Identification, activation, and selective in vivoablation of mouse NK cells via NKp46Thierry Walzer*†‡§, Mathieu Blery¶, Julie Chaix*†‡, Nicolas Fuseri¶, Lionel Chasson*†‡, Scott H. Robbins*†‡,Sebastien Jaeger*†‡, Pascale Andre¶, Laurent Gauthier¶, Laurent Daniel�, Karine Chemin¶, Yannis Morel¶,Marc Dalod*†‡, Jean Imbert**, Michel Pierres*†‡, Alessandro Moretta††, Francois Romagne¶,and Eric Vivier*†§‡‡

*Centre d’Immunologie de Marseille-Luminy, Universite de la Mediterranee, 13288 Marseille, France; †Institut National de la Sante et de la RechercheMedicale, Unite Mixte de Recherche 631, Marseille, France; ‡Centre National de la Recherche Scientifique, Unite Mixte de Recherche 6102, 13288 Marseille,France; ¶Innate-Pharma, 13009 Marseille, France; �Assistance Publique–Hopitaux de Marseille, Hopital de la Timone, 13005 Marseille, France; **InstitutNational de la Sante et de la Recherche Medicale, Unite Mixte de Recherche 599, Centre de Recherche en Cancerologie de Marseille, 13009 Marseille,France; ††Dipartimento di Medicina Sperimentale, and Centro di Eccellenza per le Ricerche Biomediche, Universita degli Studi di Genova, 16000 Genova,Italy; and ‡‡Assistance Publique–Hopitaux de Marseille, Hopital de la Conception, 13005 Marseille, France

Edited by Christophe Benoist, Harvard Medical School, Boston, MA, and approved December 31, 2006 (received for review November 1, 2006)

Natural killer (NK) cells contribute to a variety of innate immuneresponses to viruses, tumors and allogeneic cells. However, ourunderstanding of NK cell biology is severely limited by the lack ofconsensus phenotypic definition of these cells across species, bythe lack of specific marker to visualize them in situ, and by the lackof a genetic model where NK cells may be selectively ablated.NKp46/CD335 is an Ig-like superfamily cell surface receptor in-volved in human NK cell activation. In addition to human, we showhere that NKp46 is expressed by NK cells in all mouse strainsanalyzed, as well as in three common monkey species, promptinga unifying phenotypic definition of NK cells across species based onNKp46 cell surface expression. Mouse NKp46 triggers NK celleffector function and allows the detection of NK cells in situ. NKp46expression parallels cell engagement into NK differentiation pro-grams because it is detected on all NK cells from the immatureCD122�NK1.1�DX5� stage and on a minute fraction of NK-like Tcells, but not on CD1d-restricted NKT cells. Moreover, humanNKp46 promoter drives NK cell selective expression both in vitroand in vivo. Using NKp46 promoter, we generated transgenic miceexpressing EGFP and the diphtheria toxin (DT) receptor in NK cells.DT injection in these mice leads to a complete and selective NK cellablation. This model paves a way for the in vivo characterizationand preclinical assessment of NK cell biological function.

genetic models � innate immunity

Natural killer (NK) cells are large granular lymphocytes thatbelong to the innate immune system (1). NK cells are

present in lymphoid organs as well as in nonlymphoid peripheraltissues (2). They are involved in defense mechanisms againstseveral types of microbial infections and tumors (3). They alsohave a role in shaping adaptive immune responses and in thecontrol of placental development (4, 5). Strategies are emergingto apply NK cells as therapeutic agents against a broad range ofmalignancies (6, 7). However, several NK cell features havelimited our understanding of their biological function.

First, NK cells share phenotypic properties with various T cellpopulations such as CD1d-restricted NKT cells, �� T cells, anddiscrete subsets of antigen-experienced CD122�CD8� T cells(8). NK cell identification by flow cytometry thus requires othermarkers to exclude T cells.

Second, there is no consensus phenotypic definition of NKcells across species. In humans, NK cells are defined asCD56�CD3� lymphocytes. In NK cells, CD56 corresponds tothe 140-kDa isoform of the neural cell adhesion molecule(N-CAM). Yet, N-CAM is also expressed by T cell subsets,muscle cells, and neurons, but it is not expressed by murine NKcells (9). In the rat, NK cells express the activating receptorNKR-P1A, but this molecule is also expressed by T cell subsets(10). In the mouse, the widely used PK136 antibody reacts with

NK1.1, an epitope shared by the activating receptor NKR-P1Cin C57BL/6 mice and the inhibitory receptor NKR-P1B in SJLmice (11). However, NK cells from other mouse strains do notreact with the anti-NK1.1 antibody, due to allelic divergence ofthe nkrp1b/c genes (11). In NK1.1� mouse strains, the identifi-cation of NK cells is based on the expression of the integrinsubunit CD49b that is recognized by DX5 antibodies (12),despite its expression on T cell and myeloid subsets (13, 14).These various phenotypic NK cell definitions and the lack ofspecific NK cell markers impede the comparison of data acrossspecies.

Third, there is no genetic model where NK cells can beselectively deleted (15, 16). Transgenic mice lacking NK cells butwith a normal T/NKT cell compartment have been reported (17).The cause of NK cell ablation in these mice is unknown, but islinked to the expression of the ubiquitous transcription factorATF2, raising the possibility of other defects yet to be furtherinvestigated (18). In vivo NK cell depletion has thus so far reliedon anti-asialo-GM1 or anti-NK1.1 depleting antibodies. How-ever, the expression of these markers outside of the NK cellcompartment has hampered the interpretation of results ob-tained with these protocols.

Natural cytotoxicity receptors (NCRs) include the NKp30,NKp44, and NKp46 molecules, and are Ig-like transmembraneglycoproteins (19). Their transmembrane regions contain posi-tively charged amino acids allowing association with the ITAM-bearing polypeptides CD3� and FcR� for NKp46 and NKp30 orKARAP/DAP12 for NKp44 (20). NCRs are involved in therecognition of tumor targets, but the cellular ligands recognizedby the NCRs are still elusive (20). NKp46 (product of the NCR1gene) has also been shown to recognize the hemagglutininof influenza virus and the hemagglutinin-neuraminidase ofSendai virus (21). NKP46 is conserved between human andmouse, whereas no mouse orthologue of NKP44/NCR2 has

Author contributions: T.W. and M.B. contributed equally to this work; T.W., M.B., and E.V.designed research; T.W., M.B., L.C., S.H.R., P.A., L.G., L.D., and K.C. performed research; J.C.,N.F., S.J., L.G., Y.M., M.D., J.I., M.P., A.M., and F.R. contributed new reagents/analytic tools;T.W., M.B., L.G., and E.V. analyzed data; and T.W. and E.V. wrote the paper.

Conflict of interest statement: M.B., P.A., K.C., L.G., and Y.M. are employees and share-holders of Innate-Pharma. F.R. is founder, employee, and shareholder of Innate-Pharma.E.V. and A.M. are founders and shareholders of Innate-Pharma.

This article is a PNAS direct submission.

Abbreviations: NK, natural killer; N-CAM, neural cell adhesion module; NCR, naturalcytotoxicity receptor; DT, diphtheria toxin; DTR, DT receptor.

§To whom correspondence may be addressed. E-mail: walzer@ciml.univ-mrs.fr orvivier@ciml.univ-mrs.fr.

This article contains supporting information online at www.pnas.org/cgi/content/full/0609692104/DC1.

© 2007 by The National Academy of Sciences of the USA

3384–3389 � PNAS � February 27, 2007 � vol. 104 � no. 9 www.pnas.org�cgi�doi�10.1073�pnas.0609692104

Dow

nloa

ded

by g

uest

on

June

13,

202

0

been found (22), and mouse Nkp30/Ncr3 is a pseudogene inMus musculus (23).

Previous studies have shown that NKp46 expression wasrestricted to CD56�CD3�HLA-DR� lymphocytes in humanperipheral blood mononuclear cells (24). Here, we show thatNKp46 cell surface expression defines NK cells in human, in allstrains of mice tested, and in three species of monkey, leading tothe proposal that NKp46 is a unifying marker for NK cells acrossmammalian species. Moreover, NK cells can be unambiguouslyvisualized in situ and selectively activated via NKp46. Finally, weidentified a 400-bp NKP46 promoter that drives NK cell-specificgene expression, allowing the generation of the first transgenicmouse model of conditional and selective NK cell ablation invivo.

ResultsCell Surface Expression of NKp46 Defines NK Cells Across Species.NKp46 expression was previously shown to be virtually restrictedto CD56�CD3�HLA-DR� cells in humans (24). Here, wefurther show that, unlike CD56, NKp46 was not expressed by ��T cells or CD1d-restricted V�24� T cells that represent minorfractions of the human blood CD3� population [supportinginformation (SI) Fig. 6]. Thus, NKp46 cell surface expressiondefined human peripheral blood NK cells in normal individuals.These results prompted us to test whether NKp46 can also beused as a phenotypic marker of mouse NK cells, using either

monoclonal (29A1.4) or affinity-purified polyclonal antibodiesraised against a chimeric mouse NKp46-Fc fusion protein (SIFig. 7). NKp46 expression was detected in nearly all NK1.1� cellsfrom RAG�/� mice but was severely reduced when cells wereisolated from CD3��/�FcR��/�RAG�/� mice (Fig. 1A). Thelatter lack the CD3� and FcR� adapter molecules required forcell surface expression of human NKp46 (25). These data thussuggest that both human and mouse NKp46 depend on CD3� andFcR� for cell surface expression and establish the specificity ofthe mouse NKp46 staining performed with anti-NKp46 mono-clonal and polyclonal antibodies. In C57BL/6 mice, NKp46 wasnot expressed in granulocytes, dendritic cells, B cells, T cells,CD1d-�-gal-cer tetramer� NKT cells (Fig. 1B), monocytes ormacrophages (data not shown). By contrast, NK cells, defined asNK1.1�CD3�, were characterized by a high and uniform ex-pression of NKp46 (Fig. 1B). In line with previous studies inhumans (26), the cell surface expression of NKp46 is initiated atthe immature stage of NK cell development(CD122�NK1.1�DX5�) in the bone marrow (Fig. 1C) andsubsequently remains at the same level by all NK cells isolatedfrom all organs tested (Fig. 1D and data not shown). Thus,NKp46 cell surface expression defines NK cells in C57BL/6 micein contrast to NK1.1 alloantigen that is expressed by a largepercentage of CD3� cells in various organs (Fig. 1E). Next, wemeasured NKp46 expression in NK1.1� strains of mice, whereNK cell identification is the most problematic. A bright NKp46staining was observed on DX5�CD3� cells from all strains ofmice tested (BALB/C, SJL, CBA/CA, DBA/2, B6.129, NOD,NZW; Fig. 1F and data not shown). We also noticed the presenceof a substantial population of NKp46�DX5�CD3� cells (datanot shown). These cells were also present in C57BL/6 mice wherethey expressed NK1.1. They correspond to immature NK cells,enriched in young mice, especially in the liver when theyrepresented up to 50% of NK cells (Fig. 1G) (27). Thus, NKp46cell surface expression defines NK cells in all mouse strains incontrast to DX5/CD49b that is expressed by a large percentageof CD3� cells in various organs (Fig. 1E) and by only a subsetof NK cells (Fig. 1G). The cell surface expression of DX5 has alsobeen reported on lung basophils, as well as on ill-defined subsetsof CD3�NK1.1� splenocytes (28). Importantly, NKp46 is notexpressed by bone marrow basophils, or by the small fraction ofsplenic NK1.1�CD3�DX5low cells that are likely of myeloidorigin (data not shown).

Although these results indicate that NKp46 represents themost specific surface marker for mouse NK cells, �2% ofNKp46� cells also express low levels of surface CD3� (Fig. 1E).To determine whether antigenic exposure induced NKp46 ex-pression on T cells, we infected C57BL/6 mice with mousecytomegalovirus, which is associated with a robust CD8� T cellresponse. At the peak of the response, 40–50% of CD8� T cellsdisplayed an activated CD43high phenotype and acquiredNKG2A/C/E cell surface expression (29) (SI Fig. 8). However,all T cells in virus-infected mice remained NKp46� (SI Fig. 8).These results indicate that acute T cell stimulation does not leadto the cell surface expression of NKp46. Rather, NKp46�CD3�

cells could originate from chronically activated T cells repro-grammed into NK-like cells, as recently proposed for intraepi-thelial T cells in celiac patients (30). Consistent with thishypothesis, these NKp46�CD3� cells harbor a very peculiar cellsurface phenotype, as they are NK1.1�Ly49�CD1dtetramer�CD4�CD8�, and 75% of them are TCR���. TheseNKp46� �� T cells represent a minute fraction of �� T cells, as95–99% of splenic and nearly 100% of gut intraepithelial �� Tcells are NKp46� (data not shown).

NK Cells Can Be Visualized in Situ and Activated via NKp46. Visual-ization of NK cells in situ has been hampered by the lack ofreactivity and specificity of NK cell markers. Therefore, we

1.4

NK cells

96

Granulo. B cells

0.3 0.40.7

96 12 NKp46

control

0.4

RAG-/- CD3ζ-/-FcRγ-/-

RAG-/-

DC

T cells NKT cells

A

B

95 99 98 99

LN Lung Liver BloodD

98 97 99

BALB/C SJL CBA/CAF

Go

at

IgG

0 0

099

48 49

03N

Kp

46

9 88

11

G

DX5Rat IgM

spleen liver

E

0

20

40

60

80

% o

f C

D3+

w

ithin

gat

e NKp46+

NK1.1+

DX5+

4 99 99

NKp mNKiNKC

BM LN Liver PBMC Lungspleen

Fig. 1. Expression of the activating NKp46 cell surface receptor on mouse NKcells. (A) Flow cytometric measurement of NKp46 expression in gated NK1.1�

spleen cells from RAG�/� and RAG�/�CD3��/�FcR��/� mice (C57BL/6 back-ground). NKp46 expression (open histogram, thick line) or isotype control(gray histogram, thin line) is shown in the indicated subsets in A and insubsequent panels. (B) Indicated cell types were identified from C57BL/6splenocytes; Granulo., granulocytes; DC, dendritic cells. (C) Flow cytometricmeasurement of NKp46 expression on C57BL/6 NK precursors (NKp), immatureand mature NK cells, identified as described (47) as CD122�lin�NK1.1�DX5�,CD122�NK1.1�DX5�, and CD122�NK1.1�DX5�, respectively. (D) Flow cyto-metric measurement of NKp46 expression in gated NK1.1� CD3� cells fromC57BL/6 mouse lymph nodes (inguinal), lung, liver, and peripheral blood. (E)Comparison of the percentage of CD3� cells within the NKp46� (black bars),NK1.1� (gray bars), and DX5� subsets in various organs of C57BL/6 mice. (F)Flow cytometric measurement of NKp46 expression in gated splenicDX5�CD3� cells from the indicated strains. (G) Flow cytometric measurementof NKp46/DX5 expression in gated NK1.1�CD3� cells from the spleen or theliver (27), obtained from 5-week-old C57BL/6 mice. (Left) Isotype staining.Results in A–C and D–G show one representative set of data of three inde-pendent experiments. Results in E are presented as the mean � SD of at leastthree independent experiments. Similar staining results were obtained byusing either affinity purified goat anti-mouse NKp46 polyclonal antibodiesand the rat anti-NKp46 29A1.4 mAb (SI Fig. 7).

Walzer et al. PNAS � February 27, 2007 � vol. 104 � no. 9 � 3385

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

June

13,

202

0

tested the reactivity of anti-NKp46 antibodies on mouse tissuesections. A bright staining was observed on spleen sections incomparison with control antibodies. This staining was specific toNKp46, as it was abrogated by the addition of NKp46-Fcrecombinant protein (data not shown). Three-color immunoflu-orescence staining showed that NK cells were mostly localized inthe splenic red pulp at steady state. Under these conditions, NKcells were also present in parafollicular (Fig. 2) and medullar (SIFig. 9) lymph node areas. In spleen and lymph nodes, NK cellswere in close proximity to CD11cbright and CD11bbright cells thatinclude dendritic cell and macrophage subsets. NKp46 antibod-ies also stained NK cells in other organs such as the lung or theliver (SI Fig. 9). Thus, NKp46 staining allows the visualization ofNK cells in their microenvironment.

Consistent with NKp46 association with CD3� or FcR� anddata obtained with human NK cells (25), NKp46 antibody-mediated triggering induced freshly isolated mouse NK cells tosecrete IFN-� and to release their cytotoxic granules content,as measured by surface exposure of the lytic granule markerCD107a (31) (Fig. 3). Thus, resting NK cells may be activated

through NKp46 providing means to specifically activate NKcell effector functions. These results contrast with a recentstudy showing that only CD16 engagement triggers freshlyisolated human NK cells (32). However, the density of CD16surface expression is much higher than that of NKp46 onhuman NK cells. In contrast, CD16 is expressed at low densityon freshly isolated mouse NK cells and NKp46 is expressed athigher levels. Our reproducible activation of freshly isolatedmouse NK cells by anti-NKp46 reagents (both 29A1.4 mAb andantiserum), thus indicate that CD16 is not the only NK cellsurface receptor whose engagement leads to NK cell activa-tion. This finding is consistent with the association of NKp46with the same ITAM-bearing transduction polypeptides (FcR�and/or CD3�) as CD16 (19).

In Vivo Tagging of NK Cells via NKP46. The dissection of NK cellbiological functions has been complicated by the lack of selectivedeficiency models. Having shown that NKp46 was a specific NKcell marker, our goal was to use NKP46 regulatory sequences tocreate such models. To validate the feasibility of this strategy, wefirst generated a transgenic vector consisting of a 24-kb humangenomic region located between the NKP46 adjacent genesFCAR and NALP7 (Fig. 4A). From a transgenic founder (re-ferred to as huNKp46 Tg hereafter), offspring were obtained atMendelian frequencies, developed normally, and were fertile.BAB281 (anti-human NKp46) mAb that do not cross-react withmouse NKp46 (SI Fig. 7) (24) were used to assess the cell surfaceexpression of human NKp46 in these mice. Human NKp46 wasnot expressed on granulocytes, dendritic cells, B cells, T cells,and CD1d-�-gal-cer tetramer� NKT cells but expressed at a highand uniform level on NK cells (Fig. 4B). Moreover, humanNKp46 starts to be expressed at the immature stage of NK celldevelopment in the bone marrow (Fig. 4C) and remains subse-quently expressed at the same level by all NK cells isolated from

A

B

Spleen

Lymph Nodes

CD11c NKp46 CD19 CD11b NKp46 CD19CD3 NKp46 CD19

WP

RP

T

100µM

Fig. 2. NKp46 staining allows the in situ visualization of mouse NK cells. ForC57BL/6 mouse spleen (A) or inguinal lymph node (B), 7-�m serial frozensections were fixed with acetone and stained with anti-NKp46 antiserum(green), CD19 mAb (red), and either CD3, CD11c, or CD11b mAb (blue).Samples were analyzed by confocal microscopy. A representative picture foreach group is shown in the same anatomical region for each staining. WP,white pulp; RP, red pulp; T, T cell zone. (Original magnification, �16.)

CD

107a

2.9 0.3

096.6

IFN-γ

17.7 10.7

0.171.4

18.8 29.8

0.950.4

control anti-NKp46 anti-NK1.1

Fig. 3. Mouse NK cells can be activated via NKp46. C57BL/6 spleen cellsuspensions were stimulated for 4 h in wells coated with the indicated anti-bodies, in the presence of soluble anti-CD107a antibodies. Control antibodiesincluded mouse and goat IgG. Cells were then stained for surface DX5/CD3 andintracellular IFN-�. Results show the expression of CD107a and IFN-� in gatedDX5�CD3� cells, and are representative of five independent experiments.

98 96 98 99

D

-8kb +16kbA NKp46 exons

B

C

0 0.40.3

980.7 1.2

Granulo. DC B cells

T cells NKT cells NK cells

NKp46 Tg C57BL/6

3 97 96

NKp iNK mNK

LN lungliver blood

E

0

10

20

30

40

10:1 3:1 1:1B6

B6 + anti-NKp30B6 + anti-NKp46

Tg

Tg + anti-NKp30Tg + anti-NKp46

spec

ific

lysi

s (%

)

Fig. 4. NKP46 genomic sequence can be used to genetically tag mouse NKcells. (A) Schematic representation of the human genomic fragment used fortransgenesis. NKP46 exons are shown as black bars. (B) Flow cytometricmeasurement of human NKp46 expression in various gated subsets fromhuNKp46 transgenic mice. The indicated cell types were identified as de-scribed in Methods. Granulo., granulocytes; DC, dendritic cells. Human NKp46expression (open histogram, thick line) or isotype control (gray histogram,thin line) is shown in the indicated subsets in B and in subsequent panels. (C)Flow cytometric measurement of human NKp46 expression on NK precursors(NKp), immature and mature NK cells, identified as described in Fig. 1. (D) Flowcytometric measurement of human NKp46 expression in gated NK1.1� CD3�

cells from huNKp46 Tg mouse lymph nodes (inguinal), lung, liver and periph-eral blood. (E) Redirected lysis assay of LAK cells derived from B6 (C57BL/6) orhuNKp46 Tg (Tg) spleen cells against Daudi cells incubated with the indicatedantibodies. The cytolytic function of LAK cells prepared from B6 and huNKp46Tg mice were comparable. Results are representative of three experiments.

3386 � www.pnas.org�cgi�doi�10.1073�pnas.0609692104 Walzer et al.

Dow

nloa

ded

by g

uest

on

June

13,

202

0

all organs tested (Fig. 4D and data not shown). Remarkably, thepattern of human NKp46 expression in huNKp46 Tg mice wasthus similar to that of endogenous mouse NKp46 molecules inparental mice. Therefore, the cell surface expression of humanNKp46 defined NK cells in huNKp46 Tg mice demonstratingthat NKP46 regulatory sequences could be used to drive NK-specific gene expression. NK cells in huNKp46 Tg mice exhibitednormal counts, phenotype and effector function (data notshown). Importantly, redirected lysis was induced through hu-man NKp46 (Fig. 4E), indicating that human NKp46 moleculesare functional in mouse NK cells. Finally, i.v. administration ofanti-human NKp46 mAb led to a nearly complete disappearanceof NK cells from blood and all organs tested, 2 days afterinjection (SI Fig. 10). By contrast, NKT cell and �� T cell countswere not significantly affected (SI Fig. 10), indicating thathuNKp46 Tg mice can be used as a mouse model of NKcell-selective depletion.

Generation of Mouse Genetic Models for the in Vivo Dissection of NKCell Function. Next, we sought to identify a minimal promoterfrom the 24-kb huNKp46 construct. DNA alignments betweenseveral mammalian species revealed a 400-bp sequence up-stream of NKP46 exon 1 that is highly conserved throughevolution and could thus correspond to a functionally activeNKP46 promoter region (SI Fig. 11). A putative TATA box waspredicted in this region upstream of the transcription start siteand downstream of a series of consensus motifs for transcrip-tion factor binding sites. This 400-bp sequence was cloned intoa luciferase reporter vector to examine its capacity to promotetranscription in vitro. The 400-bp NKP46 sequence droveefficient luciferase expression in the NK cell line, NKL, ascompared with the ubiquitous SV40 promoter (Fig. 5A). Incontrast, this 400-bp NKP46 sequence was inefficient in theerythroleukemia K562 line, suggesting that the 400-bp se-quence upstream of NKP46 exon 1 featured both efficientpromoter activity and NK cell specificity. We thus used this400-bp NKP46 sequence to drive the transgenic expression ofa bicistronic cassette consisting of the cDNA sequences en-coding for the human diphtheria toxin (DT) receptor (DTR)and EGFP (Fig. 5B). Expression of the DTR protein shouldconfer DT sensitivity to NKp46-expressing cells leading to theselective ablation of these cells after DT administration (33).This construct, referred to as NKDTR/EGFP hereafter, wasinjected into fertilized mouse ovocytes. Of six founders, onedisplayed high transgene expression and was bred to start theNKDTR/EGFP transgenic mouse colony. Remarkably, EGFPwas expressed in NK cells in these mice, confirming that the400-bp NKP46 promoter retained the tissue-specificity of theendogenous Nkp46 regulatory sequences in vivo (Fig. 5C).DTR expression was similar to that of EGFP (data not shown).NK cells in NKDTR/EGFP transgenic mice occur at normalfrequency and display a normal phenotype (data not shown).Two injections of DT at 24-h interval led to the disappearanceof NK cells in mouse peripheral blood mononuclear cells (Fig.5D), as well as in spleen, lymph nodes, bone marrow, liver, andlung (Fig. 5E). NK cell ablation persisted for at least 7 days,before a progressive repopulation of the circulating compart-ment. The kinetics of reconstitution of circulating NK cells iscompatible with the 17-day NK cell half-life reported earlier(34). Importantly, DT injection did not affect T cell popula-tions, including NKT cells and �� T cells in all organs tested(Fig. 5F). To confirm the depletion of NK cells, splenocytesfrom control mice or NKDTR/EGFP mice treated with DT,were assayed in vitro against YAC-1 tumor cells. No cytotox-icity was detected when splenocytes were isolated from NK-DTR/EGFP mice treated with DT, in contrast to control mice(Fig. 5G). These data show that DT infusion in NKDTR/EGFPmice leads to a specific depletion of NK cells in all organs, that

is associated with the complete disappearance of in vitrocytotoxicity against YAC-1. As the small population ofNKp46� T cells expresses low level of NKp46 as compared withbone fide NK cells, it was not depleted using the above DTadministration protocols (data not shown). The generation ofNKDTR/EGFP transgenic mice thus provides an in vivo modelof selective and conditional NK cell ablation.

DiscussionUntil now, the identification of mouse NK cells relied on NK1.1or CD49b cell surface expression. However, these markers arenot specific of NK cells. Moreover, NK1.1 and CD49b patterns

Ano promoter SV40 NKp46

NKL K562 Rel

ativ

e lig

ht

un

tis

C

0.6 0.1 0.4

6 0.3

control transgenic

EGFP

NK cells

Granulo. B cells

γδ T cells

αβ T cells

NKT cells

BDT-R IRES EGFP

020

40

6080

100

D

F

% o

f co

ntr

ol

Time (days)

DT

B cells

αβ T cells

NKT cells

γδ T cells

NK cells

0

2040

60

80100

% o

f co

ntr

ol

93

NKp46 p

120

1 32 7 10 14

120

T cells

NK cells

EGFPspleen lymph

nodesbone

marrowlungliver

% o

f co

ntr

ol

20

40

60

80

100

0

E

1006 12 250 50

E/T ratio

10

20

30

40

50

0

60

% s

pec

ifc

lyis

is

GWT

NKDTR/EGFP

Fig. 5. Dissection of NKP46 regulatory sequences and generation of NKDTR/EGFP mice. (A) The putative 400-bp human NKP46 promoter was used in a dualluciferase assay in K562 or NKL cell lines. The assay was also performed withpromoterless or SV40 promoter vectors for negative and positive control,respectively. (B) Schematic representation of the transgenic construct used tomake NKDTR/EGFP transgenic mice. For details, see Methods. (C) Flow cyto-metric measurement of EGFP expression in immune cell types of NKDTR/EGFPor control mice, identified as in Fig. 1. Granulo, granulocytes. Results arerepresentative of at least three experiments. (D) NKDTR/EGFP transgenic orcontrol littermates were injected i.p. twice with DT within a 24-h interval. Thepercentage of NKp46�CD3� or CD3� cells in the blood was measured over timeafter DT injections. Results are representative of at least three experiments. (E)NKDTR/EGFP (open histograms) and wild-type littermates (filled histograms)were injected with DT. Twenty-four hours later, the number of NK cells wasmeasured in the indicated organs. Results show the mean � SD number of NKcells in DT-treated wild-type or NKDTR/EGFP mice, expressed as the percent-age of the controls (NK cell numbers in untreated wild-type mice); four miceper group. (F) NKDTR/EGFP transgenic or control mice were injected i.p. twicewith DT within a 24-h interval. Forty-eight hours after DT injections, thepercentage of NK cells, �� T cells and CD1d-restricted NKT cells was measuredin the peripheral blood. Representative results of three experiments areexpressed as the percentage of indicated cell subsets as compared with controlmice. (G) Spleen cells from mice in E were cultured o/n with IL-2 and used aseffector cells in a standard 4-h Cr51 cytotoxicity assay against YAC-1 cells.Results are expressed as the mean � SD percentage of specific lysis of YAC-1cells with four mice in each group.

Walzer et al. PNAS � February 27, 2007 � vol. 104 � no. 9 � 3387

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

June

13,

202

0

of expression are not totally overlapping. In particular, subsetsof NK cells do not express CD49b (Fig. 1G), which complicatesthe comparison of data regarding NK cells obtained in NK1.1�

and NK1.1� strains (15). By contrast, we show here that NKp46is a specific marker for all NK cells (CD49b� and CD49b�), inall mouse strains tested. This finding opens perspectives in thestudy of the role of NK cells in various conditions. For instance,there are numerous mouse disease models, either autoimmuneor infectious, that rely on specific mouse strains. The use ofNKp46 will thus make it possible to study of the role of NK cellsin these strain-dependent disorders. In particular, the cell sur-face expression of NKp46 allows to detect NK cells in autoim-mune prone strains such as NOD and NZW (data not shown),where the roles of NK cells remain to be precisely dissected orrevisited (35, 36).

Moreover, the comparison of NK cells across species has beencomplicated by various phenotypic definitions of NK cells. Here, weshow that NKp46 is selectively expressed by human and mouse NKcells, consistent with the original description of NKp46 in humans(24) and the recent generation of Ncr1gfp/gfp mice (37). In addition,we also documented the NK cell-specific expression of NKp46 inthree monkey species (Baboon, Rhesus, and Cynomolgus; SI Fig.12). Previous reports have shown that rat and bovine NK cellsspecifically express NKp46 (38, 39). We thus propose to unify thephenotypic definition of NK cells across mammalian species on thebasis of NKp46 cell surface expression.

We show that NK cells can be visualized in situ by means ofNKp46 staining on tissue sections. Our results document thepresence of NK cells in the red pulp of the spleen, as well asin the paracortex and medulla of lymph nodes consistent withprevious data (40–42). We further showed that NK cells insecondary lymphoid organs are in close proximity to CD11b�

and CD11c� cells, consistent with the cross-talks between NKand DC (43), as well as NK and macrophages (44). Theseresults provide the proof of principle that anti-NKp46 anti-bodies represent an efficient means for the identification ofNK cells in situ. In addition, GFP-tagging in NKDTR-EGFPtransgenic mice constitutes an alternative strategy of NK cellvisualization.

To evaluate the role of NK cells in vivo, previous studies haveclassically used anti-NK1.1 antibodies to deplete NK cells inC57BL/6 mice, or anti-asialo-GM1 in other strains of mice. How-ever, these markers are not NK cell-specific and might inducedepletion of various other T cell subsets, complicating the inter-pretation of such experiments. Moreover, administration of anti-bodies may induce undesirable side effects, such as NK cell acti-vation upon anti-NK1.1 cross-linking (45), or nonspecific effectsthrough interactions with Fc receptors expressed on many celltypes. To overcome these limitations, we took advantage of theidentification of a functional NKP46 promoter to generate a mousemodel of conditional NK cell ablation based on the DT/DTRsystem, successfully used before to ablate other cell types (33, 46).We showed that DT injection led to a complete and selectiveablation of NK cells.

In conclusion, we propose a unifying phenotypic definitionof NK cells across species based on the cell surface expressionof NKp46. Moreover, we foresee the extensive use of NKDTR-GFP transgenic mice to delineate the role of NK cells invarious conditions. Taken together, these innovative technicalapproaches will help to reveal the biological functions of NKcells and to develop informative preclinical models.

MethodsMice. All inbred mice (Charles River Laboratories, L’Arbresle,France) were purchased for use in this study. RAG1�/�,RAG1�/�CD3��/�FcR��/�, and transgenic mice were bred inpathogen-free breeding facilities at the Centre d’Immunologiede Marseille-Luminy (Marseille, France). All of the mice used

in this study were between 6 and 10 weeks of age. Experimentswere conducted in accordance with institutional guidelines foranimal care and use.

Flow Cytometry and Lymphocyte Preparation. Flow cytometricanalysis was done on a FACS Canto (Becton Dickinson, SanDiego, CA). For all staining procedures, cells were firstincubated with a buffer containing 10 �g/ml 2.4G2 antibody,2% normal mouse serum, and 2% horse serum to preventbinding of the goat or rat antibodies to Fc receptors. Cells weresubsequently stained with primary and secondary antibodiesdiluted in PBS 2% bovine serum. Granulocytes were identifiedin the bone marrow by high expression of GR1 and CD11b.Dendritic cells, B cells, TCR��� T cells, TCR��� T cells, andNK cells were identified in the spleen as CD11chighIA/IEhigh,CD19�, CD3�TCR��, CD3�TCR��� cells and CD3�NK1.1�,respectively. NK precursors were identified in the bone mar-row as described (47), as CD122�lin� cells, whereas immature/mature NK cells were defined as CD122�NK1.1�DX5� andCD122�NK1.1�DX5�, respectively. All antibodies were fromPharMingen. CD1d-restricted NKT cells were identified byalphagalactosylceramide-loaded CD1d tetramer staining (48).NKp46 expression was detected by using a polyclonal antibodyagainst mouse NKp46 (R & D Systems, Minneapolis, MN) orusing the rat 29A1.4 mAb raised against a mouse NKp46-Fcfusion protein from R & D Systems. More information on thismAb are available in SI Text and SI Fig. 7. Secondary stainingwas performed with goat anti-rat IgG or donkey anti-goat IgG(Invitrogen, Carlsbad, CA). Anti-human NKp46 mAb(BAB281) was from Beckman Coulter (Fullerton, CA). Lungand liver lymphocytes were prepared as described (49).

Immunofluorescence on Tissue Sections. Immunofluorescence wasperformed on 5- to 10-�m-thick serial frozen sections. Sectionswere fixed with acetone before staining with anti-CD3-APC,anti-CD19-PE and goat anti-mouse NKp46 followed by second-ary donkey anti-goat IgG-alexa 488 (Invitrogen). Slides wereanalyzed by confocal microscopy (Zeiss LSM 510, WelwynGarden City, Hertfordshire, U.K.).

NK Cell Stimulation Assay. Anti-NK1.1 (PK136) and anti- NKp46antibodies (antiserum or 29A1.4 mAb) were bound on plastic(96-well plates) overnight in carbonate buffer. Spleen lymphocytesisolated from RAG�/� or C57BL/6 mice were stimulated 4 h in thepresence of FITC-coupled anti-CD107a antibodies and Golgi-stop(PharMingen). Cells were subsequently stained with DX5/CD3antibodies and intracellular IFN-� using cytofix/cytoperm kit(PharMingen) before flow cytometric analysis.

We thank Claude Gregoire, Sophie Ugolini, Elena Tomasello, Ana Rey-nders, and C. Andrew Stewart (Centre d’Immunologie de Marseille-Luminy; CIML) for comments and advice; Mitch Kronenberg (La JollaInstitute of Allergy and Immunology, La Jolla, CA) for the generous gift ofCD1d tetramers; Adrien Kissenpfennig (CIML, Plateforme RIO), CecileGoujet (Centre National de la Recherche Scientifique, Villejuif, France),Anne Gillet, and Herve Sanchez (CIML) for help in the generation andmaintenance of the transgenic mice; the mouse functional genomics Plat-form of the Marseille-Nice Genopole for immunohistochemistry; Angel-ique Bole for help in the generation of the anti-mouse NKp46 mAb. E.V.’slaboratory is supported by European Union FP6, LSHB-CT-2004-503319-Allostem, Ligue Nationale contre le Cancer (‘‘Equipe labellisee La Ligue’’),Agence Nationale de la Recherche (‘‘Reseau Innovation Biotechnologies’’and ‘‘Microbiologie Immunologie–Maladies Emergentes’’), Institut Na-tional de la Sante et de la Recherche Medicale, Centre National de laRecherche Scientifique, and Ministere de l’Enseignement Superieur et dela Recherche. N.F. is supported by Agence Nationale de la Recherche(Reseau Innovation Biotechnologies). Innate Pharma laboratory is partlysupported by LSHB-CT-2004-503319-Allostem, and Agence Nationale dela Recherche (Reseau Innovation Biotechnologies).

3388 � www.pnas.org�cgi�doi�10.1073�pnas.0609692104 Walzer et al.

Dow

nloa

ded

by g

uest

on

June

13,

202

0

1. Moretta A, Bottino C, Mingari MC, Biassoni R, Moretta L (2002) Nat Immunol3:6–8.

2. Degli-Esposti MA, Smyth MJ (2005) Nat Rev Immunol 5:112–124.3. Stewart CA, Vivier E, Colonna M (2006) Curr Top Microbiol Immunol

298:1–21.4. Moffett-King A (2002) Nat Rev Immunol 2:656–663.5. Raulet DH (2004) Nat Immunol 5:996–1002.6. Smyth MJ, Hayakawa Y, Takeda K, Yagita H (2002) Nat Rev Cancer

2:850–861.7. Ruggeri L, Capanni M, Mancusi A, Perruccio K, Burchielli E, Martelli MF,

Velardi A (2005) Int J Hematol 81:13–17.8. Vivier E, Anfossi N (2004) Nat Rev Immunol 4:190–198.9. Lanier LL, Testi T, Bindl J, Phillips JH (1989) J Exp Med 169:2233–2238.

10. Ryan JC, Niemi EC, Goldfien RD, Hiserodt JC, Seaman WE (1991) J Immunol147:3244–3250.

11. Carlyle JR, Mesci A, Ljutic B, Belanger S, Tai LH, Rousselle E, Troke AD,Proteau MF, Makrigiannis AP (2006) J Immunol 176:7511–7524.

12. Arase H, Saito T, Phillips JH, Lanier LL (2001) J Immunol 167:1141–1144.13. Slifka MK, Pagarigan RR, Whitton JL (2000) J Immunol 164:2009–2015.14. Hamaguchi-Tsuru E, Nobumoto A, Hirose N, Kataoka S, Fujikawa-Adachi K,

Furuya M, Tominaga A (2004) Br J Haematol 124:819–827.15. Di Santo JP (2006) Annu Rev Immunol 24:257–286.16. Blom B, Spits H (2005) Annu Rev Immunol 24:287–320.17. Kim S, Iizuka K, Aguila HL, Weissman IL, Yokoyama WM (2000) Proc Natl

Acad Sci USA 97:2731–2736.18. Kim S, Song YJ, Higuchi DA, Kang HP, Pratt JR, Yang L, Hong CM,

Poursine-Laurent J, Iizuka K, French AR, et al. (2006) Blood 107:1024–1030.19. Moretta A, Biassoni R, Bottino C, Mingari MC, Moretta L (2000) Immunol

Today 21:228–234.20. Moretta L, Moretta A (2004) EMBO J 23:255–259.21. Mandelboim O, Lieberman N, Lev M, Paul L, Arnon TI, Bushkin Y, Davis DM,

Strominger JL, Yewdell JW, Porgador A (2001) Nature 409:1055–1060.22. Moretta A, Bottino C, Vitale M, Pende D, Cantoni C, Mingari MC, Biassoni

R, Moretta L (2001) Annu Rev Immunol 19:197–223.23. Hollyoake M, Campbell RD, Aguado B (2005) Mol Biol Evol 22:1661–1672.24. Sivori S, Vitale M, Morelli L, Sanseverino L, Augugliaro R, Bottino C, Moretta

L, Moretta A (1997) J Exp Med 186:1129–1136.25. Pessino A, Sivori S, Bottino C, Malaspina A, Morelli L, Moretta L, Biassoni R,

Moretta A (1998) J Exp Med 188:953–960.26. Sivori S, Falco M, Marcenaro E, Parolini S, Biassoni R, Bottino C, Moretta L,

Moretta A (2002) Proc Natl Acad Sci USA 99:4526–4531.27. Takeda K, Cretney E, Hayakawa Y, Ota T, Akiba H, Ogasawara K, Yagita H,

Kinoshita K, Okumura K, Smyth MJ (2005) Blood 105:2082–2089.

28. Voehringer D, Shinkai K, Locksley RM (2004) Immunity 20:267–277.29. Robbins SH, Terrizzi SC, Sydora BC, Mikayama T, Brossay L (2003) J Immunol

170:5876–5885.30. Meresse B, Curran SA, Ciszewski C, Orbelyan G, Setty M, Bhagat G, Lee L,

Tretiakova M, Semrad C, Kistner E, et al. (2006) J Exp Med 203:1343–1345.31. Rubio V, Stuge TB, Singh N, Betts MR, Weber JS, Roederer M, Lee PP (2003)

Nat Med 9:1377–1382.32. Bryceson YT, March ME, Ljunggren HG, Long EO (2005) Blood 107:159–166.33. Jung S, Unutmaz D, Wong P, Sano G, De los Santos K, Sparwasser T, Wu S,

Vuthoori S, Ko K, Zavala F, et al. (2002) Immunity 17:211–220.34. Jamieson AM, Isnard P, Dorfman JR, Coles MC, Raulet DH (2004) J Immunol

172:864–870.35. Poirot L, Benoist C, Mathis D (2004) Proc Natl Acad Sci USA 101:8102–

8107.36. Johansson S, Berg L, Hall H, Hoglund P (2005) Trends Immunol 26:613–618.37. Gazit R, Gruda R, Elboim M, Arnon TI, Katz G, Achdout H, Hanna J, Qimron

U, Landau G, Greenbaum E, et al. (2006) Nat Immunol 7:517–523.38. Storset AK, Kulberg S, Berg I, Boysen P, Hope JC, Dissen E (2004) Eur

J Immunol 34:669–676.39. Westgaard IH, Berg SF, Vaage JT, Wang LL, Yokoyama WM, Dissen E,

Fossum S (2004) J Leukoc Biol 76:1200–1206.40. Andrews DW, Farrell HE, Densley EH, Scalzo AA, Shellam GR, Degli-Esposti

MA (2001) J Immunol 166:1796–1802.41. Bajenoff M, Breart B, Huang AY, Qi H, Cazareth J, Braud VM, Germain RN,

Glaichenhaus N (2006) J Exp Med 203:619–631.42. Dokun AO, Chu DT, Yang L, Bendelac AS, Yokoyama WM (2001) J Immunol

167:5286–5293.43. Walzer T, Dalod M, Robbins SH, Zitvogel L, Vivier E (2005) Blood 106:2252–

2258.44. Baratin M, Roetynck S, Lepolard C, Falk C, Sawadogo S, Uematsu S, Akira

S, Ryffel B, Tiraby JG, Alexopoulou L, et al. (2005) Proc Natl Acad Sci USA102:14747–14752.

45. Arase N, Arase H, Park SY, Ohno H, Ra C, Saito T (1997) J Exp Med186:1957–1963.

46. Kissenpfennig A, Henri S, Dubois B, Laplace-Builhe C, Perrin P, Romani N,Tripp CH, Douillard P, Leserman L, Kaiserlian D, et al. (2005) Immunity22:643–654.

47. Vosshenrich CA, Samson-Villeger SI, Di Santo JP (2005) Curr Opin Immunol17:151–158.

48. Matsuda JL, Gapin L, Fazilleau N, Warren K, Naidenko OV, Kronenberg M(2001) Proc Natl Acad Sci USA 98:12636–12641.

49. Masopust D, Vezys V, Usherwood EJ, Cauley LS, Olson S, Marzo AL, WardRL, Woodland DL, Lefrancois L (2004) J Immunol 172:4875–4882.

Walzer et al. PNAS � February 27, 2007 � vol. 104 � no. 9 � 3389

IMM

UN

OLO

GY

Dow

nloa

ded

by g

uest

on

June

13,

202

0