Scand. J. Immunot. 35, 279-287, 1992

TNF Impairs In Vivo and In Vitro Natural Killer (NK) Susceptibility of B16 Melanoma Cells G. PALMIERI*§. S. MORRONE*, P.-L. LOLLINIfJ, C. DE GIOVANNIf, G, NICOLETTItt, P. NANNIt, L. FRATI* & A. SANTONI* Interuniversity Center for Cancer Research (CIRC); *Department of Experimental Medicine, University of Rome, tCanccrology Institute. Universily of Bologna, Bologna, and Jl.S.T. Genova, Sezione Aggregata di Bologna, Bologna, Italy; §Institute for Biomedical Technologies-C.N.R., Rome, Italy Palmieri G, Morrone S, Lollini P-L, De Gioviinni C, Nicalctti G, Nanni P, Frati L, Satitoni A. TNF Impairs In Vivo and in Vitro Natural Killer (NK) Susceptibility of BI6 Melanoma Cells Scand J Immunol 1992:35:279-87 Tamour necrosis factor ot {TNF) is a multipotent cytokine which affects many biological properties of both normal and neoplastic cells. Mere we show ihat treatment with TNF reduces B16-A melanoma cell susceptibility to normal and in vivo- and in vitro-activated NK cellmediated killing. This resistance is associated with an enhanccmeni of B16-A metastatic potential in normal syngeneic mice, but not in anii-asialo GM [-treated animals, further supporting the NK dependence of TNF-induced enhancement of metastatic abitiiy. A significant increase of MHC class I expression on B16-A murine melanoma cells i.s observed afler TNF treatment. In all these effects TNF Interacts positively with interferon 7 (IFN -;). Taken together, these results indicate that TNF treatment negatively alTecis (he susceptibility of B16-A murine melanoma to NK effectors in vivo and in vitro. This decreased susceptibillly may be related, at least in part, to enhanced expression of MHC class I antigens on tumour cells. G. Palmieri. Dipartimento di Medicina Sperimentate. Ill Cattedra di Patotogia Generate. Viate Regina Elena 324. 00161 Rome. Italy

Tumorigenesis is a multistep process, and many factors regulating tumour host relationship have not been clarified yet [I. 2]. Tumour take or rejection represents the final outcome hetween tumour cell growth potential and the cellular and humoral immune responses ofthe host [3]. Tumour necrosis facior y (TNF) is an important mediator of inflammatory responses [4, 5], mainly produced and secreted by maerophages [4], T lymphocytes [6] and natural killer (NK) cells [7] in response to a variety of stimuli. This cytokine exerts pleiotropic effects on diverse tissues; in particular TNF is endowed with antitumour aetivity, causes the haemorrhagic necrosis of many tumours in vivo, and is cytotoxic and cytostatic against in vitro established tumour cell lines. TNF possesses profound immunomodulatory capacity, in that it enhances T and B lymphocyte proliferation [8, 9], partici-

pates in the generation of CTL and lymphokineactivated killer (LAK) activity [10, 11] and potentiates macrophage activity [12]. Moreover, il is a potent inducer of MHC antigen expression on both normal and tumour cells [13, 14]. The ability of TNF to activate different effector mechanisms and to interact with other cytokines, coupled with its capacity to directly and indirectly aflect tumour growth, have led to its use in clinical trials either alone or in combination with other cytokines in the treatment of some tumours [15, 16]. NK cells represent a heterogeneous lymphoid cell population involved in the control of tumour progression and metastasis, exerting spontaneous cytotoxic activity against tumour and virusinfected cells [17 21]. NK activity is promptly enhanced by several cytokines, i.e. 1 FNs [22], IL-2 [23] and lL-6 [24]. NK cells constitute the main effector cell popuiation responsible for LAK 279

280

G. Palmieri et al.

activity [25] and also produce and secrete TNF. a possible mediator of cytotoxic activity, in response to target stimulation [7], Tumour susceptibility to NK kilting can be modulated by genetic variations or soluble factors. In this regard, IFN -/ increases the aggressivity of tumour eells of different histologicitl origin [26-29] and decreases target susceptibility to both NK- and LAK-mediated killing [28. 3032]. In many systems, decreased NK susceptibility has been associated with up-regulation of MHC class I antigens which have been suggested to constitute (or contribute to the recognition of) an inhibitory signal in NK-target cell interaction [33]. We have analysed the effects of TNF treatment on the susceptibility of the B16-A murine melanoma cell line [34, 35] to NK-mediated eytotoxicily in vitro and on tumour metastatic potential. Moreover, we investigated whether changes in reactivity to natural effector mechanisms are associated with modulation of MHC class 1 antigen expression induced by TNF.

MATERIALS AND METHODS Mice. Male C57BL,6 and BALB/c mice were purchased from Charles River (Calco, Italy). Cell lines and treatments. The B16-A cell line derived from B16 melanoma was cultured as previously described [.^5]. Cells were treaied for 96 h at 37 C with murine rIFN }• and human rTNF (kindly provided by G. R. Adolf, Ernst-Boehringer-lnstitut. Vienna, Austria), both produced in E. coli. with specific activities of 6x 10^ U/mg and 10-20 x lO*" U/mg, respectively. The dose of TNF IO"* U/ml was chosen as the most effective after preliminary dose-response experiments (not shown), while the different doses of IFN y employed have been chosen on the basis of previous sludics [29]. Treatment with TNF and IFN •/ only slightly impaired the proliferative capacity of B16-A cells (data not shown). E.xperimentat metastasis. Male C57BL/6 mice (8 12 weeks old) were inoculated with B16-A tumour cells in the lateral tail vein. They were killed 21 days later and their lungs were removed and fixed. Colonies were scored with the aid of a dissecting microscope. Statistical analy,sis was performed using the non-parametric Wilcoxon's rank sum test. In vivo treatment with anti-asiato GMI (aGMlj antiserum. To abolish in vivo NK activity, mice received a single i.v. injection of rabbil anti-aGMl serum (Wako. Neusfeld, Germany) (0.4 ml, dilution 1:30) the day before cell inoculation [36]. This treatment completely abrogates splenic NK activity against YAC-1 tumour cells without affecting the T-dependent proliferative response to mitogens. Preparation of effector cells. Spleens from 5 7-week-

old C57BL/6 mice were excised and cell suspensions obtained by teasing through a fine steel mesh. Poly I:Cinduced spleen cells were obtained by i.p. injeclion of Poly I:C (Sigma Chemical Company. St. Louis, Mo, USA) (1 mg/kg) 36 h before killing [37]. IL-2-activated effector cells were derived from overnight culture of splenocytes (lO^'ml) in RPMI-1640 supplemented with WA: fetal calf serum (FCS), 2 mM Lglutamine, 50 lU/ml penicillin and 50 /;g/ml streptomycin, here referred to as complete medium (CM), supplemenlcd with 5 x 10"''M 2-mercaptoethanol and 1000 U/ ml human rIL-2 (Cetus Corporation, Emeryville, Calif, USA) at 37 C in 5% CO; [23, 38]. Alloantigen-specific effector cells were generated by cocuhuring lO^ml BALB/c (H-2'') responder spleen cells with 10''/ml irradiated (2000 rad) C57BL/6 (H-2'') stimulator spleen ceils for 5 days at 37 C in 5% COi. In vitro treatment with anti-aGMl plus comptement. Efi"ector cells (10 x 10^) were incubated with I ml antiaGMl serum diluted I ;40 in CM for 45 min on ice. After two washings, samples were incubated with I ml rabbit complement diluted 1:15 (Low-Tox-M, Cedarlane Laboratories, Hornby, Ontario, Canada) for 45 min at .17 C Cells were then washed Iwice, resuspended in the original volume of CM plus 10 mM HEPES (Flow Laboratories, Irvine, UK) and tested for cytotoxic activity against YAC-1 or B16-A tumour cells in a 4-h ^'Cr release assay. Cvtotoxic activity. Tumour cells (1 x 10'') were labelled with 100 ;iCi " C r (Amersham, Bucks.. UK) Tor I h a l 3 7 C.''Cr-labelledtargelcells(5x 10^ in 100/d) were plated together with serial dilutions of effectorcells (100^1) in 96-well U-bottomed mierotitre plates in CM plus 10 mM HEPFS. After 4 h at 37 C, 100 /il of supernatant was collected and released radioactivity was measured in a y counter. Per cent of cytotoxicity and SD were calculated on triplicate samples [37]. Statistical significance was assessed by the Student's t test. Cvtoftuorinietrtc

antilysis.

Cell aliquots (lO*") were

stained with 25/il of anti-H-2 K'^ (H-142-23) and antiH-2 D*" (H-I4I-3I) monoclonal antibodies (MoAbs) (Serotec, Bicester, UK) diluted 1:10 for 30 min at 4 C. After washings, samples were incubaled with fluorescein isothiocyanate (FITC)-conjugaled goal antimouse Ig (20 /il diluted 1:10) (Becton Dickinson. Mountain View, Calif., USA) and analysed with a FACS flow cytofluorimeler (Becton Dickinson). Fluorescence is expressed in logarithmic arbitrary units (a.u.). Smoothed histograms were compared by KolmorogovSmirnov statistical analysis [39, 40].

RESULTS TNF treatment of B16-A melanoma cells decrea.ses their susceptibility to NK cytotoxicity We analysed whether TNF treatment afFected B16-A susceptibility to NK cytotoxicity in vitro. BI6-A cells were treated for four days with TNF or IFN y. or a combination ofthe two cytokines.

TNF Indtices Resistance to NK Cells

B

281

C

200 1

100 1 50 1 E T ralio

25:1

FIG. 1. TN F, alone or in combination with IFN y. decreases NK susceptibility of B! 6-A tumour cells. Cells were untreated (O), treaied wilh TNF lO'* U/ml (•), IFN y 0.5 U/ml (A), IFN y 5 U/ml (A), or bolh TNF and IFN 7O.5 U/ml (•), and cytotoxicity of normal (A), poly I:C-induced (B), and lL-2-stimuIaled (C) spleen cells was measured in a 4-h "^'Cr release assay. In parallel experiments performed with alloantigen-activated spleen cells (5-day MLC) as effectors, per cent of cytotoxicity (E/T ratio: 100:1) were the following: 4.6% against BI6-A control, 3l.l"/> against TNF-trealed BI6-A.

and their sensitivity to NK killing was assayed using normal spleen cells, spleen cells from polyl:C-treated mice, and overnighl-IL-2-slimulated spleen eells as effectors (Fig, I). Treatment with TNF alone (lO** U/ml) significantly reduces susceptibility to lysis by stimulated effectors in a 4-h ^'Cr release assay (Fig. I, panels B and C). As previously reported [26. 28], B16-A cells treated with 5 U/ml IFN y become almost completely resistant to NK cell-mediated eytotoxicity. B16-A melanoma is almost completely resistant to normal spleen eytotoxicity. and then a further increase in the resistanee after TNF or IFN 7 treatment was hardly detectable (Fig. I. panel A). TNF and IFN y share many biological activities and have been shown to interact positively in many systems [41-44]. The combination ofa suboptimal dose of IFN y (0.5 U/ml) and lO"" U/ml TNF induced a resistance to poly I:C- or IL-2activated cell lysis well above the levels attained by treatment with either cytokine alone. The same pattern of susceptibility was observed in a cytotoxicity assay performed at 18 h (data not shown). To determine whether NK cells were responsible for the in vitro eytotoxic activity observed against B16-A cells, anti-aGM 1 plus complement treatment of spleen eells was performed. This treatment results, as previously described [38,45J, in the complete abrogation of cytotoxic activity against both YAC-I (prototype NK-suseeptible target) and BI6-A melanoma (data not shown).

TNF treatment of B16-A melanoma cells enhances their metastatic potentiat The primary role of natural effector mechanisms in the control of metastasis spreading has been reported by many authors [18-21]. On this basis, we analysed the effect of TN F treatment on the metastatic ability of B16-A tumour cells. Thus. C57B1/6 mice were injected iv. with B16-A tumour cells treated with TNF or IFN 7 in vitro for four days; the number of pulmonary metastases was evaluated after 21 days. A significant increase in the number of experimental lung metastasis was observed in mice injected with TN F (10" U/ml)-treated B16-A cells as compared with untreated tumour cells (Table I), which display low metastatic potential in a syngeneic immunocompetent host [29]. In line with previous studies, we found that treatment with IFN y is also effective in increasing the number of metastases of BI6 tumour cells [26, 28. 29]. Combined treatment with TNF and IFN y (5 U/ml) results in a further enhancement of the metastatic ability of BI6-A melanoma as compared with that of eells treated with either eytokine alone, Anti-aGMl treatment in vivo abrogates the enhancement ofTNF-treatedB16-A cells' metu.static ability The analysis of the metastatic potential of

282

C. Palmieri ct al. TABLE I. TNF alone or in combination wilh IFN y enhanees BI6-A metastatic ability Lung colonies*

Treatmenlt

Incidence

Control TNF lO-* U ml IFN-;5U..ml TNF lO'' U/ml -t-IFN y 5 U/ml

Median 0.5

5/10 10/10 10/10 10/10

Range 0-7 4 56 5-51 55 176

•C57BL/6 mice received 10' BI6-A cells i.v. tCell Irealments were performed at 37 C for % h. ]: Significantly higher ihaii control, / ' < 0.001 using Wilcoxon's non-parametric test. Data from one representative experiment oul of three.

TNF-treated B16-A tumour cells in anti-aGMItreated mice provides more direct evidence for the involvement of NK effectors in the mechanisms of host resistance, in that treatment with antiaGMl causes a virtually selective impairment of NK activity while sparing T-cell-mediated functions [36]. As expected, in vivo treatment with anti-aGM I antiserum enhances the number of metastases induced by untreated tumour cells: depiction of NK activity results in a similar number of metaslases induced by TNF-treated B16-A as compared with control cells, indicating that impaired NK activity is primarily responsible for this difference (Table II). Results obtained with IFN )'-(reated cells are consistent wilh previous reports showing that NK-ccll-mcdiated mechanisms are involved in the impaired resistance of the host to IFN-treated B16 melanoma. The

enhanced metastatic potential due to the combined treatment with TNF plus IFN 7 is also abrogated by anii-aGM I treatment, TNF treatment enhances MHC cla.ss I e.xprcssion on B16-A melanoma cells TNF has been shown to modulate the expression of histocompatibitity antigens on several cell types[13, 14]. Up-regulalionof MHCantigcn expression on target cell has been shown to decrease susceplibiliiy lo NK lysis in many systems [46-49]. We have investigated whether TNF treatment affects MHC class I antigen expression on B16-A melanoma cells, which display moderate surface levels ofthese antigens [35],Thus.cellswere treated in vitro with TNF for four days and ihe expression of H-2K and H-2D was ihen evaluated by immunolluorescence and

TABLE II. Experimental lung metastasis of B16-A cells in NK-depressed mice* Mice treated witht

PBS Cell treatmentf Control T N F a 10* U/ml I F N - / 5 U/ml TNFc( I O * U / m i + I F N - / 5 U/ml

Anti-aGM 1

Incidence

Median

Range

Incidence

Median

Range

2/7 8/8

0 lOfj

8/8 8/8

57*1

0 9 3 29 I -59 26 82

6/6 7/7 7/7 8/8

176 181 173 162

149-222 128 269 141-253 85-205

I3§

* Mice were injected with 2.5 x lO"* BI6-A cells. t Mice received a single i.v. injection of PBS o r a n l i - a G M I antiserum (0.4 ml, dilution 1:30) the day before cell injection. JCell treatments were performed at 37 C for lour days. § Significantly higher than control (P