The EMBO Journal vol.9 no. 12 pp. 38 15 - 3819, 1990
Inhibition of lymphocyte mediated cytotoxicity by perforin antisense oligonucleotides
Hans Acha-Orbea, Leonardo Scarpellino, Sylvie Hertig1, Marc Dupuis1 and Jurg Tschopp1 Ludwig Institute for Cancer Research, Lausanne Branch and 1Institute of Biochemistry, University of Lausanne, Ch. des Boveresses 155, 1066 Epalinges, Switzerland Communicated by P.Golstein
The granule/perforin exocytosis model of CTL mediated cytolysis proposes that CTL, upon recognition of the specific targets, release the cytolytic, pore-forming protein perforin into the intercellular space which then mediates the cytotoxic effect. However, direct evidence for the involvement of perforin is still lacking, and indeed, recent results even seem incompatible with the model. To determine directly the role of perforin in CTL cytotoxicity, perforin antisense oligonucleotides were exogenously added during the stimulation of mouse spleen derived T cells and human peripheral blood lymphocytes (PBL), respectively. Perforin protein expression in lymphocytes was reduced by up to 65%, and cytotoxicity of stimulated T cells by as much as 69% (5.7-fold). These results provide the first experimental evidence for a crucial role of perforin in lymphocyte mediated cytotoxicity. Key words: cytolytic T/lymphocytes/cytotoxicity/perforin/ antisense oligonucleotides
Introduction The granule/perforin exocytosis model of CTL mediated cytolysis proposes that CTL, upon the recognition of their specific target, release the contents of cytolytic granules. In addition to a family of proteases named granzymes, these organelles contain a pore-forming protein designated perforin (Young, 1989; Podack, 1986; Henkart and Yue, 1988; Shinkai et al., 1988; Tschopp and Nabholz, 1990). Circumstantial evidence supports this model. (i) Cytoplasmic granules with cytolytic activity have been found in activated CTL, but not in their cytolytically inactive precursors. (ii) Degranulation of granule associated granzymes has been observed at the interface between killer and target cells (Peters et al., 1989). (iii) With few exceptions, both granule mediated and CTL mediated cytolysis are calcium dependent. However, several observations have been difficult to reconcile with the exocytosis/perforin model. (i) Certain highly lytic CTLs primed in vivo lack detectable quantities of perforin. Target cells lysed by these T cells derived from the peritoneal cavity of alloimmunized mice are devoid of perforin induced 'complement'-like lesions (Berke and Rosen, 1988). (ii) Antibodies against perforin known to abrogate lysis induced by isolated granules, do not block lysis induced by CTLs in general (Berke, 1989). (iii) CTL Oxford University Press
mediated cytolysis of some target cells can proceed in the absence of calcium in the medium, although perforin's lytic potential is strictly calcium dependent (Trenn et al., 1987; Ostergaard et al., 1987). (iv) Similarly, the presence of calcium seems to be an absolute prerequisite for granule secretion as no granule associated proteins such as granzyme proteases are released in calcium-free medium (Trenn et al., 1987; Ostergaard et al., 1987). (v) Targets are lysed by an apoptotic mechanism (Russel, 1983), apparently by inducing the disintegration of target cell nuclear DNA in nucleosomesized fragments of 190 bp in size, whereas target cell lysis caused by purified perforin is characterized by the disintegration of the cellular membrane without DNA breakdown (Duke et al., 1989; Holden et al., 1987). For these reasons, a perforin mediated target cell lysis based on hole formation has been seriously questioned. To determine directly whether perforin is involved in CTL mediated cytotoxicity, we used the antisense oligonucleotide approach. This strategy to study specific gene function has been used successfully in the past to explore, for example, the function of c-myc (Gewirtz and Calabretta, 1988) or myeloblastin (Bories et al., 1989). Here we present data showing that inhibition of perforin expression in T lymphocytes activated in vitro results in a proportional diminution of CTL cytotoxicity. -
Results Experimental design Established CTL cell lines cultured in the presence of IL-2 contain large amounts of granules and granule proteins, including perforin. Although most lines require weekly stimulation with the appropriate antigen presenting cell, the expression of perforin protein and message is never arrested. We reasoned that depletion of this granule-stored perforin pool would be difficult or even impossible, and that CTLs whose perforin protein or message expression can be induced from zero or near zero levels would provide a more suitable system for the antisense approach. Previous reports have shown that resting T cells in general do not produce measurable levels of perforin message (Garcia-Sanz et al., 1987; Liu et al., 1990). Induction of perforin message or protein requires stimulation of cells by PHA, anti-T cell -receptor complex antibodies, lectins, or antigen presented by antigen presenting cells. Thus, we chose two systems of non-activated T cells, human peripheral blood lymphocytes and mouse spleen lymphocytes to perform our experiments. These cells were stimulated with immobilized anti-T cell -receptor complex antibodies for 16-48 h in the presence or absence of perforin sense or antisense oligonucleotides. Cytotoxic activity was measured by coincubation of these activated T cells with 5'Cr-labelled P815 target cells which were coated with anti-CD3 antibodies. This readout system guarantees detection of all T cell receptor 3815
H.Acha-Orbea et al.
bearing CTLs, but excludes the measurement of lytic activity of NK cells. CTLs and oligonucleotide uptake An 18mer antisense oligodeoxynucleotide 5'-GAG CAG ACG GGC TGC CAT-3' complementary to a sequence starting at the ATG initiation codon of human perforin and an antisense 16 bp oligonucleotide of mouse perforin (5'-AAC AGG CAC GTG GCC-3') starting at nucleotide +3 of the coding region of the mRNA were synthesized. As controls, sense oligonucleotides of the complimentary sequences were synthesized. Neither sense nor antisense sequences were identical to any known nucleotide sequence available in the EMBL database. Exogenously added antisense oligodeoxynucleotides would be only minimally effective for mRNA hybridization and consequent translational arrest if they were rapidly degraded or not internalized. To verify uptake by PBL, the human perforin antisense oligodeoxynucleotide was labelled with P at its 5'-end and the percentage uptake determined. In the case of human PBLs, 1.0% of the perforin antisense oligonucleotide was associated with the PBL cellular pellet after 4 h incubation (Figure 1). This fraction decreased to 0.6% after 24 h incubation. -
Inhibition of perforn protein expression Human PBL were activated by anti-CD3 monoclonal antibodies and addition of IL-2 in the presence of either antisense or sense perforin oligomer, and protein expression was analysed by immunoblotting. Neither oligomer decreased the level of induction of granzyme B protease, the expression of which is highly upregulated upon lymphocyte activation (Schmid and Weissmann, 1987) (Figure 2). In contrast, activated PBLs treated with 40 /4M antisense perforin showed reduction in the level of perforin protein expression of up to 65 % relative to cells stimulated in the presence of sense oligomer. Inhibition of protein expression varied from 46 to 65 %, but complete absence of perforin protein was never observed. The quantity of perforin induced was, however, very low, and its detection was at the borderline of the sensitivity of our reagents, resulting in a great variation within a given experiment. 2.0
Although both antibodies used against granzyme B and perforin were made against recombinant protein and could be used at similar dilutions in Western blots and for immunomicroscopy (Hameed and Tschopp, unpublished), only the granzyme B antibodies yielded strong signals after the short time stimulation of PBL. In the experiments with mouse lymphocytes, it was not even possible to measure the low level of perforin protein levels with the available antibodies, although perforin is readily detected in CTL cell lines using the same reagents. Inhibition of lymphocyte mediated cytotoxicity Preliminary experiments to determine whether the decreased expression of perforin affected CTL effector functions indicated that induction of cytotoxicity in human PBL was not very efficient when the latter were stimulated with immobilized anti-CD3 antibodies alone for 24 h. Subsequent incubation with medium containing r-IL-2 (100 U/ml) for an additional 16 h in the absence of the stimulating antibodies highly increased the lytic activity of PBL. Up to 60% specific lysis of antibody-coated P815 target cells were obtained at a 40:1 CTL:target ratio (Figure 3a). The antisense perforin oligonucleotide was added during the activation period in three aliquots, at the beginning of CD3 induced stimulation, after 16 h and upon the change of r-IL-2-containing medium. After the activation period, antisense oligonucleotide treated cells showed a marked decrease in cytotoxicity in a dose dependent manner in the experiment displayed in Figure 3a. At a concentration of 40 itM oligomer, lysis of P815 target cells was decreased from 54% to 29%. In contrast, the level of cytotoxicity of PBL cultured in the presence of sense oligonucleotide was as high as in untreated cells. Reduced cytotoxicity was also observed in PBL activated during 16 h solely with r-IL-2 (see Figure 3b). Cells exposed to a single dose of antisense oligomers were markedly less cytotoxic (up to 62 % reduction), whereas sense oligomer sa
sa sense
OD 540 0.05
66 -
1.8 1.6 -
antisense
1.4 -
0.05
a) 1.2-
0. 1.0
-
2 0.8
-
0
27
+ +
0
0.2 I
0
I
I
5
I
I
I
10
15
20
25
Time (hr Fig. 1. Uptake of oligonucleotides in humar PBLs. Samples of 5-32plabelled anti-human perforin oligomer were incubated for 0, 1, 4, 8 and 24 h with PBL undergoing activation irn the presence of 100 U/ml IL-2 or non-activated PBLs percentage upta Ike of oligomer was determined as described in Materials and mliethods. 3816
?
Io~~~~~~
0.6 -
0.4 -
4:0 0
+.±±
Fig. 2. Inhibition of human perforin protein expression by antisense oligodeoxynucleotides. Immunoblot of protein extracts of unstimulated (-) and stimulated (+) human PBL treated with antisense (as) or sense (s) oligomer. Blots were developed with anti-granzyme B (left panel) and anti-perforin (middle panel) antibodies. Arrowheads refer to the proteins. A doublet was observed for granzyme B, most likely reflecting differences in glycosylation or granzyme X (Meier et al., 1990), a recently described granzyme B-like enzyme in lymphocytes. Molecular size markers are indicated. The right panel shows densitometric scans of the bands corresponding to human perforin as detected by Western blot analysis (see middle panel). Numbers refer relative units of the integrated peaks.
to
Perforin mediated cytotoxicity
had no effect on cytotoxicity exhibited by human lymphocytes. With mouse spleen cell derived lymphocytes, the oligonucleotides were added during the 17-21 h activation period of the lymphocytes with immobilized anti-CD3 antibody. As shown in Figure 4, the antisense oligonucleotide inhibited cytotoxic activity in a dose dependent fashion. When lytic units were calculated, at 18 ltM oligonucleotide concentration, a 5.7-fold inhibition of cytotoxicity was observed with antisense as compared with sense oligonucleotide. This inhibition decreased to 4.1-fold at 6 ,uM, 3.3-fold at 2 AM, 1.6-fold at 0.7 jiM. No inhibition was observed at 0.25 ItM. A slight inhibition caused by the three independently synthesized preparations of sense oligonucleotides was observed with murine lymphocytes. Addition of sense or antisense oligonucleotides did not have any detectable effects on blast T cell recovery after stimulation. Activation of cells for longer periods of time (48 h) resulted in 9-fold higher cytotoxic activation, but neither sense nor antisense oligonucleotides caused significant inhibitory activation in these cells. This is probably due to the relatively short half-life of the oligonucleotides in culture. In addition, it proves that the effect of the antisense oligomer is not due to toxic effects. Inhibition of lymphocyte cytotoxicity varied markedly from one experiment to another. Table I lists the results of various experiments. Each experiment was performed with a different preparation of cells. The percentage of specific 60
inhibition ranged from 69% to 3%. In no case, however, was the inhibition by the sense oligomers superior to that attained with the antisense oligomers.
Discussion Recent studies have shown that the introduction of synthetic antisense oligonucleotides can result in a specific inhibition of the expression of cellular genes. In the present study, we demonstrate the use of this strategy for the selective deletion of a granule associated protein in lymphocytes. The involvement of perforin in CTL mediated cytolysis has been hotly debated. Whereas ample evidence exists that perforin in CTL clones or cells may be, at least in part, 30
40-
18 iM
30-
6 tM
20 20-4 10
~~~~~~~~10
.Cn
C.)
cn40 -4010.7 ±M 2 iM
a
50-*OO 30 -
E
40
Effector to target ratio C
30 -_0 30 E 200*
Fig. 4. Inhibition of mouse lymphocyte mediated cytotoxicity by antiperforin oligomers. Mouse spleen cell derived lymphocytes were activated with anti-CD3 antibodies immobilized on plastic during 24 h in the presence of perforin antisense (closed triangles, dotted line) or
0 C
10
-
en
0 ._
0
I
1
1
b
0)
CoU)I50
sense (open rectangles) oligonucleotides. Lysis of P815 cells in the absence of anti-CD3 antibody is shown by closed circles. Lysis of unstimulated spleen cells in the presence of anti-CD3 antibody was
Inhibition of lymphocyte mediated cytotoxicity by perforin antisense oligonucleotides
Hans Acha-Orbea, Leonardo Scarpellino, Sylvie Hertig1, Marc Dupuis1 and Jurg Tschopp1 Ludwig Institute for Cancer Research, Lausanne Branch and 1Institute of Biochemistry, University of Lausanne, Ch. des Boveresses 155, 1066 Epalinges, Switzerland Communicated by P.Golstein
The granule/perforin exocytosis model of CTL mediated cytolysis proposes that CTL, upon recognition of the specific targets, release the cytolytic, pore-forming protein perforin into the intercellular space which then mediates the cytotoxic effect. However, direct evidence for the involvement of perforin is still lacking, and indeed, recent results even seem incompatible with the model. To determine directly the role of perforin in CTL cytotoxicity, perforin antisense oligonucleotides were exogenously added during the stimulation of mouse spleen derived T cells and human peripheral blood lymphocytes (PBL), respectively. Perforin protein expression in lymphocytes was reduced by up to 65%, and cytotoxicity of stimulated T cells by as much as 69% (5.7-fold). These results provide the first experimental evidence for a crucial role of perforin in lymphocyte mediated cytotoxicity. Key words: cytolytic T/lymphocytes/cytotoxicity/perforin/ antisense oligonucleotides
Introduction The granule/perforin exocytosis model of CTL mediated cytolysis proposes that CTL, upon the recognition of their specific target, release the contents of cytolytic granules. In addition to a family of proteases named granzymes, these organelles contain a pore-forming protein designated perforin (Young, 1989; Podack, 1986; Henkart and Yue, 1988; Shinkai et al., 1988; Tschopp and Nabholz, 1990). Circumstantial evidence supports this model. (i) Cytoplasmic granules with cytolytic activity have been found in activated CTL, but not in their cytolytically inactive precursors. (ii) Degranulation of granule associated granzymes has been observed at the interface between killer and target cells (Peters et al., 1989). (iii) With few exceptions, both granule mediated and CTL mediated cytolysis are calcium dependent. However, several observations have been difficult to reconcile with the exocytosis/perforin model. (i) Certain highly lytic CTLs primed in vivo lack detectable quantities of perforin. Target cells lysed by these T cells derived from the peritoneal cavity of alloimmunized mice are devoid of perforin induced 'complement'-like lesions (Berke and Rosen, 1988). (ii) Antibodies against perforin known to abrogate lysis induced by isolated granules, do not block lysis induced by CTLs in general (Berke, 1989). (iii) CTL Oxford University Press
mediated cytolysis of some target cells can proceed in the absence of calcium in the medium, although perforin's lytic potential is strictly calcium dependent (Trenn et al., 1987; Ostergaard et al., 1987). (iv) Similarly, the presence of calcium seems to be an absolute prerequisite for granule secretion as no granule associated proteins such as granzyme proteases are released in calcium-free medium (Trenn et al., 1987; Ostergaard et al., 1987). (v) Targets are lysed by an apoptotic mechanism (Russel, 1983), apparently by inducing the disintegration of target cell nuclear DNA in nucleosomesized fragments of 190 bp in size, whereas target cell lysis caused by purified perforin is characterized by the disintegration of the cellular membrane without DNA breakdown (Duke et al., 1989; Holden et al., 1987). For these reasons, a perforin mediated target cell lysis based on hole formation has been seriously questioned. To determine directly whether perforin is involved in CTL mediated cytotoxicity, we used the antisense oligonucleotide approach. This strategy to study specific gene function has been used successfully in the past to explore, for example, the function of c-myc (Gewirtz and Calabretta, 1988) or myeloblastin (Bories et al., 1989). Here we present data showing that inhibition of perforin expression in T lymphocytes activated in vitro results in a proportional diminution of CTL cytotoxicity. -
Results Experimental design Established CTL cell lines cultured in the presence of IL-2 contain large amounts of granules and granule proteins, including perforin. Although most lines require weekly stimulation with the appropriate antigen presenting cell, the expression of perforin protein and message is never arrested. We reasoned that depletion of this granule-stored perforin pool would be difficult or even impossible, and that CTLs whose perforin protein or message expression can be induced from zero or near zero levels would provide a more suitable system for the antisense approach. Previous reports have shown that resting T cells in general do not produce measurable levels of perforin message (Garcia-Sanz et al., 1987; Liu et al., 1990). Induction of perforin message or protein requires stimulation of cells by PHA, anti-T cell -receptor complex antibodies, lectins, or antigen presented by antigen presenting cells. Thus, we chose two systems of non-activated T cells, human peripheral blood lymphocytes and mouse spleen lymphocytes to perform our experiments. These cells were stimulated with immobilized anti-T cell -receptor complex antibodies for 16-48 h in the presence or absence of perforin sense or antisense oligonucleotides. Cytotoxic activity was measured by coincubation of these activated T cells with 5'Cr-labelled P815 target cells which were coated with anti-CD3 antibodies. This readout system guarantees detection of all T cell receptor 3815
H.Acha-Orbea et al.
bearing CTLs, but excludes the measurement of lytic activity of NK cells. CTLs and oligonucleotide uptake An 18mer antisense oligodeoxynucleotide 5'-GAG CAG ACG GGC TGC CAT-3' complementary to a sequence starting at the ATG initiation codon of human perforin and an antisense 16 bp oligonucleotide of mouse perforin (5'-AAC AGG CAC GTG GCC-3') starting at nucleotide +3 of the coding region of the mRNA were synthesized. As controls, sense oligonucleotides of the complimentary sequences were synthesized. Neither sense nor antisense sequences were identical to any known nucleotide sequence available in the EMBL database. Exogenously added antisense oligodeoxynucleotides would be only minimally effective for mRNA hybridization and consequent translational arrest if they were rapidly degraded or not internalized. To verify uptake by PBL, the human perforin antisense oligodeoxynucleotide was labelled with P at its 5'-end and the percentage uptake determined. In the case of human PBLs, 1.0% of the perforin antisense oligonucleotide was associated with the PBL cellular pellet after 4 h incubation (Figure 1). This fraction decreased to 0.6% after 24 h incubation. -
Inhibition of perforn protein expression Human PBL were activated by anti-CD3 monoclonal antibodies and addition of IL-2 in the presence of either antisense or sense perforin oligomer, and protein expression was analysed by immunoblotting. Neither oligomer decreased the level of induction of granzyme B protease, the expression of which is highly upregulated upon lymphocyte activation (Schmid and Weissmann, 1987) (Figure 2). In contrast, activated PBLs treated with 40 /4M antisense perforin showed reduction in the level of perforin protein expression of up to 65 % relative to cells stimulated in the presence of sense oligomer. Inhibition of protein expression varied from 46 to 65 %, but complete absence of perforin protein was never observed. The quantity of perforin induced was, however, very low, and its detection was at the borderline of the sensitivity of our reagents, resulting in a great variation within a given experiment. 2.0
Although both antibodies used against granzyme B and perforin were made against recombinant protein and could be used at similar dilutions in Western blots and for immunomicroscopy (Hameed and Tschopp, unpublished), only the granzyme B antibodies yielded strong signals after the short time stimulation of PBL. In the experiments with mouse lymphocytes, it was not even possible to measure the low level of perforin protein levels with the available antibodies, although perforin is readily detected in CTL cell lines using the same reagents. Inhibition of lymphocyte mediated cytotoxicity Preliminary experiments to determine whether the decreased expression of perforin affected CTL effector functions indicated that induction of cytotoxicity in human PBL was not very efficient when the latter were stimulated with immobilized anti-CD3 antibodies alone for 24 h. Subsequent incubation with medium containing r-IL-2 (100 U/ml) for an additional 16 h in the absence of the stimulating antibodies highly increased the lytic activity of PBL. Up to 60% specific lysis of antibody-coated P815 target cells were obtained at a 40:1 CTL:target ratio (Figure 3a). The antisense perforin oligonucleotide was added during the activation period in three aliquots, at the beginning of CD3 induced stimulation, after 16 h and upon the change of r-IL-2-containing medium. After the activation period, antisense oligonucleotide treated cells showed a marked decrease in cytotoxicity in a dose dependent manner in the experiment displayed in Figure 3a. At a concentration of 40 itM oligomer, lysis of P815 target cells was decreased from 54% to 29%. In contrast, the level of cytotoxicity of PBL cultured in the presence of sense oligonucleotide was as high as in untreated cells. Reduced cytotoxicity was also observed in PBL activated during 16 h solely with r-IL-2 (see Figure 3b). Cells exposed to a single dose of antisense oligomers were markedly less cytotoxic (up to 62 % reduction), whereas sense oligomer sa
sa sense
OD 540 0.05
66 -
1.8 1.6 -
antisense
1.4 -
0.05
a) 1.2-
0. 1.0
-
2 0.8
-
0
27
+ +
0
0.2 I
0
I
I
5
I
I
I
10
15
20
25
Time (hr Fig. 1. Uptake of oligonucleotides in humar PBLs. Samples of 5-32plabelled anti-human perforin oligomer were incubated for 0, 1, 4, 8 and 24 h with PBL undergoing activation irn the presence of 100 U/ml IL-2 or non-activated PBLs percentage upta Ike of oligomer was determined as described in Materials and mliethods. 3816
?
Io~~~~~~
0.6 -
0.4 -
4:0 0
+.±±
Fig. 2. Inhibition of human perforin protein expression by antisense oligodeoxynucleotides. Immunoblot of protein extracts of unstimulated (-) and stimulated (+) human PBL treated with antisense (as) or sense (s) oligomer. Blots were developed with anti-granzyme B (left panel) and anti-perforin (middle panel) antibodies. Arrowheads refer to the proteins. A doublet was observed for granzyme B, most likely reflecting differences in glycosylation or granzyme X (Meier et al., 1990), a recently described granzyme B-like enzyme in lymphocytes. Molecular size markers are indicated. The right panel shows densitometric scans of the bands corresponding to human perforin as detected by Western blot analysis (see middle panel). Numbers refer relative units of the integrated peaks.
to
Perforin mediated cytotoxicity
had no effect on cytotoxicity exhibited by human lymphocytes. With mouse spleen cell derived lymphocytes, the oligonucleotides were added during the 17-21 h activation period of the lymphocytes with immobilized anti-CD3 antibody. As shown in Figure 4, the antisense oligonucleotide inhibited cytotoxic activity in a dose dependent fashion. When lytic units were calculated, at 18 ltM oligonucleotide concentration, a 5.7-fold inhibition of cytotoxicity was observed with antisense as compared with sense oligonucleotide. This inhibition decreased to 4.1-fold at 6 ,uM, 3.3-fold at 2 AM, 1.6-fold at 0.7 jiM. No inhibition was observed at 0.25 ItM. A slight inhibition caused by the three independently synthesized preparations of sense oligonucleotides was observed with murine lymphocytes. Addition of sense or antisense oligonucleotides did not have any detectable effects on blast T cell recovery after stimulation. Activation of cells for longer periods of time (48 h) resulted in 9-fold higher cytotoxic activation, but neither sense nor antisense oligonucleotides caused significant inhibitory activation in these cells. This is probably due to the relatively short half-life of the oligonucleotides in culture. In addition, it proves that the effect of the antisense oligomer is not due to toxic effects. Inhibition of lymphocyte cytotoxicity varied markedly from one experiment to another. Table I lists the results of various experiments. Each experiment was performed with a different preparation of cells. The percentage of specific 60
inhibition ranged from 69% to 3%. In no case, however, was the inhibition by the sense oligomers superior to that attained with the antisense oligomers.
Discussion Recent studies have shown that the introduction of synthetic antisense oligonucleotides can result in a specific inhibition of the expression of cellular genes. In the present study, we demonstrate the use of this strategy for the selective deletion of a granule associated protein in lymphocytes. The involvement of perforin in CTL mediated cytolysis has been hotly debated. Whereas ample evidence exists that perforin in CTL clones or cells may be, at least in part, 30
40-
18 iM
30-
6 tM
20 20-4 10
~~~~~~~~10
.Cn
C.)
cn40 -4010.7 ±M 2 iM
a
50-*OO 30 -
E
40
Effector to target ratio C
30 -_0 30 E 200*
Fig. 4. Inhibition of mouse lymphocyte mediated cytotoxicity by antiperforin oligomers. Mouse spleen cell derived lymphocytes were activated with anti-CD3 antibodies immobilized on plastic during 24 h in the presence of perforin antisense (closed triangles, dotted line) or
0 C
10
-
en
0 ._
0
I
1
1
b
0)
CoU)I50
sense (open rectangles) oligonucleotides. Lysis of P815 cells in the absence of anti-CD3 antibody is shown by closed circles. Lysis of unstimulated spleen cells in the presence of anti-CD3 antibody was
