Stand. J. Immunol.. Vol. 6, 1977.

Complement-Dependent Cytotoxicity Against Hepatoma Cells Mediated by IgM Antibodies in Serum from Tumor-Bearms Rats p. LANDO, F. BLOMBERG, M. RAIT^LL, K. BERZINS & P. PERLMANN Department of Immunology, University of Stockholm, Stockholm, Sweden

, P., Blomberg, F., Raftell, M,, Berzins, K. & Perlmann, P. ComplementDependent Cj'totoxicit}' Against Hepatoma Cells Mediated by IgM Antibodies in Serum from Tumor-Bearing Rats. Siwid. J. ImmurwL 6. IO81-]fl92. 1977. Complement-dependent cytotoxic antibodies in syngeneic serum from rat.s carrying transplanted aminoazo dye-induced hepatomas (D23 and D3S) and from rats immunized with irradiated tumor cell or homogtnates were studied by a shortterm ^'Cr release assay. The tumor-bearer sera (TBS) were subjected to chromatography on unsolubilizecl protein A and Sepharose 4B. The cytolytic activity of D23 TBS was recovered in tbe IgM molecular weigbt region, wbereas no such activity was obtained in tbe IgG fraction. As judged from immunodiffusion experiments, the IgM molecular weight fraction did not contain any aggregated or complexed IgG. Moreover, in immunofluorescence tests against viable hepatoma cells, using a specific anti-rat IgG conjugate, tbe TBS were negative. Cross-testing of D23 and D5? TBS against tbe two bepatomas and cross-absorption of tbe sera witb tumor plasma membranes revealed no tumor specificity in these cytotoxicity reactions. Furtbermore, neitber fetal nor early postnatal liver cells could absorb out tbe activity. Absorptions witb adult liver plasma membranes, however, abrogated the cytotoxic activity, and even more effective In tbis respect were bomogenates from kidney and small intestine, thus indicating that the cytotoxic antibodies are directed against 'normal' adult antigen(s) present also in tissues other than liver. P. Lando, Department Stockbolm, Sweden

of Immunology,

The occurrence of tumor-specific antigens on chemically induced rat hepatomas has been established by Baldwin (2). During different phases of tumor growth, these antigens, and antibodies against them, have been found in the blood of tumor-bearing rats (12). In their extensive studies of the immune response in syngeneic rats again.st tumor-specific and tumorassociated antigens, Baldwin et al. (6, U ) used immunofluorescence staining and various modifications of the microcytotoxicity assay. In previous studies we have attempted to characterize the plasma membranes of two different -i-dimethylaniinoazobenzene (DMAB)-induced

University of Stockholm,

Pack, 106 9^

rat hepatomas by means of antisera prepared in rabbits. Enzyme-active membrane antigens present in the tumors and in normal rat hepatocytes were analyzed by means of crossed immunoelectrop ho resis, combined with zyinogram techniques (22). These studies have shown that the antigenic make-up of the tumor membranes is distinctly different from that of normal liver cells and, also, that the two tumors studied differed significantly from each other. However, no information as to the involvement of the various antigens in the rats" own immune response against their tumor was obtained in these investigations. To study this problem.

1082 P. Lando, F. Blomberg, M. Raftell, K. Berzins 6 P. Perlmann

we have adapted an in vitro chromium release as.say in which labeled hepatoma cells are used as target cells for detection and quantitation of cytotoxic rat antibodies. This assay is more accurate and objective than the inimunofluorescence test or the microcytotoxicity assay and is suitable for monitoring antigen purification by quantitative inhibition experiments. In this study we have tested sera from rats bearing syngeneic ttimors, and sera from rats imtnunized against these tumors, for the presence of cytotoxic antibodies. The targets were two DMAB-induced hepatomas (D23 and D33) obtained from Dr. Baldwin's laboratory (L) and also used in our earlier studies. It was found that tumor-bearer sera were particularly c)'totoxic. This paper deals primarily with the characterization of the antibodies.

MATERIALS AND METHODS Tumors. Hepatomas D23 and D33 and inbred Wistar rats were the kind gift of Prof. R. W. Baldwin (Nottingham, England). Tlie tumors were originally induced in the rats by oral administration of 4-dimethylaminoa2Obenzene (DMAB). The primar)' tumors were identified as hepatocellular carcinomas (1). They were propagated by serial subcutaneous transplantation for more than 100 passages in the inbred Wistar strain (D23 in females and D33 in males). The tumors were harvested after 2-3 weeks of growth, freed of necrotic material, and cut into small pieces. Preparation of single-cell suspensions. Tlie solid tumor pieces were digested with 0.25% (wt/vol) trypsin in Tris-buffered Hanks' solution {1 volume of isotonic Tris-buffer (pH 7.4) + 1 volume of Hanks' balanced salt solution) in the presence of 0.1% (wt/vol) DNase, in accordance with the method of Baldwin et al. (4). Trypsin was obtained from bovine pancreas, Type III, twice crystalized, and DNasc I from bovine pancreas, chromatographically prepared (Sigma Chemical Co., St. Louis, Mo., USA). The digestion was performed repeatedly until only connective tissue remained. The released cells were washed in medium, and their viability was determined

by Try pan-blue exclusion. Only preparations with more than 90% viable cells were used. Tumor-bearer serum (TBS). At the time of harvesting the tumors, blood was also drawn by cardiac puncture while the rats were still under ether anesthesia. Serum was prepared and kept frozen (-20°C) until used. Postexcision serum. Ten-day-old tumor transplants were surgically excised, and 10 days thereafter the rats were bled as above. Uat anti-tnmor antisera. Attempts were made to raise antisera by injection of antigens into the Wistar rats. [1] 0.5-1.0 X lO* irradiated (1.5 X 10-1 rad) tumor cells in Tris-Hank were injected three times intramuscularly with 2-week intervals; [2] the same type of cell suspension, but emulsified in half its volume of complete Freunds adjuvant (CFA; Difco Laboratories, Detroit, Mich., USA), was injected as above; [3] ttimor homogenates (6 mg protein solution in phosphate-buffered saline (PBS)) were emulsified in CFA and injected intranuiscularly six times every second week. Ten days after the last injection the animals were bled, and the serum was collected as abo\''e. Inuctivation of sera. Sera to be used for the cytotoxicity experiments were incubated for 1 h at 56°C to inactivate endogenous complement and were cleared by centrifugation for 1 h at 100,000 g. Chromium ('^^Cr)-labeling of hepatoma cells. 5 X 10* trypsinized cells in 0,5 ml RPMI 1640 medium containing 20mM Hepes (Biocult, Paisley, Scotland) and supplemented with 5% (vol/vol) inactivated fetal bovine serum (FBS, Flow Co., Irvine, Scotland) were incubated with 0.1 ml Nas^iCrOj (0.5-1.6 mCi/ ml; 3-20 pug Cr/ml, Radiochemical Centre, Amersham, Bucks, England) for 1 h at 37°C. The cells were then carefully washed three times in the same medium with 5% FBS. Cell viability was determined by Trypan-blue exclusion, and if it was more than 90%, the cells were used in the cytotoxicity assay. Complement-depetident cytotoxicity. The assay was performed in 15-ml conical polystyrene tubes. 1 X 10' chromium-labeled tumor cells in 0.25 ml RPMI-Hepes + 5% FBS were

IgM Cytotoxicity Against Hepatomas

mixed with 0.25 ml of various dilutions of test sera in the same medium. When serum fractions in PBS were used, the FBS concentration in the cell suspension was increased to 10%. 0.25 ml guinea pig serum diluted 1:50 in modified barbital buffer (13) was used as the source of complement. The mixtures were incubated for 1 h at 37°C and the tubes were then centrifuged for 10 min at 200-300 i^. 0.5 ml of the supernatant was taken off. The supernatant and the remaining 0.25 ml containing the pellet were counted in a well-type scintillation counter (Nuclear Chicago Division, Model 1185), The percentage lysis was calculated according to the formula: % lysis = ———^—- X IOO, where ' (Y-I-X)-2B Y = cpm in the supernatant, X = cpm in the pellet part, B = background, and 1.5 is the dilution factor. The results presented are mean values of duplicates. The mean spontaneous release (medium controls") in the experiments was 12.6% ± 3,T>f (SD). The differences between duplicates were 6.0% or less, Heterologous antiserttm against rat IgG was raised in rabbits by repeated intramuscular and intraperitoneal injections of a purified preparation (4-5 mg protein) emulsified in half its volume of CFA, Rat IgG for immunization was prepared by ammonium sulfate precipitation of rat serum, followed by ion-exchange chromatography on DEAE-Sephadex A-25 (Pharmacia Fine Chemicals, Uppsala, Sweden). The antisentm gave no precipitate when reacted with rat IgM in double immunodiffusion. Double immunodiffusion by the method of Ouchterlony (20) was performed on 5 X 5cm glass plates carrying 4 ml of 1% (wt/vol) agarose (Behringwerke, Marburg/Lahn, West Germany) in PBS. The plates were washed in saline, dried under filter paper and stained for protein with Coomassie brilliant blue R as described by Chua & Bennoun (14). Immunofluorescence microscopy was performed on single-cell suspensions of viable tumor cells (3), All incubations were carried out at +4°C. 5 X 10« cells were incubated for 20 min with 0.1 ml rat test serum (TBS, postexcision serum, or immtme serum), washed.

1083

and thereafter incubated with 0,1 ml fluorescein isothiocyanate (FITC)-labeled rabbit antirat IgG, diluted 1:20, After being washed, the cells were mounted on microscope slides in glyceroIjTris-Hanks' (1:1),They were examined in a Leitz Orthoplan Fluorescence Microscope (incident light) equipped with a Bg 12 excitation filter (1.5 mm), a KP 495 interference filter, and a K 510 barrier filter. A fluorescence index (FI) was calculated by the formula: % unstained cells % unstained cells with normal serum - with test serum FI •

,

'/f unstained cells with normal scrum

An FI of 0.30 was taken to represent a positive reaction (5), Affrnity chromatography of TBS on protein A-linked Sepharo.ie CL-4B (16). 0,75 g protein A-Sepharose CL-4B (Pharmacia Fine Chemicals) was swollen in PBS, packed in a column (0.9 X 15 cm), and washed with PBS, 5 ml inactivated TBS was fractionated on the column. Thereafter the column was extensively washed and the bound components eluted with O.IM glycine-HCl buffer (pH 2.8). The eluted fractions were irnmediately neutralized with lM NaOH and were then diaiyzed overnight against PBS, Finally, all fractions were concentrated to the original volume (5 ml) on coUoditmi membranes under negative pressure. They were stored at -2O''C until used. l-urther fractionation on Sepharose 4B of the TBS fraction noi bound to protein A. 0.5 ml of TBS retained on protein A was separated on a Sepharose 4B column (1,6 X 100 cm) (Pharmacia Fine Chemicals) equilibrated in PBS. Before use the column was calibrated with Blue Dextran (Pharmacia Fine C!hemicals), human IgM (^^sj.iabeled (17) cryoglobulin from the serum of a patient with Waldenstrom's macroglobuUnemia), and human IgG (Kabi, Stockholm, Sweden). 2,5-ml fractions were collected, and 0.25-m! samples of each tube were tested for complement-dependent cytotoxicity against hepatoma cells. Absorption experiments. D23 TBS was absorbed with various amounts of spleen cells, fetal and early postnatal liver cells, erythrocyte ghosts, plasma membranes, and various tissue

1084 P. Undo. F. Blomberg. M. Raftell, K. Berzim & P. Perlmann

homogenates from Wistar rats. D23 TBS was also absorbed with spleen cells and liver plasma membranes from the inbred rat strain Dark Agouti (DA), in which the tumors do not grow. D23 TBS and D33 TBS were, in addition, absorbed with tumor plasma membranes from both hepatomas, Spleen cells were prepared mechanically from small pieces of Wistar or DA spleen by pressing them through a stainless steel net while adding RPMI-Hepes (5% FBS) medium. Erythrocytes were removed by centrifugation on a dense layer of Fieoll-Isopaque, density 1.077 g/ml (10). After centrifugation the .spleen cells were washed twice with RPMI-Hepes (5% FBS) to remove the Ficoll-Isopaque. Viability of the spleen cells was determined by Ttypanblue exclusion. Fetal and early postnatal liver cells from Wistar rats were prepared by pressing small tissue pieces through a stainless steel net in RPMI-Hepes (5^^ FBS) medium. The cell suspension was washed twice in medium by centrifugation at 1200 g for 10 min, Viability of the separated liver cells was determined by Trypan-blue exclusion. Erj'throcytes from Wistar blood were prepared by stirring coagulated blood with glass rods. The released erythrocytes were washed four times in PBS. Constant serum dilutions were secjuentially Incubated with different amounts of cells for 1 h at 37''C and then centrifuged for 10 min at 1200 g after each absorption step and, finally, for 1 h at 100,000 g. Ghosts were prepared from Wistar erythrocytes by lysis in distilled water and repeated wa-shing in 0.9% NaCI. Liver plasma membrane fractions were isolated as described by Emmelot et al. (L5). Tumor plasma membrane fractions were prepared as previously described (21). The plasma membranes were washed twice in 0.9% NaCI. Tissue homogenates from testis, heart, skeletal muscle, kidney, lung, and small intestine were prepared from Wistar rats. The homogenates were washed twice in PBS to free the fractions from material soluble in saline. Sera diluted 1:5 were repeatedly absorbed with constant amounts of different tissue ho-

Table 1. Indirect immiinofluoresfence of viable cells Test sera D23 postexcision serat D23 tumor-bearer serat

Fluorescence index* 0.26 0.24

0.51 0.54 0.39 0.07 0.00 0.07

Rat anti-D23 antisera; pools from four rats immunized with: Irradiated D23 cells Irradiated D23 cells + CFA*' D2?> homogenate

0.00 0.54 0.12

* Fluorescence index (FI) greater than 0.30 was considered a positive reaction. t Four individual sera were tested. ** CFA = complete Freund's adjuvant.

niogenates or various amounts of membranes at - F 4 ° C for 1 h. Finally, all absorbed sera were centrifuged for 1 h at 100,000 g and then kept frozen at -20 °C until used.

RESULTS Indirect immunofluorescence on viable D23 cells was carried out to detect IgG antibodies in TBS, postexcision serum, or rat anti-D23 antiserum. Significant staining was obtained with postexcision serum and one type of antiserum that was raised by injection of irradiated D23 cells emulsified in CFA (Table I)Cytotoxicity assays When the various sera were tested for complement-dependent antibody-mediated cytotoxicity against D23 cells, D23 TBS and the im munofluarescence-positive rat anti-D23 anti serum were found to be highly reactive (Fig 1). All the other rat sera, however, wert negative. The strong cytotoxic activity of all TBS and the negative results obtained with these sera in the immunofluorescence tests suggested a possible involvement of different antibodies in the two types of reactions. This assumption was further tested by fractionation experiments.

IgM Cytotoxicity Against Hepatomas 1085 % Lysis

% Lvsis

40.

35-

3 5-

30-

2 5-

D23 TBS

a-D23irf,*FCA

* a - 0 23 Horn

ililutions

-setum dilutions

Fig. 1. Complement-dependent cytotoxicity of various rat sera against D23 cells. The sera tested were D25 tumor-bearer serum (D23 TBS), D25 postexcision serum (D23 PE), antiserum against irradiated D23 cells (a-D2i^ irr.), antiserum against the same ceils but emulsified in complete Freund s adjuvant (a-D2^ irr. + CFA), and antiserum against D23 homogenate (a-D23 Horn).

Vractionation of D23 TBS Three individual TBS were subjected to chromatography on protein A linked to Sepharose 4B. This protein from Staphylococcus aurens Is known to bind to the Fc region of most mammalian IgG (18). When normal Wistar serum and D23 TBS were subjected to chromatography on protein A-Sepharose and the fractions obtained were analyzed in single radial immunodiffusion by the method of Mancini et al. (19), 25%-3O% of the IgG was found to bind to protein A, whereas the passing fraction contained both IgG and IgM (unpublished results). This corresponds well with the results obtained by Steele et al. (23), who found that 20%-30^ of total rat igG botitid to $. aureus, strain Cowan I. After sepa-

ration of D23 TBS the bound and unbound fractions were separately tested for cytotoxic reactivity against D23 cells (l'ig. 2). All cytotoxicity was recovered iti the material that did not bind to protein A. In double immunodiffusion experiments the fractions reactive and nonreactive with protein A both gave a precipitate with anti-rat IgG antiserum. Thus, IgG was present in botb fractions, but the IgG fraction that reacted with protein A had no cytotoxic reactivity. Moreover, no anti-tumor reactivity could be detected when the two IgG-containing TBS fractions were tested in indirect immunofluorescence against viable tumor cells. Subsequently, the cytotoxic TBS fraction was separated on a Sepharose 4B column to deter-

1086 P. Undo, P. Blomberg, M. Raftell, K. Berzins & P. Perlmann

Fig. 2. Ounplement-depcndent cytotoxicity against D2^ celis of pooled D23 tumor-bearer serum (D23 TBS) fractionated on protein A—Sepharose.

mine in what molecular weight range the active molecules could be recovered. The column was calibrated with human I^M and JgG (Fig. 3). Samples from each column fraction were tested for complement-dependent cytotoxicity. Significant reactivity was found only in high molecular weight fractions. Their elution volume corresponded well with the elution volume of human IgM. No significant c>'totoxic reactivities were seen either with the material eluted in the void volume or in the major protein peak eluted at a volume corresponding to human JgG. Furthermore, fractions 16, 28, 31, and 39 (Fig. 3) were tested in double immunodiffusion against rabbit anti-rat JgG and rabbit anti-rat IgM antisexa. The anti-rat IgG antiserum reacted only with fraction 39, which corresponded to the JgG peak of the TBS. whereas JgM was detected only in the cytotoxic fractions. From these experiments it was concJuded that the cytotoxic reactions in the JgM molecular weight range were due to JgM antibodies and not aggregated or complexed IgG.

A l a o nm

FracliOIIS

Fig. 3. Sepharose 4B fractionation of pooled D23 tumor-bearer serum (D23 TBS) not binding to protein A. The continuous line shows absorbance at 280 nm. The bars represent percentage lysis in complement-dependent cytotoxicity against D23 cells. The column was calibrated widi human (H) IgG and IgM as indicatedDots on the abscissa indicate fractions tested by double immuoodiffusion.

IgM Cytotoxicity Against Hepatomas 1087

033 TBS

D 23 TBS

dilution

% Lysis

3-10'

% Lysis

—,1 3.10-'

, a-w*

D 33 TBS

tar ral serum

3 I0-'

3 10 '

3 W •'

3 10

sefum dilutions

^ serum dilution

tat sflrutn

310'-*

3 10'

-»• serum dilution;

Fig. j . Cytotoxicity of three individual D23 and D33 tumor-bearer sera (TBS) against D23 cells (4a) or D33 cells {-Ib).

Specificity of the cytotoxic activity in tumor-hearer sera To investigate whether the complementdependent cytotoxicity in TBS from a rat carrying one type of hepatoma was specific for

that particular type of tumor, D23 and D33 XBS were tested cross-wise against both tumors (Figs. 4a and b). The sera were found to react about equally well with both hepatomas. This speaks against any major involvement of tumor-

1088 P. Lando, F. Blomberg, M. Raftell, K. Berzins & P. Perlmann - medium control lysis

Fig. 5. Cytotoxicity against D23 cells of D23 tumoi bearer strum diluted 3-10"' after absorption witli fetal liver cells (5a; percentage lysis in medium control (15.4% ± 2.8%) is subtracted from percentage lysis of the test sera) and Wistar and DA spleen cell.s (5b; percentage lysis in medium control (12.5% + 1%) is subtracted from percentage iysis in the test sera). The height of each bar represents the mean of test duplicates (rangc-s indicated by vertical lines).

IU°

9 10*

10 10

Fvtel |ivei calls

% 'ysis -n^dtum 'Ontral lysis

4 10°

Specific antigens in these reactions, D23 TBS was more cytotoxic than D33 TBS against both tumors. This probably reflects a difference in immunogenicity between the two hepatomas, possibly depending on differences between the tumors in antigen expression. The question of whether cytotoxicity was due to etnbryonic antigens or 'normal' liver antigens (for example, transplantation or organspecific antigens) also present on hepatocytes in adult animals was examined in a series of absorption experiments. First, D23 TBS was absorbed with 15-day-old fetal liver cells (Fig. 5a) and spleen cells from Wistar or DA rats

10 IO"

(Fig. 5b). None of these preparations significantly abrogated the cytotoxicity. Furthermore, neither erythrocytes nor erythrocyte ghosts or early postnatal liver cells (these data are not shown) contained antigens that could absorb out the reactive antibodies. Second, the samt D23 TBS was also absorbed with tumor plasma membranes from hepatoma D23 or D33 and with liver plasma membranes from Wistar or DA rats, respectively. In addition, D33 TBS was also absorbed with the two hepatoma plasma membrane fractions (Figs. 6 and 7)AU plasma membrane fractions contained antigens that abrogated the cytotoxic activity of

IgM Cytotoxieity Against Hepatomas

1089

X lysis-medium conuol lysis

023 TBS dilute* StfT*

3-10'

Fig. 6. Cytotoxicity against D23 cells of tumor-bearei sera (TBS) after absorption with hepatoma plasma membranes (D23P and D33P, respectively). Percentage lysis in medium control (8.6% ± 0.6%) is subtracted from percentage lysis in the test sera. The height of each bar represents the mean of test duplicates (ranges indicated by vertical lines).

all TBS tested. The tumor membranes were most effective, as determined on a membrane protein basis (Fig. 6). However, significant

Fig. 8. Cytotoxicity against D23 cells of D23 tumorbearer serum diluted 3 • ' U"^ after absorption with 2 X 1.6 mg protein/ml of various Wistar tissue homogenates. Percentage lysis in medium control (15.-1% + 2.8%) is subtracted from percentage lysis in the test sera. The height of each bar represents the mean of test duplicates (ranges indicated hy vertical lines).

iim controt lysis

Fig. 7. Cytotoxicity against D23 cells of D23 tumorbearer serum diluted 3- 10"* after absorption with Wistar liver membranes. Percentage lysis in medium control (15.4% + 2.8%) is subtracted from percentage lysis in the test sera. The height of each bar represents the mean of test duplicates (ranges indicated by vertical lines).

absorption, although comparatively less, was also obtained with liver plasma membranes. Fig. 7 shows typical results obtained with plasma membranes from Wistar hepatocytes. Similar results were obtained with DA membranes (data not shown). Since D23 and D33 hepatoma plasma membranes, when used for cross-absorption, were equally efficient in abrogating cytotoxicity of both D23 TBS and and D33 TBS against D23 cells, the reactions lacked any obvious tumor specificity. Absorption experiments were also undertaken to ascertain whether the cytotoxic IgM antibodies reacted with antigens specific for liver. Various tissue homogenates (see Materials and Methods) were prepared and used for absorption of D23 TBS. As can be seen In Fig. 8, kidney and small intestine completely abrogated the c}'totoxicitj. Homogenates of heart, skeletal

1090 P. Undo, F. Blomberg, M. Raftell, K. Berzins & P. Perlmann

muscle, and lung also had some effect. On the other hand, homogenates of testis did not absorb any activit)-. To rule out any anti-complementary effect of the tissue homogenates, absorption of a rabbit anti-chicken erythrocyte antiserum was undertaken. No abrogation of the complement-dependent cytotoxicit)' of the antiserum was found when tested against chromium-labeled chicken erythrocytes. DISCUSSION To establish the occurrence of anti-hepatoma antibodies in tumor-bearer sera (TBS), postexcision serum, and rat anti-D23 antiserum from syngeneic rats, we first tested these sera by immunofluorescence. The postexcision sera and one rat anti-D23 antiserum (raised against irradiated cells in CFA) gave significant positive staining with a conjugate that was specific for IgG and did not react with IgM. These results are in agreement with the findings of Baldwin et al. (8), who also found IgG antibodies in postexcision sera. TBS taken from rats with fully developed hepatomas were all negative in our tests. For further investigations of the antibodies in various sera we used a more objective method than the immunofluorescence technitjue: the complement-dependent '''Cr release assay. In this test the immune serum mentioned above was again clearly positive, The reactivities of postexcision sera were less significant in the cytotoxicity assay than in the immunofluorescence test. This discrepancy probably reflects differences between the sera in antibody specificity or in class and subclass distribution, or all of these. However, in contrast to the results of the immunofluorescence studies, all TBS tested for cytotoxicity were highly reactive. Since the immunofluorescence conjugate was IgG-specific, this discrepancy suggested involvement of an antibody class other than IgG in the cytotoxic reactions. Therefore, attempts were made to further characterize these antibodies. When TBS was subjected to affinity chromatography on protein A linked to Sepharose 4B, both the bound and unbound frartions were demonstrated to contain IgG.

However, the bound fraction was not cytotoxic. Furthermore, by additional fractionation on Sepharose 4B, it was shown that the material in the active fractions had a molecular weight comparable to that of IgM and that the IgG fractions had no cytotoxic effect. As judged from immunodiffusion experiments, using rabbit anti-rat IgG and anti-rat IgM antisera, the cytotoxic fractions did contain IgM but no aggregated or complexed igG. The cytotoxic IgM antibodies found by us were not tumor-specific, since D23 TBS and D33 TBS were equally reactive against botii tumors and cyt(Hoxicity was abolished by crossabsorptions with tumor membranes. It should be emphasized that the two tumors have been shown to carry distinct tumor-specific antigens (2), but no test to establish this has been performed in our laboratory. However, using rabbit antisera and enzyme-active membrane antigens as markers, we have also shown that these tumors have plasma membrane characteristics that differ in several respects (9, 22). Thus, for example, D23 plasma membranes contain a tumor-associated arylamidase-active antigen that is not present in plasma membranes from liver or hepatoma D33. The absorption studies ind icated that the cytotoxic IgM antibodies were not directed against fetal antigens occurring on these tumor cells (6) but rather against antigens also present on normal liver plasma membranes of both syngeneic Wistar rats and allogeneic DA rats. These normal antigens did not seem to be histocompatibility antigens, since plasma membranes from both syngeneic and allogeneic rat liver absorbed out the antibody activity, whereas absorption with syngeneic or allogeneic spleen cells, known to carry large amounts of transplantation antigens, had no effect. Moreover, the absorption experiments with various tissue homogenates indicated that the cytotoxic antibodies were not directed against hver-specific antigen (s). Rather, they were reactive with antigens also present in other adult tissues, in particular small intestine and kidney. Bowen et al, (II) established by means of indirect immunofluorescence, combined with absorption studies, that tumor bearers of chemi-

IgM Cytotoxieity Against Hepatomas

cally induced hepatomas (such as D23 and D33) in their sera have IgG antibodies directed against antigens distinct for each tumor (TSTA). tJsing the same technique, we were unable to detect any hepatoma-reactive IgG antibodies in TBS. The absence of tumorspecific reactions in our cytotoxicity experiments may be due to the fact that the hepatomas grown in our rats were transplanted for more than LOO passages. In contrast, Baldwin et al. (7), when finding tumor-specific responses, used their tumors after about 25 passages. During repeated transplantation the tumors might lose TSTA because of immunoselection or modulation. This could also be the reason for the faster growth of the tumors observed in our experiments. Bowen et al. (12) harvested tlie hepatomas I9-2K days after transplantation and found free anti-tumor IgG antibodies in D23 TBS. In contrast, sera from our tumor bearers had to be collected as early as after 14 to 21 days, since the tumor burden already •It that stage was hazardous for the rats. Thus, it is possible that the faster-growing hepatomas express less TSTA which can induce a tumorspecific immune response of IgG type or that the tumor outgrowth is too rapid to permit any detectable TgG re.sponse, or that a combination of these occurs.

ACKNOWLEDGEMENTS We want to thank Margaretha Phil, Ingegiird Andersson, and Peter Rosenlind for excellent technical assistance. This work was supported by grant 642-B76-04X from the Swedish Cancer Society. REFERENCES 1. Baldwin, R. W. Modification of cell antigens during aminoazo dye carcinogenesis in rat liver. Brit. J. Cancer 18, 285, 1961. 2. Baldwin, R. W. Role of immunosurveillance against chemically induced rat tumors. Transplant. Ret: 28. 62, 1976. ^. Baldwin, R. W. & Barker, C. R. Demonstration of tumour-specific humoral antibody against aminoazo dye-induced rat hepatomata. Brit. J, Cancer 21. 793, 19