Scand. ]. Immunol., Vol, 5, 1976.

Cell-Mediated Cytotoxicity for Melanoma Tumor Cells: Detection by a {'H}Proline Release Assay J. G. SAAL, E. P. RIEBER & G. RIETHMOLLER Department of Experimental Surgery and Immunology, University of Tubingen, Tubingen, BRD

"'

Saal, J. G., Riebcr. E, P. & Rietbmuller, G. Cell-Mediated Cytotoxicity for Melanoma Tumor Cells: Detection by a [^HJProline Release Assay. Scand. J. Immunol. 5, 4'55-^66, 1976. An in vitro lymphocyte-mediated c>'totoxicity assay using [^H]proline-labeled target cells Is described. The assay, modified from an original procedure of Bean et al., assesses the release of t'^HJproline by filtering the total culture fluid containing both trypsinized tumor cells and effector cells. Filtration is performed with a semiautomatic harvesting device using low suction pressure and largediameter glass filters. Pretreatment of filters with whole serum diminishe.s adsorption of cell-free radioactive material considerably and thus increases the sensitivity of the assay. Nearly 100% of the radioactivity could be recovered with this harvesting device. The technique allowed the detection of cytolytic activities of lymphocytes after 6 h of incubation. Lymphoc>'tes from patients with pHmar>' malignant melanoma showed a significantly higher cytolytic reactivity (F < O.OOI) than normal donors' lymphocytes against three different melanoma cell lines. In a series of parallel experiments on 36 patients and 18 normal donors, this modification of the [^H]proline test was compared with three different assays: the conventional microcytotoxicity test of Takasugi and Klein, the original [^H]protine microcytotoxicity test of Bean et al,, and the viabilitj' count of tumor cells. /. G. Saal, M.D., DeparimenS of Experimental Surgery and Immunology, venity of Tubingen, Calwerstrasse 7, D-74 Tubingen, BRD

Cell-mediated immunity (CMI) to human tumors as tested by in vitro q-totoxicity tests has remained a remarkably controversial subject with regard to the specificity and biological significance of the reported findings. The state of affairs was succinctly summarized in a recent review (15) emphasizing the technical difficulties of the tests, the finding of high cytotoxic reactivities of healthy donors' lymphocytes against tumor cells, and the lack of histologic type-specificity encountered with lymphocytes from tumor patients. Most of the reported work on CMI to human tumor cells is based on the microcytotoxicity test (MCT) described by Takasugi & Klein (30). This as-

Uni-

say does not measure direct cytolysis but relies on the counting of adherent tumor cells remaining after cultivation with blood lymphocytes. The loss of adherent tumor cells is generally equated with their damage or destruction by effector cells. It became clear, however, that the number of residual tumor ceils after several days of cultivation results from a combination of detachment, lysis, and proliferation of cells (7). Especially detachment may be caused by nonlytic interactions with lymphocytes, and it also occurs during mitosis of cells (20). Thus, the number of residual adherent cells cannot be taken as an adequate criterion for the cytotoxic effect of lymphocytes. This also holds

456

/. G. Saal, E. P. Rjcber & G. WethmSller

true for those modifications of the MCT in whicli the time-consuming and subjective visual counting of adherent cells is avoided by prelabeling tumor cells with different isotopes such as [3H]thymidine (17), [>25l]iododeoxyuridine (9), [^HJproline (2), 99"'Tc ( I ) , and ««Rb (18). A number of studies, using the MCT in particular, led to the concept of tumor-specific immunity in humans, by demonstrating the presence of lymphocytes in tumor patients which were specifically cytotoxic for tumor cells cf the corresponding histogenic origin (10-13). Tumor-specific reactions of lymphocytes in tumor patients, however, have been questioned during a conference and workshop on cell-mediated iimnune reactions to human tumor-associated antigens (1974)* and more recently by others using the same assay system (6, 16, 23, 31). In view of the multiple factors influencing the results of microcytotoxicity tests it seemed worthwhile to apply an isotope release assay to a human tumor system. Because of the unsuitably high spontaneous release of °Klr and the encountered toxicity of [125[]UDR (21), the nontoxic {^Hjproline was chosen as radioactive label. [3H]proline had already been used by Bean et al. (2) in an MCT for counting adherent tumor cells, and it had been shown that [3H]proline can be incorporated to high specific activity by tumor cells. The spontaneous release of [^H]proline has been reported to be low, and reutilization of this label by lymphocytes or unlabeled tumor cells was found to be minimal in the presence of excess cold proline. In this report it is demonstrated that ['H]proline can be used in a release assay in which the liberated radioactivity is separated from the cell-bound activity by filtration through a semiautomatic harvesting device. The [^HJproline assay was compared with other microcytotoxicity tests and proved to be a simple, sensitive, and quantitative technique * ACS-NCI Workshop on CMC to Bladder Carcinoma, Nov. 11-19. 1974. SIoan-Kettering Institute. New York, N.Y.

for the detection of cytolysis of human tumor cells, although multiplication of tumor cells may still interfere with the assessment of cytolysis. In contrast to microc)'totoxicity tests, this assay coulfl also be applied to target cells growing in suspension. MATERIALS AND METHODS Target cells. Three cell lines derived from human malignant melanoma were used as target cells: Mel-Ku-77, derived from a primary skin lesion, grov/ing In suspension, used as target cells during the 21st and 4ath passage; Mel-Ei-7S, derived from a primary skin lesion, growing as adherent cells, used as targets during the 71st and 106th passage; and Mel-Im82, propagated from a local metastasis, growing adherent, used as targets during the I Ith and 38th passage. All tumor cells used as target cells showed distinct iiielanin production in vitro. Target cells were propagated in RPMI 1640 medium (Gibco), supplemented exclusively with 10% human AB serum, antibiotics, and L-glutamine (TCM-2). Fetal calf serum (FCS) was avoided in all handling of the tumor cells. The human AB serum was selected according to low mitogenic activity. All cultures were routinely checked for mycoplasmal and bacterial contamination. Patient data. The malignant melanoma was diagnosed by the histological criteria given by Clark (8) and classified for type and level of invasion. The cytotoxicity assay was performed after excision of the primary tumor, usually within 2 to 4 weeks after operation, but before the beginning of bacillus CalmetteGuerin (BCG) therapy. The control group consisted of healthy blood donors of about the same age and sex as the patient group. The donors were not selected with regard to immunologie or genetic factors. Effector cells. Nonadherent lymphocytes were obtained from heparinized venous blood sajiiples from patients with malignant melanoma and from normal donors. The isolation of mononuclear cells was performed according to the method of Boyum (4) with several

Release Asxay

distinct modifications. In brief, 30 ml heparinized blood (20 units/ml heparin) were diluted 1:3 in sterile phosphate-buffered saline (PBS), divided into two aliquots and carefully layered on l6 ml sterile Ficoll®-Urovison® in 80-mI centrifugation tubes. The density of this solution was 1.080 g/ml. The blood was spun at 400 g at the interface for 40 min at 20*'C. The upper phase, containing nearly pure mononuclear cells, was removed and washed twice in a large volume of PBS and once in RPMI 1640 medium supplemented with 50% non inactivated FCS (TCM-1). For removal of adherent cells the mononuclear fraction, suspended in a total volume of 5.0 ml TCM-1, was incubated in a 60-mm Petri dish (Falcon) for 1 h at 57°C in a humidified atmosphere of 5% CO2-air. Thereafter, the vessel was vigorously agitated and the supernatants were aspirated. The nonadherent lymphocytes were further purified from remaining adherent cells by nylon-wool filtration in medium containing 20% FCS. The nylon fibers were purchased from Travenol Lab., Thetford, England; 0.8 g of dry nylon fibers were loaded into a 10-ml plastic syringe. Before use, the nylon wool was rinsed with 100 ml of prewarmed saline (37''C) and 20 ml RPMI 1640 medium supplemented with 20% heat-inactivated FCS. The yield of lymphocytes was 26.0% ± 4.4%. For characterization of the nonadherent cell population spontaneous rosette formation was determined according to the method of Perlmann et al. (24). About 75% of the cells formed spontaneous rosettes with sheep erythrocytes. Surface immunoglobulin determinants on lymphocytes were determined by autoradiography using a polyvalent purified rabbit Fab anti-human Ig labeled with ^^BJ Radioiodination of rabbit Fab anti-human Ig and autoradiography was performed as previously described (26). The [^^''I]anti-human Ig stained 20%-25% of the nonadherent cells. After this two-step purification, contamination with monocytes was less than 1% when tested by light microscopy and cytochemistry (peroxidase, NAs-acetate-esterase). For adaptation to the culture conditions the purified lymphocytes were

457

incubated overnight at 37°C in RPMI I64O medium supplemented with 10% human AB serum. Viability of the purified lymphocytes was greater than 95% as judged by the trypan blue exclusion test. In some experiments thymocytes were used as effector cells, Thymuses were obtained from children undergoing cardiac surgery, by courtesy of Drs. Hoffmeister, Stunkat, and Seboldt. Thymus cell suspensions were prepared from fresh organs in TCM-2 by passage through a stainless steel mesh. More than 90% of the cells proved to be viable by trypan blue dye exclusion. \^H']prolirie release test. The tritiated amino acid proline was used for labeling the target cells as originally described by Bean et al. (2). The procedure described here was developed from a modification of the [^Hjproline test as reported earlier (27). From suspension cultures 1-2 X JO^ viable tumor cells were removed and resuspended in 2.5 ml of a proline-free Eagle's medium supplemented with 10% human AB serum and labeled for 24 h with 50 ixCi L-[*^H]proline (specific activity, 5001,000 mCi,'mmol) dt>tained from AmershamBuchler, England. Tumor cells growing as monolayers were labeled with [*''H]proline in the original tissue culture flask at comparable isotope concentrations in proline-free medium. At the end of the incubation period the cells were trypsinized if necessary, washed free of the unincorporated isotope, and resuspended to a cell concentration of 5 X lO'i/ml in TCM2 containing 20 mg of cold L-proIine per liter, Of this suspension 1 X 10^ cells in 200 ^\, corresponding to about 10,000-20,000 cpm, were seeded out per well of a microtiter plate (Falcon Microtest II). The cells could be used for the test for a period of up to 3 days after the [3H]proIine pulse. Lfsually about 10% of the incorporated [^HJproIine was spontaneously released per 24 h. Immediately after three q'cles of rapid freezing and thawing, 70% of the incorporated isotope was released. As autoradiographic studies showed, the [^HJproline was incorporated in ail cells of the cell culture. After 12 h various numbers of nonadherent lymphocytes in 100 ^1 TCM-2 were added to

458

/. G. Saal, E. P. Rieber & G. Riethmiiller

groups of three replicate wells and incubated for \^arious periods of time in a humidified atmosphere of 5% CO2 and 95% air at 37°C. (The effector to target cell ratio used in most of the experiments was 50:1.) At the end of the test period 20 ^1 of EDTA-trypsin (37°C) (7.5% trypsin and 0.6% EDTA in PBS) were added to each well for 3-5 min. Thereafter, the contents of each well were harvested by use of an automatic harvesting machine on Sartorius glass filters (134 0025; Sartorius, Gottingen, Germany) which had been preincubated in either saline or bovine serum. The harvesting device was fabricated from heavy Plexiglas by a design which allowed the simultaneous filtration of the contents from 12 wells through glass filters 25 mm in diameter. The inlets to the filter chambers were conically shaped to allow a uniform distribution of the material on the filters. The material retained was washed with 25 ml of cold physiological saline under low-pressure suction. The material left on the filter disks was then solubilized with Soluene® (Packard) in the counting vlal. The radioactivity was counted in a Nuclear Chicago LS-Counter using the PPO-POPOP-Scintillator. The cytolytic activity (CTL) was calculated relative to medium control and not to normal control effector cells, according to the formula: cpm aample teat % CTL X 100 - cpm medium control XlOO.

The data obtained in both groups, namely cytolysis by lymphocytes from melanoma patients and the qtolysis mediated by healthy donors' lymphocytes, have been analyzed for normal distribution using the KolmogorowSmirnov test. Differences between the groups were analyzed for significance by Student's / test. Instead of counting the residual activity left on the filter disks, in some experiments the activity released from the tumor cells was measured in the sediment after centrifugation of the plates. This was done by centrifugation of the microtiter plates (150 ^^^ 10 min) after trypsin ization with 20 ^1 of EDTA-trypsin solution. The residual cpm per well in the cell

sediment were calculated by subtracting the cpm in the supernatant from the total cpm per well. Cytolysis was calculated according to the formula: cpm in cell sediment "Ml CTL = 100 - of sample test x 100 cpm tn cell sediment of medium control

Microcytotoxicity test. The method was performed as outlined by Takasugi & Klein (30) with some modifications: for comparison with the proline release test, the assay was done in Falcon Microtest II tissue culture plates with I X 10' tumor cells and 5 X 10^"' lymphoq'tes per well. The plates were incubated in 5% CO2 for 48 h and were then washed free of detached target cells and lymphocytes. The remaining cells were fixed in ethanol and stained with Giemsa for counting under an inverted microscope. The percentage of cytotoxicity (CTX) was calculated as follows: CTX = 100 -

mean no. tumor cells in teat wells mean no. tumor cells tn medium control wells

X 100,

microcytoloxicity test. The method used was essentially that described by Bean et al. (2) with the exception that 1 X lO** tumor cells were seeded out per well and 5 X 10^ lymphocytes were added, giving a ratio of lymphocytes to target cells of 50:1. Lymphocyte q'totoxicity was calculated according to the formula: % CTX = 100-

residual cpm in«unple test residual cpm in medium control

Trypan blue dye exclusion test of remainhig tumor cells. As control of the [^Hjproline release test the actual number of surviving tumor cells was determined. To this end, the [^Hjproline release test was performed as described, e>:cept that the number of remaining viable tumor cells in replicate wells was counted visually after staining with trypan blue. For this purpose the content of three replicate wells

l'H]Proline Release Assay

i

lo

lo

loo

3iM

4oo

800

LOG LYMPHOCYTE/TARGET CELL RATIO

Fig. 1. Relationship between lymphocyte concentration and release of radioactivity from [^HJproUnelabeled tumor cells. Various numbers of nonadherent lymphocytes of seven melanoma patients and a constant number (5 X lO^) of [^H]proline-labeled MelEi-78 tumor cells were cultivated for 48 b in Falcon Microtest II plates. As medium control ( ^ ) tumor cells were incubated without addition of lymphocytes. After low-speed centrifugation of the plates the racUoactJvity was determined in the supernatants. Each point represents the mean of triplicate tests from .seven lymphocyte preparations + SD.

was harvested with an Eppendorf pipette and pooled in tubes. A O.259'o trypsin solution was added to the wells for 5 min. Subsec]uently each well was washed twice with RPMI 1640 medium. By visual control no remaining cells could be detected. Thereafter, the tubes were spun at 150 g for 10 min. The supernatants were carefully removed and the pelleted cells were resuspended in 50 /A RPMI l640 medium (10% human AB serum), and 50 ^1 trj'pan blue was added. After mixing, trypan-bluenegative and -positive tumor cells were counted in a Neubauer counting chamber under XlOO magnification. Cytolysis was calculated according to the formula:

459

Fig. 2. Filtration on Sephadex G-200 of the supernatants of ['''Hjproline-labeled melanoma cells cultured witb and without nonadherent lymphocytes. IO* Mel-Ei-78 tumor cells (['HJproIine-labeled) were incubated for 48 h with 5 X UP nonadherent lymphocytes from melanoma patients. The supernatants {I ml) were then gel-filtered on a Sephadex G-200 column (2 X 32 cm). The [^HJproline release test was performed on replicate cultures in parallel. ( # ) ^ supernatant medium controls without lymphocytes after 48 h; ( O ) = supernatants of culture with lymphocytes from Patient F.R. after 48 h of coculture. (% Cytolysis by [•''Hlprcline test: 48 h ^ 38%.)

RESULTS Dose-response relationship between effector lymphocytei and release of radioactivity Various numbers of lymphocytes from melanoma patients were incubated with ^H-labeled tumor ceils. After 48 h the total radioactivit)' in the supernatant of every well was determined. As demonstrated in Fig. 1, the radioactivity in the supernatants increased almost linearly with the logarithm of the number of lymphocytes added. To exclude the possibility that the lymphocytes cause [^JproUne release from tumor cells by other effects than hy cytolytic activity, thymocytes were added instead of lymphocytes. Thymocytes from seven different donors were tested; in no case was a significant cytolytic effect detected after 48 h (% CTL = 2.7 ± 3.6 SD). Gel filtration of released radioactive material For furtber analysis of the lytic interaction between effector cells and pHjproline-labeled

total no. of tumor cells in lest sample - no, of tryiion-bluB-positive tumor cells In test sample CTL, = 100 - total no. of tumor cells in medium control-no. oftrypan-blue-poaitivecpna in medium control'XlOO.

460

/ . G. Saai, E. P. R/eber 6 G. Riethmiiller

Table I. Cytotoxic activity of lymphocytes from melanoma patients assessed by two different filtration procedures Lymphocyte

100

= 0,7

% Cytotoxicity* [^H]proline release assay

1

uonor

SCH GOR SCHN

ZER SAl

MAI FUC Medium control residual cpm

Salinepreincubated glass filters 35 25 IS 25

30 34 19 8905

Serumsaturated glass filters 54 52 26 33 52

46 38 8621

• % Cytotoxicity calculated as indicated in Methods. Effector to target ratio. 50:1. Target cell: Mel-Ei-78. Alt data are means of triplicates.

numb*'of tuinof celli >ICr/w«l

Fig. 3. Recovery of radioactivity by the filtration method. Correlation of the number of seeded tumor cells with the total radioactivity per well, the radioactive material recovered on the filter, and the radioactivity measured in the supernatant. Increasing numbers of [^Hjproline-labelcd Mel-Ei-78 tumor cells were seeded in six replicate wells. The total input of [^HJproline cpm per well was measured simultaneously {stippled bars). The tumor celLs were incubated for 30 h at 37''C and harvested on glass filters after trypsinization, using the automatic harvesting device (filled bars). In parallel, tbe (^Hjproline cpm released into the supernatant were measured in replicate wells (open bars). Supernatants were collected from the wells by aspiration and centrifugc-d in Eppendorf tubes. Results are arithmetic means of the cpm from six replicate welts + SD.

K)

50 Tiypon blue

Fig. 4. Correlation between [^H]proline release test and visual count of remaining tumor cells using trj'pan blue dye exclusion test; ?>6 melanoma patients and 14 normal donors were tested in parallel experiments, r designates tbe correlation coefficient.

target cells it was of interest to know in which form the radioactive material was released into the medium. In a first approach, supernatants from tumor cell cultures with and without lymphocytes were passed through Sephadex G-200 columns (Fig. 2). From cell cultures without added lymphocytes the bulk of radioactivity migrated as low molecular weight material together with the dinitrophenyl-alanine indicator. After cultivation with cytotoxic lymphocytes, ahout 30% of the total released radioactivity was eluted as high molecular weight material appearing in the void volume of the Sephadex column. The amount of this material increased wi*'h time of incubation and was correlated with the degree of cytolysis as measured by the [•''H]proline assay (J. G. Saal, unpublished experiments). The increased radioactivity found in the high molecular fractions therefore indicates cell lysis, presumably due to the interaction with cytotoxic lymphocytes.

l'H'\ProUne Release Assay

461

seen in Fig. 3, independent of the number of filtered tumor cells, the radioactivity retained on the filters and the radioactivity found in the supernatants before filtration added up to nearly 100% of the radioactivity seeded out per well.

48hrs Fig. 5. Cytolytic activity of nonadherent lymphocytes after 6, 24, and 48 h of incubation with Mel-Ei-78 tumor cells measured with the [^H]proline release test (effector to target cell ratio, 50:1). The cells were harvested on Sartorius glass filters preincubated with calf serum. Each point represents the mean lymphocytolysis of seven selected melanoma patients ( • ) and eight healthy donors ( A ) - Vertical bajs indicate standard deviations.

Separation of released and cell-hound radioactivity by filtration Tlie har\'esting of cell cultures from microwells could be facilitated by the use of a semiautomatic harvesting device, which allowed rapid separation of released radioactive material from morphologically intact cells. Since largediameter glass filters (25 mm) were used, lowpressure suction could be applied, allowing gentle handling of the cells with larger volumes of washing fluid. Adsorption of released high molecular weight material to the filters was minimized by preincubating the glass filters in serum. As demonstrated in Table I, distinctly higher cytotoxic values were obtained with serum-saturated filters than with salinewetted filters. Damage of tumor cells during processing in the filtration machine was mainly excluded by determining the recovery of radioactivity- from wells with undamaged tumor cells. As can be

Correlation of visual control of remaining tumor celb and a [^H'jproline assay In control experiments the loss of tumor cells was correlated with the released radioactivity. For this purpose the remaining tumor cells were counted visually by the trypan blue dye exclusion test in a series of experiments on melanoma patients and healthy donors. In parallel, the [^HJproline assay was performed on replicate wells. As demonstrated in Fig. 4, a close correlation (r = 0.7) was found between the cell count and the release determined by the [^H}proiine assay.

Short-term assay of {^H'}proline release The kinetics of the cytotoxic reactivity was studied in parallel experiments on lymphocytes from eight normal donors and seven selected melanoma patients with highly cytotoxic lymphocytes. As can be seen in Fig. 5, already after 6 h of incubation a significant cytotoxic reactivity was detected in the presence of patients' lymphocytes. After from 6 to 48 h of incubation the relative increase of cytotoxic reactivity of normal donors" lymphoq'tes was liigher than that of patients' lymphocytes.

Reproducih/Uty of the [^H']proHne assay The day-to-day variation in the {3H]proline assay was investigated in experiments where lymphocytes from individual donors were tested on several subsequent days. The premise was that the immune reactivity of the donor was unlikely to change during such a short period. As Table II shows, the levels of cytotoxic reactivity of two melanoma patients were fairly constant over at least 4 consecutive days. The standard deviation of the individual tests were ± 3.09rs or ± 2.4% CTL.

462

/. G. Saal. E. P. Rieber & G. RiethrniilUr

Table II, Lymphocyte-induced cytolysis obtained with lymphocytes from individual melanoma patients tested on consecutive davs Donor

KE

Day of study

1 June 2 June

3 June Mean ± SD PF

17 August 18 August 19 August 20 August

Mean ± SD

% Cytolysts* (mean of triplicates) 12.0 15.0 18.3 15.1 ± 3.0 2J.0 18.7 20.0 26.5 22.8 ± 2.4

* Cytolysis was tested at an effector to target cell ratio of 100:1 after 48 h of Incubadon. Target cell: Mel-Ku-77.

Cytotoxic activity of lymphocytes from. melanoma patients and normal donors against three different melanoma cell lines Having established the various conditions of the [•'•HjproUne assay, the q'totoxic reactivit)' of normal donors and melanoma patients was tested against two melanoma cell lines growing as monolayers and one growing in suspension. As demonstrated in Table III, a significantly higher rytotoxic reactivity of melanoma patients' lymphocytes than of normal donors' lymphocytes was found against all three cell lines. The level of cytotoxicity among the three cell

lines varied considerably. With lymphocj'tes of individual normal donors, occasionally high cytotoxic reactivities were found. Comparison of different in vitro cytotoxicity tests At this stage of the experiments involving the [3H]proline assay, it seemed worthwhile to reinvestigate the controversy on tumor-related cytotoxicity in patients, mentioned in the introduction. The question was whether the difference between patients and healthy donors with regard to tumor-cell-directed cytotoxicity, as described above, was only detectable by a test procedure which preferentially measured cytolysis. A comparative study using other microcytotoxicity tests in parallel was therefore undertaken. Jn simultaneous experiments the cytotoxic reactivity of lymphocj'tes from 36 melanoma patients and 18 healthy donors was assessed against one individual melanoma tumor cell line. For comparison, the following test principles were used: the MCT according to Takasugi & Klein (30), the [^H]pro]ine microcytotoxicity test according to Bean et al. (2), and the trypan blue dye exclusion test of the total well contents as described above. Generally, positive correlations among all tests were found. However, as demonstrated in Figs. 4 and 6, different degrees of significance of correlation were seen between particular tests. A high correlation coefficient was noted be-

Table HI. Lymphocyte-mediated cytolysis (CTL) of normal subjects' and melanoma patients' lymphocytes on three different melanoliia cell lines as measured by the [•''H] proline release test Target cell*

Normal subjects - No. of donors

Mel-Ku-77t Mel-Im-82 Mel-Ei-78

11 20 14

CTL(%)

Melanoma patients

mean + SD

No, of donors

CTL(y