ava i l ab l e a t www.sc i enced i rec t . com

www.e l sev i e r. com/ l oca te /yc l im

Clinical Immunology (2008) 129, 145–154

In vitro expanded human invariant natural killerT-cells promote functional activity of naturalkiller cellsMaría Moreno a,b, Johan W. Molling a,1, Silvia von Mensdorff-Pouilly b,René H.M. Verheijen c, B. Mary E. von Blomberg a,Alfons J.M. van den Eertwegh d, Rik J. Scheper a, Hetty J. Bontkes a,⁎

a Department of Pathology, VU University Medical Center, Amsterdam, The Netherlandsb Department of Obstetrics and Gynaecology, VU University Medical Center, Amsterdam, The Netherlandsc Department of Woman and Baby, Division of Surgical and Oncological Gynaecology, University Medical Center Utrecht,Utrecht, The Netherlandsd Department of Medical Oncology, VU University Medical Center, Amsterdam, The Netherlands

Received 15 April 2008; accepted with revision 2 July 2008Available online 15 August 2008

Abbreviations: α-GC, α-galactosyllymphocytes; DC, dendritic cells; IFN-iant natural killer T-cells; moDC, monoNK, natural killer cells; PBMC, peripheTNF-α, tumour necrosis factor-α.⁎ Corresponding author. Current affil

tology, VU University Medical Center,Amsterdam, The Netherlands. Fax: +3

E-mail address: hj.bontkes@vumc.n1 Current affiliation: Nijmegen Cente

(NCMLS), PO Box 9101, 6500 HB, Nijme

1521-6616/$ – see front matter © 200doi:10.1016/j.clim.2008.07.004

Abstract Invariant natural killer T (iNKT) cells play a pivotal role in cancer immunity throughtrans-activation of effector cells via swift cytokine secretion. In mice, iNKT cell activation byα-galactosylceramide (α-GC) induces potent NK cell-mediated anti-tumour effects. Here weinvestigated whether human iNKT cells could enhance NK cell functional activity in vitro. iNKTcell activation by α-GC treatment of peripheral blood mononuclear cells (PBMC) was notsufficient to enhance NK cell effector functions. However, addition of in vitro expanded iNKTcells to PBMC enhanced NK cell-mediated cytotoxicity in an α-GC-dependent manner. NK cellactivation by iNKT cells was primarily mediated by soluble factors, and could be enhanced bythe NK cell activating cytokine IL-21. These results suggest that adoptive transfer of ex vivoexpanded iNKT cells will enhance NK cell function and is expected to enhance the efficacy ofcancer immunotherapy, particularly in combination with IL-21 and α-GC.© 2008 Elsevier Inc. All rights reserved.

KEYWORDSHuman;Invariant NKT cells;α-galactosylceramide;NK cells;Cytotoxicity;Interferon-γ

ceramide; CTL, cytotoxic Tγ, interferon-γ; iNKT, invar-cyte-derived dendritic cells;ral blood mononuclear cells;

iation: Department of Hema-De Boelelaan 1117, 1081HV

1 20 4442601.l (H.J. Bontkes).r for Molecular Life Sciencesgen, The Netherlands.

8 Elsevier Inc. All rights reserved

Introduction

Invariant natural killer T (iNKT) cells are T lymphocytescharacterized by an invariant TCR α chain gene rearrangement(Vα24-Jα18 paired with Vβ11 in humans) and co-expression ofNK cell receptors [1]. This invariant TCR recognizes glycolipidligands, such as the putative endogenous ligand isoglobotri-hexosylceramide (iGb3), bacterial glycosphyngolipids or theartificial ligand a-galactosylceramide (α-GC), in the context ofthe monomorphic CD1d antigen-presenting molecule [2–7].

.

146 M. Moreno et al.

Upon stimulation, iNKT cells have the capacity to rapidlyproduce large amounts of cytokines e.g. IFN-γ and IL-4 [1],leading to the trans-activation of other immune cells, such asdendritic cells (DC), natural killer (NK), Tand B cells [8–11]. Inmice, iNKT cells play an important role in tumour immuno-surveillance [8,12–17]. In vivo iNKT cell activation by α-GC orα-GC-loaded DC can induce potent anti-tumour immuneresponses in various tumour models by NK cell trans-activation[18–22].

In humans, we and others have demonstrated thatcancer patients have reduced circulating iNKT cell numbers[23–27]. In addition, we recently observed in a prospectivestudy that a relatively large circulating iNKT cell poolpredicts favourable clinical outcome in individuals withsquamous cell carcinoma of the head and neck [28]. Thesefindings suggest that cancer patients might benefit fromimmunotherapy aimed at the expansion of their peripheralblood iNKT cell pool.

We have recently demonstrated that human iNKTcells canenhance tumour antigen specific cytotoxic T lymphocyte(CTL) responses (Moreno et al. J. Immunol. In press2). Next toCTL, human NK cells have a high cytolytic potential againsttumour cells. NK cells discriminate between normal andtumour cells due to the loss of NK cell inhibiting MHC class Imolecules [29]. Since MHC class I loss renders tumour cellsinvisible to MHC class I restricted CTL [30], it is relevant toexamine whether human iNKTcells not only enhance tumourantigen specific CTL responses but also induce NK cellfunctional activity. Here, we investigated the effect ofhuman iNKT cell activation on phenotypic and functional NKcell activation.

Materials and methods

Cell lines and culture media

IMDM (Cambrex, Verviers, Belgium) was supplemented with10% foetal calf serum (Perbio, Helsingborg, Sweden) forculture of peripheral blood mononuclear cells (PBMC),monocyte-derived DC (moDC) and the NK-sensitive K562(ATCC, Manassas, VA, USA) cell line, or 8% human pooledserum (Sanquin, Amsterdam, The Netherlands) for theculture of iNKT cells. Both media were supplemented with100IU/ml sodium penicillin (Yamanouchi Pharma, Leider-dorp, The Netherlands), 100μg/ml streptomycin sulphate(Radiumfarma-Fisiopharma, Naples, Italy) and 0.01mM β-mercaptoethanol (Merck, Darmstadt, Germany).

Isolation of peripheral blood mononuclear cells(PBMC), NK cell purification, iNKT cell expansion

PBMC were isolated from buffycoats from healthy donors ob-tained from Sanquin BloodBank (Amsterdam, TheNetherlands)

2 Interferon-g producing human invariant Natural Killer T-cellspromote tumour associated antigen specific cytotoxic T cellresponses. María Moreno, Johan W Molling, Silvia von Mensdorff-Pouilly, René HM Verheijen, Erik Hooijberg, Duco Kramer, Anneke WReurs, Alfons JM van den Eertwegh, B Mary E von Blomberg, Rik JScheper, Hetty J Bontkes. J. Immunol. In press.

by density gradient centrifugation using Lymphoprep (Nyco-med, Oslo, Norway). The number of cells was determined andPBMC were cryopreserved in separate batches until use. TogeneratemoDC, 50 × 106 PBMCwere allowed to adhere in a T75flask for 60–90min. Non-adherent cells were washed away byrinsing the flask with PBS. Adherent monocytes were thencultured in the presence of 1000IU/ml GM-CSF (Schering-Plough, Kenilworth, NJ, USA) and 10ng/ml IL-4 (R and DSystems, Abingdon, UK) for 7days. Maturation was induced by48h culture in the presence of 30% v/v monocyte conditionedmedium (MCM) and 50ng/ml TNF-α (Strathmann Biotech,Hanover, Germany) as described [31].

Vα24+ Vβ11+ iNKT cells were enriched from PBMC bypositive selection using the iNKT cell specific antibody 6B11(BD Biosciences, San Jose, CA, USA) and anti-mouse Ig-coated magnetic beads mAb (Milteny Biotec, BergischGladbach, Germany) by MACS sorting. The non-selectedPBMC (iNKT cell-depleted PBMC, containing b 0.00% iNKTcells among CD3+ lymphocytes) were cryopreserved untiluse. The selected iNKT cells were expanded by weeklystimulation with irradiated autologous PBMC at a 1:1 iNKT:PBMC ratio in the presence of 100ng/ml α-GC (KRN7000,kindly provided by Dr Shigeyuki Yamano, KIRIN Brewery,Gunma, Japan), 40IU/ml recombinant human (rh) IL-2, 5ng/ml rhIL-7 and 5ng/ml rhIL-15 (all purchased from StrathmannBiotech, Hanover, Germany).

NK cell enrichment was performed by negative selectionusing Easysep human NK cell enrichment kit (StemCellTechnologies, Grenoble, France) according to the manufac-turer's recommendations. Achieved purity was higher than99.5%.

Evaluation of NK cell activation by iNKT cells

Donor PBMC (5 × 106 cells/ml) were cultured in the presence orabsence of 100ng/ml α-GC. In vitro expanded iNKTcells wereadded to iNKTcell-depleted PBMC at different percentages, asspecified, and cultured for 5days. In some experiments, IL-21(25ng/ml) was added to the cultures on day 2.

In order to analyse the involvement of soluble factors, 2 ×106 NK cells were co-cultured with 2 × 105 iNKTcells, 2 × 105

α-GC-pulsed moDC or both in a trans-well system (0.4μmpore size, Corning Life Science, Schiphol-Rijk, The Nether-lands) or in the presence of 400IU/ml of IFN-γ (Biosource,Camarillo, CA, USA), or 4μg/ml IFN-γ neutralizing antibody(R and D Systems, Abingdon, UK), for 5days.

NK cell effector function analysis

After 5days of iNKTcell-mediated stimulation of either PBMC orNK cells, NK cell effector functions were analysed. Cytolyticactivity of the NK cells to K562 cells was assessed using astandard chromium release assay. Briefly, 1 × 106 K562 targetcells were labelled with 100μCi of Na2[51Cr]O4 (Amersham,Bucks, U.K.) for 2h at 37°C and washed extensively. EffectorPBMC were harvested, washed and added to 5 × 103 target cellsat the indicated E:T ratios in triplicate wells of a round-bottom96-well plate (Nunc). After 18h incubation at 37°C, 50μl of thesupernatant was harvested, and its radioactive content wasmeasured. The percentage specific lysis was defined as follows:[(experimental release − spontaneous release) / (maximumrelease − spontaneous release)] × 100%.

147In vitro expanded human invariant natural killer T-cells promote functional activity of natural killer cells

Alternatively, NK cell cytotoxic capacity was determinedin a CD107a (LAMP1) translocation assay [32]. NK cells (1 ×105 cells) were cultured in the presence of PE-labelled mouseanti-human CD107a (BD Biosciences, San Jose, CA, USA) and4μM monensine in the absence (background translocation) orpresence of K562 target cells (0.5 × 105 cells). After 5h, cellswere washed and counterstained with CD3 and CD56 for atleast 30min at 4°C. Flow cytometry was performed on aFACSCALIBUR™ apparatus and data were analysed usingCellQuest™ software (BD Biosciences, San Jose, CA, USA).The percentage of CD107a positive NK cells and the meanfluorescence index (MFI) of CD107a expression induced byK562 were calculated as follows:

% CD107a+ = % CD107a+ upon K562 stimulation − % CD107a+

background; MFI = mean fluorescence intensity upon K562stimulation / background mean fluorescence intensity.

Cytokine detection

For determination of cytokine secretion, supernatants wereharvested after 5days of culture. Supernatants were analysedfor IL-2, IFN-γ andTNF-αbyELISAusing the appropriate PeliKineCompact human ELISA kit (Sanquin, Amsterdam, The Nether-lands) according to the manufacturer's recommendations.

Surface marker expression on iNKT and NK cells

FITC-, PE-, PerCP-Cy5- or APC-labelled isotype controls andmouse anti-human 6B11, CD3, CD4, CD16, CD25, CD69 (BDBioscience, San Jose, CA, USA), Vα24, Vβ11 (Immunotech,Marseille, France) and CD56 (IQ products, Groningen, TheNetherlands) were used to determine the phenotype of NK andiNKT cells by flow cytometry. Cells were incubated with theantibodies for at least 30min at 4°C. Flow cytometry wasperformed on a FACSCALIBUR™ apparatus and data wereanalysed using CellQuest™ software (BD Biosciences, SanJose, CA, USA).

IFN-γ and TNF-α intracellular cytokinestaining (ICCS)

Intracellular IFN-γ and TNF-α staining was performed usingthe BD cytofix/cytoperm plus kit (BD Bioscience, San Jose,CA, USA) according to the manufacturer's instructions. After5days of iNKT cell-stimulation of NK cells, GolgiPlug wasadded to each well (0.1% v/v) for 5h. iNKT and NK cellswere washed and stained with FITC-labelled 6B11 andPerCP-Cy5-labelled CD3, and APC-labelled CD56 respec-tively, followed by intracellular staining with PE-labelledanti-IFN-γ or PE-labelled anti-TNF-α (both from BDBioscience, San Jose, CA, USA).

Statistics

Normality of distribution of the data was analysed withKolmogorov–Smirnov test and Q–Q plots. The non-para-metric Wilcoxon Signed Ranks test was used to analyseincrease in cytokine (IFN-γ and TNF-α) production. Other-wise, a parametric paired Student's T-test was used toanalyse the differences between groups. Non-parametriccorrelation Spearman's test was performed to investigate therelation between cytokine secretion and up-regulation of

activation markers. p b 0.05 was considered significant. Alldata were analysed using SPSS 14.0 software.

Results

iNKT cell activation with α-GC is not sufficient toaugment NK cell activity

In vitro stimulation of PBMC, containing various numbers ofiNKT cells (range 0.01–0.19% iNKT cells among CD3+ lympho-cytes), with α-GC for up to 5days induced iNKTcell activation,as demonstrated by an increased proportion of iNKT cellsexpressing CD69 and CD25 (p b 0.001 and p = 0.017,respectively) (Figs. 1A, B). A limited, but statisticallysignificant NK cell activation was observed as well (p = 0.024and 0.027, respectively) (Figs. 1C, D). However, this limitedphenotypic NK cell activation may not be biologicallysignificant since no enhanced NK cell-mediated cytotoxicityagainst K562was observed (Fig. 1E). Moreover, IFN-γ, as well asTNF-α, secretion was only modestly increased following α-GCtreatment of total PBMC [Mean (range) 53 (0–194) vs 185 (0–768) pg/ml IFN-γ, and 11 (0–30) vs 25 (5–62) pg/ml TNF-α, insupernatants of unstimulated vs α-GC-stimulated PBMC, p N0.05]. Of note, no increase in IL-2 secretion could be detected.Thus, while addition of α-GC to total PBMC was sufficient toinduce phenotypic activation of iNKTcells accompanied with amodest increase in cytokine production, this activation wasinsufficient to establish enhanced NK cell lytic activity.

Addition of expanded iNKT cells to iNKT-depletedPBMCs enhances NK cell cytotoxicity in anα-GC-dependent manner

We have previously developed a method to generate highpurity pro-inflammatory iNKT cell lines from healthy donorsin vitro [33]. Using an adapted version of this method, weexpanded iNKT cells from healthy donor PBMC to investigatetheir capacity to activate NK cells. After 2 expansion rounds,purity of iNKT cells was more than 90%, and the cellsexhibited an activated phenotype, demonstrated by CD69expression (median 7%, range 1–34%, n = 5) and an increasedCD25 expression (median 46%, range 32–66%, n = 5),compared to resting iNKT cells (Fig. 1B, open bars).Expression of both activation markers could be furtherincreased by antigen stimulation (data not shown).

To test the capacity of these pre-activated iNKT cells toactivate NK cells, iNKT-depleted PBMC were co-cultured withincreasing percentages (within the physiological range, i.e.0.02%, 0.10% and 0.50%) of in vitro expanded iNKTcells (N 90%pure) for 5days. Addition of iNKTcells resulted in an iNKT celldose-dependent increase in IFN-γ (Fig. 2A) and TNF-α (data notshown) secretion. Higher cytokine levels were detected incultures containing α-GC, although there was a large variationbetween donors. Ex vivo stimulation of total PBMC with α-GCleads to much lower levels of IFN-γ (see previous paragraph;185 (0–768) pg/ml), which is indicative for enhanced cytokinesecretion by expanded iNKT cells compared to ex vivostimulated iNKTcells. Addition of pre-activated iNKT cells didnotmodify percentage of NK cells but enhancedCD25 andCD69expression on NK cells to a similar extend as adding α-GC tototal PBMC (data not shown). Of note, after 5days the

Figure 1 Invariant NKTcell activation by PBMC stimulation with α-GC. PBMC were stimulated with 100ng/ml α-GC. Activation statusof (A, B) iNKTcells (defined as CD3+ 6B11+ cells) and (C, D) NK cells (defined as CD3− CD56+ cells) after 5-day culture. (A, C) Exampledot-plots of CD69 and CD25 expression on iNKT and NK cells after 5days without and with α-GC. The numbers in the right upperquadrant represent the % of positive cells. (B, D) The graphs show the mean ± SEM of CD69 and CD25 expression of 5 independentexperiments performed with individual donors, in the absence (open bars) or presence (solid bars) of α-GC. Paired Student's T-test:⁎p b 0.05. (E) NK cell effector function was assessed by K562 cytotoxic assays. Results from one representative experiment out of 7performed are shown.

148 M. Moreno et al.

proportion of iNKT cells in culture remained also unchanged.However, in addition to this phenotypic activation, NK cellfunctional activity was enhanced as well. Despite somevariation in K562 lysis (5.4–19.9%) due to differences in NKcell frequencies between donors, the enhanced effect of theiNKTcells was observed in all donors (1.2–3.3 fold increase incytotoxicity at E:T = 12.5:1). In the absence of α-GC, the

addition of expanded iNKTcells to PBMC had a limited effect onNK cell cytotoxicity (Fig. 2B). On the other hand, NK cell abilityto kill K562 was substantially augmented when PBMC were co-culturedwith in vitro expanded iNKTcells in the presence of α-GC, in an iNKT cell dose-dependent manner (Fig. 2C). Thiseffect is not mediated directly by iNKTcells, as they have beenshown to be unable to kill K562 [25,34].

Figure 3 Effect of α-GC-stimulated iNKT cells and IL-21combination on NK cell activation. K562 cytolysis performedwith (A) unstimulated and (B) α-GC-stimulated iNKT-depletedPBMC without (open squares) or with (solid triangles) 0.20%in vitro expanded iNKT cells, in the absence (solid lines) orpresence (dashed lines) of IL-21. Results shown are from onerepresentative experiment out of 3 performed.

Figure 2 α-GC-dependent NK cell activation by expanded iNKT cells. iNKT-depleted PBMC were co-cultured with 0.02, 0.10 and0.50% in vitro expanded iNKTcells, in the absence or presence of iNKTcell ligand. After 5days, (A) IFN-γ secretion in supernatants wasdetermined by ELISA. Wilcoxon Signed Ranks test: ⁎p b 0.05. (B, C) NK cell effector function was assessed by K562 cytotoxic assays.Results from one representative experiment out of 5 performed are shown. iNKT-depleted PBMC (open squares) were co-cultured with0.02, 0.10 and 0.50% in vitro expanded iNKT cells (solid symbols) (B) in the absence or (C) presence of α-GC.

149In vitro expanded human invariant natural killer T-cells promote functional activity of natural killer cells

IL-21 is a strong NK cell activating cytokine. Indeed,in vivo activation of murine iNKT cells by α-GC injection,followed by IL-21 administration, enhanced NK cell activity[13]. Therefore, we analysed whether sequential activationof human iNKT (with α-GC) and NK cells (with IL-21) couldfurther enhance NK cell-mediated cytotoxicity. InvariantNKT-depleted PBMC were co-cultured with 0.20% in vitroexpanded iNKTcells (N 90% pure), in the presence or absenceof α-GC for 5days. At day 2, IL-21 was added to the co-cultures. The proportions of iNKT and NK cells remainedunaltered. The percentage in NK cell-mediated K562 killingvaried between 2.2 and 6.7%, which was modestly increasedwhen PBMC were stimulated with in vitro expanded iNKTcells followed by IL-21 in the absence of α-GC in all donorstested (1.3–3.9 fold increase at E:T=12.5:1) (Fig. 3A). In thepresence of α-GC, combination of both treatments induced astrong increase in NK cell cytolytic activity (5.5–10.3 foldincrease) (Fig. 3B). These data show that NK cell cytotoxicfunction can be enhanced by sequential activation of humaniNKT (with α-GC) and NK cells (with IL-21).

IFN-γ is sufficient but not necessary for iNKTmediated NK cell activation

To analyse whether the enhanced activation of NK cells by iNKTcells is dependent on cell–cell contact, soluble factors or both,co-cultures were set up in trans-wells. Lower wells containedpurified NK cells alone (− iNKT). Autologous in vitro expandediNKT cells were added to the lower wells (+ iNKTlow) or to theupper wells (+ iNKTup) alone or together with α-GC-pulsedmoDC (+ iNKTlow + DClow and + iNKTup + DCup, respectively).Appropriate moDC controls were included (+ DClow and + DCup).Analysis of the culture supernatants showed a significantincrease in IFN-γ (Wilcoxon Signed Ranks test, p = 0.018)(Fig. 4A), and to a lesser extent TNF-α (Wilcoxon Signed Rankstest, p = 0.018) (data not shown), secretion in wells containingboth iNKT cells and α-GC-pulsed moDC, compared to NK cellsalone. Intracellular cytokine analysis revealed that both

cytokines were produced by iNKT cells and not by NK cells(data not shown). Interestingly, IL-2 levels in supernatantsdropped under these conditions (data not shown), suggestingIL-2 consumption by NK cells, iNKTcells or both. In parallel withIFN-γ secretion, NK cell activation status (Figs. 4B, C),particularly CD69 expression, was augmented when NK cellswere cultured with iNKT cells in the presence of α-GC-pulsedmoDC. Likewise, NK cell lytic capacity against K562 induced byiNKT cells, determined by CD107a translocation, was alsoenhanced under these conditions. This was illustrated by anincrease both in the percentage of NK cells that translocatedCD107a (1.3–3.9 fold increase) (Fig. 4D) and in the amount ofdegranulation on a single cell level (1.4–2.5 fold increase in

Figure 4 NK cell activation by expanded human iNKTcells is mediated by soluble factors, the production of which is dependent oniNKTcell — α-GC-loaded moDC contact. Trans-well experiments were performed by culturing purified NK in the lower wells (− iNKT).Autologous in vitro expanded iNKTcells were added to the lower wells (+iNKTlow) or to the upper wells (+iNKTup) alone or together withα-GC-pulsed moDCs (+iNKTlow+DClow and +iNKTup+DCup, respectively). Appropriate moDC controls were included (+DClow and +DCup).After 5 days, (A) IFN-γ secretion in supernatants was assessed by ELISA (mean±SEM of 7 individual experiments). Wilcoxon Signed Rankstest: ⁎pb0.05. NK cell activation was determined as up-regulation of (B) CD69 and (C) CD25 expression (mean±SEM of 5 individualexperiments). Paired Student's T-test: ⁎pb0.05. NK cell lytic capacity against K562 was determined as (D) percentage of CD107apositive cells and (E) CD107a MFI within NK cell population (mean±SEM of 3 individual experiments).

150 M. Moreno et al.

CD107a surface expression level) (Fig. 4E). NK cell activationcould likely beattributed to the iNKTcells because addition ofα-GC-pulsedmoDCalone, either in the lower or in the upperwells,had no effect on NK cell activation.

These data suggest a role for soluble factors in iNKT cell-mediated NK cell activation. Indeed, correlation analysisshowed a significant correlation between increase in IFN-γsecretion and up-regulation of CD69 expression (p b 0.001)(Fig. 5A), as well as CD107a translocation by NK cells (p b

0.001) (Fig. 5B). Correlation was also observed for TNF-α (p =0.003 and b 0.001, respectively) (data not shown). Toconfirm that IFN-γ can indeed enhance the NK cell response,soluble recombinant IFN-γ was added to NK cell cultures.Indeed, addition of IFN-γ had similar enhancing effect asiNKT cells: up-regulation of CD69 and increase in lyticfunction in both donors tested (Figs. 6A, B). However, theenhanced effect of iNKT cells was not blocked by an IFN-γneutralizing antibody in either of the two donors tested

Figure 5 Human iNKT cell-mediated NK cell activation correlates with IFN-γ cytokine secretion. Spearman's correlation betweenincrease in IFN-γ secretion and (A) CD69 up-regulation on NK cells and (B) enhanced NK cell lytic capacity against K562.

151In vitro expanded human invariant natural killer T-cells promote functional activity of natural killer cells

(Figs. 6C, D), while IFN-γ secretion was completely inhibited(data not shown). These results suggest that beside IFN-γ,other soluble factors, such as TNF-α or IL-2, can play a role inNK cell activation.

Interestingly, higher cytokine levels, which were asso-ciated with the stronger NK cell activation, were inducedwhen NK cells were cultured in contact with both iNKT cellsand α-GC-loaded moDC compared to NK cells cultured withiNKT cells and α-GC-loaded moDC in the trans-well, indicat-ing a role for cell–cell contact between NK and iNKT cellsnext to soluble factors. This is also suggested by the

Figure 6 IFN-γ is sufficient but not necessary for iNKT mediatedα-GC-loaded moDC. 400 IU/ml soluble recombinant IFN-γ or 2×10CD69+ cells within NK cell population is shown. (B) NK cell lytic capwere cultured alone or with in vitro expanded iNKT cells in the prein the presence of 4 μg/ml blocking antibodies against IFN-γ (solidcells within NK cell population is shown. (D) NK cell lytic capacitythe results of one experiment out of two performed.

observation of slight, although not significant, increase incytokine secretion, and NK cell activation, after co-cultureof NK cells in contact with iNKTcells alone (+ iNKTlow) in someof the donors (Fig. 4).

Discussion

We investigated whether human iNKTcells activated by α-GCcould promote NK cell activation, in the context of improvingNK cell-mediated tumour immunotherapies. In murinemodels, resident iNKT cells can augment anti-tumour

NK cell activation. 2×106 NK cells were co-cultured with 2×1055 in vitro expanded iNKT cells were added. (A) Percentage ofacity against K562 was determined as CD107a MFI. NK cell cellssence or absence of α-GC-loaded moDC in a trans-well system,bars) or isotype control (open bars). (C) Percentage of CD69+

against K562 was determined as CD107a MFI. The graphs show

152 M. Moreno et al.

responses upon systemic treatment with α-GC or α-GC-loaded DC, as a result of NK cell trans-activation by iNKTcells[13,19,22]. Circulating iNKTcells of patients with cancer aresignificantly reduced compared to healthy controls [23–27].Nonetheless, the iNKTcells in these patients still possess thecapacity to proliferate and to secrete IFN-γ when properlystimulated in vitro [28]. This suggests that iNKT cells ofcancer patients, though reduced in number, may still becapable of enhancing anti-tumour responses in a physiologi-cal setting or after therapies aiming at their increase andactivation. One approach to achieve the latter could be bysystemic α-GC treatment. Here we demonstrate thatstimulation of total PBMC with α-GC strongly promotesphenotypic activation of iNKT cells as evidenced by a highlyincreased expression of CD69 and CD25. However, α-GCinduced only a modest increase in IFN-γ production.Although NK cells became phenotypically activated, func-tional activity of NK cells was not enhanced by this approach.In vivo, systemic treatment with free α-GC in mice inducesiNKT cell anergy, rendering iNKT cells unresponsive tosubsequent α-GC treatment [11]. Similarly, administrationof free α-GC in humans induces a transient and limitedincrease of cytokines (IFN-γ, TNF-α, IL-12 and GM-CSF)without enhancing NK cell cytolytic activity. Furthermore,the transient cytokine increase was only observed in patientswith normal pre-treatment iNKT cell numbers [35]. Theseresults combined with the data we present here wherestimulation with free α-GC, even of PBMC derived fromhealthy donors with relatively high numbers of iNKT cells, isnot sufficient to enhance iNKT cell numbers and NK cell-mediated cytolytic activity, exemplify that alternativetreatments to expand and activate iNKT cells in vivo arewarranted. This can be achieved either by vaccination withα-GC-pulsed mature moDC or by adoptive transfer of in vitroexpanded iNKT cells. In this study we demonstrate anenhanced, α-GC-dependent, NK cell activation status andcytotoxic effector function upon addition of expanded andpre-activated iNKTcells. Furthermore, we demonstrate thatsequential activation of in vitro expanded human iNKT cellswith α-GC, followed by IL-21 treatment to mature the iNKTcell-activated NK cells into highly cytotoxic effector cells,strongly enhanced NK cell cytotoxicity. Previously, Lin et al.[36] showed up-regulated CD69 expression on NK cells earlyafter addition of in vitro expanded iNKT cells to PBMC, butCD25 expression was not affected. Here we show that invitro expanded human iNKT cells not only promote pheno-typic activation of NK cells, but induce functional NK cellactivation as well. Both effects were dependent on α-GC.Trans-well experiments showed that iNKT cell-mediated NKcell activation is mainly mediated by soluble factors, andcorrelates with cytokine (IFN-γ and TNF-α) production.Recombinant IFN-γ induced a similar level of NK cellactivation as addition of iNKT cells. However, human iNKTcell-mediated NK cell activation was not solely dependent onIFN-γ, as neutralization of IFN-γ did not abrogate the effectsof iNKT cells (Figs. 6C, B). Interestingly, when NK cells wereco-cultured with iNKT cells in the presence of α-GC-loadedmoDC, IL-2 completely disappeared from supernatants (datanot shown), most likely due to consumption during NK cellactivation. Thus other factors such as TNF-α or IL-2 may besufficient to induce NK cell activation. A role for IL-2 wasindeed demonstrated by Metelitsa et al., who showed that

the iNKT cell-mediated enhancement of NK cell activity wasprimarily mediated by IL-2, an effect which was enhanced byIFN-γ [37]. However in contrast to our findings these authorsdemonstrated a small but significant effect of IFN-γneutralization while the addition of recombinant IFN-γ hadonly an effect when added in combination with IL-2. Thesedifferences may be explained by different dosages ofrecombinant IFN-γ added and the experimental conditions:i.e. the addition of iNKTcell culture supernatants versus theaddition of iNKT cells in trans-wells.

In addition to soluble factors, direct contact-mediatedinteractions between NK and iNKTcells appear to play a roleas well in iNKT cell-mediated NK cell activation. It is not yetclear which molecules are involved in iNKT cell-NK cellinteractions, and whether TCR ligation is required. Whileendogenous glycolipid antigens, such as iGb3 [6], could act asiNKT cell ligand, CD1d expression on NK cells has not beendemonstrated. Other co-stimulatory interactions may play arole. Of interest in this respect is the APC like properties ofactivated NK cells, expressing co-stimulatory molecules suchas CD80, CD86 and OX40 ligand as well as MHC class II [38].

We have recently shown that human iNKT cells enhanceantigen specific CTL responses as well. This effect was onlydependent on soluble factors and was completely blockedafter IFN-γ neutralization (Moreno et al., J. Immunol. Inpress1). These results suggest that the regulation of NK cellactivity and antigen specific CTL responses by iNKT cells ismediated by different mechanisms.

Recently it has been demonstrated that adoptive transferwith PBMC enriched for iNKT cells, is well tolerated andresulted in a transient increase in circulating iNKT cells anddirect ex vivo IFN-γ production in response to α-GC in anELISPOTassay [39]. Our results provide further evidence thatin vitro expanded human iNKT cells can activate NK cells,an effect which depends on α-GC and which is stronglyenhanced by IL-21. It may therefore be necessary to combineadoptive iNKT cell therapy with additional IL-21 or α-GC(pulsed DC) treatment to achieve clinical effects in vivo.

Acknowledgment

We thank Dr. S. Yamano (KIRIN Brewery, Gunma, Japan) forproviding KRN7000 (α-GC).

References

[1] D.I. Godfrey, K.J. Hammond, L.D. Poulton, M.J. Smyth, A.G.Baxter, NKT cells: facts, functions and fallacies, Immunol.Today 21 (2000) 573–583.

[2] T. Kawano, J. Cui, Y. Koezuka, I. Toura, Y. Kaneko, K. Motoki,H. Ueno, R. Nakagawa, H. Sato, E. Kondo, H. Koseki, M.Taniguchi, CD1d-restricted and TCR-mediated activation ofvalpha14 NKT cells by glycosylceramides, Science 278 (1997)1626–1629.

[3] Y. Kinjo, E. Tupin, D. Wu, M. Fujio, R. Garcia-Navarro, M.R.Benhnia, D.M. Zajonc, G. Ben-Menachem, G.D. Ainge, G.F.Painter, A. Khurana, K. Hoebe, S.M. Behar, B. Beutler, I.A. Wilson,M. Tsuji, T.J. Sellati, C.H. Wong, M. Kronenberg, Natural killer Tcells recognize diacylglycerol antigens from pathogenic bacteria,Nat. Immunol. 7 (2006) 978–986.

[4] D. Wu, G.W. Xing, M.A. Poles, A. Horowitz, Y. Kinjo, B. Sullivan,V. Bodmer-Narkevitch, O. Plettenburg, M. Kronenberg, M. Tsuji,

153In vitro expanded human invariant natural killer T-cells promote functional activity of natural killer cells

D.D. Ho, C.H. Wong, Bacterial glycolipids and analogs asantigens for CD1d-restricted NKT cells, Proc. Natl. Acad. Sci.U. S. A. 102 (2005) 1351–1356.

[5] J. Mattner, K.L. Debord, N. Ismail, R.D. Goff, C. Cantu III,D. Zhou, P. Saint-Mezard, V. Wang, Y. Gao, N. Yin, K. Hoebe,O. Schneewind, D. Walker, B. Beutler, L. Teyton, P.B. Savage,A. Bendelac, Exogenous and endogenous glycolipid antigensactivate NKT cells during microbial infections, Nature 434(2005) 525–529.

[6] D.Zhou, J.Mattner,C.Cantu III, N. Schrantz,N.Yin, Y.Gao,Y. Sagiv,K. Hudspeth, Y.P. Wu, T. Yamashita, S. Teneberg, D. Wang, R.L.Proia, S.B. Levery, P.B. Savage, L. Teyton, A. Bendelac, Lysosomalglycosphingolipid recognition by NKT cells, Science 306 (2004)1786–1789.

[7] F.M. Spada, Y. Koezuka, S.A. Porcelli, CD1d-restricted recogni-tion of synthetic glycolipid antigens by human natural killer Tcells, J. Exp. Med. 188 (1998) 1529–1534.

[8] M.J. Smyth, K.Y. Thia, S.E. Street, E. Cretney, J.A. Trapani, M.Taniguchi, T. Kawano, S.B. Pelikan, N.Y. Crowe, D.I. Godfrey,Differential tumor surveillance by natural killer (NK) and NKTcells, J. Exp. Med. 191 (2000) 661–668.

[9] H. Kitamura, A. Ohta, M. Sekimoto, M. Sato, K. Iwakabe, M.Nakui, T. Yahata, H. Meng, T. Koda, S. Nishimura, T. Kawano, M.Taniguchi, T. Nishimura, alpha-galactosylceramide inducesearly B-cell activation through IL-4 production by NKT cells,Cell Immunol. 199 (2000) 37–42.

[10] I.F. Hermans, J.D. Silk, U. Gileadi, M. Salio, B. Mathew, G. Ritter,R. Schmidt, A.L. Harris, L. Old, V. Cerundolo, NKTcells enhanceCD4+ and CD8+ T cell responses to soluble antigen in vivothrough direct interaction with dendritic cells, J. Immunol. 171(2003) 5140–5147.

[11] S. Fujii, K. Shimizu, C. Smith, L. Bonifaz, R.M. Steinman,Activation of natural killer T cells by alpha-galactosylceramiderapidly induces the full maturation of dendritic cells in vivo andthereby acts as an adjuvant for combined CD4 and CD8 T cellimmunity to a coadministered protein, J. Exp. Med. 198 (2003)267–279.

[12] N.Y. Crowe, M.J. Smyth, D.I. Godfrey, A critical role for naturalkiller T cells in immunosurveillance of methylcholanthrene-induced sarcomas, J. Exp. Med. 196 (2002) 119–127.

[13] M.J. Smyth, M.E. Wallace, S.L. Nutt, H. Yagita, D.I. Godfrey,Y. Hayakawa, Sequential activation of NKT cells and NK cellsprovides effective innate immunotherapy of cancer, J. Exp. Med.201 (2005) 1973–1985.

[14] R.R. Brutkiewicz, V. Sriram, Natural killer T (NKT) cells andtheir role in antitumor immunity, Crit. Rev. Oncol. Hematol. 41(2002) 287–298.

[15] K. Chamoto, T. Takeshima, A. Kosaka, T. Tsuji, J. Matsuzaki, Y.Togashi, H. Ikeda, T. Nishimura, NKTcells act as regulatory cellsrather than killer cells during activation of NK cell-mediatedcytotoxicity by alpha-galactosylceramide in vivo, Immunol.Lett. 95 (2004) 5–11.

[16] K. Takeda, Y. Hayakawa, M. Atsuta, S. Hong, K.L. Van, K.Kobayashi, M. Ito, H. Yagita, K. Okumura, Relative contributionof NK and NKT cells to the anti-metastatic activities of IL-12,Int. Immunol. 12 (2000) 909–914.

[17] M. Taniguchi, K. Seino, T. Nakayama, The NKT cell system:bridging innate and acquired immunity, Nat. Immunol. 4 (2003)1164–1165.

[18] Y. Akutsu, T. Nakayama, M. Harada, T. Kawano, S. Motohashi, E.Shimizu, T. Ito, N. Kamada, T. Saito, H. Matsubara, Y. Miyazawa,T. Ochiai, M. Taniguchi, Expansion of lung V alpha 14 NKT cellsby administration of alpha-galactosylceramide-pulsed dendri-tic cells, Jpn. J. Cancer Res. 93 (2002) 397–403.

[19] M. Nakui, A. Ohta, M. Sekimoto, M. Sato, K. Iwakabe, T. Yahata,H. Kitamura, T. Koda, T. Kawano, H. Makuuchi, M. Taniguchi,T. Nishimura, Potentiation of antitumor effect of NKT cellligand, alpha-galactosylceramide by combination with IL-12 on

lung metastasis of malignant melanoma cells, Clin. Exp.Metastasis 18 (2000) 147–153.

[20] M. Nakui, K. Iwakabe, A. Ohta, M. Sekimoto, M. Sato, H. Makuuchi,T. Kawano, M. Taniguchi, T. Nishimura, Natural killer T cell ligandalpha-galactosylceramide inhibited lymph node metastasis ofhighly metastatic melanoma cells, Jpn. J. Cancer Res. 90 (1999)801–804.

[21] T. Kawano, T. Nakayama, N. Kamada, Y. Kaneko, M. Harada, N.Ogura, Y. Akutsu, S. Motohashi, T. Iizasa, H. Endo, T. Fujisawa,H. Shinkai, M. Taniguchi, Antitumor cytotoxicity mediated byligand-activated human V alpha24 NKT cells, Cancer Res. 59(1999) 5102–5105.

[22] R. Nakagawa, I. Nagafune, Y. Tazunoki, H. Ehara, H. Tomura,R. Iijima, K. Motoki, M. Kamishohara, S. Seki, Mechanisms ofthe antimetastatic effect in the liver and of the hepatocyteinjury induced by alpha-galactosylceramide in mice, J.Immunol. 166 (2001) 6578–6584.

[23] J.W. Molling, W. Kolgen, H.J. van der Vliet, M.F. Boomsma,H. Kruizenga, C.H. Smorenburg, B.G. Molenkamp, J.A. Langen-dijk, C.R. Leemans, B.M. von Blomberg, R.J. Scheper, A.J. vanden Eertwegh, Peripheral blood IFN-gamma-secreting Valpha24+Vbeta11+ NKT cell numbers are decreased in cancer patientsindependent of tumor type or tumor load, Int. J. Cancer 116(2005) 87–93.

[24] S. Motohashi, S. Kobayashi, T. Ito, K.K. Magara, O. Mikuni,N. Kamada, T. Iizasa, T. Nakayama, T. Fujisawa, M. Taniguchi,Preserved IFN-alpha production of circulating Valpha24 NKTcells in primary lung cancer patients, Int. J. Cancer 102 (2002)159–165.

[25] M.V. Dhodapkar, M.D. Geller, D.H. Chang, K. Shimizu, S. Fujii,K.M. Dhodapkar, J. Krasovsky, A reversible defect in naturalkiller T cell function characterizes the progression of pre-malignant to malignant multiple myeloma, J. Exp. Med. 197(2003) 1667–1676.

[26] S.M. Tahir, O. Cheng, A. Shaulov, Y. Koezuka, G.J. Bubley,S.B. Wilson, S.P. Balk, M.A. Exley, Loss of IFN-gamma produc-tion by invariant NK Tcells in advanced cancer, J. Immunol. 167(2001) 4046–4050.

[27] J. Konishi, K. Yamazaki, H. Yokouchi, N. Shinagawa, K. Iwabuchi,M. Nishimura, The characteristics of human NKT cells in lungcancer—CD1d independent cytotoxicity against lung cancer cellsby NKT cells and decreased human NKT cell response in lungcancer patients, Hum. Immunol. 65 (2004) 1377–1388.

[28] J.W. Molling, J.A. Langius, J.A. Langendijk, C.R. Leemans,H.J. Bontkes, H.J. van der Vliet, B.M. von Blomberg, R.J.Scheper, A.J. van den Eertwegh, Low levels of circulatinginvariant natural killer T cells predict poor clinical outcome inpatients with head and neck squamous cell carcinoma, J. Clin.Oncol. 25 (2007) 862–868.

[29] L. Moretta, C. Bottino, D. Pende, R. Castriconi, M.C. Mingari,A. Moretta, Surface NK receptors and their ligands on tumorcells, Semin. Immunol. 18 (2006) 151–158.

[30] I. Algarra, A. Garcia-Lora, T. Cabrera, F. Ruiz-Cabello, F.Garrido, The selection of tumor variants with altered expres-sion of classical and nonclassical MHC class I molecules:implications for tumor immune escape, Cancer Immunol.Immunother. 53 (2004) 904–910.

[31] H.J. Bontkes, D. Kramer, J.J. Ruizendaal, E.W. Kueter, V.F.van Tendeloo, C.J. Meijer, E. Hooijberg, Dendritic cellstransfected with interleukin-12 and tumor-associated antigenmessenger RNA induce high avidity cytotoxic T cells, GeneTher. 14 (2007) 366–375.

[32] M.R. Betts, J.M. Brenchley, D.A. Price, S.C. De Rosa, D.C.Douek, M. Roederer, R.A. Koup, Sensitive and viable identifica-tion of antigen-specific CD8+ Tcells by a flow cytometric assayfor degranulation, J. Immunol. Methods 281 (2003) 65–78.

[33] H.J. van der Vliet, J.W. Molling, N. Nishi, A.J. Masterson, W.Kolgen, S.A. Porcelli, A.J. van den Eertwegh, B.M. von

154 M. Moreno et al.

Blomberg, H.M. Pinedo, G. Giaccone, R.J. Scheper, Polarizationof Valpha24+ Vbeta11+ natural killer T cells of healthyvolunteers and cancer patients using alpha-galactosylcera-mide-loaded and environmentally instructed dendritic cells,Cancer Res. 63 (2003) 4101–4106.

[34] A. Nicol, M. Nieda, Y. Koezuka, S. Porcelli, K. Suzuki, K.Tadokoro, S. Durrant, T. Juji, Human invariant valpha24+natural killer T cells activated by alpha-galactosylceramide(KRN7000) have cytotoxic anti-tumour activity throughmechanisms distinct from T cells and natural killer cells,Immunology 99 (2000) 229–234.

[35] G. Giaccone, C.J. Punt, Y. Ando, R. Ruijter, N. Nishi, M.Peters, B.M. von Blomberg, R.J. Scheper, H.J. van der Vliet,A.J. van den Eertwegh, M. Roelvink, J. Beijnen, H. Zwierzina,H.M. Pinedo, A phase I study of the natural killer T-cell ligandalpha-galactosylceramide (KRN7000) in patients with solidtumors, Clin. Cancer Res. 8 (2002) 3702–3709.

[36] H. Lin, M. Nieda, V. Rozenkov, A.J. Nicol, Analysis of the effectof different NKT cell subpopulations on the activation of CD4

and CD8 T cells, NK cells, and B cells, Exp. Hematol. 34 (2006)289–295.

[37] L.S. Metelitsa, O.V. Naidenko, A. Kant, H.W. Wu, M.J. Loza, B.Perussia, M. Kronenberg, R.C. Seeger, Human NKTcells mediateantitumor cytotoxicity directly by recognizing target cell CD1dwith bound ligand or indirectly by producing IL-2 to activate NKcells, J. Immunol. 167 (2001) 3114–3122.

[38] J. Hanna, T. Gonen-Gross, J. Fitchett, T. Rowe, M. Daniels, T.I.Arnon, R. Gazit, A. Joseph, K.W. Schjetne, A. Steinle, A.Porgador, D. Mevorach, D. Goldman-Wohl, S. Yagel, M.J.LaBarre, J.H. Buckner, O. Mandelboim, Novel APC-like proper-ties of human NK cells directly regulate T cell activation,J. Clin. Invest 114 (2004) 1612–1623.

[39] S. Motohashi, A. Ishikawa, E. Ishikawa, M. Otsuji, T. Iizasa, H.Hanaoka, N. Shimizu, S. Horiguchi, Y. Okamoto, S. Fujii, M.Taniguchi, T. Fujisawa, T. Nakayama, A phase I study of in vitroexpanded natural killer T cells in patients with advanced andrecurrent non-small cell lung cancer, Clin. Cancer Res. 12(2006) 6079–6086.