Plasticity in the organization and sequences of human
KIR
͞ILT gene families
Michael J. Wilson*
†
, Michaela Torkar*
†
, Anja Haude*, Sarah Milne
‡
, Tania Jones
§
, Denise Sheer
§
, Stephan Beck
‡
,
and John Trowsdale*
¶
*Immunology Division, Department of Pathology, Tennis Court Road, Cambridge CB2 1QP, United Kingdom;
‡
Sanger Centre, Wellcome Trust Genome
Campus, Hinxton, Cambridge CB10 1SA, United Kingdom; and
§
Imperial Cancer Research Fund, Lincolns Inn Fields, Holborn, London WC2A 3PX,
United Kingdom
Edited by Johannes van Rood, Leiden University, Leiden, The Netherlands, and approved February 14, 2000 (received for review January 3, 2000)
The
Ϸ1-Mb leukocyte receptor complex at 19q13.4 is a key polymor-
phic immunoregion containing all of the natural killer-receptor KIR
and related ILT genes. When the organization of the leukocyte
receptor complex was compared from two haplotypes, the gene
content in the KIR region varied dramatically, with framework loci
flanking regions of widely variable gene content. The ILT genes were
more stable in number except for ILT6, which was present only in one
haplotype. Analysis of Alu repeats and comparison of KIR gene
sequences, which are over 90% identical, are consistent with a recent
origin. KIR genesis was followed by extensive duplication
͞deletion as
well as intergenic sequence exchange, reminiscent of MHC class I
genes, which provide KIR ligands.
N
atural Killer (NK) cells are regulated by interaction of surface
receptors with MHC class I molecules (1, 2). In humans, these
interactions are mediated by two structurally distinct receptor
families, the C-type lectin-like (CD94
͞NKG2) heterodimers and
Ig-like receptors, generically known as KIR for killer Ig receptor
(3–5). The KIR loci are located on chromosome 19q13.4 near genes
encoding related molecules such as the Ig-like transcripts (ILTs),
also known as monocyte inhibitory receptors (MIRs) or leukocyte
Ig-like receptors (LIR), and the leukocyte associated inhibitory
receptors (LAIR) (6–12). The region has been called the leukocyte
receptor complex or cluster (LRC) (10, 13).
Over 100 highly homologous KIR sequences have been de-
posited in databases (2). Although most KIRs show 90–95%
identity, some cDNAs may represent different alleles (14, 15).
Similarly, a vast number of transcripts belonging to the ILT
family show high identity (8, 9). There are about 10 expressed
ILT loci in addition to ILT9 and 10 genes (12) for which no
functional transcripts have been found.
Different cells within an individual may each express a subset
of the available KIR repertoire. Haploid genomes may encode
different numbers of KIR genes (16). ILT polymorphism may not
be so extensive except in certain loci, such as ILT5 (17).
Published data suggested that patterns of KIR expression relate
to isotypic and allotypic variation in addition to differential regu-
lation of gene expression. To clarify the nature of polymorphism
and variation in gene number in the polygenic LRC, we undertook
a complete analysis of two different haplotypes.
Methods
P1 Artificial Chromosome (PAC) Clones.
PACs were identified from
the RPCI-1 library (18) with a KIR probe for exon 3 and an ILT2
cDNA, from the Human Genome Mapping Project Resource
Center.
DNA Sequencing and Analysis.
PAC DNA was randomly subcloned
into M13mp18 and pUC18 and amplified in 96-well microtiter
plates (19). The sequence was determined by using chain termina-
tion chemistry (20). The reads were assembled into contigs (21)
and analyzed (http:
͞͞www.sanger.ac.uk͞Teams͞HGP͞Humana).
Dot matrix comparisons were done by using
DOTTER
(22),
http:
͞͞www.sanger.ac.uk͞software͞. Amino acid motifs were
identified by using the
PSORT
(23) program and the
PROSITE
database (24). The genomic sequences are available from
ftp:
͞͞ftp.sanger.ac.uk͞pub͞human͞sequences͞Chr19͞
unfinished
sequence).
Directed Sequencing.
Directed sequencing to detect genes on the
first haplotype was done with locus-specific primers (16). For
intergenic regions, a redundant KIR exon 1 primer was used with
an upstream exon 7 primer. ILT cDNAs were aligned, and
primers were designed by using differences in the 3
Ј ends (Table
1). Long-range PCR was used (Boehringer Mannheim).
Results
Strategy.
To gain insight into the genomic diversity of the KIR
and ILT loci, we collected extensive mapping and sequence data
from the LRC. The genomic organization of the ILT and KIR
clusters was determined on two different haplotypes by using a
PAC genomic library made from a single individual (18). Contigs
were assembled by conventional mapping. They fell into two
distinct sets, and the PACs were divided into haplotypes on the
basis of additional polymorphic markers or partial sequencing
(Fig. 1). PAC clones were designated haplotype 1 or 2 by a
supercript number. Selected clones were sequenced to study
gene arrangement after comparison by Southern blot and KIR
͞
ILT-specific PCR to determine that no rearrangements or
deletions had occurred. We obtained 270 kb of complete se-
quence of two clones (52N12
1
and 1060P11
2
) spanning part of
the ILT region and all of the KIR region (Fig. 1).
Gene Content.
The Fc
␣R locus was telomeric of the KIR cluster
close to another Ig-like NK gene, NKp46. Strikingly, all ILT and
KIR genes in the sequenced section were in head-to-tail orien-
tation from centromere to telomere. This, together with the
conservation of gene structures and sequence homologies be-
tween the different receptor families, indicates that the LRC has
evolved as a result of extensive duplication. Comparison with
databases revealed a genomic clone extending telomeric of the
KIR region. A gene X was identified telomeric of NKp46 but in
the opposite orientation. This gene was expressed in testis and
not in tissue of lymphoid origin by Northern blot analysis (data
not shown), suggesting that it marks the telomeric boundary of
the LRC. In total, there are at least 24 structurally and func-
This paper was submitted directly (Track II) to the PNAS office.
Abbreviations: LRC, leukocyte receptor complex; NK, natural killer; PAC, P1 artificial
chromosome.
†
M.J.W. and M.T. contributed equally to this work.
¶
To whom reprint requests should be addressed. E-mail: jt233@mole.bio.cam.ac.uk.
The publication costs of this article were defrayed in part by page charge payment. This
article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C.
§1734 solely to indicate this fact.
Article published online before print: Proc. Natl. Acad. Sci. USA, 10.1073
͞pnas.080588597.
Article and publication date are at www.pnas.org
͞cgi͞doi͞10.1073͞pnas.080588597
4778 – 4783
͉ PNAS ͉ Aprl 25, 2000 ͉ vol. 97 ͉ no. 9
tionally related Ig-like receptors in the LRC region, spanning
approximately 1 Mb of human chromosome 19q13.4.
Sequences of
KIR Genes.
Sequencing of the
Ϸ160-kb PAC clone
1060P11
2
, which spans the KIR region, revealed numerous tightly
clustered loci, most of which were less than 3 kb apart (Fig. 2 A).
Analysis of the sequence data from 1060P11
2
revealed 10 KIR
genes, most of which have matching cDNA sequences. The excep-
tions are the recently described centromeric KIR locus KIR3DL3
[aka KIRCI (12), named according to convention (25)], the gene
Fig. 2.
(A) Feature map of the complete sequence of the KIR region on PAC 1060P11
2
. The
exons of the KIR genes are shown corresponding to cDNAs or by homology with other genes
(KIR3DL3, KIRX). Arrows point 5
Ј to 3Ј. The positions of Alu, LINE, and LTRs are indicated on
the lines below the genes. KR indicates G
ϩC-rich minisatellites specific to KIR genes. (B) Dot
matrix analysis of the sequence from PAC 1060P11
2
. The plot shows the 10 KIR genes on the
PAC on both x and y axes. Regions of similarity are identified as a concentration of dots
forming diagonal lines (22). The G
ϩC-rich minisatellites can be visualized as boxes, missing
in KIR2DL4. Insertions
͞deletions are visible in KIR2DL4 and KIR2DS5.
Fig. 1.
Organization of the leukocyte receptor complex on human chromosome 19q13.4. Two contigs were deduced from PACs positive with combinations
of individual probes. The complete sequences of PACs 1060P11
2
and 52N12
1
are displayed in Figs. 2 and 4, respectively. A sequence of a section of the KIRs from
a haplotype similar to that of 72N6
1
is available (GenBank database accession no. AC011501). The KIR gene family is flanked by the ILT cluster I and the Fc
␣ R gene.
The numbers of KIR genes differ depending on the haplotype. The ILT genes are grouped in two clusters separated by the LAIR genes. A link between the two
clusters remains to be established. Marker 204 is based on a PCR product designed for other sequences in the database. Although all PAC sizes are to scale,
distances between genes outside the sequenced region are approximate. At the telomeric end, the gene designated X is not related structurally to the Ig-like
receptors. 52N12
1
and 72N6
1
represent one haplotype, and the contig containing 167B21
2
is part of the second haplotype. This was based on sequence
polymorphism in the PCR-amplified ILT3 gene and other differences. Arrows point 5
Ј to 3Ј.
Wilson et al.
PNAS
͉ Aprl 25, 2000 ͉ vol. 97 ͉ no. 9 ͉ 4779
IMMUNOLOGY
KIRX, and a gene telomeric of KIR3DS1, which we have named
KIR2DL5. KIRX has only exons 1–5, i.e., those encoding the
matching extracellular domains of other KIRs. KIR2DL5 has a
similar gene structure and is most homologous to KIR2DL4.
KIR2DL5 has a predicted ORF that would encode potentially for
a protein with two ITIM motifs in its cytoplasmic tail. We found
transcripts corresponding to 2DL5 by reverse transcription–PCR
using cDNA derived from PBLs. All of the two-Ig-domain KIRs,
with the exception of KIR2DL4 and KIR2DL5, have a pseudo exon
3 that has remarkable similarity to the first Ig domain of the three
Ig domain KIRs but are all marked by a 3-bp deletion and a
nucleotide change leading to an in-frame stop codon at the same
position as described for the KIR2DL3 gene (26).
Dot-Matrix Analysis of
KIR Genes.
Dot matrix analysis of the
1060P11
2
sequence (Fig. 2B) revealed a remarkable organization of
reiterated sequences. The only unique sequences over 100 bp were:
(i) upstream of KIR3DL3, (ii) outside the reiterated region, and (iii)
upstream of KIR2DL4. The KIR gene sequences, including inter-
genic regions, are highly conserved. The sequences comprise a
continuous loop that extends seamlessly from gene to gene. The
reiterativeness of the loop is broken only by 14 kb upstream of the
KIR2DL4 locus, which displays some unique features, characterized
by L1 repeats. The high level of homology could facilitate exchange
of exons between different KIR loci, by some form of illegitimate
crossing over or gene conversion (27). Such mechanisms may be
behind the variation in number of Ig exons in some members of the
extended KIR family (see below).
Comparison of Two
KIR Haplotypes.
Using the sequence from
1060P11
2
, we analyzed the other haplotype (haplotype 1) by PCR
using locus-specific (16) and intergenic primer sets (Table 1). All
PCR products were subcloned and completely sequenced to de-
termine the gene arrangement of this haplotype (Fig. 3A). Genomic
databases concurred with our data on this haplotype, which is the
most common (16). The two haplotypes revealed a remarkable
difference in organization. Certain framework loci, such as
KIR3DL3 at the centromeric end, KIRX- KIR2DL4 in the middle,
and KIR3DL2 at the telomeric end, are present on both haplotypes,
consistent with their genotype frequencies of 100% in the popula-
tions studied (12, 16). The position of the KIR3DL1 locus was
occupied by an activating gene, KIR3DS1 on the second haplotype.
Fig. 3.
(A) Differences in the organization of the KIR region in two different haplotypes. On the top line is the plot of the KIR region from 1060P11
2
(Fig. 2 A).
Partial sequence was obtained from PACs 72N6
1
, 78E4
1
, 1015E9
1
, and 1015 M9
1
for the other haplotype (Fig. 1) by using PCR as well as comparison with other
data (see Fig. 1). The gene distances are not to scale and have been exaggerated to display contiguity, suggested by sequence comparisons. Thus, 2DL2 is shown
as a composite of two genes on the second haplotype, namely 2DL3 and 2DL1. Overall, the genes are so similar that the precise lineup of the alleles remains
uncertain, but 3DS1 and 3DL1 are shown as alleles, as supported by other data (see below and ref. 16). (B) Framework genes and variable bubbles in the KIR region.
Comparison of the gene organization data in Fig. 2 A, from the two haplotypes, with other data (16) is consistent with invariant framework loci, flanking regions
where the gene number shows marked flexibility, indicated as open boxes flanked by square boxes. A similar model has been proposed to explain (mostly
interspecies) gene expansion
͞contraction in the MHC, proposing independent expansion of class I genes within a framework of ancestral loci (48).
Table 1. Oligonucleotides used for detection of ILT, LAIR, and
KIR genes by PCR and determination of polymorphic sites by
sequence analysis
Sequence
Specificity
1
5
Ј
GCCACAATCACTCATCAGAGTA 3
Ј
ILT1 specific fwd
2
5
Ј
GTATCGCTGTTACTATGGTAGCG 3
Ј
ILT2 specific fwd
3
5
Ј
ATGATCCCCACCTTCACGGCT 3
Ј
ILT3 specific fwd
4
5
Ј
CAGCTTCCATGCCTTCTGGG 3
Ј
ILT3 specific rev
5
5
Ј
TCAGTATTACAGCCGCGCTCGG 3
Ј
ILT4 specific fwd
6
5
Ј
GGTGCTATGGTTATGACTCGCGCG 3
Ј
ILT6 specific fwd
7
5
Ј
ATGACCGCCGCCCTCACAGCCT 3
Ј
ILT9 specific fwd
8
5
Ј
GCAGAGCAGGGCATCATGGTGT 3
Ј
ILT10 specific fwd
9
5
Ј
AGCCTCCGAGTGTCCACACTG 3
Ј
LIR6 specific fwd
10
5
Ј
TTGGCAGACAGTCCAGATAACATC 3
Ј
LIR6 specific rev
11
5
Ј
TGTCGTGGCCCGCGGAGGC 3
Ј
LIR8 specific fwd
12
5
Ј
TGACTGACACAGCAGGGTCACG 3
Ј
ILT redundant rev
13
5
Ј
CGTGACCCTGCTGTGTCAGTCA 3
Ј
ILT4 sequencing primer
14
5
Ј
CGAAGCCATGAGTTGCACACTG 3
Ј
Marker 204 fwd
15
5
Ј
CAACCACAGCATCTGTAGGCTCC 3
Ј
Marker 204 rev
16
5
Ј
GGGGAGCCATGTGACTTTCGTG 3
Ј
LAIR fwd
17
5
Ј
GTCACTCTGCTCAGACCATTTAG 3
Ј
LAIR rev
18
5
Ј
TGATTGGGACCTCAGTGGTCA 3
Ј
KIR exon 7 redundant fwd
19
5
Ј
CCCAACRCAYRCCATGMTGA 3
Ј
KIR exon 1 redundant rev
The KIR oligonucleotides were used for linking adjoining genes on haplo-
type 1. KIR genes were detected by using primer sets as described (16).
4780
͉ www.pnas.org
Wilson et al.
The two genes are highly homologous in both the exon and intron
sequences, consistent with an allelic relationship. On haplotype 2,
a set of three activating KIR genes (KIR2DS5, DS1, or DS2) was
present between KIR3DS1 and 3DL2, whereas only one activating
locus, 2DS4, was found at the corresponding position in haplotype
1. The sequences of these genes are very similar to each other. It was
therefore difficult to determine to which locus on haplotype 2
KIR2DS4 is allelic, and it may represent a distinct locus.
Further Centromeric, Haplotype 2 Displayed a Single
KIR Gene, KIR2DL2.
The cDNA sequence for this locus shows similarities to the extra-
cellular domain of 2DL3 and the transmembrane and cytoplasmic
part of 2DL1. Interestingly, haplotype 1 contained KIR2DL3 and
2DL1 at the position corresponding to 2DL2 (Fig. 3 A). Detailed
comparison of the genomic sequences revealed the precise rela-
tionship between the two haplotypes in this region, as shown in Fig.
3A. A deletion extending from exon 5 of KIR2DL3 and exon 6 of
KIR2DL1 would result in the formation of a composite gene, 2DL2,
with loss of the intervening KIRZ locus. This is the simplest scenario
to account for the difference in organization, but such is the
similarity of the sequences of all of the loci in this region, a more
complex rearrangement should not be ruled out. The data suggest
that there is flexibility in the presence
͞absence of certain KIR
genes, whereas certain ‘‘framework’’ genes are always present, as
depicted on Fig. 3B. A similar situation prevails in other variable
regions, such as the MHC, where genes like C4 or DRB may be
present in variable numbers on different haplotypes (28, 29), and
other genes such as DRA are invariably present as singletons. In the
case of the MHC framework, ‘‘orthologous’’ regions are conserved
across species such as mice (30). KIR genes have not been identified
in rodents. As mentioned above, two of the framework genes,
present in all or most haplotypes, are positioned at the end of the
complex (KIR3DL3, KIR3DL2) adjacent to unreiterated sequence.
Similarly, the internal framework locus, KIR2DL4, also sits next to
unique sequence. This property may help prevent loss of these
genes by DNA looping, to which the other genes with variable
presence may be prone.
Minisatellites in
KIR Genes.
A feature of all of the KIR genes, with
the exception of KIR2DL4, is the presence of a sequence resembling
a classic moderately G
ϩC-rich minisatellite. The repeat unit of
19–20 residues is typically GGGCCTGGAGGGAGATAT. Taking
into account all of the repeats, none of the nucleotides is totally
invariant at any of the positions, although the first section is
generally more conserved. The number of repeat units varies from
around 30–60 (
Ϸ600–1,200 bp). Apart from the residues immedi-
ately flanking the consensus-splicing signals, the first introns of the
KIR genes are wholly taken up by the minisatellites. They could be
connected in some way to the variation in numbers of different KIR
loci. Other G
ϩC-rich minisatellites are associated with instability
via meiotic recombination processes such as gene conversion or
unequal crossing over (31). Another feature of minisatellites is their
association with recombination hotspots (32). Recombination in
the KIR region has not been studied so far. In view of the linkage
disequilibrium of various combinations of KIR alleles, this topic
deserves to be explored.
Repeat Analysis Is Consistent with the Recent Origin of the
KIR Region.
In addition to the microsatellites referred to above, the sequences
were analyzed for repeats, including Alus, LINEs, and SINEs.
Together they account for over 30% of the sequence exceeding the
coding sequence
Ϸ5-fold. The Alu family is the most frequently
represented at a density of 0.49 Alu per kilobase, which is not
atypical. Outstanding is the highly significant Alu S
͞J ratio. This
ratio can be used as a measure of the age or plasticity of a given
sequence. According to their evolutionary origin, Alus can be
divided into two main classes, J-Alu (old) and S-Alu (new), and
various subclasses of S (33). Random Alus in GenBank are at a
universal S
͞J ratio of 3.00 (34). The S͞J ratio over the KIR region
approaches 70! This is consistent with a recent origin of the region
since J-Alus retrotransposed between
Ϸ55–31 million years ago.
The lack of a KIR region in mice is consistent with its recent
conception. The S-Alus are similarly located in all KIR genes,
suggesting that all KIR loci were derived by duplication of a single
primordial locus. The simplest explanation for development of the
region is that of the emergence of a KIR gene in human ancestors
after the mouse
͞human divide, followed by multiple duplications of
the gene or its derivatives. This could have been followed by
sequence exchange by gene conversion or nonreciprocal crossovers.
In other words, the KIR region is a young region that has undergone
considerable genetic turbulence. Indeed, the specificity of KIR for
subsets of HLA-A, -B, or -C allotypes requires that these receptor
͞
ligand combinations developed subsequent to the divergence of
HLA loci from each other (35). Analysis of the KIR region in other
primates will enable more precise dating of the origin of the
region (36).
ILT Genes.
The ILT genes fell into two clusters. We sequenced a
set of six ILT loci in the region proximal to the KIR loci on a
148-kb PAC clone, 52N12
1
. We called this group, in close
proximity to the LAIR2 gene, ILT cluster I. We performed three
color fluorescence in situ hybridization analyses (not shown)
using PAC clones corresponding to the Fc
␣R͞KIR border
(800P9), the second ILT region (598H20), and NKG7, a marker
centromeric of the ILTs, demonstrating that the order of these
groups of genes, centromere to telomere, is NKG7-ILT cluster
II-ILT cluster I-KIR. The FISH results taken together with the
molecular data and information from databases show the ILT
cluster II to be within 150 kb of ILT cluster I (Fig. 1). The LAIR1
locus was grouped together with the ILT cluster II (Fig. 1). ILT
cluster I encoded on PAC 52N12
1
included the ILT1, LIR6,
ILT2, and ILT3 genes (9, 37). The ILT9 and 10 genes and exons
1–4 of KIR3DL3 were present on this PAC clone (Fig. 1 A). Thus,
52N12
1
overlaps with clone 1060P11
2
(Fig. 4A).
A comparison of the exon
͞intron organization of the ILT genes
revealed that ILT2 conformed to the prototypic structure of
inhibitory ILT genes as described recently for ILT3 (12). The
activating ILTs encoded in cluster I showed some variation. Al-
though ILT9 and 10 have an exon
͞intron organization character-
istic of activating ILTs, an extra exon was found in the LIR6 gene
encoding for part of a stalk region that is extended in the LIR6
transcript in comparison with any other activating ILT. ILT1 had a
largely extended intron 3
Ј of the last Ig domain exon, which results
in the large gene size (15 kb). Upstream of the predicted translation
start site in the LIR6 and in the ILT2 gene was an extra 5
Ј
untranslated region exon present in matching transcripts (LIR6a
and MIRcl7, respectively). Corresponding exons may be predicted
in the ILT1 and ILT9 genes but not for ILT3 and ILT10.
Comparison of the ILT genes (Fig. 4A) revealed that, unlike
the KIRs, their introns are not highly homologous and do not
contain many repeat elements with some exceptions, such as
ILT1. These data suggest an older origin of the ILT region than
the KIRs, consistent with the existence of rodent counterparts,
the paired Ig-like receptors (PIRs), which are in a syntenic region
on mouse chromosome 7 (38–40). Analysis of the repeat com-
position of the ILT sequence reveals a modern–ancient S
͞J Alu
ratio of
Ϸ5, a value that supports the contention of a greater
maturity than the KIR complex. Sequences from the two hap-
lotypes revealed a degree of variation in the ILT coding regions,
for example: ILT3 exon 12 A
͞G and ILT4 Ig3 T͞C, both of which
correspond to known polymorphic cDNA for these genes.
Haplotype-Specific Variation in
ILT Cluster II.
A contig of six clones
was formed outside the sequenced region described above that
all encoded the ILT4 locus. On the basis of a sequence poly-
morphism identified in the Ig3 domain of the ILT4 gene, they
Wilson et al.
PNAS
͉ Aprl 25, 2000 ͉ vol. 97 ͉ no. 9 ͉ 4781
IMMUNOLOGY
were grouped according to their haplotype (Fig. 5 A). The link
between this cluster and the two haplotypes identified in the
ILT
͞KIR region remains to be determined. The ILT5 gene was
at the centromeric end of this cluster on two PAC clones from
each haplotype. ILT8 was close to ILT5. Telomeric of ILT8 is
LIR8, which is present on PAC 478J11 and 933O4. On three PAC
clones, we identified a BglII (2 kb) and a XbaI fragment (3 kb)
by Southern blot analysis that could not be accounted for by any
of the identified ILT genes. We mapped this fragment to the
telomeric end of cluster II. This locus could be the ILT7 gene,
which we have not been able to amplify by PCR.
We amplified ILT6 from three clones (15J7, 933O4, 1042N13)
that correspond to a single haplotype (Fig. 5A). Because the only
overlap between these clones is between LIR8 and ILT4, the
ILT6 gene must be located either between these two loci or in
close proximity to one of them. Two clones of the second
haplotype also span this region but do not contain the ILT6 gene.
We identified a haplotype-specific 3.5-kb XbaI fragment by
Southern blot analysis that is present only on the three ILT6-
positive PAC clones. We postulate that the ILT6 gene is absent
on one haplotype in this PAC library and therefore shows
presence
͞absence variation similar to what has been observed
for some of the KIR loci. To examine this hypothesis further, we
amplified ILT6 from 25 genomic samples and found 2
͞25
homozygous negative for this locus (Fig. 5 B). This is consistent
with a gene frequency for the presence of the ILT6 gene of about
0.72, (0.51 homozygotes, 0.41 heterozygotes, and 0.08 homozy-
gote negative, applying the Hardy–Weinberg equation).
Discussion
All genes encoded in the LRC on human chromosome 19q13.4
are members of the Ig superfamily. Parts of their gene structures
are remarkably conserved, and all are in the same head-to-tail
orientation, with the possible exception of ILT complex II. The
many ILT
͞LAIR͞KIR gene fragments we found in the ILT
region (Fig. 4) could be the fallout of abortive rearrangements
at repeated duplications throughout evolution.
Rodents have genes equivalent to NKG7 (41) and NKp46 (42)
as well as the probable ILT orthologues, the paired Ig-like
Fig. 4.
(A) Organization of ILT cluster I. Sequence of the six ILT genes clustered within
150 kb centromeric of the KIR from PAC 52N12
1
. All genes are in the same 5
Ј to 3Ј
orientation from centromere to telomere and, with exception of the ILT1 gene, are less
than 6 kb in size. Fragments of ILT and LAIR exons were found throughout the entire
sequence (indicated as open blocks). The exon
͞intron organization of the two inhib-
itory ILT genes, ILT2 and ILT3, is conserved, although two exons accounting for two
additional Ig domains are present in the ILT2 gene. Some variation was observed
between the genes encoding for putative activating ILTs: ILT1 contains an 11-kb intron
between the fourth Ig domain and the stalk exon, and an extra stalk exon was found
in the LIR6 gene that is not in any other activating member. In four genes, we predicted
an extra 5
Ј untranslated region exon as seen in some LIR6 and ILT2 transcripts. (B) Dot
matrix analysis of the sequence from PAC 52N12
1
. The plot shows the region encom-
passing the five ILT genes and one LIR locus on both axes (see Fig. 4 A for details). In
contrast to the KIR loci (Fig. 2 A), the sequences of the intergenic regions are not
conserved.
Fig. 5.
(A) Haplotypic variation for presence of ILT6 in ILT cluster II. The
centromeric cluster of ILT genes (Fig. 1) contained six genes in most haplo-
types. On the haplotype shown (Top), the ILT6 gene is missing. (B) PCR analysis
for the presence of ILT6. The genotypes of 25 individuals were analyzed by PCR
by using ILT6- and ILT4-specific primers. The two samples that were negative
for ILT6 are indicated by arrows.
4782
͉ www.pnas.org
Wilson et al.
receptors ( PIRs), which are located on the syntenic mouse
chromosome 7 (40). However, no rodent KIR genes have been
identified. There are at least 14 mouse Ly-49 genes, which fulfill
a function similar to the KIRs (43, 44). These are encoded on
mouse chromosome 6 in the NK complex. Only a single human
LY49 pseudogene has been identified in the syntenic region on
human chromosome 12 (45). Multiple duplication events could
have led to rapid expansion of the KIR gene family in primates
(2) and, conversely, the Ly49 genes in rodents. This argues for
convergent evolution of function of these receptors.
Recently, a ‘‘hybrid’’ cDNA molecule, KIR2DL1v, was identified
that appeared to comprise a 2DL1 sequence with the proximal part
of the second Ig domain to the TM region being replaced by a
sequence resembling KIR2DS1 (46). Our data show how the two
putative parent genes are unlikely to be alleles. They are separated
by over 50 kb of intervening DNA and are in different variable KIR
regions. These data could be explained by some form of nonrecip-
rocal recombination such as gene conversion, which is known to
operate in the MHC (27). The arrangement of KIR genes consisting
of highly related sequences in the same orientation may provide the
ideal substrate for gene conversion.
Close examination of the KIR region shows that the sequences
upstream of the transcribed region are remarkably similar, with
the exception of the 2DL4 gene, suggesting common promoters.
Different groups of 2–9 KIR genes are expressed in NK clones,
but the sequences of the KIR promoter sequences are homoge-
neous, except for that of the KIR2DL4. Therefore, it seems likely
that regulation of expression of KIR transcripts is facilitated by
a stochastic process. The sequence upstream of KIR2DL4 may be
significant because this gene is unique among its peers in being
expressed in 100% of NK clones (15). 2DL4 is the only KIR gene
lacking the repeat region in intron 1.
Examination of the LRC from two different haplotypes
revealed variation in the KIR cluster. KIR genes present on all
haplotypes represent framework loci that flank regions of
variability. In these heterogeneous regions, there are at least
11 possible KIR loci. If we accept this model as a first
approximation of the different arrangements of KIR genes,
there are at least two main positions where haplotypes may
differ (Fig. 3). This hypothesis would account for a large
number of different haplotypes, and it concurs with the
variation in the number of KIR genes observed in different
genotypes (ref. 16; unpublished data). The frequencies with
which certain combinations of KIR loci are found in different
individuals exceeded levels expected from random association
(16), indicative of linkage disequilibrium of alleles on haplo-
types. The independent segregation of KIR3DL1 and KIR3DS1
suggested that these two specificities were alleles. This is the
case for the two haplotypes on Fig. 3.
Taking into account further variation in sequences, particu-
larly ILT4 and ILT5 (17), as well as the presence
͞absence of the
ILT6 gene, the LRC is clearly highly variable. Another region of
the genome that exhibits extensive variation is the MHC, the
products of which are ligands for some of the KIR and ILT
molecules. The MHC class I, class II, and C4 genes exhibit high
levels of variability for presence
͞absence (28, 47). The common
functional link to both sets of loci is resistance to pathogens. Like
the MHC, the LRC has all the hallmarks of a dynamic genomic
region. Selection for variation in KIR gene arrangement could be
infection, in which case we may expect to find some haplotype
frequencies skewed in different diseases.
We thank the Medical Research Council, the Wellcome Foundation, the
European Economic Community (CT961105), and the Imperial Cancer
Research Fund for support, and A. Ziegler, A. Volz, and A. Jeffries for
helpful discussions, as well as Prof. Pieter de Jong for the genomic DNA
(BACPAC Resources, Oakland, CA).
1. Karre, K. & Colonna, M., eds. (1998) Curr. Top. Microbiol. Immunol. 230.
2. Parham, P. (1997) Immunol. Rev. 155.
3. Lanier, L. L. (1998) Annu. Rev. Immunol. 16, 359–394.
4. Brown, M., Scalzo, A., Matsumoto, K. & Yokoyama, W. (1997) Immunol. Rev.
155,
53–65.
5. Plougastel, B. & Trowsdale, J. (1998) Genomics 49, 193–199.
6. D’Andrea, A., Chang, C., Franz-bacon, K., McClanahan, T., Phillips, J. H. &
Lanier, L. L. (1995) J. Immunol. 155, 2306–2310.
7. Suto, Y., Maenaka, K., Yabe, T., Hirai, M., Tokunaga, K., Tadokoro, K. & Juji,
T. (1996) Genomics 35, 270–272.
8. Borges, L. (1997) J. Immunol. 159, 5192–5196.
9. Colonna, M., Nakajima, H., Navarro, F. & Lopez-Botet, M. (1999) J. Leukocyte
Biol. 66, 375–381.
10. Wagtmann, N., Rojo, S., Eichler, E., Mohrenweiser, H. & Long, E. O. (1997)
Curr. Biol. 7, 615–618.
11. Meyaard, L., Adema, G. J., Chang, C., Lanier, L. L. & Phillips, J. H. (1997)
Immunity 7, 283–290.
12. Torkar, M., Norgate, Z., Colonna, M., Trowsdale, J. & Wilson, M. (1998) Eur.
J. Immunol. 28, 3959–3967.
13. Wende, H., Colonna, M., Ziegler, A. & Volz, A. (1998) Mamm. Genome 10,
154–160.
14. Selvakumar, A., Steffens, U. & Dupont, B. (1997) Immunol. Rev. 155, 183–195.
15. Valiante, N., Lienert, K., Shilling, H., Smits, B. & Parnham, P. (1997) Immunol.
Rev. 155, 155–164.
16. Uhrberg, M., Valiante, N. M., Shum, B. P., Shilling, H. G., Lienert-
Weidenbach, K., Corliss, B., Tyan, D., Lanier, L. L. & Parham, P. (1997)
Immunity 7, 753–763.
17. Colonna, M., Navarro, F., Bellon, T., Llano, M., Garcia, P., Samaridis, J.,
Angman, J., Cella, M. & Lopez-Botet, M. (1997) J. Exp. Med. 186, 1809–1818.
18. Ioannou, P. A., Amemiya, C. T., Garnes, J., Kroisel, P. M., Shizuya, H., Batzer,
M. A. & de Jong, P. J. (1994) Nat. Genet. 6, 84–89.
19. Beck, S. & Alderton, R. P. (1993) Anal. Biochem. 212, 498–505.
20. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl. Acad. Sci. USA 74,
5463–5467.
21. Sanger, F. (1998) Genome Res. 8, 1097–1108.
22. Sonnhammer, E. L. & Durbin, R. (1995) Gene 167, 1–10.
23. Nakai, K. & Horton, P. (1999) Trends Biochem. Sci. 24, 34–36.
24. Bairoch, A. (1990) Prosite: A Dictionary of Protein Sites and Patterns, 5th Ed.
(De´partement de Biochimie Me´dicale, Universite´ de Gene`ve, Geneva).
25. Long, E. O., Colonna, M. & Lanier, L. L. (1996) Immunol. Today 17, 100.
26. Wilson, M. J., Torkar, M. & Trowsdale, J. (1997) Tissue Ant. 49, 574–579.
27. Hogstrand, K. & Bohme, J. (1999) Immunol. Rev. 167, 305–317.
28. Trowsdale, J., Ragoussis, J. & Campbell, R. D. (1991) Immunol. Today 12,
443–446.
29. Campbell, R. D. & Trowsdale, J. (1993) Immunol. Today 14, 349–352.
30. Amadou, C., Kumanovics, A., Jones, E. P., Lambracht-Washington, D.,
Yoshino, M. & Lindahl, K. F. (1999) Immunol. Rev. 167, 211–222.
31. Jeffreys, A. J., Neil, D. L. & Neumann, R. (1998) EMBO J. 17, 4147–4157.
32. Jeffreys, A. J., Murray, J. & Neumann, R. (1998) Mol. Cell 2, 267–273.
33. Jurka, J. & Milosavljevic, A. (1991) J. Mol. Evol. 32, 105–121.
34. Jurka, J. & Smith, T. (1988) Proc. Natl. Acad. Sci. USA 85, 4775–4778.
35. Parham, P. (1994) Semin. Immunol. 6, 373–382.
36. Zietkiewicz, E., Richter, C., Malalowski, W., Jurka, J. & Labuda, D. (1994)
Nucleic Acids Res. 22, 5608–5612.
37. Cosman, D., Fanger, N., Borges, L., Kubin, M., Chin, W., Peterson, L. & Hsu,
M.-L. (1997) Immunity 7, 273–282.
38. Yamashita, Y., Fukuta, D., Tsuji, A., Nagabukuro, A., Matsuda, Y., Nishikawa,
Y., Ohyama, Y., Ohmori, H., Ono, M. & Takai, T. (1998) J. Biochem. 123,
358–368.
39. Hayami, K., Fukuta, D., Nishikawa, Y., Yamashita, Y., Inui, M., Ohyama, Y.,
Hikida, M., Ohmori, H. & Takai, T. (1997) J. Biol. Chem. 272, 7320–7327.
40. Kubagawa, H., Burrows, P. D. & Coopers, M. D. (1997) Proc. Natl. Acad. Sci.
USA 94, 5261–5266.
41. Berg, S. F., Westgaard, I. H., Fossum, S. & Dissen, E. (1999) Immunogenetics
49,
815–818.
42. Falco, M., Cantoni, C., Bottino, C., Moretta, A. & Biassoni, R. (1999)
Immunol. Lett. 68, 411–414.
43. McQueen, K. L., Freeman, J. D., Takei, F. & Mager, D. L. (1998) Immuno-
genetics 48, 174–183.
44. Brown, M. G., Fulmek, S., Matsumoto, K., Cho, R., Lyons, P. A., Levy, E. R.,
Scalzo, A. A. & Yokoyama, W. M. (1997) Genomics 42, 16–25.
45. Barten, R. & Trowsdale, J. (1999) Immunogenetics 49, 731–734.
46. Shilling, H. G., Lienert-Weidenbach, K., Valiante, N. M., Uhrberg, M. &
Parham, P. (1998) Immunogenetics 48, 413–416.
47. Campbell, R. D. & Trowsdale, J. (1997) Immunol. Today 18 (Suppl.).
48. Amadou, C. (1999) Immunogenetics 49, 362–367.
Wilson et al.
PNAS
͉ Aprl 25, 2000 ͉ vol. 97 ͉ no. 9 ͉ 4783
IMMUNOLOGY
Dostları ilə paylaş: |