nk cell education in tumor immune surveillance: dnam-1/kir ... › content › ... · maría r....

Research Article

NK Cell Education in Tumor Immune Surveillance:DNAM-1/KIR Receptor Ratios as PredictiveBiomarkers for Solid Tumor OutcomeConcepci�on F. Guillam�on1, María V. Martínez-S�anchez1, Lourdes Gimeno1, Anna Mrowiec1,Jer�onimo Martínez-García2, Gerardo Server-Pastor3, Jorge Martínez-Escribano4,Amparo Torroba5, Bel�en Ferri5, Daniel Abell�an1, Jos�e A. Campillo1, Isabel Legaz6,

María R. L�opez-�Alvarez7, María Rosa Moya-Quiles1, Manuel Muro1, and Alfredo Minguela1

Abstract

Natural killer cell (NKc)-based therapies offer promisingoutcomes in patients with tumors, but they could improvewith appropriate selection of donors and optimization ofmethods to expand NKcs in vitro. Education through licensinginteractions of inhibitory killer cell immunoglobulin-likereceptors (iKIR) and NKG2A with their cognate HLA class-Iligands optimizes NKc functional competence. This work hasevaluated the role of licensing interactions in NKc differenti-ation and the survival of cancer patients.Wehave analyzedKIRand KIR-ligand genes, and the expression of activating (CD16and DNAM-1/CD226) and inhibitory (NKG2A and iKIRs)receptors on peripheral blood NKcs in 621 healthy controlsand 249 solid cancer patients (80 melanoma, 80 bladder, and89 ovarian). Licensing interactions upregulated the expressionof activating CD226, reduced that of iKIR receptors, and

shifted the CD226/iKIR receptor ratio on NKc membranes toactivating receptors. A high tumor burden decreased CD226expression, reduced the ratio of CD226/iKIR, and negativelyaffected patient survival. The progression-free survival (38.1vs. 67.0 months, P < 0.002) and overall survival (56.3 vs. 99.6months, P < 0.00001) were significantly shorter in patientswith lower expression of CD226 onNKcs. Hence, transformedcells can downmodulate these licensing-driven receptor rear-rangements as a specific mechanism to escape NKc immunesurveillance.Our results suggest the importance of theCD226/iKIR receptor ratio ofNKcs induced by licensing interactions ascritical determinants for solid cancer immune surveillance,and may provide predictive biomarkers for patient survivalthat may also improve the selection of donors for NKc immu-notherapy. Cancer Immunol Res; 6(12); 1537–47. �2018 AACR.

IntroductionThe role of natural killer cell (NKc) education in cancer

immune surveillance is yet to be established. NKcs become fullycompetent or hyporesponsive during a process known as "licens-ing," where NKG2A or inhibitory killer cell immunoglobulin-likereceptors (iKIR) interact with their respective human leukocyteantigen class-I (HLA-I) ligands (1–7). Seminal research has shownthat only NKcs expressing iKIRs for self-HLA-I ligands are func-

tionally active (8) as they are "educated" upon iKIR/self-HLA-Iinteractions, and only then they are able to exert alloreactivityagainst mismatched allogeneic targets that do not express self-HLA iKIR ligands ("missing-self"; refs. 8–10). NKc education isalso important in an autologous setting and contributes toimmune surveillance against solid tumors (11). This is particu-larly relevant when transformed cells do not downregulate, butinstead upregulate HLA expression, as has been reported inmyelomas where the overexpression of HLA-I makes effectivecancer immune surveillance highly dependent on multiple anddiverse licensing interactions (12).

Licensing interactions include KIR2DL1/HLA-C2 allotypes(Lys80), KIR2DL2-3/HLA-C1 allotypes (Asn80; ref. 13), andKIR3DL1/Bw4 HLA-I alleles (14). Furthermore, the methionine/threonine (M/T) dimorphism at position �21 of the leadersequence of HLA-B allotypes, presented by HLA-E, provides dif-ferential education for NKcs. The ancestral haplotype encoding�21M promotes NKc education mainly through their specificinteraction with NKG2A, whereas haplotypes encoding �21T dothis mostly through KIR binding (5). Additionally, experimentalmodels indicate that other nonclassical or nonhistocompatibilityligands might also be involved in NKc education (6). HumanNKcs can also express up to 6 different activating KIR (aKIRs):KIR2DS1-5 and KIR3DS1. Only the interactions betweenKIR2DS1/HLA-C2 allotypes (15, 16), KIR2DS4/HLA-A�1102,and to a limited number of HLA-C1/-C2 alleles andKIR3DS1�014/HLA-Bw4 (17–19) have been reported, although

1Immunology Service, Hospital Clínico Universitario Virgen de la Arrixaca(HCUVA), Instituto Murciano de Investigaci�on Biosanitaria (IMIB), Murcia, Spain.2Oncology, HCUVA, Murcia, Spain. 3Urology, HCUVA, Murcia, Spain. 4Derma-tology, HCUVA, Murcia, Spain. 5Pathology Services, HCUVA, Murcia, Spain.6Forensic Medicine, Universidad deMurcia, Murcia, Spain. 7Centre for PreventiveMedicine, Animal Health Trust, Lanwades Park, Kentford, Newmarket, Suffolk,United Kingdom.

Note: Supplementary data for this article are available at Cancer ImmunologyResearch Online (http://cancerimmunolres.aacrjournals.org/).

C.F. Guillam�on,M.V.Martínez-S�anchez, and L. Gimeno contributed equally to thisarticle.

Corresponding Author:Alfredo Minguela, Servicio de Inmunología, Virgen de laArrixaca University Hospital Ctra. Madrid-Cartagena, El Palmar, 30120-Murcia,Spain. Phone: 349-683-95379; E-mail: [email protected]

doi: 10.1158/2326-6066.CIR-18-0022

�2018 American Association for Cancer Research.

CancerImmunologyResearch

www.aacrjournals.org 1537

on July 23, 2020. © 2018 American Association for Cancer Research. cancerimmunolres.aacrjournals.org Downloaded from

Published OnlineFirst September 21, 2018; DOI: 10.1158/2326-6066.CIR-18-0022

http://crossmark.crossref.org/dialog/?doi=10.1158/2326-6066.CIR-18-0022&domain=pdf&date_stamp=2018-11-15

http://cancerimmunolres.aacrjournals.org/

their roles in NKc education are less well known, with theexception of KIR2DS1, which renders NKc hyporesponsive inHLA-C2 homozygous individuals (15, 16). This suggests thateducation via aKIRs is also dependent on the presence of theirligands and could represent a mechanism that has evolved toprevent autoreactivity; however, it could have a detrimental effecton tumor immune surveillance by NKcs (20).

NKc education is a quantitative process in which the strengthof the signal delivered by particular licensing interactions isproportional to the efficiency of the responses (21, 22). However,it is not fully understood how specific licensing interactions arelinked to the functional development of NKcs. Studies havereported that the educationofNKc is related to ahigher expressionof the activating molecule CD226 (DNAM-1; ref. 23), down-modulation of iKIR receptors (24), and has helped us to under-stand how activating signals can predominate over inhibitoryones at the immune synapse to promote more effective cytolyticresponses by licensed NKcs. Thus, according to the "confiningmodel," NKc education could induce receptor confinementby compartmentalization to prompt efficient NKc contacts andactivation (6).

The clinical implications of the rearrangement of inhibitoryand activating receptors on licensed NKcs have been minimallystudied. However, the disruption of the expression profiles ofthese receptors in defectiveNKcs can lead to higher relapse rates inacute myeloid leukemia (AML; ref. 25) and higher susceptibilityto solid and hematologic cancers (26–30). Experimental modelshave established the importance of CD226 in cancer immunesurveillance. DNAM-1–deficient mice developed more DNAM-1ligand expressing fibrosarcomas and papillomas in response tochemical carcinogens. In addition, these animals showedincreased tumor growth and mortality after tumor transplanta-tion (31). In fact, CD226 enables sustained stable interactionsbetween NK and cancer cells, which is essential for the efficientkilling of transformed cells (23, 32).

Experimental studies and clinical trials on NKc-based immu-notherapies have shown promising antitumor effects (33), butthese therapies are laborious and not free from risks; the useof biomarkers to improve both the selection of patients basedon their suitability for immunotherapy and NKcs donors,will help to overcome the problems associated with thesetherapies.

Our results show that modulation of activating and inhibitoryreceptor expression induced by licensing interactions of NKcs can

result in alterations that may play key roles in cancer immunesurveillance and the survival of patients with solid tumors. Theseresults could be used as prognostic biomarkers for risk stratifica-tion of patients and for better selection of donors for NKcsimmunotherapy.

Materials and MethodsSamples and study groups

This prospective observational study included 621 healthyCaucasian volunteers (control group) and 249 consecutiveCaucasian patients with melanoma (n ¼ 80), bladder (n ¼80), or ovarian (n ¼ 89) tumors (Table 1).

Peripheral blood anticoagulated with EDTA was obtainedfrom all patients (n ¼ 249) and from 42 controls between 2014and 2016 from Hospital Clínico Universitario Virgen de laArrixaca and Hospital General Universitario Santa Lucía(Murcia, Spain). HLA-I (KIR ligands) and KIR genotypes, aswell as the expression of activating (CD16, CD226, and aKIR)and inhibitory (iKIR and NKG2A) receptors on NK and T cellswere analyzed in all samples. Healthy individuals with knownHLA-I and KIR genotypes (n ¼ 579) were added to the controlgroup for genetic studies. Patients were followed until July2017. Ten-year progression-free survival (PFS) and overallsurvival (OS) were analyzed according to the expression ofCD226 and iKIR receptor on NKcs. The institutional reviewboard (IRB-00005712) approved the study. Written informedconsent was obtained from all patients and controls in accor-dance with the Declaration of Helsinki.

HLA-A, -B, and -C and KIR genotypingHLA-A, -B, -C and KIR genotyping was performed in DNA

samples extracted by a QIAmp DNA Blood Mini kit (QIAgen,GmbH) using sequence-specific oligonucleotide PCR (PCR-SSO)and Luminex technology by Lifecodes HLA-SSO and KIR-SSOtyping procedure (Immucor Transplant Diagnostic, Inc.), as pre-viously described (12, 34).HLA-C genotyping alloweddistinctionbetween HLA-CAsn80 (group-C1) and HLA-CLys80 (group-C2)alleles (13). HLA-A and -B genotyping allowed us to distinguishalleles bearing the Bw4 motif according to the amino-acidsequences at positions 77–83 in the a1 domain of the HLAclass-I heavy chain (14), as well as methionine/threonine(M/T) dimorphism at position �21 of the leader sequence ofHLA-B allotypes (5). KIR genotyping identified iKIRs (2DL1-L3/

Table 1. Demographic and clinical characteristics of patients and controls

Cancer patients (n ¼ 249)Melanoma (n ¼ 80) Bladder (n ¼ 80) Ovarian (n ¼ 89)

Sex (M/F, %M)a 43/37 (53.7%) 68/12 (85%) 0/89 (0%)Age (mean � SD) 60.6 � 15.4 71.8 � 10.2 58.7 � 10.7Months from diagnosis, mean � SD 4.6 � 14.5 14.9 � 32.8 7.0 � 19.6Histology (n)b 35/45 36/44 69/20Staging (n)c 51/29 2/78 26/63Treatmentd 10/9/61 51/20/9 32/49/8Progression (yes/no, %) 27/80 (33.7%) 24/80 (30.0%) 37/89 (41.6%)Death (yes/no, %) 7/73 (8.7%) 12/80 (15.0%) 20/89 (22.5%)1Control group: This genetic study included 621 healthy controls (57� 15 years, 50%males); 42 of them (59� 13.1 years, 37%males)were used for the cellular study byflow cytometry. Mean age and sex of total cancer group (n ¼ 249) were 63.1 � 13.5 years, 44.4% males. Controls and patients were all of Caucasian origin.bFor melanoma (nodular/others), bladder (muscle invasive/superficial), ovarian (high-degree serous, undifferentiated, and carcinomas/others).cFor melanoma (Breslow depth <4 mm/>4 mm), bladder cancer (grade I or II/III), and ovarian carcinoma (grade I or II/III or IV).dFor melanoma (interferon-alpha/vemurafenib/others), bladder cancer (BCG/cystectomy with or without chemotherapy/no treatment), and ovarian carcinoma(surgery þ chemotherapy/neoadjuvant chemotherapy þ surgery/palliative).

Guillam�on et al.

Cancer Immunol Res; 6(12) December 2018 Cancer Immunology Research1538




2DL5 and 3DL1-L3) and aKIRs (2DS1-S5 and 3DS1; refs. 35, 36),as well as KIR2DL4, which has both inhibitory and activatingpotential (37). It was not possible to distinguish between KIR2-DL5A and KIR2DL5B. KIR genotypes were classified into A and Bgroups (38).

Flow cytometryThe expression of CD226 (DNAM-1), NKG2A and KIR recep-

tors (KIR2DL1, 2DS1, 2DL2/S2, 2DL3, and 3DL1) on bothCD3–CD16�/þCD56þþ (CD56bright) and CD3–CD16þCD56þ

(CD56dim) NKc subsets and CD3þCD4þ and CD3þCD8þ T cellswas evaluated as a percentage of positive cells and as meanfluorescence intensity (MFI) using LSR-II and DIVA Software(BD), as previously described (12). Photomultiplier (PMT)voltages were adjusted daily using rainbow calibration particles(BioLegend; Supplementary Fig. S1). Fluorescence compensationwas finely adjusted using as reference negative events foreach fluorochrome. The staining protocol included 11-color/12-monoclonal antibody (mAb): CD3AmCyam(clone SK7, BD),CD4 PE-CF594 (RPA-T4, BD), CD8 APCCy7 (SK1, BD), CD16PacBlue (3G8, BD), CD56 BV711 (NCAM16.2, BD), CD158a,hPECy7 (EB6, BD, recognizes both KIR2DL1 and 2DS1),CD158b1/b2,j PE-Cy5 (GL183, Beckman-Coulter, recognizesKIR2DL2, 2DL3, and 2DS2), CD158a FITC (143211, R&D Sys-tems, KIR2DL1), CD158b2 APC (180701, R&D Systems,KIR2DL3), CD158e APC (DX9, R&D Systems, KIR3DL1), CD226PE (11A8, BioLegend), and NKG2A biotin (REA110, MiltenyiBiotech). Streptavidin AF700 (Life Technologies, Molecularprobes) was used to detect biotinylated NKG2A.

Supplementary Fig. S2A shows the gating strategy used for theanalysis. Briefly, an FSC/SSC dotplot was used to identify lym-phocytes; a CD3/CD16dotplot to select CD3þ T cells; a CD4/CD8dotplot to identify principal T-cell subsets; a CD16/CD56 dotplotto identify CD56bright and CD56dim NK cells; a CD56/CD226dotplot to select CD226þ cells; a CD3/NKG2A dotplot to identifyNKG2Aþ cells; a KIR2DL1S1/KIR2DL1 dotplot to distinguishKIR2DL1þ, KIR2DS1þ, double-positive KIR2DL1/S1, and dou-ble-negative KIR2DL1/S1NK cells; and a KIR2DL2S2/KIR2DL3þKIR3DL1 dotplot to distinguish KIR2DL2S2þ, KIR2DL3þ,KIR3DL1þ, and triple-negative KIR2DL2S2/L3/3DL1 NK cells.Internal negative cells were used as isotype controls to set positivecutoffs. Gates were combined to define the following NK cellsubsets: (i) CD56bright; (ii) CD56dim KIR-negative NKG2Aþ NKcells; (iii) CD56dim KIR-negative NKG2A-negative (nonlicensed);(iv) CD56dim NK cells single-positive for either KIR2DL1, 2DS1,2DL2/S2, 2DL3, or 3DL1; and (v) KIR double-positiveKIR2DL1_L2/S2, KIR2DL1_L3, KIR2DL1_3DL1, and KIR2DL2/S2/L3_3DL1 CD56dim NK cells.

Analysis of CD226/iKIR receptor ratios after NKc expansionin vitro

Peripheral blood mononuclear cells (PBMC) were obtainedfrom sodium heparin anticoagulated peripheral blood samplesfrom HLA-C1C1 (n ¼ 3) and HLA-C2C2 (n ¼ 3) healthy donors.NKc expansion was performed in vitro following a protocol basedon the one previously published for obtaining cells duringNKc therapy clinical trials by Leivas and colleagues (39). Briefly,105 PBMCs were dispensed in U-bottom 96-well microplates in atotal volume of 200 mL of complete medium (RPMI, antibiotics,glutamine, 10% of AB-negative human serum), supplementedwith IL2 (200U/mL), IL15 (20 ng/mL), and 100Gy irradiated 5�

104 K562 cells, and incubated at 37�C in 5% CO2. Completemedium containing IL2 (200 U/mL) and IL15 (20 U/mL) wasadded every 48 hours. Expression of inhibitory and activatingreceptors on NKcs on days 0, 2, 4, 9, and 11 was analyzed by flowcytometry following the method described previously. Mycoplas-ma-free K562 cell line (ATCC CCL-243) was authenticated usingthe AmpFLSTR Identifiler PCR Amplification Kit (Thermo FisherScientific) and used after 24 hours of exponential growth incomplete media.

Statistical analysisAll data were collected in a database (Excel2003; Microsoft

Corporation) and analyzed using SPSS-15.0 (SPSS Inc.). Pearsonc2 and two-tailed Fisher exact tests and Mann–Whitney, ANOVA,and post hoc tests were used to analyze categorical (i.e., sex, tumorstaging) or continuous (i.e., age, CD226 MFI) variables, respec-tively. Kaplan–Meier estimator and log-rank test were used toanalyze patient survival (PFS and OS). Time to events (progres-sion or death) was estimated as months from the date of diag-nosis. Receiver operating characteristic (ROC) curve analysis wasused to explore patient survival and to determine the optimalcutoff values for CD226-MFI on total NKcs as well as for CD226/KIR2DL1, CD226/KIR2DL2, CD226/KIR2DL3, and CD226/KIR3DL1 MFI ratios on NKc subsets. Sex, age, tumor type, tumorstaging, and CD226-MFI on total NK cells were evaluated in aCOX regression analysis to confirm positive associations. Thestrength of association was estimated by odd ratio (OR) and95% confidence interval (95% CI). P values <0.05 were consid-ered statistically significant. The Bonferroni correction (Pc) wasapplied when needed.

ResultsClinical and biological endpoints of patients

This study included 621 healthy controls and 249 patients, allof them of Caucasian origin. Table 1 summarizes the sex, age, andclinical characteristics of patients from different cancer groups.Patients were enrolled at different times from diagnosis, with amean of 4.6 � 14.5, 14.9 � 32.8, and 7.0 � 19.6 months formelanoma, bladder, and ovarian cancers, respectively. Histologyshowed a predominance of nodular melanomas (43.8%), withBreslow depth <4 mm (63.75%) or >4 mm (36.25%), muscleinvasive bladder cancers (45%), and high-degree serous ovariancarcinomas (77.5%), and stage III to IV (70.7%). Treatment wasadjusted according to cancer characteristics and patient condition.Progressive disease was observed in 33.7%, 30%, and 41.6% anddeath in 8.7%, 15%, and 22.5% of melanoma, bladder, andovarian cancer patients, respectively.

Genetic background of KIR and KIR ligand (HLA-I) in controlsand cancer patients

First, we compared the frequency of KIR and KIR ligand genesbetween cancer patients and the control group. Patients from allgroups and healthy controls showed similar distribution of mostKIR and KIR-ligand (HLA-C1, -C2, -Bw4, and -A3/A11) genes(Table 2). Melanoma patients showed a significantly higherfrequency of KIR2DS5 (melanoma vs. controls: 41.3% vs.29.1%, P ¼ 0.014; Pc > 0.05; OR ¼ 1.78). The frequencies ofKIR2DL5 (66.3% vs. 51.9%, P¼ 0.01, Pc < 0.05; OR¼ 1.82) andHLA-C1 ligands (73.8%vs. 85.0%, P<0.01; Pc <0.05;OR¼ 0.49)significantly differed between bladder cancer patients and

NK Cell Education in Tumor Immune Surveillance

www.aacrjournals.org Cancer Immunol Res; 6(12) December 2018 1539




controls. Regarding iKIR/HLA-I licensing interactions, bladdercancer patients showed a significant lower frequency ofKIR2DL3/C1 when compared with controls (60.0% vs. 76.5%,P < 0.01; Pc < 0.05; OR ¼ 0.46). There were no significantdifferences in KIR haplotypes (A or B) on centromeric andtelomeric regions between control and cancer groups.

Solid cancer did not alter NKc repertoireSecond,we explored the influence of both solid tumors andKIR

ligands on the NKc repertoire by analyzing the frequency ofCD56bright and CD56dim NKcs and the expression of activatingor inhibitory receptors on NKc subsets. Controls and cancerpatients show similar percentages of total NKcs (SupplementaryFig. S2B, left) orNKc subsets (Supplementary Fig. S2C). However,patients with both HLA-C2 and Bw4 ligands showed a significant(P < 0.05) increase in the total frequency of NKcs when comparedwith patients lacking Bw4 and/or HLA-C2 ligands (Supplemen-tary Fig. S2B, right). In addition, patients bearing HLA-C2 ligandsshowed a significant (P < 0.05) increase in the percentage of singleKIR2DL1þ NKcs when compared with patients lacking C2 alleles(Supplementary Fig. S2D).

iKIR/HLA-I licensing interactions differentially modulateexpression of CD226 and iKIR receptors on NKcs

We next evaluated the influence of iKIR/HLA-I and NKG2A/HLA-E (21M) licensing interactions on the expression (MFI) ofCD16, CD56, CD226, NKG2A, and KIR receptors on single andmultiple KIR-positive NKc subsets for every individual. Prelimi-nary results showed that the expression of CD226, NKG2A, andKIR receptors did not differ between cancer types and the healthycontrol group for both iKIR/HLA-I (Supplementary Fig. S3A) andNKG2A/HLA-E (Supplementary Fig. S3B) interactions. Therefore,pooled data from all subjects (n ¼ 249 þ 42 ¼ 291; Fig. 1) wasconsidered for subsequent analyses.

The presence of specific ligands induced an upregulation ofactivating CD226 receptor and a downmodulation of the corre-sponding iKIR. These changeswere proportional to the number ofspecific ligands present in the genome and depended on thedescribed affinities of each licensing interaction (iKIR/HLA-I;ref. 36; Fig. 1A). Thus, HLA-C2 significantly increased the expres-sion of CD226 (44,7% increase in HLA-C2C2 vs. HLA-C2 neg-ative, P < 0.001) and reduced the expression of KIR2DL1 (43.1%decrease, P < 0.001) on KIR2DL1þNKcs. However, the expressionof the activating KIR2DS1 receptor on single-KIR2DS1þNKcs didnot vary in the presence of its specific HLA-C2 ligand. HLA-C1significantly increased the expression of CD226 on bothKIR2DL2/S2þ (43% increase in HLA-C1C1 vs. HLA-C1 negative,P < 0.001) and KIR2DL3þ (17% increase, P < 0.05) NKc subsets,although this increment was more subtle in cells expressingreceptors with lower affinities forHLA-C1 (KIR2DL3). In contrast,HLA-C1 induced a significant downmodulation of the KIR2DL3receptor on KIR2DL3þ NKcs (26.8% decrease in HLA-C1C1 vs.HLA-C1 negative, P < 0.05). This effect was not observed inKIR2DL2/S2þNKcs, probably due to the inability of the antibodyGL183 to distinguish between KIR2DS2 and KIR2DL2 receptors.Bw4 ligands induced a significant increase in the expression ofCD226 (27.2% increase in the presence of two Bw4 ligands vs.absence of Bw4, P < 0.001) on KIR3DL1þ NKcs, and a modestbut significant reduction in KIR3DL1 expression (18.2%decrease,P < 0.05).

NKcs bearing different combinations of two or more iKIRs(KIR2DL1, 2DL2/S2, 2DL3, or 3DL1) showed significantlyhigher expression of CD226 than their respective single iKIR-positive NKc subsets (P < 0.001; Supplementary Fig. S4A).However, the expression of iKIR receptors did not vary betweensingle and multiple iKIRþ NKc subsets (SupplementaryFig. S4B). The expression of CD226 on double-positiveKIR2DL1þ2DL2/S2þ and KIR2DL1þ2DL3þ NKc subsets wasnot affected by the presence of HLA-C1 or C2 ligands; thesecells have receptors for both ligands (Supplementary Fig. S4C).However, these double-positive iKIR NKc subsets showed aclear and significant downmodulation of both KIR2DL1(43.0% decrease, P < 0.001) and KIR2DL3 (28.3% decrease,P < 0.01) receptors in the presence of double dose (C2C2 orC1C1) compared with absence of their respective ligands.Again, it is highly likely that the monoclonal antibody GL183was not sensitive enough to detect significant modulations inthe expression of KIR2DL2/S2 on these double-positive NKcs.The presence of HLA-C2, the ligand with the highest affinity foriKIR receptors in double-positive KIR2DL1þ3DL1þ NKcs, sig-nificantly increased the expression of CD226 (15.1% increasein HLA-C2C2 vs. HLA-C2 negative, P < 0.05) and reduced thatof KIR2DL1 (26.4% decrease, P < 0.01). However, there was noclear impact of the HLA-Bw4 ligand on the expression ofCD226 or KIR3DL1. The expression of CD226 or iKIR receptorson triple-positive KIR2DL2/S2/L3þ3DL1þ NKcs did not changesignificantly in the presence of their ligands (SupplementaryFig. S4C).

The expression of CD16, CD56, or NKG2A did not varysignificantly in the presence of specific HLA-C or Bw4 ligands inany NKc subset. As expected, NKG2A was clearly detectable onCD56bright NKcs, but was faint or negative on CD56dim NKcs(Supplementary Fig. S1).

We investigated the impact of themethionine/threonine (M/T)dimorphism at position �21 of the leader sequence of HLA-B

Table 2. Frequencies of KIR and KIR ligands in controls and cancer patients

Control Melanoma Bladder Ovarian

KIR genes (n ¼ 621) (n ¼ 80) (n ¼ 80) (n ¼ 89)KIR2DL1 606 (97.6%) 79 (98.8%) 75 (93.8%) 86 (96.6%)KIR2DL2 369 (59.4%) 46 (57.5%) 51 (63.8%) 50 (56.2%)KIR2DL3 558 (89.9%) 75 (93.8%) 66 (82.5%) 80 (89.9%)KIR2DL5 322 (51.9%) 46 (57.5%) 53 (66.3%)a 47 (55.7%)KIR3DL1 589 (94.8%) 77 (96.3%) 76 (95.0%) 83 (93.3%)KIR2DS1 248 (39.9%) 35 (43.8%) 39 (48.8%) 36 (40.4%)KIR2DS2 373 (60.1%) 46 (57.5%) 50 (62.5%) 53 (59.6%)KIR2DS3 198 (31.9%) 27 (33.8%) 33 (41.3%) 31 (35.2%)KIR2DS4 589 (94.8%) 78 (97.5%) 76 (95.0%) 83 (93.3%)KIR2DS5 181 (29.1%) 33 (41.3%)b 30 (37.5%) 21 (23.9%)KIR3DS1 256 (41.2%) 36 (45.0%) 39 (48.8%) 38 (43.2%)

KIR ligandsC1 528 (85.0%) 64 (80.0%) 59 (73.8%)c 78 (87.6%)C2 407 (65.5%) 52 (65.0%) 55 (68.8%) 50 (56.2%)Bw4 487 (78.4%) 62 (77.5%) 66 (82.5%) 68 (76.4%)A3/A11 199 (32.0%) 21 (26.2%) 28 (35.0%) 28 (31.8%)

Licensing iKIR/ligand interactionsKIR2DL1/C2 398 (64.1%) 51 (63.8%) 50 (62.5%) 50 (56.2%)KIR2DL2/C1 308 (49.6%) 33 (41.2%) 37 (46.3%) 45 (50.6%)KIR2DL3/C1 475 (76.5%) 61 (76.2%) 48 (60.0%)d 66 (77.5%)KIR3DL1/Bw4 464 (74.7%) 60 (75.0%) 62 (77.5%) 63 (70.8%)

Fisher exact test:aControl vs. bladder, P ¼ 0.01, Pc < 0.05, OR ¼ 1.82.bControl vs. melanoma, P ¼ 0.014, Pc > 0.05, OR ¼ 1.78.cControl vs. bladder, P < 0.01, Pc < 0.05, OR ¼ 0.49.dControl vs. bladder, P < 0.01, Pc < 0.05, OR ¼ 0.46.

Guillam�on et al.





allotypes on the expression of CD226 and NKG2A (Fig. 1B). The21M allotype (ligand for NKG2A when presented by HLA-E)positivelymodulated the expression of CD226 in bothCD56bright

(P < 0.01) and CD56dim KIR-negative (P < 0.05) NKcs whencompared with 21T homozygous patients. However, the expres-sion of the inhibitory NKG2A receptor on NKcs did not change inthe presence of 21M HLA-B allotypes.

Solid tumors downmodulate CD226/iKIR receptor ratios set onNKcs by iKIR/HLA-I licensing interactions.

We evaluated the influence of licensing interactions on theexpression of CD16, CD56, CD226, iKIR, andNKG2a duringNKcdifferentiation. We also considered the impact that sex or age ofpatients, tumor burden or the in vitro expansion of the NKcs mayhave on the stability of the CD226/iKIR receptor ratios set on themembrane of NKcs by iKIR/HLA-I licensing interactions.

During thedifferentiationofNKcs fromCD56bright toCD56dim,they downmodulate the expression of CD56 and NKG2A, whileupregulating the expression of CD16 and KIR receptors (Fig. 2A).Maximum CD226 expression in CD56bright NKcs (12,054 �264 MFI) remained high in CD56dim NKcs only in the presenceof iKIR/HLA-l licensing interactions (KIR3DL1/Bw4þ vs.KIR3DL1/Bw4�, CD226-MFI: 11,655 � 250 vs. 8,500 � 450,P < 0.001). The expression of CD226 was at its lowest in CD56dim

KIR-negative NKcs (7,355 � 157 MFI). In contrast, licensinginteractions significantly decreased the expression of the relevantiKIRs (KIR3DL1/Bw4þ vs. KIR3DL1/Bw4�, KIR3DL1-MFI: 14,000

� 525 vs. 17,300 � 1,200, P < 0.001), but did not influence theexpression of CD16, CD56, or NKG2A in CD56dim NKcs.

Licensing interactions appear to increase the ratios of CD226/iKIR receptors on NKc membranes. Sex or age did not influencethe ratios of CD226/iKIR receptors (Supplementary Fig. S5, left).However, terminally ill patients with the highest tumorburden, compared with patients who survived (Fig. 2B), hadsignificantly decreased CD226/KIR2DL1 (22.0% decrease, P <0.05), CD226/KIR2DL2/S2 (65.9% decrease, P < 0.01), CD226/KIR2DL3 (27.7% decrease, P < 0.05), and CD226/KIR3DL1(52.9% decrease, P < 0.05) ratios when cognate HLA-I ligandswere present, whereas the CD226/iKIR receptor ratios remainedalmost unchanged on uneducated NK cells. CD226/KIR2DL1significantly decreased (P < 0.05) in patients with the worstoutcome in all three types of cancer, whereas CD226/KIR2DL2/S2 (P < 0.05) and CD226/KIR3DL1 (P < 0.05) ratiosonly did in patients with ovarian and bladder cancers (Supple-mentary Fig. S5, right).

The stability of the ratios of the CD226/iKIR receptors duringin vitro expansion is essential for NKc-based tumor immuno-therapy. In view of the results obtained from the analysis ofCD226/iKIR receptor ratios in cancer patients, we evaluated theCD226/KIR2DL1 receptor ratios on NKcs from HLA-C1C1 andHLA-C2C2 healthy volunteers after 11 days of in vitro expansion(Fig. 2C). As expected, our results showed that the CD226/KIR2DL1 ratio was higher on NKcs from the HLA-C2C2 than theHLA-C1C1 healthy donors (14.38 vs. 2.94), a difference that was

Figure 1.

Licensing interactions inversely modulate CD226 (DNAM-1) and iKIR receptors on NK cells. Analysis of the expression of CD226 and iKIR receptors included249 patients (melanoma, n ¼ 80; bladder, n ¼ 80; or ovarian, n ¼ 89 tumors) and 42 healthy controls. A, MFI expression of CD226 and the correspondingKIR receptor is shown for each single KIRþ NKc subset in relation to the doses of HLA-C1, C2, or Bw4 ligands. � , P < 0.05; �� , P < 0.001 (ANOVA andDMS post hoc tests: No-ligand vs. 1 or 2 doses of the correspondent ligand). B, MFI expression of CD226 and NKG2A on CD56bright and CD56dim KIR-negativeNKc subsets in relation to the methionine/threonine (M/T) dimorphism at position �21 of the leader sequence of HLA-B allotypes. � , P < 0.05; �� , P < 0.01(ANOVA and DMS post hoc tests: TT vs. MT þ MM).






maintainedduring the in vitro expansion. Additionally,NKcs fromHLA-C2C2 donors showed a higher increase in CD226/KIR2DL1ratio than HLA-C1C1 donors after in vitro stimulation (36.06 vs.7.05 on day 2), which was likely due to a greater upregulationof the activating CD226 molecule (CD226-MFI: 56,038 vs.21,863; Fig. 2C).

Survival of solid cancer patients is related to expression ofCD226 and CD226/iKIR receptor ratios on NKcs

Finally, we explored the impact of CD226 expression on PFSand OS in all the patients with solid cancer. The optimal cutofffor CD226 expression on total NKcs was estimated by ROCanalysis and set at an MFI of 7,385 (area under the curve ¼0.67, sensitivity ¼ 87.0%, and specificity ¼ 65.0%). Solid cancerpatients with CD226 MFI < 7,385 showed significantly shorterPFS (38.1 vs. 67.0 months, P < 0.002) and OS (56.3 vs. 99.6months, P < 0.00001; Fig. 3A).

We also analyzed the impact on OS due to the ratios ofexpression of CD226/KIR2DL1 (cutoff ¼ 8.0, AUC ¼ 0.65sensitivity ¼ 57.0%, specificity ¼ 81.0%), CD226/KIR2DL2(cutoff ¼ 0.32, AUC ¼ 0.76 sensitivity ¼ 69.0%, specificity ¼75.0%), and CD226/KIR3DL1 (cutoff ¼ 0.7, AUC ¼ 0.67sensitivity ¼ 55.0%, specificity ¼ 73.0%) receptors and theirrelevance to the expression of CD226 on total NKcs (Fig. 3B).Patients showing high expression of CD226 (MFI > 7,385) andhigh CD226/iKIR receptor ratios had the highest 10-yearOS (93%, 90%, and 87% for CD226/KIR2DL1, CD226/KIR2DL2/S2, and CD226/KIR3DL1, respectively). However,patients showing high expression of CD226 but lowCD226/iKIR receptor ratios showed significantly lower 10-yearOS for CD226/KIR2DL1 (57%, P < 0.05), CD226/KIR2DL2/S2(53%, P < 0.05), and CD226/KIR3DL1 (52%, P < 0.05) whencompared with patients showing both high CD226 and highiKIR/CD226 receptor ratios. Patients with low expression of

Figure 2.

Solid tumors downmodulate CD226/iKIR receptors ratio set on the NKc membrane by iKIR/HLA-I licensing interactions after NKc differentiation.A, Expression of CD16, CD56, NKG2A, CD226, and KIR3DL1 receptors on CD56bright (CD16–/þCD56þþ) NKcs and in three subsets of CD56dim (CD16þCD56þ)NKcs: KIR3DL1þBw4þ (licensed), KIR3DL1þBw4– (unlicensed), and KIR-negative (KIR2DL1/S1–, KIR2DL2/S2/L3–, and KIR3DL1–). Data represent mean valuesfrom 249 cancer patients and 42 healthy controls. �� , P < 0.001 (Mann–Whitney test: CD226 and KIR3DL1 expression on KIR3DL1þBw4þ licensed NKcs vs.KIR3DL1þBw4– unlicensed NKcs). B, CD226/KIR2DL1, CD226/KIR2DL2/S2, CD226/KIR2DL3, and CD226/KIR3DL1 receptor ratios in relation to the dose oftheir respective HLA-C1, -C2, or -Bw4 ligands for total, alive, or dead cancer patients. P < 0.05 (ANOVA and DSM post hoc tests: alive vs. dead patients inthe presence of single or double dose of their respective HLA-I ligands).C,CD226/KIR2DL1 receptor ratio (left), CD226 (center), and KIR2DL1 (right)MFI on KIR2DL1þ

NKcs from HLA-C2C2 and HLA-C1C1 donors during NKc in vitro expansion (11 days). Representative results from one of three independent experiments.

Guillam�on et al.





CD226 on total NKcs (MFI < 7,385) and high CD226/iKIRreceptor ratios had a 10-year OS of 40% for CD226/KIR2DL1,65% for CD226/KIR2DL2, and 60% for CD226/KIR3DL1.Patients who had both low expression of CD226 and lowCD226/iKIR receptor ratios presented the lowest 10-year OS(32%, 22%, and 20% for CD226/KIR2DL1, CD226/KIR2DL2/S2, and CD226/KIR3DL1, respectively), and it was significantlylower in the case of CD226/KIR2DL2/S2 (P < 0.05). TheCD226/KIR2DL3 receptor ratio (cutoff ¼ 2.89, AUC ¼ 0.60sensitivity ¼ 65.0%, specificity ¼ 60.0%) did not significantlyaffect PFS or OS.

These results highlight the importance of CD226 expression ontotal NKcs and the effect it may have on the antitumor responseagainst solid cancers and suggest that excessive inhibitory signalsfrom iKIRs may attenuate the activating signals triggered throughCD226 and hamper NKc immune surveillance.

We analyzed patient outcome in the three types of solid cancerindependently. PFS (Fig. 3C) was significantly shorter in mela-noma and ovarian cancer patients with CD226/KIR2DL1 ratio <8.0 (melanoma: 52.9 vs. 92.1 months, P < 0.05; ovarian: 25.6 vs.61.0 months, P < 0.02) and CD226/KIR3DL1 ratio < 0.7(melanoma 46.4 vs. 90.8 months, P < 0.01; ovarian: 22.9% vs.

Figure 3.

Survival in patients with solid cancer is related to the CD226 expression on total NK cells as well as to the CD226/iKIR receptor ratio on NK cell subsets. A,PFS and OS of total solid cancer patients (n ¼ 249) with CD226 MFI on total NK cells below (black circles) or above (open circles) the optimal cutoff ¼ 7,358(ROC analysis). B,OS of total patients with CD226-MFI > 7,385 (left) or < 7,385 (right) segregated by high or low CD226/iKIR MFI-ratios on KIR2DL1þ (cutoff¼ 8.0;black bars), KIR2DL2/S2þ (cutoff ¼ 0.32; gray bars) and KIR3DL1þ (cutoff ¼ 0.7; checkered bars). � , P < 0.05 (Kaplan–Meier and log-rank, comparing OSfrom high and low CD226/iKIR MFI ratios within CD226-MFI > 7,385 or < 7,385). C, PFS for melanoma (n ¼ 80), bladder (n ¼ 80), or ovarian (n ¼ 89)patients with CD226/KIR2DL1, CD226/KIR2DL2/S2, and CD226/KIR3DL1 MFI ratios below (black circles) or above (open circles) their respective optimal cutoffs.D, OS of patients with muscle invasive bladder cancer (n ¼ 36), either with CD226/KIR2DL1, CD226/2DL2/S2, or CD226/3DL1 MFI ratios below (black circles)and above (open circles) their respective optimal cutoffs. Patients with superficial tumors showed 100% survival any case. All P values for Kaplan–Meieranalysis were estimated by log-rank.






71.0 months, P < 0.01). CD226/KIR2DL2 ratios did not signif-icantly affect the PFS of melanoma and ovarian cancer patients.The PFS of bladder cancer patients showed significant differenceswhen superficial (CIS, Ta or T1, grade III) and muscle-invasive(>T2) bladder cancers were analyzed separately. Patients withsuperficial cancer (n ¼ 44) showed 100% 10-year OS, whereaspatients with invasive carcinomas (n ¼ 36) showed signi-ficantly reduced OS for CD226/KIR2DL1 ratio < 8.0 (25.6 vs.43.8 months, P ¼ 0.02), CD226/KIR2DL2 ratio < 0.32 (41.9 vs.106.9 months, P ¼ 0.05), and CD226/KIR3DL1 ratio < 0.7(33.2 vs. 95.3 months, P > 0.05) than patients with higher ratios(Fig. 3D). Solid cancer patients with CD226/KIR2DS1 ratioshigher or lower than the optimal cutoff ¼ 0.91 (AUC ¼ 0.66sensitivity¼ 57.0%, specificity¼ 70.0%) did not show significantdifferences in PFS and OS (Supplementary Fig. S6).

The predictive value of CD226-MFI<7,385 was maintained inbladder cancer patients presented with the protective KIR2DL3/HLA-C1 genotype or the susceptibility KIR2DL5 genotype. Thus,bladder cancer patients with CD226-MFI<7,385 showed shorterOS than patients showing higher values: KIR2DL3/C1 (43.0 vs.98.0 months, P < 0.0001) and KIR2DL5þ (37.0 vs. 98.0 months,P < 0.0001; Supplementary Fig. S7).

Cox regression analysis confirmed that the type of tumor(P < 0.0001; OR ¼ 2.031), the tumor staging of each type ofcancer (P < 0.001; OR ¼ 4.911) and the expression of CD226 ontotal NKcs (P < 0.0001;OR¼ 0.2411)were independent variablesthat significantly affected the OS of patients with solid cancer(Table 3).

We did not perform a longitudinal study to evaluate variationsof CD226/iKIR receptor ratios during the follow-up to assigncausality with patients prognosis; however, we compared theCD226/iKIR receptor ratio in alive patients immediately afterdiagnosis (<1month), before the first year (<12months), or afterthe first year (>12months) post-diagnosis, with ratios observed inhealthy controls and in patients who died during the follow-up(Supplementary Fig. S8). No significant differences were foundbetween healthy controls and alive patients at different timespost-diagnosis, but the CD226/KIR2DL1, CD226/KIR3DL1(P < 0.05), and CD226/KIR2DL2S2 (P < 0.001) ratios weresignificantly lower in samples fromdead patients when comparedwith alive patients.

DiscussionA prospective follow-up of a large group of patients with

different types of solid cancers (melanoma, bladder, and ovary)has revealed that, independent of sex, age or even tumor staging,the expression of CD226 (DNAM-1) and ratios of CD226/iKIR

receptors on themembranes ofNKcs are predictive biomarkers forPFS and OS of patients with solid cancers.

Our results have established that NKc education can differen-tially modulate the expression of activating and inhibitory recep-tors on the membranes of NKcs via iKIR/HLA-I or NKG2A/HLA-Elicensing interactions. The presence of cognate ligands induces theoverexpression of the activating CD226 receptor and inhibits thatof specific iKIRs on themembranes of eachNKc subset. Consistentwith the "rheostat model," such modulation of the expressedreceptors on the membranes of NKcs is proportional to (i) thenumber of different iKIR expressed by each NKc, (ii) the numberof ligands present in the genome, and (iii) the affinity of eachiKIR/HLA-I interaction (36). Importantly, CD226/iKIR receptorratios set on the membranes of NKcs by licensing interactionsseem to be crucial for immune surveillance against these types oftumors, with KIR2DL1þ and KIR3DL1þ NKc subsets playing keyroles in melanoma and ovarian cancers. In the case of immunesurveillance for the most aggressive forms of bladder cancers, therole of KIR2DL2 is particularly important and is probably relatedto the lower genetic frequency of the specificHLA-C1 ligand in thistype of tumor.

Our data showed how solid tumors can interfere with therearrangement of CD226 and iKIR receptors on licensed NKcsto escape immune surveillance, whereas the CD226/iKIR receptorratios remain unaltered in "uneducated" NKcs, thus emphasizingthe importance of NKc education in cancer immuneresponse(1–5). Of particular note, we could not establish a clear role foractivating KIR receptors either in the genetic susceptibility or forthe survival of patients, and this is a subject for further studies.

CD226 is a coactivating receptor that acts synergistically withCD96, TIGIT, and CRTAM and is involved in the regulation ofNKc adhesion, immune synapse formation, cross-talk withdendritic and T cells, lymphocyte signaling, cytokine secretion,and effective cytotoxicity (40, 41). Along with KLRG1 (42),CD226 plays an important role in NKc education and differ-entiation. The expression of CD226 on NKc positively corre-lates with the quantity and quality of iKIR/HLA-I licensinginteractions and the magnitude of NKc responses by promotingchanges in the conformational state of LFA-1 that colocalizeswith CD226 at the immune synapse to stabilize NK–target cellconjugates (23, 43). However, results from experimental mod-els seem to indicate that the expression of CD226 is associatedwith, but not required for, NKc education (44). Even if CD226does not directly intervene in the education of NKcs, and itsoverexpression is only a downstream effect of the educationalprocess, our results support the idea that CD226 expression islinked to human NKc education. Extensive evidence fromhuman pathology indicates that appropriate expression ofCD226 or other activating receptors, such as NKG2D, is crucialfor an effective NKc response and proper immune surveillancein different types of cancers, such as AML, myeloma, B-celllymphocytic leukemia, melanoma, pancreas, or ovarian carci-noma (25–30). However, if an appropriate expression ofCD226 is important in effective cancer immune surveillance,the expression of inhibitory receptors that imprint the educa-tion of NKcs is also critical. The survival of solid cancer patientsin our cohort was directly proportional to the expression ofCD226 on NKcs and was also affected by the ratios of CD226/iKIR receptors.

Our results have shed light on (i) the fine tuning of theexpression of activating and inhibitory receptors induced

Table 3. Cox analysis of OS in solid cancer patients

Covariables OR 95% CI P

Sex 0.879 0.358–2.156 0.778Age 1.018 0.987–1.051 0.258Tumor type (Mel, Bla, and Ova) 2.031 1.387–2.973 0.000Tumor staginga 4.919 1.941–12.463 0.001NK cell CD226 (MFI) > 7,385 0.241 0.120–0.486 0.000

OR, odd ratio; 95% CI, 95% confidence interval; Mel: melanoma; Bla: bladder;Ova: ovarian.aEstimated as tumor staging for melanoma (Breslow depth > or <1 mm), bladdercancer (superficial vs. invasive), and ovarian carcinoma (staging I and II vs. IIIand IV).

Guillam�on et al.





by licensing interactions during NKc differentiation, (ii) theimportance of the set stoichiometric ratios of CD226/iKIR recep-tors for optimized antitumor responses, and (iii) how thesechanges can influence cancer immune surveillance. Althoughprofound changes in the expression of CD16, CD56, and NKG2Areceptors accompanied NKc differentiation from CD56bright toCD56dim, licensing interactions did not seem to modify theexpression patterns of these molecules. Consistent with resultsdescribed in mice (44), our data showed that the maximumexpression of CD226 observed on CD56bright NKcs was signifi-cantly lost on CD56dim NKcs if licensing interactions did notprevent its downregulation. During differentiation, CD56dim

NKcs gain CD16 and cytotoxic granules, which makes themhighly cytolytic; to regulate this potentially hazardous function,iKIRs are gained concomitantly. We have demonstrated thatlicensing interactions can restrain the expression of iKIRs to setthe ratios of CD226/iKIR receptors on NKc membranes so thatactivating signals prevail over inhibitory ones.

The ratios of CD226/iKIR receptors on the membranes of allNKc subsets remained stable and were not influenced by sex,age, or their in vitro expansion in the same conditions used forobtaining NKcs for clinical trials for cancer immunotherapies.However, tumors were able to reset the upmodulated ratios ofCD226/iKIR receptors induced by licensing interactions toescape the cytolytic activity of NKcs. In our cohort, this wasclearly demonstrated in terminally ill patients with high tumorburdens and has previously been described in patientswith ovarian cancer and AML (45, 46). Furthermore, our dataindicate that the ratios of CD226/iKIR receptors set on NKcs bylicensing interactions are a dynamic event that depends on theinput signals that NKcs receive in each scenario, as previouslysuggested (21). An understanding of the biochemical pathwaysthat differentially regulate the expression of activating andinhibitory receptors, i.e., CD226 and iKIRs, can be applied toenhance NKc immunotherapeutic potential, by favoring theexpression of activating receptors and/or hindering that ofinhibitory ones during in vitro expansion.

Themain limitation of our study is the cross-sectional nature ofthe evaluation of the CD226/iKIR receptor ratios that hinderscausality association with cancer prognosis. Although a longitu-dinal study should be performed to rule out other factors intrinsicto individual patients or cancer itself, analyses of CD226/iKIRreceptor ratios in alive patients at different times post-diagnosiscompared with ratios observed in healthy controls and in patientswho died during the follow-up suggest that tumor burden is amajor factor involved in the downmodulation of CD226/iKIRreceptor ratios.

Our results may also have clinical applications in the selec-tion of donors for haploidentical NKc therapies. Current mod-els for this type of NKc therapy are based on KIR and KIR-ligandgenetics, the type of tumor, the conditioning regimen, and thecharacteristics of grafts. However, controversies still exist as towhether NKcs should be educated in the donor and the patientshould lack specific KIR ligands (2, 47–50), which highlightsthe need for additional information that is not easy to obtain;unfortunately, functional studies are not easily available. Withthis in mind, the analytical method described here reveals a wayto select donors with the highest CD226/iKIR receptor ratios onpotentially alloreactive NK cell subsets.

In summary, the results described in this article provide relevantinformation on the modulation of molecules that regulate the

function of NKc during their differentiation, and how licensinginteractions finely tune them to induce CD226/iKIR receptorratios with a predominance of activating receptors over inhibitoryones. Transformed cells can downmodulate these licensing-driven receptor rearrangements as a specific mechanism to escapeNKc immune surveillance, explaining why both CD226 expres-sion and CD226/iKIR receptor ratios on peripheral bloodNKc arepredictive biomarkers for the survival of patients with differenttypes of solid cancers and genetic backgrounds. These resultsmight also have important implications in the selection of donorsand in the design of NKc-based therapies. Biochemical pathwayscould bemanipulated during in vitro expansion ofNKcs to achievemore favorable CD226/iKIR receptor ratios on the membranes ofNKcs to enhance their antitumor potential.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: M.V. Martínez-S�anchez, L. Gimeno, G. Server-Pastor,J. Martínez-Escribano, D. Abell�an, A. MinguelaDevelopment of methodology: C.F. Guillam�on, L. Gimeno, G. Server-Pastor,D. Abell�an, J.A. Campillo, I. Legaz, M.R. Moya-QuilesAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): C.F. Guillam�on, L. Gimeno, J. Martínez-García,G. Server-Pastor, J. Martínez-Escribano, A. Torroba, B. Ferri, D. Abell�an,A. MinguelaAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): L. Gimeno, A. Mrowiec, J. Martínez-García,G. Server-Pastor, D. Abell�an, M. Muro, A. MinguelaWriting, review, and/or revision of the manuscript: L. Gimeno, A. Mrowiec,J.Martínez-García, J.Martínez-Escribano,D.Abell�an, J.A. Campillo,M.R. L�opez-�Alvarez, M.R. Moya-Quiles, M. Muro, A. MinguelaAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): L. Gimeno, J. Martínez-García, J. Martínez-Escribano, D. Abell�an, A. MinguelaStudy supervision: L. Gimeno, J. Martínez-García, A. Minguela

AcknowledgmentsThis work was funded by MINECO—Instituto de Salud Carlos III

(PI1302297). C.F. Guillam�on was funded by the Fundaci�on para el estudioy el desarrollo de la inmunogen�etica en Murcia (FEYDIM). M.V. Martínez-S�anchez was funded by the Asociaci�on Pablo Ugarte (APU). We thankDr. Valentine P. Iyemere of Bionodum Limited for critical review of themanuscript.

The authors are also grateful to B. Rodriguez Martin-Gil for blood samplecollection and patient management; J.M. Bolarin for statistical advice; as wellas the oncologists (Drs. J.L. Alonso, P. Sanchez-Henarejos, A. Soto, A. Puertes,M.D. Gimenez, and M. Guirao), urologists (Drs. J.F. Escudero, P.A. L�opez-Gonz�alez, and P. L�opez-Cubillana), dermatologists (Drs. J. Frías, P. S�anchez-Pedre~no, R. Corbal�an, M. Lova, A.M. Victoria, J. Hern�andez-Gil, C. Brufau, A.Ramírez, M.E. Gimenez, E. Cutillas, C. Pereda, R. Rojo, P. Mercader, J.M.R�odenas, A. Pe~na, and J. Pardo), gynecologists (Drs. F. Barcel�o and B. Gomez-Monreal), and surgeons (Dr. P. Cascales-Campos, J. Gil, and E. Gil) at theClinic University Hospital Virgen de la Arrixaca (Murcia, Spain), and theoncologists (Dr. P. Cerezuela and E. Braun) at the University HospitalSanta Lucía (Cartagena, Spain) for their kind collaboration in enrolling andattending the patients.

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby markedadvertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

ReceivedMarch 6, 2018; revised June 15, 2018; accepted September 17, 2018;published first September 21, 2018.






References1. Kim S, Poursine-Laurent J, Truscott SM, Lybarger L, Song YJ, Yang L, et al.

Licensing of natural killer cells by host major histocompatibility complexclass I molecules. Nature 2005;436:709–13.

2. Anfossi N, Andr�e P, Guia S, Falk CS, Roetynck S, Stewart CA, et al. HumanNK cell education by inhibitory receptors for MHC class I. Immunity2006;25:331–42.

3. Thomas LM. Current perspectives on natural killer cell education andtolerance: emerging roles for inhibitory receptors. ImmunoTargets Ther2015;4:45–53.

4. Boudreau JE, Liu XR, Zhao Z, Zhang A, Shultz LD, Greiner DL, et al. Cell-extrinsic MHC class I molecule engagement augments human NK celleducation programmed by cell-intrinsic MHC class I. Immunity 2016;45:280–91.

5. Horowitz A, Djaoud Z, Nemat-Gorgani N, Blokhuis J, Hilton HG, B�eziat V,et al. Class I HLA haplotypes form two schools that educate NK cells indifferent ways. Sci Immunol 2016;1:1–35.

6. He Y, Tian Z. NK cell education via nonclassical MHC and non-MHCligands. Cell Mol Immunol 2017;14:321–30.

7. B�eziat V, Hilton HG, Norman PJ, Traherne JA. Deciphering the killer-cellimmunoglobulin-like receptor system at super-resolution for natural killerand T-cell biology. Immunology 2017;150:248–64.

8. Velardi A, Ruggeri L, Mancusi A, Aversa F, Christiansen FT. Natural killercell allorecognition of missing self in allogeneic hematopoietic trans-plantation: a tool for immunotherapy of leukemia. Curr Opin Immunol2009;21:525–30.

9. Ruggeri L, Capanni M, Urbani E, Perruccio K, ShlomchikWD, Tosti A, et al.Effectiveness of donor natural killer cell aloreactivity in mismatchedhematopoietic transplants. Science 2002;295:2097–100.

10. K€arre K. Natural killer cell recognition of missing self. Nat Immunol2008;9:477–80.

11. Tanimine N, Tanaka Y, Kobayashi T, Tashiro H, Miki D, Imamura M, et al.Quantitative effect of natural killer-cell licensing on hepatocellular carci-noma recurrence after curative hepatectomy. Cancer Immunol Res 2014;2:1142–7.

12. Martínez-S�anchez MV, Periago A, Legaz I, Gimeno L, Mrowiec A,Montes-Barqueros NR, et al. Overexpression of KIR inhibitory ligands(HLA-I) determines that immunosurveillance of myeloma depends ondiverse and strong NK cell licensing. Oncoimmunology 2015;5:e1093721.

13. Mandelboim O, Reyburn HT, Val�es-G�omez M, Pazmany L, Colonna M,Borsellino G, et al. Protection from lysis by natural killer cells of group 1and 2 specificity is mediated by residue 80 in human histocompatibilityleukocyte antigen C alleles and also occurs with empty major histocom-patibility complex molecules. J Exp Med 1996;184:913–22.

14. Foley BA, De Santis D, Van Beelen E, Lathbury LJ, Christiansen FT, Witt CS.The reactivity of Bw4þHLA-B andHLA-A alleles with kir3dll: implicationsfor patient and donor suitability for haploidentical stem cell transplanta-tions. Blood 2008;112:435–43.

15. Chewning JH, Gudme CN, Hsu KC, Selvakumar A, Dupont B. KIR2DS1-positive NK cells mediate alloresponse against the C2 HLA-KIR ligandgroup in vitro. J Immunol 2007;179:854–68.

16. Fauriat C, Ivarsson MA, Ljunggren H, Malmberg K, Michae J. Education ofhuman natural killer cells by activating killer cell immunoglobulin-likereceptors. Blood 2010;115:1166–74.

17. Katz G, Markel G, Mizrahi S, Arnon TI, Mandelboim O. Recognition ofHLA-Cw4 but not HLA-Cw6 by the NK cell receptor killer cell Ig-likereceptor two-domain short tail number 4. J Immunol 2001;166:7260–7.

18. Graef T, Moesta AK, Norman PJ, Abi-Rached L, Vago L, Older Aguilar AM,et al. KIR2DS4 is a product of gene conversion with KIR3DL2 thatintroduced specificity for HLA-A� 11 while diminishing avidity for HLA-C. J Exp Med 2009;206:2557–72.

19. O'Connor GM, Yamada E, Rampersaud A, Thomas R, Carrington M,McVicar DW. Analysis of binding of KIR3DS1�014 toHLA suggests distinctevolutionary history of KIR3DS1. J Immunol 2011;187:2162–71.

20. Nowak J, Ko�sci�nska K, Mika-Witkowska R, Rogatko-Koro�s M, Mizia S,Jaskuøa E, et al. Role of donor activating KIR-HLA ligand-mediated NK celleducation status in control of malignancy in hematopoietic cell transplantrecipients. Biol Blood Marrow Transplant 2015;21:829–39.

21. Brodin P, H€oglund P. Beyond licensing and disarming: a quantitative viewon NK-cell education. Eur J Immunol 2008;38:2934–7.

22. Goodridge JP, €Onfelt B, Malmberg KJ. Newtonian cell interactions shapenatural killer cell education. Immunol Rev 2015;267:197–213.

23. Enqvist M, Ask EH, Forslund E, Carlsten M, Abrahamsen G, B�eziat V, et al.Coordinated expression of DNAM-1 and LFA-1 in educated NK cells.J Immunol 2015;194:4518–27.

24. Yawata M, Yawata N, Draghi M, Little A-M, Partheniou F, Parham P.Roles for HLA and KIR polymorphisms in natural killer cell repertoireselection and modulation of effector function. J Exp Med 2006;203:633–45.

25. Khaznadar Z, Boissel N, Agaugu�e S, Henry G, Cheok M, Vignon M, et al.Defective NK cells in acute myeloid leukemia patients at diagnosisare associated with blast transcriptional signatures of immune evasion.J Immunol 2015;195:2580–90.

26. CarlstenM,Bj€orkstr€omNK,NorellH, BrycesonY, VanHall T, BaumannBC,et al. DNAX accessory molecule-1 mediated recognition of freshly isolatedovarian carcinoma by resting natural killer cells. Cancer Res 2007;67:1317–25.

27. Guillerey C, De Andrade LF, Vuckovic S, Miles K, Ngiow SF, Yong MCR,et al. Immunosurveillance and therapy of multiple myeloma are CD226dependent. J Clin Invest 2015;125:2077–89.

28. Labani-Motlagh A, Israelsson P, Ottander U, Lundin E, Nagaev I, NagaevaO, et al. Differential expression of ligands for NKG2D and DNAM-1receptors by epithelial ovarian cancer-derived exosomes and its influenceon NK cell cytotoxicity. Tumor Biol 2016;37:5455–66.

29. Peng Y, Xi C, Zhu Y, Yin L, Wei J, Zhang J-J, et al. Altered expression ofCD226 and CD96 on natural killer cells in patients with pancreatic cancer.Oncotarget 2016;7:66586–94.

30. MacFarlane AW, Jillab M, Smith MR, Alpaugh RK, Cole ME, Litwin S, et al.NK cell dysfunction in chronic lymphocytic leukemia is associated withloss of the mature cells expressing inhibitory killer cell Ig-like receptors.Oncoimmunology 2017;6:e1330235.

31. Iguchi-Manaka A, Kai H, Yamashita Y, Shibata K, Tahara-Hanaoka S,Honda S, et al. Accelerated tumor growth in mice deficient in DNAM-1receptor. J Exp Med 2008;205:2959–64.

32. Kim JS, Shin BR, Lee HK, Lee JH, Kim KH, Choi JE, et al. Cd226�/� naturalkiller cells fail to establish stable contacts with cancer cells and showimpaired control of tumor metastasis in vivo. Oncoimmunology 2017;6:e1338994.

33. Fang F, Xiao W, Tian Z. NK cell-based immunotherapy for cancer. SeminImmunol 2017;31:37–54.

34. Legaz I, L�opez-�Alvarez MR, Campillo JA, Moya-Quiles MR, Bolarín JM, DeLa Pe~na J, et al. KIR gene mismatching and KIR/C ligands in liver trans-plantation: consequences for short-term liver allograft injury. Transplan-tation 2013;95:1037–44.

35. Moretta L, Moretta A. Killer immunoglobulin-like receptors. Curr OpinImmunol 2004;16:626–33.

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

37. FaureM, Long EO. KIR2DL4 (CD158d), anNK cell-activating receptor withinhibitory potential. J Immunol 2002;168:6208–14.

38. Uhrberg M, Valiante NM, Shum BP, Shilling HG, Lienert-Weidenbach K,Corliss B, et al. Human diversity in killer cell inhibitory receptor genes.Immunity 1997;7:753–63.

39. Leivas A, Perez-Martinez A, Blanchard MJ, Martín-Clavero E, Fern�andez L,Lahuerta JJ, et al. Novel treatment strategy with autologous activated andexpanded natural killer cells plus anti-myeloma drugs for multiple mye-loma. Oncoimmunology 2016;5:e1250051.

40. Xiong P, Sang HW, Zhu M. Critical roles of co-activation receptor DNAXaccessory molecule-1 in natural killer cell immunity. Immunology2015;146:369–78.

41. Martinet L, Ferrari De Andrade L, Guillerey C, Lee JS, Liu J, Souza-Fonseca-Guimaraes F, et al. DNAM-1 expression marks an alternative program ofNK cell maturation. Cell Rep 2015;11:85–97.

42. Corral L, Hanke T, Vance RE, Cado D, Raulet DH. NK cell expression ofthe killer cell lectin-like receptor G1 (KLRG1), the mouse homolog ofMAFA, is modulated by MHC class I molecules. Eur J Immunol2000;30:920–30.

Guillam�on et al.





43. Thomas LM, Peterson ME, Long EO. Cutting edge: NK cell licensingmodulates adhesion to target cells. J Immunol 2013;191:3981–5.

44. Wagner AK, Kadri N, Sn€all J, Brodin P, Gilfillan S, Colonna M, et al.Expression of CD226 is associated to but not required for NK cell educa-tion. Nat Commun 2017;8:15627.

45. da Silva RF, Petta CA, Derchain SF, Alici E, Guimar~aes F. Up-regulation ofDNAM-1 and NKp30, associated with improvement of NK cells activationafter long-term culture of mononuclear cells from patients with ovarianneoplasia. Hum Immunol 2014;75:777–84.

46. Sandoval-Borrego D, Moreno-Lafont MC, Vazquez-Sanchez EA, Gutierrez-Hoya A, L�opez-Santiago R, Montiel-Cervantes LA, et al. Overexpression ofCD158 and NKG2A inhibitory receptors and underexpression of NKG2Dand NKp46 activating receptors on NK cells in acute myeloid leukemia.Arch Med Res 2016;47:55–64.

47. LeungW, Iyengar R, Turner V, Lang P, Bader P, Conn P, et al. Determinantsof antileukemia effects of allogeneic NK cells. J Immunol 2004;172:644–50.

48. Rogatko-Koro�s M, Mika-Witkowska R, Bogunia-Kubik K, Wysocza�nska B,Jaskuøa E, Ko�sci�nska K, et al. Prediction of NK cell licensing level inselection of hematopoietic stem cell donor, initial results. Arch ImmunolTher Exp 2016;64:63–71.

49. Gagne K, Brizard G, Gueglio B, Milpied N, Herry P, Bonneville F, et al.Relevance of KIR gene polymorphisms in bone marrow transplantationoutcome. Hum Immunol 2002;63:271–80.

50. McQueenKL,Dorighi KM,Guethlein LA,WongR, Sanjanwala B, ParhamP.Donor-recipient combinations of group A and B KIR haplotypes and HLAclass I ligand affect the outcome of HLA-matched, sibling donor hemato-poietic cell transplantation. Hum Immunol 2007;68:309–23.






2018;6:1537-1547. Published OnlineFirst September 21, 2018.Cancer Immunol Res Concepción F. Guillamón, María V. Martínez-Sánchez, Lourdes Gimeno, et al. OutcomeReceptor Ratios as Predictive Biomarkers for Solid Tumor NK Cell Education in Tumor Immune Surveillance: DNAM-1/KIR

Updated version

10.1158/2326-6066.CIR-18-0022doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerimmunolres.aacrjournals.org/content/suppl/2018/09/21/2326-6066.CIR-18-0022.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerimmunolres.aacrjournals.org/content/6/12/1537.full#ref-list-1

This article cites 50 articles, 17 of which you can access for free at:

Citing articles

http://cancerimmunolres.aacrjournals.org/content/6/12/1537.full#related-urls

This article has been cited by 2 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerimmunolres.aacrjournals.org/content/6/12/1537To request permission to re-use all or part of this article, use this link



http://cancerimmunolres.aacrjournals.org/lookup/doi/10.1158/2326-6066.CIR-18-0022

http://cancerimmunolres.aacrjournals.org/content/suppl/2018/09/21/2326-6066.CIR-18-0022.DC1

http://cancerimmunolres.aacrjournals.org/content/6/12/1537.full#ref-list-1

http://cancerimmunolres.aacrjournals.org/content/6/12/1537.full#related-urls

http://cancerimmunolres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerimmunolres.aacrjournals.org/content/6/12/1537


nk cell education in tumor immune surveillance: dnam-1/kir ... › content › ... · maría r....

Documents