersen PK, Behrens TW. Genetics LD and haplotype maps. Bioinformatics Wanderer AA, Kolodner RD. Mutation of
of autoimmune diseases disorders of 2005;21:263-5. a new gene encoding a putative pyrin-like
immune homeostasis. Nat Rev Genet 20. Abecasis GR, Cherny SS, Cookson protein causes familial cold autoinflam-
2006;7:917-28. WOC, Cardon LR. Merlin rapid analy- matory syndrome and Muckle-Wells syn-
8. Alkhateeb A, Fain PR, Thody A, Ben- sis of dense genetic maps using sparse gene drome. Nat Genet 2001;29:301-5.
nett DC, Spritz RA. Epidemiology of viti- flow trees. Nat Genet 2002;30:97-101. 34. Aganna E, Martinon F, Hawkins PN,
ligo and associated autoimmune diseases 21. Cordell HJ, Clayton DG. A unified et al. Association of mutations in the
in Caucasian probands and their relatives. stepwise regression procedure for evaluat- NALP3/CIAS1/PYPAF1 gene with a broad
Pigment Cell Res 2003;16:208-14. ing the relative effects of polymorphisms phenotype including recurrent fever, cold
9. Laberge G, Mailloux CM, Gowan K, et within a gene using case/control or family sensitivity, sensorineural deafness, and
al. Early disease onset and increased risk data: application to HLA in type 1 diabe- AA amyloidosis. Arthritis Rheum 2002;
of other autoimmune diseases in familial tes. Am J Hum Genet 2002;70:124-41. 46:2445-52.
generalized vitiligo. Pigment Cell Res 22. Hlaing T, Guo R-F, Dilley KA, et al. 35. Feldman J, Prieur AM, Quartier P, et al.
2005;18:300-5. Molecular cloning and characterization of Chronic infantile neurological cutaneous
10. Spritz RA, Gowan K, Bennett DC, DEFCAP-L and -S, two isoforms of a novel and articular syndrome is caused by muta-
Fain PR. Novel vitiligo susceptibility loci member of the mammalian Ced-4 family tions in CIAS1, a gene highly expressed in
on chromosomes 7 (AIS2) and 8 (AIS3), of apoptosis proteins. J Biol Chem 2001; polymorphonuclear cells and chondrocytes.
confirmation of SLEV1 on chromosome 276:9230-8. Am J Hum Genet 2002;71:198-203.
17, and their roles in an autoimmune dia- 23. Kufer TA, Fritz JH, Philpott DJ. 36. Hawkins PN, Lachmann HJ, McDer-
thesis. Am J Hum Genet 2004;74:188-91. NACHT-LRR proteins (NLRs) in bacterial mott MF. Interleukin-1receptor antago-
11. Nath SK, Kelly JA, Namjou B, et al. infection and immunity. Trends Microbiol nist in the MuckleWells syndrome.
Evidence for a susceptibility gene, SLEV1, 2005;13:381-8. N Engl J Med 2003;348:2583-4.
on chromosome 17p13 in families with 24. Petrilli V, Papin S, Tschopp J. The in- 37. Goldbach-Mansky R, Dailey NJ, Can-
vitiligo-related systemic lupus erythema- flammasome. Curr Biol 2005;15:R581. na SW, et al. Neonatal-onset multisystem
tosus. Am J Hum Genet 2001;69:1401-6. 25. Su AI, Wiltshire T, Batalov S, et al. inflammatory disease responsive to inter-
12. Johansson CM, Zunec R, Garcia MA, A gene atlas of the mouse and human leukin-1 inhibition. N Engl J Med 2006;
et al. Chromosome 17p12-q11 harbors protein-encoding transcriptomes. Proc 355:581-92.
susceptibility loci for systemic lupus ery- Natl Acad Sci U S A 2004;101:6062-7. 38. Stack JH, Beaumont K, Larsen PD, et
thematosus. Hum Genet 2004;115:230-8. 26. Kummer JA, Broekhuizen R, Everett al. IL-converting enzyme/caspase-1 in-
13. Nordlund JJ. Vitiligo vulgaris. In: Nor- H, et al. Inflammasome components hibitor VX-765 blocks the hypersensitive
dlund JJ, Boissy RE, Hearing VJ, King RA, NALP 1 and 3 show distinct but separate response to an inflammatory stimulus in
Oetting WS, Ortonne J-P, eds. The pig- expression profiles in human tissues, monocytes from familial cold autoinflam-
mentary system. 2nd ed. Malden, MA: suggesting a site-specific role in the in- matory syndrome patients. J Immunol
Blackwell Publishing, 2006:551-98. flammatory response. J Histochem Cyto- 2005;175:2630-4.
14. Martin ER, Monks SA, Warren LL, chem (in press). Copyright 2007 Massachusetts Medical Society.

n engl j med 356;12 www.nejm.org march 22, 2007 1225

Downloaded from www.nejm.org on June 15, 2010 . Copyright 2007 Massachusetts Medical Society. All rights reserved.