Int. J. Cancer: 50,659-664 (1992) 0 1992 Wiley-Liss, Inc.
Publication of the International Union Against Cancer Publicationde I’Union Internationale Contre le Cancer
RELATIONSHIP BETWEEN SENSITIVITY TO NATURAL KILLER CELLS AND MHC CLASS-I ANTIGEN EXPRESSION IN COLON CARCINOMA CELL LINES HervC M. BLOTTIERE’,Rahima ZENNADI, Cedric BURG,Jean-Yves DOUILLARD, Khaled MEFLAHand Jacques LE PENDU CJF INSERM 90.11, Facultt! de Mkdecine, I rue Gaston Veil, 44035 Nantes, France. The sensitivity of colorectal tumors to NK-cell-mediated cytotoxicity and their expression of major histocompatibility complex (MHC) class-I antigens were studied in an attempt to determine whether such antigens play a role in the susceptibility of colorectal tumors to NK-cell lysis. In a rat colon-carcinoma model, 2 clones differing in their sensitivity to NK-cell-mediated cytotoxicity were tested for class-I expression; it was seen that the more sensitive cells (REGb) expressed less class-I products than did the resistant cells (PROb). However, when MHC class-I antigen expression was increased by IFN-y treatment, no change in NK-cell lysis was found with the PROb cells, while an increase in cytotoxicity was obtained with the REGb cells. After in vivo or in vitro selection of NK-resistant REGb cells, we observed in the selected cells an important decrease in RT-I class-I antigen expression. Fifteen different human colorectal cell lines were also studied for HLA class-I expression and NK-cell susceptibility, and no quantitative correlation between these 2 features was seen. However, cell lines which were deficient in HLA class-I antigens were more sensitive than class-I-positive cells.
Natural killer (NK) cells, which represent 10-15% of peripheral blood lymphocytes (PBL) and are principally associated with large granular lymphocytes (LGL), play an important role in the elimination of tumors (Robertson and Ritz, 1990). In contrast to cytotoxic T lymphocytes, these cells are able to lyse tumor cells in an MHC-unrestricted manner. Yet, not all tumor cells are equally sensitive to NK cells and some previous observations led to the suggestion that a deficit in MHC class I could be associated with increased sensitivity to NK-cell lysis (Karre et al., 1986; Harel-Bellan et al., 1986; Storkus et al., 1989). In the “missing self” hypothesis, Ljunggren and Karre (1990) proposed a multiple-choice model as an explanation. However, this remains controversial since several investigators, using different systems, failed to find a correlation between MHC class-I-antigen expression and NK-cell susceptibility (Chervenak and Wolcott, 1988; Stam et al., 1989; Leiden et af., 1989; Bertschmann et al., 1990). Moreover, in a mouse model, the BL6 melanoma cell line, increased sensitivity of the cells after transfection with class-I H-2K gene has been reported (Gorelik et al., 1990). The loss or diminution of HLA class-I expression is a relatively frequent event in colon carcinoma (Momburg et al., 1989), although Moller et al. (1991) have reported that it could not be correlated with recurrence rate or survival of patients. The aim of the present study was to investigate the influence of the level of expression of MHC class-I antigens on NKsusceptibility of colorectal tumor target cells. This study was performed using 2 clones of rat colon carcinoma originating from the same parental cell line (Caignard et al., 1985). These clones differ in their capacity to form tumors in vivo in the syngeneic host and in their sensitivity to NK cells (Pelletier et al., 1988). Moreover, variants with NK-resistant properties were obtained and studied. To test the value of the rat model, 15 human colorectal cell lines were also studied for HLA class-I-antigen expression and NK-cell susceptibility, in an attempt to see if the “missing self” hypothesis could be applied to colon tumors.
MATERIAL AND METHODS
Tumor-cell lines PROb and REGb rat colon adenocarcinoma cell clones, derived from a single tumor induced by 1,2 dimethylhydrazine and presenting with different tumorigenic properties in syngeneic rats, were obtained from Dr. F. Martin (Dijon, France). When grafted subcutaneously, PROb cells were able to grow and to metastasize, whereas REGb cells started to grow and were then rejected within 3 weeks (Caignard et al., 1985). Cells were cultured in RPMI 1640 supplemented with 10% FCS and 2 mM L-glutamine. Cells were passaged at confluence by dispersal with 0.025% trypsin and 0.02% EDTA. All human colorectal adenocarcinoma cell lines were cultured as described for PROb, with the exception of Caco-2 which was grown in DMEM supplemented with 20% FCS, 2 mM L-glutamine and 1% non-essential amino acids. All cell lines were free of Mycoplasma contamination. Cytotoxic assay Peripheral blood lymphocytes (PBL) of healthy donors or normal rats were separated by centrifugation on Ficoll gradient. Cytotoxicity was performed in either a 4-hr ”Cr-release assay or an 18-hr “’In-release assay as described (Blottikre et al., 1991). In order to increase MHC class-I expression, 2 x lo6target cells were treated with 1 X lo3 U/ml of interferon (IFN) y for 48 hr before being tested for W6/32 reactivity and NK susceptibility. Cytofluorimettyanalysis Cell staining, with mouse antibody OX-18 directed against the rat MHC class-I non-polymorphic antigens or W6/32 directed against human MHC class-I non-polymorphic antigens, was performed in microtiter plates kept on ice, 5 x lo5 cells/well being incubated with the antibodies at the appropriate concentrations in 0.1% gelatin/PBS for 30 min. After 3 washes, a second incubation was performed with FITC-labeled anti-mouse immunoglobulins (Vector, Burlingame, CA). After 3 more washes, fluorescence analysis was performed on an ATC 3000 flow cytometer (ODAM, Wissembourg, France) at 488 nm and 400 mwusing a 500-nm long-pass dichroic mirror and a 515-nm long-pass interference filter. Negative control were performed using an isotype-matched, irrelevant antibody. Selection of NK- or LAK-resistant cell lines Two different methods were performed in order to obtain NK or lymphokine-activated killer (LAK) resistant cells. The first approach was done in viva Rats were treated with anti-rat monoclonal antibody (MAb) CD5 (OX-19) which permits slow growth of the regressive clone REGb. Fifty-six days after cell inoculation, the tumor was removed under sterile conditions and put into culture. The cells, named ‘To whom correspondence and reprint requests should be addressed. Received: October 5,1991.
660
BLOTTIERE ET AL.
REGb-5, were then tested for MHC class-I expression and NK susceptibility. In a second approach, REGb cells were subjected in vitro to a continuous exposure to LAK cells. LAK cells were generated from rat splenocytes after 5 days in culture with IL-2 (Vujanovic et ai., 1988). REGb cells were detached and 500,000 cells were cultured together with 12.5 x lo6 LAK cells for 24 hr, then non-adherent cells were removed and adherent surviving cells were allowed to grow to confluence before being exposed again to LAK cells. These cells, named REGb-L, were tested for MHC class-I expression and LAK-cell susceptibility after 4 and 6 cycles of selection. RESULTS
MHC class-I and NK cells in the rat model The cytotoxicity of peripheral blood lymphocytes was assessed against PROb and REGb cells at 3 different effector: target ratios, 50:1, 25:1, 12.5:l. As reported by Pelletier et al. (1988), REGb cells were more sensitive to NK lysis than PROb cells (Fig. 1). For instance, at a 50:l effector:target ratio, 18.49 k 6.2% REGb cells were killed whereas 14.27 k 3.86% PROb cells were lysed ( p = 0.008). Both clones expressed class-I MHC antigens but this expression was stronger on PROb cells. IFNy treatment of PROb and REGb cells resulted in a considerable increase in expression on both cell lines (Fig. 2). However, NK-cell susceptibility of PROb cells was not modified by this pre-treatment (Fig. l), whereas on REGb cells a statistically significant increase in NK-cell lysis was observed (p = 0.006).
Selection of NK- or LAK-resistant cell variants Using 2 different approaches, subpopulations of REGb cells that were more resistant to effector cells than the parental clone were obtained. After in vivo T-cell depletion using anti-CD5 MAb OX-19, REGb tumor cells were able to grow slowly and after 2 months were put into culture. The cells obtained, REGb-5, were tested for NK-cell sensitivity and MHC class-I expression. REGb-5 cells were more resistant to NK-cell lysis than the initial REGb cells, with values very similar to those obtained for PROb cells (Fig. 3). As shown above, REGb cells expressed fewer MHC
401
il( 20
10
Y
.
5011
12.5/1
25/1 Effector to target
ratio
FIGURE 1 - Susceptibility of PROb (squares) and REGb (circles) to NK-cell-mediatedcytotoxicity before (open symbols) and after (closed symbols) IFN-7 treatment of the tumor cells. The results of one representative experiment out of 4 are presented. Values represent the mean of quadriplicate determinations in an 18-hr "'In-release assay as described in "Material and Methods".
class-I antigens than PROb cells; REGb-5 cells expressed far fewer MHC class-I antigens than REGb cells (Fig. 4a). An in vitro selection of REGb cells was also performed using LAK cells obtained after culture of rat splenocytes in the presence of IL-2 for 5 days. REGb cells were exposed to LAK cells for 24 hr. Then, the surviving cells were allowed to grow to confluence before being exposed again to LAK cells. After 4 and 6 cycles of selection, cells were tested for sensitivity to LAK cells. As shown in Table I, REGb cells exposed to LAK cells after 4 or 6 cycles of selection were more resistant than the initial REGb cells; however, they remain slightly more sensitive than PROb cells. When tested for MHC class-I expression, REGb-L cells displayed a lower expression than REGb cells as shown above for REGb-5 cells (Fig. 4b).
Relationship between NK-cell susceptibility and level of MHC class-I expression in human colorectal tumor cell lines Fifteen human colorectal carcinoma cell lines were tested for sensitivity to NK-cell lysis in an 18-hr "'In-release assay (Table 11). The same pattern of cytotoxicity was obtained when a 4-hr "Cr-release assay was used (data not shown). However, we decided to use "'In instead of "Cr because of the easiness of labeling with this radioactive compound. Cell lines were classified according to their level of susceptibility from resistant cells, such as Colo 205, HT-29 and SW-1222, to highly sensitive cells, such as HCT-8R or SW-48. The level of HLA class-I expression was also tested using h4Ab W6132 which recognizes a monomorphic determinant on HLA-A, B and C (Table 11). Five cell lines were found negative for W6132 reactivity. On the other cell lines, the reactivity varied from moderate (LS-174 T or SW-480) to strong (Colo 205 or SW-1116). A regression analysis showed that expression of MHC class-I antigens and cell susceptibility to NK-cell lysis (Fig. 5 ) were independent (r = 0.365; p = 0.149). However, when we compared NK-cell sensitivity of MHC class-I-negative cell lines (mean = 21.8 ? 11.0) with that of MHC class-I-positive lines (mean = 39.7 2 12.9), a statistically significant difference was found (p = 0.015). DISCUSSION
In spite of the MHC-unrestricted function of NK cells, it has been proposed that expression of MHC class-I antigens is a determinant factor in susceptibility of cells to NK-cell lysis, since target cells deficient in MHC class-I antigens are the most sensitive (Karre et al., 1986; Harel-Bellan ef aL, 1986). However, the relationship between MHC class-I expression and NK-cell susceptibility remains controversial, possibly depending on the nature of the tumors. Most of the work performed on tumors of hematopoietic origin favors the notion of a protective effect, as shown by IFN-y treatment which induces or increases MHC class-I expression (Ljunggren et al., 1989), after selection of mutants (Karre et al., 1986) or after MHC class-I gene transfection (Storkus et a[., 1989). However, several studies demonstrated that this protective effect was not always applicable, as shown after selection of variants in a murine lymphoma model (ASL1) and with P815 mastocytoma (Chervenak and Wolcott, 1988; Bertschmann et aL, 1990) or after gene transfection in the human erythroleukemia cell line K562 and the T-cell leukemia line MOLT-4 (Leiden et al., 1989). In the case of solid tumors, the same controversial results have been obtained with studies showing that MHC class-I antigen had no influence on susceptibility to NK-cell lysis. This was observed in studies of small-cell lung carcinoma (Stam et al., 1989), brain tumor (Pena et al., 1990) and melanoma
661
NK AND MHC CLASS-I ANTIGENS IN COLON TUMORS f
1
100
10
10
1000
100
1000
Fluorescence intensity
FIGURE 2 - Cytofluorimetric analysis of MHC class-I antigen expression on PROb (a) and REGb (b) using MAb OX-18. Negative control for untreated cells was performed using an isotype-matched irrelevant antibody (a); cells treated with IFN-y (c) and untreated (b) are shown. Negative control for IFN-y-treated cells (not shown) was similar to that of untreated cells.
0
1 25/1
50/1
Effector to target
12.5/1 ratio
FIGURE 3 - Susceptibility of PROb ( O ) ,REGb (0)and in vitro-selected REGb-5 cells ( 0 )to NK-cell-mediated rolysis. The results of one representative experiment out of 4 are presented. Values represent the mean of triplicates in an 18-hr In-release assay as described in the text. (Gorelik et al., 1990) and in other experiments demonstrating that MHC class-I molecules can confer resistance to melanomas or fibrosarcomas (Versteeg et al., 1989; Algarra et al., 1989). In colorectal cancer, the question remains unsolved. Modifications in expression of MHC class-I antigens are a common event in such tumors (Momburg et al., 1989). Either complete or selective loss occurs, resulting in a reduced or complete absence of reactivity with HLA-A-B-C framework MAb W6/ 32. In the present report, we have studied this question in an attempt to clarify the relationship between MHC class-I antigens and NK-cell-mediated lysis in colorectal tumors. The question was first addressed in an experimental model of colon carcinoma, in which 2 cell lines differ in their sensitivity to NK-cell lysis and in their tumorigenicity in the syngeneic host. Martin and colleagues have demonstrated, after NK-cell depletion with anti-asialo-GM1 MAb, that NK cells were involved in the rejection of the non-tumorigenic cell line (Shimizu et al., 1987). In this model, we observed that the “missing self’ hypothesis was plausible since the most resistant cells expressed more RT-1 class-I products than did the sensitive ones. However, IFN-y treatment of both cell lines,
which increased expression of MHC class-I antigens, resulted in an increase in susceptibility to NK effector cells in one cell line and had no effect on the other. Moreover, NK-resistant variants were obtained which exhibited lower expression of RT-1 class-I antigens than did the parental clone. This was quite surprising since, at first sight, the 2 selection procedures that we used should not have acted against the expression of MHC class-I antigens. The in vitro procedure was based upon the use of LAK cells which should kill their targets independently of MHC class-I antigens. As for the in vivo procedure, it was based upon T-cell depletion of the syngeneic host. Of course, this depletion could have been incomplete and some cytotoxic T cells recognizing a tumor-specific antigen presented by class-I molecules could have played a role in the selection process. Nevertheless this is most unlikely since we have observed that rats completely depleted of CD8-positive cells (recognizing antigens presented by class-I products) remained capable of rapidly rejecting REGb tumors (data not shown). Whatever the explanation for the paradoxical decrease in NK sensitivity associated with decreased class-I expression of our selected variants, the above results show that, in this rat model, the level of MHC class-I antigens expressed by the cells cannot be related to their susceptibility
662
BLOTTIERE
ET AL.
B
1
10
10
1600
100
100
1000
Fluorescence Intensity
and non-selected REGb cells using MAb OX-18. vivo-selected cells REG-5 (a), OX-18 on REGb-5 an isotype-matched irrelevant antibody on the in and OX-18 on the non-selected REGb cells (c). Negative controls for the parental REGb cells were similar to those given by the selected variants (not shown). Details of selection procedures are described in the text. The difference in fluorescence intensity between ( a ) and ( b ) is due to different settings of the cytofluorimeter. TABLE 1 - COMPARISON OF SENSITIVITYOF RAT COLON TUMOR CELLS, AFTER IN VITRO SELECTION USING LAK CELLS, WITH THAT OF PARENTALTUMOR CELLS Cell clone
Expt. 1 Expt.2
PROb REGb REGb-L (4)’ PROb REGb REGb-L (6)*
Effector:targetratios 12.5:l 25:l 501
15’ 46 29 25 37 28
29 65 41 28 45 33
42 70 58 37 51 39
‘Mean of percentage of cytotoxicity determined in triplicate in an 18-hr “‘In-release assay. The differences in killing between the 2 experiments are due to differences in activation of LAK cells.-2REGb-L cells obtained after the number of selection cycles indicated in parentheses.
to NK-cell lysis. The differences in susceptibility between the various cells studied must therefore reside in other factors. IFN-y treatment, for example, which increases expression of class-I antigens, is known to have many effects such as induction of adhesion molecules which participate in NK-cell activity (Naganuma et al., 1991). Our selection procedures could also have altered the expression of adhesion molecules. However, no difference in expression of the a5 and p1 chains of integrins between the parental REGb and its selected variants REGb-L or REGb-5 were observed (data not shown). To evaluate whether the results obtained in the rat model applied to all colon tumors, and to estimate the importance of MHC class-I expression relative to that of other factors possibly involved in the resistance of the target cells to NK-mediated lysis, a panel of 15 human colorectal cell lines was used. Since these cell lines are all independent, and hence highly heterogeneous, only those factors strongly affecting the NK susceptibility of the cells should be discernible. Analysis of HLA class-I expression and NK-cell sensitivity showed that there was no quantitative relationship between these 2 parameters. However, the 5 cell lines which were completely negative for HLA class-I molecule expression, furthermore not inducible by IFN y (data not shown), were significantly more sensitive to NK lysis than class-I-positive cell lines.
TABLE 11 - COMPAKISON BE‘IWttN NKCELL S t Y S I ’ I I V l T Y
AUD MHC
(‘LASS1 ANTI(iEN EXPRESSION I N COLON CARCINOMA CELL LINES Cell line
Cole 205 HT-29 sw-1222 LS-174-T SW-707 LS-180 SW-480 HRT-18
sw-948 Caco-2 SW-620 HCT-8R SW-48
(n)‘
% of NK cytotoxicity’
Class71 expression
8.5 2 0.6 13.1 2 3.4 13.4 f 2.6 15.3 f 2.0 16.7 k 7.4 19.9 f 3.9 25.4 f 3.8 25.8 ? 7.5 26.4 f 1.6 32.6 2 2.4 34.3 t 15.2 37.4 f 11.1 42.3 f 3.5 48.7 2 5.7 57.2 2 2.1
3303 204 110 46 120 117 66 0 346 0 0 210 247 0 0
‘Number of experiments.-*Mean ercentage of cytotoxicity ? SE of experiments performed in quadrip1icate.-! Relative intensity of fluorescence determined with a cytofluorimeter. The values given were obtained after subtraction of the fluorescence intensities obtained using an irrelevant isotype-matched control antibody.
Thus, when MHC class-I antigens were expressed at the cell surface, quantitative differences did not affect the susceptibility of the cells. This was true for the rat cell variants which all originated from the same parental cell line, thus limiting their heterogeneity, as well as for the human cell lines. It appears that a complete lack of MHC class-I antigens may be necessary, but not sufficient, for susceptibility of colon carcinoma cells to NK-mediated cell lysis. Hypotheses have been proposed in order to explain how the presence of MHC class-I molecules at the target cell surface confers resistance. For Ljunggren and Karre (1990), there are 2 possible mechanisms: (i) “target interference”, in which an as yet unidentified natural killer target structure is masked by MHC class-I molecules, resulting in protection of the target; and (ii) “effector inhibition”, which implies that MHC class-I antigens bind to a molecule present on NK cells, resulting in an inhibition of cytotoxicity. A hypothesis similar to the “target
NK AND MHC CLASS-I ANTIGENS IN COLON TUMORS
r = 0,365
40
i
p = 0,149
20
31
lo
1
‘
=
m
o !
0
I
I
100
300
200
400
HLA class I expression (relative
fluoresence
Intensity)
FIGURE 5 - HLA class-I expression related to NK-cell-mediated cytotoxicity of human colorectal tumor cell lines showing the absence of a linear relationship (r = 0.36.5;p = 0.149) between these 2 parameters. Values are presented in Table 11.
interference” mechanism described above has been proposed by Storkus and Dawson (1991). Such hypotheses satisfactorily explain why enhancement of MHC class-I expression through IFN y treatment or gene transfection renders the cells more resistant to NK lysis. However, they do not explain how cells which strongly express MHC class-I molecules can also b e highly sensitive to NK-mediated lysis. This concerns, for instance, cell lines Caco-2 or SW 620, or cells which became more sensitive to NK lysis after MHC class-I induction with
663
IFN y, such as R E G b cells (see Fig. 1 and 2), or cells which became more sensitive t o NK lysis after H-2Khgene transfection (Gorelik et al., 1990). Such contradictory results are partially answered by Ljunggren and Kirre (1990) who suggest that class-I molecules can be overruled by other, triggering, target structures. However n o such molecules have yet been identified. In conclusion, we have observed that MHC class-I antigens probably play a role in modulating the susceptibility of colon carcinoma cells to NK cells. However this phenomenon of susceptibility is clearly a multifactorial event (Storkus and Dawson, 1991). Thus, the influence of M H C class-I products, depending on the particular cell line under study, may be overruled by other factors. The state of differentiation of the target cells could b e one of these factors (Gidlund et al., 1981) as well as alterations in glycosylation (Ahrens and Ankel, 1988), or the level of expression of adhesion molecules (Storkus and Dawson, 1991). Studies are now being performed to determine which of these factors also significantly influence the susceptibility of colorectal cancer cell lines toward NK cells.
ACKNOWLEDGEMENTS
H.M.B. is the recipient of a fellowship from the Rtgion des Pays d e Loire, France. Support for this study was provided by grants 6169 and 6871 from the Association pour la Recherche sur le Cancer (ARC), France, and by the Langlois Foundation, Rennes, France.
REFERENCES AHRENS, P.B. and ANKEL,H., Natural killer cells discriminate between high-mannose and complex-ty e asparagine-linked oligosaccharides. Biochimie, 70,1619-1625 (19887. ALGARRA, I., OHLEN,c., PEREZ,M., LJUNGGREN, H.G., KLEIN, G., GARRIDO, F. and K ~ R R EK., , NK sensitivity and lung clearance of MHC class-I-deficient cells within a heterogeneous fibrosarcoma. Int. J. Cancer, 44,675-680 (1989). BERTSCHMANN, M., GSERMANN, D. and KELLER,R., MHC class-I antigen deficiency, malignancy and susceptibilityof P815 mastocytoma to NK and macrophage killing. Znt. J. Cancer, 46,739-744 (1990). Z., HERLYN, D. and DOUILLARD, J.Y., BLOTTIPRE, H.M., STEPLEWSKI, Human anti-murine immunoglobulinresponses and immune functions in cancer patients receiving murine monoclonal antibody therapy. Hum. Antibod. Hybridomas, 2,16-25 (1991). CAIGNARD, A., MARTIN, M.S., MICHEL, M.F. and MARTIN, F., Interaction between two cellular subpopulations of a rat colonic carcinoma when inoculated to the syngeneic host. Int. J. Cancer, 36, 273-279 (1985). CHERVENAK, R. and WOLCOTT, R.M., Target cell expression of MHC antigens is not (always) a turn-off signal to natural killer cells. J. Immunol., 140,3712-3716 (1988). GIDLUND, M., ORN,A,, PATTENGALE, P.K., JANSSON, M., WIGZELL, H. and NILSSON, K., Natural killer cells kill tumor cells at a given stage of differentiation.Nature (Lond.),292,848-850 (1981). GORELIK, E., JAY, G., KWIATKOWSKI, B. and HERBERMAN, R.B., Increased sensitivity to MHC-non-restricted lysis of BL6 melanoma cells by transfection with class I H-2Kb gene. J. Immunol., 145, 1621-1632 (1990). HAREL-BELLAN, A,, QUILLET, A., MARCHIOL, C., DEMARS, R., TURSZ, T. and FRADELIZI, D., Natural killer susceptibilityof human cells may be regulated by genes in the HLA region on chromosome 6. Proc. nut. Acad. Sci. (Wash.),83,5688-5692 (1986). KARRE, K., LJUNGGREN, H.G., PIONTEK,G. and KIESSLING, R., Selective rejection of H-2-deficient lymphoma variants suggests alternative immune defense strategy. Nature (Lond.),319,675-678 (1986). J., LEIDEN, J.M., KARPINSKI, B.A., GOTTSCHALK, L. and KORNBLUTH,
Susceptibilityto natural killer cell-mediated cytolysis is independent of the level of target cell class I HLA expression. J. Immunol., 142, 2140-2147 (1989). LJUNGGREN, H.G. and G R R E ,K., In search of the “missing self’: MHC molecules and NK cell recognition. Imrnunol. Today, 11, 237-243 (1990). M., KLING,G., KOURILSKY, P. LJUNGGREN, H.G., PGBo, S., COCHET, and KARRE,K., Molecular analysis of H-Zdeficient lymphoma lines. Distinct defects in biosynthesis and association of MHC class-I heavy-chain and P2-microglobulin observed in cell with increased sensitivity to NK lysis. J. Immunol., 142,2911-2917 (1989). G., HERMOLLER,P., MOMBURG, F., KORETZ,K., MOLDENHAUER, FARTH, C., Omo, H.F., HAMMERLING, G.J. and SCHLAG, P., Influence of major histocompatibilitycomplex class I and I1 antigens on survival in colorectal carcinoma. Cancer Res., 51,729-736 (1991). MOMBURG, F., ZIEGLER, A,, HARPPRECHT, J., MOLLER,P., MOLDENHAUER, G. and ~ M M E R L I G.J., N G ,Selective loss of HLA-A or HLA-B antigen expression in colon carcinoma. J. Immunol., 142, 352-3.58 (1989). NAGANUMA, H., KIESSLING, R., PATARROYO, M., HANSSON, M., HANDGRETINGER, R. and GRONBERG, A., Increased susceptibility of IFN-y-treated neuroblastoma cells to lysis by lymphokine-activated killer cells: participation of ICAM-1 induction on target cells. Int. J. Cancer, 47,527-532 (1991). PELLETIER, H., OLSSON, N.O., FADY,C., REISSER, D., LAGADEC, P. and JEANNIN, J.F., Differential sensitivity to natural cell-mediated cytotoxicity of two rat colon adenocarcinoma variants differing in their tumorigenicity:identification of the effector cells as natural killer cells. Cancer Immunol. Immunother., 26,263-268 (1988). C., SOLANA, R., SERRANO, R., CARRACEDO, J. and PENA,J., ALONSO, RAMIREZ, R., Natural killer susceptibilityis independent of HLA class I antigen expression on cell lines obtained from human solid tumors. Europ. J. Immunol., 20,2445-2449 (1990). ROBERTSON, M.J. and R I ~ J.,, Biology and clinical relevance of human natural killer cells. Blood, 76,2421-2438 (1990).
664
BLOTTIEREET AL.
SHIMIZU, T., PELLETIER, H., HAMMANN, A., OLSON, N.O., MARTIN, M.S. and MARTIN, F., Effects of a single injection of anti-asialo GM1 serum on natural cytotoxicity and the growth of a regressive colonic tumor in syngeneic rats. Int. J. Cancer, 40,676-680 (1987). STAM,N.J., UST, W.M., VOORDOUW, A.C., PASTOORS, L.B., VAN DER HOEVEN,F.A., MELIEF,C.J. and PLOEGH,H.L., Lack of correlation between levels of MHC class I antigen and susceptibility to lysis of small cellular lung carcinoma (SCLC) by natural killer cells. J. Immunol., 142,4113-4117 (1989). STORKUS, W.J., ALEXANDER, J., PAYNE,J.A., CRESSWELL, P. and
DAWSON,J.R., The d / a 2 domains of class I HLA molecules confer resistance to natural killing. J. Immunol., 143,3853-3857 (1989). STORKUS,W.J. and DAWSON,J.R., Target structures involved in natural killing (NK): characteristics, distribution and candidate molecules. CriL Rev. Immunol., 10,393416 (1991). VERSTEEG, R., PELTENBURG, L.T.C., PLOMP,A.C. and SCHRIER, P.I., High expression of the c-myc oncogene renders melanoma cells prone to lysis by natural killer ce1ls.J. Immunol., 143,43314337 (1989). VUJANOVIC, N.L., HERBERMAN, R.B. and HISERODT, J.C., Lymphokineactivated killer cells in rats: analysis of tissue and strain distribution, ontogeny, and target specificity. Cancer Rex, 48,878-883 (1988).
Publication of the International Union Against Cancer Publicationde I’Union Internationale Contre le Cancer
RELATIONSHIP BETWEEN SENSITIVITY TO NATURAL KILLER CELLS AND MHC CLASS-I ANTIGEN EXPRESSION IN COLON CARCINOMA CELL LINES HervC M. BLOTTIERE’,Rahima ZENNADI, Cedric BURG,Jean-Yves DOUILLARD, Khaled MEFLAHand Jacques LE PENDU CJF INSERM 90.11, Facultt! de Mkdecine, I rue Gaston Veil, 44035 Nantes, France. The sensitivity of colorectal tumors to NK-cell-mediated cytotoxicity and their expression of major histocompatibility complex (MHC) class-I antigens were studied in an attempt to determine whether such antigens play a role in the susceptibility of colorectal tumors to NK-cell lysis. In a rat colon-carcinoma model, 2 clones differing in their sensitivity to NK-cell-mediated cytotoxicity were tested for class-I expression; it was seen that the more sensitive cells (REGb) expressed less class-I products than did the resistant cells (PROb). However, when MHC class-I antigen expression was increased by IFN-y treatment, no change in NK-cell lysis was found with the PROb cells, while an increase in cytotoxicity was obtained with the REGb cells. After in vivo or in vitro selection of NK-resistant REGb cells, we observed in the selected cells an important decrease in RT-I class-I antigen expression. Fifteen different human colorectal cell lines were also studied for HLA class-I expression and NK-cell susceptibility, and no quantitative correlation between these 2 features was seen. However, cell lines which were deficient in HLA class-I antigens were more sensitive than class-I-positive cells.
Natural killer (NK) cells, which represent 10-15% of peripheral blood lymphocytes (PBL) and are principally associated with large granular lymphocytes (LGL), play an important role in the elimination of tumors (Robertson and Ritz, 1990). In contrast to cytotoxic T lymphocytes, these cells are able to lyse tumor cells in an MHC-unrestricted manner. Yet, not all tumor cells are equally sensitive to NK cells and some previous observations led to the suggestion that a deficit in MHC class I could be associated with increased sensitivity to NK-cell lysis (Karre et al., 1986; Harel-Bellan et al., 1986; Storkus et al., 1989). In the “missing self” hypothesis, Ljunggren and Karre (1990) proposed a multiple-choice model as an explanation. However, this remains controversial since several investigators, using different systems, failed to find a correlation between MHC class-I-antigen expression and NK-cell susceptibility (Chervenak and Wolcott, 1988; Stam et al., 1989; Leiden et af., 1989; Bertschmann et al., 1990). Moreover, in a mouse model, the BL6 melanoma cell line, increased sensitivity of the cells after transfection with class-I H-2K gene has been reported (Gorelik et al., 1990). The loss or diminution of HLA class-I expression is a relatively frequent event in colon carcinoma (Momburg et al., 1989), although Moller et al. (1991) have reported that it could not be correlated with recurrence rate or survival of patients. The aim of the present study was to investigate the influence of the level of expression of MHC class-I antigens on NKsusceptibility of colorectal tumor target cells. This study was performed using 2 clones of rat colon carcinoma originating from the same parental cell line (Caignard et al., 1985). These clones differ in their capacity to form tumors in vivo in the syngeneic host and in their sensitivity to NK cells (Pelletier et al., 1988). Moreover, variants with NK-resistant properties were obtained and studied. To test the value of the rat model, 15 human colorectal cell lines were also studied for HLA class-I-antigen expression and NK-cell susceptibility, in an attempt to see if the “missing self” hypothesis could be applied to colon tumors.
MATERIAL AND METHODS
Tumor-cell lines PROb and REGb rat colon adenocarcinoma cell clones, derived from a single tumor induced by 1,2 dimethylhydrazine and presenting with different tumorigenic properties in syngeneic rats, were obtained from Dr. F. Martin (Dijon, France). When grafted subcutaneously, PROb cells were able to grow and to metastasize, whereas REGb cells started to grow and were then rejected within 3 weeks (Caignard et al., 1985). Cells were cultured in RPMI 1640 supplemented with 10% FCS and 2 mM L-glutamine. Cells were passaged at confluence by dispersal with 0.025% trypsin and 0.02% EDTA. All human colorectal adenocarcinoma cell lines were cultured as described for PROb, with the exception of Caco-2 which was grown in DMEM supplemented with 20% FCS, 2 mM L-glutamine and 1% non-essential amino acids. All cell lines were free of Mycoplasma contamination. Cytotoxic assay Peripheral blood lymphocytes (PBL) of healthy donors or normal rats were separated by centrifugation on Ficoll gradient. Cytotoxicity was performed in either a 4-hr ”Cr-release assay or an 18-hr “’In-release assay as described (Blottikre et al., 1991). In order to increase MHC class-I expression, 2 x lo6target cells were treated with 1 X lo3 U/ml of interferon (IFN) y for 48 hr before being tested for W6/32 reactivity and NK susceptibility. Cytofluorimettyanalysis Cell staining, with mouse antibody OX-18 directed against the rat MHC class-I non-polymorphic antigens or W6/32 directed against human MHC class-I non-polymorphic antigens, was performed in microtiter plates kept on ice, 5 x lo5 cells/well being incubated with the antibodies at the appropriate concentrations in 0.1% gelatin/PBS for 30 min. After 3 washes, a second incubation was performed with FITC-labeled anti-mouse immunoglobulins (Vector, Burlingame, CA). After 3 more washes, fluorescence analysis was performed on an ATC 3000 flow cytometer (ODAM, Wissembourg, France) at 488 nm and 400 mwusing a 500-nm long-pass dichroic mirror and a 515-nm long-pass interference filter. Negative control were performed using an isotype-matched, irrelevant antibody. Selection of NK- or LAK-resistant cell lines Two different methods were performed in order to obtain NK or lymphokine-activated killer (LAK) resistant cells. The first approach was done in viva Rats were treated with anti-rat monoclonal antibody (MAb) CD5 (OX-19) which permits slow growth of the regressive clone REGb. Fifty-six days after cell inoculation, the tumor was removed under sterile conditions and put into culture. The cells, named ‘To whom correspondence and reprint requests should be addressed. Received: October 5,1991.
660
BLOTTIERE ET AL.
REGb-5, were then tested for MHC class-I expression and NK susceptibility. In a second approach, REGb cells were subjected in vitro to a continuous exposure to LAK cells. LAK cells were generated from rat splenocytes after 5 days in culture with IL-2 (Vujanovic et ai., 1988). REGb cells were detached and 500,000 cells were cultured together with 12.5 x lo6 LAK cells for 24 hr, then non-adherent cells were removed and adherent surviving cells were allowed to grow to confluence before being exposed again to LAK cells. These cells, named REGb-L, were tested for MHC class-I expression and LAK-cell susceptibility after 4 and 6 cycles of selection. RESULTS
MHC class-I and NK cells in the rat model The cytotoxicity of peripheral blood lymphocytes was assessed against PROb and REGb cells at 3 different effector: target ratios, 50:1, 25:1, 12.5:l. As reported by Pelletier et al. (1988), REGb cells were more sensitive to NK lysis than PROb cells (Fig. 1). For instance, at a 50:l effector:target ratio, 18.49 k 6.2% REGb cells were killed whereas 14.27 k 3.86% PROb cells were lysed ( p = 0.008). Both clones expressed class-I MHC antigens but this expression was stronger on PROb cells. IFNy treatment of PROb and REGb cells resulted in a considerable increase in expression on both cell lines (Fig. 2). However, NK-cell susceptibility of PROb cells was not modified by this pre-treatment (Fig. l), whereas on REGb cells a statistically significant increase in NK-cell lysis was observed (p = 0.006).
Selection of NK- or LAK-resistant cell variants Using 2 different approaches, subpopulations of REGb cells that were more resistant to effector cells than the parental clone were obtained. After in vivo T-cell depletion using anti-CD5 MAb OX-19, REGb tumor cells were able to grow slowly and after 2 months were put into culture. The cells obtained, REGb-5, were tested for NK-cell sensitivity and MHC class-I expression. REGb-5 cells were more resistant to NK-cell lysis than the initial REGb cells, with values very similar to those obtained for PROb cells (Fig. 3). As shown above, REGb cells expressed fewer MHC
401
il( 20
10
Y
.
5011
12.5/1
25/1 Effector to target
ratio
FIGURE 1 - Susceptibility of PROb (squares) and REGb (circles) to NK-cell-mediatedcytotoxicity before (open symbols) and after (closed symbols) IFN-7 treatment of the tumor cells. The results of one representative experiment out of 4 are presented. Values represent the mean of quadriplicate determinations in an 18-hr "'In-release assay as described in "Material and Methods".
class-I antigens than PROb cells; REGb-5 cells expressed far fewer MHC class-I antigens than REGb cells (Fig. 4a). An in vitro selection of REGb cells was also performed using LAK cells obtained after culture of rat splenocytes in the presence of IL-2 for 5 days. REGb cells were exposed to LAK cells for 24 hr. Then, the surviving cells were allowed to grow to confluence before being exposed again to LAK cells. After 4 and 6 cycles of selection, cells were tested for sensitivity to LAK cells. As shown in Table I, REGb cells exposed to LAK cells after 4 or 6 cycles of selection were more resistant than the initial REGb cells; however, they remain slightly more sensitive than PROb cells. When tested for MHC class-I expression, REGb-L cells displayed a lower expression than REGb cells as shown above for REGb-5 cells (Fig. 4b).
Relationship between NK-cell susceptibility and level of MHC class-I expression in human colorectal tumor cell lines Fifteen human colorectal carcinoma cell lines were tested for sensitivity to NK-cell lysis in an 18-hr "'In-release assay (Table 11). The same pattern of cytotoxicity was obtained when a 4-hr "Cr-release assay was used (data not shown). However, we decided to use "'In instead of "Cr because of the easiness of labeling with this radioactive compound. Cell lines were classified according to their level of susceptibility from resistant cells, such as Colo 205, HT-29 and SW-1222, to highly sensitive cells, such as HCT-8R or SW-48. The level of HLA class-I expression was also tested using h4Ab W6132 which recognizes a monomorphic determinant on HLA-A, B and C (Table 11). Five cell lines were found negative for W6132 reactivity. On the other cell lines, the reactivity varied from moderate (LS-174 T or SW-480) to strong (Colo 205 or SW-1116). A regression analysis showed that expression of MHC class-I antigens and cell susceptibility to NK-cell lysis (Fig. 5 ) were independent (r = 0.365; p = 0.149). However, when we compared NK-cell sensitivity of MHC class-I-negative cell lines (mean = 21.8 ? 11.0) with that of MHC class-I-positive lines (mean = 39.7 2 12.9), a statistically significant difference was found (p = 0.015). DISCUSSION
In spite of the MHC-unrestricted function of NK cells, it has been proposed that expression of MHC class-I antigens is a determinant factor in susceptibility of cells to NK-cell lysis, since target cells deficient in MHC class-I antigens are the most sensitive (Karre et al., 1986; Harel-Bellan ef aL, 1986). However, the relationship between MHC class-I expression and NK-cell susceptibility remains controversial, possibly depending on the nature of the tumors. Most of the work performed on tumors of hematopoietic origin favors the notion of a protective effect, as shown by IFN-y treatment which induces or increases MHC class-I expression (Ljunggren et al., 1989), after selection of mutants (Karre et al., 1986) or after MHC class-I gene transfection (Storkus et a[., 1989). However, several studies demonstrated that this protective effect was not always applicable, as shown after selection of variants in a murine lymphoma model (ASL1) and with P815 mastocytoma (Chervenak and Wolcott, 1988; Bertschmann et aL, 1990) or after gene transfection in the human erythroleukemia cell line K562 and the T-cell leukemia line MOLT-4 (Leiden et al., 1989). In the case of solid tumors, the same controversial results have been obtained with studies showing that MHC class-I antigen had no influence on susceptibility to NK-cell lysis. This was observed in studies of small-cell lung carcinoma (Stam et al., 1989), brain tumor (Pena et al., 1990) and melanoma
661
NK AND MHC CLASS-I ANTIGENS IN COLON TUMORS f
1
100
10
10
1000
100
1000
Fluorescence intensity
FIGURE 2 - Cytofluorimetric analysis of MHC class-I antigen expression on PROb (a) and REGb (b) using MAb OX-18. Negative control for untreated cells was performed using an isotype-matched irrelevant antibody (a); cells treated with IFN-y (c) and untreated (b) are shown. Negative control for IFN-y-treated cells (not shown) was similar to that of untreated cells.
0
1 25/1
50/1
Effector to target
12.5/1 ratio
FIGURE 3 - Susceptibility of PROb ( O ) ,REGb (0)and in vitro-selected REGb-5 cells ( 0 )to NK-cell-mediated rolysis. The results of one representative experiment out of 4 are presented. Values represent the mean of triplicates in an 18-hr In-release assay as described in the text. (Gorelik et al., 1990) and in other experiments demonstrating that MHC class-I molecules can confer resistance to melanomas or fibrosarcomas (Versteeg et al., 1989; Algarra et al., 1989). In colorectal cancer, the question remains unsolved. Modifications in expression of MHC class-I antigens are a common event in such tumors (Momburg et al., 1989). Either complete or selective loss occurs, resulting in a reduced or complete absence of reactivity with HLA-A-B-C framework MAb W6/ 32. In the present report, we have studied this question in an attempt to clarify the relationship between MHC class-I antigens and NK-cell-mediated lysis in colorectal tumors. The question was first addressed in an experimental model of colon carcinoma, in which 2 cell lines differ in their sensitivity to NK-cell lysis and in their tumorigenicity in the syngeneic host. Martin and colleagues have demonstrated, after NK-cell depletion with anti-asialo-GM1 MAb, that NK cells were involved in the rejection of the non-tumorigenic cell line (Shimizu et al., 1987). In this model, we observed that the “missing self’ hypothesis was plausible since the most resistant cells expressed more RT-1 class-I products than did the sensitive ones. However, IFN-y treatment of both cell lines,
which increased expression of MHC class-I antigens, resulted in an increase in susceptibility to NK effector cells in one cell line and had no effect on the other. Moreover, NK-resistant variants were obtained which exhibited lower expression of RT-1 class-I antigens than did the parental clone. This was quite surprising since, at first sight, the 2 selection procedures that we used should not have acted against the expression of MHC class-I antigens. The in vitro procedure was based upon the use of LAK cells which should kill their targets independently of MHC class-I antigens. As for the in vivo procedure, it was based upon T-cell depletion of the syngeneic host. Of course, this depletion could have been incomplete and some cytotoxic T cells recognizing a tumor-specific antigen presented by class-I molecules could have played a role in the selection process. Nevertheless this is most unlikely since we have observed that rats completely depleted of CD8-positive cells (recognizing antigens presented by class-I products) remained capable of rapidly rejecting REGb tumors (data not shown). Whatever the explanation for the paradoxical decrease in NK sensitivity associated with decreased class-I expression of our selected variants, the above results show that, in this rat model, the level of MHC class-I antigens expressed by the cells cannot be related to their susceptibility
662
BLOTTIERE
ET AL.
B
1
10
10
1600
100
100
1000
Fluorescence Intensity
and non-selected REGb cells using MAb OX-18. vivo-selected cells REG-5 (a), OX-18 on REGb-5 an isotype-matched irrelevant antibody on the in and OX-18 on the non-selected REGb cells (c). Negative controls for the parental REGb cells were similar to those given by the selected variants (not shown). Details of selection procedures are described in the text. The difference in fluorescence intensity between ( a ) and ( b ) is due to different settings of the cytofluorimeter. TABLE 1 - COMPARISON OF SENSITIVITYOF RAT COLON TUMOR CELLS, AFTER IN VITRO SELECTION USING LAK CELLS, WITH THAT OF PARENTALTUMOR CELLS Cell clone
Expt. 1 Expt.2
PROb REGb REGb-L (4)’ PROb REGb REGb-L (6)*
Effector:targetratios 12.5:l 25:l 501
15’ 46 29 25 37 28
29 65 41 28 45 33
42 70 58 37 51 39
‘Mean of percentage of cytotoxicity determined in triplicate in an 18-hr “‘In-release assay. The differences in killing between the 2 experiments are due to differences in activation of LAK cells.-2REGb-L cells obtained after the number of selection cycles indicated in parentheses.
to NK-cell lysis. The differences in susceptibility between the various cells studied must therefore reside in other factors. IFN-y treatment, for example, which increases expression of class-I antigens, is known to have many effects such as induction of adhesion molecules which participate in NK-cell activity (Naganuma et al., 1991). Our selection procedures could also have altered the expression of adhesion molecules. However, no difference in expression of the a5 and p1 chains of integrins between the parental REGb and its selected variants REGb-L or REGb-5 were observed (data not shown). To evaluate whether the results obtained in the rat model applied to all colon tumors, and to estimate the importance of MHC class-I expression relative to that of other factors possibly involved in the resistance of the target cells to NK-mediated lysis, a panel of 15 human colorectal cell lines was used. Since these cell lines are all independent, and hence highly heterogeneous, only those factors strongly affecting the NK susceptibility of the cells should be discernible. Analysis of HLA class-I expression and NK-cell sensitivity showed that there was no quantitative relationship between these 2 parameters. However, the 5 cell lines which were completely negative for HLA class-I molecule expression, furthermore not inducible by IFN y (data not shown), were significantly more sensitive to NK lysis than class-I-positive cell lines.
TABLE 11 - COMPAKISON BE‘IWttN NKCELL S t Y S I ’ I I V l T Y
AUD MHC
(‘LASS1 ANTI(iEN EXPRESSION I N COLON CARCINOMA CELL LINES Cell line
Cole 205 HT-29 sw-1222 LS-174-T SW-707 LS-180 SW-480 HRT-18
sw-948 Caco-2 SW-620 HCT-8R SW-48
(n)‘
% of NK cytotoxicity’
Class71 expression
8.5 2 0.6 13.1 2 3.4 13.4 f 2.6 15.3 f 2.0 16.7 k 7.4 19.9 f 3.9 25.4 f 3.8 25.8 ? 7.5 26.4 f 1.6 32.6 2 2.4 34.3 t 15.2 37.4 f 11.1 42.3 f 3.5 48.7 2 5.7 57.2 2 2.1
3303 204 110 46 120 117 66 0 346 0 0 210 247 0 0
‘Number of experiments.-*Mean ercentage of cytotoxicity ? SE of experiments performed in quadrip1icate.-! Relative intensity of fluorescence determined with a cytofluorimeter. The values given were obtained after subtraction of the fluorescence intensities obtained using an irrelevant isotype-matched control antibody.
Thus, when MHC class-I antigens were expressed at the cell surface, quantitative differences did not affect the susceptibility of the cells. This was true for the rat cell variants which all originated from the same parental cell line, thus limiting their heterogeneity, as well as for the human cell lines. It appears that a complete lack of MHC class-I antigens may be necessary, but not sufficient, for susceptibility of colon carcinoma cells to NK-mediated cell lysis. Hypotheses have been proposed in order to explain how the presence of MHC class-I molecules at the target cell surface confers resistance. For Ljunggren and Karre (1990), there are 2 possible mechanisms: (i) “target interference”, in which an as yet unidentified natural killer target structure is masked by MHC class-I molecules, resulting in protection of the target; and (ii) “effector inhibition”, which implies that MHC class-I antigens bind to a molecule present on NK cells, resulting in an inhibition of cytotoxicity. A hypothesis similar to the “target
NK AND MHC CLASS-I ANTIGENS IN COLON TUMORS
r = 0,365
40
i
p = 0,149
20
31
lo
1
‘
=
m
o !
0
I
I
100
300
200
400
HLA class I expression (relative
fluoresence
Intensity)
FIGURE 5 - HLA class-I expression related to NK-cell-mediated cytotoxicity of human colorectal tumor cell lines showing the absence of a linear relationship (r = 0.36.5;p = 0.149) between these 2 parameters. Values are presented in Table 11.
interference” mechanism described above has been proposed by Storkus and Dawson (1991). Such hypotheses satisfactorily explain why enhancement of MHC class-I expression through IFN y treatment or gene transfection renders the cells more resistant to NK lysis. However, they do not explain how cells which strongly express MHC class-I molecules can also b e highly sensitive to NK-mediated lysis. This concerns, for instance, cell lines Caco-2 or SW 620, or cells which became more sensitive to NK lysis after MHC class-I induction with
663
IFN y, such as R E G b cells (see Fig. 1 and 2), or cells which became more sensitive t o NK lysis after H-2Khgene transfection (Gorelik et al., 1990). Such contradictory results are partially answered by Ljunggren and Kirre (1990) who suggest that class-I molecules can be overruled by other, triggering, target structures. However n o such molecules have yet been identified. In conclusion, we have observed that MHC class-I antigens probably play a role in modulating the susceptibility of colon carcinoma cells to NK cells. However this phenomenon of susceptibility is clearly a multifactorial event (Storkus and Dawson, 1991). Thus, the influence of M H C class-I products, depending on the particular cell line under study, may be overruled by other factors. The state of differentiation of the target cells could b e one of these factors (Gidlund et al., 1981) as well as alterations in glycosylation (Ahrens and Ankel, 1988), or the level of expression of adhesion molecules (Storkus and Dawson, 1991). Studies are now being performed to determine which of these factors also significantly influence the susceptibility of colorectal cancer cell lines toward NK cells.
ACKNOWLEDGEMENTS
H.M.B. is the recipient of a fellowship from the Rtgion des Pays d e Loire, France. Support for this study was provided by grants 6169 and 6871 from the Association pour la Recherche sur le Cancer (ARC), France, and by the Langlois Foundation, Rennes, France.
REFERENCES AHRENS, P.B. and ANKEL,H., Natural killer cells discriminate between high-mannose and complex-ty e asparagine-linked oligosaccharides. Biochimie, 70,1619-1625 (19887. ALGARRA, I., OHLEN,c., PEREZ,M., LJUNGGREN, H.G., KLEIN, G., GARRIDO, F. and K ~ R R EK., , NK sensitivity and lung clearance of MHC class-I-deficient cells within a heterogeneous fibrosarcoma. Int. J. Cancer, 44,675-680 (1989). BERTSCHMANN, M., GSERMANN, D. and KELLER,R., MHC class-I antigen deficiency, malignancy and susceptibilityof P815 mastocytoma to NK and macrophage killing. Znt. J. Cancer, 46,739-744 (1990). Z., HERLYN, D. and DOUILLARD, J.Y., BLOTTIPRE, H.M., STEPLEWSKI, Human anti-murine immunoglobulinresponses and immune functions in cancer patients receiving murine monoclonal antibody therapy. Hum. Antibod. Hybridomas, 2,16-25 (1991). CAIGNARD, A., MARTIN, M.S., MICHEL, M.F. and MARTIN, F., Interaction between two cellular subpopulations of a rat colonic carcinoma when inoculated to the syngeneic host. Int. J. Cancer, 36, 273-279 (1985). CHERVENAK, R. and WOLCOTT, R.M., Target cell expression of MHC antigens is not (always) a turn-off signal to natural killer cells. J. Immunol., 140,3712-3716 (1988). GIDLUND, M., ORN,A,, PATTENGALE, P.K., JANSSON, M., WIGZELL, H. and NILSSON, K., Natural killer cells kill tumor cells at a given stage of differentiation.Nature (Lond.),292,848-850 (1981). GORELIK, E., JAY, G., KWIATKOWSKI, B. and HERBERMAN, R.B., Increased sensitivity to MHC-non-restricted lysis of BL6 melanoma cells by transfection with class I H-2Kb gene. J. Immunol., 145, 1621-1632 (1990). HAREL-BELLAN, A,, QUILLET, A., MARCHIOL, C., DEMARS, R., TURSZ, T. and FRADELIZI, D., Natural killer susceptibilityof human cells may be regulated by genes in the HLA region on chromosome 6. Proc. nut. Acad. Sci. (Wash.),83,5688-5692 (1986). KARRE, K., LJUNGGREN, H.G., PIONTEK,G. and KIESSLING, R., Selective rejection of H-2-deficient lymphoma variants suggests alternative immune defense strategy. Nature (Lond.),319,675-678 (1986). J., LEIDEN, J.M., KARPINSKI, B.A., GOTTSCHALK, L. and KORNBLUTH,
Susceptibilityto natural killer cell-mediated cytolysis is independent of the level of target cell class I HLA expression. J. Immunol., 142, 2140-2147 (1989). LJUNGGREN, H.G. and G R R E ,K., In search of the “missing self’: MHC molecules and NK cell recognition. Imrnunol. Today, 11, 237-243 (1990). M., KLING,G., KOURILSKY, P. LJUNGGREN, H.G., PGBo, S., COCHET, and KARRE,K., Molecular analysis of H-Zdeficient lymphoma lines. Distinct defects in biosynthesis and association of MHC class-I heavy-chain and P2-microglobulin observed in cell with increased sensitivity to NK lysis. J. Immunol., 142,2911-2917 (1989). G., HERMOLLER,P., MOMBURG, F., KORETZ,K., MOLDENHAUER, FARTH, C., Omo, H.F., HAMMERLING, G.J. and SCHLAG, P., Influence of major histocompatibilitycomplex class I and I1 antigens on survival in colorectal carcinoma. Cancer Res., 51,729-736 (1991). MOMBURG, F., ZIEGLER, A,, HARPPRECHT, J., MOLLER,P., MOLDENHAUER, G. and ~ M M E R L I G.J., N G ,Selective loss of HLA-A or HLA-B antigen expression in colon carcinoma. J. Immunol., 142, 352-3.58 (1989). NAGANUMA, H., KIESSLING, R., PATARROYO, M., HANSSON, M., HANDGRETINGER, R. and GRONBERG, A., Increased susceptibility of IFN-y-treated neuroblastoma cells to lysis by lymphokine-activated killer cells: participation of ICAM-1 induction on target cells. Int. J. Cancer, 47,527-532 (1991). PELLETIER, H., OLSSON, N.O., FADY,C., REISSER, D., LAGADEC, P. and JEANNIN, J.F., Differential sensitivity to natural cell-mediated cytotoxicity of two rat colon adenocarcinoma variants differing in their tumorigenicity:identification of the effector cells as natural killer cells. Cancer Immunol. Immunother., 26,263-268 (1988). C., SOLANA, R., SERRANO, R., CARRACEDO, J. and PENA,J., ALONSO, RAMIREZ, R., Natural killer susceptibilityis independent of HLA class I antigen expression on cell lines obtained from human solid tumors. Europ. J. Immunol., 20,2445-2449 (1990). ROBERTSON, M.J. and R I ~ J.,, Biology and clinical relevance of human natural killer cells. Blood, 76,2421-2438 (1990).
664
BLOTTIEREET AL.
SHIMIZU, T., PELLETIER, H., HAMMANN, A., OLSON, N.O., MARTIN, M.S. and MARTIN, F., Effects of a single injection of anti-asialo GM1 serum on natural cytotoxicity and the growth of a regressive colonic tumor in syngeneic rats. Int. J. Cancer, 40,676-680 (1987). STAM,N.J., UST, W.M., VOORDOUW, A.C., PASTOORS, L.B., VAN DER HOEVEN,F.A., MELIEF,C.J. and PLOEGH,H.L., Lack of correlation between levels of MHC class I antigen and susceptibility to lysis of small cellular lung carcinoma (SCLC) by natural killer cells. J. Immunol., 142,4113-4117 (1989). STORKUS, W.J., ALEXANDER, J., PAYNE,J.A., CRESSWELL, P. and
DAWSON,J.R., The d / a 2 domains of class I HLA molecules confer resistance to natural killing. J. Immunol., 143,3853-3857 (1989). STORKUS,W.J. and DAWSON,J.R., Target structures involved in natural killing (NK): characteristics, distribution and candidate molecules. CriL Rev. Immunol., 10,393416 (1991). VERSTEEG, R., PELTENBURG, L.T.C., PLOMP,A.C. and SCHRIER, P.I., High expression of the c-myc oncogene renders melanoma cells prone to lysis by natural killer ce1ls.J. Immunol., 143,43314337 (1989). VUJANOVIC, N.L., HERBERMAN, R.B. and HISERODT, J.C., Lymphokineactivated killer cells in rats: analysis of tissue and strain distribution, ontogeny, and target specificity. Cancer Rex, 48,878-883 (1988).
