AlSOa

b

c

d

e

A

ARAA

KKHNLK

1

ipaaar

lKmeugiah

Human Immunology 71 (2010) 976–981

Contents lists available at ScienceDirect

0d

ssociations between genes for killer immunoglobulin-like receptors and theirigands in patients with solid tumorsuliman Al Omar a, Derek Middleton a,b, Ernie Marshall c, Dawn Porter d, George Xinarianos e,laide Raji e, John K. Field e, Stephen E. Christmas a,*

Division of Immunology, School of Infection and Host Defence, University of Liverpool, Liverpool, United KingdomTransplant Immunology, Royal Liverpool and Broadgreen University Hospital Trust, Liverpool, United KingdomClatterbridge Centre for Oncology, NHS Foundation Trust, Wirral, Merseyside, United KingdomSt Helens and Knowsley NHS Trust, Merseyside, United KingdomCancer Research Centre, Roy Castle Building, University of Liverpool, Liverpool, United Kingdom

R T I C L E I N F O

rticle history:eceived 10 May 2010ccepted 22 June 2010vailable online 30 June 2010

eywords:iller immunoglobulin-like receptorsuman leukocyte antigensatural killer cellsung canceridney cancer

A B S T R A C T

Killer immunoglobulin-like receptor (KIR) and human leukocyte antigen (HLA) genotypes were analyzedfrompanels of lung (non–small-cell lung cancer [NSCLC] and small-cell lung cancer [SCLC]), colon, and kidneycancer patients and compared with normal control subjects. No significant differences were noted betweenKIR gene frequencies in patients compared with normal subjects. When combinations of KIR genes and theirHLA ligands were considered, there were significant decreases in frequencies of both KIR2DL2 and KIR2DL3in homozygotes for their ligand HLA-C1, and an increase in the frequency of KIR3DL1 and its ligand HLA-Bw4in kidney cancer patients compared with controls. Both associations were partly attributable to changes inligand frequencies alone. NSCLC patients showed a significant increase in the frequency of KIR2DL1 and itsligand HLA-C2 and a corresponding decrease in frequency of KIR2DL3 and its ligand HLA-C1 in homozygotes.In NSCLC, the Ile80 form of HLA-Bw4 was decreased in KIR3DL1� HLA-Bw4� patients, whereas in SCLC theIle80 formwas increased and the Thr80 form decreased in KIR3DS1� HLA-Bw4� patients. These findings areconsistent with increased co-expression of high-affinity inhibitory KIRs and their ligands, potentially result-ing in decreased natural killer cell function, and hencewith natural killer cells having a protective role in lungand kidney cancer but not colon cancer.

� 2010 American Society for Histocompatibility and Immunogenetics. Published by Elsevier Inc. All rights

reserved.

tettKKgi

dcfewiag

cl

. Introduction

Natural killer (NK) cells provide first-line defense against viralnfections and malignantly transformed cells. They comprise ap-roximately 10% of lymphocytes in human peripheral blood andre distinguished from other cells by expression of CD56 in thebsence of CD3. The two subsets, CD56dim and CD56bright NK cells,re mainly responsible for cytotoxicity and cytokine secretion,espectively [1].

NK cells express a range of receptors of several different fami-ies, which govern the killing activity of NK cells, one ofwhich is theiller cell Immunoglobulin-like Receptor (KIR) family. KIRs areembers of the immunoglobulin superfamily of receptors and arencoded on chromosome 19q13.4. The KIR gene cluster comprisesp to 17 highly homologous and closely linked genes and pseudo-enes. KIRs can be classified into 2 different groups: firstly thenhibitory group is characterized by a long (L) intracytoplasmic tailnd 2 or 3 extracellular Ig domains. Secondly, the activating groupas 2 or 3 extracellular Ig domains and a short (S) intracytoplasmic

T* Corresponding author.

E-mail address: sechris@liv.ac.uk (S.E. Christmas).

198-8859/10/$32.00 - see front matter � 2010 American Society for Histocompatibilityoi:10.1016/j.humimm.2010.06.019

ail [2]. The KIR activating receptors function via the adapter mol-cule DAP12, whereas KIR inhibitory receptors inhibit NK cell func-ion via Immunotyrosine inhibitory motifs [3,4]. Ligand specifici-ies for only 4 of the inhibitory KIRs have been clearly defined:IR2DL2 and KIR2DL3 for the HLA-CAsn80 (C1) group of alleles,IR2DL1 for the HLA-CLys80 (C2) group and KIR3DL1 for the Bw4roup of HLA-B (and some A) alleles [5]. Specificities of the remain-ng inhibitory and activatory receptors are less well characterized.

The presence or absence of particular KIR genes has led to theefinition of 2 distinct haplotypes, A and B, based on their geneontent. Four of the KIR genes: KIR3DL3, 3DP1, 2DL4 and 3DL2 areound in all haplotypes. Haplotype A is characterized by the pres-nce of KIR2DL1, 2DL3, 3DL1 and 2DS4, and contains 8 KIR genes,hereas haplotype B is characterized by one ormore of the follow-

ng genes: KIR2DL5, 2DS1, 2DS2, 2DS3, 2DS4, 2DS5 and 3DS1 [6]. Inddition, there is considerable allelic diversity in many of the KIRenes [6].Cells that express class I human leukocyte antigen (HLA) mole-

ules are protected fromNK cell killing, while cells that downregu-ate or lose class I HLA molecules are susceptible to NK cell killing.

herefore, KIR inhibitory receptors block NK cell killing activity by

and Immunogenetics. Published by Elsevier Inc. All rights reserved.

bHas

attdcioiwpaoft

kcctsst[tcK

2

2

rSsHTRgokpcp

2e

adeBrwduW

2

dM

appocwpKv

mfCp-

2

tlctsctuediu

ww

3

3

dAbtwHaltpp

3

llaictCo(swCn

S. Al Omar et al. / Human Immunology 71 (2010) 976–981 977

inding class I HLA molecules and delivering inhibitory signals.owever, lack of expression of class I HLA molecules leads toctivation of NK cell killing activity through delivery of activatingignals [7].

NK cells play a crucial role in protection against viral infectionnd a lack of NK cells leads to susceptibility to herpesvirus infec-ions [8]. Genotyping studies of KIRs and their ligands have shownhem to be involved in protection from or susceptibility to someiseases. In a single patient, the expression of KIR2DL1 on all NKells contributed to recurrent cytomegalovirus (CMV) infection bynhibiting NK cell killing activity [9]. Moreover, the co-expressionf KIR3DS1 and its ligand HLA Bw4Ile80 slowed disease progressionn individuals infected with HIV [10]. Kidney transplant patientsith a higher number of activating KIR genes were significantlyrotected from CMV infection [11]. KIRs may also be involved inutoimmune diseases. In psoriasis patients, KIR2DS1, either aloner in combination with its ligand HLA-Cw6, has been found at highrequency, whereas KIR2DL1 has been found at lower frequencyhan in healthy subjects [12].

For many years, NK cells have been known to be capable ofilling tumor cell lines, and there is also evidence that they areytotoxic toward freshly isolated tumor cells [13]. Several statisti-ally significant associations between frequencies of KIR genes andheir ligands and either protection or susceptibility to leukemias orolid tumors have been reported (reviewed in [14]). In view of thetronger inhibitory interaction between KIR2DL1 and its ligands,he HLA-C2 group of alleles than between KIR2DL2/3 and C115,16], the aims of the present study were to investigate whetherhe combination of KIR2DL1 andC2wasmorehighly represented inancer patients than normal subjects in a population in the Unitedingdom.

. Subjects and methods

.1. Patient and control samples

Human peripheral blood samples were obtained from 255 un-elated healthy Caucasian volunteers from Liverpool and Belfast.amples from 186 non–small-cell lung cancer (NSCLC) and 102mall-cell lung cancer (SCLC) patients were obtained from St.elens andKnowsleyNHS Trust and the Cardiothoracic CentreNHSrust, Liverpool, some directly and others via the Roy Castle Canceresearch Centre, University of Liverpool. This was a heterogeneousroup of patients with some samples being taken at diagnosis andthers at follow-up. DNA samples from 128 colon cancer and 40idney cancer patients were obtained from the University of Liver-ool Cancer Tissue Bank. All samples were obtained with informedonsent of the donor to participate in the study, which was ap-roved by the Cheshire Research Ethics Committee.

.2. Peripheral blood mononuclear cell preparation and DNAxtraction

Bloodwas taken into preservative-free heparin (30 u/ml of blood),nd peripheral blood mononuclear cells (PBMC) were isolated byensity gradient centrifugation. Blood was gently layered onto anqual volume of Ficoll-Paque Plus (GEHealthcare, Chalfont St Giles,uckinghamshire, UK) and centrifuged at 400 g for 20 minutes atoom temperature. PBMC were harvested from the interface andashed twice with phosphate-buffered isotonic saline and spunown at 400 g for 10 minutes. DNA was harvested from �107 PBMCsingDNeasybloodandtissuekit (Qiagenproductno.69504,Crawley,est Sussex, UK) according to themanufacturer’s protocol.

.3. KIR and HLA typing

DNA for HLA and KIR testing was either isolated as above orirectly from patients’ blood using the Qiagen Q1Aamp DNA Blood

ini Kit, which uses aMini Spin column for purification of the DNA p

fter whole-blood digestion with protease K. KIR gene typing waserformed by a sequence-specific oligonucleotide probe methodreviously described [17]. Briefly, two separate amplifications, onef the first and second domains and one of the transmembrane/ytoplasmic regions, were performed. Digoxigenin-labeled probesere added to the amplicons in a chemiluminescence method. Arobe was also used that detects the nondeleted version ofIR2DS4, enabling detection of both the deleted and nondeletedersions of KIR2DS4 [18].HLA class I typing (A, B, and C loci) was performed using poly-

erase chain reaction–sequence specific oligonucleotide proberom Dynal RELI HLA-A, -B, and -C kits (Dynal Biotech, Wirral, UK).ontrols and patients were classified as to whether they wereositive for C1 or C2 groups of HLA-C or both, and for HLA-Bw4 orBw6 or both.

.4. Statistical analysis

The �2 test for comparison of proportions was used to comparehe frequency distribution of each gene allele, genotype, and hap-otype between cancer cases and controls. The comparisonwith theontrol was done separately for each cancer subtype. Fisher’s exactest was used for statistical comparison with small numbers ofubjects (n � 5). Odds ratios and 95% confidence intervals werealculated for gene alleles, genotypes or haplotypes that were sta-istically significant different between cases and controls usingncorrected p value. The significance of the associations was alsoxamined using p value corrected with the step-up adaptive proce-ure false discovery rate. However, thiswas not considerednecessaryn situations in which there were a priori reasons for testing a partic-lar association, for example a receptor–ligand relationship.Statistical analysis was done using StatsDirect statistical soft-

are and Stata statistical software version 10.1. All statistical testsere two-tailed performed at the 5% level of significance.

. Results

.1. KIR gene frequencies

Frequencies of individual KIR genes were compared betweenifferent groups of cancer patients and normal controls (Table 1).lthough there were some differences in KIR gene frequenciesetween groups of cancer patients and healthy controls, none ofhese were statistically significant (p � 0.05). Patients and controlsere then allocated KIR haplotypes according to their genotype.aplotype A was defined by the presence of 2DS4 as the onlyctivating gene, together with 3DL2, 2DL1 and 2DL3, whereas hap-otype B was defined by the presence of any activating gene otherhan 2DS4. However, there were no significant differences in theroportions of patients of different KIR haplotypes in any of theatient groups compared with normal subjects (data not shown).

.2. Combinations of KIR and HLA genes

The frequencies of combinations of KIR genes and their HLAigands were then analyzed to test whether lower-affinity KIR–igand combinations occurred less frequently than the higher-ffinity 2DL2-C2 combination that would lead to greater NK cellnhibition. The frequency of KIR2DL3 together with C1 was signifi-antly reduced in kidney cancer patients compared with the con-rol group (p � 0.04; Table 2). The combination of either of the1-specific KIRs 2DL2 and 2DL3 with homozygous C1 was found innly 25% of kidney cancer patients comparedwith 47.8% of controlsp � 0.014; data from Table 2). Kidney cancer patients were thenelected whose only possible HLA-C–related inhibitory interactionas between 2DL2 or 2DL3 and C1, i.e., they lacked either 2DL1 or2. This group again had a significantly lower frequency than didormal controls (p � 0.03; data from Table 2). Similarly, in NSCLC

atients, there was a significantly reduced frequency of patients

wfiN

ltc(haww

3

ttipfwc(ch

TF

K

2222ND332222222333

N . Ther*

TK

K

2222222223322

Ca

b

c

d

e

e

f

h

g

h

*

S. Al Omar et al. / Human Immunology 71 (2010) 976–981978

ho possessed 2DL3 and homozygous C1 (p � 0.04; Table 2). Therequency of patients having the combination of the strongly inhib-tory interaction of 2DL1 and C2 was significantly increased inSCLC (p � 0.044; Table 2).In viewof the significant association of the reduced frequency of

ow affinity inhibitory KIR-ligand combinations in kidney cancer inhe absence of any significant effect of KIR genotype alone, frequen-ies of KIR ligands per se were examined in the patient groupsTable 3). The frequency of C1 homozygous patients lacking theigh-affinity ligand C2was decreased,whereas the presence of a highffinity inhibitory ligandC2was correspondingly increased comparedith that in normal subjects (p � 0.016). No other HLA-C frequenciesere significantly different from those in normal controls.

able 1requencies of KIR genes in patient and healthy control groups

IR gene Lung cancer (NSCLC*) Lung cancer (SCLC*)(N � 186) (N � 102)n (%) n (%)

DL1 181 (97.3%) 101 (99.0%)DP1 181 (97.3%) 101 (99.0%)DL3 169 (90.8% 94 (92.2%)DS4 176 (94.6%) 98 (96.1%)D 79 (42.5%) 38 (37.3%)el 151 (81.2%) 79 (77.5%)DL1 176 (94.6% 98 (96.1%)DS1 68 (36.6%) 43 (42.2%)DS1 69 (37.1%) 41 (40.2%)DS2 97 (52.2%) 44 (43.1%)DS3 58 (31.2%) 27 (26.5%)DS5 52 (28.0%) 32 (31.4%)DL2 96 (51.6%) 43 (42.2%)DL5 91 (48.9%) 47 (46.1%)DL4 186 (100%) 102 (100%)DL2 186 (100%) 102 (100%)DL3 186 (100%) 102 (100%)DP1 186 (100%) 102 (100%)

umbers of subjects with the gene are shown, with the percentages in parenthesesNSCLC, SCLC.

able 2IR gene frequencies in the presence and absence of their ligands

IR-ligand Lung cancer (NSCLC) Lung cancer (SCLC(N � 186) (N � 102)n (%) n (%)

.DL1�C2� 112 (60.2%)a 59 (57.8%)DL1�C2� 69 (37.1%) 42 (41.2%)DL1�C2/C2 22 (11.8%) 16 (15.7%)DL2�C1� 81 (43.5%) 36 (35.3%)DL2�C1� 15 (8.1%)c 7 (6.9%)DL2�C1/C1 38 (20.4%) 18 (17.6%)DL3�C1� 151 (81.2%) 78 (76.5%)DL3�C1� 18 (9.7%) 19 (18.6%)f

DL3�C1/C1 58 (31.2%)h* 39 (38.2%)DL1�Bw4� 110 (59.1%) 58 (56.9%)DL1�Bw4� 66 (35.5%) 40 (39.2%)DS1�C2 42 (22.6%) 23 (22.5%)DS2�C1 82 (44.1%) 37 (36.3%)

I, confidence interval; OR, odds ratio.p � 0.044, OR � 1.48 (95% CI � 1.01–2.17) for NSCLC vs controls.p � 0.04, OR � 0.44 (95% CI � 0.21–0.91) for kidney cancer vs controls.p � 0.01, OR � 3.6 (95% CI � 1.38–9.57) for NSCLC vs controls.p � 0.009, OR � 5.9 (95% CI � 1.34–24.4) for kidney cancer vs controls.p � 0.002�, OR � 0.08 (95% CI � 0.01–0.61) for kidney cancer vs controls.p � 0.04, OR � 0.47 (95% CI � 0.23–0.98) for kidney cancer vs controls.p � 0.034, OR � 2.1 (95% CI � 1.1–4) for SCLC vs controls.p�0.023, OR � 2.31 (95% CI � 1.15–4.82) for kidney cancer vs controls.p � 0.058, OR � 0.48 (95% CI � 0.23–1.01) for kidney cancer vs controls.

p � 0.0066*, OR � 0.58 (95% CI � 0.39–0.86) for NSCLC vs controls.p value remains statistically significant after Benjamin and Hochberg adaptive False Disc

.3. HLA-Bw4 associations

Although the Bw4 epitope is also present in some HLA-A alleles,his was not considered in this analysis. A significant reduction inhe incidence of homozygosity for HLA-Bw6 (Bw6/Bw6) was foundn kidney cancer patients (25%) comparedwith the controls (43.9%;� 0.023; Table 3). Alleles with the Bw4 epitope of HLA-B can be

urther divided into those with threonine at residue 80 and thoseith isoleucine at that position. Of the kidney cancer patients, 62%arriedHLA-Bw4Thr80 comparedwith only 40%of healthy controlsp � 0.0087; Table 3). A total of 72.5% of kidney cancer patientsarried the KIR3DL1-Bw4 combination, which was significantlyigher than in the control group (53.3%; p � 0.023). When

Kidney cancer Colon cancer Controls(N � 40) (N � 128) (N � 255)n (%) n (%) n (%)

37 (92.5%) 124 (96.8%) 247 (96.9%)37 (92.5%) 125 (97.6%) 247 (96.9%)34 (85%) 118 (92.1%) 232 (91.0%)39 (97.5%) 121 (94.5%) 242 (94.9%)14 (35%) 50 (39%) 95 (37.3%)36 (90%) 102 (79.6%) 205 (80.4%)39 (97.5%) 120 (93.7%) 242 (94.9%)14 (35%) 57 (44.5%) 97 (38.0%)15 (37.5%) 52 (41.6%) 97 (38.0%)18 (45%) 68 (53.1%) 121 (47.5%)8 (20%) 39 (30.4%) 64 (25.1%)

13 (32.5%) 47 (36.7%) 78 (30.6%)18 (45%) 67 (52.3%) 119 (46.7%)17 (42.5%) 68 (53.1%) 118 (46.3%)40 (100%) 128 (100%) 255 (100%)40 (100%) 128 (100%) 255 (100%)40 (100%) 128 (100%) 255 (100%)40 (100%) 128 (100%) 255 (100%)

e were no significant differences between any of the groups and healthy controls.

Kidney cancer Colon cancer Controls(N � 40) (N � 128) (N � 255)n (%) n (%) n (%)

26 (65%) 71 (55.5%) 129 (50.6%)11 (27.5%)b 53 (41.4%) 118 (46.3%)8 (20%) 15 (11.7%) 26 (10.2%)

13 (32.5%) 61 (47.6%) 113 (44.3%)5 (12.5%)d 5 (3.9%) 6 (2.4%)1 (2.5%)e* 33 (25.8%) 61 (23.9%)

27 (67.5%)e 105 (82.0%) 208 (81.6%)7 (17.5%) 13 (10.2%) 25 (9.8%)

11 (27.5%)g 49 (38.3%) 112 (43.9%)29 (72.5%)h 75 (58.6%) 136 (53.3%)10 (25%) 45 (35.2%) 108 (42.4%)11 (27.5%) 25 (19.5%) 52 (20.4%)13 (32.5%) 61 (47.6%) 116 (45.5%)

)

overy Rate correction.

KiscAtrca

3

pcaco(ctccn

4

mcmwtg

sc

iitcbnwKppti2slKdarc

lsHca8ml

TF

C

CCCCCBB

BBB

a

b

c

d

d

* e Disc

TF

K

BBB

a

b

c

S. Al Omar et al. / Human Immunology 71 (2010) 976–981 979

IR3DL1� and Bw4� patients and controls were analyzed accord-ng to the presence of Thr or Ile in position 80, only NSCLC patientshowed a significant decrease in the percentage expressing Ile80ompared with controls (Table 4; 25.5% vs 39.7%; p � 0.018).mong the patient groups, only the SCLC group had an increase inhe frequency of the KIR3DS1�Bw4Ile80� combination and a cor-esponding decrease in the KIR3DS1�Bw4Thr80� combinationompared with controls (60% vs 28% and 52% vs 76.9%; p � 0.0087nd p � 0.027, respectively; data not shown).

.4. KIR gene frequencies in the presence or absence of ligand

NK cell licensing is dependent upon expression of KIRs in theresence of class I HLA ligands; the frequencies of patients andontrols with or without the ligand specific for KIR2DL1, 2DL2/3nd 3DL1 were therefore analyzed. As shown in Table 2, for kidneyancer a significantly lower proportion of patients were incapablef NK cell licensing for the high-affinity KIR2DL1-C2 combinationp � 0.04) and a significantly higher proportion of patients wereapable of NK cell licensing for the KIR3DL1-Bw4 ligand interactionhan were normal controls (p � 0.023). Other statistically signifi-ant differences were also observed between lung and kidney can-er patients and controls for the low-affinity KIR2DL2/3-C1 combi-ation (Table 2).

. Discussion

One escape mechanism used by tumor cells to avoid T-cell–ediated attack is to downregulate class I HLA molecules at theell surface, thereby rendering them more susceptible to NK cell–ediated lysis [19]. However, complete lack of HLA expressionould facilitate tumor cell killing bymost or all NK cells, andmanyumors showonly partial loss ofHLA expression or loss of heterozy-osity [20]. In this situation, and in light of the variegated expres-

able 3requencies of KIR ligands in patient groups and healthy control subjects

lass I HLA Lung cancer (NSCLC) Lung cancer (SC(N � 186) (N � 102)n (%) n (%)

1 164 (88.2%) 85 (83.3%)2 113 (60.8%) 60 (58.8%)1/C1 72 (38.7%) 42 (41.2%)2/C2 22 (11.8%) 17 (16.7%)1/C2 91 (48.9%) 43 (42.2%)w4Thr80 92 (49.5%)c 45 (44.1%)w4Ile80 28 (15%) 20 (19.6%)Ile80� Thr80 6 (3.22%) 6 (5.9%)w4/Bw4 24 (12.9%) 12 (11.8%)w6/Bw6 72 (38.7%) 43 (42.2%)w4/Bw6 90 (48.4%) 46 (45.1%)

p � 0.016, OR � 2.42 (95% CI � 1.16–5.05) for kidney cancer vs controls.p � 0.016, OR � 0.41 (95% CI � 0.20–0.86) for kidney cancer vs controls.p � 0.050, OR � 1.44 (95% CI � 0.97–2.11) for NSCLC vs controls.p � 0.0087*, OR � 2.46 (95% CI � 1.24–4.89) for kidney cancer vs controls.p � 0.024, OR � 0.43 (95% CI � 0.20–0.91) for kidney cancer vs controls.p Value remains statistically significant after Benjamin and Hochberg adaptive Fals

able 4requencies of Bw4Ile80 and Bw4Thr80 in KIR3DL1� Bw4� patients and controls

IR3DL1 ligand Lung cancer (NSCLC) Lung cancer(N � 103) (N � 58)n (%) n (%)

w4Ile80 26 (25.2%)a 19 (32.7%)w4Thr80 83 (80.5%) 45 (77.6%)w4Ile80� Bw4Thr80 6 (5.8%) 6 (10.3%)

p � 0.018, OR � 0.52 (95% CI � 0.30–0.90) for NSCLC vs controls.

p � 0.0001, OR � 3.1 (95% CI � 1.7–5.6) for colon cancer vs controls.p � 0.0001, OR � 0.3 (95% CI � 0.17–0.56) for colon cancer vs controls.

ion of KIRs by individual NK cells [21], it is likely that some NK celllones would be capable of tumor cell lysis.Fromthedatapresented inTable1and in the text, it is clear thatno

ndividual KIR genes or haplotypes were over- or under-representedn groups of solid tumor patients compared with unaffected con-rols. This is similar to the findings of an earlier study of bladder,olorectal, and laryngeal cancer patients [18]. However,when com-inations of KIRs and their HLA ligands were considered, someotable differences emerged. First, in kidney cancer patients, thereas a significantly decreased incidence of the combination ofIR2DL2 or 2DL3 and their ligand, HLA-C1. This also applied inatients lacking either KIR2DL1 or its ligand C2 and was at leastartly a result of a decreased frequency of C1/C1 homozygotes inhis group. The C2 group of ligands has been found to give strongernhibition of NK cell function via KIR2DL1 than the C1 group withDL2/3 [15,16]. Hence, a decreased incidence of C1 and a corre-ponding increased incidence of C2 might reduce the likelihood ofysis of tumor cells expressing HLA-C2 by NK cells expressingIR2DL1. However, in studies of malignant melanoma, the inci-ence of C1was increased [22] and C2 homozygotes decreased [23]lthough, for the metastatic form, C2 homozygotes were over-epresented [22]. The KIR2DL1-C2 combination was also signifi-antly increased in NSCLC patients compared with controls.An increased incidence of the Thr80 subgroup of HLA-Bw4 al-

eles was also noted in both kidney cancer and NSCLC patients,imilar to that found in a study of malignant melanoma [23]. ThisLA-B configuration can lead toprotection against lysis by someNKell clones that is not found with molecules expressing Ile80 [24],lthough in other situations stronger inhibition was seen with Ile0–expressing ligands [25]. Again, an increased incidence of theore protective group of Bw4 allelesmight reduce the likelihood of

ysis of HLA-Bw4� tumor cells by NK cells expressing KIR3DL1, the

Kidney cancer Colon cancer Controls(N � 40) (N � 128) (N � 255)n (%) (%) n (%)

32 (80%) 113 (88.3%) 229 (89.8%)29 (72.5%)a 73 (57%) 133 (52.2%)11 (27.5%)b 54 (42.2%) 122 (47.8%)8 (20%) 15 (11.7%) 26 (10.2%)

21 (52.5%) 57 (44.5%) 109 (42.7%)25 (62.5%)d* 52 (40.6%) 103 (40.4%)8 (20%) 35 (27.3%) 52 (20.4%)3 (7.5%) 7 (5.46%) 10 (3.9%)5 (12.5%) 20 (15.6%) 28 (11.0%)

10 (25%)d 48 (37.5%) 112 (43.9%)24 (60%) 60 (46.9%) 115 (45.1%)

overy Rate correction.

) Kidney cancer Colon cancer Controls(N � 28) (n � 75) (N � 135)n (%) (%) n (%)

7 (25%) 49 (65.3%)b 51 (37.7%)24 (85.7%) 32 (42.6%)c 95 (70.3%)3 (10.7%) 6 (8%) 11 (8.1%)

LC)

(SCLC

iaae

tfcBKitHKrstic

aacIpt�

HmfKptpl

l[cactasdctlcpcpiwp

hlptlibic

Nc

osliirlp

ttglftm

A

tapt

R

[

[

[

[

[

[

[

[

[

S. Al Omar et al. / Human Immunology 71 (2010) 976–981980

nhibitory ligand for HLA-Bw4. Both of these HLA-C and HLA-Bssociations would be compatible with a situation in which NK cellctivity could be reduced against tumor cells that had not lostxpression of the appropriate high-affinity class I HLA ligands.Therewas a significant decrease in the incidence of KIR3DL1 and

he Ile80 form of Bw4, complementary to the increase in the Thr80orm found when analyzing HLA ligands alone. The apparentlyontradictory findings that both the Thr80 [24] and Ile80 forms ofw4 [25] may give stronger inhibition following interaction withIR3DL1 may result from the extensive polymorphism of KIR3DL1tself whereby different alleles mediate different levels of inhibi-ion [26–28]. Alternatively, the range of peptides associated withLA-Bw4� class I molecules may alter the interaction withIR3DL1. Peptides have previously been shown to be vital in theecognition of HLA-A3 and A11 by KIR3DL2 [29]. The data pre-ented here would be more consistent with the Thr80 form havinghe greater inhibitory capacity. In further studies, it would be ofnterest to analyze which alleles of KIR3DL1 were expressed inancer patients compared with controls.The gene for the activating receptor KIR3DS1, now considered

n allele of KIR3DL1 [30], was only present in 35–45% of patientsnd controls, but there was nevertheless a highly significant in-rease in the proportion of KIR3DS1� SCLC patients who had thele80 form of Bw4 and a corresponding significant decrease in theroportion having the Thr80 form. The ligand for KIR3DS1 has yeto be identified and appears not to be HLA-Bw4, although it has97%homologywith the extracellular domains of KIR3DL1 [31,32].owever, the range of peptides bound by Bw4 containing allelesay potentially influence the interaction with KIR3DS1. It is there-

ore difficult to draw conclusions from the above association unlessIR3DS1 has a higher affinity for Thr80 than for Ile 80 forms of Bw4,ossibly dependent upon peptides bound, leading to enhancedumor cell killing. In contrast, in a studyof hepatitis C virus–infectedatients, the combination of the Ile80 form of Bw4 and KIR3DS1 wasess likely to progress to hepatocellular carcinoma [33].

Several associations between combinations of KIRs and theirigands and a variety of cancers have been reported (reviewed in34]). These include malignant melanoma and several hematologi-al malignancies, both of which may be susceptible to lysis byutologous [35,36] and allogeneic NK cells [37]. Along with theseancers, somepatientswith renal carcinomahave long been knowno be responsive to IL-2 therapy, which can at least be partiallyttributed to NK cells [38]. In the present study, despite the smallerample size, renal carcinoma showed the most highly significantifferences from normal controls in terms of KIRs and their ligands,onsistent with the involvement of NK cells in protection againsthis disease. Although the sample size for colon carcinoma wasarger, there were no significant differences from normal controls,onsistent with the lack of evidence that NK cells play a role inrotection against this cancer. Ideally, a larger sample of renalarcinoma patients would have been studied, but no further sam-leswere available from our tissue bank. It would also have been ofnterest to examine KIR phenotype and NK cell function, and thisill be the subject of a subsequent manuscript (Al Omar et al., inreparation).Regarding the KIR and ligand distribution in lung cancer, the

igher affinity pair of KIR2DL1 and C2 were increased and theower-affinity pair KIR2DL3 and C1 were decreased in NSCLC com-ared with controls, suggestive of NK cell involvement in protec-ion against the disease. NK cells comprise approximately 10% ofymphocytes within normal lung [39] and could potentially comento contact with lung carcinoma cells. A prognostic advantage haseen noted for those lung cancer patients with greater numbers ofnfiltrating NK cells [40]. Lung cancers frequently downregulate

lass I HLA [41], which would increase their sensitivity to attack by

[

K cells. NK cells might also have a protective effect against lungancer metastases in the vascular compartment.The analysis of KIR gene frequencies in the presence or absence

f ligand is relevant to the process of NK cell licensing [42]. Theignificantly decreased incidence of KIR2DL1 in the absence ofigand and increased incidence of KIR3DL1 in the presence of ligandn kidney cancer are consistent with a higher rate of NK cell licens-ng than in controls. The significance of this is unclear but mightesult in a general increase in sensitivity to NK cells of tumor cellsosing expression of the relevant class I HLA molecules, as a higherroportion of patients would have licensed NK cells.In future work, it would be of interest to examine NK cell infil-

ration into biopsy samples from both lung and kidney carcinomaso test for their involvement in protection against carcinoma pro-ression. It would also be relevant to examine HLA expression andoss of heterozygosity, particularly at the allelic level, in tumorsromKIR and HLA genotyped patients. It is conceivable that tumorshat retain expression of a ligand for an inhibitory KIR would re-ain resistant to NK cell–mediated killing.

cknowledgments

Colon and kidney cancer patient samples were obtained fromheUniversity of Liverpool Cancer TissueBankResearchCentre. Theuthors acknowledge Lucy Berresford for collecting lung canceratient samples from Aintree Hospital. The authors assert thathere are no competing financial interests in relation to the work.

eferences

[1] Caligiuri MA. Human natural killer cells. Blood 2008;112:461–9.[2] YokoyamaWM, Plougastel BF. Immune functions encoded by the natural killer

gene complex. Nat Rev Immunol 2003;3:304–16.[3] Burshtyn DN, Scharenberg AM,Wagtmann N, Rajagopalan S, Berrada K, Yi T, et

al. Recruitment of tyrosine phosphatase HCP by the killer cell inhibitor recep-tor. Immunity 1996;4:77–85.

[4] Lanier LL, Corliss BC, Wu J, Leong C, Phillips JH. Immunoreceptor DAP12 bear-ing a tyrosine-based activation motif is involved in activating NK cells. Nature1998;391:703–7.

[5] Uhrberg M. The KIR gene family: Life in the fast lane of evolution. Eur J Immu-nol 2005;35:10–5.

[6] BashirovaAA,MartinMP,McVicarDW,CarringtonM.Thekiller immunoglobulin-like receptor gene cluster: Tuning the genome for defense. Annu Rev GenomicsHumGenet 2006;7:277–300.

[7] Bix M, Liao NS, Zijlstra M, Loring J, Jaenisch R, Raulet D. Rejection of class IMHC-deficient haemopoietic cells by irradiated MHC-matched mice. Nature1991;349:329–31.

[8] Orange JS. Humannatural killer cell deficiencies and susceptibility to infection.Microbes Infect 2002;4:1545–58.

[9] Gazit R, Garty BZ, Monselise Y, Hoffer V, Finkelstein Y, Markel G, et al. Expres-sion of KIR2DL1 on the entire NK cell population: A possible novel immunode-ficiency syndrome. Blood 2004;103:1965–6.

10] Martin MP, Gao X, Lee JH, Nelson GW, Detels R, Goedert JJ, et al. Epistaticinteraction between KIR3DS1 and HLA-B delays the progression to AIDS. NatGenet 2002;31:429–34.

11] Stern M, ElsÅsser H, H×nger G, Steiger J, Schaub S, Hess C. The number ofactivating KIR genes inversely correlates with the rate of CMV infection/reactivation in kidney transplant recipients. Am J Transplant 2008;8:1312–7.

12] Holm SJ, Sakuraba K, Mallbris L, Wolk K, StÇhle M, SÂnchez FO. Distinct HLA-C/KIR genotype profile associates with guttate psoriasis. J Invest Dermatol2005;125:721–30.

13] Carlsten M, Malmberg KJ, Ljunggren HG. Natural killer cell-mediated lysis offreshly isolated human tumor cells. Int J Cancer 2009;124:757–62.

14] Boyton RJ, Altmann DM. Natural killer cells, killer immunoglobulin-like recep-tors and human leucocyte antigen class I in disease. Clin Exp Immunol 2007;149:1–8.

15] Carrington M, Wang S, Martin MP, Gao X, Schiffman M, Cheng J, et al.Hierarchy of resistance to cervical neoplasia mediated by combinations ofkiller immunoglobulin-like receptor and human leukocyte antigen loci. JExp Med 2005;201:1069–75.

16] Parham P. MHC class I molecules and KIRs in human history, health andsurvival. Nat Rev Immunol 2005;5:201–14.

17] Middleton D, Williams F, Halfpenny IA. KIR genes. Transpl Immunol 2005;14:135–42.

18] Middleton D, Vilchez JR, Cabrera T, Meenagh A, Williams F, Halfpenny I, et al.Analysis of KIR gene frequencies inHLA class I characterised bladder, colorectaland laryngeal tumours. Tissue Antigens 2007;69:220–6.

19] M×ller P, HÅmmerling GJ. The role of surface HLA-A, B,C molecules in tumourimmunity. Cancer Surv 1992;13:101–27.

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

S. Al Omar et al. / Human Immunology 71 (2010) 976–981 981

20] Garrido F, Ruiz-Cabello F, Cabrera T, PÊrez-Villar JJ, LÔpez-Botet M, Duggan-Keen M, Stern PL. Implications for immunosurveillance of altered HLA class Iphenotypes in human tumours. Immunol Today 1997;18:89–95.

21] Yawata M, Yawata N, Draghi M, Partheniou F, Little AM, Parham P. MHC classI-specific inhibitory receptors and their ligands structure diverse human NK-cell repertoires toward a balance of missing self-response. Blood 2008;112:2369–80.

22] Campillo JA, MartÎnez-Escribano JA, Muro M, Moya-Quiles R, MarÎn LA,Montes-Ares O, et al. HLA class I and class II frequencies in patients withcutaneous malignant melanoma from southeastern Spain: The role ofHLA-C in disease prognosis. Immunogenetics 2006;57:926–33.

23] Naumova E, Mihaylova A, Stoitchkov K, Ivanova M, Quin L, Toneva M. Geneticpolymorphism of NK receptors and their ligands in melanoma patients: Prev-alence of inhibitory over activating signals. Cancer Immunol Immunother2005;54:172–8.

24] Luque I, Solana R, Galiani MD, GonzÂlez R, GarcÎa F, LÔpez de Castro JA, PeÒa J.Threonine 80 onHLA-B27 confers protection against lysis by a group of naturalkiller clones. Eur J Immunol 1996;26:1974–7.

25] Cella M, Longo A, Ferrara GB, Strominger JL, Colonna M. NK3-specific naturalkiller cells are selectively inhibited by Bw4-positive HLA alleles with isoleu-cine. 80. J Exp Med 1994;180:1235–42.

26] Carr WH, Pando MJ, Parham P. KIR3DL1 polymorphisms that affect NK cellinhibition by HLA-Bw4 ligand. J Immunol 2005;175:5222–9.

27] Thananchai H, Gillespie G, Martin MP, Bashirova A, Yawata N, Yawata M, et al.Cutting edge: Allele-specific and peptide-dependent interactions betweenKIR3DL1 and HLA-A and HLA-B. J Immunol 2007;178:33–7.

28] Kim S, Sunwoo JB, Yang L, Choi T, Song YJ, French AR, et al. HLA allelesdetermine differences in human natural killer cell responsiveness and po-tency. Proc Natl Acad Sci U S A 2008;105:3053–8.

29] Hansasuta P, Dong T, Thananchai H, Weekes M, Willberg C, Aldemir H, et al.Recognition of HLA-A3 and HLA-A11 by KIR3DL2 is peptide-specific. Eur J Im-munol 2004;34:1673–9.

30] Trundley A, Frebel H, Jones D, Chang C, Trowsdale J. Allelic expression patterns

of KIR3DS1 and 3DL1 using the Z27 and DX9 antibodies. Eur J Immunol2007;37:780–7.

[

31] O’Connor GM, Guinan KJ, Cunningham RT, Middleton D, Parham P, GardinerCM. Functional polymorphism of the KIR3DL1/S1 receptor on human NK cells.J Immunol 2007;178:235–41.

32] MorvanM,WillemC, GagneK, KerdudouN, DavidG, SÊbille V, et al. Phenotypicand functional analyses of KIR3DL1� and KIR3DS1� NK cell subsets demon-strate differential regulation by Bw4 molecules and induced KIR3DS1 expres-sion on stimulated NK cells. J Immunol 2009;182:6727–35.

33] LÔpez-VÂzquez A, Rodrigo L, MartÎnez-Borra J, PÊrez R, RodrÎguez M, Fdez-Morera JL, et al. Protective effect of the HLA-Bw4I80 epitope and the killer cellimmunoglobulin-like receptor 3DS1 gene against the development of hepato-cellular carcinoma in patients with hepatitis C virus infection. J Infect Dis2005;192:162–5.

34] Kulkarni S, Martin MP, Carrington M. The yin and yang of HLA and KIR inhuman disease. Semin Immunol 2008;20:343–52.

35] Pierson BA, Miller JS. The role of autologous natural killer cells in chronicmyelogenous leukemia. Leuk Lymphoma 1997;27:387–99.

36] Nasca R, Carbone E. Natural killer cells as potential tools in melanoma meta-static spread control. Oncol Res 1999;11:339–43.

37] Moretta A, Locatelli F, Moretta L, Human NK. Cells: From HLA class I-specifickiller Ig-like receptors to the therapy of acute leukemias. Immunol Rev 2008;224:58–69.

38] Margolin KA. Interleukin-2 in the treatment of renal cancer. Semin Oncol2000;27:194–203.

39] GrÊgoire C, Chasson L, Luci C, Tomasello E, Geissmann F, Vivier E,Walzer T. Thetrafficking of natural killer cells. Immunol Rev 2007;220:169–82.

40] Villegas FR, Coca S, Villarrubia VG, Jimenez R, Chillon MJ, Jareno J, et al.Prognostic significance of tumor infiltrating natural killer cells subset CD57 inpatients with squamous cell lung cancer. Lung Cancer 2002;35:23–8.

41] So T, Takenoyama M, Mizukami M, Ichiki Y, Sugaya M, Hanagiri T, et al.Haplotype loss of HLA class I antigen as an escape mechanism from immuneattack in lung cancer. Cancer Res 2005;65:5945–52.

42] Jonsson AH, Yokoyama WM. Natural killer cell tolerance licensing and othermechanisms. Adv Immunol 2009;101:27–79.