Immunology Letters 93 (2004) 205–210

NK cell depletion diminish tumour-specificB cell responses

Markus Jensena,c,∗, Samir Tawadrosc, Hans-Harald Sedlacekb,Joachim L. Schultzec, Frank Bertholda

a Department of Pediatric Oncology and Hematology, University of Cologne,Joseph-Stelzmann Strasse 9, 50924, Cologne, Germany

b Medinnova GmbH, Marburg, Germanyc Molecular Tumour Biology and Tumour Immunology, University of Cologne, Cologne, Germany

Received 11 February 2004; received in revised form 24 March 2004; accepted 24 March 2004

Available online 16 April 2004

Abstract

Natural killer (NK) cells can exercise immediate cytotoxicity against malignant cells and thus far modulate the development of tumourdirected T cell immunity. To investigate the impact of NK cells on the development of tumour directed B cell immunity mice were immunisedwith IMR5-75 human neuroblastoma cells with or without prior in vivo NK cell depletion. Flow cytometry analyses gave evidence for animpaired IgG response against the cells immunised with. Dissection of Th1 (IgG2a) and Th2 (IgG1) oriented B cell responses revealed Th1responses as primarily affected, while Th2 oriented B cell responses as measured by flow cytometry and GD2 ganglioside-specific ELISAwere enforced. The data reveal an unexpected impact of NK cells on the development of tumour directed B cell responses. Consequently, NKcell function has also to be taken into account when developing B cell-based cancer immunotherapy.© 2004 Elsevier B.V. All rights reserved.

Keywords:NK cells; B cells; Innate immunity; Adaptive immunity; Immunotherapy; Cancer

1. Introduction

Innate immunity builds the first defence line againstpathogens or aberrant cells. Innate immune cells as naturalkiller (NK) cells or antigen presenting cells (APCs) actimmediately without prior sensitisation after specialisedcell surface receptors have recognised certain character-istic molecular patterns (pathogen associated molecularpatterns, PAMPs) shared by various pathogens[1,2]. Lossof HLA class I molecules from the cell surface, which isalso observed during malignant transformation, is anotherindependent signal provoking NK cell activity[3]. In con-trast, adaptive immunoresponses by T and B cells needclone-specific receptor mediated signals leading to clonalexpansion followed by efficient effector function usuallyreached at the peak of the immune response many hours

∗ Corresponding author. Tel.:+49-221-478-3663;fax: +49-221-478-5912.

E-mail address:jensen@uni-koeln.de (M. Jensen).

to days later. Once initiated adaptive immunity is highlyspecific and builds up a long persisting memory that canact fast and efficient upon antigen re-challenge[4]. Innateand adaptive immunity complement one another building acomplex defence system and are closely linked together ondifferent functional levels[1,5].

NK cells have been recently shown to be important regu-lators of tumour-specific T cell responses[6]. NK cell activ-ity is essential for development of tumour-specific cytotoxicand T helper type 1 responses[6]. In contrast, depletion ofNK cells results in abrogation of tumour-specific effectorCTL and memory T cell reaction[7]. Activated NK cells aremajor sources of the Th1 polarising cytokine IFN-�, whichis thought be an important link between NK cell activityand generation of tumour-specific CTL[7]. Interestingly ac-tivated T cells themselves can limit NK cell activity[8], e.g.by secretion of IL-21[9], suggesting a regulatory loop be-tween cytotoxic T and cytotoxic NK cell activity.

Based on these findings, we were interested to investi-gate the impact of NK cells on the development of tumour

0165-2478/$ – see front matter © 2004 Elsevier B.V. All rights reserved.doi:10.1016/j.imlet.2004.03.015

206 M. Jensen et al. / Immunology Letters 93 (2004) 205–210

directed B cell responses. To address this question, micewere immunised with tumour cells with or without in vivoNK cell depletion to induce anti-tumour antibody response.As assessed by the induction of specific antibody titers wefound evidence for a diminished B cell response in NK celldepleted animals. Dissection of Th1 (IgG2a) and Th2 (IgG1)oriented B cell answers revealed the Th1 track as primarilyaffected, while Th2 oriented B cell responses in contrastwere enforced. Our data suggest that NK cells—in addi-tion to their regulatory role on CTL induction—have alsoa steering function for the development of tumour-specificantibody responses.

2. Materials and methods

2.1. Growth of tumour xenografts in SCID mice

For the experiments, 6–8-week-old severe combinedimmunodeficiency mice (SCID) having no T- and B- lym-phocytes but an intact NK cell system were used. Ten micewere treated with 20�l anti-asialo-GM1 (anti-ASGM1)rabbit antiserum (WAKO Chemicals, Düsseldorf, Germany,dose recommended by manufacturer) i.v. 6–12 h before s.c.tumour cell challenge and again on day 4 after tumour cellinjection. Twelve mice from the control group were not NKdepleted. Mice were s.c. injected with(3.45–4.0) × 107

cells from the human neuroblastoma cell line IMR5-75[10] to induce tumour xenografts. Macroscopic tumourgrowth was documented during 8 weeks after tumour cellinjection.

2.2. Immunisation protocol

Six- to eight-week-old C57Bl/6 mice (Taconic M&B)were immunised with 2× 106 IMR5-75 human neurob-lastoma cells (26 Gy irradiated to block mitosis) intraperi-toneally once weekly for 2 or 3 weeks, respectively. Micewere additionally i.v. boostered with 1×106 IMR5-75 cells3 days before dissection. Animals from the experimentalgroup were NK cell depleted by i.v. injection of 20�lasialo-GM1-antiserum (ASGM-1, WAKO Chemicals, Os-aka, Japan) starting on day-1. Anti-ASGM-1 treatment wasrepeated every 4–5 days[7]. Control mice were injectedwith phosphate buffered saline (PBS). F004/C57 hybridmice (own breeding colonies) were generated by intercross-ing wildtype C57Bl/6 mice with transgenic mice (strainF004) producing human IgG1 heavy chain and humankappa light chain immunoglobulin constant regions insteadof murine wildtype chains[11]. F004/C57Bl/6 hybrid miceco-express murine and human IgG1 and kappa constantchains. Hybrid mice were selected as a control strain for theELISA experiments due to availability. Hybrid mice wereimmunised within the 2 weeks immunisation scheme (+/−anti-ASGM1) using cells from the GD2 expressing humansmall cell lung cancer cell line H69 (26 Gy irradiated).

2.3. Cytotoxicity assay

To control NK cell depletion after immunisation spleencells from sacrificed mice were tested for NK cell ac-tivity in a cytotoxicity assay using NK sensitive YAC-1cells as a target. YAC-1 cells were Europium labelled asdescribed elsewhere[12] and extensively washed to re-move extracellular Europium traces. The E/T-ratio wasranging from 3:1 until 200:1. After 3 h incubation at37◦C and 5% CO2 the tests were read out by transferring20�l supernatant of each test to a flat-bottom microtitreplate prepared with 200�l/well enhancement solutionTM

(Perkin Elmer, Norton, OH). Time-resolved fluorescencewas measured in a fluorometer (Delfia 1232, Wallac, Turku,Finland). All tests were run in triplicates. Specific eu-ropium release was calculated as(experimental release−spontaneous release)/(maximum release−spontaneous rele-ase) × 100 [%].

2.4. Determination of antibody response

Serum samples were taken pre- and post-immunisationto determine antibody response against cells immunisedwith. Serum samples were stepwise diluted (range 1:10until 1:5120) in phosphate buffered saline and IMR5-75cells incubated for 15 min. After washing with PBS an-tibodies were detected with goat-anti-mouse IgG-FITC(Southern Biotechnology, Malvern, PA) secondary antibod-ies for determining global IgG response or goat-anti-mouseIgG1 and IgG2a-FITC antibodies (Southern Biotech-nology) for separation of Th1 and Th2 oriented anti-body responses. Normal mouse serum (own production)was used as a negative control. Dead cells were ex-cluded using 7-amino-actinomycin-D (7-AAD, Sigma,Deisenhofen, Germany). Cells were measured in a FAC-SCalibur Flow cytometer (Becton Dickinson, Heidel-berg, Germany). Tests revealing at least 50% of cellspositive as compared to the negative control were con-sidered positive. Data analyses and determination ofGeoMean values were done using WinMDI v.2.8. software[13].

Antibody responses against GD2 ganglioside was deter-mined by ELISA assay. Purified GD2 (Calbiochem, Bad So-den, Germany) was immobilised on a 96-well flat bottomplate (Saarstedt, Nümbrecht, Germany) at a concentration of1�g/ml in PBS (phosphate buffered saline, 50�l/well) bydrying the plates for 3 h at 60◦C. Plates were blocked using20% foetal calf serum in PBS and washed three times. Serumsamples were diluted 1:20 until 1:2560 using PBS plus1% FCS. BW704[14] hybridoma cell culture supernatantsand purified 14.G2a[15] antibodies were used as positivecontrols. Bound anti-GD2 antibodies were detected by sec-ondary goat-anti-mouse IgG1-HRPO and IgG2a-HRPO an-tibodies (1�g/ml; Southern Biotechnology) in PBS (+1%FCS). OPD (o-phenylenediamine, Sigma-Aldrich, Deisen-hofen) was used as a substrate. The reaction was stopped by

M. Jensen et al. / Immunology Letters 93 (2004) 205–210 207

adding 50�l/well 1.3 M H2SO4. Plates were read at 490 nmwavelength in a plate photometer.

2.5. Statistical analyses

Statistical significance was calculated using Student’st-test.

3. Results

3.1. NK cells reduce outgrowth of IMR5-75 tumourxenografts in SCID mice

SCID mice do not own a functional T or B cell sys-tem and are commonly used as a model to induce tumourxenografts since they poorly can defend themselves fromgrowth of foreign tumour cells. To explore weather murineNK cells, still present in SCID mice, can recognise and spon-taneously attack IMR5-75 cells, 22 SCID mice were injecteds.c. with (3.45–4.0) × 107 IMR5-75 cells to induce sub-cutaneous tumour growth. Ten mice were pre-treated withASGM1 anti-serum to induce NK cell depletion, 12 micewere not pre-treated. Only 4/12 mice in the control grouphad macroscopic tumour growth approximately 4–6 weeksafter injection of cells. Ten out of 10 mice in the NK de-pleted group showed tumour growth as early as 2 weeks aftertumour cell injection (P < 0.000; Fig. 1). The data clearlydemonstrate that innate immunity as present in SCID micecan spontaneously becoming active against IMR5-75 cellsand prevent mice from tumour growth. NK cell depletionefficiently breaks down this self defense indicating sponta-neous murine NK cell action as an early event in immun-odefense of IMR5-75 cells that may result in control of Bcell responses.

0

20

40

60

80

100

controln = 12

+ASGM1n = 10

P < 0.000

% S

CID

mic

e w

ith tu

mor

gro

wth

Fig. 1. NK cell depletion in SCID mice using anti-ASGM1 antiserumincreases the probability of tumour growth after subcutaneous injectionof IMR5-75 cells: SCID mice were s.c. injected with(3.45–4.0) × 107

IMR5-75 human neuroblastoma cells. Ten mice had prior i.v. injec-tions with anti-ASGM1 antiserum, while 12 control mice received noanti-ASGM1 pre-treatment. Macroscopic tumour growth could be docu-mented in the majority of mice approximately 2 weeks after tumour cellinjection. Mice were observed for at least 8 weeks before no tumourgrowth was documented;P < 0.005.

0

10

20

30

0 50 100 150 2

% s

peci

fic ly

sis

P = 0.002 P = 0.007 P = 0.003

P = 0.003

00

E/T-ratio

Fig. 2. Europium release cytotoxicity assay: lysis of NK sensitive YAC-1cells by spleen cells isolated from C57Bl/6 mice. Filled circles: mousetreated with anti-ASGM-1 antiserum; open circles: control mouse treatedwith PBS; error bars: standard deviation; one representative experimentis shown;P-values are indicated in the graph.

3.2. Delayed IgG antibody response to IMR5-75 cells inC57Bl/6 mice

First, in vivo NK cell depletion was documented for eachmouse at the end of the immunisation period revealing asignificantly reduced NK activity in anti-ASGM1 treatedmice as measured by lysis of NK sensitive YAC-1 cells usingfreshly isolated splenocytes as effector cells (Fig. 2).

Prior vaccination with IMR5-75 tumour cells 6/8 miceshowed an anti-IMR5-75 IgG antibody titre of<1:10,while the remaining two mice exhibited titres slightly abovebackground (1:10, resp. 1:40). When analysing antibodytitres after vaccination, all control mice treated with PBSshowed significantly increased IgG antibody titres (range1:160–1:320). In contrast, in 3/4 NK cell depleted mice noincrease in IgG antibody titre was detected and the fourth

Table 1Antibody response against IMR5-75 neuroblastoma cells

Mouseno.

NK celldepletion

Pre-immunisationtitre

Post-immunisationtitrea

1 PBS <1:10 1:3202 PBS <1:10 1:3203 PBS 1:10 1:1604 PBS <1:10 1:1605 Anti-ASGM1 <1:10 <1:106 Anti-ASGM1 <1:10 <1:107 Anti-ASGM1 <1:10 1:408 Anti-ASGM1 1:40 1:40

Serum samples from mice immunised with IMR5-75 cells were diluted(stepwise 1:10 until 1:5120) prior incubation with IMR5-75 cells. Cellsurface bound antibodies were detected using goat-anti-mouse IgG-FITCsecondary antibodies. Cells were measured in a FACSCalibur® (BectonDickinson) flow cytometer. If >50% of cells were positive as comparedwith a negative control (normal mouse serum) a test was classified aspositive.

a For all the valuesP = 0.003.

208 M. Jensen et al. / Immunology Letters 93 (2004) 205–210

mouse developed a significantly lower titre (1:40) comparedto control mice. IgG antibody titres after vaccination dif-fered significantly between control and experimental group(P = 0.003, Student’st-test) (Table 1).

In four additional mice we studied IgG antibody titres af-ter a 3-week period. Interestingly, IgG titres in control micedid not further increase (1:320), while the titres in the NKdepleted mice were 1:80, respectively, 1:320 suggesting adelayed onset of the antibody response in the absence ofNK cells (pre-immunisation titre in all mice were<1:10).Alternatively, the different kinetics might be due to the in-duction of different classes of antibodies. We therefore as-sessed overall titres of anti-IMR5-75-specific antibodies ofthe IgG1 type (Th2 response) and of the IgG2a type (Th1response).

3.3. Th2-biased antibody response after NK cell depletion

In mice immunised for 2 weeks, there were only verylow level IgG1 responses in both control and NK depleted

0

5

10

15

20

25

1 10 100 1000

0

5

10

15

20

25

30

1 10 100 1000

0

5

10

15

20

25

1 10 100 1000

0

0.1

0.2

0.3

0.4

10 100 1000

0

0.1

0.2

0.3

0.4

10 100 1000

0

5

10

15

20

25

30

1 10 100 1000

IgG2a IgG1

OD

490

IgG2a IgG1

Geo

Mea

n

IgG2a IgG1

(A)

(B)

(C)

titer x 10-1

Fig. 3. Dissection of Th1 (IgG2a) and Th2 (IgG1) weighted B cell responses: (A) mice immunised with IMR5-75 cells for 2 weeks: IgG1 and IgG2aserum titre as measured by flow cytometry using IMR5-75 cells and secondary goat-anti-mouse IgG1 and IgG2a-FITC antibodies; (B) same as (A)but mice immunised for 3 weeks with IMR5-75 cells; (C) IgG1 and IgG2a antibody titre against GD2 ganglioside as measured by ELISA assayusing immobilised GD2 and secondary goat-anti-mouse IgG1- and IgG2a-HRPO antibodies.Y-axis (A): GeoMean fluorescence intensity;Y-axis (B):OD490;X-axis: serum titre× 10−1; open boxes: serum samples pre-immunisation (anti-ASGM-1 treated and control mice); open circles: serum samplespost-immunisation:control mice; closed circles: serum samples post-immunsation:anti-ASGM1 treated mice.

mice (Fig. 3A). In contrast, only control mice respondedwith a significant IgG2A antibody response to IMR5-75,suggesting a Th1-biased immune response against tumourimmunisation (Fig. 3A).

The difference in IgG2a titres was even further pro-nounced in mice immunised for 3 weeks (Fig. 3B).Most surprising, NK cell depleted mice showed higheranti-IMR5-75 IgG1 titres than control mice suggesting thatthe antibody response observed in these mice is mainly dueto a Th2-biased immune response. These data indicate thatNK cells are not only regulating the global IgG antibody re-sponse against tumour cells in this experimental model butalso modulate the isotype usage of the overall anti-tumourimmune response.

3.4. Induction of anti-GD2-specific antibodiesof the IgG1 isotype

To further elucidate the antibody response to tumour cellimmunisation in relation to NK cells we chose the GD2

M. Jensen et al. / Immunology Letters 93 (2004) 205–210 209

antigen as a model. GD2 ganglioside is a tumour associ-ated antigen highly overexpressed on IMR5-75 cells[16].Clinically, GD2 has importance in immunodiagnostics andspecific antibody therapy in neuroblastoma[16].

By ELISA we could even detect anti-GD2 antibodies inNK depleted mice immunised only for 2 weeks when overallantibody response to IMR5-75 cells was still very low. Iso-type analysis clearly revealed that these antibodies were alsoof IgG1 type, while no anti-GD2 antibodies of the IgG2a iso-type were detectable in these mice. In contrast, neither IgG1nor IgG2a anti-GD2 antibodies were detectable in controlmice (Fig. 3C). These findings were confirmed in a secondmodel. F004/C57 hybrid mice were immunised with an ir-radiated GD2 expressing human small cell lung carcinomacell line H69 following the 2 weeks immunisation schema.Again, an antibody response to GD2 was only detectable inNK cell depleted mice and anti-GD2 antibodies were alsoof the IgG1 isotype.

These data further corroborate the impact of NK cells oninduction and differentiation of the B cell response againstcellular targets in this experimental model.

4. Discussion

Here we have revealed a delayed onset as well as aTh2-biased antibody response to tumour cell immunisationin NK depleted mice. The reduction of antigen-specificimmunity in the absence of NK cells as described by Geld-hof et al. for CTL responses[7] and described here for Bcell responses is apparently not restricted to tumour anti-gens. Suppression of antibody responses associated withNK cell depletion was also reported by Korsgren et al.[17] in a mouse model of allergen induced eosinophilicairway inflammation and by Shi et al.[18] who reportedreduced titres of (pathologic) anti-acetylcholine recep-tor antibodies in an experimentally induced autoimmunemyasthenia gravis (EAMG) mouse model after NK celldepletion.

NK cells represent a key element of the innate immunesystem. Although it is generally accepted that the innateimmune system can regulate secondary adaptive immuneresponses, little is known about the mechanisms that linkNK cell function to antigen-specific T and B cell responses.According to our data and the data in murine autoimmunemodels NK cells might play a much more important rolein regulating antigen-specific B cell responses than so faranticipated.

The SCID mouse experiments (Fig. 1) give evidencefor spontaneous murine NK cell activity against humanIMR5-75 cells. Activated NK cells are a major source of theTh1 polarising cytokine IFN-� that might link Th1 orientedantibody responses with NK cell activity[7]. Another po-tential link was described by Gao et al. demonstrating NKcells programming B cells to preferentially switch to theIgG2a isotype via an IFN-� independent mechanism acting

by B cell/NK cell membrane contact[19]. Geldhof et al.demonstrated a lack of Th1 associated cytokines correlatingwith the presence of Th2 cytokines in spleen cell culturestaken from NK deficient (anti-ASGM1 treated) tumourbearing AKR mice, which is compatible with our findings ofan impaired IgG2a response, but enhanced IgG1 response.It is noteworthy that C57Bl/6 mice tend to develop Th1 ori-ented adaptive immunoresponses[7] therefore the stronglyimpaired Th1 oriented B cell response after NK depletionis an even more remarkable observation. Current and futurestudies will have to address whether NK cell depletion insyngeneic tumour models in fully competent mice will havecomparable impact on anti-tumour B cell responses.

NK cell depletion by anti-ASGM1 has been used to de-plete NK cells in many experimental systems[6,7,20,21].More recently it has been shown that anti-ASGM1 doesnot only bind NK cells but also some monocytes andmacrophages[22]. However, the impact of these findingson experiments using anti-ASGM1 to deplete NK cells isnot fully understood since it was also demonstrated thatAPC function of adherent monocytic cells was shown tobe intact after antibody treatment[7]. Moreover, B cells donot express ASGM1 and appeared not to be impaired byanti-ASGM1 treatment[22].

The impact of NK cells on B cell responses to tumour cellsmight also be dependent on the genetic background of themice strain evaluated. Previous reports have demonstratedthat NK cell lysis to surrogate targets such as YAC-1 cellsdiffer between different strains. C57Bl/6 mice used here aresimilar effective in NK cell lysis when compared to AKRmice [7,18]. However, NK cells from other strains such asC57Bl/6CrSle seem to be more efficient in lysing surrogatetarget cells[23]. Whether these differences in lytic activ-ity also correlate with the capacity of NK cells to influenceanti-tumour B cell responses is an area of further exploita-tion.

A lack of functional active NK cells in cancer patientsmight be associated with severe impairment not only ofanti-tumour CTL responses but also B cell responses. SinceTh1 oriented B cell responses leads to the production of anti-bodies with higher affinity for Fc-receptors (in humans IgG1and IgG3) than Th2 weighted responses[4] impaired tumourdirected antibody dependent cellular cytotoxicity (ADCC)will be a consequence. We therefore propose that more at-tention needs to be paid on NK cells and their function par-ticularly after primary radiotherapy or chemotherapy thatmight negatively influence success of an immunotherapeuticintervention.

Acknowledgements

We thank Mrs. Sandra Schmitz for expert technical assis-tance. JLS is supported by the Sofja-Kovalevskaja Award ofthe Alexander von Humboldt Foundation. FB is supportedby the German Cancer Aid (Deutsche Krebshilfe).

210 M. Jensen et al. / Immunology Letters 93 (2004) 205–210

References

[1] Trinchieri G. Interleukin-12 and the regulation of innate resistanceand adaptive immunity. Nat Rev Immunol 2003;3:133–46.

[2] Barton GM, Medzhitov R. Control of adaptive immune responsesby Toll-like receptors. Curr Opin Immunol 2002;14:380–3.

[3] Moretta A, Bottino C, Vitale M, Pende D, Cantoni C, Mingari MC,et al. Activating receptors and coreceptors involved in human natu-ral killer cell-mediated cytolysis. Annu Rev Immunol 2001;19:197–223.

[4] Janeway C, Travers P, Walport M, Shlomchik M. Immunobiology:the immune system in health and disease. Edinburgh, New York:Churchill Livingstone Garland Publ., 2001.

[5] Pulendran B, Palucka K, Banchereau J. Sensing pathogens and tuningimmune responses. Science 2001;293:253–6.

[6] Kelly JM, Darcy PK, Markby JL, Godfrey DI, Takeda K, Yagita H, etal. Induction of tumor-specific T cell memory by NK cell-mediatedtumor rejection. Nat Immunol 2002;3:83–90.

[7] Geldhof AB, Van Ginderachter JA, Liu Y, Noel W, Raes G, DeBaetselier P. Antagonistic effect of NK cells on alternatively activatedmonocytes: a contribution of NK cells to CTL generation. Blood2002;100:4049–58.

[8] Kos FJ. Regulation of adaptive immunity by natural killer cells.Immunol Res 1998;17:303–12.

[9] Kasaian MT, Whitters MJ, Carter LL, Lowe LD, Jussif JM, Deng B,et al. IL-21 limits NK cell responses and promotes antigen-specificT cell activation: a mediator of the transition from innate to adaptiveimmunity. Immunity 2002;16:559–69.

[10] Fichtner I, Lemm M, Becker M, Berthold F. Effects of amifostine(WR-2721, ethyol) on tumor growth and pharmacology of cytotoxicdrugs in human xenotransplanted neuroblastomas. Anticancer Drugs1997;8:174–81.

[11] Zou Y, Muller W, Gu H, Rajewsky K. Cre-loxP-mediated genereplacement: a mouse strain producing humanized antibodies. CurrBiol 1994;4:1099–103.

[12] Hombach A, Tillmann T, Jensen M, Heuser C, Sircar R, Diehl V, et al.Specific activation of resting T cells against tumour cells by bispecificantibodies and CD28-mediated costimulation is accompanied by Th1

differentiation and recruitment of MHC-independent cytotoxicity.Clin Exp Immunol 1997;108:352–7.

[13] Trotter J. WinMDI v.2.8, 2000. http://www.fz-borstel.de/ZentrEinrichtungen/flowcytes/software/software.htm.

[14] Ollert MW, David K, Schmitt C, Hauenschild A, Bredehorst R,Erttmann R, et al. Normal human serum contains a natural IgMantibody cytotoxic for human neuroblastoma cells. Proc Natl AcadSci USA 1996;93:4498–503.

[15] Murray JL, Cunningham JE, Brewer H, Mujoo K, Zukiwski AA,Podoloff DA, Kasi LP, Bhadkamkar V, Fritsche HA, Benjamin RS.Phase I trial of murine monoclonal antibody 14G2a administeredby prolonged intravenous infusion in patients with neuroectodermaltumors. J Clin Oncol 1994;12:184–93.

[16] Reuland P, Geiger L, Thelen MH, Handgretinger R, Haase B,Muller-Schauenburg W, Niethammer D, Bares R. Follow-up in neu-roblastoma: comparison of metaiodobenzylguanidine and a chimericanti-GD2 antibody for detection of tumor relapse and therapy re-sponse. J Pediatr Haematol Oncol 2001;23:437–42.

[17] Korsgren M, Persson CG, Sundler F, Bjerke T, Hansson T, Cham-bers BJ, et al. Natural killer cells determine development ofallergen-induced eosinophilic airway inflammation in mice. J ExpMed 1999;189:553–62.

[18] Shi FD, Wang HB, Li H, Hong S, Taniguchi M, Link H, et al. Naturalkiller cells determine the outcome of B cell-mediated autoimmunity.Nat Immunol 2000;1:245–51.

[19] Gao N, Dang T, Yuan D. IFN-gamma-dependent and -independentinitiation of switch recombination by NK cells. J Immunol 2001;167:2011–8.

[20] Strbo N, Oizumi S, Sotosek-Tokmadzic V, Podack ER. Perforin isrequired for innate and adaptive immunity induced by heat shockprotein gp96. Immunity 2003;18:381–90.

[21] Liu ZX, Govindarajan S, Okamoto S, Dennert G. NK cells causeliver injury and facilitate the induction of T cell-mediated immunityto a viral liver infection. J Immunol 2000;164:6480–6.

[22] ASGM1, WAKO BioProducts, 2003, p. 6.[23] Kurita H, Kamei T, Nagaoka K, Yoshida T, Taniwaki H, Ono Y, et al.

Natural killer cell activity in mice after intraperitoneal administrationof beryylium chloride. J Occup Health 2001;43:284–6.