The in vitro induction and release of a cell toxin by immune C57B1-6 mouse peritoneal macrophages.

Abstract Peritoneal macrophages from Sarcoma 1 (SAl)-sensitized CS7B1/6 mice release a toxic factor (s), termed macrophage toxic factor (MTF), into the medium when exposed to allogeneic target L cells in vitro . Medium containing MTF is cytotoxic to cultures of allogeneic L cells, syngeneic C57B1/6 fetal fibroblasts and xenogenic HeLa cells. The cell toxin (s) is insensitive to the effects of nuclease, neuraminidase and trypsin, but is partially abrogated by treatment with pronase. Two fractions of toxic activity are eluted from Sephadex G-100, one associated with a 150,000 mol wt marker (IgG), and the other associated with a 47,000 mol wt marker (ovalbumin). Goat antiserum prepared against PHA-induced mouse lymphotoxin (MLT) is capable of neutralizing the toxicity of both MTF and MLT, indicating that the factors may be similar.


INTRODUCTION
Gorer proposed that the macrophage is the primary effector cell in the destruction of allografts of ascites cells, while it plays a lesser role in the rejection of allografts of slowly growing vascularized tissue and leukotic cells (1). Sarcoma 1 (SAl) is an ascites tumor of A/JAX mouse (H-2k/d) or1 m which is consistently rejected 'g' within a period of 10-12 days, upon intraperitoneal transplantation into C57B1/6 (H-2 d) mice (2). Studies by a number of workers have demonstrated that the peritoneal macrophage appears to be the dominant cell involved in the rejection of this tumor (2-4).
While these in viva studies have not revealed the mechanism by which cell destruction occurs, contact between aggressor and target cells is considered to be the first step in the destructive reaction. Actual tumor cell cytolysis apparently occurs by a nonphagocytic mechanism resulting in destruction of both the aggressor and the target cell (2).
In vitro studies have demonstrated that purified peritoneal macrophages obtained from C57B1/6 mice immunized against SAl are capable of destroying cell monolayers containing target A strain antigens (5). As in the previously mentioned in vivo system, cell destruction occurs by a nonphagocytic mechanism which requires cell contact. It has also been demonstrated that the immune macrophage possesses 1  a surface factor, specific for A strain antigens, which promotes contact with the target cell. This factor has the biological properties of an immunoglobulin (5). While aggressor cell metabolism is necessary for the destructive reaction, the mechanism by which killing occurs has not been elaborated (5). The present report suggests that the immune peritoneal macrophage causes in vitro target cell destruction by release of nonspecific cytotoxic factors. Following incubation, the media were collected and cells and debris were removed by sedimentation at 300s in an International Pr-2 refrigerated centrifuge.

MATERIALS AND METHODS
The media were then filter-sterilized and stored at 4°C. Certain batches of pooled media were concentrated 10X in an Amicon ultrafiltration unit using a #3 membrane (30,000 mol wt filter) at 25 PSI pressure, and stored at -18°C.
Greater than 90% of the activity was recovered in the concentrated sample. When pressures of greater than 25 PSI were employed, however, there was a reduction in recovered activity. Control media were obtained from parallel cultures containing normal or starch-induced macrophages substituted in place of the immune macrophages, and/or HeLa or fetal fibroblasts (C57B1/6) in place of L cells. These media were similarly collected and stored at 4°C. Assay System. Cell-free media were tested for toxicity on 1 ml tube cultures of various target cells containing 5 X 10' cells established 24 hr prior to testing as previously described (7). Levels of toxicity were determined by exposing 1.0 ml tube cultures of target cells to various dilutions of test media. These dilutions are expressed as ml of toxic medium per 1.0 ml culture medium. The cells were incubated in the presence of the test medium for 48-72 hr. During this time, the cultures were periodically examined with the light microscope. After incubation, cell viability was assayed by measuring their ability to incorporate 14C amino acids into trichloroacetic acid precipitable protein, as previously described (7). This data is expressed Control] X 100.

Measwing i~~acro~izolcc~ilar Synthesis
Suspensions of starch and SAl induced peritoneal macrophages from C57B1/6 mice were collected and purified by a differential centrifugation technique or adsorption to glass as previously described (5). M acrophage suspensions were adjusted to either 2.5 X 10" or 8 X 10" cells/ml and 1 ml cultures were pulse labeled with . After 20-60 min at 37"C, the cells were sedimented and the nucleic acid or protein extracted and measured for radioactivity as previously described (8). In several experiments, 2 X 10" cells were allowed to establish themselves as monolayers for 24 hr in plaque bottles. The washed monolayers were then labeled and extracted as described above for suspensions.

Chnractrvi2ation
of the mmoplzage toxin Heat stai',ility. The heat stability of the cell toxin was assessed in the following tnamler. Test and control media (3.0 ml aliquots) were placed in individual screwcapped tubes (16 X 125 mm). The tubes were exposed for 15 min to various temperatures in a water bath. After heating, the tubes were plunged into a 4°C ice bath and subsequently tested on indicator L cells for toxicity.
The approximate size of the macrophnge toxin(s) w,as estimated by molecular sieving on a 1.4 X 35 cm G-100 Sephadex l)ed in a siliclad (Clay Adams, Parsippany, NJ) glass column. The column was equilibrated with 0.01 M Tris-HCl, 0.025 M NaCI, 10e5 M EDTA, pH 7.0, and calibrated by applying a 0.7 ml sample containing the following molecular weight markers (Pentex, Kankakee, Ill. j : (a) 1 jovine serum albumin (BSA) , 67,000 daltons, (b) human gamma globulin, 150,000 daltons, and (c) ovalbumin, 47,000 daltons. Then the flow rate was adjusted to 18 ml per hour and 1 ml fractions were collected at 4°C. The absorbancy of each 1.0 ml fraction was measured at 280 nm. Toxic media and control MEM were concentrated 10X in an Amicon (Lexington, MA) ultrafiltration unit using a PM 10 membrane and were then fractionated in a similar manner. Adjacent fractions were then pooled, reconstituted with STP, filter-sterilized and tested for toxicity on L cells. /lntiserzbl;iz firoducfion. Medium containing phytohemagglutinin (PHA) -&mulated Swiss Webster mouse lymphotoxin (MLT) was concentration 10X by =\micon ultrafiltration, then fractionated and concentrated by sequential ammonium sulfate precipitation (9). Toxicity was found in the precipitate which formed between the 50-80~~ salt saturation level. The precipitate was sedimented, resuspended in distilled water and passed through a G-25 Sephadex column or dialyzed against 0.15 M NaCl to remove the ammonium sulfate. This solution contained 15-30 mg/tnl protein. It was then emulsified with an equal volume of I;reund's Complete ;idjuvant. Two Swiss goats were each injected with 0.5 ml of the emulsion sub- KRAMER AiXD GRAiXGER cutaneously on each side and 0.5 ml intramuscularly in the adductor magnus. After 14 days, the goats were reinjected in a similar fashion and bled from the jugular vein 7 days after the second injection.
The blood was allowed to clot at 37°C for 2 hr. Th serum was then collected, cleared of red cells by centrifugation, heatinactivated at 56°C for 30 min, and stored at -1S"C. Normal goat serum (NGS) was collected from the same goats prior to injection. early vaculation similar to that ohservecl with toxic medium, but SO+lOO% of the cells recovered and were not killed. Additional studies were performed to dctermiiie the specificity of the toxic medium. Figure 2 shows the results of one of several such experiments.

RESULTS
It is apparent that the medium was cytotoxic for allogeneic L cells, senogeneic T-Tel,;1 cells and syngeneic C.5713 l/G fetal fihrol)lnsts. In nlany cases we ol)served that lower dilutions stimulated cellular protein synthesis above coiitrol cultures.
A series of extensive experiments verified that medium tos,icity was not due to microbial contamination and could not l)e reversed 1)~ adding fresh serum or essential nutrients. were either overlaid with a ratio of 5 L cells to 1 macrophage or left untreated.
Then the culture vessels were placed at 37°C. After 48 hr of incubation, the medium was collected and cleared of cells and debris. The toxic media was diluted with fresh MEM and tested on L cell monolayer tube cultures. In one experiment HeLa cells were used in place of L cells as the target cell. Figure 3 illustrates results which were characteristic of these etiperiments. Rledium from cultur8es of immune macrophages + L cells was highly cytotoxic, even at a dilution of 1 :5. Control cultures of immune cells alone, however, had a reproducibly low level of activity. Medium from cultures of normal cells + target cells or normal cells alone had a low level of toxicity. The data obtained with starch-induced macrophages paralleled that shown for normal macrophages.
While it is not shown, the amount of medium toxicity detectable when HeLa cells were used as targets was parallel to that found when immune cells were cultured in the absence of target cells.

Macrowtolecular
Synthesis of '~i~~zmune" and "nonimwmne" peritoneal Ynacrophages Experiments were designed in an attempt to compare the relative amounts of DNA, RNPL and protein synthesized by starch-induced and SAl tumor-"activated" peritoneal macrophages. These cells were collected from C57B1/6 mice and pulselabeled with various radioactive precursors as described in Materials and Methods. The results of a typical experiment are visualized in Table 1. These experiments were repeated with the ce!ls from three separate sets of animals and were highly reproducible.
It is clear that levels of nuclei acid synthesis were only slightly elevated in immune cells as compared to starch-induced cells ; however, on a per cell basis, the level of immune macrophage protein synthesis was approximately 1.2 -+ 2 times that of a starch-induced macrophage.

Characterization of the cell ton-in
Heat stability and Enzylize Susceptibility. Samples of toxic and control media were heated at various temperatures for 15 min and subsequently tested for toxicity on L cell monolayers.
As shown in  It was retarded, however, on the G-100 column and eluted in the fraction following the blue dextran marker. Toxic and control media were then fractionated on a 1.4 X 30 cm G-100 Sephadex column. Figure 4A shows the elution profile of three protein markers and the toxicity observed with control preparations. Figure  4B shows the elution profile of test medium.
One peak of toxicity eluted from the column with yG globulin (= 150,000 ml wt) while the second peak eluted between the BSA and OA markers ( = 47,000 ml wt  Figure  5 shows the results of these experiments. The MTF employed in these experiments was toxic at a 1:2 dilution, while the MLT was toxic at a 1 15 dilution under the same conditions. This may account for the apparent difference in the ability of the antiserum to neutralize the MTF more effectively than the MLT. The antiserum obtained from the MLT-immunized goat completely protected the target cells from the cytolytic action of both MLT and the macrophage factor(s) . In contrast, normal serum from the same goat was unable to neutralize the toxic effect. A second control experiment was performed to test the specificity of the antiserum.
The highest level of antiserum employed in the previous experiment was used in an attempt to neutralize the LT secreted by PHA-activated human lymphocytes in vitro. Medium containing HLT was diluted 1 :l with fresh MEM before testing in order to bring the toxic level of this medium to that of the MLT-containing medium. Table 3 shows the outcome of these experiments. The antiserum was unable to neutralize the toxic effects of the HLT at the levels used.

DISCUSSION
It is clear that purified monolayers of peritoneal macrophages obtained from SAl-immunized C57B1/6 mice release toxic substances into the culture medium upon interaction with specific target cells in vitro. Furthermore, these materials act nonspecifically, since they destroy cultures of syngeneic, allogeneic and xenogeneic target cells. Cell destruction is not synchronous, although the majority of the target cells are killed in a period of 2472 hr after initial exposure to the toxic medium. The cytotoxic effect is an active process and not clue to microbial contami-  It became apparent that monolayers of normal and starch-induced C57B1/6 mouse peritoneal macrophages also release low levels of cytolytic substance(s) into the culture media. We found that the levels of cell toxin increase when the macrophage monolayers are incubated with cellular antigens, various soluble antigens or PHA. This phenomenon has been previously described in vitro by Pincus (15). We have no firm explanation for factor release under these conditions. Whether the appearance of the cytotoxin(s) in the culture medium is due to active macrophage protein biosynthesis and secretion subsequent to contact with the target cell, or simply due to the release of toxic intracellular materials into the medium upon cell death is, at present, unknown.
The former situation is suggested by the following three observations. It has been shown that cell extracts from lo8 macrophages/ml are needed to destroy target cell monolayer cultures in vitro (17). Since the cell concentration in these extracts is several orders of magnitude higher than those used in our studies, the medium toxicity we have ohserved would not appear to be due to macrophage destruction. Secondly, immune macrophages have an elevated rate of protein biosynthesis.
Finally, one series of experiments indicates that the cell toxins are present in culture media before any evidence of targetaggressor cell destruction can be observed (prior to 24 hr). Fractionation of the toxic medium on Sephadex G-100 columns reveals that the activity is associated with at least two macromolecules of approximately is insensitive to DNAase, RNAase and neuraminidase, indicating that it is not nucleic acid or protein containing N-ncetyl neuraminic acid. While the activity is not destroyed by treatment with the proteolytic enzyme trypsin, it is partially ablated by pronase digestion. (These experiments have proven difficult to evaluate for the pronase digestion of control medium apparently generates toxic peptides.) Because of the amount of serum present, the interpretation of the results with tryp:sin and neuraminidase may be questionable. The macrophage and lymphocyte cytotoxic factors are apparently antigenically related, since both are inactivated by the antiserum made against the lymphocyte factor. This is a surprising result, for while the two share certain physical properties, i.e., heat stability and enzyme resistance, the macrophage factor(s) differs in molecular weight from the lymphocyte factor (9). These observations suggest that if they are indeed the same, they may under certain conditions, be able to complex with themselves or other serum proteins.
The large cell toxin released by target cell-stimulated immune C57B1/6 mouse peritoneal macrophages is similar in size to the toxin secreted in vitro by PPDstimulated. immune guinea pig alveolar macrophages reported by Heise and Weiser (12). The latter cell toxin is, however, more heat sensitive, which may reflect the fact that they come from different animal species. Several articles by Pincus have described the in vitro release of a low molecular weight cell toxin by both antigenstimulated immune and normal guinea pig macrophages (15,(18)(19)(20). This material has been recently identified as a phospholipid of a molecular weight of less than 1,000 daltons (20). The cytotoxin( s) described in the present report appears to be unrelabed to the factor described by Pincns, since the activity is clearly associated with macromolecules. Additional physical studies are required to either affirm or negate this relationship.
It has been previously demonstrated that immune macrophages from SAl-immunized C57B1/6 mice cause specific nonphagocytic contact destruction of target cells carrying SAl antigens in vitro (5). Yet the present results indicate the immune C57B1/6 mouse peritoneal macrophages are capable of releasing nonspecific cytotoxic substances into the culture medium upon exposure to specific cellular antigens. These data taken together suggest that the specificity of these reactions may lie at the level of recognition. Recognition probably occurs via a previously described specific heat-elutable material (possibly cytophilic antibody) on the surface of the immune macrophage (5). It is probable that the cytophilic antibody (CAB) provides the receptor which permits the macrophage to both recognize and attach to the target cell antigens. Subsequent to cell contact, cell destruction may then occur by a nonspecific mechanism, namely the release of nonspecific cell toxins by the macrophage. There is also a report that mouse macrophages can release a specific toxin into the culture medium upon interaction with target cells in vitro (21). The relationship of this specific toxin to filTF is at present unknown. While destruction does not involve phagocytosis, one might envision a somewhat similar process where the cell releases the cell toxins (lysozomal e*lzymes ?) at the junction oii the niacrol'll"ge-hrget cell contact points, as suggested by the study of Journey and L%mos (22). At present, there is no evidence to support the in vivo role of macrophage cyto-KRnMER AND GRANGER toxic factors in these reactions. Preliminary studies, however, have indicated that the ascites fluid taken from C57B1/6 animals which have just rejected the tumor are nonspecifically cytotosic when tested on cells in tissue culture. Whereas cellfree ascites fluid from tumor bearing RfJAX mice or C57B1/6 mice, prior to tumor rejection, was not toxic. Ascites fluid toxicity has also been reported previously in mice undergoing an immune response to Erlich's Ascites Tumor (23). In this system, the cell-free ascites fluid would cause cell damage to these cells in vitro. It remains to be demonstrated that the factors from the ascites fluid are the same as those released in vitro from the activated macrophage. In addition, recent studies have demonstrated that nonspecific target cell destruction can occur in viva in cutaneous delayed hypersensiti~rity reactions (24,25). The main effector cell in these reactions appears to be a monocytic ceI1. The specificity of these reactions then may be attributed to CAB on the immune macrophage, and the mechanism of cellular destruction may occur by release of a nonspecific cell toxin(s) .