Kinetics of Cytotoxicity Induced by Immunotoxins ENHANCEMENT BY LYSOSOMOTROPIC AMINES AND CARBOXYLIC IONOPHORES*

The kinetics of cytotoxicity induced by ricin and a series of immunotoxins consisting of ricin A-chain cou-pled to antibodies against cell-surface antigens has been studied. The inhibition of protein synthesis in cells treated with immunotoxins or ricin occurs after a lag period. The rate of protein synthesis decreases according to a mono-exponential function, indicating a first-order process. With increasing concentration of immunotoxin, a maximal rate of inhibition is reached. The inactivation rate induced by immunotoxins was much slower than that achieved with ricin, even when products were compared on a basis of an identical number of molecules bound per cell, demonstrating the real higher efficacy of ricin. The time required to reduce protein synthesis by 90%, denoted Tlo, was 1.4- 1.6 h with ricin, 60 h with anti-T65 immunotoxin on CEM human T leukemia cells (T65 positive), 65 h with anti-p97 immunotoxin on SK-MEL 28 human mela-noma cells (p97 positive), and 20 h with an IgM anti- Thy 1.2 immunotoxin on WEHI-7 mouse T leukemia cells (Thy 1.2 positive). In this latter case, when the IgM antibody was replaced by an IgG anti-Thy 1.2, a &-fold increase in the inactivation rate was obtained, demonstrating the importance of the binding moiety for the immunotoxins.

The kinetics of cytotoxicity induced by ricin and a series of immunotoxins consisting of ricin A-chain coupled to antibodies against cell-surface antigens has been studied. The inhibition of protein synthesis in cells treated with immunotoxins or ricin occurs after a lag period. The rate of protein synthesis decreases according to a mono-exponential function, indicating a first-order process. With increasing concentration of immunotoxin, a maximal rate of inhibition is reached. The inactivation rate induced by immunotoxins was much slower than that achieved with ricin, even when products were compared on a basis of an identical number of molecules bound per cell, demonstrating the real higher efficacy of ricin. The time required to reduce protein synthesis by 90%, denoted Tlo, was 1.4- Lysosomotropic amines such as ammonium chloride, chloroquine, and methylamine and carboxylic ionophores such as monensin, which are known to interfere with the uptake of certain macromolecules, strongly increased the rate of protein synthesis inhibition by all immunotoxins tested and increased 4-50,000-fold the sensitivity of cells to the immunotoxin. Enhancement in the inactivation rate was as much as 7-10-fold when either of these compounds was added, generating Tlo values comparable to those of ricin.
A new strategy in cancer chemotherapy using hybrid molecules possessing dual functions of recognition and cytotoxicity towards tumor cells is presently under study in numerous laboratories. This mode of treatment was initially envisaged by Ehrlich, who conceived the notion of using agents consisting of a binding component (the haptophore) and a toxic part (the toxophore). In recent years, conjugates (referred to here as immunotoxins), in which the haptophore consists of an antibody molecule or fragment thereof and the toxophore 60 S ribosomal subunit with an extreme potency, in accordance with the widely held view that a single molecule is sufficient to kill a cell (7).
In previous studies, immunotoxins were constructed in which the A-piece of ricin was conjugated to antibodies specific for a variety of target antigens. The results obtained with these reagents demonstrated the expected specific in vitro cytotoxicity since more than 99.9% of target cells were eliminated, whereas cells not expressing the relevant antigens were unaffected (6,8,9). However, despite these encouraging data, immunotoxins are far less toxic than the parent toxin ricin, and in in viuo experiments, complete eradication of tumor cells in animal models has not been possible with immunotoxin treatment (8,IO). In recent studies, we (9) and others (2,11,12) have pointed out that the rate of cell killing induced by immunotoxins is extremely slow, and this may account for the relative ineffectiveness of the immunotoxins in uiuo. However, since the kinetics of cell killing by immunotoxins has so far been inadequately studied, a more detailed examination is warranted.
We show here, from a kinetic analysis of a number of immunotoxins, that the time required to kill target cells can vary within a wide range depending on the model used and that two variables affected the rate of cell killing: the number of immunotoxin molecules bound per cell and the class of the antibody moiety. In addition, the previously described enhancement of target-cell killing in the presence of ammonium chloride (9) is here extended to include a variety of lysosomotropic amines and a group of carboxylic ionophores. These compounds, which improved both activity and specificity of immunotoxins, could have important therapeutic implications.

Kinetics of Cytotoxicity Induced by Immunotoxins and Ricin at a Comparable Amount of A-chain Molecules Bound per
Cell-We demonstrated that immunotoxins exhibited inactivation rates which varied to a large extent @-fold) and which were significantly slower than that of ricin on the same cells (Fig. 2).
Next we examined if the differences between ricin and immunotoxins could be explained by differences of the number of molecules bound per cell. We found, from binding analysis, that at the doses tested in Fig. 1, the number of Achain molecules bound per cell was around 2 X lo6 and 7 x lo6 with ricin on WEHI-7 and CEM cells, respectively, whereas it was 1.5 x lo6 with the anti-Thy 1.2 immunotoxin (IgM) on WEHI-7 cells, 4 x lo4 with the anti-T65 immunotoxin on CEM cells, and 4.5 x lo5 with the anti-p97 immunotoxin on SK-MEL 28 cells. In order to compare the ability of the bound toxin and immunotoxins to intoxicate cells, we performed kinetic experiments a t concentrations ensuring an identical number of A-chain molecules bound per cell. The results are given in Table I. Although the rate of protein synthesis inactivation induced by ricin slightly decreased when the number of ricin molecules bound per cell diminished, ricin achieved 30-fold and 12-fold faster inactivation rates compared to anti-T65 immunotoxin and anti-Thy 1.2 immunotoxin (IgM), respectively, demonstrating the higher efficacy of ricin uersus immunotoxins. However, the differences between the two toxins could vary dramatically with the immunotoxin examined, as illustrated with the anti-Thy 1.2 immunotoxin (IgG). The rate of protein synthesis inactivation induced by anti-Thy 1.2 immunotoxin (IgG) on WEHI-7 cells was only 2.3-fold longer than that obtained with ricin ( Table  I).
In addition, this example demonstrated the importance of the antibody-moiety class for the immunotoxin efficacy since the simple substitution of the anti-Thy 1.2 IgM by an IgG of same specificity improved by a factor of 5 the rate of cell killing measured under comparable conditions.
It should be noticed that the IgG molecules could bind to the WEHI-7 cells to a higher extent than could IgM (Fig. 4). With higher immunotoxin concentrations, the binding could be even greater than 7 x lo5, and this resulted in more rapid kinetics (data not shown). Binding differences could be due to the steric hindrance of the IgM molecule.
Influence of Lysosomotropic Amines and Monensin on the Rate of the Protein Synthesis Inhibition Induced by Immuno-toxins-We have reported that ammonium chloride increased the effectiveness and the rate of action of anti-p97 immunotoxin (9). This prompted us to examine the effect of this potentiator on the other immunotoxins and also to test if other agents known to affect the cellular uptake of macromolecules had the same effect. Lysosomotropic amines and carboxylic ionophores were considered and used at their highest concentration at which they did not decrease the ["C] leucine uptake in the control samples. The data shown in Table I1 demonstrate that the time course of protein synthesis inhibition was considerably shortened when ammonium chloride was added together with immunotoxin and kept throughout the incubation period. The TI, values were achieved at 9.6 h with the anti-P97 immunotoxin, 5.5 h with the anti-T65 immunotoxin, and 2.2 h with the anti-Thy 1.2 immunotoxin (IgM) on their corresponding target cells. Methylamine at 10 mM or chloroquine at 100 p M was even more efficient than 10 mM ammonium chloride and generated T,, values of 2 and 2.6 h, respectively, with anti-T65 immunotoxin. Similar acceleration was obtained with 50 nM monensin ( Table 11). The extent of stimulation depended on the concentration of potentiator, as shown in Fig. 5 with the anti-T65 immunotoxin on CEM cells. With from 2 to 10 mM NH4Cl and from 10 to 20 nM monensin, the TI, values diminished with increasing concentrations for both agents and then reached a saturation level with monensin. Concentrations greater than 10 mM NH4Cl alone abolished protein synthesis, precluding the determintion of the dose of NH4Cl which produces the maximal inactivation. Although the agents could not be compared at the concentrations producing the maximal effect, on a molar basis, monensin was much more effective than ammonium chloride.
Influence of Lysosomotropic Amines and Carboxylic Ionophores on the Cytotoxicity of Immunotoxins: Enhancement of Cytotoxicity-Cytotoxicity was determined quantitatively using cells treated for 24 h with the relevant immunotoxin in the presence of lysosomotropic amines or ionophores. The results are presented in Table 111. A considerable activation of the anti-T65 immunotoxin cytotoxicity on CEM cells was obtained with these drugs. The 50% inhibiting dose (ICB0) was reduced by a factor of 1,180 with 1 mM amantadine, 2,500 with 100 p~ chloroquine, 6,700 with 10 mM NH,Cl, and even 13,300 with 10 mM methylamine. However, when methyl-  amine, dimethylamine, and trimethylamine were compared, the primary amine was a much better potentiator than the secondary amine, which was again better than the tertiary amine.
With the carboxylic ionophore nigericin at 10 nM, an IC50 similar to that observed with 10 mM NH4C1 was measured. More dramatic results were obtained with 50 nM grisorixin (the IC,, was reduced by a factor of 25,000), 1 PM lasalocid (33,000), and with 50 nM monensin (50,000). Ionophores such as nonactin (10 nM), valinomycin (1 nM), and calcimycin (10 nM) produced no activation of anti-T65 immunotoxin, nor did the protease inhibitors leupeptin and pepstatin used at 1 mM.
When 10 mM NH4Cl or 50 nM monensin was added, the IC5, of anti-p97 immunotoxin on SK-MEL 28 cells was decreased by a factor of 42-and 420-fold, respectively. The smallest potentiation effect was seen with the anti-Thy 1.2 immunotoxin (IgM) acting on WEHI-7 cells. In this case, the cytotoxic activity increased only 5.7-fold with NH4Cl and 4.4fold with monensin.
When CEM cells were exposed to NH4Cl for 4 h at 37 "C and then washed prior to anti-T65 immunotoxin treatment, no stimulation was observed, demonstrating that the simultaneous presence of immunotoxin and NH4Cl is required for the effect. With monensin, a similar procedure led to a 20fold reduction in the sensitizing effect; possibly monensin cannot be completely removed by washing.

DISCUSSION
The kinetics of immunotoxin intoxication of target cells was examined by measuring the rate of protein-synthesis inhibition induced by immunotoxins of various specificities and was compared to that obtained with ricin. From this study, it appeared that the intoxication induced by both immunotoxins and ricin was characterized by 1) a lag period preceding the start of protein-synthesis inhibition; 2) a monoexponential decrease in protein-synthesis rate, indicating a first-order process; and 3) a rate of inactivation which is a function of the toxin concentration and which was maximal when all the receptors were occupied (Figs. 1 and 2).
The presence of a lag period for both ricin and the immunotoxins is consistent with the finding that they must first be processed by endocytosis to gain access to the cytosol compartment, as recently demonstrated for ricin by Sandvig and Olsnes (20) and confirmed for immunotoxins (data not shown).
Slower intoxication of cells was obtained when the amount of toxin molecules bound per cell decreased (Fig. l ) , demonstrating that the inactivation rate was a function of the number of toxin molecules bound. However, in the case of ricin, maximal inactivation rate was achieved before all ricin receptors were fully saturated. The fact that ricin was much more efficient than immunotoxins could be explained by the presence of the B-chain which could facilitate the transmembrane passage of the A-chain into the cytosol compartment and/or select a special receptor mediating rapid entry.
The nature of the binding moiety also seems to be important for the activity of immunotoxin. Thus, immunotoxin containing IgM and IgG molecules directed against the same antigen was very different in toxicity (Table I). The reason for the difference between immunotoxin containing the two Ig classes is not known, but it could be due to differences in capacity to induce capping or to the smaller size of IgG which may bring the A-chain into closer contact with the cell membrane than does IgM. This could be much more favorable for the passage of the A-chain.
We have recently observed that the rate of cell killing by immunotoxins of different specificities differed greatly,' suggesting that the target antigen chosen may account for the dissimilarities (comparisons were made on a basis of an identical number of molecules bound per cell). This also could explain in part the variation obtained with the immunotoxins examined here.
Lysosomotropic amines such as ammonium chloride, chloroquine and methylamine, which have been described as inhibiting the action of diphtheria toxin (21), strongly accelerate the inhibition of protein synthesis by all the immunotoxins tested and reduced the Tlo values to 2.2-9.6 h (Table 11).
Similarly, treatment with the carboxylic ionophore monensin, which also inhibits the action of diphtheria toxin (20, 22), reduced the TI, to a similar extent. The effects of NH,Cl and monensin were found to be dose-dependent processes. On a concentration basis, monensin was approximately lo5 times more potent than NH4CI. It should be noticed that the concentration of monensin which generated the maximal effect on immunotoxins was 20-fold lower than that required to protect cells against diphtheria toxin (22). The fact that the activity of ricin and A-chain was affected by monensin but not by NH4C1 (Fig. 6) suggests that these two activators act in different ways, as already proposed by Ray and Wu (23).
The reason for the increase in activity of immunotoxins in the presence of these potentiators could be: (a) an increase in the binding of the immunotoxin to the cell surface, (b) an increase in the efficiency of the transport system, or (c) an increase in the intrinsic toxicity of the A-chain. On the basis of binding studies, we ruled out the possibility that these agents played any direct role in the initial binding step (data not shown). It could also be demonstrated that these agents had no influence on the effect of the A-chain of ricin on ribosomes as evaluated in a cell-free protein-synthesizing system prepared from rat liver (data not shown). Taken together, these facts suggest that NH4Cl and monensin might act on the immunotoxin internalization process. In the presence of these compounds, this process may be extremely potent. Thus, ICso was reached with as little as 10 or 100 anti-T65 immunotoxin molecules bound per CEM cells when monensin or NH4Cl, respectively, was added (calculated from Fig. 6 according to the mass-action law). This efficacy on intact cells approaches the maximal effect at which one molecule is sufficient to kill a cell (24).
Carboxylic ionophores exchange monovalent cations across membranes. Exchange of K' for H' may increase the pH of acidic vesicles like lysosomes (25). An increase in intralyso-soma1 pH is also obtained with NH4C1, which thereby blocks the lysosomal pathway of protein degradation (25, 26). One may speculate that the activation could be due to an inhibition of protein degradation.
Recent studies have shown that the entry of toxin and viruses which takes place through acidified prelysosomal vesicles called endosomes or receptosomes could be blocked by amine and carboxylic ionophores (27). This suggests that such molecules could act also on this compartment.
Carboxylic ionophores and lysosomotropic amines may modify the traffic between different intracellular compartments (28-31). Therefore, it is possible that these agents conduct immunotoxin molecules to a compartment more suitable for passage of the A-chain across the limiting membrane than the compartments reached by the immunotoxin in the absence of the sensitizing agents. In accordance with this possibility, we found that immunotoxins, which under normal conditions are rapidly transferred to secondary lysosomes, are found in new enlarged vesicles, distinct from the previous compartment, when NH4C1 was present. 3 In addition to the interest for these compounds in studies of the mechanism of action of hybrid toxins, these potentiating molecules possess favorable properties for clinical applications. Thus, lysosomotropic amines stimulated the cytotoxicity, as shown by dose-response curves, without modifying that of the A-chain or of a nonspecific immunotoxin. This resulted in higher specificity factors (Table IV and Fig. 6). The specificity increased by factors ranging from 5 to 6000, depending on the model used. Cytotoxicity of immunotoxins was also improved by the carboxylic ionophores grisorixin and lasalocid and, to an even greater extent, by monensin. However, in contrast to the amines, the carboxylic ionophores enhanced A-chain and nonspecific immunotoxin activities. Nevertheless, the resulting specificity factor was improved due to the higher effect on immunotoxins ( Fig. 6 and Table  IV).
Although the in vivo utilization of potentiating drug:; remains to be explored, there is an immediate clinical application for immunotoxins used in combination with one of these compounds to eliminate metastatic tumor cells from bonemarrow samples in patients undergoing autologous transplantation in order to avoid recurrence from the graft. Promising preclinical results have already been obtained, showing that a treatment of a mixture of clonogenic CEM cells and normal human bone marrow with anti-T65 immunotoxin plus NH,Cl led to the reduction by more than 6 orders of magnitude of tumor cells, whereas bone-marrow progenitor cells were unaffected by such a treatment.4  Cells (4 x IO5 cell~lnl for leukemic cells and IO5 cel+s/nl for melanana gellsl i n culture medium were exposed to ricin at 2 x 10-U o r to IT at 10-M for the different periods of tine as indicated on the abaclaea. then they were assayed for 1W-leucxne uptake as described i n Materials and Lthods. m e e different models were examined : T65poaitive CEU cells expoeed to anti-T65 IT ( A I . or to ricin (8). m y 1.2-positive WHI-7 cells exposed to anti-Thy 1.2 IT (I@) (C) o r to ricin I D ) , p97-positive SK-KSL 28 cells exposed to anti-P97 IT (E) or to r i c i n IF). Results are expressed as the percentage of IdCleucine incarparation relative to contmls at each incubation pen& and plotted on the Ordinate on a log scale. Radioactivity in the controls slightly increased during the easey indlcatlng a good viabillty of the cells in the test conditions. Radioactivity incorporated was in the range of Bwo to 15020 cpn depending on the cells. . Blndlng or ITS Lo the target cells.

TABLE IV
Effect of NHqCl and Yonensin on the specificity ractor of 17s %e TI" values were Obtained from h n e f l c curves and calculated a5 descnbed i n Materials and Methods.