Znt. J. Cancer: 51,772-779 (1992) 0 1992 Wiley-Liss, Inc.

i

Publicationof the InternationalUnion Against Cancer Publicationde I'Union InternationaleContre le Cancer

A COMBINATION OF TWO IMMUNOTOXINS EXERTS SYNERGISTIC CYTOTOXIC ACTIVITY AGAINST HUMAN BREAST-CANCER CELL LINES Jennie R. CREWS~, Lisa A. MAIER',Yin Hua Yul, Susan HESTER~, Kathy O'BRIANTI,David S. LESLIE~, Karen DESOMBRE~, Stephen L. GEORGE3, Cinda M. BOYER',Yair ARGON^ and Robert C. BAST,J R . ' , ~ , ~ Departments of 'Medicine, 2Microbiology-Immunologyand 3Community and Family Medicine; and the Duke Comprehensive Cancer Center, Duke University Medical Center, Durham, NC 27710. USA. In previous studies, combinations of immunotoxins reactive with different cell-surface antigens have exerted additive cytotoxicity against tumor cells in culture. In this report we describe a combination of 2 immunotoxins that produce synergistic cytotoxic activity. Recombinantly derived rich A chain (RTA) was conjugated with murine monoclonal antibodies (MAbs) 3 1765, 260F9, 454Al2 and 74 I F8 that bound to cell-surface determinants of 42,55, I80 (transferrin receptor) and I85 kDa (HER-2lneu) expressed by the SKBr3 human breast-cancer cell line. When inhibition of clonogenic growth was measured in a limiting dilution assay, the combination of 26OF9-RTA and 454A 12-RTA produced synergistic cytotoxic activity against SKBr3 and 2 other breast-cancer cell lines. All other combinations produced only additive inhibition of clonogenic growth. Simultaneous binding of 260F9 and 454A12 was not supraadditive, but sub-populations of cells which lacked one or the other antigen could be detected. Kinetic studies of internalization, using antibodies conjugated with gold particles, indicated that 454A12 remained within peripheral endosomes for a longer interval in the presence of 260F9. This change in the traffic of the transferrin receptor may contribute to synergy between 26OF9-RTA and 454A 12-RTA.

o 1992 Wiley-Liss,Inc. Immunotoxin conjugates prepared from MAbs and the A chain of ricin have exerted anti-tumor activity in pre-clinical models (Bjorn et al., 1986; Bast et al., 1983; Uckun et al., 1985; Bregni et al., 1986; Pirker et aL, 1985; Blythman et al., 1981; Casellas et al., 1985). The cytotoxicity produced by immunotoxins depends on binding of the immunotoxin to antigens on the tumor cell surface, internalization of the conjugate, and translocation of the toxic moiety into the cytoplasm, where the ricin A chain catalytically inactivates the 60s ribosomal subunit, thereby inhibiting protein synthesis (Olsnes and Sandvig 1988; Endo, 1988). The initial step of immunotoxin binding depends on the presence and the density of antigenic determinants expressed on the tumor cell surface. Within a given tumor, antigen expression is heterogenous (Boyer et al., 1989). Consequently, use of multiple immunotoxins directed against different antigenic determinants might prove more effective in eliminating tumor cells than any single reagent. In addition, multiple immunotoxins could prove more effective than individual immunotoxins if internalization and/or intracellular routing of the different conjugates occurred by complementary mechanisms. Bjorn et al. (1985) have described a number of immunotoxins that have been produced using the A chain of r i c h These immunotoxins are targeted against breast-cancer cell lines and have limited reactivity with normal tissues. Three of these immunotoxins, 260F9-RTA, 317G5-RTA and 741F8-RTA, and an additional immunotoxin, 454A12-RTA, were chosen for this study. The immunotoxins 260F9-RTA, 317G5-RTA and 741F8-RTA have previously been shown to bind to SKBr3 human breast-cancer cells and to inhibit growth of this cell line (Yu et al., 1990). Additive cytotoxic activity has been demonstrated when 260F9-RTA and 317G5-RTA were used in combination (Yu et al., 1990). In the present report we have

extended these observations to describe a combination of immunotoxins that exert synergistic cytotoxic activity. MATERIAL AND METHODS

Chemicals For labeling antibodies, fluorescein isothiocyanate (FITC) was purchased from Sigma (St. Louis, MO) and goat antimouse R-phycoerythrin (RPE) conjugate was purchased from Tago (Burlingame, CA). Gold sols ( 5 and 15 nm) were obtained from Amersham (Arlington Heights, IL). Antibodies and immunotoxins The murine MAbs 31765, 260F9, 454A12, and 741F8 are IgGl immunoglobulins that react with antigenic determinants of 42, 55, 180 (transferrin receptor) and 185 kDa (HER-2/ neu), respectively (Frankel et al., 1985). MOPC21 is a murine IgGl that does not react with human breast-cancer cells. Each of these reagents was obtained from Cetus (Emeryville, CA). To construct immunotoxins, the individual antibodies were conjugated to recombinantly derived ricin A chain (RTA), as described by Yu et al. (1990). Cell lines The human breast-cancer cell lines, BT483, SKBr3 and BT20 used in the study were maintained in RPMI-1640 medium supplemented with 5,15 or 20% FBS respectively and 0.03% L-glutamine. The CAMAl cell line was maintained in Dulbecco's modified Eagle's medium supplemented with 15% FBS and 0.3% L-glutamine. All tissue-culture materials were obtained from Hazelton Biologics (Kansas City, MO). Cells were cultured at 37°C in an atmosphere of 5% COz and 95% humidified air. For experiments, cells were detached in 0.25% trypsin-0.02% EDTA. Conjugation of antibodies with fluorescein isothiocyanate (FITC) Antibody and FITC were conjugated according to the method described by Feteanu (1978). Fluorescein/protein ratios were calculated with a spectrophometer.

F / P ratios of the conjugated antibodies ranged from 2.5 to 4.1.

Clonogenic assay A limiting dilution assay was used to determine the effects of immunotoxins on clonogenic SKBr3 human breast-cancer cells. SKBr3 cells (1 x lo6) were incubated with varying dilutions of immunotoxins for 3 hr at 37°C on a rocking platform in a humidified atmosphere containing 5% C 0 2 and 95% humidified air. At the end of the incubation period, the

4To whom requests for reprints should be addressed at Box 3843, Duke University Medical Center, Durham, NC 27710, USA. Received: December 20,1991 and in revised form March 13,1992.

773

SYNERGISTIC CYTOTOXICITY OF IMMUNOTOXINS

SKBr3 cells were washed twice with medium and were serially diluted 5-fold. An aliquot of 100 pl of each dilution was plated into 6 wells of a 96-well flat-bottom microtiter plate that contained 100 p1 medium per well, and the cells were incubated for 14 days. The clonogenic growth of surviving tumor cells was evaluated by phase-contrast microscopy, scoring the number of wells with at least 1 colony containing 30 or more cells. An estimate of the number of surviving clonogenic units was calculated using a modification of the method of Spearman and Karber (Johnson and Brown, 1961). Controls for each assay consisted of SKBr3 cells treated with medium alone.

Protein synthesis inhibition assay Different concentrations of immunotoxin were incubated with 50,000 breast-cancer cells per well in 96-well flat-bottom tissue-culture microtiter plates at 37°C in an atmosphere of 5% C02 and 95% humidified air. One pCi of 35S-methionine in methionine-free medium was added during the last 6 hr of incubation. Wells were washed, cells were harvested, and incorporation of 35S-methionine was determined by liquid scintillation counting. Isobolographic analysis Analysis was performed according to the method of Steel and Peckham (1979). Isobolographic analysis examines the dose of individual agents required to produce a given antitumor effect, e.g., 99% (2 logs) of clonogenic tumor-cell elimination. For agents with linear dose-response curves, a concave isobole indicates synergy. For agents with non-linear dose-response curves, such as the immunotoxins used in this study, a concave isobole may represent synergy or additivity. Consequently, an “envelope of additivity”, as described by Steel and Peckham (1979), was used to determine the interaction between immunotoxins with non-linear dose-response curves (Fig. 1).The envelope consisted of 2 components, mode I and mode 11, which represented the theoretical limits of the additive effects of the 2 immunotoxins being considered. Mode I was calculated from the dose-response curves of each agent, on the assumption that the 2 agents operated independently. Mode TI was calculated in a similar manner, but assumed that the 2 agents interacted. When the envelope was drawn, immunotoxin combinations were considered synergistic if the isobole fell below the envelope created by mode I and mode I1 (Fig. 1). Combinations were considered additive when the isobole fell within the envelope (Fig. 1). Imrnunofluorescenceassays To determine the relative number of binding sites of each antibody, SKBr3 cells (5 x lo5)were incubated with saturating concentrations of antibody-FITC conjugates (50 pl of 1 5 0 dilution with protein concentrations ranging from 0.4 to 0.8 pg/ml) for 30 min on ice and washed twice with PBS containing 1% FBS and 0.02% sodium azide (FACS buffer). After incubation, the cells were fixed with 2% formaldehyde. Cells were analyzed for mean linear fluorescent intensity and compared with the mean fluorescent intensities of standard beads with known numbers of FITC molecules per bead. Antigen-binding sites were calculated by dividing the number of FITC molecules bound to the cells by the F / P ratio for the individual FITC-conjugated antibodies. In experiments studying additive fluorescence, SKBr3 human breast-cancer cells (5 x lo5) were incubated with 25 pl of the individual antibodies at 10 pg/ml plus 25 p1FACS buffer or with 25 pl of each of 2 antibodies, at a concentration of 10 pg/rnl. Incubation was performed on ice for 30 min. The cells were then washed twice in FACS buffer and 50 pl of 1 5 0 goat anti-mouse FITC was added to each tube. After 30 min incubation on ice, the cells were washed twice in FACS buffer and fixed in 2% formaldehyde. For 2-color immunofluorescence studies, SKBr3 cells (5 x lo5)were incubated with one MAb in microfuge tubes on

B(ug/ml)

FIGURE1 -The envelope of additivity. Example of isobolograms showing mode-I and mode-I1 isoboles for 2 agents, A and B. The shaded area between mode I and mode I1 represents the envelope of additivity. Isoboles falling within this envelope indicate an additive relationship between A and B. Isoboles falling to the left of the envelope indicate synergy.

ice for 30 min. The cells were washed twice with FACS buffer and incubated with goat anti-murine immunoglobulin-RPE conguate on ice. After 2 additional washes, FITC-MAb conjugates were added and incubated for 30 rnin on ice. All cells were analyzed in an Epics 753 flow cytometer (Coulter, Hialeah, FL).

Cooperutivity assay In order to determine whether binding of the antibody portion of one immunotoxin enhanced the cytotoxic effect of the second immunotoxin, 1 x lo6 SKBr3 cells were incubated with 5 pg/ml 260F9-RTA 5 pg/ml 454A12 or 5 pg/ml 454A12-RTA + 5 pg/ml 260F9. Controls consisted of SKBr3 cells incubated with 260F9 alone, 454A12 alone, 260F9-RTA, 454A12-RTA, 260F9 + 454A12, and 260F9-RTA + 454A12RTA. The effects of these agents on SKBr3 cells were analyzed by clonogenic assay as described previously.

+

Conjugation of gold sols to antibodies MAbs were conjugated to gold sols according to the instructions of the manufacturer (Amersham) (Slot and Geuze, 1985). Briefly, antibodies were dialyzed against 2 mM sodium borate buffer, pH 9.0, Centrifuged at 100,000 g for 1 hr, retaining the upper two thirds of the supernatant. The pH of the gold sols was adjusted to approximately 0.5 units above the PI of the antibody, and 25 ml of the sol was added to 17 to 30 pg/ml antibody. After stirring for 2 min, the pH was adjusted to 9.0 with potassium carbonate and the reaction was quenched by the addition of 10% BSA to a final concentration of 1%. To remove any unconjugated gold particles, the mixture was centrifuged at 4°C for 45 rnin at 45,OOOg for the 5-nm sols, or at 12,OOOg for the 15-nm beads. The pellet was re-suspended in the lower 10% of the supernatant and layered over a 10 to 30% glyccrol step gradicnt. This was centrifuged for 45 rnin at 125,OOOg for the 5-nm beads, or at 15,OOOg for the 15-nm beads. Fractions containing the concentrated antibody gold were then dialyzed against 1% BSA/Tris buffer, pH 8.2, to remove the glycerol and the O.D. 520 nm (O.D.Q~)was measured. Immunocytochemistty Confluent SKBr3 cells in 60-mm Petri plates were exposed to gold conjugates (0.4 to 0.6 ml) of 260F9 and 454A12 alone or together, for 30 rnin at 4°C (0.D.520 1.7 to 3.0 for 15-nm conjugates, 0.D.s200.8 to 1.4 for 5-nm conjugates). The cells

774

CREWS ETAL.

were then washed with PBS to remove any non-adherent antibody and warmed to 37°C for 0, 10, or 30 min to allow internalization. The PBS was removed and the dishes were filled with 2% glutaraldehyde in 150 mM sodium cacodylate buffer, pH 7.4, with 2.5 mM CaCI. Fixation was allowed to proceed from 1 to 24 hr. The fixed cells were scored with a blade, removed from the dishes with a rubber policeman, and sedimented by centrifugation. Pellets were post-fixed on ice with 2% osmium tetroxide and 0.5% or 1%potassium ferrocyanide in the same buffer. After washing in cacodylate, cells were transferred to sodium acetate buffer and the pellets were stained en bloc with 1%uranyl acetate in 0.2 M sodium acetate, pH 5.2. Subsequently, cells were dehydrated with increasing concentrations of ethanol, incubated in mixtures of 100% ethanol with epoxy, and embedded in beem capsules in Embed 812 (EM Science, Cherry Hill, NJ). Pale gold to silver sections were cut on a Reichert-Jung (Vienna, Austria) Ultracut E microtome and stained either with saturated uranyl acetate and lead citrate, or with lead citrate alone. The sections were examined on a Philips 300 electron microscope at 80kV.

toxins for a period of 3 hr at 37°C. Each of the 6 possible pairs of the 4 immunotoxins was tested. In each case, combined treatment with a pair of immunotoxins was more effective in eliminating SKBr3 cells than were the individual agents. For example, when 454A12-RTA was used alone, increasing concentrations eliminated 0.5 to 1.0 log clonogenic units (Fig. 3a). The 260F9-RTA (5 pg/ml) eliminated less than 1 log of SKBr3. When the same concentrations of 454A12-RTA were used in combination with 5 pg/ml of 260F9-RTA, 1 to 4 logs of clonogenic units were eliminated. Similarly enhanced cytotoxic activity was observed when 1 pg/ml to 5 pg/ml of 260F9-RTA was used in combination with 5 pg/ml of 454A12 (Fig. 3b). The interaction of the 2 immunotoxins within a pair was further evaluated by isobolographic analysis to determine whether the immunotoxins were acting additively or synergistically in their elimination of SKE3r3 tumor cells. Because each of the 4 immunotoxins studied exhibited non-linear doseresponse curves, isobolographic analysis was performed using the "envelope of additivity" described by Steel and Peckham (Johnson and Brown, 1961). The "envelopes" were calculated for each combination of immunotoxins and reflected the range

RESULTS

Binding of antibodies to a breast-cancer cell line Binding of 260F9, 31765,454A12 and 741F8 to the SKBr3 breast-cancer cell line was evaluated using FITC-antibody conjugates of the 4 antibodies. Data in Table I show the relative binding of these antibodies to SKBr3 cells. Antibody 741F8 bound to the greatest number of apparent antigenic sites on SKBr3 cells, whereas 454A12 bound to the least. Effect of immunotoxins on the clonogenic growth of breast-cancer cells The effect of the immunotoxins 260F9-RTA, 317G5-RTA, 454A12-RTA and 741F8-RTA on the growth of SKBr3 breastcancer cells was assessed using clonogenic assays. Tumor cells were incubated with immunotoxins for 3 hr at 37°C and surviving clonogenic units were determined by a limiting dilution assay. Previous studies from this laboratory have indicated a direct correlation between visual scoring and incorporation of 3H-thymidine by clonogenic cells in this assay (Anderson et al., 1989). Each of the 4 immunotoxins decreased the clonogenic units of SKBr3 cells in a dose-dependent fashion, as shown in Figure 2. Even at the highest concentration tested (10 pgiml), none of the immunotoxins when used alone eliminated all clonogenic SKBr3 cells. Consequently, it was of interest to determine whether combinations of the immunotoxins were more effective than individual immunotoxins for eliminating tumor cells.

4

3

2

1

i

X

'

1

No Treatment 317G5-RTA

0

260F9-RTA

A

454A12-RTA

A 741F8-RTA

0

I I 1 " I " " ' I I I ' I

0

2 4 6 8 Concentration of lmrnunotoxin (pglml)

10

FIGURE 2 - Comparison of immunotoxin activity against the SKBr3 cell line. Results of clonogenic assays for each of the individual immunotoxins presented here are depicted. B

A

Increased clonogenic elimination of breast-cancer cell lines with combinations of immunotoxins The effect of immunotoxins in combination on SKBr3 breast-cancer cells was cxamined using clonogenic assays in which tumor cells were exposed to combinations of 2 immunoTABLE I - RELATIVE BINDING OF ANTIBODIES TO S-13 Antibody

31705 260F9 454A12 741F8

Determinant

Antigenic sites x 10-5/cell k SD

42-kDa glycoprotein 55-kDa glycoprotein 180-kDatransferrin receptor 185-kDa HER-2ineu

2.54 t .27 1.4 ? 0.37 0.158 k 2.4 9.15 t 2.4

SKBr3 cells (5 x lo5) were incubated with antibody-FITC conjugates for 30 min on ice, washed and fixed with formalin. Immunofluorescencewas analyzed with an Epics 753 flow cytometer. Beads with a known number of FITC molecules per bead were analyzed as standards.

-

00

1

2

3

4

5

454A12-RTA (pg/rnll

FIGURE 3 - Enhanced elimination of SKBr3 cells with combinations of immunotoxins. (a) Cytotoxicity when the concentration of 454A12-RTA is varied in combination with a fixed concentration (5 kg/ml) of 260F9-RTA (0---0); cytoxicity of different doses of 454A12-RTA alone (A-A) and the cytotoxicity of 260F9-RTA alone at 5 pgiml ( 0 ) .(b) Cytotoxicity when the concentration of 260F9-RTA is varied in combination with a fixed concentration (5 pg/ml) of 454A12-RTA 0---0); cytoxicity of different doses of 260F9-RTA alone (A-A\ and cytoxicity of 454A12-RTA alone at 5 pg/ml(O).

SYNERGISTIC CYTOTOXICITY OF IMMUNOTOXINS

in which the 2 immunotoxins of a pair could act additively to produce a defined increment of tumor-cell killing. The combination of 260F9-RTA with 454A12-RTA was found to be synergistic in 4 separate assays. Figure 4a shows that the isobole for 0.9 log tumor cell elimination fell below the envelope of additivity, indicating synergy between 454A12R T A and 260F9-RTA. All other combinations of the immunotoxins produced isoboles that fell within their respective envelopes, indicating that these combinations of immunotoxins interacted additively (Fig. 4b-f ). As the clonogenic assay required 14 days to complete, we also evaluated the effect of 260F9-RTA and 454A12-RTA on SKBr3 in a 24-hr 35Smethionine incorporation assay. Synergistic interaction of the immunotoxins was observed in the short-term assay (data not shown). T o determine whether synergistic interactions of 454A12-RTA and 260F9-RTA applied only to SKBr3 cells, 3 other breast-cancer cell lines were treated with these immunotoxins. Isobolographic analysis indicated synergistic cytotoxic activity against CAMAl and BT483, but not against BT20 cells (data not shown). Level of cooperativity between imrnunotoxins and unconjugated antibodies Synergy between 260F9-RTA and 454A12-RTA might relate to the binding of the antibody from one immunotoxin enhancing the binding or internalization of the other immunotoxin, resulting in a greater effect against tumors. This possibility was investigated using clonogenic assays in which SKBr3 cells were incubated with 260F9-RTA 454A12 and with 454A12-RTA 260F9. Controls consisted of untreated SKBr3 cells, cells incubated with each free antibody alone, and cells incubated with 260F9-RTA 454A12-RTA. If cooperativity between the binding o r internalization of the immunotoxins accounted for the synergy, one would expect that the cytotoxic activity of 260F9-RTA + 454A12 and 454A12-RTA + 260FY would be greater than either immunotoxin alone and similar to that of 260F9-RTA + 454A12-RTA. A combination of 260F9RTA and 454A12 eliminated 2.2 logs of clonogenic SKBr3 and a combination of 454A12-RTA and 260F9 eliminated 2.4 logs.

+

+

+

FIGURE4 - (a) Isobolographic analysis of 0.9 log elimination of

clonogenic units with 260F9-RTA and 454A12-RTA. The envelope of additivity is outlined by mode I and mode I1 as indicated. (6). Isobolographic analysis of 0.82 log elimination of clonogenic units with 317G5-RTA and 454A12-RTA. (c) Isobolographic analysis of 1.8 log elimination of clonogenic units with 260F9-RTA and 741F8-RTA. (d) Isobolographic analysis of 1.0 log elimination of clonogenic units with 454A12-RTA and 741F8-RTA. (e) Isobolographic analysis of 1.6 log elimination of clonogenic units with 260F9-RTA and 317G5-RTA. (f) Isobolographic analysis of 1.6 log elimination of clonogenic units with 317G5-RTA and 741F8RTA.

775

When used alone, 260F9-RTA gave 2.6 logs of clonogenic elimination and 454A12-RTA gave 2.9 logs of elimination. The combination of 260F9-RTA with 454A12-RTA yielded 5.4 logs of growth inhibition. These results suggest that the synergy observed between 260F9-RTA and 454A12-RTA is not solely a function of cooperativity between the antibodies. This does not, however, rule out the possibility that mutual effects on binding or internalization might be necessary but not sufficient to observe synergistic cytotoxicity. Binding of 2 antibodies to SKBr3 cells Because the antibodies 260F9, 317G5, 454A12 and 741F8 bind to distinct cell-surface antigens, it was of interest to determine whether the additive or synergistic cytotoxic effects of the paired immunotoxins could be related to additive or supra-additive binding of the corresponding antibody pairs to SKBr3 cells. One-color immunofluorescence using antibodyFITC conjugates was used to examine the binding characteristics of combinations of 2 antibodies. Additive binding was defined as an increase in the percentage of positive cells and/or the mean fluorescent intensity of the cells above that with the best individual antibody alone. In none of the combinations, however, was consistent additive or supraadditive binding of antibody observed (results not shown). Enhanced cytotoxic activity observed with each of the 6 immunotoxin combinations could relate, in part, to the presence of sub-populations of tumor cells that lacked one or the other antigen recognized by each immunotoxin in a pair. Previous work from our laboratory had demonstrated 260F9+317G5- and 260F9-317G5+ sub-populations among SKBr3 cells (Yu et al., 1990). When 260FY-RTA and 317G5R T A were used in combination, both sub-populations could be eliminated. In order to assess the other antibody combinations, two-color immunofluorescence was performed in which one antibody of a pair was incubated with SKBr3 cells and then allowed to react with goat anti-mouse RPE. The cells were then incubated with the second antibody of the pair, which was conjugated to FITC. Red and green fluorescence was measured by flow cytometry, so that the percentage of cells with red or green fluorescence (ie., cells with antigen recognized by only one antibody of the pair), those with both red and green fluorescence (i.e., cells with antigens recognized by both antibodies of the pair), and those non-fluorescent cells that expressed neither antigen could be quantitated. Figure 5 shows that 56.13% of the SKBr3 cells were positive for 260F9 only, 1.56% of SKBr3 cells were positive for 454A12 only, 42.07% were positive for both 260FY and 454A12, and 0.24% fell below the threshold of detection of either antibody conjugate. Again, t h e p r e s e n c e of cells t h a t were 260F9+454A12- and those that were 260FY-454A12+ is significant, in that treatment with both 260F9-RTA and 454A12R T A would b e required to eliminate both sub-populations. A fraction of singly positive cells was identified with antibodies A and B in all combinations of antibodies, except for 260FY and 741F8 where no 741F8-260FY+ cells were observed (Table 11). Internalization of gold-labeled M b s The pattern of internalization for each of the MAbs was studied using electron microscopy to visualize antibodies that had been conjugated with gold particles of 5 or 15 nm. Gold-labeled antibodies were allowed to bind to SKBr3 cells at 4"C, and both the extent and route of their internalization was evaluated after warming the cells to 37°C. The 454A12 anti-transferrin receptor antibody bound over the entire surface of the cells in an apparently random distribution (Fig. 6a). Within 5 to 10 min at 3 7 T , 454A12 was endocytosed extensively and much of it was found in coated pits and vesicles (Fig.

776

CREWS ETAL.

6b, inset) as well as in peripheral endosomes (arrowheads in Fig. 6b). By 30 rnin at 3TC, most of the surface 454A12 was internalized and was seen in deep cytoplasmic vesicles (Fig. 6, c, d, arrowheads), including tubular endosomes, and in multivesicular bodies (Fig. 6, c, d, brackets). The 260F9-gold conjugate also internalized through the same structures (Fig. 7), and at 30 min was found in deep endosomes and in multivesicular bodies (Fig. 7c) near the Golgi zone of the cell (Fig. 7b). The internalization of 260F9 was, however, much slower than that of 454A12. Much of the 260F9 remained on the cell surface even after 30 min at 37°C (compare Figs. 6d and 7b). Neither antibody seemed to be routed to the Golgi stack, although the multivesicular bodies were often in the vicinity of the Golgi complex. Co-internalization of 454A12 and 260F9 was studied using antibodies that had been conjugated with gold particles of different sizes. In our initial studies, 454A12 was conjugated to 5-nm gold particles and 260F9 to 15-nm particles. Both antibodies could be found co-localized in each of the compartments along the endocytic pathway (Fig. 8). 454A12 and 260F9

"ill,

.

could be found in the same coated pits (Fig. 8, a, b), coated vesicles (Fig. 8c) and endosomes (Fig. 8, d, e), as well as in multivesicular bodies (Fig. 8d). A t early time points ( 5 to 10 min), a much larger fraction of 454A12 was internalized than of 260F9 (Fig. 8b). However, in the presence of 260F9, the intracellular distribution of 454A12 was altered. After incubation for 30 min at 37"C, much of the 454A12 was in peripheral endosomes and a larger fraction was on the cell surface. At this time interval in the absence of 260F9, most 454A12 was concentrated in deep endosomes and multivesicular bodies (compare Figs. 8c and 6c). This same altered distribution of the endocytosed 454A12 was also observed when 454A12 was conjugated with 15-nm gold particles and 260F9 with 5-nm particles (Fig. 8e). Persistence of 454A12 in peripheral endosomes was seen only in the presence of 260F9. Passage of 454A12 through endosomes was slowed to a similar degree in the presence of 260F9-gold, 260F9-RTA or unconjugated 260F9 (data not shown). In similar double-label experiments with either 31765 o r 741F8 conjugates, most of the 454A12 conjugate was cleared from the cell surface and routed into deep endosomes and multivesicular bodies as rapidly as in the absence of the other antibodies (data not shown). Therefore, the apparent delay in the kinetics of endocytosis was seen only with the antibody combinations that exhibited synergistic cytotoxic activity when conjugated with the A chain of ricin, and not with those antibody combinations that provided additive effects. DISCUSSION

.

Our study has examined the cytotoxic activity of 4 immunotoxins used individually and in combination against breastcancer cell lines. All immunotoxins exhibited inhibition of clonogenic tumor growth when used individually. When the immunotoxins were used in combination, all pairs of immunotoxins showed at least additive cytotoxic activity. Experiments reported here and in previous work from our laboratory (Yu et al., 1990) indicate that 5 of the 6 immunotoxin combinations produced log-additive inhibition of tumor-cell growth, while 454A12-RPE the sixth exhibited synergistic cytotoxic activity. The additive effects may be due in part to sub-populations of tumor cells that lacked one, but not both, antigenic determinants. In addition, additive effects could be a function of increased Quadrant Percent internalization of conjugate molecules that may exceed a threshold for tumor-cell death. A n apparent discrepancy 1 = 260F9 56.1 3 between the fraction of cells targeted by some antibodies, for 454A12 example 454A12, and the log killing may reflect effective 2 = 260F9 + 42.07 internalization of some conjugates despite relatively low levels 454A12 + of antigen expression that may be difficult to detect by flow-cytometric analysis. Different mechanisms of intracellular 3 = 260F9 1.56 trafficking also must be considered as a possible reason for 454A12 additive cytotoxic effects. 4 = 260F9 0.24 With the exception of Vallera et al. (1983) and our recent work (Yu et al., 1990), there are few reports of increased 454A12 cvtotoxic effects when combinations of immunotoxins are used. FIGURE5 - Coincident binding of ~ ~ O F ~ - F Iand T C 4 5 4 ~ 1 2 - \ i a k r a et al. (1983) found greater ehlination of human T cells with a combination of 3 immunotoxins than with any of the 3 RPE by SKBr3 cells.

+

-

-

+

TABLE I1 - COINCIDENT BINDING' OF ANTIBODIES TO S U r 3 CELLS Antibody combination Antibody A

260F9 741F8 317G5 741F8

Antibody B

Antigenic phenotype of SKBr3 cells (%) A+ B-

A- B+

A+ B+

A- B-

454A12 56.13 0.24 42.07 1.56 31765 0.51 0.03 99.33 0.13 454A12 50.50 0.10 49.33 0.10 454A12 0.01 80.57 19.26 0.16 317G5 260F9 0.42 0.05 99.37 0.16 741F8 260F9 9.49 0.0 90.35 0.16 'SKBr3 cells were incubated sequentially with unconjugated primary antibody followed by RPE-conjugated goat anti-mouse antibody, followed by FITC-conjugated secondary antibody.

SYNERGISTIC CYTOTOXICITY OF IMMUNOTOXINS

777

FIGURE6 - (a) Endocytosis of anti-transferrin receptor-gold by SKEh-3cells. Antibody 454A12, conjugated to 15 nm gold particles was allowed to bind to SKBr3 cells for 30 min at 4°C. After washing the excess antibody, the cells were further incubated at 4°C (a), or at 37°C for 10 min (b) or 30 min (c,d). Arrowheads in panels ( b 4 ) point to antibody internalized in a variety of vesicles. The insets in (b) show examples of coated (presumably clathrin) vesicles containing internalized 454A12, which in this experiment was tagged with 5 nm gold. Brackets in (c) and (d) highlight multivesicular bodies containing the gold tracer. N, nucleus; M, mitochondria; G, Golgi complex. Scale bars, 0.5 bm.

alone. Bregni et al. (1986), however, found no greater elimination of Burkitt’s-lymphoma cell lines when multiple ricin-Achain immunotoxins were used than when optimal concentrations of singlc immunotoxins were used. Stong et al. (1985) could not demonstrate enhanced elimination of lymphoblastic leukemia cells when combinations of immunotoxins were compared with a single highly effective conjugate. Synergy between immunotoxins has not, to our knowledge, been previously reported. In this study, the combination of 454A12-RTA with 260F9RTA produced synergistic cytotoxic activity against SKBr3 cells in multiple assays. Although the clonogenic efficiency varied among experiments, the presence of synergy as defined by isobolographic analysis was consistent. The term synergy is generally defined as the condition in which 2 or more agents act together to produce an effect which is greater than the constituent effects of the agents. The algebraic and graphic representation of this definition requires that the constituent

agents have linear dose-response curves. Because many agents do not have linear dose-response curves, the presence of synergy is often difficult to determine. This difficulty may be overcome by using the analysis proposed by Steel and Peckham (Johnson and Brown, 1961), which compensates for deviations from linearity found in the dose-response curves of the individual agents. Since the immunotoxins used in the present study have non-linear dose-response curves, we found the “envelope of additivity” to be a useful method for determining the presence of synergy. The Steel and Peckham analysis poses stringent criteria for synergy, although this method has not been universally accepted (Berenbaum, 1989). Berenbaum’s definition of synergy maintains that it is unnecessary to construct an envelope of additivity, and would consider all of the immunotoxin combinations tested synergistic. We chose to use the Steel and Peckham analysis in order to determine those combinations which would be considered synergistic by the most conservative criteria.

778

CREWS ETAL.

FIGURE 7 - Endocytosis of 260F9-gold by SKBr3 cells. Antibody 260F9 was conjugated either to 15 nm gold (a,b) or to 5 nm gold (panel (c). After the gold was bound to the cells at 4°C for 30 min, the cells were washed and incubated at 37°C for 10 min (a) or 30 rnin (b,c). Arrowheads highlight internalized antibody gold complex in endosomes (a$) and in multivesicular bodies (c). Note the persistence of label on the surface after 30 rnin (b). N, nucleus; M, mitochondria. Scale bars, 0.5 km in (a) and (b); 0.25 km in (c). One advantage of isobolographic analysis is the consideration of immunotoxin interactions over a broad range of immunotoxin concentrations. Although data in Figure 3a suggest an abrupt increase in cytotoxicity when 454A12-RTA concentration was increased over a 2-fold range in the presence of 5 pg/ml 260F9-RTA, synergistic interactions have been observed using isobolographic analysis at substantially lower immunotoxin concentrations in Figure 4a. Similar results were obtained with CAMAl and BT483 cells. In these studies, tumor cells were incubated with up to 5 p.g/ml of immunotoxin, the highest concentration that might be maintained in clinical studies that utilize i.p. administration of the reagents. Because the combination of 454A12-RTA with 260F9-RTA consistently produced synergistic cytotoxic activity against SKBr3, C A W 1 and BT483 cells, we attempted to determine thc possible mechanism for this synergy. The synergistic interaction between 454A12-RTA and 260F9-RTA could not be explained by supra-additive binding of these immunotoxins to SKBr3 cells when the immunotoxins were used in combination. Likewise, the synergy observed was not a function of cooperativity in binding and/or internalization betwcen one of the immunotoxins and the antibody moiety of the second, since combinations of one immunotoxin with the corresponding

FIGURE 8 - Endocytosis of 454A12-gold and 260F9-gold in double-labelled SKBr3 cells. The antibody-gold conjugates were bound to the cells at 4°C for 30 min, and then allowed to internalize at 37°C for 5 rnin (a), 10 min (b,c) or 30 rnin (d,e). In ( a x ) , 454A12 is tagged with 5 nm gold particles, while 260F9 was tagged with 15 nm gold particles. Surface pits containing either one of the antibodies or both are highlighted with brackets. Arrowheads point to internalized anti-transferrin receptor. (b,c) Coated pits and vesicles with both antibodies; (d,e) co-localization of both MAbs in deep endosomal vesicles [bracketed in (d)] and in multivesicular bodies (e). Note the persistence of 454A12-gold on the cell surface even after 30 rnin [arrows in (d)].M, mitochondria; G, Golgi complex. Scale bars indicate 0.5 km in (a), (d);0.25 pm in (b) (c) (e). 9

9

unconjugated antibody did not produce any greater clonogenic elimination than did the immunotoxin used alone. A possible explanation for the synergy between 454A12R T A and 260F9-RTA is offered by studies of internalization of gold-labelled antibody using electron microscopy. Transferrin receptor, the target of 454A12, is known to internalize in many cell types via clathrin-coated pits and vesicles (Willingham et aZ., 1984; Hopkins, 1983; Neutra, et al., 1985). Synergy might relate to internalization of 260F9 through a different route, but

SYNERGISTIC CYTOTOXICITY OF IMMUNOTOXINS

the EM studies presented here indicate that 260F9 is also transported into endosomes by the same route. Kinetic internalization studies show that 454A12 is retained within peripheral endosomes for a longer period of time in the presence of 260F9. This did not appear to be an artifact produced by conjugating the antibodies with gold, in that similar behavior of 260F9 and 454A12 was observed when the antibodies were conjugated either with 5- or 15-nm gold particles. Moreover, unconjugated 260F9 and 260F9-RTA also slowed the passage of 454A12 gold conjugates through the endosomal compartment. Whether this reflects slowing of the passage of 454A12 to deep endosomes or a possible recirculation of 454A12 to the cell surface cannot be determined from our present study. In either event, the persistence of 454A12-RTA within endosomes may facilitate transfer of RTA into the cytoplasm and delay fusion with lysosomes which leads to degradation of the cytotoxic agent. Conjugation of antibodies to gold might, of course, alter their normal pattern of internalization (Neutra et al., 1985). Whether or not this is the case, 260F9 exerted a different effect on internalization of 454A12 than did 741F8 or 31765. The ability of antibodies to induce persistence of 454A12 in peripheral endosomes did not relate simply to the number of determinants recognized by each antibody. Determinants recognized by 260F9 were %fold more plentiful than those recognized by 454A12 (Table I), but 2- to 7-fold less plentiful than binding sites for the 31765 and 741F8 antibodies that did not affect internalization of 454A12-gold conjugates. The

779

molecular weight of antigenic targets did not correlate with the ability of antibodies to affect the internalization of transferrin receptor in the presence of 454A12. The p55 recognized by 260F9 was larger than the p42 recognized by 31765, but substantially smaller than the HER-2/neu to which 741F8 bound. Clinical utility of immunotoxins is dependent upon their ability to bind to all clonogenic tumor cells but not to critical normal tissues. Use of immunotoxins in combination should permit targeting of a greater fraction of clonogenic tumor cells. Co-expression of antigen on tumor cells, but not on normal cells, could also provide a greater therapeutic index, provided that co-internalization of immunotoxins produced additive or synergistic cytotoxic activity. Our current data suggest that different combinations of immunotoxins can indeed exert additive or synergistic cytotoxic activity. ACKNOWLEDGEMENTS

We thank Dr. L. Houston for his most valuable advice and assistance in obtaining reagents. We also appreciate the contribution of Mr. P. Crews in developing the software for graphic analysis. The outstanding secretarial assistance of Mrs. K. Cash and Mrs. J. Carey as well as the bibliographic assistance of Mrs. P. Haythorn are gratefully acknowledged. This work was supported in part by Grants 5-R01-CA39930 and PO1 CA47741 from the National Cancer Institute, Department of Health and Human Services.

REFERENCES ANDERSON, I.C., SHPALL, E.J., LESLIE,D.S., NUSTAD,K., UGELSTAD, J., PETERS,W.P. and BAST, R.C., JR., Elimination of malignant clonogenic breast-cancer cells from human bone marrow. Cancer Rex, 49,4659-4664 (1989). BAST,R.C., JR., RITZ, J., LIPTON,J.M., FEENEY,M., SALLAN, S.E., NATHAN, D.G. and SCHLOSSMAN, S.F., Elimination of leukemic cells from human bone marrow using monoclonal antibody and complement. CancerRes., 43,1389-1394 (1983). BERENBAUM, M.C., What is synergy? Pharmacol. Rev., 41, 93-141 (1989). BJORN, M.J., GROETSEMA, G. and SCALAPINO, L., Antibody-Pseudomonus exotoxin A conjugates cytotoxic to human breast-cancer cells in vitro. Cancer Res., 46,3262-3267 (1986). BJORN,M.J., RING,D. and FRANKEL, A,, Evaluation of monoclonal antibodies for the development of breast-cancer immunotoxins. Cancer Res., 45,1214-1221 (1985). BLYTHMAN, H.E., CASELLAS, P., GROS,O., GROS,P., JANSEN,F.K., PAOLUCCI, F., PAU,B. and VIDAL,H., Immunotoxins: hybrid molecules of monoclonal antibodies and a toxin sub-unit specifically kill tumour cells. Nature (Lond.), 290,145-146 (1981). BOYEK,C.M., BOROWITZ, M.J., MCCAKTY, K.S., JR., KINNEY,R.B., EVERITT,L., DAWSON, D.V., RING,D. and BAST,R.C., JR., Heterogeneity of antigen expression in benign and malignant breast and ovarian epithelial cells. Int. J. Cancer, 43,55-60 (1989). BREGNI, M., DE FABRITILS, P., RASO,V., GREENBERGER, J., LIPTON,J., L., RITZ,J. and BAST,R.C., JR., Elimination NADLER,L., ROTHSTEIN, of clonogenic tumor cells from human bone marrow using a combination of monoclonal antibody: ricin A chain conjugates. Cancer Res., 46, 1208-1 2 13 (1986). CASELLAS,P., CANAT,X., FAUSEK,A.A., GROS,O., LAURENT, G., PONCELET, P. and JANSEN,F.K., Optimal elimination of leukemic T cells from human bone marrow with T101-ricin-A-chain immunotoxin. Blood, 65,289-297 (1985). ENDO,Y., Mechanism of actions of ricin and related toxins on the inactivation of eukaryotic ribosomes. Cancer Treat. Res., 37, 75-89 (1988). FETEANU, A,,Labelled antibodies in biology and medicine (2nd ed.), pp. 52-57, Abacus Press, Tunbridge Wells, UK (1978). FRANKEL, A.E., RING,D.B., TRINGALE, F. and HSIEH-MA,S.T., Tissue distribution of breast-cancer-associated antigens defined by monoclonal antibodies. J. biol. resp. Mod., 4,273-286 (1985).

HOPKINS,C.R., Intracellular routing of transferrin and transferrin receptors in epidermoid carcinoma A431 cells. Cell, 35, 321-330 (1983). JOHNSON, E.A. and BROWN, B.W., The Spearman estimator for serial dilution assays. Biometrics, 17,79-88 (1961). NEUTRA,M.R., CIECHANOVER, A., OWEN,L.S. and LODISH,H.F., Intracellular transport of transferrin- and asialoorosomucoid-colloidal gold conjugates to lysosomes after receptor-mediated endocytosis. J. Histochern. Cytochem., 33,1134-1144 (1985). OLSNES,S. and SANDVIG, K., How protein toxins enter and kill cells. Cancer Treat. Res., 37,39-73 (1988). PIRKER,R., FITZGERALD, D.J., HAMILTON, T.C., OZOLS,R.F., WILLINGHAM,M.C. and PASTAN,I., Anti-transferrin receptor antibody linked to Pseudomonas exotoxin as a model immunotoxin in human ovarian carcinoma cell lines. Cancer Rex, 45,751-757 (1985). SLOT,J.W. and GEUZE,H.J., A new method of preparing gold probes for multiple-labeling cytochemistry. Europ. J. Cell. Biol., 38, 87-93 (1985). STEEL,G.G. and PECKHAM, M.J., Exploitable mechanisms in combined radiotherapy-chemotherapy: the concept of additivity. Int. J. Radiat. Oncol. Biol. Phys., 585-91 (1979). STONG,R.C., UCKUN,F., YOULE,R.J., KERSEY,J.H. and VALLERA, D.A., Use of multiple T-cell-directed intact ricin irnmunotoxins for autologous bone-marrow transplantation. Blood, 66,627-635 (1985). UCKUN, F.M., RAMAKRISHNAN, S. and HOUSTON, L.L., Immunotoxinmediated elimination of clonogenic tumor cells in the presence of human bone marrow. J. Immunol., 134,2010-2016 (1985). VALLERA,D.A., ASH, R.C., ZANJANI, E.D., KERSEY,J.H., LEBIEN, T.W., BEVERLEY, P.C., NEVILLE, D.M., JR. and YOULE,R.J., Anti-Tcell reagents for human bone-marrow transplantation: ricin linked to three monoclonal antibodies. Science, 222,512-515 (1983). WILLINGHAM. M.C.. HANOVER. J.A.. DICKSON. R.B. and PASTAN.I.. Morphologic’characterization of the pathway of transferrin endocytosis and recycling in human KB cells. Proc. nat. Acad. Sci. (Wash.), 81, 175-179 (1984). Yu, Y.H., CREWS,J.R., COOPER,K., RAMAKRISHNAN, S., HOUSTON, L.L., LESLIE,D.S., GEORGE, S.L., LIDOR,Y ., BOYER,C.M., RING,D.B. and BAST,R.C., JR., Use of immunotoxins in combination to inhibit clonogenic growth of human breast-carcinoma cells. Cancer Rex, 50, 3231-3238 (1990).