Pancreas
Vol. 5 , No.3, pp. 267-211
8 1990 Raven Press, Ltd., New York
Binding and Biological Effects of Tumor Necrosis Factor and Gamma Interferon in Human Pancreatic Carcinoma Cells Arthur B. Raitano, Philip Scuderi, and Murray Korc Departments of Microbiology and Immunology, Internal Medicine, Biochemistry, and the Arizona Cancer Center, University of Arizona School of Medicine, Tucson, Arizona, U.S.A.
Summary: The cytotoxidcytostatic effects of recombinant human tumor necrosis factor alpha (rhTNF) and gamma interferon (rhIFN-y) were studied in five human pancreatic tumor cell lines. During a 48-h incubation, MIA PaCa-2 cells were most sensitive to rhTNF (56% cytotoxicity, 500 U/ml), T,M, cells were most sensitive to rhIFN-y (54% cytostasis, 250 U/ml), and ASPC-1 and COLO 357 cells were most sensitive to the combination of rhTNF and rhIFN-y (56 and 55% cytotoxicity, respectively, 250 U/ml of each cytokine). The PANC-1 cells were relatively insensitive to either the individual or the combined effects of these cytokines. All five cell lines exhibited specific, highaffinity receptors for 'Z51-labeled rhTNF (480-8,610 sites/cell) and rhIFN-y (2,0504,280 sitedcell). The MIA PaCa-2 cells, which were the most sensitive to the inhibitory effects of rhTNF, also possessed the largest number of '*'I rhTNF receptors; all other cell lines had a relatively low number of binding sites and low sensitivity. In contrast, no direct correlation could be made between the number of IFN-y binding sites and inhibitory sensitivity in any of the cell lines. Incubation of COLO 357 cells at 37°C with either lZ5IrhTNF or lZ5IrhINF-y led to internalization of the respective 1251-labeledligand. Our findings document the presence of cytokine receptors in human pancreatic carcinoma cells and suggest that postreceptor events rather than differences in receptor number or affinity more likely govern the responsiveness of pancreatic cancer cells to TNF and IFN-7. Key Words: Cytokines-CytotoxicityReceptors-Internalization.
Pancreatic cancer is the fifth leading cause of cancer-related death in the United States (1). The lack of effective anticancer therapies for pancreatic neoplasms has contributed to the very poor prognosis for this disease. The cloning of the genes that en-
code tumor necrosis factor (TNF) and gamma interferon (IFN-y) has allowed for the availability of these compounds for use in novel therapeutic approaches to a number of neoplastic diseases (2,3). However, the potential usefulness of these cytokines in the treatment of pancreatic cancer is not known. Tumor necrosis factor is a pleiotropic cytokine that has selective antiproliferative and/or cytotoxic effects in a variety of transformed cells in vitro (43). It causes the hemorrhagic necrosis of transplantable tumors in vivo (6,7) and has been impli-
Manuscript received May 18, 1989; revised manuscript accepted August 10, 1989. Address correspondenceand reprint requests to Dr. M. Korc, Section of Endocrinology, Med Sci I, C240, University of California, Irvine, CA 92717, U.S.A.
267
A . B . RAITANO ET AL.
268
cated as a mediator of macrophage tumoricidal activity (8,9). The cytotoxic/cytostatic effects of TNF are sometimes enhanced by the immunomodulating protein IFN-y (43). The IFN-y alone may also exhibit potent antitumor activity (4,5,10) and may increase TNF binding in target cells (11-14). Highaffinity binding sites for TNF and IFN-y have been documented on a variety of normal and transformed cells (15-20). However, the presence of receptors for these cytokines has not been reported in human pancreatic carcinoma cells. Thus, the relationship between the binding and biological responsiveness of human pancreatic carcinoma cells to these cytokines is not known. In this study, we have tested the cytotoxic/ cytostatic effects of TNF and IFN-y in a number of human pancreatic carcinoma cell lines and analyzed each cell line for the presence of TNF and IFN-y receptors. We now report that recombinant human TNF and IFN-y exert differential cytotoxidcytostatic effects in several human pancreatic carcinoma cell lines in vitro that are synergistic in some instances. Further, all of the cell lines studied possess high-affinity binding sites for both TNF and IFN-y. MATERIALS AND METHODS Cell culture The MIA PaCa-2, PANC-1, and ASPC-1 cells were obtained from the American Type Culture Collection. The T3M4(21) and COLO 357 (22) cells were obtained from Dr. R. S. Metzgar (Duke University). The C8161 melanoma cells were obtained from Dr. Frank Meyskens (University of California at Irvine). Cells were routinely propagated in monolayer culture in a humidified incubator at 37°C in a 5% C0,/95% air atmosphere. The MIA PaCa-2, COLO 357, PANC-1, and C8161 cells were grown in Dulbecco’s modified Eagle’s medium; T3M, and ASPC-1 cells were grown in Roswell Park Memorial Institute 1640 medium. Media were supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Hyclone Inc., Logan, UT, U.S.A.), 100 p M nonessential amino acids, 1 mM sodium pyruvate, 2 mM L-glutamine, 50 U/ml of penicillin, and 50 pg/d of streptomycin (Irvine Scientific, Santa Ana, CA, U.S.A.). Reagents The recombinant human tumor necrosis factor (rhTNF) (specific activity, 5.02 x lo7 U/mg) and recombinant human interferon gamma (rhIFN-y) Pancreas, Vol. 5 , No. 3, 1990
(specific activity, 2 x lo7 U/mg) were gifts of Genentech Inc. (S. San Francisco, CA, U.S.A.). Rabbit polyclonal antibodies specific for rhTNF and rhIFN-y were produced by immunizing New Zealand white rabbits with the respective cytokine emulsified in Freund’s complete adjuvant (Difco, Inc., Detroit, MI, U.S.A.). Each antiserum contained at least 400 neutralizing unitdpl as determined by the inhibition of cytotoxic effects of rhTNF on L929 cells and inhibition of the growth inhibitory effects of rhIFN-y on T3M4 cells. Actinomycin D, cyclohexamide, and MTT (3-4,5-dimethylthiazol-2-yl-2,5-diphenyltetrazoliumbromide) were obtained from Sigma Chemical Co. (St. Louis, MO, U.S.A.). Iodogen reagent (1,3,4,6tetrachloro-3,6-diphenylglycouril)was obtained from Pierce Chemical Co. (Rockford, IL, U.S.A.). Cytotoxicity assay The cytotoxicity assay was a modification of the assay developed by Mosmann (23), which relies on the uptake and metabolism of the tetrazolium salt, MTT. Only viable cells take up and metabolize the compound into dark blue formazon crystals. The crystals are solubilized in alcohol, and the amount of reagent metabolized can be quantitated in a colorimetric assay. Briefly, cell monolayers were trypsinized with 0.25% trypsin (Irvine Scientific), washed with complete medium, plated at 25,000 cells/well in 96 well microtiter plates (Costar), and allowed to adhere. Cells were then incubated with the indicated concentrations of either TNF or IFN-y or their combination for 48 h in a final volume of 100 pl. Following the incubation period, media were aspirated and replaced with fresh media containing 50 pg/ml of MTT, and the cells were incubated for an additional 4 h. The reaction was stopped by the addition of 150 p1 of acidified isopropanol (0.04 N) followed by vigorous mixing. Plates were then centrifuged to remove cell debris, and 100 pl of colored supernatant from each well was transferred to clean plates. Optical density (OD) was determined in a Titertek multiscan using a 570-nm filter. The OD readings were converted to the percentage of control growth using the following formula: % of control growth
=
experimental OD - background OD x 100 maximal OD - background OD The maximal OD was obtained from control wells containing cells incubated with fresh medium alone,
EFFECTS OF TNF AND IFN-y IN PANCREATIC CANCER CELLS
269
and the background OD was obtained by the concomitant addition of 10 pl of 1 N of HC1 and MTT dye to control wells. Cytotoxic/cytolytic effects were also expressed as the percentage of cytotoxicity, which was calculated using the formula 1 minus the percentage of control growth. Inasmuch as this assay cannot distinguish between dead (or severely damaged) cells and metabolically inactive cells, cell numbers (and viability) were also determined by trypsinization and cell counting using a hemocytometer and trypan blue dye exclusion. Similar effects on cell viability were seen with either the MTT assay or by cell counting.
gens on C8161 melanoma cells with that of unlabeled rhIFN-y. The melanoma cell line was used as a positive control inasmuch as melanocytes and many melanoma cell lines are responsive to the induction of HLA-DR antigens by IFN-y (28). Binding experiments were carried out in 12 well plates seeded with 5 X lo5 cellslwell. Monolayers were washed with complete medium and incubated at 37°C for 2 h with binding medium containing 200,000 cpm of the respective radiolabeled cytokine (maximal binding at 37°C occurred within 2 h for all cell lines). Binding was terminated by washing the cells four times with Hank's balanced salt solution containing 0.1% BSA at 4°C. Cells were then solubilized in 0.5 M NaOH, and cell-associated radioBinding experiments The rhTNF and rhIFN-y were labeled with 1251 activity was measured using a gamma counter. Nonspecific binding, determined by the addition of by modifications of the iodogen method previously a 250-500-fold excess of the respective unlabeled described for both cytokines (11,24). Briefly, the cytokine, was never greater than 20% of the total rhTNF or rhIFN-y (25 pg/lOO p1) and 1 mCi of '"I binding. To determine the receptor number and af(New England Nuclear, Boston, MA, U.S.A.) were finity, cells (5 x 1o5/we1i, 12 well plates) were inadded to a test tube coated with 10-15 pg of iodogen reagent (~4,3,4,6-tetrachloro-3,6-diphenylglycouril) cubated under equilibrium binding conditions (4°C for 4-6 h) with increasing concentrations (1-50 and incubated at 4°C for 10 min. The reaction was ng/ml) of either 1251 rhTNF or 'I rhIFN-y. They stopped by removal of the mixture from the test were analyzed by the method of Scatchard (29) ustube. Free ''%odine was removed by passage over a ing the LIGAND program (30). This analysis was PD- 10 column (Pharmacia) equilibrated with a carried out with the assumption that the molecular phosphate-buffered saline solution (pH 7.4) conmoieties of "'I rhTNF and '"I rhIFN-y that bound taining 0.1% bovine serum (BSA). The peak radioto the cells were trimers (MW = 51 kDa) and active protein fraction was always greater than 95% dimers (MW = 32 kDa), respectively (18,19,26,27). TCA precipitable and was used for all binding exrhTNF and 1251 rhIFN-y internalTo monitor 1251 periments. Recovery of "'I rhTNF was 80-90% as ization, COLO 357 cells (5 x lo5 in 12 well plates) determined by enzyme-linked immunosorbent aswere incubated at 37°C with 200,000 cpm of 12'1 say (ELISA) for TNF as previously described (25). rhTNF or 300,000 cpm of '1 rhIFN-y for either 30 Recovery of 1251rhIFN-y was 55-70% as deteror 120 min and washed as in binding experiments. mined by a commercially available ELISA for Cells were then incubated for 4 min at 4°C with 0.5 IFN-y (Amgen, Thousand Oaks, CA, U.S.A.). The M NaCl that was titrated to pH 2.5 with acetic acid specific activity of both labeled cytokines ranged (3 1). Following elution of membrane-bound (acid from 35,000 to 60,000 cpm/ng for several iodinadissociable) radioactivity, cells were solubilized in tions , 0.5 M NaOH to determine the internalized (acid When iodinated rhTNF and rhIFN-y were anaresistant) radioactivity. lyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, monomeric moieties of 17 and 16 Statistical analysis kilodaltons (kDa), respectively, were observed Statistical analysis of multiple comparisons was (2,3). However, under nonreducing conditions, a carried out using analysis of variance (ANOVA) dimeric form of rhIFN-y and dimeric and trimeric and Tukey's multiple comparison procedure. forms of rhTNF were observed as previously deWhere appropriate, data were also analyzed by a scribed (18,19,26,27). The '"I rhTNF retained full two-tailed Student's t test. Cellular viability data biological activity as determined by comparing its were analyzed by a dose-effect analysis computer cytotoxic actions on L929 cells with the actions of program that enables determination of antagonistic, unlabeled rhTNF (11). The '251-labeled rhIFN-y readditive, or synergistic effects of multiple drug tained full biological activity as determined by comtreatments (32). paring its effect on the induction of HLA-DR antiPancreas, Vol. 5 , No. 3, 1990
A . B . RAITANO ET AL.
270 RESULTS
Effects of rhTNF and rhIFN-y on cellular viability Four of five human pancreatic tumor cell lines demonstrated at least a 50% decrease in growth after a 48-h incubation with a particular cytokine or
A loo
combination of cytokines using the MTT cell growth assay (Fig. 1). The growth inhibitory effects produced by these cytokines were found to be highly significant when analyzed by ANOVA and Tukey's multiple comparison procedure. The MIA PaCa-2 cells were most sepsitive to rhTNF, exhib-
1
1-
B
loo I_
T 75
g::g
" 5 7
a
0
50-
i o u p
MIA PaCa-2 MF
-A-
0 IFNy
+ 0
100
300
200
500
0
100
CYTOKINE (UNITSIML)
C
100
75
TNF
0
IFNy TNF+IFNy
+
TNFclFNy
400
-A-
200
300
400
500
CYTOKINE (UNITSIML)
1
D
100 1
1 6COLO 357
ASPC-1 -b Q
+
0
100
300
200
TNF IFNy TNF+IFNy
400
TNF IFNy I- TNFtIFNy -&
4
500
0
CYTOKINE (UNITSIML)
100
200
300
400
500
CYTOKINE (UNITSIML)
E l75
o
o
]
v
PANC-1 *MF CF IFNy + TNF+IFNy
0
100
200
300
400
CYTOKINE (UNITSIML)
Pancreas, Vol. 5, No. 3, 1990
500
FIG. 1. Effects of rh tumor necrosis factor (TNF) and rh gamma interferon (IFN-y) on human pancreatic carcinoma cell growth. A: MIA PaCa-2 cells. B T,M, cells. C: ASPC-1 cells. D: COLO 357 cells. E PANC-1 cells. Cells were incubated for 48 h with the indicated concentrations of rhTNF (triangle), rhIFN-y (open square), or rhTNF in the presence of 250 U/ml of rhIFNy (filled square). Cytotoxidcytostatic effects were determined by the 3-4,5-dimethylylthiazol-2-yl-2,5-dipheny~tetrazolium bromide assay as described in Materials and Methods. Data are the means f SEM from four separate experiments in which three to five determinations were made for each experimental condition.
EFFECTS OF TNF AND IFN-y IN PANCREATIC CANCER CELLS
271
iting a 56% decrease in growth at a concentration of 500 U/ml (p < 0.001, Fig. 1A). This effect appeared to be cytotoxic/cytolytic, as it was associated with marked detachment of cells and a marked increase in trypan blue dye uptake. The T3M4 cells were most sensitive to rhIFN-y, demonstrating a 54% decrease in growth at a concentration of 250 U/ml (p < 0.001, Fig. 1B). The majority of T3M4 cells remained attached and excluded trypan blue dye, suggesting that the action of IFN-y on these cells was mainly cytostatic. The effects of rhTNF in MIA PaCa-2 cells or rhIFN-y in T3M4cells were not enhanced by the addition of the reciprocal cytokine. The PANC-1 cells were resistant to the inhibitory effects of either cytokine alone and to their combined activity, exhibiting only an 18% decrease in cell growth (Fig. 1E). In contrast to the other cell lines, ASPC-1 and COLO 357 cells were most sensitive to the combination of rhTNF and rhIFN-y (250 U/ml of each), exhibiting a 56 and a 55% decrease in growth, respectively (p < 0.001, Fig. 1C and D). Unlike the effects produced by single cytokine treatment, in both cell lines concomitant treatment resulted in cytotoxic/cytolytic effects, as evidenced by marked cell detachment and trypan blue dye uptake by the detached cells. Analysis of the cytotoxic effects of rhTNF and rhIFN-y in ASPC-1 and COLO 357 cells with various dilutions of the cytokines in a constant ratio of 10 to 1 of rhTNF to rhIFN-y, 10 to 1 of rhIFN-y to rhTNF, or a 1-to-1 ratio of both cytokines revealed that the effects of the cytokines were synergistic (data not shown), as determined by a dose-effect analysis computer program (32). The most potent growth inhibitory effect of a cytokine($ on each of the five human pancreatic carcinoma cell lines is summarized in Table 1.
Effects of pretreatment and addition of specific neutralizing antisera on cytoxicity Inasmuch as ASPC-1 and COLO 357 cells were most sensitive to the combined actions of rhTNF and rhIFN-y, we next examined the effects of pretreating these cells with one cytokine prior to exposure to the second agent. Pretreatment for 24 h with either rhTNF or rhIFN-y, followed by a 48-h incubation with the reciprocal cytokine, was not as effective as the simultaneous treatment with both cytokines in either ASPC-1 cells (p < 0.005, Fig. 2A) or COLO 357 cells (p < 0.02, Fig. 2B), based on analysis of the data by ANOVA and a two-tailed Student's t test. For both cell lines, treating initially with rhIFN-y tended to produce greater cytotoxicity than treating initially with rhTNF. However, this difference was not statistically significant. The combined cytotoxic effects of rhTNF and rhIFN-y were completely neutralized in the presence of specific antisera to both cytokines (Fig. 3A and B; 0 h). The addition of normal rabbit antiserum did not reduce cytotoxicity (data not shown). When the addition of the rhTNF-specific and the rhIFN-y-specific antisera was delayed by 4 h, complete neutralization was no longer attained (Fig. 3A and B; 4 h). However, the cytotoxic effect was still markedly blocked as compared to cells that were not incubated with antisera (p < 0.001). Delaying the addition of both antisera for 24 h further decreased their neutralizing effects, but the cytotoxic action of the cytokines was still partially blocked (p < 0.05; Fig. 3A and B; 24 h). Analysis of the data by ANOVA and Tukey's multiple comparison procedure indicated that the neutralizing effects of either anti-rhTNF or anti-rhIFN-y serum were not significantly different from each other or from those of their combination.
TABLE 1. Cytostaticlcytotoxic effects of rhTNF and rhIFN-y" in human pancreatic carcinoma cells
Binding of "'I rhTNF and 1'2 rhIFN-y To determine whether cultured human pancreatic carcinoma cells bind either TNF or IFN-y at the same temperature at which they exert their biological actions, binding studies were carried out at 37°C using '251-labeled rhTNF and rhIFN-y. Maximal binding of both '251-labeled cytokines to the pancreatic cancer cell lines at 37°C occurred between 90-120 min (data not shown). The cellassociated '"I radioactivity after a 2-h incubation at 37°C with the respective '251-labeled cytokine is shown in Table 2. The MIA PaCa-2 and T3M4cells exhibited relatively high levels of I2'I rhTNF bind-
Cell line ASPC-1 COLO 357 MIA PaCa-2 PANC-1' T3M4
Most effective treatment
+ +
Inhibitory effectb
rhTNF rhIFN-y rhTNF rhIFN-y rhTNF
Cytotoxic Cytotoxic Cytotoxic
rhIFN-y
Cytostatic
-
-
Tumor necrosis factor, TNF; interferon, IFN. The growth inhibitory effect of the most effective cytokine treatment was determined by microscopic examination. Criteria for cytotoxic effects included the appearance of detached and floating cells and the inability of the cells to exclude trypan blue. 'The PANC-1 cells were relatively resistant to all treatments. a
Pancreas, Vol. 5 , No.3, 1990
A . B . RAITANO ET AL.
272
>
I
I-
TNF + lFNy
PTNF -IFNy
PlFNy -TNF
I
TNF + IFNy
PTNF -1FNy
PlFNy -TNF
FIG. 2. Effects of pretreatment with cytokines. A ASPC-1 cells. B: COLO 357 cells. Cells were preincubated for 24 h in complete medium without cytokine or with 250 U/ml of either rh tumor necrosis factor (TNF) (PTNF) or rh gamma interferon (IFN-y) (PIFN-y). Cells were then washed two times and further incubated with 250 U/ml of rhTNF and rhIFN-y (TNF + IFN-y), 250 U/mlof rhIFN-y (+IFN-y), or 250 U/ml of rhTNF (+TNF), respectively. Cytotoxidcytostatic effects were determined by the 3-4,5-dimethylylthiazol-2-yl-2,5-diphenyltetrazolium bromide assay. Data are the means SEM from three separate experiments, in which three to five determinations were made for each experimental condition.
*
ing, ASPC-1 and COLO 357 cells exhibited intermediate levels of binding, and PANC-1 cells exhibited the least binding. For '"I rhIFN-y binding, ASPC-1 and MIA PaCa-2 cells bound the highest levels, while COLO 357, PANC-1, and T3M4cells
bound relatively similar amounts. The addition of 500 ng/ml of unlabeled epidermal growth factor (EGF) or of the reciprocal cytokine did not alter the specific binding of either labeled cytokine (data not shown). Internalization of "'I rhTNF and 12'1 rhIFN-y by COLO 357 cells Following binding, most peptide hormones are internalized via clathrin-coated pits into "receptosome" structures that may subsequently fuse with lysosomal compartments leading to degradation of the ligand and/or its receptor (33). We therefore sought to determine if pancreatic carcinoma cells internalize rhTNF and rhIFN-y. Accordingly, COLO 357 cells were incubated at 37°C for either 30 or 120 min with either '1 rhTNF or '"I rhIFN-y, and the surface-bound radioactivity was removed by acid wash (31). The remaining acid-resistant radioactivity therefore represented internalized ligand molecules. As can be seen in Table 3, at 30 min of incubation, approximately equal amounts of either '1 rhTNF or "'I rhIFN-y radioactivity were localized to the acid-sensitive (cell surface) and to the acid-resistant (internalized) compartments. In contrast, at 120 min of incubation, most of the radioactivity for both ligands was localized to the acidresistant (internal) compartment. Further, for both ligands, the amount of surface-bound radioactivity was less at 120 than at 30 min of incubation, suggesting there was a loss of cell surface binding sites
75 A
ASPC-1
B
COLO 357 +ANTI-TNF.IFNy +ANTI-TNF +ANTI-IFNy
+ANTI-TNF,IFNy
H +ANTI-TNF +ANTI-IFNy
50
25
0 NOAB OHRS +ANTI-TNF. +ANTI-IFNy
4 HRS
24HRS
NOAB
OHRS
4 HRS
24 H R S
+ ANTI-TNF, +ANTI-IFNy
FIG. 3. Effects of neutralizing polyclonal antisera on cytotoxicity. A ASPC-1 cells. B: COLO 357 cells. Cells were incubated for 48 h with 250 U/ml each of rh tumor necrosis factor (TNF) and rh gamma interferon (IFN-y) in the absence or presence of the indicated antisera (5 pVwell), which were added at 0,4, or 24 h following initiation of cytokine treatment. Cytotoxic/cytostatic effects were determined by the 3-4,5-dimethylylthiazol-2-yl-2,5-diphenyltet~zoliumbromide assay. Data are the means SEM from four separate experiments, in which three to five determinations were made for each experimental condition.
*
Pancreas, Vol. 5 , No.3, 1990
EFFECTS OF TNF AND IFN-y IN PANCREATIC CANCER CELLS TABLE 2. Specific "'I rhTNF and ''I rhIFN-y" binding to pancreatic carcinoma cells '1 rhTNF, cpm ASPC-1 COLO 357 MIA Paca-2 PANC-1 T3M4
1,423 f 25 2,364 f 186 8,157 f 89 582 f 59 3,678 f 150
rhIFN-y, cpm
'I
12,839 f 548 2,189 f 55 5,886 f 226 3,603 f 198 3,060 f 118
Cells, 1 x lo6, were incubated at 37°C for 2 h with 100,000 cpmlml of the indicated "'I-labeled cytokine. Data are the means f SD of triplicate determinations from a representative (of three) experiment. Tumor necrosis factor, TNF; interferon, IFN.
at the later time point due to the internalization of occupied receptors. Characteristics of TNF and IFN-y receptors in human pancreatic carcinoma cells All the cell lines specifically bound '"I-labeled rhTNF and rhIFN-y at 4"C, with maximal binding for both labeled cytokines occurring between 4-6 h, depending on the cell line (data not shown). To determine the number of binding sites per cell and the affinity of each cytokine for its respective receptor, binding studies with increasing concentrations of '251-labeledligands were next carried out at 4°C for either 4 or 6 h, to allow for maximal ligand binding. Scatchard analysis of the equilibrium binding data obtained with 'I rhTNF revealed the presence of a single order of high-affinity binding sites (dissociation constant (kd) = 0.862.0 x lO-"M) for all five pancreatic cancer cell lines, and 49043,610 binding siteskell (Table 1;representative plot, MIA PaCa-2 cells, Fig. 4A). Scatchard analysis of the equilibrhIFN-y fit on rium binding data obtained with 1251 curvilinear plots, suggesting that high- and lowaffinity components existed for the '1 rhIFN-y
273
binding sites. However, the low-affinity component of the curvilinear plot was due to binding at high ligand concentrations of "'I rhIFN-y, and was characterized by the LIGAND program as representing nonspecific binding. This effect was also evident by the relative linearity of the saturation plot for '"I rhIFN-y at high ligand concentrations (inset, Fig. 4B). Similar binding data for IFN-y depicting spurious low-affinity binding sites have been described by other investigators (18,19). The LIGAND analysis of the 1251 rhIFN-y binding data was therefore carried out with the assumption that a single order of binding sites was present in the pancreatic cancer cells. This analysis revealed the presence of high-affinity binding sites for '"I rhIFN-y (kd = 2.04.4 x lo-" M ) in all five cell lines, and 2,0504,280 binding siteskell (Table 4; representative plot, MIA PaCa-2 cells, Fig. 4B). DISCUSSION In this study, we have determined that human pancreatic tumor cell lines demonstrate a variable sensitivity to the effects of recombinant TNF and IFN-y. After a 48-h incubation, three of five cell lines (MIA PaCa-2, T3M4,and ASPC-1) were sensitive to the inhibitory effects of either TNF or IFN-y (greater than 30% growth inhibition). Concomitant treatment with both TNF and IFN-y resulted in synergistic antiproliferative effects in ASPC-1 and COLO 357 cells but not in MIA-PaCa-2 and T,M4 cells. In contrast, PANC-1 cells were relatively insensitive to either single or combined treatment with TNF or IFN-y (less than 20% growth inhibition). These data support the findings of Schmiegel et al., who also demonstrated that human pancreatic cancer cell lines exhibited varied
TABLE 3. Internalization of '''1 rhTNF and "'I rhIFN-y" by COLO 357 cells 30 min, cpm
120 min, cpm
Surface
Internalized
Surface
Internalized
274 ? 50 3,873 -t 666
322 f 12 3,983 f 199
201 2 23 1,363 rfr 157
1,059 f 87 6,707 f 134
~~
TNF~ I F N ~
a Cells, 5 x lo5, were incubated for 10 or 120 min with either 200,000 cpmlml of lZ5I rh tumor necrosis factor (TNF) or 300,000 c p d m l of '"I rh gamma interferon (IFN-y). Cells were washed and then incubated 4 min at 4°C with 0.5 M NaCl that was titrated to pH 2.5 with acetic acid. Following elution of surface-bound (acid-dissociable) radioactivity, cells were solubilized in 0.5 M NaOH to determine the internalized (acid-resistant) radioactivity. Data are the mean ? SD from a representative experiment.
Pancreas, Vol. 5 , No. 3, 1990
2 74
A . B . RAITANO ET AL. 0.084
2
FIG. 4. Scatchard analysis of 1251rh tumor necrosis factor (TNF) and l2’I rh gamma interferon (IFN-y) binding to MIA PaCa-2 cells. A I2’I rhTNF. B: ‘*’I rhIFN-y. Cells were incubated with increasing concentrations of the respective lZ’I-labeled ligand under equilibrium conditions (4°C) as described in Materials and Methods. Data shown are from a representative (of three) experiment, as analyzed by the method of Scatchard using the LIGAND program. Insets are saturation plots of the same data.
0.042
0
.ooo 0.00
2.35
1251
4.69
7.04
rhTNF BOUND (Xi0-l2M)
B
0.071
L
&
0.034
\
I
-0.003
, 0.00
I
1.28
i251
sensitivity to TNF and IFN-y and that a subset of the cell lines was sensitive to the synergistic effects of these cytokines (34). The actions of TNF and IFN-y in ASPC-1 and COLO 357 cells were completely neutralized by polyclonal antisera specific for the respective cytokines, indicating that the observed effects were not due to the release of nonspecific cytotoxins by the cells. Delaying the addition of either antiserum resulted in a progressive increase in the cytotoxic effects of the cytokines (Fig. 3). A short-term exposure (4 h) to both TNF and IFN-y was sufficient to induce an irreversible cytotoxic effect in at least a portion of these cells. However, pretreatment of Pancreas, Yol. 5, No. 3. 1990
I
I
2.55
= I 3.83
rhlFN-y BOUND (X10-l2M)
ASPC-1 and COLO 357 cells for 24 h with either TNF or IFN-y followed by single treatment with the reciprocal cytokine did not significantly enhance growth inhibition by comparison to the effects observed in the absence of pretreatment (Fig. 2). These observations suggest that the synergistic effects of the cytokines during short-term incubations required the continuous presence of both TNF and IFN-7. It is generally accepted that the first step in the initiation of the actions of a particular cytokine is dependent on its binding to a specific cell surface receptor. Specific receptors for TNF and IFN-y have been demonstrated in a variety of normal and
EFFECTS OF TNF AND IFN-7 IN PANCREATIC CANCER CELLS TABLE 4. Binding characteristics of ' ' 9rhTNF and 1251rhIFN-y" in human pancreatic carcinoma cells TNF binding sites Cell line ASPC-1 COLO 357 MIA PaCa-2 PANC-1
T3M4
No./ cell 480 1,080 8,610 1,830 1,230
Kd
(XIO-''
0.92 1.16 0.86 2.00 1.21
M)
IFN-y binding sites No./ cell
(XIO-" M)
6,280 3,850 2,050 2,280 3,470
3.64 6.41 2.60 2.04 5.53
Kd
Binding was carried out with increasing concentrationsof labeled ligand under equilibrium conditions (4°C) as described in Materials and Methods. Data are the means of three separate experiments for each cell line and for each cytokine, as analyzed by the method of Scatchard using the LIGAND program. Tumor necrosis factor, TNF; interferon, IFN.
transformed cells (15-20). Our findings are the first to document the presence of specific high-affinity receptors for both TNF and IFN-y on human pancreatic carcinoma cells. The TNF receptors in these cells had a kd on the order of 0.5-2 X lo-'' M, which is consistent with the values (1-3 x 10- M) reported for various other cells (15-17). The number of TNF receptors per cell for the five cell lines studied ranged from 480 to 8,600. The largest number of TNF receptors (8,600) and the greatest amount of cell-associated "'I rhTNF radioactivity at 37°C was seen in MIA PaCa-2 cells. This cell line was also the most sensitive among the cell lines studied to the growth inhibitory effects of TNF. All the other cell lines had low numbers of TNF receptors (
Vol. 5 , No.3, pp. 267-211
8 1990 Raven Press, Ltd., New York
Binding and Biological Effects of Tumor Necrosis Factor and Gamma Interferon in Human Pancreatic Carcinoma Cells Arthur B. Raitano, Philip Scuderi, and Murray Korc Departments of Microbiology and Immunology, Internal Medicine, Biochemistry, and the Arizona Cancer Center, University of Arizona School of Medicine, Tucson, Arizona, U.S.A.
Summary: The cytotoxidcytostatic effects of recombinant human tumor necrosis factor alpha (rhTNF) and gamma interferon (rhIFN-y) were studied in five human pancreatic tumor cell lines. During a 48-h incubation, MIA PaCa-2 cells were most sensitive to rhTNF (56% cytotoxicity, 500 U/ml), T,M, cells were most sensitive to rhIFN-y (54% cytostasis, 250 U/ml), and ASPC-1 and COLO 357 cells were most sensitive to the combination of rhTNF and rhIFN-y (56 and 55% cytotoxicity, respectively, 250 U/ml of each cytokine). The PANC-1 cells were relatively insensitive to either the individual or the combined effects of these cytokines. All five cell lines exhibited specific, highaffinity receptors for 'Z51-labeled rhTNF (480-8,610 sites/cell) and rhIFN-y (2,0504,280 sitedcell). The MIA PaCa-2 cells, which were the most sensitive to the inhibitory effects of rhTNF, also possessed the largest number of '*'I rhTNF receptors; all other cell lines had a relatively low number of binding sites and low sensitivity. In contrast, no direct correlation could be made between the number of IFN-y binding sites and inhibitory sensitivity in any of the cell lines. Incubation of COLO 357 cells at 37°C with either lZ5IrhTNF or lZ5IrhINF-y led to internalization of the respective 1251-labeledligand. Our findings document the presence of cytokine receptors in human pancreatic carcinoma cells and suggest that postreceptor events rather than differences in receptor number or affinity more likely govern the responsiveness of pancreatic cancer cells to TNF and IFN-7. Key Words: Cytokines-CytotoxicityReceptors-Internalization.
Pancreatic cancer is the fifth leading cause of cancer-related death in the United States (1). The lack of effective anticancer therapies for pancreatic neoplasms has contributed to the very poor prognosis for this disease. The cloning of the genes that en-
code tumor necrosis factor (TNF) and gamma interferon (IFN-y) has allowed for the availability of these compounds for use in novel therapeutic approaches to a number of neoplastic diseases (2,3). However, the potential usefulness of these cytokines in the treatment of pancreatic cancer is not known. Tumor necrosis factor is a pleiotropic cytokine that has selective antiproliferative and/or cytotoxic effects in a variety of transformed cells in vitro (43). It causes the hemorrhagic necrosis of transplantable tumors in vivo (6,7) and has been impli-
Manuscript received May 18, 1989; revised manuscript accepted August 10, 1989. Address correspondenceand reprint requests to Dr. M. Korc, Section of Endocrinology, Med Sci I, C240, University of California, Irvine, CA 92717, U.S.A.
267
A . B . RAITANO ET AL.
268
cated as a mediator of macrophage tumoricidal activity (8,9). The cytotoxic/cytostatic effects of TNF are sometimes enhanced by the immunomodulating protein IFN-y (43). The IFN-y alone may also exhibit potent antitumor activity (4,5,10) and may increase TNF binding in target cells (11-14). Highaffinity binding sites for TNF and IFN-y have been documented on a variety of normal and transformed cells (15-20). However, the presence of receptors for these cytokines has not been reported in human pancreatic carcinoma cells. Thus, the relationship between the binding and biological responsiveness of human pancreatic carcinoma cells to these cytokines is not known. In this study, we have tested the cytotoxic/ cytostatic effects of TNF and IFN-y in a number of human pancreatic carcinoma cell lines and analyzed each cell line for the presence of TNF and IFN-y receptors. We now report that recombinant human TNF and IFN-y exert differential cytotoxidcytostatic effects in several human pancreatic carcinoma cell lines in vitro that are synergistic in some instances. Further, all of the cell lines studied possess high-affinity binding sites for both TNF and IFN-y. MATERIALS AND METHODS Cell culture The MIA PaCa-2, PANC-1, and ASPC-1 cells were obtained from the American Type Culture Collection. The T3M4(21) and COLO 357 (22) cells were obtained from Dr. R. S. Metzgar (Duke University). The C8161 melanoma cells were obtained from Dr. Frank Meyskens (University of California at Irvine). Cells were routinely propagated in monolayer culture in a humidified incubator at 37°C in a 5% C0,/95% air atmosphere. The MIA PaCa-2, COLO 357, PANC-1, and C8161 cells were grown in Dulbecco’s modified Eagle’s medium; T3M, and ASPC-1 cells were grown in Roswell Park Memorial Institute 1640 medium. Media were supplemented with 10% heat-inactivated fetal bovine serum (FBS) (Hyclone Inc., Logan, UT, U.S.A.), 100 p M nonessential amino acids, 1 mM sodium pyruvate, 2 mM L-glutamine, 50 U/ml of penicillin, and 50 pg/d of streptomycin (Irvine Scientific, Santa Ana, CA, U.S.A.). Reagents The recombinant human tumor necrosis factor (rhTNF) (specific activity, 5.02 x lo7 U/mg) and recombinant human interferon gamma (rhIFN-y) Pancreas, Vol. 5 , No. 3, 1990
(specific activity, 2 x lo7 U/mg) were gifts of Genentech Inc. (S. San Francisco, CA, U.S.A.). Rabbit polyclonal antibodies specific for rhTNF and rhIFN-y were produced by immunizing New Zealand white rabbits with the respective cytokine emulsified in Freund’s complete adjuvant (Difco, Inc., Detroit, MI, U.S.A.). Each antiserum contained at least 400 neutralizing unitdpl as determined by the inhibition of cytotoxic effects of rhTNF on L929 cells and inhibition of the growth inhibitory effects of rhIFN-y on T3M4 cells. Actinomycin D, cyclohexamide, and MTT (3-4,5-dimethylthiazol-2-yl-2,5-diphenyltetrazoliumbromide) were obtained from Sigma Chemical Co. (St. Louis, MO, U.S.A.). Iodogen reagent (1,3,4,6tetrachloro-3,6-diphenylglycouril)was obtained from Pierce Chemical Co. (Rockford, IL, U.S.A.). Cytotoxicity assay The cytotoxicity assay was a modification of the assay developed by Mosmann (23), which relies on the uptake and metabolism of the tetrazolium salt, MTT. Only viable cells take up and metabolize the compound into dark blue formazon crystals. The crystals are solubilized in alcohol, and the amount of reagent metabolized can be quantitated in a colorimetric assay. Briefly, cell monolayers were trypsinized with 0.25% trypsin (Irvine Scientific), washed with complete medium, plated at 25,000 cells/well in 96 well microtiter plates (Costar), and allowed to adhere. Cells were then incubated with the indicated concentrations of either TNF or IFN-y or their combination for 48 h in a final volume of 100 pl. Following the incubation period, media were aspirated and replaced with fresh media containing 50 pg/ml of MTT, and the cells were incubated for an additional 4 h. The reaction was stopped by the addition of 150 p1 of acidified isopropanol (0.04 N) followed by vigorous mixing. Plates were then centrifuged to remove cell debris, and 100 pl of colored supernatant from each well was transferred to clean plates. Optical density (OD) was determined in a Titertek multiscan using a 570-nm filter. The OD readings were converted to the percentage of control growth using the following formula: % of control growth
=
experimental OD - background OD x 100 maximal OD - background OD The maximal OD was obtained from control wells containing cells incubated with fresh medium alone,
EFFECTS OF TNF AND IFN-y IN PANCREATIC CANCER CELLS
269
and the background OD was obtained by the concomitant addition of 10 pl of 1 N of HC1 and MTT dye to control wells. Cytotoxic/cytolytic effects were also expressed as the percentage of cytotoxicity, which was calculated using the formula 1 minus the percentage of control growth. Inasmuch as this assay cannot distinguish between dead (or severely damaged) cells and metabolically inactive cells, cell numbers (and viability) were also determined by trypsinization and cell counting using a hemocytometer and trypan blue dye exclusion. Similar effects on cell viability were seen with either the MTT assay or by cell counting.
gens on C8161 melanoma cells with that of unlabeled rhIFN-y. The melanoma cell line was used as a positive control inasmuch as melanocytes and many melanoma cell lines are responsive to the induction of HLA-DR antigens by IFN-y (28). Binding experiments were carried out in 12 well plates seeded with 5 X lo5 cellslwell. Monolayers were washed with complete medium and incubated at 37°C for 2 h with binding medium containing 200,000 cpm of the respective radiolabeled cytokine (maximal binding at 37°C occurred within 2 h for all cell lines). Binding was terminated by washing the cells four times with Hank's balanced salt solution containing 0.1% BSA at 4°C. Cells were then solubilized in 0.5 M NaOH, and cell-associated radioBinding experiments The rhTNF and rhIFN-y were labeled with 1251 activity was measured using a gamma counter. Nonspecific binding, determined by the addition of by modifications of the iodogen method previously a 250-500-fold excess of the respective unlabeled described for both cytokines (11,24). Briefly, the cytokine, was never greater than 20% of the total rhTNF or rhIFN-y (25 pg/lOO p1) and 1 mCi of '"I binding. To determine the receptor number and af(New England Nuclear, Boston, MA, U.S.A.) were finity, cells (5 x 1o5/we1i, 12 well plates) were inadded to a test tube coated with 10-15 pg of iodogen reagent (~4,3,4,6-tetrachloro-3,6-diphenylglycouril) cubated under equilibrium binding conditions (4°C for 4-6 h) with increasing concentrations (1-50 and incubated at 4°C for 10 min. The reaction was ng/ml) of either 1251 rhTNF or 'I rhIFN-y. They stopped by removal of the mixture from the test were analyzed by the method of Scatchard (29) ustube. Free ''%odine was removed by passage over a ing the LIGAND program (30). This analysis was PD- 10 column (Pharmacia) equilibrated with a carried out with the assumption that the molecular phosphate-buffered saline solution (pH 7.4) conmoieties of "'I rhTNF and '"I rhIFN-y that bound taining 0.1% bovine serum (BSA). The peak radioto the cells were trimers (MW = 51 kDa) and active protein fraction was always greater than 95% dimers (MW = 32 kDa), respectively (18,19,26,27). TCA precipitable and was used for all binding exrhTNF and 1251 rhIFN-y internalTo monitor 1251 periments. Recovery of "'I rhTNF was 80-90% as ization, COLO 357 cells (5 x lo5 in 12 well plates) determined by enzyme-linked immunosorbent aswere incubated at 37°C with 200,000 cpm of 12'1 say (ELISA) for TNF as previously described (25). rhTNF or 300,000 cpm of '1 rhIFN-y for either 30 Recovery of 1251rhIFN-y was 55-70% as deteror 120 min and washed as in binding experiments. mined by a commercially available ELISA for Cells were then incubated for 4 min at 4°C with 0.5 IFN-y (Amgen, Thousand Oaks, CA, U.S.A.). The M NaCl that was titrated to pH 2.5 with acetic acid specific activity of both labeled cytokines ranged (3 1). Following elution of membrane-bound (acid from 35,000 to 60,000 cpm/ng for several iodinadissociable) radioactivity, cells were solubilized in tions , 0.5 M NaOH to determine the internalized (acid When iodinated rhTNF and rhIFN-y were anaresistant) radioactivity. lyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis, monomeric moieties of 17 and 16 Statistical analysis kilodaltons (kDa), respectively, were observed Statistical analysis of multiple comparisons was (2,3). However, under nonreducing conditions, a carried out using analysis of variance (ANOVA) dimeric form of rhIFN-y and dimeric and trimeric and Tukey's multiple comparison procedure. forms of rhTNF were observed as previously deWhere appropriate, data were also analyzed by a scribed (18,19,26,27). The '"I rhTNF retained full two-tailed Student's t test. Cellular viability data biological activity as determined by comparing its were analyzed by a dose-effect analysis computer cytotoxic actions on L929 cells with the actions of program that enables determination of antagonistic, unlabeled rhTNF (11). The '251-labeled rhIFN-y readditive, or synergistic effects of multiple drug tained full biological activity as determined by comtreatments (32). paring its effect on the induction of HLA-DR antiPancreas, Vol. 5 , No. 3, 1990
A . B . RAITANO ET AL.
270 RESULTS
Effects of rhTNF and rhIFN-y on cellular viability Four of five human pancreatic tumor cell lines demonstrated at least a 50% decrease in growth after a 48-h incubation with a particular cytokine or
A loo
combination of cytokines using the MTT cell growth assay (Fig. 1). The growth inhibitory effects produced by these cytokines were found to be highly significant when analyzed by ANOVA and Tukey's multiple comparison procedure. The MIA PaCa-2 cells were most sepsitive to rhTNF, exhib-
1
1-
B
loo I_
T 75
g::g
" 5 7
a
0
50-
i o u p
MIA PaCa-2 MF
-A-
0 IFNy
+ 0
100
300
200
500
0
100
CYTOKINE (UNITSIML)
C
100
75
TNF
0
IFNy TNF+IFNy
+
TNFclFNy
400
-A-
200
300
400
500
CYTOKINE (UNITSIML)
1
D
100 1
1 6COLO 357
ASPC-1 -b Q
+
0
100
300
200
TNF IFNy TNF+IFNy
400
TNF IFNy I- TNFtIFNy -&
4
500
0
CYTOKINE (UNITSIML)
100
200
300
400
500
CYTOKINE (UNITSIML)
E l75
o
o
]
v
PANC-1 *MF CF IFNy + TNF+IFNy
0
100
200
300
400
CYTOKINE (UNITSIML)
Pancreas, Vol. 5, No. 3, 1990
500
FIG. 1. Effects of rh tumor necrosis factor (TNF) and rh gamma interferon (IFN-y) on human pancreatic carcinoma cell growth. A: MIA PaCa-2 cells. B T,M, cells. C: ASPC-1 cells. D: COLO 357 cells. E PANC-1 cells. Cells were incubated for 48 h with the indicated concentrations of rhTNF (triangle), rhIFN-y (open square), or rhTNF in the presence of 250 U/ml of rhIFNy (filled square). Cytotoxidcytostatic effects were determined by the 3-4,5-dimethylylthiazol-2-yl-2,5-dipheny~tetrazolium bromide assay as described in Materials and Methods. Data are the means f SEM from four separate experiments in which three to five determinations were made for each experimental condition.
EFFECTS OF TNF AND IFN-y IN PANCREATIC CANCER CELLS
271
iting a 56% decrease in growth at a concentration of 500 U/ml (p < 0.001, Fig. 1A). This effect appeared to be cytotoxic/cytolytic, as it was associated with marked detachment of cells and a marked increase in trypan blue dye uptake. The T3M4 cells were most sensitive to rhIFN-y, demonstrating a 54% decrease in growth at a concentration of 250 U/ml (p < 0.001, Fig. 1B). The majority of T3M4 cells remained attached and excluded trypan blue dye, suggesting that the action of IFN-y on these cells was mainly cytostatic. The effects of rhTNF in MIA PaCa-2 cells or rhIFN-y in T3M4cells were not enhanced by the addition of the reciprocal cytokine. The PANC-1 cells were resistant to the inhibitory effects of either cytokine alone and to their combined activity, exhibiting only an 18% decrease in cell growth (Fig. 1E). In contrast to the other cell lines, ASPC-1 and COLO 357 cells were most sensitive to the combination of rhTNF and rhIFN-y (250 U/ml of each), exhibiting a 56 and a 55% decrease in growth, respectively (p < 0.001, Fig. 1C and D). Unlike the effects produced by single cytokine treatment, in both cell lines concomitant treatment resulted in cytotoxic/cytolytic effects, as evidenced by marked cell detachment and trypan blue dye uptake by the detached cells. Analysis of the cytotoxic effects of rhTNF and rhIFN-y in ASPC-1 and COLO 357 cells with various dilutions of the cytokines in a constant ratio of 10 to 1 of rhTNF to rhIFN-y, 10 to 1 of rhIFN-y to rhTNF, or a 1-to-1 ratio of both cytokines revealed that the effects of the cytokines were synergistic (data not shown), as determined by a dose-effect analysis computer program (32). The most potent growth inhibitory effect of a cytokine($ on each of the five human pancreatic carcinoma cell lines is summarized in Table 1.
Effects of pretreatment and addition of specific neutralizing antisera on cytoxicity Inasmuch as ASPC-1 and COLO 357 cells were most sensitive to the combined actions of rhTNF and rhIFN-y, we next examined the effects of pretreating these cells with one cytokine prior to exposure to the second agent. Pretreatment for 24 h with either rhTNF or rhIFN-y, followed by a 48-h incubation with the reciprocal cytokine, was not as effective as the simultaneous treatment with both cytokines in either ASPC-1 cells (p < 0.005, Fig. 2A) or COLO 357 cells (p < 0.02, Fig. 2B), based on analysis of the data by ANOVA and a two-tailed Student's t test. For both cell lines, treating initially with rhIFN-y tended to produce greater cytotoxicity than treating initially with rhTNF. However, this difference was not statistically significant. The combined cytotoxic effects of rhTNF and rhIFN-y were completely neutralized in the presence of specific antisera to both cytokines (Fig. 3A and B; 0 h). The addition of normal rabbit antiserum did not reduce cytotoxicity (data not shown). When the addition of the rhTNF-specific and the rhIFN-y-specific antisera was delayed by 4 h, complete neutralization was no longer attained (Fig. 3A and B; 4 h). However, the cytotoxic effect was still markedly blocked as compared to cells that were not incubated with antisera (p < 0.001). Delaying the addition of both antisera for 24 h further decreased their neutralizing effects, but the cytotoxic action of the cytokines was still partially blocked (p < 0.05; Fig. 3A and B; 24 h). Analysis of the data by ANOVA and Tukey's multiple comparison procedure indicated that the neutralizing effects of either anti-rhTNF or anti-rhIFN-y serum were not significantly different from each other or from those of their combination.
TABLE 1. Cytostaticlcytotoxic effects of rhTNF and rhIFN-y" in human pancreatic carcinoma cells
Binding of "'I rhTNF and 1'2 rhIFN-y To determine whether cultured human pancreatic carcinoma cells bind either TNF or IFN-y at the same temperature at which they exert their biological actions, binding studies were carried out at 37°C using '251-labeled rhTNF and rhIFN-y. Maximal binding of both '251-labeled cytokines to the pancreatic cancer cell lines at 37°C occurred between 90-120 min (data not shown). The cellassociated '"I radioactivity after a 2-h incubation at 37°C with the respective '251-labeled cytokine is shown in Table 2. The MIA PaCa-2 and T3M4cells exhibited relatively high levels of I2'I rhTNF bind-
Cell line ASPC-1 COLO 357 MIA PaCa-2 PANC-1' T3M4
Most effective treatment
+ +
Inhibitory effectb
rhTNF rhIFN-y rhTNF rhIFN-y rhTNF
Cytotoxic Cytotoxic Cytotoxic
rhIFN-y
Cytostatic
-
-
Tumor necrosis factor, TNF; interferon, IFN. The growth inhibitory effect of the most effective cytokine treatment was determined by microscopic examination. Criteria for cytotoxic effects included the appearance of detached and floating cells and the inability of the cells to exclude trypan blue. 'The PANC-1 cells were relatively resistant to all treatments. a
Pancreas, Vol. 5 , No.3, 1990
A . B . RAITANO ET AL.
272
>
I
I-
TNF + lFNy
PTNF -IFNy
PlFNy -TNF
I
TNF + IFNy
PTNF -1FNy
PlFNy -TNF
FIG. 2. Effects of pretreatment with cytokines. A ASPC-1 cells. B: COLO 357 cells. Cells were preincubated for 24 h in complete medium without cytokine or with 250 U/ml of either rh tumor necrosis factor (TNF) (PTNF) or rh gamma interferon (IFN-y) (PIFN-y). Cells were then washed two times and further incubated with 250 U/ml of rhTNF and rhIFN-y (TNF + IFN-y), 250 U/mlof rhIFN-y (+IFN-y), or 250 U/ml of rhTNF (+TNF), respectively. Cytotoxidcytostatic effects were determined by the 3-4,5-dimethylylthiazol-2-yl-2,5-diphenyltetrazolium bromide assay. Data are the means SEM from three separate experiments, in which three to five determinations were made for each experimental condition.
*
ing, ASPC-1 and COLO 357 cells exhibited intermediate levels of binding, and PANC-1 cells exhibited the least binding. For '"I rhIFN-y binding, ASPC-1 and MIA PaCa-2 cells bound the highest levels, while COLO 357, PANC-1, and T3M4cells
bound relatively similar amounts. The addition of 500 ng/ml of unlabeled epidermal growth factor (EGF) or of the reciprocal cytokine did not alter the specific binding of either labeled cytokine (data not shown). Internalization of "'I rhTNF and 12'1 rhIFN-y by COLO 357 cells Following binding, most peptide hormones are internalized via clathrin-coated pits into "receptosome" structures that may subsequently fuse with lysosomal compartments leading to degradation of the ligand and/or its receptor (33). We therefore sought to determine if pancreatic carcinoma cells internalize rhTNF and rhIFN-y. Accordingly, COLO 357 cells were incubated at 37°C for either 30 or 120 min with either '1 rhTNF or '"I rhIFN-y, and the surface-bound radioactivity was removed by acid wash (31). The remaining acid-resistant radioactivity therefore represented internalized ligand molecules. As can be seen in Table 3, at 30 min of incubation, approximately equal amounts of either '1 rhTNF or "'I rhIFN-y radioactivity were localized to the acid-sensitive (cell surface) and to the acid-resistant (internalized) compartments. In contrast, at 120 min of incubation, most of the radioactivity for both ligands was localized to the acidresistant (internal) compartment. Further, for both ligands, the amount of surface-bound radioactivity was less at 120 than at 30 min of incubation, suggesting there was a loss of cell surface binding sites
75 A
ASPC-1
B
COLO 357 +ANTI-TNF.IFNy +ANTI-TNF +ANTI-IFNy
+ANTI-TNF,IFNy
H +ANTI-TNF +ANTI-IFNy
50
25
0 NOAB OHRS +ANTI-TNF. +ANTI-IFNy
4 HRS
24HRS
NOAB
OHRS
4 HRS
24 H R S
+ ANTI-TNF, +ANTI-IFNy
FIG. 3. Effects of neutralizing polyclonal antisera on cytotoxicity. A ASPC-1 cells. B: COLO 357 cells. Cells were incubated for 48 h with 250 U/ml each of rh tumor necrosis factor (TNF) and rh gamma interferon (IFN-y) in the absence or presence of the indicated antisera (5 pVwell), which were added at 0,4, or 24 h following initiation of cytokine treatment. Cytotoxic/cytostatic effects were determined by the 3-4,5-dimethylylthiazol-2-yl-2,5-diphenyltet~zoliumbromide assay. Data are the means SEM from four separate experiments, in which three to five determinations were made for each experimental condition.
*
Pancreas, Vol. 5 , No.3, 1990
EFFECTS OF TNF AND IFN-y IN PANCREATIC CANCER CELLS TABLE 2. Specific "'I rhTNF and ''I rhIFN-y" binding to pancreatic carcinoma cells '1 rhTNF, cpm ASPC-1 COLO 357 MIA Paca-2 PANC-1 T3M4
1,423 f 25 2,364 f 186 8,157 f 89 582 f 59 3,678 f 150
rhIFN-y, cpm
'I
12,839 f 548 2,189 f 55 5,886 f 226 3,603 f 198 3,060 f 118
Cells, 1 x lo6, were incubated at 37°C for 2 h with 100,000 cpmlml of the indicated "'I-labeled cytokine. Data are the means f SD of triplicate determinations from a representative (of three) experiment. Tumor necrosis factor, TNF; interferon, IFN.
at the later time point due to the internalization of occupied receptors. Characteristics of TNF and IFN-y receptors in human pancreatic carcinoma cells All the cell lines specifically bound '"I-labeled rhTNF and rhIFN-y at 4"C, with maximal binding for both labeled cytokines occurring between 4-6 h, depending on the cell line (data not shown). To determine the number of binding sites per cell and the affinity of each cytokine for its respective receptor, binding studies with increasing concentrations of '251-labeledligands were next carried out at 4°C for either 4 or 6 h, to allow for maximal ligand binding. Scatchard analysis of the equilibrium binding data obtained with 'I rhTNF revealed the presence of a single order of high-affinity binding sites (dissociation constant (kd) = 0.862.0 x lO-"M) for all five pancreatic cancer cell lines, and 49043,610 binding siteskell (Table 1;representative plot, MIA PaCa-2 cells, Fig. 4A). Scatchard analysis of the equilibrhIFN-y fit on rium binding data obtained with 1251 curvilinear plots, suggesting that high- and lowaffinity components existed for the '1 rhIFN-y
273
binding sites. However, the low-affinity component of the curvilinear plot was due to binding at high ligand concentrations of "'I rhIFN-y, and was characterized by the LIGAND program as representing nonspecific binding. This effect was also evident by the relative linearity of the saturation plot for '"I rhIFN-y at high ligand concentrations (inset, Fig. 4B). Similar binding data for IFN-y depicting spurious low-affinity binding sites have been described by other investigators (18,19). The LIGAND analysis of the 1251 rhIFN-y binding data was therefore carried out with the assumption that a single order of binding sites was present in the pancreatic cancer cells. This analysis revealed the presence of high-affinity binding sites for '"I rhIFN-y (kd = 2.04.4 x lo-" M ) in all five cell lines, and 2,0504,280 binding siteskell (Table 4; representative plot, MIA PaCa-2 cells, Fig. 4B). DISCUSSION In this study, we have determined that human pancreatic tumor cell lines demonstrate a variable sensitivity to the effects of recombinant TNF and IFN-y. After a 48-h incubation, three of five cell lines (MIA PaCa-2, T3M4,and ASPC-1) were sensitive to the inhibitory effects of either TNF or IFN-y (greater than 30% growth inhibition). Concomitant treatment with both TNF and IFN-y resulted in synergistic antiproliferative effects in ASPC-1 and COLO 357 cells but not in MIA-PaCa-2 and T,M4 cells. In contrast, PANC-1 cells were relatively insensitive to either single or combined treatment with TNF or IFN-y (less than 20% growth inhibition). These data support the findings of Schmiegel et al., who also demonstrated that human pancreatic cancer cell lines exhibited varied
TABLE 3. Internalization of '''1 rhTNF and "'I rhIFN-y" by COLO 357 cells 30 min, cpm
120 min, cpm
Surface
Internalized
Surface
Internalized
274 ? 50 3,873 -t 666
322 f 12 3,983 f 199
201 2 23 1,363 rfr 157
1,059 f 87 6,707 f 134
~~
TNF~ I F N ~
a Cells, 5 x lo5, were incubated for 10 or 120 min with either 200,000 cpmlml of lZ5I rh tumor necrosis factor (TNF) or 300,000 c p d m l of '"I rh gamma interferon (IFN-y). Cells were washed and then incubated 4 min at 4°C with 0.5 M NaCl that was titrated to pH 2.5 with acetic acid. Following elution of surface-bound (acid-dissociable) radioactivity, cells were solubilized in 0.5 M NaOH to determine the internalized (acid-resistant) radioactivity. Data are the mean ? SD from a representative experiment.
Pancreas, Vol. 5 , No. 3, 1990
2 74
A . B . RAITANO ET AL. 0.084
2
FIG. 4. Scatchard analysis of 1251rh tumor necrosis factor (TNF) and l2’I rh gamma interferon (IFN-y) binding to MIA PaCa-2 cells. A I2’I rhTNF. B: ‘*’I rhIFN-y. Cells were incubated with increasing concentrations of the respective lZ’I-labeled ligand under equilibrium conditions (4°C) as described in Materials and Methods. Data shown are from a representative (of three) experiment, as analyzed by the method of Scatchard using the LIGAND program. Insets are saturation plots of the same data.
0.042
0
.ooo 0.00
2.35
1251
4.69
7.04
rhTNF BOUND (Xi0-l2M)
B
0.071
L
&
0.034
\
I
-0.003
, 0.00
I
1.28
i251
sensitivity to TNF and IFN-y and that a subset of the cell lines was sensitive to the synergistic effects of these cytokines (34). The actions of TNF and IFN-y in ASPC-1 and COLO 357 cells were completely neutralized by polyclonal antisera specific for the respective cytokines, indicating that the observed effects were not due to the release of nonspecific cytotoxins by the cells. Delaying the addition of either antiserum resulted in a progressive increase in the cytotoxic effects of the cytokines (Fig. 3). A short-term exposure (4 h) to both TNF and IFN-y was sufficient to induce an irreversible cytotoxic effect in at least a portion of these cells. However, pretreatment of Pancreas, Yol. 5, No. 3. 1990
I
I
2.55
= I 3.83
rhlFN-y BOUND (X10-l2M)
ASPC-1 and COLO 357 cells for 24 h with either TNF or IFN-y followed by single treatment with the reciprocal cytokine did not significantly enhance growth inhibition by comparison to the effects observed in the absence of pretreatment (Fig. 2). These observations suggest that the synergistic effects of the cytokines during short-term incubations required the continuous presence of both TNF and IFN-7. It is generally accepted that the first step in the initiation of the actions of a particular cytokine is dependent on its binding to a specific cell surface receptor. Specific receptors for TNF and IFN-y have been demonstrated in a variety of normal and
EFFECTS OF TNF AND IFN-7 IN PANCREATIC CANCER CELLS TABLE 4. Binding characteristics of ' ' 9rhTNF and 1251rhIFN-y" in human pancreatic carcinoma cells TNF binding sites Cell line ASPC-1 COLO 357 MIA PaCa-2 PANC-1
T3M4
No./ cell 480 1,080 8,610 1,830 1,230
Kd
(XIO-''
0.92 1.16 0.86 2.00 1.21
M)
IFN-y binding sites No./ cell
(XIO-" M)
6,280 3,850 2,050 2,280 3,470
3.64 6.41 2.60 2.04 5.53
Kd
Binding was carried out with increasing concentrationsof labeled ligand under equilibrium conditions (4°C) as described in Materials and Methods. Data are the means of three separate experiments for each cell line and for each cytokine, as analyzed by the method of Scatchard using the LIGAND program. Tumor necrosis factor, TNF; interferon, IFN.
transformed cells (15-20). Our findings are the first to document the presence of specific high-affinity receptors for both TNF and IFN-y on human pancreatic carcinoma cells. The TNF receptors in these cells had a kd on the order of 0.5-2 X lo-'' M, which is consistent with the values (1-3 x 10- M) reported for various other cells (15-17). The number of TNF receptors per cell for the five cell lines studied ranged from 480 to 8,600. The largest number of TNF receptors (8,600) and the greatest amount of cell-associated "'I rhTNF radioactivity at 37°C was seen in MIA PaCa-2 cells. This cell line was also the most sensitive among the cell lines studied to the growth inhibitory effects of TNF. All the other cell lines had low numbers of TNF receptors (
