Intracellular Targeting with Low pH-triggered Bispecific Antibodies*

Bispecific antibodies were designed to deliver a reversibly bound ligand into target cells and then spontaneously release it upon passage into acidified vesicles. These reagents were assembled by coupling monoclonal antibodies that recognize acid-sensitive epitopes on diphtheria toxin to cell type-specific monoclonal antibodies. The dual binding capacity of the bispecific antibodies was confirmed by delivery of125I-diphtheria toxin to target molecules present on intact cells. Bispecific antibodies directed against transferrin receptors on human cells were loaded with toxin and tested for cytotoxicity. The mutant diphtheria toxins CRM107 and CRM45 were used since their inability to bind cell receptors renders them ordinarily nontoxic. Their full cytotoxic potential, however, was restored via bispecific antibody-mediated delivery and release within low pH intracellular vesicles. Cytotoxicity was shown to be specific by blocking receptor sites and to be acidification-dependent by protection using NH4Cl to raise endosomal pH. Kinetics for inhibition of cellular protein synthesis was identical for native diphtheria toxin and the bispecific antibody·CRM107 combination. The rate of inhibition (t 1/2 = 20 min) indicated that release of CRM107 from the antibody combining site was fast, and its toxic action was unimpeded by this delivery mechanism.

One of the earliest methods devised for the selective delivery of unmodified bioactive molecules into target cells used bispecific antibodies (1)(2)(3). By virtue of their dual specificity (4), these agents can noncovalently bind an effector moiety and simultaneously attach to specific antigenic sites on the cell membrane. Those initial studies used the plant toxin ricin or its enzymatically active A chain, whereas subsequent work used saporin (5) as a functional agent. Even though these complexes were internalized and ultimately killed target cells, neither the mechanism for antibody-toxin dissociation nor the intracellular compartment in which this release occured was known.
Diphtheria toxin acts via a well defined intracellular pathway that culminates with an obligatory transfer of its toxic moiety from acidic endosomes to the cytosol (6,7). That translocation is actuated by structural transitions in the toxin that are triggered in the pH 4.5-5.5 range (8) found within those organelles (9 -11). Acid-induced conformational changes in the transmembrane domain of DT 1 (12) are critical to the passage of its catalytic domain across vesicular membranes and into the cytosol, where it inactivates elongation factor 2 and kills the cell. These dynamic properties of DT have allowed us to design novel bispecific antibodies that not only accurately deliver toxin into target cells but also incorporate a precise mechanism for triggering its release upon entering endosomes, thereby exerting its full toxic potential.
We have selected murine monoclonal antibodies that tightly bind DT at neutral pH but spontaneously release the toxin as it undergoes a conformational change at pH 4.5-5.5 (see preceding paper (37)). Those antibodies were covalently linked to a second, cell-reactive antibody or receptor ligand to form hybrid molecules with dual specificity. The hybrid reagent carries antibody-bound toxin and selectively attaches to the surface of cells bearing a chosen target site. Toxin is released in its lethal form when the bispecific antibody-toxin complex is taken into cells and encounters the low pH in endosomes.
Monoclonal Antibodies-The attributes of the 7D3 monoclonal antibody directed against the human transferrin receptor have been reported (13). Monoclonal anti-DT antibodies were generated and characterized as described in the previous paper (37). A hybridoma cell line producing a human anti-DT monoclonal antibody designated F10 (14) was obtained from the American Type Culture Collection (Rockville, MD). Hybridoma clones were injected into pristane-primed BALB/c mice or Swiss nu/nu athymic mice (Taconic, Germantown, NY), ascites fluid was collected, and the 150-kDa anti-DT monoclonal antibody was isolated by (NH 4 ) 2 SO 4 precipitation and size fractionation on a high resolution Sephacryl S-300 column. The antibodies were Ͼ90% pure as judged by polyacrylamide gel electrophoresis. Immunoblot reactivity with the A versus B fragment of nicked DT was used to localize the region of antibody binding (15). Interaction with CRM mutant proteins and genetically engineered DT constructs helped to further characterize antibody specificity (see Table I).
Production of Bispecific Antibodies and Conjugates-A method for unidirectionally disulfide-linking two monoclonal antibodies has been described (3). The purified antibodies were each reacted with a 6-fold molar excess of SPDP, and the uncoupled reagent was removed by passage through a Sephadex G-25 column equilibrated with 0.1 M sodium acetate, 0.1 M NaCl, pH 4.5. This resulted in the addition of approximately 3 mol of SPDP/mol of antibody. To produce an appropriate bispecific antibody combination, the SPDP-substituted anti-transferrin receptor antibody was reduced with 50 mM dithiothreitol for 30 min, and the protein fraction was isolated by passage through a Sephadex G-25 column equilibrated with PBS. This thiolated antibody was then mixed with an SPDP-substituted anti-DT antibody, and the com-* This work was supported by National Institutes of Health Grants CA49856 and CA49254. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ bination was allowed to react overnight. The product, a 300-kDa disulfide-linked heterodimeric antibody with two binding sites for DT plus two sites directed against the transferrin receptor, was separated from unreacted antibody by chromatography on high resolution Sephacryl S-300. Fractions were analyzed by polyacrylamide gel electrophoresis. No difference was seen when the bispecific antibody and its parent anti-DT were compared using 125 I-DT binding titrations (37), indicating that they retained full activity.
The 7D3-CRM107 disulfide-linked covalent conjugate was formed in an analogous manner. The thiolated 7D3 monoclonal antibody was reacted with CRM107 that had been sparingly substituted with SPDP at neutral pH. The resulting conjugate was purified by size exclusion chromatography.
Isolation of Toxins-Diphtheria toxin was isolated (16) from a partially purified preparation purchased from Connaught Laboratories. The production and purification of ricin A chain from whole ricin was carried out as previously reported (17,18). CRM107 and CRM45 (19 -21) were obtained from cultures of the appropriate mutant strain of Corynebacterium diphtheriae as described earlier (19). Toxin was isolated from the culture supernatant either by (NH 4 ) 2 SO 4 precipitation or by phenyl-Sepharose chromatography (22) and purified further as described (16). CRM107 was nicked by limited proteolytic cleavage with trypsin or urokinase (23).
Cell Lines-The human acute lymphoblastic leukemia CEM cell line and human colon adenocarcinoma LS 174T cell line were obtained from the American Type Culture Collection. The human mesothelioma H-Meso cell line (24) was used previously for cytotoxicity studies (13,25). Cells were cultured at 37°C with 5% CO 2 in tissue culture flasks containing RPMI 1640 medium with 10% fetal calf serum, L-glutamine, and penicillin-streptomycin (Life Technologies, Inc.). Adherent cells were released from the culture flask using 0.05% trypsin, 0.02% EDTA for 10 min at 37°C, and then medium plus serum was added to inactivate the trypsin.
Binding of Toxin to Cells-Diphtheria toxin (ϳ4.5 Ci/g) was labeled with 125 I using the iodogen method (26). Radioactivity was measured using a Beckman model 5500 gamma counter. Two different protocols were used to deliver 125 I-DT to cells using the bispecific antibodies.
In the first procedure, CEM cells (1 ϫ 10 7 ) were removed from the culture medium and suspended in 100 l of PBS. Bispecific antibodies were added at 10 Ϫ7 M and incubated with the cells for 30 min at 2°C, and unbound reagent was then washed away with cold PBS. The bispecific antibody-coated cells were exposed to 125 I-DT for 30 min at 2°C, washed with cold PBS, and analyzed for radioactivity to measure the amount bound to cells. In some experiments, the 7D3 monoclonal antibody was included at 10 Ϫ5 M during the first step to block attachment of the bispecific antibody.
In the second method, 125 I-DT was added either to 100 l of PBS or 100 l of PBS containing the designated hybrid antibody at 10 Ϫ8 M and allowed to combine at 22°C for 15 min. CEM cells (1 ϫ 10 7 ) were added to each preparation, incubated for 1 h at 2°C, washed with cold PBS, and then analyzed for radioactivity.
Cytotoxicity Assays-For cytotoxicity assays, culture medium was removed, and the cells were resuspended at 5 ϫ 10 5 /ml in leucine-free minimal essential medium (Life Technologies, Inc.) supplemented with L-glutamine and penicillin-streptomycin. Aliquots were added to 96well microtiter plates so as to give a final volume of 200 l/well. Other additions made to the wells did not exceed a total volume of 20 l. The bispecific antibody and CRM toxins were added either separately or mixed first so that the specified final concentration was achieved by diluting the preformed complex in the microtiter well. Both methods of treatment gave essentially identical results. After these components were delivered at room temperature, the cells were incubated for the designated time at 37°C in 5% CO 2 . When the incubation was completed, 2.5 Ci of [ 3 H]leucine was added to each well, and the incubation was extended for 30 min at 37°C. The cells were then collected onto glass fiber filters using a Mash II cell harvester (Life Technologies, Inc.). Incorporation of [ 3 H]leucine into cellular protein was measured by scintillation counting of the glass fiber discs. When protein synthesis is reduced to Ͻ10% that of the control level, the cytotoxic effect may be underestimated, so these values were not weighted as strongly in the kinetics study. All of the cytotoxicity assay wells were set up in duplicate, and the results were averaged. The coefficient of variation of the assay was usually Ͻ 0.1.

Hybrid Antibody-mediated Delivery of 125 I-Toxin to Cellular
Targets-Bispecific antibodies were formed by linking various anti-DT monoclonal antibodies (Table I) to an anti-human transferrin receptor monoclonal antibody designated 7D3. Dual specificity was verified by measuring the capacity of these reagents to attach simultaneously to transferrin receptors on the cell and bind 125 I-toxin. The CEM cell line, which expresses high levels of transferrin receptor, served as a target, and two different protocols were followed to deliver 125 I-DT via the hybrid antibodies. Cells were coated first with bispecific antibody by pretreatment at 2°C for 30 min with an ensuing wash to remove unbound molecules. This surface-bound hybrid antibody with its empty toxin binding sites was allowed to capture the 125 I-DT subsequently added. Alternatively, bispecific antibody and 125 I-DT were precomplexed so that they could be bound to cell surface transferrin receptors as a single reagent. In each case, the use of anti-transferrin receptor/ anti-DT bispecific antibodies resulted in the delivery of 5-10 times more 125 I-DT to cells compared with their basal 125 I-DT binding level (data not shown). This enhancement was transferrin receptor-specific, since preoccupying the target epitope with an excess of unmodified 7D3 blocked hybrid antibody attachment and subsequent 125 I-DT binding.
Bispecific Antibody-mediated Cytotoxicity of CRM107-Hybrid antibody-mediated cytotoxicity was evaluated using two mutant forms of DT, CRM107 and CRM45, that have little or no inherent capacity for attaching to cells (27). CRM107 was bound to bispecific antibodies targeted against the transferrin receptors on a human mesothelioma cell line (H-Meso), and its inhibition of [ 3 H]leucine incorporation into protein was measured (Table II). A 2-h incubation time was sufficient to obtain extensive cytotoxic effects. Whereas CRM107 alone is incapable of entering cells and inhibiting protein synthesis, it became a very effective and rapid-acting cytotoxin when used in combination with the anti-transferrin receptor/anti-DT bispecific antibody, 7D3/5E8. This potent lethal action was dependent upon hybrid antibody-mediated delivery to transferrin receptors, since much less toxicity was observed when these target sites were blocked by including an excess of free anti-human transferrin receptor antibody (7D3) ( Table II). The minor fraction of [ 3 H]leucine incorporation that was not restored indicates that the blockade of receptors was not absolute. Separate experiments showed that the hybrid antibody was without effect in the absence of CRM107 and that mouse NS-1 cells, which lack the human transferrin receptor, were unaffected by toxin targeted by the 7D3-based bispecific antibody (data not shown).
The transferrin receptor was targeted because it normally cycles into acidic endosomes where Fe 3ϩ release from transferrin is induced by the low pH (28). It was assumed that bispecific antibody-mediated cytotoxicity would likewise be dependent on the acidic environment of intracellular compartments, and this was demonstrated by adding NH 4 Cl (29) to the cells. The weak base, which is known to raise vesicle pH, greatly reduced the ability of the bispecific antibody-CRM107 combination to kill H-Meso cells (data not shown). Low vesicular pH is essential for action, since it both triggers release of CRM107 from the antibody and induces translocation of toxin into the cytosol, where it enzymatically inactivates elongation factor 2. Dose response curves were generated to better characterize the multicomponent system. H-Meso cells were treated with a fixed high level of the bifunctional antibody, 7D3/5E8, plus increasing amounts of CRM107 (Fig. 1A). Maximal cytotoxicity was attained as the ratio of CRM molecules to toxin binding sites approached 2:1. Likewise, cells were exposed to a set high concentration of CRM107 plus varying amounts of the bispecific antibody. Cytotoxicity on both of the cell lines tested reached a maximum level when 7D3/5E8 was at 10 Ϫ10 M or higher (Fig. 1B).
The transferrin receptor-directed hybrid antibody⅐CRM107 complex was also assayed on human colon adenocarcinoma cells to determine if the same high cytotoxic potency found for the H-Meso line extended to other malignant cell types. The combined action of CRM107 plus hybrid at 10 Ϫ8 M produced extensive killing of these cells within 2 h, and its potency was comparable to 10 Ϫ7 M native DT (Table III). These results indicated that bispecific antibody delivery not only rendered CRM107 cytotoxic to cells but also suggested that its entry via the transferrin pathway was as efficient as DT uptake by its usual mechanism.
Kinetics of Cytotoxicity-A fundamental premise underlying the acid-triggered bispecific antibody carrier concept is that the mechanism of toxin action proceeds normally after specific targeting has been achieved. The kinetics for inhibition of protein synthesis gives a critical measure of toxin efficiency by monitoring how rapidly it gains access to and inactivates elongation factor 2 in the cytosol. This parameter was therefore used to evaluate bispecific antibody-delivered CRM107 (Fig. 2).
The time course of protein synthesis inhibition in H-Meso cells was monitored following the addition of native DT alone or the 7D3/5E8 bispecific antibody complexed with either CRM107 or nicked CRM107. Each of these agents gave a comparable kinetics pattern, which was characterized by a 15-min lag period followed by a rapid first order inactivation phase, with t 1/2 ϭ 20 min (Fig. 2). The fact that hybrid antibodydelivered CRM107 killed cells as rapidly as native DT suggests that its release from the antibody combining site was unimpeded and that translocation into the cytosol proceeded efficiently. Proteolytic cleavage of the bispecific antibody-delivered CRM107 was apparently not rate-limiting, since prior nicking of the toxin had little effect on its rate of action (Fig. 2). Similar biphasic kinetics patterns were given by the 7D3/3E1 and 7D3/4B7 hybrid antibodies plus CRM107, with a t 1/2 ϭ 24 and 36 min, respectively. The 5E8 and 3E1 antibodies recognized an epitope localized on the receptor binding domain of DT, whereas 4B7 bound to the catalytic domain (Table I).

Comparison of Acid-sensitive and Acid-insensitive Bispecific
Antibodies-Different hybrid reagents were constructed to compare the effectiveness of antibodies that attach to pHsensitive toxin epitopes with those that bind to acid-insensitive regions of DT. Two antibodies, F10 (14) and 5F5, which did not release from immobilized DT when the pH was lowered to 4.0 (37), were used for this purpose. The 7D3 anti-transferrin receptor antibody was coupled either to these acid-stable anti-  bodies, forming 7D3/F10 and 7D3/5F5, or to acid-releasable antibodies, producing 7D3/5E8 and 7D3/5D5. The identically targeted bispecific antibodies were then tested for killing cells using CRM107 or CRM45 (Tables IV and V). The acid-stable 7D3/F10 agent plus CRM107 showed insignificant inhibition when tested after 3 or 24 h incubation, whereas the pH-sensitive 7D3/5E8-CRM107 combination caused extensive cell damage (Table IV). Similarly, the pH-insensitive 7D3/5F5 hybrid antibody failed to mediate CRM45 cytotoxicity, whereas this toxin gave 95% inhibition in 2 h when delivered by the pHsensitive 7D3/5D5 reagent (Table V). Extending the incubation time an additional 15 h did not improve the cytotoxic activity of the 7D3/5F5-CRM45 combination (data not shown). The 5E8 and F10 antibodies recognized an epitope localized on the receptor binding domain of DT, whereas 5D5 bound to the transmembrane domain and 5F5 bound to the catalytic domain (Table I). These large differences in cytotoxicity demonstrate the importance of the toxin-bearing component of the bispecific antibody. Each construct was directed to the identical cell-receptor site by 7D3 and therefore followed the same transferrin receptor internalization pathway. Translocation into the cytosol and cell killing was rapid when the toxin interaction was with an acid-sensitive epitope; however, if the hybrid antibody bound to an acid-insensitive epitope, the complex was not cytotoxic even after 24 h incubation. This implies that to be effective toxin must quickly release from the antibody combining sites while inside low pH endosomes. Only when the toxin is freed in these acidic compartments can it translocate into the cytosol and inactivate elongation factor 2.

Comparison of Covalently-linked and Bispecific
Antibodydelivered CRM107-A covalently-coupled anti-transferrin receptor-CRM107 conjugate (7D3-CRM107) was constructed by standard disulfide linkage methods. Dose response curves were generated by treating CEM cells for 16 h at 37°C with varying concentrations of either native DT, the 7D3-CRM107 covalent conjugate, or the 7D3/5E8 bispecific antibody in the presence of 10 Ϫ7 M CRM107 (Fig. 3). CEM cells are not particularly sensitive to DT as reflected by the IC 50 ϭ 2 ϫ 10 Ϫ9 M obtained with native toxin and the prolonged incubation time required for cytotoxicity. The transferrin receptor-directed 7D3-CRM107 disulfide conjugate was slightly more effective, giving an IC 50 ϭ 1 ϫ 10 Ϫ9 M. In contrast, hybrid-delivered CRM107 (IC 50 ϭ 4 ϫ 10 Ϫ12 M) was 250-fold more potent than the covalent conjugate, based upon the concentration of hybrid antibody added. Neither the 7D3/5E8 bispecific antibody alone nor CRM107 alone at 1 ϫ 10 Ϫ7 M had an effect upon the cells (Fig. 3). When tested on H-Meso cells, the IC 50 values for the 7D3-CRM107 disulfide conjugate and 7D3/5E8-delivered CRM were 2.5 ϫ 10 Ϫ9 and 7 ϫ 10 Ϫ12 M, respectively, a 360-fold differential. These results indicate that covalent coupling actually impedes toxin action since the disulfide-linked 7D3-CRM107 conjugate was much less potent than the corresponding 7D3/5E8 hybrid antibodydelivered CRM107. This observation may have important implications for the relative therapeutic effectiveness of these two types of anti-tumor agents when tested in animal model systems. DISCUSSION We originally designed hybrid antibodies with dual specificity to deliver sensitive bioactive molecules into target cells in a nondestructive manner. These reagents used an antibody to hold the molecule reversibly within its combining site while the second antibody or ligand component accurately targeted the complex to selected cells (1-3). This delivery method circumvents the potential steric inactivation of delicate molecules that may result from coupling to a carrier moiety by covalent link-   age, using either chemical means or genetically engineered fusion. Subsequent enzymatic or chemical cleavage from the carrier to restore activity is obviated since dissociation from the antibody combining site leaves the molecule unchanged.
In the present embodiment of this technology, we have incorporated a precise means of triggering the rapid release and full potency of the active molecule once it has reached its intracellular destination (Fig. 4). A binding-defective CRM mutant served as an active agent with a readily detectable cytotoxic activity. Its spontaneous release inside of the cell is based upon using monoclonal antibodies that bind to acid-sensitive epitopes that are altered upon passage of the toxin into acidified subcellular compartments (37). Efficient transfer of toxin from these vesicles and into the cytosol of target cells is signaled by a rapid inhibition of protein synthesis.
The acid-triggered bispecific antibodies were shown to mediate the cytotoxicity of CRM107 and CRM45, which are ordinarily inactive because they cannot bind to cells. These bispecific antibody-delivered CRMs killed most of the target cells within 2 h. When examined in more detail, the kinetics for inhibition of cellular protein synthesis was identical for the bispecific antibody⅐CRM107 combination and native DT, showing a t 1/2 ϭ 20 min for the rapid phase. This indicated that CRM released quickly from the antibody carrier and that its transmembrane domain facilitated the translocation of its enzymatic domain into the cytosol in a normal manner. The effectiveness of the CRM45 deletion mutant is especially pertinent since this protein consists solely of those two domains. Previous reports showed that truncated molecules such as CRM45 did not possess full cytotoxic activity when covalently linked to monoclonal antibody or other binding moieties (27,30,31). Such attenuation might be caused by steric encroachment of the closely coupled antibody on the translocation domain since our results revealed no inherent translocation deficiency in the CRM45 molecule.
Target cells were protected from the toxic action of hybrid antibody-delivered CRM by using NH 4 Cl to raise endosomal pH. This perturbation precluded both the acid-triggered release of toxin from the antibody combining site and the pHinduced translocation step needed for efficient cytotoxicity. Moreover, CRM107 delivered into cells using acid-insensitive bispecific antibodies was inactive, even after prolonged incubation. These antibodies bind to toxin epitopes that do not undergo a conformational change at low pH, and the toxin there-fore remains bound inside the endosome. Failure to release within endosomal vesicles may result in the re-routing of toxin to ensuing compartments like the golgi or to lysosomes where it is destroyed. These combined neutralization and acid-stable antibody results confirmed that CRM toxins must release from the antibody in acidified endosomes to function properly. Concomitantly, the system provides strong evidence for the acidinduced breakup of certain specific antigen-antibody complexes within the low pH vesicles of living cells.
Treatment of target cells with hybrid antibodies carrying ricin A chain instead of CRM107 showed an extremely delayed cytotoxic response that was unaffected by neutralization of endosomal pH by NH 4 Cl (data not shown). This slow action was in sharp contrast to the rapid acting, pH-sensitive, DT-based system and prevailed even though delivery via the transferrin pathway in the same cells was mediated by an identical targeting antibody. Unlike the direct pH-dependent endosomal route taken by DT, many toxins like ricin, Shiga toxin, and Pseudomonas exotoxin apparently follow more intricate and time consuming pathways before reaching their targets in the cytosol. The point of egress for these toxins may be the endoplasmic reticulum or Golgi apparatus rather then the FIG. 4. Delivery and acid-triggered release of toxin using bispecific antibodies. Diphtheria toxin and CRMs undergo a conformational change at pH Յ 5. Acid-sensitive monoclonal antibodies that recognize the epitopes involved in this transition were disulfide-linked to targeting antibodies that attach to cell surface sites. The resulting bispecific antibody bound the CRM toxin at neutral pH and selectively delivered it to the cell membrane where the complex was taken into cells by endocytosis. Acid conditions within the endosome triggered release of CRM from the antibody and induced its translocation across the vesicle membrane. Once in the cytosol, the catalytic domain of the toxin killed cells by inactivating elongation factor 2 and curtailing protein synthesis. endosome (32,33).
The bispecific antibody approach is very flexible since antibodies or ligands to virtually any membrane site can be easily coupled with the toxin-bearing, pH-sensitive antibody to form a variety of highly specific delivery agents. In addition to using the anti-transferrin receptor monoclonal antibody to target cells, we have also disulfide-linked diferric transferrin (18) to an acid-releasable anti-DT antibody, and this bifunctional construct also mediated CRM cytotoxicity via the same internalization pathway (data not shown). Endocytotic cycling though an acid compartment is required for cytotoxic effectiveness using these DT-based systems, but that condition seems to be the rule for surface receptor sites rather than the exception.
Anthrax toxin operates through an acid-dependent mechanism that is similar to that used by DT (34). One component of anthrax toxin undergoes a pH-induced conformational change and oligomerization to assist the translocation of its effector portion out of the endosome and into the cytosol (35). Thus this toxin or its components might also be targeted to cells and released inside by using an acid-triggered bispecific antibody approach analogous to the one developed for DT.
Whereas the present study utilized bispecific antibodies formed by linking two whole antibodies, we have also made exclusively heterobispecific F(abЈ) 2 hybrid antibodies (36) using the same targeting and acid-sensitive monoclonal antibodies described in this paper. These smaller F(abЈ) 2 reagents (ϳ100 kDa) have proven to be slightly superior delivery vehicles on a molar basis compared with the bivalent whole antibody constructs. Their effectiveness dispels the notion that cross-linking of two surface receptors is necessary for the cellular uptake of bispecific antibody-delivered toxins, since these reagents are monovalent for both the toxin and the cell surface epitope. The production of analogous bispecific single chain Fv reagents (ϳ50 kDa) for toxin delivery is feasible, and their use for therapeutic purposes might offer an advantage.
By providing both the selective delivery to target cells and a pH-controlled intracellular release of biological agents, acidtriggered bispecific antibodies might offer some advantages for therapeutic use. Moreover, the hybrid antibody and active ligand could be administered either as separate components or as a single preloaded complex. However, to maintain a high occupancy of antibody combining sites in vivo, either high affinity antibodies or elevated levels of ligand would have to be used.
The ability to deliver the individual functional domains of DT independently into the same cell using this bispecific antibody delivery system has opened an interesting prospect. The catalytic, transmembrane, and receptor binding domains of DT can be produced by genetic engineering, and separate acidreleasable hybrid antibodies that bind to these constituent domains have been produced. Therefore, we intend to test if those toxin components can cooperate to produce cytotoxicity when they are delivered into the cell as separate entities and are released within the same acidified endosome.