of March 21, 2018.This information is current as Subsets

Framework Haplotypes, Each with Multiple Diversity with a Minimum of Six Basicby Gene Content: Evidence for Genomic Killer Ig-Like Receptor Haplotype Analysis

Eric Mickelson, Richard J. O'Reilly and Bo DupontKatharine C. Hsu, Xiao-Rong Liu, Annamalai Selvakumar,

http://www.jimmunol.org/content/169/9/5118doi: 10.4049/jimmunol.169.9.5118

2002; 169:5118-5129; ;J Immunol 

Referenceshttp://www.jimmunol.org/content/169/9/5118.full#ref-list-1

, 17 of which you can access for free at: cites 48 articlesThis article

        average*  

4 weeks from acceptance to publicationFast Publication! •    

Every submission reviewed by practicing scientistsNo Triage! •    

from submission to initial decisionRapid Reviews! 30 days* •    

Submit online. ?The JIWhy

Subscriptionhttp://jimmunol.org/subscription

is online at: The Journal of ImmunologyInformation about subscribing to

Permissionshttp://www.aai.org/About/Publications/JI/copyright.htmlSubmit copyright permission requests at:

Email Alertshttp://jimmunol.org/alertsReceive free email-alerts when new articles cite this article. Sign up at:

Print ISSN: 0022-1767 Online ISSN: 1550-6606. Immunologists All rights reserved.Copyright © 2002 by The American Association of1451 Rockville Pike, Suite 650, Rockville, MD 20852The American Association of Immunologists, Inc.,

is published twice each month byThe Journal of Immunology

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

Killer Ig-Like Receptor Haplotype Analysis by Gene Content:Evidence for Genomic Diversity with a Minimum of Six BasicFramework Haplotypes, Each with Multiple Subsets1

Katharine C. Hsu,*† Xiao-Rong Liu,* Annamalai Selvakumar,* Eric Mickelson, ‡

Richard J. O’Reilly,* † and Bo Dupont2*†

Killer Ig-like receptor (KIR) genes constitute a multigene family whose genomic diversity is achieved through differences in genecontent and allelic polymorphism.KIR haplotypes containing a single activatingKIR gene (A-haplotypes), andKIR haplotypes withmultiple activating receptor genes (B-haplotypes) have been described. We report the evaluation ofKIR gene content in extendedfamilies, sibling pairs, and an unrelated Caucasian panel through identification of the presence or absence of 14KIR genes and2 pseudogenes. Haplotype definition included subtyping for the expressed and nonexpressedKIR2DL5 variants, for two alleles ofpseudogene3DP1, and for two alleles of2DS4, including a novel 2DS4 allele, KIR1D. KIR1D appears functionally homologous tothe rhesus monkeyKIR1D and likely arose as a consequence of a 22 nucleotide deletion in the coding sequence of2DS4, leadingto disruption of Ig-domain 2D and a premature termination codon following the first amino acid in the putative transmembranedomain. Our investigations identified 11 haplotypes within 12 families. From 49 sibling pairs and 17 consanguineous DNA samples,an additional 12 haplotypes were predicted. Our studies support a model forKIR haplotype diversity based on six basic genecompositions. We suggest that the centromeric half of theKIR genomic region is comprised of three major combinations, whilethe telomeric half can assume a short form with either2DS4 or KIR1D or a long form with multiple combinations of severalstimulatory KIR genes. Additional rare haplotypes can be identified, and may have arisen by gene duplication, intergenic recom-bination, or deletions. The Journal of Immunology, 2002, 169: 5118–5129.

B ridging innate and adaptive immunity, the NK cell is animportant effector lymphocyte that participates in theearly immune response to pathogens through the produc-

tion of cytokines and chemokines (1). Furthermore, the NK cellhas also been found to mediate cytolytic activity against virallyinfected cells and malignant cells (1, 2). Following the principle ofthe “missing self ” hypothesis, NK recognition of “self ” MHC Agson putative target cells leads to inhibition of effector functions.Accordingly, target cell loss of self-MHC class I expression re-leases NK cell effector functions by removing the MHC-mediatedinhibition (3). Regulation of NK cell function is accomplishedthrough a diverse complement of receptors mediating, activating,and inhibiting signals in response to ligand interactions.

In humans, receptors that signal activation include the NK cy-totoxicity receptors (4), whose ligands remain unclear, andNKG2D, which has been shown to recognize MHC class I chain-related proteins A and B and UL16-binding proteins (5). Inhibitoryreceptors include the heterodimer molecules CD94:NKG2A whichrecognize complexes of HLA-E and peptides encoded from theHLA-A, -B, -C, or -G leader sequences (6, 7). Fulfilling both in-

hibitory and activating roles are members of theCD158 gene fam-ily (8), commonly referred to as killer Ig-like receptors (KIR),3

which are found on all NK cells and on some CD8� T cell subsetswith activated or memory phenotype (9–15).

The strategy of regulating NK cells by pairs of activating andinhibitory KIRs is used not only in humans, but also in other pri-mates such as apes and Old World monkeys in which orthologousKIR genes have been identified (16–18). In addition,KIR-likegenes have also been found in nonprimate higher vertebrates, suchas the cow (19). An intriguing finding has been that in rodents, asystem of NK receptors has evolved in parallel with the humanKIR, such that the C-type lectin Ly-49 receptor family in themouse performs related functions to those of KIR in humans (20).

TheLy49 gene family is located within the NK complex on mousechromosome 6 (21), which also contains other NK receptor genessuch asCD94 and theNKG2 genes. The syntenic chromosomal re-gion in humans is located on chromosome 12, but only a single, non-functionalLy49 gene has been identified (22, 23). TheKIR gene fam-ily is located on human chromosome 19 within the leukocyte receptorcomplex in region 19q13.4 together with the genes encoding other Iggene superfamily members such as the Ig-like transcripts (ILT) andleukocyte-associated Ig-like receptors (24). A similar complement ofIg superfamily members called novel immune-type receptor genes,have recently been described in cartilaginous and early jawed fish(25–27). TheseKIR/ILT-like gene clusters have been localized in thezebra fish to a syntenic region corresponding to mouse chromosome7 and human 19q13. The novel immune-type receptor genes may thusrepresent early ancestors for the leukocyte receptor complex genes.

*Immunology Program, Sloan-Kettering Institute for Cancer Research and†Alloge-neic Bone Marrow Transplantation Service, Department of Medicine, MemorialSloan-Kettering Cancer Center, New York, NY 10021; and‡Human ImmunogeneticsProgram, Clinical Research Division, Fred Hutchinson Cancer Research Center, Se-attle, WA 98109

Received for publication May 30, 2002. Accepted for publication August 30, 2002.

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby markedadvertisement in accordancewith 18 U.S.C. Section 1734 solely to indicate this fact.1 This work was supported by Grants CA 08748, CA23766, and U24 AI49213 fromthe National Institutes of Health.2 Address correspondence and reprint requests to Dr. Bo Dupont, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10021. E-mail address:b-dupont@ski.mskcc.org

3 Abbreviations used in this paper: KIR, killer Ig-like receptor; ILT, Ig-like transcript;BLCL, B-lymphoblastoid cell line; SSP, sequence-specific primer; UTR, untranslatedregion; LD, linkage disequilibrium.

The Journal of Immunology

Copyright © 2002 by The American Association of Immunologists, Inc. 0022-1767/02/$02.00

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

The ligand specificity for HLA-A, -B, -C, and -G has been dem-onstrated for certain KIR, while specificity for other KIR remainsunknown (28). The KIR region exhibits an extensive degree ofdiversity, which it achieves through a combination of variable genecontent and polymorphism (29–38). Population studies have gen-erally used KIR gene typing methods on genomic DNA to deter-mine the presence or absence of each KIR gene and have demon-strated that between individuals, KIR gene content can varywidely. Recently, these studies were extended to include evalua-tion of allelic differences at polymorphic sites to document an ad-ditional dimension of KIR diversity achieved through polymor-phism (36). Indeed, estimates of the extent of KIR genotypediversity within the population suggest that far �0.24% of unre-lated individuals can expect to have identical genotypes (36).

However, underlying the diversity of the KIR genomic regionare patterns that appear conserved within the population. The chro-mosomal arrangement of KIR genes, for instance, maintains a cer-tain regularity as exemplified by the regular spacing of KIR genes�2.4 kb from each other and the presence of the framework genes3DL3 and 3DL2 at either terminus of the region and 2DL4 in themiddle (24). There is suggestion of conservation of haplotypes aswell, with the early description of two main haplotype groups Aand B found within the population (29–30). Estimated to be asfrequent as 47–59% within the Caucasian population, haplotype Acomprises a common complement of KIR genes 3DL3, -2DL3,-2DL1, -2DL4, -3DL1, -2DS4, and -3DL2. In contrast, haplotype Bhas been defined as a more varied haplotype group that encom-passes genotypes containing more activating receptor genes, in-cluding KIR2DS1, -2DS2, -2DS3, and -2DS5. The definition ofthese haplotypes, particularly haplotype B, has been limited due toan inability to accurately resolve haplotypes, short of sequencingthe genomic region, which has only been achieved for three hap-lotypes (Refs. 24 and 39 and accession no. AC011501.7). Familystudies are a useful means by which to clarify haplotypes, if thefamilies are large and informative for haplotype segregation.

In this study, we report the gene content evaluation of 12 fam-ilies, 6 of which are multigenerational with large sibships, and theelucidation of specific haplotypes within these families. In addi-tion, the investigation presented in this study includes typing forthe two major subtypes of a pseudogene, for the expressed andnonexpressed forms of KIR2DL5, and for a new KIR gene thatpreviously was included in the typing for KIR2DS4. Typing forthese loci provides additional insight into distinct haplotypic genecombinations based on gene content. These findings were thenextended to a set of 49 sibling pairs and a panel of unrelated do-nors. The studies indicate the existence of a limited number ofgene combinations centromeric to KIR2DL4 and greater variabilitytelomeric to this anchor gene. We present a model for the genomicorganization of the human KIR region where the gene content canbe defined by six major haplotypic gene combinations, each withmultiple permutations.

Materials and MethodsPopulations

The normal Caucasian population consisted of 85 randomly selected un-related individuals. Sixty-one of these individuals were relatives of patientsfrom the New York tristate area referred to Memorial Sloan-Kettering Can-cer Center (New York, NY) for allogeneic hemopoietic stem cell trans-plantation, and the remaining 24 individuals were parents from the Centerd’Etude du Polymorphisme Humain (CEPH) cell bank. An additional 49unrelated Caucasian individuals, each being a sibling of one of the paneldonors from the New York tristate area, were included for comparison ofKIR genotypes between siblings. These 49 individuals were not included inthe analysis of the unrelated panel. Family studies were performed onB-lymphoblastoid cell lines (BLCL) derived from the family members of

12 CEPH families (consisting of at least two parents and two children). SixCEPH families, consisting of multiple generations and sibships containingat least eight children, were selected for extended typing studies for a totalof 98 individuals. Seventeen BLCLs derived from offspring of consanguin-eous parents and obtained from the International Histocompatibility Work-shop Repositories were included as a potential source of KIR homozygouscells (40). BLCL were maintained in RPMI 1640 supplemented with 10%FBS at 5% CO2, 37°C.

Sample preparation

Genomic DNA was extracted from 5 � 106 PBMCs, bone marrow mono-nuclear cells, or BLCL tissue culture cells using the Puregene DNA Iso-lation kit (Gentra Systems, Minneapolis, MN) according to the manufac-turer’s instructions.

Polymerase chain reaction

PCR-sequence-specific primers (SSP) for the detection of KIR gene loci ingenomic DNA are detailed in Table I. To detect the gene encoding allknown alleles at a given locus and to achieve consistent results, alternativeprimer sets to those previously published (29, 32) were designed for thedetection of KIR2DS1, 2DS5, 2DL1, 2DL4, 2DL5, 3DS1, 3DL1, 3DL2,3DL3, and pseudogenes 3DP1 and 2DP1. Pseudogene KIR3DP1 has pre-viously been designated KIRX (24), KIR2DS6 (39), KIR48 (41), andCD158c (8). Pseudogene KIR2DP1 has previously been designated KIRZ(24), KIR15 (41), and KIRY (20). KIR genes are reported in this studyaccording to the guidelines by the Human Gene Nomenclature Committee(HGNC) (42). Results obtained with these primer sets were compared withresults using primers designed by other groups (29, 32). Primers were alsodesigned in the 5�-untranslated region (UTR) and 3�-UTR as forward orreverse primers, respectively, for use with a KIR gene-specific primer part-ner (Table I). For many KIR genes, two different primer sets were used toestablish the presence or absence of the gene. The primer pair for pseu-dogene 3DP1 amplifies two different-sized amplicons (344 and 1817 bp)corresponding to KIR3DP1 and KIR3DP1v. In addition, primer sets weredesigned to accurately identify 2DS4 and distinguish it from a novel gene(designated KIR1D), which contains a 22-bp deletion in D2, leading to aframe shift and premature stop codon, with a predicted protein producttruncated just within the transmembrane domain. Primers were also de-signed for KIR2DL5 subtyping to distinguish between expressed and non-expressed variants of 2DL5 (44). Amplifications using primers labeled2DL5.1 identify the expressed variants 2DL5.1 and 2DL5.3, and amplifi-cations using primers labeled 2DL5.2 identify the nonexpressed variants2DL5.2 and 2DL5.4 (nomenclature 2DL5.1-2DL5.4 as previously de-scribed; Ref. 44). All primers were confirmed to be KIR gene-specific bybasic local alignment sequence tool searches and verification with theNational Center for Biotechnology Information KIR database (http://www.ncbi.nlm.nih.gov/IEB/Research/GVWG/MHC/MHC.cgi) (3) andwere subsequently validated by amplicon sequencing. Nucleotide sequenc-ing of amplicons was performed using dye terminators and automated se-quencing (ABI 377 instrument; PE Applied Biosystems, Foster City, CA)in the Sloan-Kettering Institute Sequencing Core Facility (New York, NY).

Amplification conditions were optimized for each primer pair set asfollows: 50-�l volume reactions were prepared to include 0.1 �g of testDNA, PCR buffer, 0.2 mM of dNTPs, 2.5 U Taq DNA polymerase (RocheMolecular Biochemicals, Mannheim, Germany), 0.4 �M (with the excep-tion of KIR2DS1 primers which were used at an 0.8-�M final concentra-tion). Internal controls for KIR2DL4 were used to confirm PCR amplifi-cations. All amplifications were performed in PerkinElmer GeneAmp 9700or 2400 thermocyclers (Wellesley, MA) programmed with a 2-min dena-turing step, followed by 30 cycles of 92°C for 10s, 65°C for 30s, 68°C for1 min 30 s, followed by 72°C for 10 min. Annealing temperatures weremodified for primers amplifying KIR2DL2 (63°C), 2DS3 (63°C), KIR1D(63°C), 2DS4 (61°C), and 2DS5 (63°C). Extension time was modified forlong-range amplification of gene KIR2DL1 (10 min).

2DS4 variant identification and cloning

Amplification of KIR2DS4 using previously published primer pairs (29)was performed on genomic DNA samples. PCR products were gel-purifiedusing the GFX PCR DNA and Gel Band Purification kit (Amersham Phar-macia Biotech, Piscataway, NJ) and were subsequently cloned intopCR2.1-TOPO using the TOPO-TA Cloning kit (Invitrogen, Carlsbad,CA). Plasmids were isolated from individual bacterial clones using theQiAmp Spin Miniprep kit (Qiagen, Chatsworth, CA), and vector-specificprimers were used for sequence analysis of plasmid inserts using dye ter-minators and automated sequencing (ABI 377 instrument; PE Applied Bio-systems) for identification of a 22-bp deletion. Primer pairs were thendesigned to specifically identify the intact KIR2DS4 gene and KIR1D from

5119The Journal of Immunology

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

genomic DNA samples (Table I). NK cells isolated from individuals ho-mozygous for KIR1D and individuals heterozygous for 2DS4 and KIR1Dwere used for mRNA isolation (MicroFastTrack 2.0 Kit; Invitrogen) andcDNA isolation (cDNA Cycle kit; Invitrogen). KIR2DS4-specific primers(forward 5�-CCATGTCGCTCATGGTCATCAT-3�, reverse 5�-ACATGTTCTGATTGGGACC-3�) were then used to generate a 730-bp 2DS4 am-plicon spanning from the ATG start codon to the premature TGA stopcodon. PCR products were gel-purified using the GFX PCR DNA and GelBand Purification kit and cloned into pCR2.1-TOPO. Plasmids were iso-lated from individual bacterial clones using the QiAmp Spin Miniprep kit,and KIR1D cDNA clones were distinguished from 2DS4 clones by se-quencing and by PCR analysis.

Haplotype analysis

KIR haplotypes were determined by segregation patterns in families. Inassigning genes to specific haplotypes, the following assumptions weremade: 1) all haplotypes contained KIR3DL3, 2DL4, and 3DL2; 2) haplo-types contained either 2DL2 or 2DL3, but not both; and 3) haplotypescontained either pseudogene 3DP1 or 3DP1v, but not both. All assump-tions were supported by previous analyses of linkage disequilibrium (LD)(29, 31, 32, 34, 35) and by sequenced KIR haplotypes (Refs. 24 and 39 andaccession no. AC011501.7).

Haplotype deduction for sibling pairs not included in the family studieswas performed by first applying the haplotypes defined in the family stud-ies and only defining “new” haplotypes when this approximation was notpossible. The haplotypes in the second sibling were then deduced using theassumption that 75% of siblings will share one KIR haplotype and 25% ofsiblings will share two haplotypes. Haplotype deductions in the unrelatedCaucasian panel and in the panel of consanguineous DNA samples wasperformed using first the KIR haplotypes defined in the family studies, thenadding “new” haplotypes derived from the sibling pair analysis and finallyadding extra haplotypes if needed.

Statistical analysis

The observed KIR Ag frequency was determined by the ratio of genepresence within the population to the total population number. Gene fre-

quency was calculated by the formula: gene frequency � 1�2�(1�ƒ),where ƒ is the observed Ag frequency in the population. Two-locus LDparameters () were calculated according to Mattiuz et al. (45).

ResultsKIR genomic typing of unrelated individuals

A PCR-SSP typing method was devised for the identification of 16known KIR genes and pseudogenes, KIR2DS1-5, 2DL1-5, 3DS1,3DL1-3, and pseudogenes 3DP1 and 2DP1 (Table I). The methodincludes alternative primer sets designed for inclusion of all knownalleles for KIR2DS1, 2DS3, 2DS5, 2DL4, 2DL4, 2DL5, 3DS1, 3DL1,and 3DL2. Confirmation of new primer set specificity was achieved inseveral ways: positive amplification reactions were compared withresults obtained using previously published primer sets (29, 32); am-plification products were sequenced for specificity; and finally, typingof 85 unrelated Caucasian individuals yielded comparable estimatedgene frequencies to those in previously published findings (29, 32, 34,35). In this cohort, 36 different genotypes were identified, 23 of whichwere unique, with the remaining genotypes each observed from 2 to14 times (Fig. 1). The most common genotype (observed 14 times)was KIR3DL3-2DL3-2DP1-2DL1-3DP1-2DL4-3DL1-1D-3DL2, cor-responding to homozygosity for the major subtype of the previouslyreported A haplotype (see below). Observed nine times was the ge-notype KIR3DL3-2DL3-2DP1-2DL1-3DP1-2DL4-3DL1-2DS4-1D-3DL2, corresponding to heterozygosity for the two major subtypes ofhaplotype A. Genotypes represented more than once accounted for73% of all genotyped samples. Included in this typing is the detectionof a new gene KIR1D and typing for the two different variants ofpseudogene 3DP1.

Table I. KIR amplification primers

KIR Locus Sense (5�33�) Antisense (5�33�) PositionaAmpliconSize (bp)

2DL4 CCC CTC AAC AGA TAC CAG CGT GTG GCA GGC AGT GGG GAC CTT AGA CA 962–3�-UTR 99b 2712DL4 GTA TCG CCA GAC ACC TGC ATG CTG GCA GGC AGT GGG GAC CTT AGA CA 701–3�-UTR 99 10823DL3 AAC ACG GAA CTT CCA AAT GCT GAG CG GCA GGC AGT GGG GAC CTT AGA CA 1233–3�-UTR 99 2432DP1 GCA AGA CAC CCC CAA CAG ATA CCA GA GCA GGC AGT GGG GAC CTT AGA CA 969–3�-UTR 99 2783DL1 AAG ACA CCC CCT ACA GAT ACC ATC T GCA GGC AGT GGG GAC CTT AGA CA 1256–3�-UTR 99 2773DP1 ATC CTG TGC GCT GCT GAG CTG AG GCC TAT GAA AAC GGT GTT TCG GAA TAC 5�-UTR 58c–213 3443DP1v ATC CTG TGC GCT GCT GAG CTG AG GCC TAT GAA AAC GGT GTT TCG GAA TAC 5�-UTR 58–249 18172DL3 CCT TCA TCG CTG GTG CTGd CAG GAG ACA ACT TTG GAT CAd 791–1051 8122DL5 GCT CTT CTT TCT CCT TCA TTG CTG C GCA GGC AGT GGG GAC CTT AGA CA 765–3�-UTR 99 10252DL5.1e CTC CCG TGA TGT GGT CAA CAT GTA AA GGG GTC ACA GGG CCC ATG AGG AT 5�-UTR 109–260 18832DL5.2e GTA CGT CAC CCT CCC ATG ATG TA GGG GTC ACA GGG CCC ATG AGG AT 5�-UTR 119–260 18933DS1 CAG CGC TGT GGT GCC TCG C CTG TGA CCA TGA TCA CCA T 233–355 2492DS4 ATC CTG CAA TGT TGG TCG CTG GAT AGA TGG TAC ATG TC 135–485 1902KIR1D ATC CTG CAA TGT TGG TCG CTG GAT AGA TGG AGC TGC AG 135–465 18852DS1 TCT CCA TCA GTC GCA TGA A/Gd AGG GCC CAG AGG AAA GTG/T 251–567 18383DL2 TTG CTG CAG GGG GCC TGG CCA CT CAT GGG CCT CCC CTT CCC TGG AC 43–787 45302DS2 CTT CTG CAC AGA GAG GGG AAG TAd CAC GCT CTC TCC TGC CAAd 173–434 17612DS2 ATC CTG TGC GCT GCT GAG CTG AG CAC GCT CTC TCC TGC CAAd 5�-UTR 58–434 52212DL2 CAT GAT GGG GTC TCC AAAd CCC TGC AGA GAA CCT ACAd 228–521 18082DL2 ATC CTG TGC GCT GCT GAG CTG AG CCC TGC AGA GAA CCT ACAd 5�-UTR 58–521 54382DS3 ATC CTG TGC GCT GCT GAG CTG AG GCA TCT GTA GGT TCC TCC Td 5�-UTR 58–593 59202DS3 GAC ATG TAC CAT CTA TCC ACf GCA TCT GTA GGT TCC TCC Td 465–593 1302DL1 CTG TTA CTC ACT CCC CCT ATC AGGd AGG GCC CAG AGG AAA GTC Ad 307–566 17702DL1 CTG TTA CTC ACT CCC CCT ATC AGGd CAG AAT GTG CAG GTG TCG 307–740 93652DS5 CTG CAC AGA GAG GGG ACG TTT AAC C TCC AGA GGG TCA CTG GGC 176–355 179

a Nucleotide positions of amplified products were calculated from the start codon according to previously reported cDNA sequences (43).b Nucleotide sequence common to all KIR loci (except for KIR3DL2 and pseudogene 3DP1 or 3DP1v) located 99 bp 5� to stop codon.c Nucleotide sequence common to all KIR loci (except for KIR3DL3) located 58 bp 3� to start codon.d Uhrberg et al. (29).e 2DL5.1 typing amplifies the expressed variants 2DL5.1 and 2DL5.3; 2DL5.2 typing amplifies the nonexpressed variants 2DL5.2 and 2DL5.4 (44).f Norman et al. (32).

5120 KIR HAPLOTYPE DEFINITION BY GENE CONTENT IN FAMILIES

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

A novel 2DS4 variant resembling MmKIR1DPCR-SSP typing for gene 2DS4 using a previously publishedprimer set (29) revealed a nearly ubiquitous gene frequency withinthe population. However, sequencing of several genomic DNAamplicons using this primer set revealed the presence of a 2DS4gene variant, characterized by a 22-bp deletion in the second ex-tracellular Ig-like domain (D2). Although the remainder of thenucleotide sequence is identical to KIR2DS4, the amino acidsequence resulting from the deletion-generated frame shift iscomprised of a novel stretch of 88 aa, loss of the D2 Ig domain,and termination at a final length of 239 aa, one amino acid into theputative transmembrane domain, with no discernible cytoplasmicdomain (Fig. 2). Search of GenBank nucleotide sequences re-vealed the presence of this gene in the genomic DNA sequence ofa human chromosome 19 haplotype (accession no. AC011501.7),placing this novel KIR gene in the KIR gene cluster on humanchromosome 19q13.4. A GenBank amino acid search identifiedsignificant amino acid homology (72%) to a variant of the Mm-KIR1D receptor found in rhesus monkeys (accession no.AF334633) which has nucleotide homology to Mm-KIR2DL4, butcontains only one complete Ig domain and no cytoplasmic tail asa consequence of a frame shift (17). Because of the amino acidhomology between this novel human gene and Mm-KIR1D, wepropose designation of the human gene as KIR1D.

Primer sets were generated for PCR-SSP identification ofKIR2DS4 and KIR1D from genomic DNA, and amplification prod-ucts were sequenced to confirm primer specificity. cDNA forKIR1D was isolated from an individual exhibiting KIR1D PCRpositivity (accession no. AY102624). Although both 2DS4 and

KIR1D transcripts are transcribed in heterozygous individuals, ef-forts to isolate the full cDNA transcripts revealed the majority oftranscripts to be 2DS4. In contrast, KIR1D transcripts were morereadily isolated from individuals homozygous for KIR1D (Fig. 3).From a separate individual, a KIR1D variant was identified; whileexhibiting the same 22-bp deletion, this variant lacked the stemregion (corresponding to exon 5), indicating its likely identity as asplice variant (accession no. AY102633). A GenBank search alsorevealed a recently cloned KIR gene that is identical to humanKIR1D except for a single nucleotide substitution at position 156giving rise to a codon change from arginine to lysine in the exon4-encoded gene product (accession no. AF417554).

LD patterns support KIR1D and KIR2DS4 as alleles

The percentage of Caucasian individuals exhibiting KIR1D is high(78.8%) with an estimated gene frequency of 0.54. Accurately re-vised typing for KIR2DS4 revealed a lower Ag frequency (35.3%)than previously determined with an estimated gene frequency of0.20. LD analysis identified strong negativity between the twogenes, supporting their possible identities as alleles. Strong posi-tive LD with KIR1D was noted with KIR3DL1 and 2DL3, whereasthere was only a very weak LD between KIR2DS4 and these genes,indicating that the previously noted positive LD between KIR2DS4and KIR3DL1 and KIR2DS4 and 2DL3 was likely due to lack ofdistinction between KIR1D and 2DS4. Strong negative LD be-tween KIR1D and 2DS1 is also seen. Other LD relationships areindicated in Table II.

FIGURE 1. KIR genotypes observed in a population of unrelated Caucasian individuals. Filled boxes indicate the presence of KIR gene; open boxesindicate the absence of KIR gene. Each of the 36 unique genotypes is identified by letter code. The number of individuals exhibiting each genotype isindicated. Haplotype combinations were deduced from family segregation and sibling studies. Genotype AG, corresponding to homozygosity for haplotypeA-1D, was the most frequently identified genotype in the population study.

5121The Journal of Immunology

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

A pseudogene 3DP1 variant defines a partial KIR haplotype

KIR typing for pseudogene 3DP1 revealed the presence of two pre-dominant subtypes. These can be identified as two different ampli-cons, whose 1.5-kb size difference is due to the presence or absenceof exon 2 and its flanking intron sequences. KIR3DP1, characterizedby the absence of exon 2, was found in the genomic DNA sequenceof a fully sequenced human chromosome 19 KIR haplotype (acces-sion no. AC011501.7); KIR3DP1v, characterized by the presence ofexon 2, was found in a separate chromosome 19 KIR haplotypegenomic sequence (accession no. AL133414). LD for KIR3DP1 andKIR3DP1v reveal the two exhibiting strong negative LD, supportinga possible allelic relationship. Whereas KIR3DP1 has a high genefrequency within the population (0.72), KIR3DP1v exhibits a signif-icantly lower gene frequency (0.17). KIR3DP1 displays a strong pos-itive LD with pseudogene KIR2DP1 and with 2DL1, supporting theirinclusion in a partial haplotype KIR2DP1-KIR2DL1-KIR3DP1. Incontrast, KIR3DP1v displays an equally strong negative LD withKIR2DP1 and 2DL1; moreover, the identical frequency of these genepairs in relation to each other further supports their definition of aKIR3DP1v partial haplotype distinguished by the absence ofKIR2DP1 and 2DL1. In other words, when KIR3DP1v is present,KIR2DP1 and 2DL1 are absent. This partial haplotype lackingKIR2DP1-2DL1 en bloc can be observed in PCR-based genotypingonly when the individual is homozygous for KIR3DP1v, a relativelyrare combination noted in five of the genotypes from the Caucasoidpanel studied (see Fig. 1, genotypes Z, AA, AB, AC, and AD) and inextended family study haplotype elucidation. In addition, KIR3DP1vwas found to be in strongly positive LD with KIR2DL2 and KIR2DS2extending the partial haplotype to resemble KIR2DL2-(absent 2DP1-absent 2DL1)-3DP1v.

Family segregation studies define haplotypes

Family studies performed to date have not included typing forKIR1D or for the pseudogenes 3DP1, 3DP1v, and 2DP1, whosetyping significantly helped to define distinct haplotypes. Twelvefamilies consisting of both parents and two children were analyzedfor gene content; of these, six families were selected for extendedfamily analysis based on heterozygosity at the KIR3DL1-3DS1 lo-

cus and the likelihood of haplotype resolution. Through typing ofmultiple generations and sibships numbering no fewer than eightand in several cases more than 10, KIR haplotypes could be reli-ably resolved with nearly no ambiguities. Examples of genotypeanalysis for individuals from two families and the resolved paren-tal and grandparental haplotypes are shown in Fig. 4. From 12families, 11 different haplotypes were resolved by gene content

FIGURE 3. Genomic DNA and cDNA typing for KIR2DS4 and KIR1Din KIR1D homozygous and heterozygous individuals. Results of cDNAtyping are identical to genomic DNA typing. Lanes 1, 3, 5, 7, 9, and 11 aretyping for an individual heterozygous for KIR2DS4 and KIR1D. Lanes 2,4, 6, 8, 10, and 12 are typing for an individual homozygous for KIR1D. Forcomparison, both individuals exhibit genomic and cDNA positivity forKIR2DL3, an inhibitory KIR. Markers are indicated as base pairs.

FIGURE 2. KIR1D is a deletion variant of 2DS4 resulting in significant amino acid change due to a frame shift. A, Nucleotide sequence alignment ofKIR2DS4 cDNA (allele L76672 is used for comparison), KIR1D genomic DNA (haplotype sequence AC011501.7), and KIR1D cDNA (accession no.AY102624) demonstrates a loss of 22 nucleotides from exon 4 in KIR1D. B, Amino acid alignment reveals no homology between KIR1D and 2DS4 afteramino acid 151, but demonstrates 72% homology to a MmKIR1D variant found in the rhesus monkey (accession no. AF334633).

5122 KIR HAPLOTYPE DEFINITION BY GENE CONTENT IN FAMILIES

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

(Fig. 5A), with 2 representing haplotype A (haplotypes 1 and 2), aspreviously defined by Shilling et al. (36) as the presence ofKIR2DL3 and the lack of all activating genes with the exception of2DS4, presented in this study as comprising both 2DS4 and KIR1D.

Haplotype analysis-centromeric half

Segregation analysis of haplotypes supported the following find-ings: as predicted by their negative LD relationship, KIR2DL2 and2DL3 segregated exclusively onto separate chromosomes. In noneof the families could both KIR genes be assigned to the samehaplotype, supporting their relationship as possible alleles. Pseu-dogene 2DP1, KIR2DL1, and pseudogene 3DP1 segregated enbloc and could be associated with either KIR2DL3 or 2DL2, al-though more often with the former (Fig. 5A, haplotypes 1, 2, 5, 6,7 vs haplotypes 8 and 9). In contrast, the presence of the pseudo-gene variant 3DP1v was associated with the absence of pseudo-gene 2DP1 and KIR2DL1 and was observed exclusively withKIR2DL2 in the family studies. This “deletion” partial haplotypewas observed in three of the resolved haplotypes and occurred in9 of the 24 parents (37.5%) (Fig. 5A, haplotypes 3, 4, and 10).Haplotype 11 is an unusual KIR haplotype, seen only in onefamily.

Haplotype analysis-telomeric half-KIR1D redefines haplotype A

Telomeric to KIR2DL4, the following findings were observed:KIR1D and 2DS4 segregated exclusively onto separate chromo-somes (e.g. Fig. 4, C and D). Typing for KIR1D revealed that whatwas previously defined as haplotype A can be divided into twogroups, one containing KIR1D, designated as haplotype A-1D (Fig.5A, haplotype 1), and the other containing KIR2DS4, designated ashaplotype A-2DS4 (Fig. 5A, haplotype 2). In the family analyses,haplotype A-1D was the more frequently observed, representing 20of a total of 48 parental haplotypes (41.7% gene frequency). Incontrast, haplotype A-2DS4 was present in only 6 of a total of 48

haplotypes (12.5%). These frequencies are comparable to the fre-quencies in the panel of 85 unrelated Caucasians (38.8% and11.8%, respectively). Homozygosity for haplotype A-1D was seenin 14 of 85 unrelated Caucasian individuals (16.67%), whereashomozygosity for haplotype A-2DS4 was seen in only 1 of 85unrelated individuals (1.1%). From the same panel, haplotypeA-1D had the highest haplotype frequency of 38.8%, while hap-lotype A-2DS4 had a haplotype frequency of 11.8%.

In one extended family, the exclusive segregation of KIR2DS1from haplotypes containing KIR1D or 2DS4 was unequivocallyobserved (Fig. 4, C and D). This mutual exclusion is supported bythe highly significant negative LD between the genes, suggestingtheir possible relationship as alleles. Likewise, KIR2DS3 and 2DS5were never identified on the same haplotype in our studies. Ac-companying KIR2DS3 and 2DS5 is 2DL5, which was not found tosegregate separately from either of the two activating receptorgenes, except in one haplotype (Fig. 5B, haplotype 12), which waslater found in a sibling pair that displayed full KIR identity. Sub-typing for the expressed (labeled 2DL5.1) and nonexpressed formsof KIR2DL5 (labeled 2DL5.2) identified a haplotype containingboth forms (Fig. 4, A and B, and Fig. 5A, haplotype 9). Otherhaplotypes contained either the nonexpressed or the expressedKIR2DL5 subtype, paired with either 2DS3 or 2DS5. Finally,KIR2DL2 and 2DS2 were found to segregate together for all re-solved haplotypes. To address the issue of nonspecific amplifica-tion or coamplification, 20 amplicons for both KIR2DL2 andKIR2DS2 were sequenced, repeatedly confirming gene specificityof the primers.

Sibling studies

KIR genotyping of 49 sibling pairs allowed for the informed de-duction of 10 additional haplotypes (Fig. 5B). Of the 49 pairs, 17pairs (34.7%) yielded identical genotypes, which could implyidentical KIR haplotype pair combinations. Although this ratio was

Table 2. LD for KIR locia

2DL2 2DL3 2DP1 2DL1 3DP1 3DP1v 3DL1 3DS1 2DL5 2DS3 2DS5 2DS1 2DS4 1D

2DS2 0.22 �0.254 �0.18 �0.194 �0.194 0.109 �0.147 0.031 0.07 0.077 0.031 0.031 �0.005 �0.067p �0.0005 �0.001 �0.025 �0.025 �0.025 �0.0005 NS NS �0.025 �0.0005 NS NS NS NS

2DL2 �0.254 �0.18 �0.194 �0.194 0.11 �0.147 0.031 0.07 0.077 0.012 0.031 �0.005 �0.067p �0.001 �0.025 �0.025 �0.025 �0.0005 NS NS �0.025 �0.0005 NS NS NS NS

2DL3 0.166 0.18 0.18 �0.099 0.072 0.026 �0.072 0.003 �0.088 �0.094 0.024 0.069p �0.0005 �0.0005 �0.0005 �0.005 �0.05 NS NS NS �0.025 �0.025 NS NS

2DP1 0.19 0.19 �0.224 0.051 0.018 �0.03 0.038 �0.062 �0.046 0.03 0.03p �0.0005 �0.0005 �0.0005 NS NS NS NS NS NS NS NS

2DL1 0.205 �0.242 0.046 0.028 �0.01 0.041 �0.045 �0.061 0.036 0.021p �0.0005 �0.0005 NS NS NS NS NS NS NS NS

3DP1 �0.242 0.046 0.028 �0.01 0.041 �0.45 �0.061 0.036 0.021p �0.0005 NS NS NS NS NS NS NS NS

3DP1v �0.029 �0.019 �0.033 �0.021 �0.006 �0.01 0.018 �0.043p NS NS NS NS NS NS NS NS

3DL1 �0.163 �0.151 �0.078 �0.067 �0.163 0.043 0.117p �0.025 NS �0.025 NS �0.025 NS �0.0005

3DS1 0.165 0.081 0.102 0.139 �0.043 �0.058p �0.0005 �0.0005 �0.0005 �0.0005 �0.05 NS

2DL5 0.089 0.132 0.165 �0.053 �0.075p �0.0005 �0.0005 �0.0005 �0.05 NS

2DS3 �0.014 0.03 �0.034 0.011p NS NS NS NS

2DS5 0.143 �0.043 �0.102p �0.0005 NS �0.005

2DS1 �0.043 �0.102p �0.05 �0.01

2DS4 �0.153p �0.0005

a LD coefficients for two-locus associations were calculated from unrelated Caucasian individuals according to Mattiuz et al (45). Pairs with p � 0.05 are indicated.

5123The Journal of Immunology

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

FIGURE 4. Continues.

5124 KIR HAPLOTYPE DEFINITION BY GENE CONTENT IN FAMILIES

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

higher than the 25% expected from normal Mendelian segregation,it could be explained by high frequency of haplotype A-1D in thesesibling pairs (12 of 17). Given the high frequency of this haplotypein the population, it might be expected that in some cases whereboth siblings exhibit this haplotype, they may not be identical hap-lotypes by descent. Eleven pairs (22.5%) exhibited no shared hap-lotypes, while 21 pairs (42.9%) exhibited one shared haplotype. Ofthe 10 new KIR haplotypes deduced among the sibling pairs, onehaplotype contained both expressed and nonexpressed variants ofKIR2DL5 (Fig. 5B, haplotype 22). Combinations from the deducedsibling haplotypes in addition to the 11 identified through familystudies could account for all 36 genotypes seen in the unrelatedCaucasian panel (Fig. 1). Fifteen unrelated individuals among thesibling group were found to be homozygous for 1 of the 24 iden-tified or deduced haplotypes. This included eight instances of hap-lotype 1, two of haplotype 5, and one each of haplotypes 2, 4, 8,14, and 17.

Haplotype frequency in the Caucasian panel

The unrelated Caucasian panel described in Fig. 1 was analyzedfor deduced KIR haplotypes based on the haplotypes characterized

from family studies and sibling pair analysis (Fig. 5, A and B). All36 genotypes could be resolved into corresponding pairs of hap-lotypes as shown in Fig. 1. Of the 170 haplotypes exhibited in 85individuals, the most commonly observed haplotype was haplo-type 1, occurring 66 times (38.8%). In contrast, haplotype 2 wasfound 20 times (11.8%). Comprising the “classical haplotype A”frequency, the sum of frequencies for haplotype 1 (haplotypeA-1D) and haplotype 2 (haplotype A-2DS4) yielded a total haplo-type A frequency of 50.6%, consistent with previous estimates (29,36). Less frequent, but very common, were the haplotypes char-acterized by the deletion partial haplotype (KIR2DS2-2DL2-3DP1v): haplotype 3 had a frequency of 7.7% while haplotype 4had a frequency of 6.5%. In total, “short” haplotypes with 3DL1-2DS4 or 3DL1-1D accounted for 64.7% of all KIR haplotypes inthis Caucasian population.

Consanguineous individuals

Consanguineous cell lines have proved useful for HLA allele iden-tification and HLA haplotype analysis. A panel of 17 cell linesderived from individuals from consanguineous families was ana-lyzed for KIR gene content (Fig. 6). Among these individuals, two

FIGURE 4. KIR typing of multigeneration CEPH families identifies KIR haplotypes by haplotype segregation. A and C, Genotype profiles, includingKIR2DL5 subtyping, for two extended families are shown: PGF, paternal grandfather; PGM, paternal grandmother; MGF, maternal grandfather; MGM,maternal grandmother; F, father; M, mother; C, child. B and D, Identified haplotypes are assigned letter codes: a-d are parental haplotypes, w-z aregrandparental haplotypes. Filled boxes correspond to the haplotype framework genes, as previously described (24). Typing for presumed alleles andKIR2DL5 subtyping are indicated within boxes.

FIGURE 5. Haplotypes identified in family studies are shown in A. Haplotype 1 and 2 comprise the previously described haplotype A. Haplotypesdeduced from sibling studies and consanguineous individuals are shown in B. Two unusual haplotypes (11 and 23) were each identified only once.

5125The Journal of Immunology

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

more unique genotypes were found, yielding two additional uniquehaplotypes (Fig. 5B, haplotypes 18 and 23). Based on the conser-vative estimates for the haplotypes described in this study, five ofthe consanguineous individuals appear to be homozygous for hap-lotype A-1D, while one is homozygous for haplotype A-2DS4.These cell lines could very well be KIR homozygous by descent.For cell lines with larger gene content, it was not possible throughthis analysis to document homozygosity.

Haplotype B is comprised of multiple stereotypical KIRcombinations

Haplotype B has been defined as containing more variable KIRgene combinations and characterized by KIR2DS genes other thanKIR2DS4. According to this definition, 21 different B haplotypeswere resolved. These haplotypes included most of the partial hap-lotypes described by other groups (31, 36); however, in contrast toone previous report (31), there was no representation of the 2DS3-2DS5-2DS1 complement in our studies. Evaluation of the haplo-types revealed specific patterns that are supported by KIR locus LDanalysis. Although KIR3DS1 is most commonly associated withthe presence of multiple 2DS genes, KIR3DL1 could occasionally,but not as commonly, also be found to be associated with thesegenes. KIR2DS3 and 2DS5 were each almost always paired with2DL5, whose subtyping revealed the nonexpressed 2DL5.2 variantexhibiting a stronger positive LD with 2DS3 ( p �� 0.0005) incomparison to 2DS5 ( p � 0.05). In addition, the nonexpressed2DL5.2 variant exhibited a strong positive LD ( p �� 0.0005) with2DL2. The expressed 2DL5.1 variant showed strong positive LDwith 3DS1 ( p �� 0.0005), 2DS3 ( p � 0.0005), and 2DS5 ( p ��0.0005) and no significant LD with 2DL2. These analyses supportearlier reports of the placement of KIR2DL5.2 adjacent to 2DL2 onthe chromosome and 2DL5.1 adjacent to 3DS1 (44). Although pairKIR2DL5-2DS3 could be seen with KIR1D, 2DS4, or 2DS1, pairKIR2DL5-2DS5 was seen only with 2DS1 in this analysis, consis-tent with the strongly positive LD between 2DS5 and 2DS1.KIR2DS2 displayed a strong positive LD with 2DS3, but its mainassociation appeared to be with 2DL2. Although not exclusive,these patterns were supported by the overwhelming majority ofidentified and deduced haplotypes. Only rarely would haplotypesbe identified that did not adhere to these patterns (see Fig. 5, hap-lotypes 11 and 12).

DiscussionThe human immune system has adopted a strategy of immensediversity, exemplified by the MHC, whose variable gene contentand allelic polymorphism combine to individualize the HLA ge-

notype. Like the HLA, the KIR region is rapidly emerging as anequally diverse region that uses the similar evolutionary strategiesof variable gene content and polymorphism to achieve similardepths of diversity. Addressing the need to identify unique KIRgenotypes are increasingly refined methods of KIR gene detection.Initial studies first revealed variations in KIR gene content fromindividual to individual, whose KIR gene combinations roughlycorresponded to two KIR haplotype groups, A and B (29, 30).These gene content studies have been extended to population anal-yses, which while revealing a great diversity in KIR genotypes,still supported the model of two general haplotype groups (31–35).Providing a still higher resolution of the KIR haplotypes has beentyping analyses of specific KIR gene alleles (36–38) and sequenceanalysis of three haplotypes (Refs. 24 and 39 and accession no.AC011501.7). Although these two approaches have provided de-tailed analyses of KIR haplotypes, one based on sequencing, theother based on allele detection, they have also supported the pres-ence of two main haplotype groups.

Haplotype A has been defined as containing KIR3DL3, -2DL2,-2DL1, -2DL4, -3DL1, -2DS4, and -3DL2, and haplotype B hasbeen defined as having more KIR genes than the A haplotype;generally, these additional genes are the activating KIR genescharacterized by short cytoplasmic tails. The ability to unambig-uously define haplotypes has previously been approached by fam-ily studies; however, these studies have typically included fairlysmall family cohorts, consisting of three to six members. In thisstudy, we report the elucidation of haplotypes through gene con-tent analysis of extended family pedigrees whose large sibshipsand multiple generations could reliably identify specific haplo-types. Eleven different haplotypes were resolved through thesestudies, many of which have not previously been described. Con-tributing significantly to the resolution of these haplotypes was thetyping for two KIR loci in particular, the 2DS4/KIR1D locus andthe pseudogene 3DP1 locus. In this study, we report for the firsttime the cDNA isolation of human KIR1D, whose typing has notpreviously been distilled from that of 2DS4 and which appears tobe the more prevalent of the two putative alleles of the 2DS4 locus.In addition, typing for the two known variants of pseudogene3DP1 has revealed the presence of a “deletion” partial haplotypethat is distinguished by the presence of the pseudogene 3DP1vwith KIR2DS2 and 2DL2, and the absence of 2DL1 and pseudo-gene 2DP1.

KIR1D has not before been identified in humans and was recentlyidentified in the rhesus monkey as Mm-KIR1D (17). Although thenucleotide sequences do not show significant homology, the amino

FIGURE 6. KIR locus genotypes in consanguineous individuals. Genotype AG, corresponding to homozygosity for haplotype A-1D, is represented mostfrequently. These individuals may be homozygous for the KIR region by descent.

5126 KIR HAPLOTYPE DEFINITION BY GENE CONTENT IN FAMILIES

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

acid sequences between the two gene products display a high degreeof homology, suggesting they may be functional orthologs. In addi-tion, while the Mm-KIR1D gene appears to be a deletion variant ofMm-KIR3DL7, the human KIR1D is likely a deletion variant of 2DS4.Interestingly, the respective deletions in the rhesus and human genesresult in frame shifts that yield homologous predicted protein prod-ucts. As in the rhesus monkey, it also appears that several putativesplice variants of human KIR1D are present, with at least one isolatedclone having a deletion of the exon encoding the stem region. KIR1Dencodes a predicted protein product that has a complete D1 Ig domainand a portion of the D2 domain, before the frame shift deletion ab-rogates any transmembrane or cytoplasmic region. Splice variants ofother KIR/ILT genes have been identified (46, 47), some of which arelacking the sequences encoding the transmembrane and cytoplasmicregions, leading one to speculate about the possibility of a secretedtruncated protein product. However, the presence of an encoded se-creted protein such as KIR1D has not heretofore been described. Theligands for the KIR1D proteins, both in the human and the rhesusmonkey, remain unknown; however, the likelihood that that it is anMHC-like molecule is remote. Of interest is the difficulties in isolatingKIR1D relative to 2DS4 transcripts in a heterozygous individual, de-spite the high prevalence of the gene within the population. The sig-nificance of this apparent transcriptional discord remains unclear. Incontrast, transcripts were easily obtainable from an individual foundto be homozygous for KIR1D.

A new model for KIR haplotypes

In this study, we describe extended family studies with very largesibships to define specific haplotypes, following them throughthree generations within the families. Analysis of the resolved hap-lotypes invites a reassessment of the initial two haplotype groups,

first proposed by Uhrberg et al. (29) and further supported by thegenomic sequencing of the complete KIR region for a representa-tive of the A and B haplotypes (24). From our studies, it is evidentthat what was traditionally viewed as haplotype A is actually acombination of two different genotype groups, one containingKIR2DS4 and the other containing KIR1D in association withpseudogene 3DP1. The latter subtype is the more common, com-prising 73% of haplotype A cases, and observed in 39% of unre-lated Caucasian individuals. In contrast, the 2DS4-positive haplo-type A has a haplotype frequency of 12%. The combined frequencyof both haplotype A subtypes of 51% is consistent with previouslypublished estimates (29, 36). It is interesting to note that becausethe most common genotype in the Caucasian population reflectshomozygosity for the haplotype A containing KIR1D (14%), theseindividuals are lacking all activating KIR receptors, with the excep-tion of KIR2DL4, a ubiquitously found receptor characterized by along cytoplasmic tail, but exhibiting activating function (48, 49).

Although our studies are consistent with the broad classificationof KIR regions into A and B haplotypes, the inclusion of typing infamilies for the KIR pseudogenes and the discovery of the humanKIR1D as a likely allele of the 2DS4 locus has provided a moredetailed view of the KIR genomic region, as also recently de-scribed by Shilling et al. (36). Our data can best be accommodatedby an alternative model for KIR haplotypes (Fig. 7). The KIR hap-lotype can be considered as two halves, the centromeric half bor-dered upstream by KIR3DL3 and comprised of those KIR genesupstream of anchor gene KIR2DL4 and the telomeric half bordereddownstream by 3DL2 and comprised of those KIR genes down-stream of 2DL4. The centromeric half may be characterized by thepresence of KIR2DL3 or 2DL2, but not both, and rarely, neither. Inthe rare case where neither KIR2DL3 nor 2DL2 was identified, the

FIGURE 7. KIR haplotype model. Three partial haplotypes comprise gene combinations centromeric to KIR2DL4. Pseudogene 3DP1v is paired withKIR2DL2 and defines a partial haplotype lacking both KIR2DL1 and pseudogene 2DP1 (haplotypes 3 and 6). KIR2DL2 is paired with 2DS2. The areatelomeric to KIR2DL4 is more variable (haplotypes 4–6), but with generally predictable rules of combination, including the allelic relationship betweengenes 3DL1 and 3DS1 and between genes 2DS1, 1D, and 2DS4. KIR2DL5 is paired with 2DS3 or 2DS5. As indicated by an asterisk (�) haplotypes withboth expressed and nonexpressed 2DL5 variants likely position the nonexpressed 2DL5.2 variant adjacent to 2DL2. The nonexpressed variant is more likelypaired with 2DS3 than with 2DS5.

5127The Journal of Immunology

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

possibility exists for the presence of a new KIR gene (S. Chida andD. Geraghty, personal communication). In our studies, KIR2DL2was found always to segregate with 2DS2, the latter occupying themore upstream position of the two. Alteration of our model toreflect the gene position of KIR2DS2 is consistent with a reinter-pretation of the RP5 chromosomal KIR genomic sequence (acces-sion no. AL133414) and with gene order studies performed by S.Chida and D. Geraghty (personal communication). Following thelocus occupied by either KIR2DL3 or 2DL2 is either pseudogene3DP1v or the trio of pseudogene 2DP1-KIR2DL1-pseudogene3DP1. In our studies, KIR2DL3 was exclusively associated withthe trio 2DP1-2DL1-3DP1 to define a common partial haplotype(Fig. 7, haplotype models 1 and 4). Although this trio could also beseen to associate with KIR2DL2 and 2DS2 (Fig. 7, haplotype mod-els 2 and 5), 2DL2 and 2DS2 are more commonly found to asso-ciate with the pseudogene 3DP1v, in a partial haplotype defined bythe en bloc absence of 2DP1-2DL1-3DP1 (Fig. 7, haplotype mod-els 3 and 6). This partial haplotype is consistent with an availablegenomic sequence of this region of the 2DL2-positive chromo-some (24).

The telomeric half of the complex is alternately defined by thepresence of KIR3DL1 or 3DS1, or rarely, neither. Most commonly,KIR3DL1 is associated with a “short” haplotype, one that is lack-ing most activating KIR genes. In these “short” haplotypes, thelocus adjacent to KIR3DL2 may be occupied by 2DS4, or morecommonly, KIR1D (Fig. 7, haplotype models 1–3). In the “long”haplotypes, a number of activating KIR genes may be seen, almostalways accompanied by KIR2DL5. KIR2DL5 has four known vari-ants, two of which (2DL5.1 and 2DL5.3) are expressed and, in thispaper, collectively labeled 2DL5.1. The two nonexpressed variants(2DL5.2 and 2DL5.4) are collectively labeled 2DL5.2. Previousgene order studies (44) and LD analysis performed in this reportsupport the likelihood that the nonexpressed 2DL5 variants arelocated in the centromeric half adjacent to 2DL2 and are mostoften accompanied by 2DS3, while the expressed 2DL5 variantsremain in the telomeric half accompanied by 2DS3 or 2DS5 (Fig.7, haplotype models 5 and 6). In our studies, haplotypes containingboth expressed and nonexpressed KIR2DL5 variants have beenidentified. Finally, in the long haplotypes, the locus adjacent toKIR3DL2 can be occupied by KIR1D, 2DS4, or 2DS1 (Fig. 7,haplotype models 4–6). Accordingly, the vast majority of our datafits the model presented in Fig. 7, which classifies KIR haplotypesinto one of six major groups. Haplotype 1 in Fig. 7 is synonymouswith the canonical haplotype A, with the remainder comprising theclassical B haplotype. The documentation of the three short hap-lotypes can readily be seen in the family studies. The high fre-quency of these haplotypes with their small gene content in thetelomeric KIR region is evident also in the sibling analysis and inthe unrelated panel studies.

Not all 30 permutations within the three major subtypes of thelong B haplotypes (Fig. 7, haplotypes 4–6) were identified in thefamily studies, in part due to the limitations of the family materi-als, but also likely in part due to differences in prevalence for eachpermutation. For instance, haplotype 5 (Fig. 5) identified in thepedigree studies was the most common long haplotype found inthe population analysis, occurring in 21 of 85 individuals (24.7%)with a haplotype frequency of 12.4%. Because KIR2DL5 was ob-served to pair with either 2DS3 or 2DS5, with one rare exception(Fig. 5B, haplotype 12), it is therefore very possible that KIR2DS3and 2DS5 are alleles of the same locus, but our studies cannotresolve this. Although we did not observe KIR2DL5, 2DS3, and2DS5 segregating together, it is theoretically possible for such ahaplotype to exist if both 2DL5 subtypes are present. Similarly, wedid not observe KIR1D, 2DS1, and 2DS4 together either as pairs or

as triplets, which in combination with their negative LD, suggeststheir relationship as alleles. Although the nucleotide sequencingdata would support KIR1D and 2DS4 being alleles, we cannotunequivocally claim 2DS1 as an allele of the same locus. Identi-fication of possible recombinant haplotypes or direct sequencing ofappropriate KIR genomic regions will be needed to resolve thisissue. The finding that the pairing of 2DL5/2DS5 most frequentlysegregated with 2DS1 was also very characteristic for well-definedhaplotypes in the families. We have no explanation for the mech-anisms underlying these phenomena, but some functional interac-tions between the gene products may have preserved this relation-ship during evolution.

It is to be expected that “new” unusual “recombinant” haplo-types or haplotypes with gene duplication of fragments of the KIRregion will be identified which cannot be accommodated by ourmodel. This would be expected for a genomic region with multiplehighly homologous genes, as has been observed for the HLA chro-mosomal region (50). The emerging picture of the gene contentwithin the KIR region is becoming very similar to the findingsmade for the HLA class II region. In this study, the HLA-DR regiondemonstrates a variable gene content with functional class II genesand pseudogenes. Recently, several other genomic regions withinthe human genome have displayed similar variable gene compo-sitions (50).

AcknowledgmentsWe thank Drs. Yatin M. Vyas, Hina Maniar, and Linda Butros for helpfuladvice and suggestions and Karen Reede for preparing the manuscript. Wealso thank Seow Fong for performing some of the statistical computations.We also thank Drs. Shohei Chida and Dan Geraghty (Fred HutchinsonCancer Research Center, Seattle, WA) for permission to cite their unpub-lished findings regarding KIR gene order.

References1. Biron, C. A., K. B. Nguyen, and G. C. Pein. 2002. Innate immune responses to

LCMV infections: natural killer cells and cytokines. Curr. Top. Microbiol. Im-munol. 263:7.

2. Trinchieri, G. 1989. Biology of natural killer cells. Adv. Immunol. 47:187.3. Ljunggren, H. G., and K. Karre. 1990. In search of the missing self: MHC mol-

ecules and NK recognition. Immunol. Today 11:237.4. Moretta, A., C. Bottino, M. Vitale, D. Pende, C. Cantoni, M. C. Mingari,

R. Biassoni, and L. Moretta. 2001. Activating receptors and coreceptors involvedin human natural killer cell-mediated cytolysis. Annu. Rev. Immunol. 19:197.

5. Steinle, A., P. Li, D. L. Morris, V. Groh, L. L. Lanier, R. K. Strong, and T. Spies.2001. Interactions of human NKG2D with its ligands MICA, MICB, and ho-mologs of the mouse RAE-1 protein family. Immunogenetics 53:279.

6. Braud, V. M., D. S. Allan, A. O’Callaghan, K. Soderstrom, A. D’Andrea,G. S. Ogg, S. Lazetic, N. T. Young, J. I. Bell, and J. H. Phillips. 1998. HLA-Ebinds to natural killer cell receptors CD94/NKG2A, B and C. Nature 391:795.

7. Brooks, G. G., F. Borrego, P. E. Posch, A. Patamawenu, C. J. Scorzelli,M. Ulbrecht, E. H. Weiss, and J. E. Coligan. 1999. Specific recognition ofHLA-E, but not classical, HLA class I molecules by soluble CD94/NKG2A andNK cells. J. Immunol. 162:305.

8. Andre, P., R. Biassoni, M. Colonna, D. Cosman, L. L. Lanier, E. O. Long,M. Lopez-Botet, A. Moretta, L. Moretta, P. Parham, et al. 2001. New nomen-clature for MHC receptors. Nat. Immunol. 2:661.

9. Wagtmann, N., R. Biassoni, C. Cantoni, S. Verdiani, M. S. Malnati, M. Vitale,C. Bottino, L. Moretta, A. Moretta, and E. O. Long. 1995. Molecular clones ofthe p58 NK cell receptor reveal immunoglobulin-related molecules with diversityin both the extra- and intracellular domains. Immunity 2:439.

10. Colonna, M., and J. Samaridis. 1995. Cloning of immunoglobulin-superfamilymembers associated with HLA-C and HLA-B recognition by human natural killercells. Science 268:405.

11. D’Andrea, A., C. Chang, K. Franz-Bacon, T. McClanahan, J. H. Philips, andL. L. Lanier. 1995. Molecular cloning of NKB1: a natural killer cell receptor forHLA-B allotypes. J. Immunol. 155:2306.

12. Philips, J. H., J. E. Gumperz, P. Parham, and L. L. Lanier. 1995. Superantigen-dependent, cell-mediated cytotoxicity inhibited by MHC class I receptors on Tlymphocytes. Science 268:403.

13. Mingari, M. C., C. Vitale, A. Cambiaggi, F. Schiavetti, G. Melioli, S. Ferrini, andA. Poggi. 1995. Cytolytic T lymphocytes displaying natural killer (NK)-like ac-tivity: expression of NK-related functional receptors for HLA class I molecules(p58 and CD94) and inhibitory effect on the TCR-mediated target cell lysis orlymphokine production. Int. Immunol. 7:697.

14. Uhrberg, M., N. M. Valiante, N. T. Young, L. L. Lanier, J. H. Phillips, andP. Parham. 2001. The repertoire of killer cell Ig-like receptor and CD94:NKG2A

5128 KIR HAPLOTYPE DEFINITION BY GENE CONTENT IN FAMILIES

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from

receptors in T cells: clones sharing identical �� TCR rearrangement expresshighly diverse killer cell Ig-like receptor patterns. J. Immunol. 166:3923.

15. Vely, F., M. Peyrat, C. Couedel, J. Morcet, F. Halary, F. Davodeau, F. Romagne,E. Scotet, X. Saulquin, E. Houssaint, et al. 2001. Regulation of inhibitory andactivating killer-cell Ig-like receptor expression occurs in T cells after terminationof TCR rearrangements. J. Immunol. 166:2487.

16. Khakoo, S. I., R. Rajalingam, B. P. Shum, K. Weidenbach, L. Flodin, D. G. Muir,F. Canavez, S. L. Cooper, N. M. Valiante, L. L. Lanier, and P. Parham. 2000.Rapid evolution of NK cell receptor systems demonstrated by comparison ofchimpanzees and humans. Immunity 12:687.

17. Hershberger, K. L., R. Shyam, A. Miura, and N. L. Letvin. 2001. Diversity of thekiller cell Ig-like receptors of rhesus monkeys. J. Immunol. 166:4380.

18. Rajalingam, R., M. Hong, E. J. Adams, B. P. Shum, L. A. Guethlein, andP. Parham. 2001. Short KIR haplotypes in pygmy chimpanzee (bonobo) resemblethe conserved framework of diverse human KIR haplotypes. J. Exp. Med. 193:135.

19. McQueen, K. L., B. T. Wilhelm, K. D. Harden, and D. L. Mager. 2002. Evolutionof NK receptors: a single Ly49 and multiple KIR genes in the cow. Eur. J. Im-munol. 32:810.

20. Barten, R., M. Torkar, A. Haude, J. Trowsdale, and M. J. Wilson. 2001. Diver-gent and convergent evolution of NK-cell receptors. Trends Immunol. 22:52.

21. Brown, M. G., S. Fulmek, K. Matsumoto, R. Cho, P. A. Lyons, E. R. Levy,A. A. Scalzo, and Y. M. Yokoyama. 1997. A 2-Mb YAC contig and physical mapof the natural killer gene complex on mouse chromosome 6. Genomics 42:16.

22. Westgaard, I. H., S. F. Berg, S. Orstavik, S. Fossum, and E. Dissen. 1998. Iden-tification of a human member of the Ly-49 multigene family. Eur. J. Immunol.28:1839.

23. Barten, R., and J. Trowsdale. 1999. The human Ly-49L gene. Immunogenetics49:731.

24. Wilson, M. J., M. Torkar, A. Haude, S. Milne, T. Jones, D. Sheer, S. Beck, andJ. Trowsdale. 2000. Plasticity in the organization and sequences of human KIR/ILT gene families. Proc. Natl. Acad. Sci. USA 97:4778.

25. Strong, S. J., M. G. Mueller, R. T. Litman, N. A. Hawke, R. N. Haire,A. L. Miracle, J. P. Rast, C. T. Amemiya, and G. W. Litman. 1999. A novelmultigene family encodes diversified variable regions. Proc. Natl. Acad. Sci. USA96:15080.

26. Yoder, J. A., M. G. Mueller, S. Wei, B. C. Corliss, D. M. Prather, T. Willis,R. T. Litman, J. Y. Djeu, and G. W. Litman. 2001. Immune-type receptor genesin zebrafish share genetic and functional properties with genes encoded by themammalian leukocyte receptor cluster. Proc. Natl. Acad. Sci. USA 98:6771.

27. Hawke, N. A., J. A. Yoder, R. N. Haire, M. G. Mueller, R. T. Litman,A. L. Miracle, T. Stuge, L. Shen, N. Miller, and G. W. Litman. 2001. Extraor-dinary variation in a diversified family of immune-type receptor genes. Proc.Natl. Acad. Sci. USA 98:13832.

28. Lanier, L. L. 1998. NK receptors. Annu. Rev. Immunol. 16:359.29. Uhrberg, M., N. M. Valiante, B. P. Shum, H. G. Shilling, K. Lienert-Weidenbach,

B. Corliss, D. Tyan, L. L. Lanier, and P. Parham, P. 1997. Human diversity inkiller cell inhibitory receptor genes. Immunity 7:753.

30. Valiante, N. M., M. Uhrberg, H. G. Shilling, K. Lienert-Weidenbach,K. L. Arnett, A. D’Andrea, J. H. Phillips, L. L. Lanier, and P. Parham. 1997.Functionally and structurally distinct NK cell receptor repertoires in the periph-eral blood of two human donors. Immunity 7:739.

31. Crum, K. A., S. E. Logue, M. D. Curran, and D. Middleton. 2000. Developmentof a PCR-SSOP approach capable of defining the natural killer cell inhibitoryreceptor (KIR) gene sequence repertoires. Tissue Antigens 56:313.

32. Norman, P. J., H. A. Stephens, D. H. Verity, D. Chandanayingyong, andR. W. Vaughan. 2001. Distribution of natural killer cell immunoglobulin-likereceptor sequences in three ethnic groups. Immunogenetics 52:195.

33. Norman, P. J., C. V. Carrington, M. Byng, L. D. Maxwell, M. D. Curran,H. A. Stephens, D. Chandanayingyong, D. H. Verity, K. Hameed,D. D. Ramdath, and R. W. Vaughan. 2002. Natural killer cell immunoglobulin-like receptor (KIR) locus profiles in African and South Asian populations. GenesImmun. 3:86.

34. Witt, C. S., C. Dewing, D. C. Sayer, M. Uhrberg, P. Parham, andF. T. Christiansen. 1999. Population frequencies and putative haplotypes of thekiller cell immunoglobulin-like receptor sequences and evidence for recombina-tion. Transplantation 68:1784.

35. Toneva, M., V. Lepage, G. Lafay, N. Dulphy, M. Busson, S. Lester, A. Vu-Trien,A. Michaylova, E. Naumova, J. McCluskey, and D. Charron. 2001. Genomicdiversity of natural killer cell receptor genes in three populations. Tissue Antigens57:358.

36. Shilling, H. G., L. A. Guethlein, N. W. Cheng, C. M. Gardiner, R. Rodriguez,D. Tyan, and P. Parham. 2002. Allelic polymorphism synergizes with variablegene content to individualize human KIR genotype. J. Immunol. 168:2307.

37. Gardiner, C. M., L. A. Guethlein, H. G. Shilling, M. Pando, W. H. Carr,R. Rajalingam, C. Vilches, and P. Parham. 2001. Different NK cell surface phe-notypes defined by the DX9 antibody are due to KIR3DL1 gene polymorphism.J. Immunol. 166:2992.

38. Rajalingam, R., C. M. Gardiner, F. Canavez, C. Vilches, and P. Parham. 2001.Identification of seventeen novel KIR variants: fourteen of them from two non-Caucasian donors. Tissue Antigens 57:22.

39. Martin, A. M., E. M. Freitas, C. S. Witt, and F. T. Christiansen. 2000. Thegenomic organization and evolution of the natural killer immunoglobulin-likereceptor (KIR) gene cluster. Immunogenetics 51:268.

40. Yang, S. Y., E. Milford, U. Hammerling, and B. Dupont. 1989. Description of thereference panel of B-lymphoblastoid cell lines for factors of the HLA system: theB-cell line panel designed for the Tenth International Histocompatibility Work-shop. In Immunobiology of HLA, Volume I: Histocompatibility Testing 1987.B. Dupont, ed. Springer-Verlag, New York, p.11.

41. Vilches, C., R. Rajalingam, M. Uhrberg, C. M. Gardiner, N. T. Young, andP. Parham. 2000. KIR2DL5, a novel killer-cell receptor with a D0–D2 config-uration of Ig-like domains. J. Immunol. 164:5797.

42. Wain, H. M., E. A. Bruford, R. C. Lovering, M. J. Lush, M. W. Wright, andS. Povey. 2002. Guidelines for human gene nomenclature. Genomics 79:464.

43. Selvakumar, A., U. Steffens, and B. Dupont. 1997. Polymorphism and domainvariability of human killer cell inhibitory receptors. Immunol. Rev. 155:183.

44. Vilches, C., C. M. Gardiner, and P. Parham. 2000. Gene structure and promotervariation of expressed and nonexpressed variants of the KIR2DL5 gene. J. Im-munol. 165:6416.

45. Mattiuz, P. L., D. Ihde, A. Piazza, R. Ceppellini, and W. F. Bodmer. 1971. Newapproaches to the population genetic and segregation analysis of the HLA system.In Histocompatibility Testing 1970, P. I. Terasaki, ed. Munskaard, Copenhagen,p. 193.

46. Selvakumar, A., U. Steffens, N. Palanisamy, R. S. Chaganti, and B. Dupont.1997. Genomic organization and allelic polymorphism of the human killer cellinhibitory receptor gene KIR103. Tissue Antigens 49:564.

47. Dohring, C., J. Samaridis, and M. Colonna. 1996. Alternatively spliced forms ofhuman killer inhibitory receptors. Immunogenetics 44:227.

48. Selvakumar, A., U. Steffens, and B. Dupont. 1996. NK cell receptor gene of theKIR family with two IG domains but highest homology to KIR receptors withthree IG domains. Tissue Antigens 48:285.

49. Rajagopalan, S., J. Fu, and E. O. Long. 2001. Cutting edge: induction of IFN-�production but not cytotoxicity by the killer cell Ig-like receptor KIR2DL4(CD158d) in resting NK cells. J. Immunol. 167:1877.

50. Trowsdale, J. 2001. Genetic and functional relationships between MHC and NKreceptor genes. Immunity 15:363.

5129The Journal of Immunology

by guest on March 21, 2018

http://ww

w.jim

munol.org/

Dow

nloaded from