Eur. J. Immunol. 1990. 20: 171-177
Knut Sturmhofel and Gunter J. Hammerling Institute for Immunology and Genetics, German Cancer Research Center, Heidelberg
Reduction of natural killer susceptibility by gene transfection
171
Reconstitution of H-2 class I expression by gene transfection decreases susceptibility to natural killer cells of an EL4 class I loss variant Several reports have suggested that an inverse correlation exists between major histocompatibility complex class I expression and the susceptibility to natural killer (NK)-mediated lysis. For example, the increased class I expression induced by interferon-y was always accompanied by an increased resistance to NK lysis. Likewise, class I loss variants were often more NK susceptible than their normal counterparts. To investigate whether the inverse correlation between class I expression and NK susceptibility was fortuitous or whether the class I molecules were directly responsible for this effect we resorted to gene transfection studies. From the murine thymoma line EL4 an H-2Db- and Kb-negative variant S3 was selected. This variant was highly susceptible to NK lysis. S3 was found to have a defect in P2-microglobulingene expression.Therefore, restoration of Dband Kb expression could be achieved by transfection with the P2-microglobulin gene. This resulted in a strong decrease in susceptibility to NK lysis to the level of the H-2+ parental EL4. Transfection with class I1 genes had no effect. Blocking of the class I molecules on the H-2+ cells with anti-H-2b F(ab’)2 fragments increased the susceptibility to NK cells to the level of the H-2- variant S3. These data demonstrate that the class I molecules on the targets are directly responsible for regulation of NK susceptibility but the mechanism is not clear. Possibly the class I molecules interfere with the unknown NK target structures.
1 Introduction Tumor cells can be rejected by either MHC-restricted T lymphocytes or by non-MHC-restricted NK cells.The latter ones are viewed as a more primitive defense system against tumors [l]. No definitive information is presently available about the receptors on NK cells and their putative target structures [2-41. In spite of the fact that NK cells are non-MHC restricted, a number of reports in human and murine systems have indicated that MHC class I molecules on the target cells may play an important role for the regulation of susceptibility to NK-mediated lysis. For instance, class I loss variants of tumor cells were frequently found to be lysed more efficiently by NK cells in vitro and in vivo than their class I+ counterparts [5-121. It was also shown by many investigators that IFN-y-mediated enhancement of class I expression on the target cells was paralleled by a reduction of their NK susceptibility [13-191. This inverse correlation between NK susceptibility and MHC class I expression could have been merely fortuitous because numerous reports also exist in which no significant correlation between MHC expression and reduction of NK susceptibility was observed [20-261. Recently, in studies with human NK cells it was reported that the transfection of human NK target cells with class I genes resulted in a
[I 79811 Correspondence: Knut Sturmhofel,Institut fur Immunologie und Genetik, Deutsches Krebsforschungszentrm, Im Neuenheimer Feld 280, D-6900 Heidelberg, FRG Abbreviations: ADCC: Antibody-dependent cytotoxic cells &m: Pz-Microglobulin MFV: Mean fluorescence value neoR:
Neo(G418) resistance 0
VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1990
decrease of their NK susceptibility, showing that in these cases indeed the HLA molecules were responsible for this effect [27-291. In order to investigate whether MHC molecules could also affect NK lysis in other species we reconstituted the expression of class I molecules on a class I- murine tumor variant by gene transfection and determined the effect on NK lysis. For this purpose the C57BL/6-derived thymoma line EL4 was chosen which was relatively resistant t o NK cells. An H-2 loss variant of the EL4 cells was found to be highly susceptible to NK lysis whereas H-2+ transfectants were again resistant to NK cells.These observations suggest that the murine H-2 class I molecules can directly influence NK lysis.
2 Materials and methods 2.1 Cell lines and media EL4.Rob (a gift of I? Robinson, German Cancer Research Center, Heidelberg, FRG) is a clone of the murine thymoma EL4 of C57BL/6 origin. It was typed CD2- and FcR-. Variants negative for surface class I molecules were selected from EL4.Rob by two treatments with anti-Db and -Kb antibodies plus rabbit anti-mouse C (Low-tox rabbit anti-mouse C; Cedarlane, Hornby, Ontario, Canada) followed by several selections by FCM sorting (Ortho 50H, Ortho Diagnostics, Westwood, MA). The H-2- cells were cloned by LD. The clone S3 was negative for surface expression of Db and Kb as shown by FCM analysis. Clone C2S6.6 was negative for Kb and weak for Db. The H-2variants C6, C10 and C19 were gifts of l? Brams (Receptron, Concord, CA) and were also EL4.Rob derived. EL4.Rob grew as a semi-adherent line,whereas all other cell lines and OO14-2980/90/0101-0171$02.50/0
172
Eur. J. Immunol. 1990. 20: 171-177
K. Sturmhofel and G. J. Hammerling
transfectants grew adherent in DMEM supplemented with 7% FCS (Gibco, Grand Island, NY) and L-glutamine (Gibco). The cells were kept in culture for seveal months without changes of MHC class I expression or NK susceptibility. Recombinant mouse IFN-y was kindly provided by Dr. Svetly (Boehringer, Vienna, Austria). 2.2 Antibodies and production of 2&&6S-F(abr)2 fragments For FCM analysis and immunoprecipitation the following mouse mAb and sera were used: €322-249: anti-Db [30]; K10-56.1: anti-Kb [31]; 28-8-6s: anti-Db and -Kb [32]; T19-191: anti-Db (G. J. Hammerling, unpublished); 6/68: anti-Thy-1.2 [33]; Lym11.2: anti-P2-microglobulin (p2m) [34]; rabbit anti-rnouse-fl2m serum, recognizing the free P2m protein [35]; 17/227: anti-1-Ab [36]. F ( a b ' ) ~fragments of the mAb 28-8-6s were produced according to the method of Parham [37] and Lamoyi and Nisonoff [38]: 2 mg/ml of antibody was digested with 45 pg/ml of pepsin for 8-9 h at 37°C. The cleavage products were separated over a Sephadex G-150 (Pharmacia, Freiburg, FRG) column. The F(abr)2 peak was additionally purified over a protein ASepharose column to eliminate contaminations of undigested complete antibody. The purity of the F(ab')2 preparation was verified on a non-reducing SDS-PAGE gel which was silver stained. 2.3 Detection of surface molecules by FCM analysis About 1 x lo6 cells were incubated with 50 p1 SN of the appropriate mAb for ' h h at 4 "C, washed twice and stained with a 1/25 diluted goat anti-mouse-IgG-FITC conjugate (Sigma, St. Louis, MO) for l/2 h at 4°C. After washing, the cells were directly analyzed in a "FACScan" (Becton Dickinson, Mountain View, CA). The fluorescence intensity was measured in a 4-decade log-scale. A mean fluorescence value (MFV) within the first decade represented negative cells.
2.5 Immunoprecipitation of cytoplasmic
am proteins
For detection of cytoplasmic P2m products [35S]methioninelabeled proteins were precipitated with specific mAb and separated on 13%-18% gradient SDS-PAGE gels as recently described [44]. In short, cells were kept for 40 min in methionine-free medium at 37 "C, then incubated with 200 pCi (7.4 x lo6 Bq)/5 ml [3sS]methioninefor 3-4 h and finally lysed with a buffer containing 1% Triton-X, 100 mM NaCI, 5 mM MgC12, 20 mM Hepes (Gibco), 1mM proteinase inhibitor PMSF (Sigma). The lysate was preincubated with protein A-Sepharose for l/2 h. P2m proteins were precipitated from the lysate with the mAb Lymll.2 or rabbit anti-mouse P2m antiserum followed by protein A-Sepharose (1 h). 2.6 NK assay NK susceptibility of the tumor cells was measured in a standard SICr-releaseassay as described [45]. NK cells from three pooled spleens of 5-8 week-old CBA/J mice (Wiga, Hannover, FRG) were used. The NK cells were induced in vivo by i.p. injection of 150 pg poly(1) * poly(C) (Sigma) 18-22 h before the assay. The SC were isolated, erythrocytes lysed and the cells incubated on a 16-cm tissue culture dish for 11/2h for removal of plastic-adherent cells. The nonadherent cells were used as effectors. Target cells were labeled with Naz51Cr04 (NEN, Dreieich, FRG) for 11/2 h, washed twice and adjusted to 5 x 1@cells/100 pl as described [43]. Effectors and targets were incubated at different effector : target ratios, starting with 200 : 1, for 5 h at 37 "C. Then the cells were sedimented by centrifugation, 100 p1 of SN was collected and released W r was counted. Percent NK lysis was calculated by the formula: YOLysis = (cpm of probe - spontaneous release)/(maximum release - spontaneous release).
3 Results 3.1 Loss of class I surface expression on S3 is caused by a Ptm defect
2.4 Gene transfection and genes Two different protocols for gene transfection were applied. The first method was transfection with DNA-calcium phosphate precipitates according to Wigler et al. [39]. Transfection by electroporation [40] proved to be more efficient for the S3 cells. Briefly, a Bio-Rad-"gene pulser" (Bio-Rad Labs., Richmond, CA) was used with a capacity of 960 pF and voltages in the range of 190-250 V. Cells (1 X 107-2 X lo7) in 200 p1 PBS were cotransfected with about 30 pg of the appropriate MHC gene and 5 pg of the neoR gene. The neoR gene confers the resistance to geniticin. The neoR gene was either on a separate plasmid (pGB002, [41]) or combined with the @2mbgene on one plasmic (P2m/neoR). The Pzm/neoR plasmid was constructed by I? Robinson and contained a 10.9-kb long genomic Pzm DNA fragment and the neoR gene within the pACYC184 vector. Selection was started 1 or 2 days after the transfection process with 1.0 mg/ml geniticin (G418, Gibco) and was continued until resistant clones came up after about 14 days. The transfected MHC genes were Db [42]; Dd [43]; 1-A!, 1-A; [36].
To test the hypothesis that class I surface molecules on the target cells influence the NK susceptibility class I loss variants were selected from the ELA.Rob thymoma cell line. The H-2- variant S3 was selected from EL4.Rob as described by treatment with anti-H-2b antibodies plus C, followed by cell sorting and cloning.The total loss of Dband Kb surface molecules on the S3 cells was found to be due to a defect of P2m expression. The respective evidence was obtained from the following observations: (a) the transfection of the P2mbgene into S3 resulted in Db- and Kb-positive cells (see below) whereas the transfection of class I genes alone such as the Dbgene without P2m did not yield surface class If cells. (b) In immunoprecipitations of [%]methionine-labeled S3 cells no P2m protein could be precipitated either with the mAb Lymll.2, which binds to the P2mb protein associated with the class I H chain, or with a rabbit anti-mouse P2m antiserum which precipitates the free, unassociated P2m chain (Fig. 1). The EL4.Rob cells and all P2m transfectants which were positive for cell surface class I containing high amounts of cytoplasmic P2m protein (Fig. 1). Some neo-resistant transfectants, obtained after
Eur. J. Immunol. 1990. 20: 171-177
A s
B 82
s
02
Reduction of natural killer susceptibility by gene transfection
C
D
s
s
E
s
similar defect of P2m expression seemed to be responsible for loss of surface class I expression in the independently selected clones C6, C19, C19.This was suggested by similar immunoprecipitation data and transfections with p2m/neoR (data not presented). Another class I loss variant, clone C2S6.6, was found to be defective for Kb expression only. It was still positive for Db.
02
kD
k h
46-
173
-46
3.2 Reconstitution of Db/Kbexpression on S3 cells by gene transfection
R2m 14
-14
--2m
Figure I. Immunoprecipitation of the H-2-positive and -negative parental lines and of P2m transfectants. All lanes are derived from the same immunoprecipitation experiment. Lanes A (EL4.Rob) and B (S3) are from the same SDS-gel, lanes C, D (37.2.1,37.2.2, respectively, both positive for surface Db and Kb), and lane E (37.2.4, neoR,but negative for surface class I) were from another gel run in parallel. Lanes irrelevant for the present study were cut from the photograph. In lanes indicated with “S” precipitation was performed with the rabbit anti-mouse P2m antiserum recognizing only free P2m chains. In lanes marked with “By proteins were precipitated with the mAb Lymll.2 recognizing Pzm associated with H-2. The band at 12 kDa represents the P2m protein. In all precipitations with the rabbit anti-p2m antiserum, a 14-kDa band was also detected which appeared to be unrelated to P2m [34].The bands at about 46kDa represent the H chains of class I molecules.
transfection of the p2m/neoRplasmid, did not express the P2m gene. This is presumably due to inefficient integration of the transfected genes. These cells remained negative for surface class I expression. P2m or surface class I expression could not be induced with IFN-y as discussed later. A
S3 was transfected either with the genes for P2mb alone or with the genes for Dband P2mb.Transfectionwas performed with the DNAkalcium phosphate precipitation technique or by electroporation. D b and Kb surface expression of the resulting clones was measured by FCM analysis. All class I+ transfectants expressed at least as much class I as the parental EL4.Rob cells. S3 cells transfected with P2rn plus Db gene did not express higher levels of Db than the cells transfected with the P2m gene alone. The MFV of class I expression of the EL4.Rob and the S3 cells and the transfectants are summarized in Table 1. This table also shows the normalized values for NK lysis which will be discussed below. Representative FCM diagrams for the H-2 class I expression of EL4.Rob and three transfectants are shown in Fig. 2.
The NK susceptibility of the transfectants was measured in an NK assay as described. In each experiment the S3 and the EL4.Rob cells were included as high and low controls, respectively. A representative experiment is shown in Fig. 3. Here the NK lysis for the H-2- variant S3 was 46% and for the H-2- transfectant E45.5 about 50%. In contrast, the parental EL4.Rob cells and the H-2+ transfectants E45.41 and E45.44 were lysed t o only 15%-21%. This approximately two- to three-fold difference between H-2+ and H-2- EL4 cells was observed in all experiments
Table 1. Description of ELA.Rob, S3 and different Pzm transfectants: their class I expression and % NK lysis relative to S3 (summary of different experiments) Cell line EL3.Rob s3 32.2.1 37.2.2 37.2.4 E6.10 E6.8 E35.31 E45.41
E15.44 E35.1 E45.5 E19.7
MHC class I antigens expressed
MFV Dbb’
MFV
-
Db, Kb
Yo
Dh f12m/neoR Dh +p2m/neoR Dh + f12m/neoR Dh +p2m/neoR Db +P2m/neoR P2m/neoR P2m/neoR Pzm/neoR p2m/neoR f12m/neoR Dd + P2m/neoR
Db, Kb Dh, Kb
136 3 173 135 4 159 3 172 130 144 4 3 107
Transfect.a) procedure
Transfected genes
-
-
CaP CdP CaP E.P. E.P. E.P. E.P. E.P. E.P. E.P. E.P.
+
--
_ _
Db. Kh
- -
Dh, Kh Dh, Kb Dh, Kb
_ -
- -
Db. Kb. Dd
Kb”
2 87 75 3 82
2 121 70 101 4
2 77
Mean‘) % NK lysis relative to S3 56 (14) 100 (14) 39 (6) 4s (3) 101 (2) 61 (6) 117 (3) 23 (2) 53 (5) 26 (3) 95 (1) 110 (1) 27 (2)
MFV Dd: 84
a) Cap: DNA-calcium phosphate precipitation; E.P.: electroporation. b) MFV calculated from 5 to 20 FACS analyses. c) Percent NK lysis of S3 was set to 100% in each experiment. For each target the value of NK lysis was related to S3.The arithmetical mean of relative NK lysis was calculated from all experiments performed; number of experiments given within parentheses.
Eur. J. Immunol. 1990. 20: 171-177
K. Sturmhofel and G. J. Hammerling
174
The lysis of the other target cells was calculated in relation to the lysis of S3. Table 1 indicates also the number of experiments performed for each cell line. A consistent picture is obtained: whenever the cells express class I antigens on the cell surface they are less susceptible to NK lysis than their H-2- counterparts.
lcq fluorescence
3QQ
1
300
1
-
.:' .
€45.44
. .
. . .. ..
.. .. ..
..
1- .: '.. ,: i' : .- 2
is,
ii
i '
i
lcq -lf
Figure 2. Measurement of Dbsurface expression of EL4.Rob and three P2m transfectants by FCM analysis. Cells were labeled with the anti-Dbantibody B22-249. I n each diagram the H-2- S3 cells are included as negative control (dotted line). E45.41 and E45.44 are class I+ Pzm transfectants, E45.5 is a neo-resistant but surface class I- transfectant.
In Fig. 4 these NK values are plotted over the mean fluorescence intensity obtained for total class I expression. The data suggest an inverse linear relationship between the amount of class I expression and susceptibility to NK lysis. Additionally, the clone C2S6.6 which expressed no Kb but intermediate amounts of Dbon the cell surface exhibited an NK susceptibility intermediate to S3 and EL4.Rob and therefore fitted well into this correlation. The figure also presents the independently derived class I- EL4 variants C6, C10 and C19 which were as efficiently lysed as S3. We further investigated whether transfection of class I1 genes into the class 11- S3 line would alter its NK susceptibility. S3 was cotransfected with genes for the I-Aba and @ chains. Surface expression of I-Ab was demonstrated with the mAb 17/227 which recognizes only the a/@-assembled I-A molecule.The resulting clones were still negative for Db and Kbon the surface but positive for class 11. No difference in NK lysis between S3 and the I-A transfectants was found (Fig. 5).
3.3 IFN-y treatment does not influence NK lysis In many studies it was reported that an increase in the amount of class I antigens induced by IFN-y was always accompanied by a decrease of NK susceptibility [9, 14, 16, 19,201.To examine whether IFN-y would decrease NK lysis in the absence of altered class I expression the cells were
40
%
'$
..__ -..
60%-
e
b
effectortarget ratio
.I?
k
40% -
e
Figure3. NK experiment with the targets EL4.Rob ( O ) , S3 (0) and the same P2rn transfectants as in Fig. 2: E45.41 (m), E45.44 (A),(both class I) and E45.5 (0) (class I-). The E/T was downtitrated in four steps from 200 : 1 to 0.5 : 1. The % NK lysis was calculated as described in Sect. 2.6 and was not normalized as in Table 1.
and with all transfectants. The degree of NK lysis ranged from about 29%-49% for the S3 cells with an arithmetic mean of 37% depending on the preparation of the NK cells. However, in the very same experiments the lysis of the H-2+ transfectants was always reduced by the factor of two or more compared to the H-2- S3 cell 1ine.The respective NK data are compiled in the right hand part of Table 1in which the degree of NK lysis was normalized: for each experiment . the percentage of NK lysis of the S3 cells was set to 100%.
g 20%K 0%
'
0
50
100
150
200
I
I
I
250
300
350
t o t a l class I expression /MFV
Figure 4. Mean % NK lysis relative to S3 (S3 = 100%)is plotted vs. amount of total class I expression.The mean values for % NK lysius and class I expression of the different target cells are compiled in Table 1. The amount of total class I cell surface expression was calculated by addition of the MFV data of Dband Kb. ( 0 )EL4.Rob; (0)S3; (x) different class I- transfectants; (+) different class I+ transfectants; (A) clone E19.7 transfected with Dd plus Pzm, expressing D, Kb and Dd. Additionally the H-2clones C6, C10 and C19 (V) and the Kb-negative but Db-positive clone C2S6.6 (0)are included in this graph. The dashed line represents the calculated linear trend for the relationship between increasing amount of class I expression and decreasing NK lysis.
Eur. J. Immunol. 1990. 20: 171-177
Reduction of natural killer susceptibility by gene transfection
175
Table 2. Influence of IFN-y Treatment on class I expression and NK lysis of class I-negative and -positive cellsa)
s3 ELA.Rob 37.2.1
6o
3 85 150
2 65 64
39 19 17
4 93 152
One experiment out of three; all experiments gave similar results. Brget cells were incubated with rFN-y (60U/ml) for 68 h. MFV intensity: Db labeled with mAb B22249, Kb with mAb K10-56.1. Value for % NK lysis is given in absolute numbers not related to S3 as in Table 1.
2 72 69
r
0
effector:target ratio
0.4
10
50
amount of antibody [ ,ugg/ml]
Figure 5. NK susceptibility of S3 class I1 transfectants. H-2 class I- and class 11- S3 cells (0)were cotransfected with the genes for IA: and IAI. The obtained transfectants were tested in a standard NK assay. The results of five transfectants are presented: IAdp-positive clones: (V) E23.2.17; (0)E23.2.18; (A) E23.11.6; (0)E23.11.13; IAdb-negative clone: ($) E23.11.15. EL4.Rob ( 0 )and the P2m transfectant E45.41 (B) were used asclass 11-, but class I+ controls.
Figure 6. Effect of blocking the target cell class I molecules by anti-class I antibody on NK susceptibility.The class I+ transfectant 37.2.1 (B) and the S3 cells (0) were incubated with different amounts of F(ab’)*fragments of the anti-DbKbmAb 28-8-6s (solid line) prior to the NK assay. For control the target cells were treated with a non-class I antibody in the same way (dashed line; anti-Thy-1.2 mAb 6/68).
treated with 60 U/ml of mouse rIFN-y for about 60-70 h. For the class I- S3 line no induction of class I expression and no change of NK susceptibility was observed (Table 2). This unaltered reaction of S3 was not due to an unresponsiveness to IFN-y because Ly-6.2 surface expression and the level of mRNA for Db and Kb were increased (data not shown). Likewise, IFN-y did not increase class I expression on the parental EL4.Rob or on the transfectant 37.2.1, probably because class I was already maximally expressed in these cells. Again no change in NK susceptibility was observed (Table 2).
EL4.Rob or class I+ transfectants were incubated with 28-8-63 F(ab’)z fragments prior to the NK assay the NK susceptibility was drastically enhanced and reached the level of the H-2- variant S3 (Fig. 6). As expected, the 28-8-68F(ab’)2 fragments did not increase the lysis of the S3 cells. An antibody against the non-MHC surface antigen Thy-1.2 was used as control; 6/68 is of the IgM class and therefore not recognized by ADCC. EU.Rob, S3 and the transfectants are all Thy-1.2+ (data not shown). Preincubation of the target cells with the 6/68-antibody did not influence NK susceptibility (Fig. 6).
3.4 Masking of class I antigens by an anti-class I antibody renders target cells more susceptible to NK lysis
4 Discussion
To demonstrate that indeed the class I molecules are involved in the reduction of NK susceptibilitywe attempted to block the class1 molecules of the target cells with a class I-specific mAb. In order to avoid lysis mediated by antibody-dependent cytotoxic cells (ADCC), F(ab’)~fragments of the anti-Db/KbmAb 28-8-68were prepared.When
In numerous studies an inverse correlation between NK lysis and MHC class I expression on the target cells was found. First, class I loss variants were frequently more sensitive to NK-mediated lysis [5-12,461 and second, enhancement of class I expression by IFNy was always accompanied by a decrease in NK susceptibility [13-191. The present report demonstrates that the inverse correlation between H-2 class I expression and NK lysis is not
176
Eur. J. Immunol. 1990. 20: 171-177
K. Sturmhofel and G . J. Hammerling
fortuitous. In all experiments the NK lysis of the H-2+ parental EL4.Rob cells and transfectants was always decreased by a factor of two to three in comparison to the H-2- EL4 variant S3, or the H-2- transfectants.These data demonstrate that the class I molecules on the target cells are directly involved in the reduction of NK lysis. In addition, the Kb-negative, but Db-positive variant C2S6.6 exhibited an intermediate amount of class I expression and an intermediate degree of NK lysis. It should also be noted that the independently derived EL4.Rob class I loss variants C6, C10 and C19 were as highly NK susceptible as the S3 variant. Their loss of class I expression was also caused by a defect in Pzm expression. Recently, it has been reported for several H-2- variants of two murine tumor cell lines that they had lost class I expression due to distinct defects in Pzm expression and that they were all highly NK susceptible [lo]. In contrast, cell surface expression of class I1 molecules obtained after transfection of I-A genes did not alter the NK susceptibility of the S3 cell 1ines.Thisis in agreement with observations made with human B cell lymphomas [12,281. The H-2- variant S3 failed t o express class I surface antigens after in vitro treatment with IFN-y. However, S3 could respond to IFN-y as demonstrated by the increase of class I-specific mRNA and cell surface expression of the Ly-6.2 antigens. Therefore, the S3 cells offered the opportunity to test directly the influence of IFN-y on NKmediated lysis. No change of NK-mediated lysis was observed after treatment with IFN-y (Table 2). Likewise, on the H-2+ parental EL4.Rob cells and transfectants no significant increase of class I expression could be observed after IFN-y treatment, probably because the expression of class I antigens on the surface of these cells was already saturated. Again, NK lysis of the IFN-treated cells was not changed. These observations support the notion that the decreased NK susceptibility obtained after treatment with IFN-y, which was observed in many studies [14-171, is indeed mediated by enhancement of MHC class I expression. While this work was in progress, other investigators reported for the human NK system that the transfection of HLA class I genes into human lymphoblastoid B cell lines reduced their NK susceptibility [27-291. To our knowledge no such transfection studies have been performed in other species. Such additional studies are even more important in the light of the observations by Storkus et al. [29] that only the transfection with HLA but not with H-2 genes (DP and Kb) reduced the lysis of the human NK targets by human NK effector cells. In contrast, the present study shows that H-2 can indeed reduce NK susceptibility. It is not clear why the effect of H-2 is only found in a murine but not in a human NK system. Possibly, this reflects a difference in the nature of mouse and human NK target structures. The mechanism by which class I molecules influence NK-mediated lysis is not clear. It is possible that the MHC molecules on the target surface interact with the putative target structures for the NK effector cells. This interaction could change the conformation and the accessibility of the target structures in such a way that they can no longer be recognized by NK cells. This model is also supported by our finding that blocking of the class I antigens on the target cells with anti-Db/Kbantibodies rendered these cells highly
sensitive to NK-mediated lysis. Possibly, the binding of the antibodies with the class I molecules prevented their interaction with the NK target structures. In other reports no enhancement of NK lysis was observed after the incubation of the target cells with anti-class I antibodies [12, 201.These discrepancies could be due to a lower affinity of the antibodies used, or to the possibility that the respective antibodies were directed against class I epitopes which were not involved in the interaction between MHC molecules and NK target structures. The concept of MHC molecules interfering with putative NK target structures is supported by the evidence that MHC class I molecules interact with other cell surface structures such as, for example, the receptors for insulin, epidermal growth factor and glucagon [47-511. Different domains of the MHC class I molecules are known to have different functions. Thus, the al and a2 domains form the peptide-binding groove [52] whereas the a3 domain has been found to interact with the CD8 molecule [53,54]. It will be of interest to determine which part of the MHC class I molecule is responsible for the proposed interaction with the NK target structures. It is also conceivable that the efficiency of such interactions could depend on the respective MHC class I alleles. The proposed model of the interference of MHC with NK target structures would also explain why the effect of MHC class I is only partial and does not lead to a complete reduction of NK lysis, and why in several reports no influence of class I on NK lysis was observed [20-261. It is conceivable that the amount of class I is not sufficient to block all NK target structures, or that the target structures are heterogeneous and not all of them can interact with class I molecules or the respective class I alleles. We thank N. Bulbuc for skilful technical assistance. Received September 27, 1989.
5 References 1 Herberman, R. B., Reynolds, C.W. and Ortaldo, J. R., Annu. Rev. Immunol. 1985. 4: 651. 2 Ortaldo, J. R. and Reynolds, C. W., J. lmmunol. 1989. 138: 4545, 3 Fitzgerald-Bocarsly, I?, Herberman, R. B., Hercend, T., Hiserodt, J., Kumar, V., Lanier, L., Ortaldo, J. R., Pross, H., Reynolds, C. W.,Welsh, R. M. and Wigzell, H., in Ades, E. W. and Lopez, C . (Eds.), Natural Killer Cells and Host Defense, Karger, Basel 1989, p. 13. 4 Ortaldo, J. R . , Kantor, R., Segal, D., Bolhuis, R. H. and Bino, T.,in Ades, E.W. and Lopez, C. (Eds.), Natural Killer Cells and Host Defense: 5th International Natural Killer Cell Workshop, Karger, Basel 1989, p. 221. 5 Taniguchi, K., Karre, K. and Klein, G., lnt. J. Cancer 1985.36: 503. 6 Karre, K., Lunggren, H.-G., Piontek, G . E. and Kiessling, R. W., Nature 1986. 319: 675. 7 Ljunggren, H.-G., OhlCn, C., Hoglund, I!,Yamasaki,T., Klein, G . and Karre, K . , J. Immunol. 1988. 110: 671. 8 Kawano, Y.-I., Xiniguchi, K., Karre, K., Toshitani, A . and Nomoto, K., Cell. Immunol. 1988. 111: 341. 9 Ohlen, C., Bejarano, M.-T., Gronberg, A., Torsteinsdottir, S., Franksson, L., Ljunggren, H.-G., Klein, E., Klein, G. and Karre, K., J. lmmunol. 1989. 142: 3336.
Eur. J. Immunol. 1990. 20: 171-177
Reduction of natural killer susceptibility by gene transfection
10 Ljunggren, H.-G., Paabo, S., Cochet, M., Kling, G., Kourilsky, P. and Karre, K., J. Irnmunol. 1989. 142: 2911. 11 Harel-Bellan, A., Quillet, A., Marchiol, C., DeMars, R.,Tursz, T. and Fradelizi, D., Proc. Natl. Acad. Sci. USA 1986. 83: 5688. 12 Storkus,W. J., Howell, D. N., Salter, R. D., Dawson, J. R. and Cresswell, P., J. Immunol. 1987. 138: 1657. 13 Hansson, M., Kiessling, R. W., Anderson, B. and Welsh, R. M., J. Immunol. 1980. 125: 2225. 14 McMillan,T. J., Everett, C. A. and Hart, I. R., Int. J. Cancer 1987. 40: 659. 15 Welsh, R. M., Karre, K., Hansson, M., Kunkel, L. A. and Kiessling, R. W., J. Immunol. 1981. 126: 219. 16 Piontek, G. E., Taniguchi, K., Ljunggren, H.-G., Gronberg, A., Kiessling, R. W., Klein, G. and Karre, K., J. Immunol. 1985. 135: 4281. 17 Gronberg, A., Ferm, M. T., Ng, J., Reynolds, C. W. and Ortaldo, J. R., J. Immunol. 1988. 140: 4397. 18 Zoller, M., Strubel, A., Hammerling, G. J., Andrighetto, G., Raz, A. and Ben-Ze'ev, A., lnt. J. Cancer 1988. 41: 256. 19 Hammerling, G. J., Klar, D., Maschek, U., Piilm, W. and Sturmhofel, K., in David, C. S. (Ed.), H-2 Antigens: Genes, Molecules, Function, NATO AS1 Series A: Life Sci. vol. 144, Plenum Publ. Corpo., New Yok 1987, p. 705. 20 Gorelik, E., Gunji,Y and Herberman, R. B., J. Zmmunol. 1988. 140: 2096. 21 Chervenak, R. and Wolcott, M., J. Immunol. 1988. 140: 3712. 22 Dennert, G., Landon, D., Lord, E. M., Bahler, D. W. and Frelinger, J. G., J. lmmunol. 1988. 140: 2472. 23 Leiden, J. M., Karpinski, B. A., Gottschalk, L. and Kornbluth, J., J. Immunol. 1989. 142: 2140. 24 Stam, N. J., Kast,W. M.,Voordouw, A. C., Pastoors, L. B.,Van der Hoeven, F., Melief, C. J. M. and Ploegh, H. L., J. Irnmunol. 1989. 142: 4113. 25 Nishimura, M. I., Stroynowski, I., Hood, L. and OstrandRosenberg, S., J. Immunol. 1988. 141: 4403. 26 Gopas, J., Segal, S., Hammerling, G. J., Bar-Eli, M. and Rager-Zisman, B., Immunol. Lett. 1988. 17: 261. 27 Quillet, A., Presse, F., Marchiol-Fournigault,C., Harel-Bellan, A., Benbunan, M., Ploegh, H. L. and Fradelizi, D., J. Immunol. 1988. 141: 17. 28 Shimizu, Y. and DeMars, R., Eur. J. Immunol. 1989. 19: 447. 29 Storkus, W. J., Alexander, J., Payne, J. A., Dawson, J. R. and Cresswell, P., Proc. Natl. Acad. Sci. USA 1989. 86: 2361. 30 Lemke, H., Hammerling, G. J. and Hammerling, U., Immunol. Rev. 1979. 47: 175. 31 Hammerling, G. J., Riisch, E., B d a , N., Kimura, S. and
177
Hammerling, U., Proc. Nut[. Acad. Sci. USA 1982. 79: 4737. 32 Ozato, K. and Sachs, D. H., J. Immunol. 1981. 126: 317. 33 Hammerling, G. J., Lemke, H., Hammerling, U., Hohmann, C., Wallich, R. and Rajewski, K., Curr. Top. Microbiol. Immunol. 1978. 81: 100. 34 Margulies, D. H., Parnes, J. R., Johnson, N. A. and Seidman, J. G., Proc. Natl. Acad. Sci. USA 1983. 80: 2328. 35 Robinson, P. J., Graf, L. and Sege, K., Proc. Natl. Acad. Sci. USA 1981. 78: 1167. 36 Landais, D., Beck, B. N., Buerstedde, J.-M., Degraw, S., Klein, D., Koch, N., Murphy, D., Pierres, M., Tada, T., Yamamoto, K., Benoist, C. and Mathis, D., J. Immunol. 1986. 137: 3002. 37 Parham, P., J. Immunol. 1983. 131: 2895. 38 Lamoyi, E. and Nisonoff, A., J. Immunol. Methods 1983.56: 235. 39 Wigler, M., Pellicer, A., Silverstein, S., Axel, R., Urlaub, G. and Chasin, L., Proc. Natl. Acad. Sci. USA 1979. 76: 1373. 40 Doffinger, R., Pawlita, M. and Sczakiel, G., Nucleic Acids Res. 1988.16: 11840. 41 Brady, G., Jantzen, H. M., Bernard, H. U., Brown, R., Schiitz, G. and Hashimoto-Gotoh, T., Gene 1984. 27: 223. 42 Allen, H.,Wraith, D., Pala, P., Askonas, B. and Flavell, R. A., Nature 1984. 309: 279. 43 Margulies, D. H., Evans, G. A., Ozato, K., Camerini-Otero, R. D. ,Bnaka, K., Appella, E. and Seidman, J. G., J. Immunol. 1983. 130: 463. 44 Klar, D. and Hammerling, G. J., EMBO J. 1989. 8: 475. 45 Wigzell, H. and Ramstedt, U., in Weir, M. (Ed.), Handbook of Experimental Immunology; vol.2 CellularImmunology, Blackwell Scientific Publications, Oxford 1986, chapter 60.1. 46 Kawano, Y.-I., Taniguchi, K., Toshitani, A. and Nomoto, K., J. Immunol. 1986. 136: 4729. 47 Kittur, D., Shimizu,Y.,DeMars, R. and Edidin, M., Proc. Natl. Acad. Sci. USA 1987. 84: 1351. 48 Verland, S., Simonsen, M., Gammeltoft, S., Allen, H., Flavell, R. A. and Olsson, L., J. lmmunol. 1989. 143: 945. 49 Edidin, M., lmmunol. Today 1983. 4: 269. 50 LaFuse,W. and Edidin, M., Biochemistry 1980. 19: 49. 51 Phillips, M. L., Moule, M. L., Delovitch,T. L. and Yip, C. C., Proc. Natl. Acad. Sci. USA 1986. 83: 3474. 52 Bjorkman, P. J., Saper, M. A., Samraoui, B., Bennett, W. S., Strominger, J. L. and Wiley, D. C., Nature 1987. 329: 506. 53 Potter,T. A., Rajan,T.V, Dick, R. F., I1 and Bluestone, J. A., Nature 1989. 337: 73. 54 Salter, R. D., Norment, A. M., Chen, B. P., Clayberger, C., Krensky, A. M., Littman, D. R. and Parham, P., Nature 1989. 338: 345.
Knut Sturmhofel and Gunter J. Hammerling Institute for Immunology and Genetics, German Cancer Research Center, Heidelberg
Reduction of natural killer susceptibility by gene transfection
171
Reconstitution of H-2 class I expression by gene transfection decreases susceptibility to natural killer cells of an EL4 class I loss variant Several reports have suggested that an inverse correlation exists between major histocompatibility complex class I expression and the susceptibility to natural killer (NK)-mediated lysis. For example, the increased class I expression induced by interferon-y was always accompanied by an increased resistance to NK lysis. Likewise, class I loss variants were often more NK susceptible than their normal counterparts. To investigate whether the inverse correlation between class I expression and NK susceptibility was fortuitous or whether the class I molecules were directly responsible for this effect we resorted to gene transfection studies. From the murine thymoma line EL4 an H-2Db- and Kb-negative variant S3 was selected. This variant was highly susceptible to NK lysis. S3 was found to have a defect in P2-microglobulingene expression.Therefore, restoration of Dband Kb expression could be achieved by transfection with the P2-microglobulin gene. This resulted in a strong decrease in susceptibility to NK lysis to the level of the H-2+ parental EL4. Transfection with class I1 genes had no effect. Blocking of the class I molecules on the H-2+ cells with anti-H-2b F(ab’)2 fragments increased the susceptibility to NK cells to the level of the H-2- variant S3. These data demonstrate that the class I molecules on the targets are directly responsible for regulation of NK susceptibility but the mechanism is not clear. Possibly the class I molecules interfere with the unknown NK target structures.
1 Introduction Tumor cells can be rejected by either MHC-restricted T lymphocytes or by non-MHC-restricted NK cells.The latter ones are viewed as a more primitive defense system against tumors [l]. No definitive information is presently available about the receptors on NK cells and their putative target structures [2-41. In spite of the fact that NK cells are non-MHC restricted, a number of reports in human and murine systems have indicated that MHC class I molecules on the target cells may play an important role for the regulation of susceptibility to NK-mediated lysis. For instance, class I loss variants of tumor cells were frequently found to be lysed more efficiently by NK cells in vitro and in vivo than their class I+ counterparts [5-121. It was also shown by many investigators that IFN-y-mediated enhancement of class I expression on the target cells was paralleled by a reduction of their NK susceptibility [13-191. This inverse correlation between NK susceptibility and MHC class I expression could have been merely fortuitous because numerous reports also exist in which no significant correlation between MHC expression and reduction of NK susceptibility was observed [20-261. Recently, in studies with human NK cells it was reported that the transfection of human NK target cells with class I genes resulted in a
[I 79811 Correspondence: Knut Sturmhofel,Institut fur Immunologie und Genetik, Deutsches Krebsforschungszentrm, Im Neuenheimer Feld 280, D-6900 Heidelberg, FRG Abbreviations: ADCC: Antibody-dependent cytotoxic cells &m: Pz-Microglobulin MFV: Mean fluorescence value neoR:
Neo(G418) resistance 0
VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1990
decrease of their NK susceptibility, showing that in these cases indeed the HLA molecules were responsible for this effect [27-291. In order to investigate whether MHC molecules could also affect NK lysis in other species we reconstituted the expression of class I molecules on a class I- murine tumor variant by gene transfection and determined the effect on NK lysis. For this purpose the C57BL/6-derived thymoma line EL4 was chosen which was relatively resistant t o NK cells. An H-2 loss variant of the EL4 cells was found to be highly susceptible to NK lysis whereas H-2+ transfectants were again resistant to NK cells.These observations suggest that the murine H-2 class I molecules can directly influence NK lysis.
2 Materials and methods 2.1 Cell lines and media EL4.Rob (a gift of I? Robinson, German Cancer Research Center, Heidelberg, FRG) is a clone of the murine thymoma EL4 of C57BL/6 origin. It was typed CD2- and FcR-. Variants negative for surface class I molecules were selected from EL4.Rob by two treatments with anti-Db and -Kb antibodies plus rabbit anti-mouse C (Low-tox rabbit anti-mouse C; Cedarlane, Hornby, Ontario, Canada) followed by several selections by FCM sorting (Ortho 50H, Ortho Diagnostics, Westwood, MA). The H-2- cells were cloned by LD. The clone S3 was negative for surface expression of Db and Kb as shown by FCM analysis. Clone C2S6.6 was negative for Kb and weak for Db. The H-2variants C6, C10 and C19 were gifts of l? Brams (Receptron, Concord, CA) and were also EL4.Rob derived. EL4.Rob grew as a semi-adherent line,whereas all other cell lines and OO14-2980/90/0101-0171$02.50/0
172
Eur. J. Immunol. 1990. 20: 171-177
K. Sturmhofel and G. J. Hammerling
transfectants grew adherent in DMEM supplemented with 7% FCS (Gibco, Grand Island, NY) and L-glutamine (Gibco). The cells were kept in culture for seveal months without changes of MHC class I expression or NK susceptibility. Recombinant mouse IFN-y was kindly provided by Dr. Svetly (Boehringer, Vienna, Austria). 2.2 Antibodies and production of 2&&6S-F(abr)2 fragments For FCM analysis and immunoprecipitation the following mouse mAb and sera were used: €322-249: anti-Db [30]; K10-56.1: anti-Kb [31]; 28-8-6s: anti-Db and -Kb [32]; T19-191: anti-Db (G. J. Hammerling, unpublished); 6/68: anti-Thy-1.2 [33]; Lym11.2: anti-P2-microglobulin (p2m) [34]; rabbit anti-rnouse-fl2m serum, recognizing the free P2m protein [35]; 17/227: anti-1-Ab [36]. F ( a b ' ) ~fragments of the mAb 28-8-6s were produced according to the method of Parham [37] and Lamoyi and Nisonoff [38]: 2 mg/ml of antibody was digested with 45 pg/ml of pepsin for 8-9 h at 37°C. The cleavage products were separated over a Sephadex G-150 (Pharmacia, Freiburg, FRG) column. The F(abr)2 peak was additionally purified over a protein ASepharose column to eliminate contaminations of undigested complete antibody. The purity of the F(ab')2 preparation was verified on a non-reducing SDS-PAGE gel which was silver stained. 2.3 Detection of surface molecules by FCM analysis About 1 x lo6 cells were incubated with 50 p1 SN of the appropriate mAb for ' h h at 4 "C, washed twice and stained with a 1/25 diluted goat anti-mouse-IgG-FITC conjugate (Sigma, St. Louis, MO) for l/2 h at 4°C. After washing, the cells were directly analyzed in a "FACScan" (Becton Dickinson, Mountain View, CA). The fluorescence intensity was measured in a 4-decade log-scale. A mean fluorescence value (MFV) within the first decade represented negative cells.
2.5 Immunoprecipitation of cytoplasmic
am proteins
For detection of cytoplasmic P2m products [35S]methioninelabeled proteins were precipitated with specific mAb and separated on 13%-18% gradient SDS-PAGE gels as recently described [44]. In short, cells were kept for 40 min in methionine-free medium at 37 "C, then incubated with 200 pCi (7.4 x lo6 Bq)/5 ml [3sS]methioninefor 3-4 h and finally lysed with a buffer containing 1% Triton-X, 100 mM NaCI, 5 mM MgC12, 20 mM Hepes (Gibco), 1mM proteinase inhibitor PMSF (Sigma). The lysate was preincubated with protein A-Sepharose for l/2 h. P2m proteins were precipitated from the lysate with the mAb Lymll.2 or rabbit anti-mouse P2m antiserum followed by protein A-Sepharose (1 h). 2.6 NK assay NK susceptibility of the tumor cells was measured in a standard SICr-releaseassay as described [45]. NK cells from three pooled spleens of 5-8 week-old CBA/J mice (Wiga, Hannover, FRG) were used. The NK cells were induced in vivo by i.p. injection of 150 pg poly(1) * poly(C) (Sigma) 18-22 h before the assay. The SC were isolated, erythrocytes lysed and the cells incubated on a 16-cm tissue culture dish for 11/2h for removal of plastic-adherent cells. The nonadherent cells were used as effectors. Target cells were labeled with Naz51Cr04 (NEN, Dreieich, FRG) for 11/2 h, washed twice and adjusted to 5 x 1@cells/100 pl as described [43]. Effectors and targets were incubated at different effector : target ratios, starting with 200 : 1, for 5 h at 37 "C. Then the cells were sedimented by centrifugation, 100 p1 of SN was collected and released W r was counted. Percent NK lysis was calculated by the formula: YOLysis = (cpm of probe - spontaneous release)/(maximum release - spontaneous release).
3 Results 3.1 Loss of class I surface expression on S3 is caused by a Ptm defect
2.4 Gene transfection and genes Two different protocols for gene transfection were applied. The first method was transfection with DNA-calcium phosphate precipitates according to Wigler et al. [39]. Transfection by electroporation [40] proved to be more efficient for the S3 cells. Briefly, a Bio-Rad-"gene pulser" (Bio-Rad Labs., Richmond, CA) was used with a capacity of 960 pF and voltages in the range of 190-250 V. Cells (1 X 107-2 X lo7) in 200 p1 PBS were cotransfected with about 30 pg of the appropriate MHC gene and 5 pg of the neoR gene. The neoR gene confers the resistance to geniticin. The neoR gene was either on a separate plasmid (pGB002, [41]) or combined with the @2mbgene on one plasmic (P2m/neoR). The Pzm/neoR plasmid was constructed by I? Robinson and contained a 10.9-kb long genomic Pzm DNA fragment and the neoR gene within the pACYC184 vector. Selection was started 1 or 2 days after the transfection process with 1.0 mg/ml geniticin (G418, Gibco) and was continued until resistant clones came up after about 14 days. The transfected MHC genes were Db [42]; Dd [43]; 1-A!, 1-A; [36].
To test the hypothesis that class I surface molecules on the target cells influence the NK susceptibility class I loss variants were selected from the ELA.Rob thymoma cell line. The H-2- variant S3 was selected from EL4.Rob as described by treatment with anti-H-2b antibodies plus C, followed by cell sorting and cloning.The total loss of Dband Kb surface molecules on the S3 cells was found to be due to a defect of P2m expression. The respective evidence was obtained from the following observations: (a) the transfection of the P2mbgene into S3 resulted in Db- and Kb-positive cells (see below) whereas the transfection of class I genes alone such as the Dbgene without P2m did not yield surface class If cells. (b) In immunoprecipitations of [%]methionine-labeled S3 cells no P2m protein could be precipitated either with the mAb Lymll.2, which binds to the P2mb protein associated with the class I H chain, or with a rabbit anti-mouse P2m antiserum which precipitates the free, unassociated P2m chain (Fig. 1). The EL4.Rob cells and all P2m transfectants which were positive for cell surface class I containing high amounts of cytoplasmic P2m protein (Fig. 1). Some neo-resistant transfectants, obtained after
Eur. J. Immunol. 1990. 20: 171-177
A s
B 82
s
02
Reduction of natural killer susceptibility by gene transfection
C
D
s
s
E
s
similar defect of P2m expression seemed to be responsible for loss of surface class I expression in the independently selected clones C6, C19, C19.This was suggested by similar immunoprecipitation data and transfections with p2m/neoR (data not presented). Another class I loss variant, clone C2S6.6, was found to be defective for Kb expression only. It was still positive for Db.
02
kD
k h
46-
173
-46
3.2 Reconstitution of Db/Kbexpression on S3 cells by gene transfection
R2m 14
-14
--2m
Figure I. Immunoprecipitation of the H-2-positive and -negative parental lines and of P2m transfectants. All lanes are derived from the same immunoprecipitation experiment. Lanes A (EL4.Rob) and B (S3) are from the same SDS-gel, lanes C, D (37.2.1,37.2.2, respectively, both positive for surface Db and Kb), and lane E (37.2.4, neoR,but negative for surface class I) were from another gel run in parallel. Lanes irrelevant for the present study were cut from the photograph. In lanes indicated with “S” precipitation was performed with the rabbit anti-mouse P2m antiserum recognizing only free P2m chains. In lanes marked with “By proteins were precipitated with the mAb Lymll.2 recognizing Pzm associated with H-2. The band at 12 kDa represents the P2m protein. In all precipitations with the rabbit anti-p2m antiserum, a 14-kDa band was also detected which appeared to be unrelated to P2m [34].The bands at about 46kDa represent the H chains of class I molecules.
transfection of the p2m/neoRplasmid, did not express the P2m gene. This is presumably due to inefficient integration of the transfected genes. These cells remained negative for surface class I expression. P2m or surface class I expression could not be induced with IFN-y as discussed later. A
S3 was transfected either with the genes for P2mb alone or with the genes for Dband P2mb.Transfectionwas performed with the DNAkalcium phosphate precipitation technique or by electroporation. D b and Kb surface expression of the resulting clones was measured by FCM analysis. All class I+ transfectants expressed at least as much class I as the parental EL4.Rob cells. S3 cells transfected with P2rn plus Db gene did not express higher levels of Db than the cells transfected with the P2m gene alone. The MFV of class I expression of the EL4.Rob and the S3 cells and the transfectants are summarized in Table 1. This table also shows the normalized values for NK lysis which will be discussed below. Representative FCM diagrams for the H-2 class I expression of EL4.Rob and three transfectants are shown in Fig. 2.
The NK susceptibility of the transfectants was measured in an NK assay as described. In each experiment the S3 and the EL4.Rob cells were included as high and low controls, respectively. A representative experiment is shown in Fig. 3. Here the NK lysis for the H-2- variant S3 was 46% and for the H-2- transfectant E45.5 about 50%. In contrast, the parental EL4.Rob cells and the H-2+ transfectants E45.41 and E45.44 were lysed t o only 15%-21%. This approximately two- to three-fold difference between H-2+ and H-2- EL4 cells was observed in all experiments
Table 1. Description of ELA.Rob, S3 and different Pzm transfectants: their class I expression and % NK lysis relative to S3 (summary of different experiments) Cell line EL3.Rob s3 32.2.1 37.2.2 37.2.4 E6.10 E6.8 E35.31 E45.41
E15.44 E35.1 E45.5 E19.7
MHC class I antigens expressed
MFV Dbb’
MFV
-
Db, Kb
Yo
Dh f12m/neoR Dh +p2m/neoR Dh + f12m/neoR Dh +p2m/neoR Db +P2m/neoR P2m/neoR P2m/neoR Pzm/neoR p2m/neoR f12m/neoR Dd + P2m/neoR
Db, Kb Dh, Kb
136 3 173 135 4 159 3 172 130 144 4 3 107
Transfect.a) procedure
Transfected genes
-
-
CaP CdP CaP E.P. E.P. E.P. E.P. E.P. E.P. E.P. E.P.
+
--
_ _
Db. Kh
- -
Dh, Kh Dh, Kb Dh, Kb
_ -
- -
Db. Kb. Dd
Kb”
2 87 75 3 82
2 121 70 101 4
2 77
Mean‘) % NK lysis relative to S3 56 (14) 100 (14) 39 (6) 4s (3) 101 (2) 61 (6) 117 (3) 23 (2) 53 (5) 26 (3) 95 (1) 110 (1) 27 (2)
MFV Dd: 84
a) Cap: DNA-calcium phosphate precipitation; E.P.: electroporation. b) MFV calculated from 5 to 20 FACS analyses. c) Percent NK lysis of S3 was set to 100% in each experiment. For each target the value of NK lysis was related to S3.The arithmetical mean of relative NK lysis was calculated from all experiments performed; number of experiments given within parentheses.
Eur. J. Immunol. 1990. 20: 171-177
K. Sturmhofel and G. J. Hammerling
174
The lysis of the other target cells was calculated in relation to the lysis of S3. Table 1 indicates also the number of experiments performed for each cell line. A consistent picture is obtained: whenever the cells express class I antigens on the cell surface they are less susceptible to NK lysis than their H-2- counterparts.
lcq fluorescence
3QQ
1
300
1
-
.:' .
€45.44
. .
. . .. ..
.. .. ..
..
1- .: '.. ,: i' : .- 2
is,
ii
i '
i
lcq -lf
Figure 2. Measurement of Dbsurface expression of EL4.Rob and three P2m transfectants by FCM analysis. Cells were labeled with the anti-Dbantibody B22-249. I n each diagram the H-2- S3 cells are included as negative control (dotted line). E45.41 and E45.44 are class I+ Pzm transfectants, E45.5 is a neo-resistant but surface class I- transfectant.
In Fig. 4 these NK values are plotted over the mean fluorescence intensity obtained for total class I expression. The data suggest an inverse linear relationship between the amount of class I expression and susceptibility to NK lysis. Additionally, the clone C2S6.6 which expressed no Kb but intermediate amounts of Dbon the cell surface exhibited an NK susceptibility intermediate to S3 and EL4.Rob and therefore fitted well into this correlation. The figure also presents the independently derived class I- EL4 variants C6, C10 and C19 which were as efficiently lysed as S3. We further investigated whether transfection of class I1 genes into the class 11- S3 line would alter its NK susceptibility. S3 was cotransfected with genes for the I-Aba and @ chains. Surface expression of I-Ab was demonstrated with the mAb 17/227 which recognizes only the a/@-assembled I-A molecule.The resulting clones were still negative for Db and Kbon the surface but positive for class 11. No difference in NK lysis between S3 and the I-A transfectants was found (Fig. 5).
3.3 IFN-y treatment does not influence NK lysis In many studies it was reported that an increase in the amount of class I antigens induced by IFN-y was always accompanied by a decrease of NK susceptibility [9, 14, 16, 19,201.To examine whether IFN-y would decrease NK lysis in the absence of altered class I expression the cells were
40
%
'$
..__ -..
60%-
e
b
effectortarget ratio
.I?
k
40% -
e
Figure3. NK experiment with the targets EL4.Rob ( O ) , S3 (0) and the same P2rn transfectants as in Fig. 2: E45.41 (m), E45.44 (A),(both class I) and E45.5 (0) (class I-). The E/T was downtitrated in four steps from 200 : 1 to 0.5 : 1. The % NK lysis was calculated as described in Sect. 2.6 and was not normalized as in Table 1.
and with all transfectants. The degree of NK lysis ranged from about 29%-49% for the S3 cells with an arithmetic mean of 37% depending on the preparation of the NK cells. However, in the very same experiments the lysis of the H-2+ transfectants was always reduced by the factor of two or more compared to the H-2- S3 cell 1ine.The respective NK data are compiled in the right hand part of Table 1in which the degree of NK lysis was normalized: for each experiment . the percentage of NK lysis of the S3 cells was set to 100%.
g 20%K 0%
'
0
50
100
150
200
I
I
I
250
300
350
t o t a l class I expression /MFV
Figure 4. Mean % NK lysis relative to S3 (S3 = 100%)is plotted vs. amount of total class I expression.The mean values for % NK lysius and class I expression of the different target cells are compiled in Table 1. The amount of total class I cell surface expression was calculated by addition of the MFV data of Dband Kb. ( 0 )EL4.Rob; (0)S3; (x) different class I- transfectants; (+) different class I+ transfectants; (A) clone E19.7 transfected with Dd plus Pzm, expressing D, Kb and Dd. Additionally the H-2clones C6, C10 and C19 (V) and the Kb-negative but Db-positive clone C2S6.6 (0)are included in this graph. The dashed line represents the calculated linear trend for the relationship between increasing amount of class I expression and decreasing NK lysis.
Eur. J. Immunol. 1990. 20: 171-177
Reduction of natural killer susceptibility by gene transfection
175
Table 2. Influence of IFN-y Treatment on class I expression and NK lysis of class I-negative and -positive cellsa)
s3 ELA.Rob 37.2.1
6o
3 85 150
2 65 64
39 19 17
4 93 152
One experiment out of three; all experiments gave similar results. Brget cells were incubated with rFN-y (60U/ml) for 68 h. MFV intensity: Db labeled with mAb B22249, Kb with mAb K10-56.1. Value for % NK lysis is given in absolute numbers not related to S3 as in Table 1.
2 72 69
r
0
effector:target ratio
0.4
10
50
amount of antibody [ ,ugg/ml]
Figure 5. NK susceptibility of S3 class I1 transfectants. H-2 class I- and class 11- S3 cells (0)were cotransfected with the genes for IA: and IAI. The obtained transfectants were tested in a standard NK assay. The results of five transfectants are presented: IAdp-positive clones: (V) E23.2.17; (0)E23.2.18; (A) E23.11.6; (0)E23.11.13; IAdb-negative clone: ($) E23.11.15. EL4.Rob ( 0 )and the P2m transfectant E45.41 (B) were used asclass 11-, but class I+ controls.
Figure 6. Effect of blocking the target cell class I molecules by anti-class I antibody on NK susceptibility.The class I+ transfectant 37.2.1 (B) and the S3 cells (0) were incubated with different amounts of F(ab’)*fragments of the anti-DbKbmAb 28-8-6s (solid line) prior to the NK assay. For control the target cells were treated with a non-class I antibody in the same way (dashed line; anti-Thy-1.2 mAb 6/68).
treated with 60 U/ml of mouse rIFN-y for about 60-70 h. For the class I- S3 line no induction of class I expression and no change of NK susceptibility was observed (Table 2). This unaltered reaction of S3 was not due to an unresponsiveness to IFN-y because Ly-6.2 surface expression and the level of mRNA for Db and Kb were increased (data not shown). Likewise, IFN-y did not increase class I expression on the parental EL4.Rob or on the transfectant 37.2.1, probably because class I was already maximally expressed in these cells. Again no change in NK susceptibility was observed (Table 2).
EL4.Rob or class I+ transfectants were incubated with 28-8-63 F(ab’)z fragments prior to the NK assay the NK susceptibility was drastically enhanced and reached the level of the H-2- variant S3 (Fig. 6). As expected, the 28-8-68F(ab’)2 fragments did not increase the lysis of the S3 cells. An antibody against the non-MHC surface antigen Thy-1.2 was used as control; 6/68 is of the IgM class and therefore not recognized by ADCC. EU.Rob, S3 and the transfectants are all Thy-1.2+ (data not shown). Preincubation of the target cells with the 6/68-antibody did not influence NK susceptibility (Fig. 6).
3.4 Masking of class I antigens by an anti-class I antibody renders target cells more susceptible to NK lysis
4 Discussion
To demonstrate that indeed the class I molecules are involved in the reduction of NK susceptibilitywe attempted to block the class1 molecules of the target cells with a class I-specific mAb. In order to avoid lysis mediated by antibody-dependent cytotoxic cells (ADCC), F(ab’)~fragments of the anti-Db/KbmAb 28-8-68were prepared.When
In numerous studies an inverse correlation between NK lysis and MHC class I expression on the target cells was found. First, class I loss variants were frequently more sensitive to NK-mediated lysis [5-12,461 and second, enhancement of class I expression by IFNy was always accompanied by a decrease in NK susceptibility [13-191. The present report demonstrates that the inverse correlation between H-2 class I expression and NK lysis is not
176
Eur. J. Immunol. 1990. 20: 171-177
K. Sturmhofel and G . J. Hammerling
fortuitous. In all experiments the NK lysis of the H-2+ parental EL4.Rob cells and transfectants was always decreased by a factor of two to three in comparison to the H-2- EL4 variant S3, or the H-2- transfectants.These data demonstrate that the class I molecules on the target cells are directly involved in the reduction of NK lysis. In addition, the Kb-negative, but Db-positive variant C2S6.6 exhibited an intermediate amount of class I expression and an intermediate degree of NK lysis. It should also be noted that the independently derived EL4.Rob class I loss variants C6, C10 and C19 were as highly NK susceptible as the S3 variant. Their loss of class I expression was also caused by a defect in Pzm expression. Recently, it has been reported for several H-2- variants of two murine tumor cell lines that they had lost class I expression due to distinct defects in Pzm expression and that they were all highly NK susceptible [lo]. In contrast, cell surface expression of class I1 molecules obtained after transfection of I-A genes did not alter the NK susceptibility of the S3 cell 1ines.Thisis in agreement with observations made with human B cell lymphomas [12,281. The H-2- variant S3 failed t o express class I surface antigens after in vitro treatment with IFN-y. However, S3 could respond to IFN-y as demonstrated by the increase of class I-specific mRNA and cell surface expression of the Ly-6.2 antigens. Therefore, the S3 cells offered the opportunity to test directly the influence of IFN-y on NKmediated lysis. No change of NK-mediated lysis was observed after treatment with IFN-y (Table 2). Likewise, on the H-2+ parental EL4.Rob cells and transfectants no significant increase of class I expression could be observed after IFN-y treatment, probably because the expression of class I antigens on the surface of these cells was already saturated. Again, NK lysis of the IFN-treated cells was not changed. These observations support the notion that the decreased NK susceptibility obtained after treatment with IFN-y, which was observed in many studies [14-171, is indeed mediated by enhancement of MHC class I expression. While this work was in progress, other investigators reported for the human NK system that the transfection of HLA class I genes into human lymphoblastoid B cell lines reduced their NK susceptibility [27-291. To our knowledge no such transfection studies have been performed in other species. Such additional studies are even more important in the light of the observations by Storkus et al. [29] that only the transfection with HLA but not with H-2 genes (DP and Kb) reduced the lysis of the human NK targets by human NK effector cells. In contrast, the present study shows that H-2 can indeed reduce NK susceptibility. It is not clear why the effect of H-2 is only found in a murine but not in a human NK system. Possibly, this reflects a difference in the nature of mouse and human NK target structures. The mechanism by which class I molecules influence NK-mediated lysis is not clear. It is possible that the MHC molecules on the target surface interact with the putative target structures for the NK effector cells. This interaction could change the conformation and the accessibility of the target structures in such a way that they can no longer be recognized by NK cells. This model is also supported by our finding that blocking of the class I antigens on the target cells with anti-Db/Kbantibodies rendered these cells highly
sensitive to NK-mediated lysis. Possibly, the binding of the antibodies with the class I molecules prevented their interaction with the NK target structures. In other reports no enhancement of NK lysis was observed after the incubation of the target cells with anti-class I antibodies [12, 201.These discrepancies could be due to a lower affinity of the antibodies used, or to the possibility that the respective antibodies were directed against class I epitopes which were not involved in the interaction between MHC molecules and NK target structures. The concept of MHC molecules interfering with putative NK target structures is supported by the evidence that MHC class I molecules interact with other cell surface structures such as, for example, the receptors for insulin, epidermal growth factor and glucagon [47-511. Different domains of the MHC class I molecules are known to have different functions. Thus, the al and a2 domains form the peptide-binding groove [52] whereas the a3 domain has been found to interact with the CD8 molecule [53,54]. It will be of interest to determine which part of the MHC class I molecule is responsible for the proposed interaction with the NK target structures. It is also conceivable that the efficiency of such interactions could depend on the respective MHC class I alleles. The proposed model of the interference of MHC with NK target structures would also explain why the effect of MHC class I is only partial and does not lead to a complete reduction of NK lysis, and why in several reports no influence of class I on NK lysis was observed [20-261. It is conceivable that the amount of class I is not sufficient to block all NK target structures, or that the target structures are heterogeneous and not all of them can interact with class I molecules or the respective class I alleles. We thank N. Bulbuc for skilful technical assistance. Received September 27, 1989.
5 References 1 Herberman, R. B., Reynolds, C.W. and Ortaldo, J. R., Annu. Rev. Immunol. 1985. 4: 651. 2 Ortaldo, J. R. and Reynolds, C. W., J. lmmunol. 1989. 138: 4545, 3 Fitzgerald-Bocarsly, I?, Herberman, R. B., Hercend, T., Hiserodt, J., Kumar, V., Lanier, L., Ortaldo, J. R., Pross, H., Reynolds, C. W.,Welsh, R. M. and Wigzell, H., in Ades, E. W. and Lopez, C . (Eds.), Natural Killer Cells and Host Defense, Karger, Basel 1989, p. 13. 4 Ortaldo, J. R . , Kantor, R., Segal, D., Bolhuis, R. H. and Bino, T.,in Ades, E.W. and Lopez, C. (Eds.), Natural Killer Cells and Host Defense: 5th International Natural Killer Cell Workshop, Karger, Basel 1989, p. 221. 5 Taniguchi, K., Karre, K. and Klein, G., lnt. J. Cancer 1985.36: 503. 6 Karre, K., Lunggren, H.-G., Piontek, G . E. and Kiessling, R. W., Nature 1986. 319: 675. 7 Ljunggren, H.-G., OhlCn, C., Hoglund, I!,Yamasaki,T., Klein, G . and Karre, K . , J. Immunol. 1988. 110: 671. 8 Kawano, Y.-I., Xiniguchi, K., Karre, K., Toshitani, A . and Nomoto, K., Cell. Immunol. 1988. 111: 341. 9 Ohlen, C., Bejarano, M.-T., Gronberg, A., Torsteinsdottir, S., Franksson, L., Ljunggren, H.-G., Klein, E., Klein, G. and Karre, K., J. lmmunol. 1989. 142: 3336.
Eur. J. Immunol. 1990. 20: 171-177
Reduction of natural killer susceptibility by gene transfection
10 Ljunggren, H.-G., Paabo, S., Cochet, M., Kling, G., Kourilsky, P. and Karre, K., J. Irnmunol. 1989. 142: 2911. 11 Harel-Bellan, A., Quillet, A., Marchiol, C., DeMars, R.,Tursz, T. and Fradelizi, D., Proc. Natl. Acad. Sci. USA 1986. 83: 5688. 12 Storkus,W. J., Howell, D. N., Salter, R. D., Dawson, J. R. and Cresswell, P., J. Immunol. 1987. 138: 1657. 13 Hansson, M., Kiessling, R. W., Anderson, B. and Welsh, R. M., J. Immunol. 1980. 125: 2225. 14 McMillan,T. J., Everett, C. A. and Hart, I. R., Int. J. Cancer 1987. 40: 659. 15 Welsh, R. M., Karre, K., Hansson, M., Kunkel, L. A. and Kiessling, R. W., J. Immunol. 1981. 126: 219. 16 Piontek, G. E., Taniguchi, K., Ljunggren, H.-G., Gronberg, A., Kiessling, R. W., Klein, G. and Karre, K., J. Immunol. 1985. 135: 4281. 17 Gronberg, A., Ferm, M. T., Ng, J., Reynolds, C. W. and Ortaldo, J. R., J. Immunol. 1988. 140: 4397. 18 Zoller, M., Strubel, A., Hammerling, G. J., Andrighetto, G., Raz, A. and Ben-Ze'ev, A., lnt. J. Cancer 1988. 41: 256. 19 Hammerling, G. J., Klar, D., Maschek, U., Piilm, W. and Sturmhofel, K., in David, C. S. (Ed.), H-2 Antigens: Genes, Molecules, Function, NATO AS1 Series A: Life Sci. vol. 144, Plenum Publ. Corpo., New Yok 1987, p. 705. 20 Gorelik, E., Gunji,Y and Herberman, R. B., J. Zmmunol. 1988. 140: 2096. 21 Chervenak, R. and Wolcott, M., J. Immunol. 1988. 140: 3712. 22 Dennert, G., Landon, D., Lord, E. M., Bahler, D. W. and Frelinger, J. G., J. lmmunol. 1988. 140: 2472. 23 Leiden, J. M., Karpinski, B. A., Gottschalk, L. and Kornbluth, J., J. Immunol. 1989. 142: 2140. 24 Stam, N. J., Kast,W. M.,Voordouw, A. C., Pastoors, L. B.,Van der Hoeven, F., Melief, C. J. M. and Ploegh, H. L., J. Irnmunol. 1989. 142: 4113. 25 Nishimura, M. I., Stroynowski, I., Hood, L. and OstrandRosenberg, S., J. Immunol. 1988. 141: 4403. 26 Gopas, J., Segal, S., Hammerling, G. J., Bar-Eli, M. and Rager-Zisman, B., Immunol. Lett. 1988. 17: 261. 27 Quillet, A., Presse, F., Marchiol-Fournigault,C., Harel-Bellan, A., Benbunan, M., Ploegh, H. L. and Fradelizi, D., J. Immunol. 1988. 141: 17. 28 Shimizu, Y. and DeMars, R., Eur. J. Immunol. 1989. 19: 447. 29 Storkus, W. J., Alexander, J., Payne, J. A., Dawson, J. R. and Cresswell, P., Proc. Natl. Acad. Sci. USA 1989. 86: 2361. 30 Lemke, H., Hammerling, G. J. and Hammerling, U., Immunol. Rev. 1979. 47: 175. 31 Hammerling, G. J., Riisch, E., B d a , N., Kimura, S. and
177
Hammerling, U., Proc. Nut[. Acad. Sci. USA 1982. 79: 4737. 32 Ozato, K. and Sachs, D. H., J. Immunol. 1981. 126: 317. 33 Hammerling, G. J., Lemke, H., Hammerling, U., Hohmann, C., Wallich, R. and Rajewski, K., Curr. Top. Microbiol. Immunol. 1978. 81: 100. 34 Margulies, D. H., Parnes, J. R., Johnson, N. A. and Seidman, J. G., Proc. Natl. Acad. Sci. USA 1983. 80: 2328. 35 Robinson, P. J., Graf, L. and Sege, K., Proc. Natl. Acad. Sci. USA 1981. 78: 1167. 36 Landais, D., Beck, B. N., Buerstedde, J.-M., Degraw, S., Klein, D., Koch, N., Murphy, D., Pierres, M., Tada, T., Yamamoto, K., Benoist, C. and Mathis, D., J. Immunol. 1986. 137: 3002. 37 Parham, P., J. Immunol. 1983. 131: 2895. 38 Lamoyi, E. and Nisonoff, A., J. Immunol. Methods 1983.56: 235. 39 Wigler, M., Pellicer, A., Silverstein, S., Axel, R., Urlaub, G. and Chasin, L., Proc. Natl. Acad. Sci. USA 1979. 76: 1373. 40 Doffinger, R., Pawlita, M. and Sczakiel, G., Nucleic Acids Res. 1988.16: 11840. 41 Brady, G., Jantzen, H. M., Bernard, H. U., Brown, R., Schiitz, G. and Hashimoto-Gotoh, T., Gene 1984. 27: 223. 42 Allen, H.,Wraith, D., Pala, P., Askonas, B. and Flavell, R. A., Nature 1984. 309: 279. 43 Margulies, D. H., Evans, G. A., Ozato, K., Camerini-Otero, R. D. ,Bnaka, K., Appella, E. and Seidman, J. G., J. Immunol. 1983. 130: 463. 44 Klar, D. and Hammerling, G. J., EMBO J. 1989. 8: 475. 45 Wigzell, H. and Ramstedt, U., in Weir, M. (Ed.), Handbook of Experimental Immunology; vol.2 CellularImmunology, Blackwell Scientific Publications, Oxford 1986, chapter 60.1. 46 Kawano, Y.-I., Taniguchi, K., Toshitani, A. and Nomoto, K., J. Immunol. 1986. 136: 4729. 47 Kittur, D., Shimizu,Y.,DeMars, R. and Edidin, M., Proc. Natl. Acad. Sci. USA 1987. 84: 1351. 48 Verland, S., Simonsen, M., Gammeltoft, S., Allen, H., Flavell, R. A. and Olsson, L., J. lmmunol. 1989. 143: 945. 49 Edidin, M., lmmunol. Today 1983. 4: 269. 50 LaFuse,W. and Edidin, M., Biochemistry 1980. 19: 49. 51 Phillips, M. L., Moule, M. L., Delovitch,T. L. and Yip, C. C., Proc. Natl. Acad. Sci. USA 1986. 83: 3474. 52 Bjorkman, P. J., Saper, M. A., Samraoui, B., Bennett, W. S., Strominger, J. L. and Wiley, D. C., Nature 1987. 329: 506. 53 Potter,T. A., Rajan,T.V, Dick, R. F., I1 and Bluestone, J. A., Nature 1989. 337: 73. 54 Salter, R. D., Norment, A. M., Chen, B. P., Clayberger, C., Krensky, A. M., Littman, D. R. and Parham, P., Nature 1989. 338: 345.
